Abstract The present review examines the experimentalevidence supporting the existence of central mechanismsable to modulate the synaptic effectiveness of sensory fi-bers ending in the spinal cord of vertebrates. The firstsection covers work on the mode of operation and thesynaptic mechanisms of presynaptic inhibition, in partic-ular of the presynaptic control involving axo-axonic syn-apses made by GABAergic interneurons with the termi-nal arborizations of the afferent fibers. This includes re-viewing of the ionic mechanisms involved in the genera-tion of primary afferent depolarization (PAD) by GABA-ergic synapses, the ultrastructural basis underlying thegeneration of PAD, the relationship between PAD andpresynaptic inhibition, the conduction of action poten-tials in the terminal arborizations of the afferent fibers,and the modeling of the presynaptic inhibitory synapse.The second section of the review deals with the function-al organization of presynaptic inhibition. This includesthe segmental and descending presynaptic control of thesynaptic effectiveness of group-I and group-II muscle af-ferents, the evidence dealing with the local character ofPAD as well as the differential inhibition of PAD in se-lected collaterals of individual muscle-spindle afferentsby cutaneous and descending inputs. This section alsoexamines observations on the presynaptic modulation oflarge cutaneous afferents, including the modulation ofthe synaptic effectiveness of thin myelinated and unmy-elinated cutaneous fibers and of visceral afferents, as

well as the presynaptic control of the synaptic actions ofinterneurons and descending tract neurons. The third sec-tion deals with the changes in PAD occurring duringsleep and fictive locomotion in higher vertebrates andwith the changes of presynaptic inhibition in humansduring the execution of a variety of voluntary move-ments. In the final section, we examine the non-synapticpresynaptic modulation of transmitter release, includingthe possibility that the intraspinal endings of primary af-ferents also release colocalized peptides in a similar wayas in the periphery. The outcome of the studies presentlyreviewed is that intraspinal terminals of sensory fibersare not hard-wired conductors of the information gener-ated in their peripheral sensory receptors, but dynamicsystems that convey information that can be selectivelyaddressed by central mechanisms to specific neuronaltargets. This central control of information flow in pe-ripheral afferents appears to play an important role in thegeneration of integrated movements and processing ofsensory information, including nociceptive information.

Key words Presynaptic inhibition · Primary afferentdepolarization · Axo-axonic synapses · Volume-transmitted inhibition · GABA · Monoamines ·Neuropeptides · Spinal cord

Introduction

It appears to be a not too far fetched assumption that thecentral nervous system of a mammal or any other verte-brate is continuously exposed to a barrage of afferent im-pulses coming from the various sense organs to such anextent that it exceeds its overall information-processingcapabilities. That is, the central action of the “surplus”afferent impulses has to be reduced or abolished by inhi-bition. Such a suppression of afferent inflow may not on-ly be necessary for the adaptation of the stimulus-inten-sity level (which may range through several orders ofmagnitude), but may also serve many other purposes,such as concentration on relevant afferent inputs by sup-

This review is dedicated to Karl Frank and Sir John C. Eccles inappreciation of their seminal contributions to the discovery of pre-synaptic inhibition in the mammalian spinal cord and in grateful-ness for all the personal encouragement they gave to both authors.

P. Rudomin (✉)Centro de Investigación y de Estudios Avanzados del InstitutoPolitécnico Nacional, Apartado Postal 14–740, México DF 07000,México,e-mail: rudomin@fisio.cinvestav.mx,Tel.: +52-5-747-7099, Fax: +52-5-747-7105

R.F. SchmidtPhysiologisches Institut der Universität, Röntgenring 9, D-97070 Würzburg, Germany

Exp Brain Res (1999) 129:1–37 © Springer-Verlag 1999

R E V I E W A RT I C L E

Pablo Rudomin · Robert F. Schmidt

Presynaptic inhibition in the vertebrate spinal cord revisited

Received: 16 December 1998 / Accepted: 1 June 1999

pressing those which are trivial to the organism at thatmoment (“focusing the attention”). Inhibition of afferentactivity can be exerted at any place in the centripetalpathways, but it appears most economical to do it at theearliest possible sites, even before the unwanted afferentactivity has produced any appreciable disturbance in thecentral nervous system. There are three prime locationsin the afferent pathway that can be used for this task: (1)the sensory receptor, (2) the primary afferent terminal,and (3) the second-order cell. Whereas in invertebratesdirect inhibition of peripheral receptors is a rather com-mon feature, it has been abandoned in the somatosensoryreceptor systems of vertebrates in favor of presynapticinhibition. This offers some additional advantages to re-ceptor inhibition that suppresses all information in theperiphery, whereas presynaptic inhibition may suppressinformation flow in some intraspinal branches, but not inother branches of same afferent and may serve to addressexcitation to selected neuronal targets, according to spe-cific needs in motor control and sensory discrimination.Thus, the presynaptic inhibitory synapses of afferent ter-minals and the postsynaptic inhibitory synapses of sec-ond-order cells are the prime targets for all nervous ac-tivity aimed at inhibiting afferent input in vertebrates.

As will be reviewed later in this article, there is plentyof evidence that presynaptic inhibition via axo-axonicGABAergic synapses acts very strongly on thickly mye-linated primary afferents, but it is probably not or onlyweakly exerted on the terminals of thinly myelinated andunmyelinated afferent fibers as well as in vagal afferentssubserving homeostatic functions (Rudomin 1967; seealso Richter et al. 1986). Regarding the presynaptic in-hibitory control of these fine afferents, increasing evi-dence is accumulating in the literature, and will be re-viewed here, that volume-transmitted (autocrine andparacrine transmitted) substances act at presynaptic sitesin the control of the transmitter release of these centralterminals (see “Non-synaptic presynaptic modulation oftransmitter release”). Recently, in this journal, a reviewon presynaptic inhibition in invertebrate nervous systemshas appeared (Clarac and Cattaert 1996). Also, Miller(1998) has published an excellent review on the molecu-lar aspects of transmitter release, mechanisms of presyn-aptic inhibition, and ionotrophic receptors in the termi-nals of vertebrate neurons. There is, in addition, a recentbook on presynaptic inhibition in the vertebrate nervoussystem, in which most of the available information is re-viewed by leading investigators in the field (Rudomin etal. 1998). And, even more recently, a review has ap-peared in this journal dealing with dorsal-root reflexesand their potential role in enhancing central neurotrans-mission and in triggering peripheral inflammation andhyperalgesia (Willis, 1999). Therefore, the present article– which is far from exhaustive – is mostly concernedwith the spinal cord of higher vertebrates, in order toprovide the readership of this journal a historical per-spective and an update on the various mechanisms andfunctional organization of the presynaptic control oftransmitter release.

The mode of operation and the synaptic mechanismof presynaptic inhibition

The emergence of the concept of presynaptic inhibition

The concept of presynaptic inhibition as we understand ittoday began to take shape in 1957 when Frank and Fuor-tes described a depression of monosynaptic excitatorypotentials (EPSPs) that occurred in the absence of anypostsynaptic potential change or any change in motoneu-ronal excitability. Two alternative explanations were of-fered by Frank (1959) for this finding: either the post-synaptic inhibitory changes took place far out on thedendrites of the motoneuron and could not be “seen”(quotation marks used by Frank 1959) by the intracellu-lar microelectrode, or the inhibitory nerve impulses in-teracted with the excitatory volley before the latter hadarrived at the surface of the motoneuron and blockedtransmission in the terminal fibers. Since both of theseexplanations would locate inhibitory action at a site re-mote from the motoneuronal soma, Frank (1959) desig-nated the phenomenon “remote inhibition”.

The pioneering observations of Frank and Fuorteswere taken up and extended by Eccles and his associatesin Canberra and by Lundberg and his associates inGoteborg (Eccles et al. 1961; Eide et al. 1968). Theseauthors added the observations that, during “remote” in-hibition, there was no detectable change in the timecourse of the depressed monosynaptic Ia-induced EPSPand no conductance change of the motoneuronal mem-brane, whereas at the same time the size and the timecourse of the monosynaptic EPSP generated by volleysin descending tract were unaltered, as shown in Fig. 1A,B (see also Rudomin et al. 1975, 1991).

The lack of changes in the time course of the inhibit-ed Ia monosynaptic EPSP has often been considered as atest for presynaptic inhibition. This view is based on theassumption that the falling phase of the monosynapticEPSP is a reliable indicator of the contribution of remoteinhibitory conductances. However, according to Carlenet al. (1980), if the EPSP were entirely generated at a re-mote site and the conditioning conductance increase alsotook place at the same site, the major fraction of theEPSP time course would nevertheless depend on theproperties of the proximal membrane and remote chang-es would be less detectable. Yet, as shown by McCrea etal. (1990), large reductions in composite monosynapticEPSP amplitude following conditioning stimulation can-not be accounted for by increased postsynaptic conduc-tances in motoneurons. Therefore, the finding that mono-synaptic EPSPs evoked in the same motoneurons at asimilar electronic distance by descending volleys are notdepressed by those conditioning stimuli that depress theIa EPSPs without changing their time course may beconsidered to be compelling evidence for the presynapticorigin of the depression of the Ia EPSPs.

2

Relating primary-afferent depolarization (PAD) to presynaptic inhibition

A variety of hypotheses had been put forward to explainthe mechanism of primary-afferent depolarization (PAD)and its possible correlation with presynaptic inhibitionbefore Eccles and his associates postulated that presyn-aptic depolarization is responsible for the EPSP depres-sion because it reduces the size of the presynaptic im-pulse and, hence, decreases the liberation of the excitato-ry transmitter. Initially, Barron and Matthews (1935,1938a, 1938b) postulated that the prolonged (200 ms)positive potential (P-wave) that could be recorded fromthe surface of the cord dorsum following an afferent vol-ley (Gasser and Graham 1933) was produced by thesame potential generator as the negative dorsal-root po-

tential (DRP) led off from a dorsal rootlet. It was con-cluded that these potentials reflected a prolonged depo-larization of the dorsal-root fibers, and this has been ac-cepted by all subsequent investigators. However, therewas much diversity in the attempts to explain the mannerin which DRPs were produced (for reviews, see Schmidt1971; Burke and Rudomin 1977).

In the 1960s, the work of Eccles and his colleaguesreestablished the association of primary-afferent depolar-ization with reflex depression, adding detailed postulatesboth on the mode of generation of PAD by the activationof axo-axonic synapses and on the mechanism and modeof operation of the resulting presynaptic inhibition (seeEccles 1957, 1964 for reviews).

Presynaptic depolarization and inhibition by potassium ions

Barron and Matthews (1938b) suggested that the intra-spinal endings of the afferent fibers can be depolarizedby accumulation of potassium ions in the extracellularspace because of the increased interneuronal activity, butthis possibility was not thoroughly investigated until thelate 1970s with the introduction of potassium-sensitivemicroelectrodes (Walker 1971).

Tetanic stimulation of peripheral nerves induced anincrease in extracellular potassium in the spinal cord,from a resting level of about 3 mM, to around 9 mM. Po-tassium transients exhibited a spatial distribution in thespinal cord with maxima in the dorsal horn and in the in-termediate nucleus. The extracellular accumulation ofpotassium appeared to have a ceiling that could not beelevated by a further increase in the frequency or intensi-ty of stimulation, suggesting that the processes of potas-sium release from intracellular sources and its removalfrom the extracellular space were in equilibrium (Brug-gencate et al. 1974; Kriz et al. 1974; see also Somjen andLothman 1974).

These observations revived the original proposal ofBarron and Matthews, and it was proposed that accumu-lated potassium in the extracellular space of the spinalcord produces depolarization and increased excitabilityof primary-afferent terminals (Kriz et al. 1974, 1975; So-mjen and Lothman 1974). Subsequent pharmacologicaland physiological observations indicated, however, thatextracellular accumulation of potassium ions was not theonly source of PAD in the spinal cord. Picrotoxin, aGABAA antagonist, was found to depress PAD andDRPs, even though there was an increase in potassiumconcentration. In contrast, pentobarbital increased theDRPs and reduced potassium accumulation. Moreover,stimulation of cutaneous fibers was found to inhibit thePAD produced in Ia fibers by stimulation of group-I af-ferents, even though this procedure further increased theextracellular concentration of potassium ions (Bruggen-cate et al. 1974; Jiménez et al. 1984).

A more direct approach to the question of whether orto what extent accumulation of the extracellular potassi-

3

Fig. 1A–C Differential inhibition of Ia and descending EPSPs.A Effects of posterior biceps-semitendinosus-nerve- (PBSt-) con-ditioning stimulation on Ia EPSPs produced by stimulation of thegastrocnemius (GS) nerve. CDP Cord dorsum potential, IC-EC in-tracellular potential recorded from motoneuron minus extracellu-lar field, Diff conditioned Ia EPSP obtained by subtracting, fromtrace 2, the potential changes produced by PBSt stimulation alone(not illustrated). Dots in trace 1 were obtained by multiplyingEPSP in lowest trace by a factor of 1.51. B Effects of same PBSt-conditioning stimulation on monosynaptic EPSPs produced bystimulation of the ventro-medial fasciculus (VMF). Same formatas A. Dots in trace 1 are EPSP of lowest trace multiplied by a fac-tor of 1. The diagram in C shows experimental paradigm and neu-ronal pathways mediating pre- and postsynaptic inhibition. Pre-synaptic inhibition is mediated by last-order GABAergic interneu-rons synapsing with Ia GS terminals (Class II interneurons). Post-synaptic inhibition is mediated by glycinergic (Class I) interneu-rons; see the section “Selective presynaptic control of group-I af-ferents” for further explanations. Reproduced with kind permis-sion from Rudomin et al. (1991) and Rudomin (1992)

um plays a role in presynaptic inhibition came from thestudies of Jiménez et al. (1983, 1984). These investiga-tors measured the PAD produced in single muscle affer-ents, as well as the changes in potassium concentrationproduced at the same site, following separate stimulationof muscle and cutaneous afferents. They found no directcorrelation between the magnitude of the PAD evokedby stimulation of sensory nerves and the presumed con-centration of potassium ions accumulated around the siteof excitability testing.

Further evidence excluding potassium accumulationas a major contributor to the generation of PAD was ob-tained by examining the action of rubrospinal and reti-culospinal fibers on the intraspinal threshold of cutane-ous afferents (Jiménez et al. 1987). Following repetitivestimulation of these descending fibers, an increased ex-tracellular concentration of potassium was measured inthe intermediate nucleus and in the motor pool, but notin the dorsal horn where cutaneous afferents terminate.Yet, there was a strong PAD of cutaneous fibers, whichwas abolished following the i.v. injection of picrotoxin,as expected if this PAD were mediated by activation ofGABAA receptors.

Recent evidence suggests, however, that the cyclicPAD that occurs during fictive locomotion in the new-born rat may be accounted for, at least in part, by extra-cellular accumulation of potassium ions (see Kremer andLev-Tov 1997, and the section “Presynaptic inhibitionand behavior”). However, it must be kept in mind thatthis situation could be different in the adult because ofthe development of different types of GABAergic recep-tors that may play a more important role in reflex inte-gration during fictive locomotion, as suggested by thefindings of Wu et al. (1992).

The histological basis for presynaptic inhibition: axo-axonic synapses

The anatomical evidence presently available supports theassumption of three distinct mechanisms for primary af-ferent control in the mammalian spinal cord: axo-axonicinteractions, dendro-axonic interactions, and paracrinecommunication. This section will deal with the first ofthese three, and the last two will be discussed in the sec-tion “Non-synaptic presynaptic modulation of transmit-ter release”. Very little is known about the functionalrole of dendro-axonic interactions and, therefore, theywill only briefly be mentioned here.

When Gray (1962) published the first illustrations ofaxo-axonic synapses, his figures were greeted with somerelief in Canberra, where Eccles and his associates hadbeen postulating the existence of such contacts for someyears. It took another seven years before primary affer-ent terminals were positively identified as synapticallylinked to the motoneuron membrane and themselvespostsynaptic to a further terminal (Conradi 1969). Thisfinding is now thoroughly established (for a recent re-view, see Alvarez 1998). In addition to group-Ia afferent

terminals, there is now evidence for axo-axonic synapsesmade by GABAergic interneurons on the intraspinal ter-minals of group-Ib and -II muscle afferents, as well as onthe terminals of large cutaneous primary afferents in thedorsal horn of the lumbar spinal cord and in the trigemi-nal nuclei. The main observations are summarized below(for additional literature, see Alvarez 1998).

Axo-axonic synapses on group-Ia primary afferent fibers

Exploring the quantitative ultrastructure of Ia boutons inthe ventral horn, Pierce and Mendell (1993) found that,of 35 Ia boutons, 86% received at least one axo-axonalcontact and that the extent of this contact was directlyproportional to the size of the bouton (see also Fyffe andLight 1984). In Clarke’s column, the number of presyn-aptic contacts on Ia boutons is smaller, only about 10%(Walmsley et al. 1987, 1995), but, as the latter authorsshowed with computer simulations of presynaptic inhibi-tion based directly on the serial-section-EM-reconstruc-tions, it is likely that the effect of a presynaptic contacton membrane potential and action-potential amplitudeextends beyond the contacted bouton to other boutonsalong the branches of the Ia axon (however, see Gossard1996; Quevedo et al. 1997).

Destombes et al. (1996) analyzed the distribution ofGABA-like immunoreactivity in axon terminals apposedto cat motoneurons in the lumbar column. They foundthat, on somatic and proximal dendritic membranes of α-motoneurons, about 20% of F type boutons (boutonswith flattened and/or irregularly shaped vesicles assumedto contain inhibitory transmitter; see Uchizono 1975)contained GABA. Furthermore, all P terminals, i.e., ter-minals presynaptic to the large M boutons, which are as-sumed to arise from Ia afferents (see Fyffe and Light1984; Destombes et al. 1992), also contained GABA.The M-type boutons were observed mainly on proximaldendrites and were usually contacted by one or two Pterminals. None of the M boutons showed any labelingfor GABA. In agreement with these observations on α-motoneurons, it has been reported that GABA is alsopresent in presynaptic synapses located on the terminalsof Ia afferents contacting unidentified interneurons inlaminae VI (Maxwell et al. 1990).

Axo-axonic synapses on group-Ib primary afferent fibers

Lamotte d’Incamps et al. (1997, 1998) labeled the intra-spinal portion of functionally identified Ib fibers fromthe posterior biceps and semitendinosus muscle of theanesthetized cat with intraaxonal injection of a fluores-cent dye (tetramethyl rhodamin dextran). On two collat-erals of a single Ib fiber that were reconstructed in theintermediate zone using confocal microscopy, there wereat least 69 axo-axonic contacts of GABAergic interneu-rons, most of them located in or close to the terminals.An example is shown in Fig. 2. On two collaterals of this

4

Ib fiber, 59 axo-axonic contacts were counted. A 3-D re-construction of one collateral showed that the contactswere not distributed uniformly over the intraspinal arbo-rization: they were mostly located on, or close to, termi-nals (as in the example shown in the figure) and, inter-estingly, some branches received less contacts than oth-ers. This work demonstrates the presence of axo-axoniccontacts on Ib fibers in the spinal cord and suggests thatIb fibers bear adequate synaptic equipment to receivesubstantial presynaptic inhibition, as suggested by elec-trophysiological observations (see the section “Selectivepresynaptic control of group-I afferents”).

Axo-axonic synapses on group-II primary afferent fibers

Quite recently, Maxwell and Riddell (1999) have beenable to label with HRP functionally identified group-IImuscle afferents supplying the gastrocnemius or thesemitendinosus muscles. They examined the termina-tions in the dorsal horn (laminae IV and V) as well as inthe intermediate gray matter (laminae VI and VII) andfound that all group-II boutons were postsynaptic to oth-er axon terminals. An example is shown in Fig. 3A.Quite often, more than one axon was presynaptic to asingle group-II axon terminal. Presynaptic boutons wereoccasionally components of synaptic triads (a singlebouton was presynaptic to a group-II axon and a dendritethat was also postsynaptic to the group-II axon). Immu-nostaining for GABA and glycine showed that all axons

in presynaptic apposition to group-II boutons containedGABA (Fig. 3B). Glycine was co-localized in many ofthese boutons. These findings support the proposal thatPAD is generated in group-II fibers by activation ofGABAergic interneurons (see “Presynaptic inhibition ofgroup-II muscle afferents”, below). At the present time,there is no documented action of glycine on group-II af-ferents. However, glycine could have some action on theinterneurons receiving inputs from the group-II afferents(Maxwell et al. 1997), as suggested by the observationsof Jonas et al. (1998) in the isolated spinal cord of thenewborn rat. According to these investigators, unitaryIPSPs generated at interneuron-motoneuron synapses re-sult from the co-release of glycine and GABA, which ac-tivates functionally distinct receptors in their postsynap-tic target cells.

Axo-axonic synapses on cutaneous primary-afferentfibers

For primary-afferent endings and their synaptic glomeru-li in the dorsal horn, a large body of work has concen-trated on their structure, synaptology, typology, and asso-ciation with various sensory modalities (for recent re-views, see Carlton 1994; Ribeiro-da-Silva 1995). Assummarized by these authors and by Alvarez (1998), itturned out that axo-axonic synapses do not representmore than 1–4% of all synaptic interactions in laminae IIand III. And at the same time, it has been established thatthe density of axo-axonic as well as dendro-axonic con-tacts differs among primary afferent terminals and thatseveral types of synaptic configurations can be discrimi-nated from each other.

Using gold immunostaining of GABA, glutamate, andglycine, Maxwell et al. (1995) found, in laminae II andIV, large glutamate-containing boutons from presumedlow-threshold cutaneous afferents that were postsynapticto other axon terminals. Analysis of serial sections re-vealed that the presynaptic boutons contacting the affer-ent fibers were immunoreactive for GABA and could,therefore, mediate presynaptic inhibition of large cutane-ous afferents.

The synaptic interconnections between afferent fibersand interneurons in the dorsal horn of the rat seem to berather complex. According to Todd (1996), the superfici-al dorsal horn of rat spinal cord contains two types ofsynaptic glomeruli, which are centered around the termi-nals of unmyelinated and myelinated primary afferents,respectively. Both types of glomeruli contain GABA-

5

Fig. 2 Axo-axonic contact on an Ib fiber from the posterior bi-ceps-semitendinosus of the cat. The single Ib fiber was labeled byintra-axonal injection of a fluorescent dye (tetramethylrhodamin-dextran, fluorescing in red) in the intraspinal portion of the fiber.S1–L7 spinal cord sections were processed for demonstration ofGABA immunoreactivity with isothiocyanate fluorescein (fluo-rescing in green) and scanned using a dual-beam laser confocalmicroscope to detect the two fluorochroms. Two successive confo-cal planes, spaced 0.5 µm apart, are show on the left part of thefigure (calibration bar = 5 µm). A GABAergic structure (green) isin close apposition with the fiber (red), as suggested by the yellowcolor, indicating that the two fluorochroms were detected withinthe same voxel (0.3*0.3*0.5 µm3). The yellow zone was visible ontwo serial confocal planes, showing that it was not due to a ran-dom vicinity of the afferent with the GABAergic structure. Theright part of the figure is the superimposition of ten successiveconfocal planes at 2 µm intervals (excitation of rhodamin only),showing a portion of the Ib collateral in a branching zone. The ar-row points to the location of the contact shown in the left part ofthe figure. Courtesy of D. Zytnicki et al.; cf. Lamotte d’Incamps etal. (1997) and (1998)

ergic axons and dendrites, which are thought to originatefrom local inhibitory interneurons. Some of the dendritescontain synaptic vesicles and may be presynaptic to thecentral axon at dendro-axonic synapses. The function ofthese dendro-axonic synapses has not been yet deter-mined, but they could produce PAD just as axo-axonicsynapses do. It is also remarkable that, in type-I glomer-uli (made by unmyelinated afferents), all of the peripher-al axons and most vesicle containing dendrites wereGABA-immunoreactive and that only one of 32 axonsand none of the vesicle-containing dendrites was gly-cine-immunoreactive. In contrast, most of the peripheralaxons and some of the vesicle-containing dendrites intype-II glomeruli possessed both GABA- and glycine-immunoreactivity. On the other hand, unmyelinated (pre-sumably nociceptive) afferents in type-I glomeruli ap-pear to receive few axo-axonic synapses from GABA-ergic interneurons.

Axo-axonic synapses in the dorsal column nuclei

It is generally accepted that presynaptic inhibition playsa major role in the sensory processing within the gracilisand cuneate nuclei (Wall 1958; Andersen et al. 1964a,1964b, 1964c, 1964d). The axon terminals of the inter-neurons are believed to be in synaptic contact with the

terminals of the primary afferent fibers as well as withdendrites of relay neurons receiving synapses from theafferent fibers (Walberg 1965). The terminals of the cor-ticofugal fibers ending in the cuneate appear to synapsewith interneurons (Andersen et al. 1964e), which, inturn, would make presynaptic contacts with terminals ofprimary afferent fibers. Additional evidence supportingthese early observations has been provided more recentlyby Ellis and Rustioni (1981), who found that the termi-nals of primary afferents ending within the cuneate nu-cleus are postsynaptic to small boutons containing flat-tened vesicles (Fyffe et al. 1986).

The generation of PAD by GABAergic axo-axonicsynapses

Mode of operation of the axo-axonic synapse generatingPAD and presynaptic inhibition

It is now generally agreed that the PAD associated withpresynaptic inhibition is produced by axo-axonicGABAergic synapses having GABAA-type receptors.The histological evidence supporting this view was pre-sented in the previous section. Pharmacologically, it wasrecognized long ago that presynaptic inhibition and PADare both reduced by the GABA antagonists picrotoxinand bicuculline (Eccles et al. 1963c; Schmidt 1963; Da-vidoff 1972; Barker and Nicoll 1973), and, based ofthese results, it has been suggested that the effects ofGABA are mediated through GABAA-type receptors,i.e., GABA receptors coupled to chloride (Cl––ion) chan-nels, as shown in Fig. 4 (for reviews, see Nicoll and Al-ger 1979; Nistri 1983; Davidoff and Hackman 1983,1984).

The prevailing hypothesis about the cellular mecha-nisms of presynaptic inhibition is that GABA inhibits therelease of neurotransmitter, most likely glutamate, fromprimary afferent axons, either by blocking action-poten-tial invasion into their terminals or by reducing the am-

6

Fig. 3A, B An axo-axonic synapse on a terminal of a group-II(GpII) muscle afferent. The muscle afferent was characterized ac-cording to electrophysiological criteria, intra-axonally labeledwith horseradish peroxidase, and prepared for combined light andelectron microscopy. A The group-II bouton can be recognized bythe characteristic reaction product contained within it. It is post-synaptic to a small axon terminal (*), which contains irregularvesicles. The inset is taken form a serial section and shows detailsof the synapse. B An immunogold labeling of GABA was per-formed on an adjacent section. The presence of small gold parti-cles on the presynaptic terminal (*) shows that it was immunore-active for GABA. Scale bar 1 µm. From unpublished work by D.J. Maxwell and J.S. Riddell

plitude of propagated action potentials, thereby blockingor reducing Ca2+ influx. Actually, the blocking of action-potential invasion or the reduction in its amplitude isthought to result mainly from the depolarization of theterminals and, to a lesser extent, from the change in con-ductance produced by GABA (see Miller 1998). Directproof for this hypothesis at the axo-axonic synapses ofvertebrate primary afferents is still lacking, but convinc-ing results are available from the secretory nerve termi-nals from rat posterior pituitary, where it has been shownthat activation of GABAA receptors gates a Cl–-channeland that action potentials are little affected when activa-tion of the GABAA receptor produces only conductancechanges and no voltage changes (Jackson and Zhang1995; Zhang and Jackson 1995a, 1995b). In the spinalcord of the bullfrog, both pre- and postsynaptic inhibi-tion at synapses between muscle-spindle afferents andmotoneurons are caused by activation of GABAA recep-tors, and the presynaptic inhibitory effects are accompa-nied by reduction of the action potentials invading theafferent terminals (Peng and Frank 1989a).

As reviewed by Miller (1998), other proposed mecha-nisms for presynaptic inhibition are that the reduction ofpresynaptic Ca2+ currents is achieved through activationof GABAB receptors via a second messenger (Dolphin etal. 1990), or by activation of K+ currents also mediatedthrough GABAB receptors (Gage 1992). However,GABAB receptors seem to play a minor, if any role at, allfor PAD generation (Stuart and Redman 1992). Other

speculations include GABA-induced depolarization inac-tivating high-threshold Ca2+ channels (Graham and Red-man 1994) or low-voltage-activated Ca2+ channels(Walmsley et al. 1995). In either case, the end resultwould be a reduction in Ca2+ influx, thereby contributingto presynaptic inhibition.

Thompson and Wall (1996) found that, in the rat spi-nal cord, picrotoxin only reduced the DRPs produced bydorsal-root stimulation to 60% of their control ampli-tude. The GABAB antagonist, CGP 36742, and antago-nists specific to the serotonin receptor 5-HT1A(MDL73005EF) and to the 5-HT3 receptor (granisetron)had no significant effect on the DRPs. However, methy-sergide, a 5-HT2 antagonist, significantly reduced theDRP to 71% of its control level. Combined picrotoxinand methysergide reduced the DRP to 20% of controllevel. On the basis of these results, it was proposed thatboth GABA- and 5-HT2-receptor mechanisms play arole in generating the DRP, although it is possible thatmethysergide acts as a partial agonist (Peroutka 1988).

Lopez-Garcia and King (1996) observed in the 10- to14-day-old rat lumbar dorsal horn in vitro that 5-HT aswell as GABA depressed the spontaneous dorsal-rootpotentials and that, in most cases, excitatory synaptictransmission from primary afferents to dorsal-horn neu-rons was depressed by 5-HT. However, there was noclear association between the magnitude of the PAD gen-erated by 5-HT and the depression of the evoked synap-tic potentials. Since 5-carboxamido-tryptamine, an ago-nist for 5-HT1 receptors, mimicked the depression ofsynaptic transmission in the dorsal horn without produc-ing PAD, it was suggested that neither PAD nor changesin basic membrane properties of the neurons account forthe observed 5-HT-induced depression in synaptic trans-mission, which could be mediated by a 5-CT-sensitivereceptor.

In the cat spinal cord with intact neuraxis, Jan-kowska and collaborators have shown that 5-HT de-presses synaptic actions of group-II muscle afferentsand that there might be a presynaptic component in thiseffect (see the section “Presynaptic inhibition of group-II muscle afferents”), because PAD was evoked bystimulation in raphe (and also in Locus Coeruleus).However, it was not possible in these experiments toexclude the possibility that PAD was caused by fiberspassing or terminating in these nuclei and not by sero-tonin- or noradrenaline-releasing neurons (Jankowska,personal communication).

Dorsal-root reflexes evoked by primary-afferentdepolarization (PAD)

Primary-afferent depolarization evoked by large and syn-chronous afferent inputs is often so intense that it trig-gers action potentials in the terminal areas of the afferentfibers being depolarized. These action potentials are an-tidromically conducted to the dorsal roots and in the pe-riphery. Somewhat misleadingly, they have been termed

7

Fig. 4 Schematic representation of a presynaptic inhibitory axonestablishing synaptic contact with a primary afferent terminal,which in turn makes synaptic contact with a motoneuronal den-drite. A microelectrode inserted in the terminal of the primary af-ferent records the depolarization produced by GABA. This depo-larization is due to Cl– efflux. The depolarization activates K+

channels with consequent K+ efflux. The outwardly directed Cl–

gradient is reestablished by the operation of the Na+-K+-2Cl– co-transporter. Osmotic water fluxes are indicated with dashed ar-rows. Reproduced with kind permission from Alvarez-Leefmanset al. (1998)

dorsal-root reflexes, DRR (Barron and Matthews 1938a;Toennies 1938).

Dorsal-root reflexes can be best recorded in cutane-ous nerves following stimulation of cutaneous afferentfibers. They can also be recorded in muscle nerves whenmuscle afferent fibers are stimulated, and they can beevoked in muscle nerves by cutaneous volleys and viceversa (for literature, see Schmidt 1971). DRRs can beseen in awake, behaving animals during locomotion (fora review, see Rossignol et al. 1998), and they are presentin humans (Shefner et al. 1992). Taken together, thesefindings suggest that the DRR may participate in thecontrol of somatosensory afferent inflow to the centralnervous system, but direct proof for such a role of DRRsis still not available (see Rossignol et al. 1998).

There is, however, recent evidence that DRRs maycontribute to the development of neurogenic inflamma-tion in the course of experimental arthritis and otherpathophysiological tissue states. As recently summarizedby Willis et al. (1998) and Willis (1999), he and his asso-ciates have presented evidence that antidromic discharg-es in fine articular afferent fibers (Aδ- as well as C-fibers) release peptides and possibly other substances in-to the inflamed joint tissue, and that half of the swellingof the knee joint that is seen in acute arthritis in the rat isapparently due to the contribution of DRRs. From thesefindings, it appears that this may be a major, but previ-ously unrecognized positive-feedback mechanism ofneurogenic inflammation and edema. This mechanismmay also be involved in the development of neurogenicinflammation in uninjured tissue, e.g., in a knee joint.

GABA-mediated presynaptic inhibition without PAD

There are several reports in the literature that presynapticendings of primary afferents, in addition to GABAA re-ceptors, also have GABAB receptors and that activationof these receptors reduces transmitter output without anaccompanying PAD. For instance, in the spinal cord ofthe bullfrog (±)-baclofen (a specific GABAB-receptoragonist) reversibly reduced the amplitude of EPSPsevoked monosynaptically in motoneurons without anychange in their time course or in the time course or theamplitude of the action potentials in the sensory afferents(Peng and Frank 1989b). Similar effects have been de-scribed in the cat spinal cord (Curtis et al. 1981; Lev-Tovet al. 1988; Jiménez et al. 1991). In the lamprey spinalcord, it was shown that GABAB receptors at presynapticterminals of sensory afferents can be activated by GABAand that this activation leads to a reduction in the dura-tion of the presynaptic action potential and a consequentreduction in transmitter release via a G-protein-coupledmechanism (Alford and Grillner 1991).

Similarly, a reduction of presynaptic Ca2+ inflow witha consequent shortening of the preterminal action poten-tial and a reduction in transmitter release seems to be responsible for the presynaptic inhibitory effects of (–)-baclofen on spinal motor reflexes, as recently dem-

onstrated by Curtis and his associates (Curtis et al. 1997;Curtis 1998; Curtis and Lacey 1998). This action ofbaclofen is exerted via GABAB receptors, which uponactivation appear to interfere with the operation of volt-age-activated calcium channels directly associated withthe transmitter-release mechanism. The effects ofbaclofen both on action-potential duration and on trans-mitter release are not accompanied by any PAD, andboth effects can be blocked by baclofen antagonists(without any effect on PAD).

The effectiveness of the GABAB-mediated, “non-PAD” presynaptic mechanism relative to the one exertedvia PAD remains to be determined. Any such evaluationhas to take into account the relatively minor side effectsof baclofen when administered systemically to humansin doses that attenuate spasticity of spinal origin and thefact that administration of baclofen to mammals produc-es minimal behavioral effects, with little or no impair-ment of motor function (Bowery 1993; Mott and Lewis1994), which is in marked contrast to the convulsive ef-fects of picrotoxin and bicuculline. However, it must beconsidered that, under defined experimental conditionsin decerebrate cats, baclofen produces a significant de-crease in the soleus stretch-reflex stiffness accompaniedby an increase in stretch-reflex threshold (Capaday1995), both effects being convincingly attributed by theauthor to a baclofen-induced tonic presynaptic inhibitionof the intraspinal terminals of the muscle-spindle affer-ents.

Nonpassive Cl– distribution across the primary afferent membrane

The hypothesis outlined above that PAD results from anefflux of Cl– ions implies that the steady-state intracellu-lar Cl– concentration in the primary afferent terminals ishigher than that predicted from a passive distribution andthat this is due to an active transport mechanism thatmaintains the outwardly directed Cl– electrical gradientneeded for depolarization to ensue. As was first shownby Alvarez-Leefmans et al. (1988) and Alvarez-Leef-mans (1990), the underlying mechanism is an inwardlydirected Cl–-transport system coupled to Na+ and K+,i.e., not a chloride pump driven directly by energy de-rived from ATP splitting, but a secondary active-trans-port system (see Fig. 4).

Most studies of the mechanism underlying PAD havebeen performed in the cell body of sensory neurons, and,as reviewed by Alvarez-Leefmans et al. (1998), the useof this model has been shown to be fully justified sinceGABAA receptors are present both in the intraspinal ter-minals and in the cell bodies of primary sensory neuronsof vertebrates and since, in both locations, they producea depolarization underlying an electrodiffusional Cl– ef-flux. Moreover, drugs known to affect Cl–-transportmechanisms in the cell body seem to affect similarmechanisms in the terminals (see reviews by Nicoll andAlger 1979; Alvarez-Leefmans 1990).

8

The experiments of Alvarez-Leefmans and his associ-ates on frog dorsal-root ganglion cells quoted abovedemonstrated that the intracellular activity of chlorideions is about 2–3× higher than expected for an electro-chemical equilibrium distribution in sensory neurons.Their analysis of the mechanisms by which this outward-ly directed Cl– gradient is established and maintainedshowed that the DRG membrane has a low Cl– restingpermeability and that the movements of Cl– occurredlargely through a non-electrodiffusional pathway, i.e., anelectroneutral carrier mechanism. Furthermore, theyfound that the steady-state intracellular activity levels ofchloride ions depended on the simultaneous presence ofextracellular Na+ and K+. Similarly, the active reaccumu-lation of Cl– after intracellular Cl– depletion required thesimultaneous presence of Na+ and K+ in the extracellularmedium. These results can best and unequivocally ex-plained with the presence of a Na+-K+-Cl–-co-transporterin the soma and in the terminals of primary afferents.

Factors contributing to restoring ionic gradients and osmotic equilibrium after activation of the axo-axonic synapse

The present concepts on the mechanism of operation ofaxo-axonic synapses generating PAD and presynaptic in-hibition are summarized in Fig. 4. The primary afferentfiber is depolarized by the release of GABA from thepresynaptic axon. The depolarization of up to 25 mVfrom the resting potential is due to Cl– efflux through an-ion channels coupled to GABAA receptors. The magni-tude of this depolarization is sufficient to produce open-ing of voltage-activated K+ channels (for literature onthis point, see Alvarez-Leefmans et al. 1998) with a con-sequent K+ efflux, thus moving the membrane potentialback to its resting value.

The net result of the activation of the axo-axonic syn-apse should be a loss of intracellular Cl– and K+ accom-panied by osmotic water efflux. Given the small volume-to-surface ratio of the terminals, this net solute and waterefflux means that they should have a tendency to shrinkupon GABA action. This is most likely prevented by theNa+-K+-Cl–-co-transport, which becomes activated in re-sponse to either or both: reduction in intracellular K+ andthe shrinkage itself, thereby restoring ionic gradients andosmotic balance in the terminal. True, the activation ofthe Na+-K+-Cl–-co-transporter implies intracellular Na+

gain. But the Na+ is immediately extruded via the Na+-K+-pump, thereby keeping the intracellular Na+ at a con-stant level.

Regarding the Na+-K+-Cl–-co-transporter itself, threefurther details need to be mentioned here: first of all, it issensitive to “loop” diuretics such as furosemide (see Fig.4) and insensitive to Na+-K+-ATPase inhibitors. Second,the Na+-K+-Cl–-co-transport can be regulated by varioushormones, neurotransmitters and growth factors via in-tracellular messengers (for literature, see Alvarez-Leef-mans et al. 1998). And third, recent molecular cloning

and functional expression studies have revealed that theNa+-K+-Cl–-co-transporter belongs to a family of struc-turally related salt-transport proteins (cf. Plotkin et al.1997).

Conduction of action potentials in terminal arborizations

The issue of whether the action potentials are securelypropagated along the entire intraspinal axonal trajecto-ries of the afferent fibers, or if the nerve impulse propa-gating in the parent axon does not invade the entire arbo-rization and the extent to which this is modified duringPAD, has been a matter of debate for more than four de-cades. In contrast to graded modulation at the level ofsingle synapses, conduction failure would be a mecha-nism controlling synaptic release at single synapses or asmall number of synapses (Lüscher 1998).

Experimental (Parnas 1972; Smith 1980; Stockbridgeand Stockbridge 1988; Lüscher et al. 1994a, 1994b) aswell as theoretical studies (Goldstein and Rall 1974; Par-nas and Segev 1979; Stockbridge 1988; Lüscher andShiner 1990a, 1990b; Manor et al. 1991a, 1991b) sug-gest that changing core conductor geometry at axonbranch points or other inhomogeneities alters impulseconduction due to electrical impedance mismatch andthe action potential may eventually fail to propagate be-yond the enlargement or branch point. However, withoutspecific control mechanisms of impulse propagation,morphological variations in the axonal structure alonewould not be a very useful mechanism for modulatinginformation transmission in a systematic and predictablemanner.

Already in 1955, Howland et al. provided evidencethat, during the development of the dorsal-root poten-tials, invasion of the intraspinal arborizations of the af-ferent fibers by action potentials would be curtailed.Conduction failure in the spinal arborizations of the af-ferent fibers was used to explain the inhibition of spinalreflexes that is associated with DRPs. Quantal analysisof monosynaptic EPSPs produced in spinal motoneuronsby single Ia fibers has shown occasional failures, whichappear to be increased during presynaptic inhibition. Ithas been suggested that transmission failure is the resultof failure of transmitter release from the presynapticbouton (Redman 1990) or, alternatively, of propagationfailure of the action potential somewhere along the tra-jectory of the afferent fiber (Henneman et al. 1984;Lüscher 1990). However, until now there is no direct ev-idence favoring propagation failure as a normal featureof impulse transmission in the intraspinal arborizationsof afferent fibers. Also, evidence for conduction failureduring presynaptic inhibition is far from having been es-tablished (see below).

Using spike-triggered averaging techniques, Lüscheret al. (1983) analyzed functional connections betweensingle Ia fibers or group-II spindle-afferent fibers andhomonymous motoneurons in the cat spinal cord (Cla-mann and Lüscher 1985). In animals with acutely

9

transected spinal cords, there was an increase in the pro-jection probabilities of Ia afferents, especially from smallafferent fibers to small motoneurons. At the same time,the EPSPs elicited by large Ia fibers increased in ampli-tude. To explain these findings, Henneman et al. (1984)suggested that transmission failures occurred throughoutparticular anatomical synaptic contact systems and thatsome of these transmission failures were probably re-lieved immediately after spinal cord transection. Theyfurther suggested that the relief of transmission failurewould be, at least in part, the loss of presynaptic inhibi-tion from a more cranial input.

Solodkin et al. (1991) approached the question ofconduction failure from Ia afferents to motoneurons dur-ing presynaptic inhibition by simultaneously recordingfrom pairs of motoneurons the monosynaptic EPSPs pro-duced by the same Ia afferent, made to discharge at10–20 Hz by stretching the homonymous muscle. Theyreasoned that, if there was intermittent conduction blockat branch points in the terminal arborizations, failures inthe Ia EPSPs recorded from both motoneurons should becorrelated. Observations made in several pairs of moto-neurons failed to show any correlation between the fluc-tuations of pairs of single-fiber Ia EPSPs in control con-ditions as well as during presynaptic inhibition. Thissuggested that, under those particular conditions, therewas no conduction failure in the Ia branches innervatingboth motoneurons. However, as pointed out by these in-vestigators, detection of an existing correlation betweenthe fluctuations of the two sets of monosynaptic EPSPs,if any, could be obscured by non-linear interactions be-tween the EPSPs and the background synaptic activity ofthe motoneurons introduced via pathways that were acti-vated by stretching the muscle.

More recently, Castillo et al. (1998) have examined inthe cat the extent to which PAD might prevent invasionof action potentials to axonal and/or terminal branches ofsingle group-Ia afferents ending in different spinal seg-ments. To this end, they measured the changes in the in-traspinal threshold of one L6 collateral in the intermedi-ate nucleus and one L3 collateral within Clarke’s columnof single group-I gastrocnemius afferent fibers. Stimula-tion of one collateral with suprathreshold pulses in-creased the intraspinal threshold of the other collateralwhen applied at conditioning-testing time intervalsshorter than 2–3 ms, because of the refractoriness pro-duced by the action potential invading the distal collater-al (see Curtis et al. 1995, 1997). During the PAD pro-duced by conditioning stimulation of the posterior bicepsand semitendinosus (PBSt) nerves, the threshold in thedistal collateral was still increased at short time intervalsbetween stimuli, suggesting that this collateral continuedto be invaded by the action potentials generated in theother collateral. Similar results have been reported byCurtis et al. (1995). It could be argued that changes inthe intraspinal threshold allow measurement of eventsoccurring in the relative coarse terminals, while conduc-tion failure in the unmyelinated terminals would remainundetected. Nevertheless, it is clear from these experi-

ments that conduction in axonal branches in the dorsalcolumns, and probably also in the pre-terminal fibers ofgroup-I muscle afferents, is not impaired during PAD.

On the other hand, Wall (1995) reported that, in the ratspinal cord, conduction of action potentials in myelinatedcutaneous afferents coursing through the dorsal columnsmay be subject to a central inhibitory control mediated byGABAergic mechanisms. In the rat spinal cord, many cu-taneous afferent fibers extend their arborizations beyondthe area in which cells can be shown to respond to theiractivation. Impulses fail to propagate in the long-rangereach of myelinated fibers caudally in the dorsal columns,probably because of a GABA-operated Cl– shunt in theterminals. When this mechanism was suppressed byGABAA-receptor antagonists, impulse blockade was re-lieved. According to Wall and his colleagues (Wall 1994,1995; Wall and McMahon 1994), this would be the ex-pression of an additional control mechanism regulated byGABAergic interneurons mediating PAD that is locatedin axons proximal to the synaptic area.

In summary, at the present time, there is no convinc-ing evidence that presynaptic inhibition in group-I mus-cle afferents results from conduction failure in the intra-spinal arborizations. Alternative explanations involve re-duction of transmitter release due to the conductance in-crease and the depolarization of the afferent terminalsproduced by activation of GABAergic interneurons. Fur-ther experimental tests are required to appraise the con-tribution of conduction failure in the unmyelinated ter-minal arborizations of group-I muscle and of cutaneousafferents during presynaptic inhibition.

Modeling of the presynaptic inhibitory synapse

The functional consequences of PAD and GABAergicpresynaptic inhibition have been explored in a variety ofmodels. Most of them include the effects of the chlorideshunt produced by activation of GABAA receptors viaaxo-axonic synapses on the amplitude of the action po-tential invading the intraspinal arborizations of the affer-ent fibers. Computer simulations have shown that the ac-tion potential may be modified as it propagates along,even through simply branched axons, and this under-scores the relevance of the geometry of an axon in influ-encing the efficacy of transmitter release at different re-lease sites along its branches (Parnas and Segev 1979;Waxman and Wood 1984; Franciolini 1987; Swadlow etal. 1980). In addition to the geometry of an axon and itsbranches, the myelination pattern and the location anddensity of voltage dependent sodium and potassiumchannels also play an important role in this process.

At the present time, there is little information con-cerning the number and location of the presynaptic axo-axonic contacts along the central axons (however, seeFyffe and Light 1984; Maxwell et al. 1990; Pierce andMendell 1993; Nicol and Walmsley 1991). Nevertheless,this has prompted the development of several modelsaimed at understanding the impulse transmission capa-

10

bilities of the intraspinal arborizations of the afferent fi-bers. Some of them are mostly theoretical (Segev 1990),while others are based on information derived from mor-phological reconstructions (Graham and Redman 1994;Walmsley et al. 1995; Walmsley and Nicol 1998).

Walmsley et al. (1995, see also Walmsley and Nicol1998) modeled presynaptic inhibition as a steady-statedepolarizing chloride conductance. Simulation of pre-synaptic inhibition caused a depolarization of the entirenodal region receiving the axo-axonic GABAergic con-tacts, together with a significant decrease in the ampli-tude of the action potential at all boutons of the branchednode, but the action potential at the prior and subsequentnodes was unaffected by the large shunt. In addition tothe peak amplitude, the shunt also reduced the durationof the action potential in the branched node, and thismay contribute to a reduction of transmitter release.However, no evidence was obtained to support the possi-bility of complete conduction block of the action poten-tial propagating along the axon collateral branches (seethe section “Conduction of action potentials in terminalarborizations” and Henneman et al. 1984; Lüscher1998). According to Graham and Redman (1994), trans-mitter release would be significantly reduced for a de-crease in the amplitude of the action potential to 90 mVor less with no transmission for action potentials smallerthan 50 mV. Their simulations indicated that this wouldrequire a chloride shunt of 3–8 nS at each bouton, which,compared with previous estimates of quantal GABAconductances of 150–300 pS, represents an extremelylarge, maintained conductance.

According to Walmsley and Nicol (1998), if repolar-ization were assisted by an active potassium conduc-tance, then a smaller leak conductance would be re-quired. This would potentially increase the effects of achloride shunt on membrane-potential depolarization andaction-potential reduction (see also Walmsley et al.1995). Another possibility is that presynaptic inhibitionoperates primarily through a sustained subthreshold (lessthan 15 mV) depolarization rather than a substantial re-duction in the amplitude of the action potential. Sus-tained depolarization could act by partial inactivation ofvoltage-gated calcium channels responsible for transmit-ter release (see also Zytnicki et al. 1990; Zytnicki andL’Hôte 1993). In this context, direct recordings frommammalian presynaptic terminals will provide importantinformation in subsequent modeling (Forsythe 1994).

The small proportion of Ia boutons found to be con-tacted by presynaptic boutons in Clarke’s column(Walmsley et al. 1987, 1995) contrasts markedly with thehigh proportion of presynaptic contacts found on Ia bou-tons in the ventral horn by Pierce and Mendell (1993)and Fyffe and Light (1984), and this may reflect distinctdifferences in the control of Ia transmission in these tworegions of the spinal cord. The theoretical simulations ofWalmsley et al. (1995) and Walmsley and Nicol (1998)have shown that the effectiveness of presynaptic contactsmay be restricted to individual unmyelinated nodal re-gions along the parent axon. In this case, nodal regions

connected by myelinated internodal segments may be ef-fectively isolated from each other with respect to the ac-tion of presynaptic contacts, thus enabling segregatedpresynaptic control over different branches arising fromthe same fiber, as suggested by the studies of Eguibar etal. (1994, 1997), Quevedo et al. (1997), and Lomelí et al.(1998) for pairs of collaterals of individual fibers endingeither in the intermediate nucleus at the L6 level, or oneof them ending within L6 and another ascending to theL3 segment in Clarke’s column.

Functional organization of presynaptic inhibition

Selective presynaptic control of group-I afferents

Functional organization of PAD

The first intrafiber recordings of PAD in the dorsal col-umns made by Eccles and Krnjevic (1959) alreadyshowed that not all muscle afferents are affected in thesame manner by segmental conditioning inputs. Record-ings made from group-I fibers, considered as Ia or Ib onthe basis of their conduction velocity and peripheralthreshold, revealed the selectivity of the pathways lead-ing to PAD. Ia afferents from both flexors and extensorsappeared to be depolarized mainly from flexors, while Ibafferents, independent of their origin, were depolarizedfrom both flexors and extensors. Another difference wasthat Ia fibers were depolarized both by Ia and Ib affer-ents, but Ib afferents only by Ib volleys (Eccles et al.1962c, 1962d; Schmidt 1971; Lundberg 1964, 1982;Brink et al. 1984).

Studies made in the eighties by measuring the intra-spinal threshold of single muscle afferents (Rudomin etal. 1983, 1986) or by intrafiber recording of PAD fromsingle, functionally identified muscle afferents (Jiménezet al. 1988) have expanded the initial observations ofEccles et al. (1962b, 1963a), Lundberg and Vyklicky(1966), and Lundberg (1964, 1982) by showing thatstimulation of group-I afferents and of the vestibular nu-clei produced PAD in muscle spindle (Ia) afferents,while stimulation of cutaneous nerves, of the bulbar re-ticular formation, the red nucleus, and the pyramidaltract produced no PAD, but was instead able to inhibitthe PAD produced by group-I and by vestibulospinal fi-bers (type A PAD pattern; see Figs. 5 and 6A). In con-trast, most of the intraspinal terminals of the tendon or-gans were depolarized by stimulation of group-I muscleafferents and also by stimulation of rubrospinal, reticulo-spinal, and corticospinal fibers. Interestingly, stimulationof cutaneous nerves appeared to have a dual action on Ibafferents: it produced PAD in one set of fibers (type BPAD pattern; see Fig. 6B) and inhibited the PAD in an-other set (type C PAD pattern; see Fig. 6C).

In a more recent and detailed study, also made by in-trafiber recording of PAD in functionally identified af-ferents, Enríquez et al. (1996a, 1996b) found insteadthat only 52% of muscle-spindle afferents in the medialgastrocnemius had a type-A PAD pattern, while 26%

11

had a type-B and 13% a type-C PAD pattern. In con-trast, only 11% of the fibers from tendon organs had atype-A PAD pattern, 35% a type-B, and 54% a type-CPAD pattern.

The proportion of different PAD patterns in Ia and Ibafferents was not an invariant feature of these ensembles.It could be changed after crushing the medial gastrocne-mius nerve close to the muscle (Enríquez et al. 1996b).Two to twelve weeks after the nerve crush, the propor-tion of fibers reconnected to muscle spindles with a type-A PAD pattern was reduced to 35%. The proportion ofafferent fibers in which stimulation of the bulbar reticu-lar formation, but not of cutaneous afferents producedPAD was increased to 65% (type-C PAD pattern). On theother hand, all afferents reconnected with tendon organswere depolarized by stimulation of cutaneous nerves and

also by stimulation of the bulbar reticular formation.That is, they expressed a type-B PAD pattern.

These alterations in the PAD patterns after a peripher-al nerve crush have been explained assuming separatechanges in the effectiveness of the spinal pathways lead-ing to PAD of muscle spindles and of tendon organs. Anindependent control of the synaptic effectiveness of Iaand Ib fibers could have functional relevance in thosecases where both inputs converge onto the same spinalinterneurons (for a review, see the section “Selective pre-synaptic control of group-I afferents” and Rudomin1990a, 1990b; Rudomin et al. 1998). The non-linear in-teractions between DRPs produced by stimulation of Iaand Ib afferents reported by Brink et al. (1983) could re-sult from convergence of these inputs on the first-orderinterneurons mediating PAD (Rudomin et al. 1983).

Spinal interneurons mediating PAD

Spike-triggered averaging of spinal potentials. Ecclesand collaborators (1962a) attempted an electrophysiolog-ical characterization of the PAD-mediating interneurons,selected because they responded mono- or disynapticallyto activation of group-I muscle afferents and presented aprolonged discharge, a feature considered to explain thelong duration of PAD. They concluded that the samelast-order GABAergic interneurons mediate the PAD ofgroup-Ia and group-Ib fibers. Subsequent studies haveinstead suggested that the prolonged duration of PAD isnot necessarily due to a sustained interneuronal activity(Rudomin and Muñoz-Martínez 1969; Nicoll and Alger1979; Quevedo et al. 1997), but rather to the slow dy-namics of GABA release and/or uptake (see Nicoll andAlger 1979).

12

Fig. 5A–D Primary-afferent depolarization (PAD) in a functional-ly identified Ia muscle-spindle afferent. The inset diagram showsthe stimulation and recording arrangements: a glass micropipettewas placed in the dorsal columns and inserted into a fiber left incontinuity with the gastrocnemius muscle. A Dot display and his-togram of background afferent activity to show the silent periodproduced during a muscle twitch, as expected for a muscle spin-dle. This fiber conducted at 110 m .s–1 and was, therefore, classi-fied as Ia. B Intrafiber potential changes (IC-EC) produced in thesame fiber by stimulation of group-I posterior biceps-semitendino-sus-nerve (PBSt) afferent fibers with three pulses at 300 Hz, 2×Tstrength. C Effects produced by sural-nerve (SU) stimulation witha single pulse of 2×T strength. D Interaction between SU andPBSt stimulation. CDP Cord dorsum potentials. Measurement ofintrafiber potential was performed with the gastrocnemius muscleslack to prevent generation of action potentials in the receptor.Note that PBSt stimulation produced a slow depolarization lastingmore than 100 ms and that SU stimulation produced no transmem-brane potential changes in the fiber, but was able to inhibit the de-polarization produced by PBSt stimulation. Reproduced with kindpermission from Jiménez et al. (1988) and Rudomin (1990b)

A more precise identification of the interneurons me-diating PAD requires, in addition, disclosure of theirconnections with afferent fibers. Solodkin et al. (1984)and Rudomin et al. (1987) used spike-triggered averag-ing of dorsal root (DRPs) and of ventral root potentials(VRPs) to reveal presumed connections of individual in-terneurons with afferent fibers and motoneurons, respec-tively. They found two different classes of interneuronsthat responded mono- or disynaptically to stimulation ofgroup-I muscle afferents and polysynaptically to stimu-

lation of low-threshold cutaneous afferents (class-I andclass-II interneurons; see diagram in Fig. 1). The actionpotentials of class-I interneurons were time-locked toinhibitory ventral root potentials (iVRPs), which ap-peared without concurrent DRPs (Rudomin 1990b). TheiVRPs had, in some cases, monosynaptic latencies rela-tive to the interneuronal spike and were abolished fol-lowing the i.v. administration of strychnine, suggestinginvolvement of glycinergic synapses (Rudomin et al.1990). Since class-I interneurons could be antidromical-ly activated from Clarke’s column (Rudomin et al.1987), they would belong to the same category of neu-rons shown by Jankowska and collaborators to mediatethe non-reciprocal inhibition of Ib origin (Hongo et al.1983).

Activity of class-II interneurons appeared in synchro-ny with longer-lasting iVRPs and also with negativeDRPs. In some cases, these root potentials had onset la-tencies relative to the interneuronal spike consistent withmonosynaptic activation. These iVRPs and DRPs werestrychnine-resistant and picrotoxin-sensitive, suggestingactivation via GABAergic synapses (Rudomin et al.1990). Since class-II interneurons responded to stimula-tion of muscle and cutaneous afferents, it was assumedthey mediated the PAD of group-Ib fibers. However, asshown by Enríquez et al. (1996a), a fraction of muscle-spindle afferents is also depolarized by cutaneous inputs.It is, therefore, possible that some of the class-II inter-neurons described originally instead mediated the PADof muscle-spindle afferents.

The simultaneous occurrence of monosynaptic iVRPsand DRPs has suggested that the same class-II interneu-ron may have direct depolarizing connections with af-ferent fibers and direct inhibitory connections withmotoneurons. That is, that pre- and postsynaptic inhibi-tion are not mutually exclusive, as it was assumed whenpresynaptic inhibition was first discovered (Granit et al.1964; Kellerth and Szumski 1966a, 1966b), but may co-exist (Rudomin et al. 1987; see also Cook and Cangiano1972; Carlen et al. 1980). Anatomical evidence indicat-ing that the same interneuronal GABAergic bouton hasconnections with afferent terminals and motoneurondendrites has been published (see the section “The his-tological basis for presynaptic inhibition: axo-axonicsynapses”).

Mediation of pre- and postsynaptic inhibition by thesame class-II interneuron may have interesting function-al implications in the control of movement. Activation ofclass-II interneurons via descending commands will pro-duce a long-lasting GABAergic inhibition of motoneu-rons that may not be counteracted by excitation arisingfrom the presynaptically inhibited muscle spindle, ten-don organ, or cutaneous afferents (see the section “Su-praspinal control of the spinal presynaptic inhibitorypathways”). This would endow a priority status to the in-hibition (or excitation) mediated by descending fibers,compared with the actions initiated by peripheral affer-ents, and may be of relevance for the execution of move-ments initiated by supraspinal commands.

13

Fig. 6A–C Primary-afferent depolarization (PAD) patterns ofgroup-I muscle afferents. Continuous measurements of the intra-spinal threshold of three different group-I gastrocnemius afferentfibers ending within the intermediate nucleus at the L6 segmentallevel. A Stimulation of the superficial peroneus (SP) and sural(SU) nerves, as well as stimulation of the bulbar reticular forma-tion (RF) and nucleus raphe magnus (NRM), had no effect on theresting intraspinal threshold of the fiber, but inhibited the PADproduced by stimulation of the posterior biceps-semitendinosus(PBSt) nerve. B The intraspinal resting threshold of the fiber wasreduced because of PAD by stimulation of SU, SP, RF, and NRM.The PAD produced by these conditioning stimuli occluded withthe PAD produced by PBSt stimulation. C The fiber was depolar-ized by RF and NRM stimulation. This PAD summated with thePBSt-induced PAD. In contrast, stimulation of the SU and SPnerves had no effect on the resting intraspinal threshold, but inhib-ited the PBSt-induced PAD. The PBSt nerve was stimulated withthree pulses at 300 Hz, 35 ms before the intraspinal threshold test-ing pulse. SU and SP stimulation was one pulse, and NRM and RFstimulation was a train of 15 pulses at 400 Hz, all applied 75 msbefore the threshold testing pulse. Stimulus strengths are indicat-ed. Conduction velocities of fibers in A, B, and C were 93.6, 82.3,and 76.1 m.s–1, respectively. Peripheral thresholds were 1.18, 1.4,and 1.28×T. Further explanations in text. Reproduced with kindpermission from Quevedo et al. (1995)

Direct activation of PAD-mediating interneurons. Directactivation, by means of intraspinal microstimulation, ofthe last-order interneurons mediating the PAD of Ia andIb afferents, has provided important information on thefunctional organization of the pathways mediating pre-synaptic inhibition. Comparison of the onset latency ofthe segmentally induced PAD with the latency of themonosynaptic PAD produced by intraspinal microstimu-lation (Jankowska et al. 1981a), supports the originalproposal that the shortest pathway mediating the PAD ofgroup-I afferents of segmental origin has two interposedinterneurons, and that the last-order interneurons medi-ating the PAD are located within the intermediate nucle-us in laminae V and VI (Eccles et al. 1962; Schmidt1971).

Studies by Rudomin et al. (1983; see also Quevedoet al. 1995) have further suggested that cutaneous, ru-brospinal, corticospinal and raphespinal fibers inhibitthe segmentally-induced PAD in afferent fibers with atype-A PAD pattern by acting on the first-order inter-neurons, while reticulospinal fibers inhibit PAD by act-ing on the second, last-order interneurons in the PADpathway, as illustrated in Fig. 7. In contrast, reticulospi-nal fibers appear to have excitatory connections withthe last-order interneurons mediating the PAD of affer-ent fibers with type-B and type-C PAD patterns (Rudo-min et al. 1983).

By measuring the intraspinal threshold of two collat-erals of the same afferent fiber ending in a close proxim-ity within the intermediate nucleus, Quevedo et al.(1997) found that, in some fibers, intraspinal microstim-ulation produced a monosynaptic GABAergic PAD inone, but not in the other collateral. This suggested thatthe PAD elicited by single last-order interneurons had alocal character and did not spread to nearby collaterals of the same afferent fiber. However, in other fibers, intraspinal microstimulation produced a monosynapticGABAergic PAD in both collaterals, suggesting that, inthese cases, the last-order PAD-mediating interneuronshad synaptic contacts with both collaterals (see also thesection “The histological basis for presynaptic inhibi-tion: axo-axonic synapses”). Thus, it seems that singlelast-order interneurons have connections with a restrict-ed number of collaterals of individual afferents and thatsingle collaterals receive connections from more thanone interneuron. This may be considered as the basic cir-cuitry required for independent control of informationflow in selected collaterals of individual afferents (seebelow).

14

Fig. 7A–E Differential effects produced by segmental and supra-spinal stimuli on the monosynaptic primary-afferent depolariza-tion (PAD) produced by intraspinal microstimulation. Continuousmeasurements of the intraspinal threshold of a single Ia gastrocne-mius afferent tested within the intermediate nucleus at the L6 seg-mental level. A Effects produced by stimulation of cutaneousnerves (SU and SP) and supraspinal nuclei (NRM and RF) on theresting intraspinal threshold. B Cutaneous and descending stimuliinhibit the PAD produced by posterior biceps-semitendinosusnerve (PBSt) stimulation. C Effects of these same stimuli on themonosynaptic PAD produced by intraspinal microstimulation (onepulse, 4.5 µA applied 3 ms before the threshold testing pulse).Note inhibition of monosynaptic PAD by stimulation of the RFand lack of effects by the NRM and the cutaneous nerves. Periph-eral threshold of the afferent fiber, 1.23×T. Conduction velocity,92 m.s–1. D Histological section of the brainstem showing stimula-tion sites. E Proposed location of the inhibitory actions exertedalong the pathways mediating PAD. SP Superficial peroneusnerve, SU sural nerve, Cx motor cortex, RF bulbar reticular forma-tion, NRM nucleus raphe magnus. See text for further explana-tions. Reproduced with kind permission from Quevedo et al.(1995)

Differential control of PAD

Eguibar et al. (1994, 1997) investigated the effects pro-duced by conditioning stimulation of cutaneous andmuscle afferents as well as of the cerebral cortex on thePAD elicited in pairs of intraspinal collaterals of thesame afferent fiber ending in the intermediate nucleus.They found that, in many single-Ia afferents, stimulationof the PBSt nerve produced PAD of about the same mag-nitude in the two collaterals. Yet, conditioning stimula-tion of cutaneous nerves and of the motor cortex inhibit-ed, in a differential manner, the PAD elicited in both ofthem. The relative magnitude of the inhibition of thePAD in both collaterals could be changed by varying thestrength of the segmental input used to produce the back-ground PAD, the strength and source of the inhibitoryconditioning stimuli, or the site of cortical stimulation. Itwas possible, in several fibers, to completely suppressthe background PAD in one collateral, while the PADelicited in the other collateral remained practically thesame.

A similar differential inhibition of PAD has beenfound in pairs of collaterals of single muscle spindles,one ending at the L6-segmental level within the interme-diate nucleus and the other ascending to the L3 level toend in Clarke’s column, as illustrated in Fig. 8A. Quiteoften, the differential inhibition of PAD in the L3 and L6collaterals could be reversed during cold conductionblock of the thoracic spinal cord (see Fig. 11 and the sec-tion “Supraspinal control of the spinal presynaptic inhib-itory pathways”). This has suggested that descendingmechanisms play an important role in the differentialcontrol of the PAD in individual collaterals of muscle-spindle afferents during the execution of specific motortasks (Lomeli et al. 1998).

Stimulation of sensory nerves and/or the motor cortexoften produced PAD of different magnitudes in pairs ofcollaterals of single-Ib fibers ending within the interme-diate nucleus at the L6 level, or one collateral ending atthe L6 level and the other at the L3 level, within Clarke’scolumn. In some fibers, stimulation of cutaneous nervesor of the motor cortex could reduce the intraspinalthreshold of one collateral, practically without affectingthe intraspinal threshold of the other collateral (Eguibaret al. 1994, 1997). A separate control of the synaptic ef-fectiveness of the segmental and ascending collaterals ofthe same muscle afferent appears possible because, asshown previously by Jankowska and collaborators (Jan-kowska and Padel 1984; Harrison and Jankowska 1984),different sets of interneurons mediate PAD at these twodifferent spinal levels, as illustrated in Fig. 8B.

It thus seems that focal control of presynaptic inhibi-tion allows the intraspinal arborizations of muscle affer-ents to function as a dynamic assembly that can be frac-tionated to convey information to selected neuronal tar-gets. This could be a means by which different spinalpostsynaptic targets coupled by sensory input from acommon source can be uncoupled. Focal control of pre-synaptic inhibition in axonal terminals has been also ob-

15

Fig. 8A, B Selective inhibition of primary-afferent depolarization(PAD) in segmental and ascending collaterals of a single muscle-spindle afferent. Two separate stimulating micropipettes wereplaced at the L3 and L6 segmental levels within Clarke’s columnand intermediate nucleus, respectively. Antidromic action poten-tials produced by stimulation through each micropipette were re-corded from a fine gastrocnemius nerve filament. Interaction be-tween the antidromic potentials, because of refractoriness, wastaken as evidence for activation of two collaterals of the same af-ferent fiber. A Conditioning stimulation of sensory nerves [super-ficial peroneus (SP), sural (SU), and posterior articular (PAN)nerves] and of supraspinal structures [reticulospinal (RF) andraphespinal (NRM) fibers] inhibited the PAD produced in both col-laterals by stimulation of the posterior biceps-semitendinosus(PBSt) nerve. However, the inhibition of PAD was stronger in theL3 than in the L6 collateral. The PBSt stimulus was a train of fourpulses, 400 Hz, 1.6×T, applied 25 ms before the threshold testingpulse. SU and SP nerves were stimulated with one pulse applied50 ms before the threshold testing pulse; PAN, NRM, and RF witha train of eight pulses at 700 Hz, preceding the threshold testingpulse by 75 ms. Stimulus strengths are indicated (Lomelí, Linares,and Rudomin, unpublished observations). B Neuronal connectionsexplaining the local character of inhibition of PAD in the L3 andL6 collaterals (Coll L3 and Coll L6, respectively) of individualmuscle-spindle afferents. PAD produced by stimulation of group IPBSt afferents is mediated by at least two interposed interneurons.Separate groups of GABAergic interneurons produce PAD of L3and L6 collaterals. Stimulation of cutaneous afferents and ofraphespinal (NRM) and corticospinal (Cx) fibers inhibit PAD byacting on the first-order interneurons mediating PAD of musclespindles, through separate sets of inhibitory interneurons. Reti-culospinal (RF) fibers reduce PAD by inhibiting the last-orderGABAergic interneurons. Reproduced with kind permission fromLomelí et al. (1998)

served in reticulospinal neurons in the lamprey and inthe nervous system of invertebrates (for a review, seeNusbaum et al. 1997).

Presynaptic inhibition of group-II muscle afferents

The physiology and functional organization of the path-ways leading to PAD and presynaptic inhibition ofgroup-II afferents have been mainly disclosed by Jan-kowska and collaborators in a series of remarkable pa-pers (see Jankowska and Riddell 1998 for a more de-tailed review). These studies have revealed that synapticactions of group-II muscle afferents are modulated bytwo different neuronal systems. Neurons of the firstsystem are spinal interneurons and operate by GABA-mediated PAD and presynaptic inhibition. Neurons ofthe second system belong to supraspinal monoaminergicneurons, which operate by releasing noradrenaline (NA),serotonin (5-HT), or dopamine (DA) and which depressor facilitate synaptic transmission from group-II affer-ents (Bras et al. 1989, 1990; Jankowska, personal com-munication). The contribution of presynaptic mecha-nisms in the depression of synaptic transmission elicitedby the monoamines has not been yet elucidated.

GABAergic modulation

Group-II afferents are strongly depolarized by group-IImuscle afferents as well as cutaneous and joint afferentsand only weakly by group-I muscle afferents (Harrisonand Jankowska 1989; Riddell et al. 1993). These sameconditioning stimuli depress the monosynaptic field po-tentials produced by group-II volleys (Fig. 9A, B). Thus,it seems that PAD of group-II afferents is associated withdepression of their synaptic effectiveness. Group-II fi-bers terminating in a given segment are most effectivelydepolarized by afferents with a preferential projection tothe same segment and only weakly by those afferentswhich terminate mainly elsewhere (Riddell et al. 1995).

The origin of PAD of group-II afferents suggests thatthe first-order interneurons in the PAD pathway must bemost effectively activated by group-II, cutaneous andjoint afferents and only weakly by group-I afferents.Some interneurons must be also activated by stimuli ap-plied to the sensorimotor cortex and/or the locus coerule-us and raphe nuclei, which also evoke PAD of group-IIafferents (Carpenter et al. 1963b; Riddell et al. 1993).Since multimodal input characterizes the majority of in-terneurons mediating PAD of group-II afferents, they areexpected to be coexcited by different peripheral afferentsand descending tract fibers (Jankowska and Riddell1995).

DRPs of group-II origin are depressed in decerebratepreparations, in parallel with a depression of the reflexactions of group-II afferents on motoneurons (Carpenteret al. 1963a). This suggests that the PAD pathways aresubjected to a descending inhibitory control. However,

the pathways mediating this effect have not been identi-fied.

With respect to the actions of individual PAD inter-neurons, it is still unknown whether they affect transmis-sion from only one or from several categories of affer-ents. The similar origin of PAD of group-II muscle affer-ents and of fibers in the joint, interosseus (Jankowska etal. 1993), and pudendal nerves (Angel et al. 1994; Bussand Shefchyk 1999) is compatible with mediation ofPAD of these fibers by the same interneurons (Jan-kowska and Riddell 1995). Observations that PAD ofgroup-II afferents is preferentially evoked by afferentsterminating in the same region of the spinal cord sug-gest, in addition, that different subsets of PAD interneu-rons are functionally specialized and that the interneu-rons have mainly localized actions. However, in view ofdifferences in the effectiveness with which PAD isevoked in these afferents by group-I volleys, it seemsmore likely that separate interneurons are involved in thegeneration of PAD in group-Ia, -Ib, and -II afferents.

Monoaminergic modulation

Modulation of the synaptic actions of group-II muscleafferents by monoamines has been investigated by ana-lyzing the effects of local application of monoaminergiccompounds on field potentials and on responses of singleneurons, both evoked monosynaptically. These neuronswere tested in the dorsal horn and in the intermediatezone. Neurons with axons in the dorsal spinocerebellar

16

Fig. 9A–F Selective inhibition of synaptic actions of group-II af-ferents. Changes in the field potentials of group-II origin by shorttrains of conditioning stimuli applied to peripheral nerves (A, B)or within the raphe (C) and locus coeruleus (D) nuclei and by lo-cally applying the 5-HT1A,1C,2-agonist 5-methoxytriptamine andNA-alpha2-agonist tizanidine (E, F). Upper row Superimposed,control field potentials evoked by group-II afferents in the dorsalhorn (larger) and conditioned potentials evoked by the same stim-uli (smaller). Lower row Superimposed control and conditionedfield potentials evoked by group-I (early components) and group-II (late components) muscle afferents; note that only the late com-ponents are depressed. From Jankowska and Riddell (1998). (Re-produced with kind permission from Rudomin P, Romo R, Men-dell LM (eds) Presynaptic inhibition and neural control. OxfordUniversity Press, New York)

tract (DSCT) and spinocervical tract (SCT) were alsotested (Bras et al. 1989, 1990; Jankowska et al. 1995,1997).

Locally applied monoamines appear to affect trans-mission from peripheral afferents rather selectively.They depress responses evoked by nociceptive skin stim-ulation, leaving responses evoked by innocuous stimulipractically unaffected (see below). They also depresssynaptic actions of group-II muscle afferents, but haveno effect on actions of group-I afferents or facilitatethem (Bras et al. 1989, 1990; Jankowska personal com-munication).

The effectiveness with which monoaminergic drugsdepress transmission from group-II muscle afferents de-pends on the site of application. Synaptic actions ofgroup-II muscle afferents on interneurons located in thedorsal horn (laminae III-IV) are preferentially depressedby 5-HT, and those in the intermediate zone (laminae V-VI) by NA. Effects of specific 5-HT and NA agonists areeven more selective (Fig. 9E, F). DA appears to haveless-selective effects since it depresses group-II field po-tentials at the two locations to a similar extent (see Jan-kowska and Riddell 1998). The laminar selectivity ofmonoaminergic actions appears to be the same in differ-ent segments (Jankowska et al. 1994). According to Jan-kowska and Riddell (1998), the topographically relatedeffectiveness of monoaminergic and PAD-depressive ac-tions may express different functions of the monoamin-ergic and PAD-modulatory systems. The actions of de-scending monoaminergic systems that are confined tospecific laminae might be used to select different func-tional types of neurons, irrespective of the origin of theirinput, while PAD neurons might be more involved in se-lecting subpopulations of neurons according to their in-put, as it seems to be the case for group-I fibers (Rudo-min et al. 1983, 1986).

It is still not known which receptor subtypes are in-volved in the effects of monoamines on various function-al types of neurons, nor whether the responses of theseneurons are modulated pre- or postsynaptically. Howev-er, the reported observations are already of interest be-cause they document the existence of differential mono-aminergic control of functionally different types of neu-rons in pathways from group-II muscle afferents.

Descending control

Electrical stimulation of supraspinal structures contain-ing noradrenergic and serotonergic tracts ending in thespinal cord (locus coeruleus/subcoeruleus, Kölliker-Fuseand raphe nuclei) have been found to depress synapticactions of group-II muscle afferents as effectively as lo-cally injected monoamines (Fig. 9C, D) or as PAD gen-erated by stimulation of peripheral nerves. Stimulationof these supraspinal structures has a dual effect: (1) itproduces PAD of group-II afferents involving activationof descending fibers that activate GABAergic spinal in-terneurons, resembling the actions exerted on group-I fi-

bers; and (2) it activates slowly conducting tract fibers,which may depress transmission from group-II fibers byreleasing monoamines.

The PAD and monoaminergic systems of control ofgroup-II actions activated from the brainstem appear tobe tightly linked together, as judged from the overlap ofthe sites from which it is possible to evoke PAD ofgroup-II afferents and selectively depress their actions(Noga et al. 1992; Riddell et al. 1993). However, sincethe PAD and activation of monoaminergic descendingactions are evoked by electrical stimulation, it is alwayspossible that, under more physiological conditions, thesetwo systems are separately activated, involving separatepopulations of spinal interneurons (see also Quevedo etal. 1995).

Presynaptic inhibition of cutaneous afferents

Dorsal horn

Physiological and anatomical evidence supports the exis-tence of last-order GABAergic interneurons that modu-late the synaptic effectiveness of large cutaneous affer-ents in the dorsal horn (Eccles et al. 1963b, see also thesection “The histological basis for presynaptic inhibi-tion: axo-axonic synapses”). Schmidt and collaboratorsshowed some time ago that some of the pathways medi-ating PAD of these afferents are modality specific (Jäniget al. 1967, 1968a, 1968b). Namely, the largest PAD inrapidly adapting cutaneous afferents is produced by vol-leys in other rapidly adapting cutaneous afferents, andthe most effective PAD in slowly adapting cutaneous af-ferents was found to be produced by stimulation of otherslowly adapting cutaneous afferents (see also Brown andHayden 1972). These findings have suggested that theensuing presynaptic inhibition could play an importantrole in sensory discrimination by enhancing contrast andeliminating surplus excitation (Jänig et al. 1967, 1968a,1968b).

In addition to the PAD produced by stimulation of cu-taneous nerves, cutaneous afferents are also depolarizedby stimulation of group-Ib, -II, and -III muscle afferents(Eccles et al. 1963b; Buss and Shefchyk 1999) and alsoby stimulation of the sensorimotor cortex (Carpenter etal. 1963b; Andersen et al. 1964d), the bulbar reticularformation, the raphe nuclei (Martin et al. 1979; Quevedoet al. 1995) and the red nucleus (Jiménez et al. 1987).

Although it has been established that PAD in large cu-taneous fibers is not due to extracellular accumulation ofpotassium ions, but rather to more specific GABAergicmechanisms (Jiménez et al. 1987), the situation is not soclear with finely myelinated and unmyelinated afferents(see below), particularly because these afferents appearnot to be targets of synapses made by GABAergic inter-neurons (see the section “Rareness of axo-axonic synap-ses at central terminals of fine afferent fibers”).

17

Dorsal-column nuclei

Excitability testing and intrafiber recordings in presyn-aptic axons of large cutaneous afferents in the dorsal-col-umn nuclei in response to afferent stimulation have indi-cated that presynaptic depolarization occurs simulta-neously with the P wave (Wall 1958; Andersen et al.1964b, 1964d). Direct stimulation of the contralateralpyramidal tract, as well as stimulation of the sensorimo-tor cortex, also evoked P-wave responses in the dorsal-column nuclei (Andersen et al. 1964e; Magni et al.1959). Since these waves have been considered a sign ofPAD, it was concluded that cortical stimulation depolar-ized the same set of fibers as cutaneous volleys and alsoproduced presynaptic inhibition (Andersen et al. 1964a).

In spite of all of the electrophysiological and the ana-tomical observations supporting the existence of PADand presynaptic inhibition in the terminals of cutaneousafferents ending in the dorsal column nuclei (see the sec-tion “The histological basis for presynaptic inhibition:axo-axonic synapses”), Canedo (1997) has suggestedthat the inhibitory effects produced by cortical stimula-tion on sensory responses of cuneate interneurons andcuneothalamic cells are mostly postsynaptic and that pre-synaptic mechanisms play only a minor role in the con-trol of impulse transmission in this nucleus. This conclu-sion was based on the finding that high-frequency repeti-tive stimulation of the dorsal columns (at 100 Hz), whichvery effectively produces PAD of cutaneous afferents,leads to successive monosynaptically evoked action po-tentials, only with occasional failures in cuneate neurons.That is, there was no evidence of a cumulative depres-sion that could be ascribed, at least in part, to a presyn-aptic mechanism. However, it must be pointed out thatthe existence of postsynaptic inhibitory mechanisms af-fecting impulse transmission from cutaneous afferents tocuneate neurons following stimulation of the cerebralcortex does not exclude the possible contribution of pre-synaptic control mechanisms. Other experimental para-digms appear necessary to sort out this issue.

Interneurons mediating PAD

In 1962, Wall proposed that neurons in the LissauerTract were involved in the generation of PAD, becauseDRPs were abolished by sectioning the dorsal columnbetween the segment of the dorsal root that was stimulat-ed and that of the dorsal root used to record the DRP. Hesuggested that the “depolarization of the neurons in thesubstantia gelatinosa reflects back onto the afferent fi-bers to cause a cathodal block in conduction or a reduc-tion in transmitter release by reducing the size of the in-coming action potential” (see also Cervero et al. 1978).In 1963, Eccles et al. examined, in the cat, the depolariz-ing actions of cutaneous and group-Ib, -II, and -III mus-cle afferent volleys on the central terminals of cutaneousfibers. They proposed instead that two interneurons wereinterposed in the presynaptic inhibitory pathway and that

the depolarizing synapses on the primary afferent fiberswere located at a depth of 1.5–2.0 mm from the dorsalsurface of the lumbar cord, i.e., in laminae III and IV. In1981, Jankowska et al. (1981b) used intraspinal micro-stimulation to directly activate the last-order interneu-rons, producing PAD in single cutaneous afferents with-out concurrent activation of nearby afferent fibers (seethe section “Selective presynaptic control of group-I af-ferents”). They concluded, in agreement with the origi-nal suggestion of Eccles et al. (1963b), that the cell bod-ies of the last-order interneurons producing PAD of largecutaneous afferents were located in the middle and later-al parts of laminae III and IV.

In a more recent publication, Wall and Lidierth (1997)described the DRPs and PAD produced in the rat spinalcord by direct stimulation of the superficial Lissauertract (LT) at the border between the L2 and L3 spinalsegments. This procedure produced a prolonged negativeDRP in the L2 dorsal root with a latency of 15 ms. Re-cording from nearby dorsal roots showed this DRP to beunaccompanied by stimulation of afferent fibers in thoseroots. In contrast, stimulation of a neighboring dorsalroot produced a DRP with a shorter latency (about 5 ms).The LT-DRP was unaffected by neonatal capsaicin treat-ment that destroyed most unmyelinated fibers (see alsoCervero and Plenderleith 1984). Measurements of intra-spinal excitability of myelinated fibers showed that theLT-DRP was accompanied by PAD. The LT-DRP inter-acted with the DRPs evoked from the sural or gastrocne-mius nerves or motor cortex, suggesting convergencesomewhere in the pathways that produce the DRPs. Itthus seems that neurons in the substantia gelatinosa arenot interposed in the shortest pathways leading to PADof large cutaneous afferents. The functional role of themodulation of PAD of large cutaneous afferents by activ-ity in substantia gelatinosa neurons remains to be estab-lished.

PAD of thinly myelinated and unmyelinated fibers

The question of whether thinly myelinated and unmyeli-nated fibers in cutaneous nerves are subjected to PADand presynaptic inhibition in the same manner as largemyelinated fibers has attracted the attention of severalinvestigators. Dorsal root reflexes, which are a sign ofPAD, have been recorded from Aδ fibers by Toennies in1938, by Brooks and Koizumi in 1956, and by Casey andOakley in 1972. Further evidence on PAD in Aδ fiberswas obtained by Whitehorn and Burgess in 1973. Theyshowed that the intraspinal terminals of Aδ fibers frommechanoreceptors increased their excitability followingmechanical stimulation of the skin, but not during nox-ious stimulation. On the other hand, noxious mechanical,but not noxious heat stimuli, very effectively increasedthe intraspinal threshold of nociceptive Aδ fibers, whileinnocuous stimuli were less effective (see also Sastry1978; Lisney 1979). As with large cutaneous fibers,stimulation of the nucleus raphe magnus and of the bul-

18

bar reticular formation also reduced the intraspinalthreshold of cutaneous Aβ mechanoreceptors and of Aδnociceptors, suggesting PAD (Martin et al. 1979).

Several groups of investigators have also reported in-creased excitability of C-fiber intraspinal terminals fol-lowing conditioning stimulation of A fibers in cutaneousnerves. They suggested that these terminals are subjectto presynaptic inhibition (Carstens et al. 1979; Hentalland Fields 1979; Fitzgerald and Woolf 1981; Calvillo etal. 1982). Iontophoretic application of GABA has beenshown to reduce the intraspinal threshold of some Aδand C-fibers. Since this effect was blocked by bicucul-line, it was suggested that the increase in the intraspinalexcitability involved a GABAergic axo-axonic mecha-nism (Randic et al. 1980). However, as discussed in thesection “Rareness of axo-axonic synapses at central ter-minals of fine afferent fibers”, anatomical evidence sup-porting this view is rather scarce both for Aδ and C fi-bers, suggesting that other mechanisms may be also in-volved.

Effects of peripheral nerve lesions on PAD

In contrast with what has been reported for group-I fi-bers (see the section “Selective presynaptic control ofgroup-I afferents”), 2–12 weeks after a peripheral nervecrush, cutaneous afferents show very little PAD (Horch1978; Horch and Lisney 1981; Wall and Devor 1984).Although the reasons for these differences are not clear,it should be noted that continuous application of nervegrowth factor to the lesioned nerve reduces the depres-sion of PAD following cutaneous nerve injury (Fitzger-ald et al. 1985). Depression of PAD in cutaneous nervesis not produced if afferent activity is suppressed bychronically applied tetrodotoxin (Devor 1983), whiledamage to a cutaneous nerve reduces the amount of sub-stance P, somatostatin, and calcitonin gene-related pep-tide in the dorsal horn of the spinal cord (Takahashi andOtsuka 1975; Jessell et al. 1979; Barbut et al. 1981; Tes-sler et al. 1984; Wall et al. 1981). It thus seems that thechanges in PAD of cutaneous fibers following peripheralnerve injury are associated with changes in the availabil-ity of trophic factors transported from the periphery tothe intraspinal arborizations of the afferent fibers (seealso Devor 1983) and that expression of PAD in cutane-ous afferents, unlike expression of PAD in group-I mus-cle afferents, may require the presence of these peptides,but this remains to be investigated.

Presynaptic inhibition of visceral afferents

In 1967, Rudomin showed that the terminals of superiorlaryngeal nerve ending in the solitary-tract nucleus weredepolarized by conditioning stimulation of low-thresholdafferents in the cervical vagus (among them, pulmonarystretch receptors) and in the aortic nerve (mainly frombaroreceptors). During PAD, there was a depression of

the presynaptic and postsynaptic extracellular field po-tentials produced in the solitary-tract nucleus by stimula-tion of the superior laryngeal nerve, as expected for pre-synaptic inhibition. Stimulation of trigeminal afferentsthat produced PAD of superior laryngeal nerve terminals,or direct stimulation of the solitary tract nucleus, was un-able to change the excitability of vagal and aortic nerveterminals. This suggested that vagal afferents concernedwith blood-pressure regulation were not subjected to pre-synaptic modulation of their synaptic effectiveness bymechanisms involving activation of GABAergic inter-neurons (Rudomin 1968).

The finding that small-diameter vagal afferents maybe devoid of presynaptic modulatory mechanisms medi-ated by last-order GABAergic interneurons has been thesubject of discussion (see Barillot 1970), but it is nowaccepted that myelinated fibers from lung stretch-recep-tor afferents are subjected to presynaptic inhibition,while aortic myelinated baroreceptors are not (Richter etal. 1986). It thus seems that the synaptic effectiveness ofthe vagal afferents carrying “homeostatic” information isnot directly modulated by GABAergic interneurons. Thisis an issue that needs to be re-examined with more detail,using techniques developed more recently (Eguibar et al.1997), because of its implications on the control of sen-sory information. The absence of presynaptic GABA-ergic modulation in homeostatic afferents makes sense,if it is considered that automatic control mechanisms ofthis type require precise information on the variable tobe controlled, whereas activity of the PAD-mediating in-terneurons modifies the information content of the sig-nals transmitted by the axon terminals that receive axo-axonic synapses (Rudomin and Madrid 1972), a featurethat may be relevant for motor control and sensory dis-crimination.

Segmental presynaptic modulation is by no means re-stricted to cutaneous and articular afferents or to motorafferents involved in limb movements. There is increas-ing evidence that presynaptic inhibition also plays a rolein the control of the synaptic efficacy of segmental affer-ent inputs to external sphincter motoneurons during mic-turition. During micturition in the cat, urethral sphinctermotoneurons are hyperpolarized (Shimoda et al. 1992;Fedirchuk and Shefchyk 1993). In addition to the de-crease in tonic urethral sphincter activity during micturi-tion, there is also a suppression of phasic excitatorysphincter reflexes. The polysynaptic EPSPs evoked byelectrical stimulation of sensory pudendal and superficialperineal cutaneous afferent are decreased in amplitudeduring micturition (Fedirchuk et al. 1994), and transmis-sion from perineal afferents appears to be decreased toall sacral spinal targets by postsynaptic inhibition as wellas by presynaptic inhibition.

Angel et al. (1994) examined the intraspinal excitabil-ity of sensory pudendal and superficial perineal afferentsand found that these fibers had PAD patterns somewhatdifferent from those of cutaneous fibers. Namely, cutane-ous afferents innervating skin distant from the perineum(i.e., foot) were as effective in producing PAD as affer-

19

ents innervating perineal regions. Group-II hindlimbmuscle afferents were also powerful sources of PAD,while group-I afferents had much less potent actions. In-creases in perineal afferent excitability were also seenduring stimulation of cutaneous or mixed hindlimb affer-ents, and, in some cases, there was decreased excitability(PAH) when using high strengths of nerve-conditioningstimuli. In many sensory pudendal afferent fibers, therewas increased excitability during micturition. Bladderdistention alone failed to produce excitability changes inperineal afferents. Excitability changes were only seenwith activation of the entire micturition circuitry(Shefchyk 1998). The presynaptic control of sensorytransmission is envisaged as a mechanism that contrib-utes to reflex modulation during voiding. In addition tothe PAD of perineal afferents, it seems necessary to con-sider the potential contributions of presynaptic inhibitionto the spinal modulation of visceral (bladder or colon)sensory information during bladder filling, micturition,and defecation (see Buss and Shefchyk 1999).

Supraspinal control of the spinal presynaptic inhibitorypathways

Lundberg and his colleagues in Goteborg established thatfour different descending systems from the brain stemexert similar inhibitory effects on interneuronal transmis-sion in reflex pathways to motoneurons (see the detailedreview by Lundberg 1982). The descending systems thatinhibit transmission to motoneurons have equally stronginhibitory effects on interneuronal transmission to as-cending pathways, and, as particularly relevant to thetopic of this review, to primary afferent terminals. Ofthese four descending pathways, two turned out to bemonoaminergic (noradrenergic and serotonergic, respec-tively), the other two being the dorsal reticulospinalsystem and the ventral reticulospinal tract.

The tonic inhibitory control exerted by these path-ways comes most clearly to light when comparing the ef-fects of afferent volleys on flexor and extensor motoneu-rons and on PAD in decerebrate animals before and afterspinalization, the latter procedure dramatically “freeing”the spinal-cord mechanisms from their powerful brain-stem inhibitory control (enhanced in the decerebratestate by removal of descending inhibition from highercenters). In particular, the transmission in the interneuro-nal pathways mediating PAD was investigated by com-paring DRPs in decerebrate and spinal states (Carpenteret al. 1963b).

These studies led to the conclusion that the tonic de-cerebrate inhibition of transmission in pathways to pri-mary afferents is selective for transmission from theflexor reflex afferents (FRA afferents) and operates nei-ther on pathways mediating PAD from group-Ia and -Ibafferents, nor on the specialized pathways mediatingPAD from low threshold cutaneous afferents. However,Quevedo et al. (1993) have shown that cooling the tho-racic spinal cord to block impulse conduction increasedthe DRPs produced by stimulation of low-threshold af-ferents in cutaneous nerves and by stimulation of the

20

Fig. 10 Effects of reversible spinalization on the inhibition of pri-mary-afferent depolarization (PAD) of a single Ia afferent. Chang-es in the intraspinal threshold of a single afferent ending in the in-termediate nucleus, at the L6 segmental level. Before (uppertrace) and during (lower trace) conduction block of the thoracicspinal cord by means of a cooling probe. Posterior biceps-semiten-dinosus nerve (PBSt) stimulus was a train of four pulses, 400 Hz,1.6×T, applied 25 ms before the threshold testing pulse. Sural (SU)and superficial peroneus (SP) nerve: one pulse applied 50 ms be-fore the threshold testing pulse. Posterior articular nerve (PAN)and reticulospinal fibers (RF) were stimulated with a train of eightpulses at 700 Hz, preceding the threshold testing pulse by 75 ms.Stimulus strengths are indicated in figure. Note that spinalizationhad little effect on the magnitude of the PBSt-induced PAD, butvery effectively increased the inhibition of PAD. The diagramshows the neuronal circuitry, explaining the effects of spinalizat-ion. Increased effectiveness in the inhibition of PAD would resultfrom removal of a descending tonic inhibition. Reproduced withkind permission from Quevedo et al. (1993)

posterior articular nerve. In contrast, spinalization re-duced the DRPs produced by stimulation of group-Imuscle afferents from flexors and increased the inhibi-tion exerted by cutaneous and joint afferents on the PADof Ia fibers (see Fig. 10). It thus seems that reflex actionsof cutaneous and joint afferents on pathways mediatingPAD are under a tonic descending inhibitory control,which is removed by spinalization (see diagram in Fig.10). Moreover, as illustrated in Fig. 11, spinal block isable to reverse the inhibitory action of cutaneous affer-ents on the PAD elicited by PBSt stimulation on segmen-tal and ascending collaterals of individual muscle spindleafferents. This suggests that tonic descending influencesplay an important role in the differential modulation ofthe information conveyed by the intraspinal arborizationsof afferent fibers (Lomeli et al. 1998).

Presynaptic control of synaptic actions of interneuronsand descending-tract neurons

Intraspinal terminals of descending fibers

The possibility that presynaptic inhibition mediated byaxo-axonic GABAergic synapses not only affects the in-traspinal terminations of afferent fibers, but also the ax-on terminals of spinal interneurons has been exploredwith some detail by several investigators, but so far thereis no clear, convincing evidence in this regard. Since thefirst observations of Jankowska and Roberts (1972), itbecame clear that studies of presynaptic inhibition exert-ed on interneuron axon terminals can be complicated bythe possibility that synaptic potentials initiated in the

neuron dendrites, or in the cell body, may spread to theaxon terminals and affect their excitability as well astransmitter release, particularly if the interneurons haveshort axons (see also Nicolls and Wallace 1978).

No such complication occurred in the case of studiesof changes in synaptic effectiveness of brainstem neu-rons with long axons ending at lumbosacral levels. Eideet al. (1968) showed that monosynaptic EPSPs evoked inspinal motoneurons by stimulation of vestibulospinal fi-bers were not depressed when conditioned with group-Ivolleys from flexors, which depressed very effectivelythe monosynaptic Ia EPSPs produced in the same moto-neurons (Fig. 1). These findings were subsequently con-firmed by Rudomin et al. (1975), who showed, in addi-tion, that vestibulospinal and rubrospinal terminals werenot depolarized following stimulation of sensory nerves,by intraspinal microstimulation, or by iontophoretic ap-plication of GABA, which very effectively depolarizedIa fibers ending in the vicinity (Rudomin et al. 1981; seealso Curtis et al. 1982; Curtis and Malik 1984; Curtis etal. 1984). An additional finding has been that, unlike Iaand Ib afferents, the terminal arborizations of reticulospi-nal fibers have a rather low density of GABAB receptors,as it may be inferred from the finding that the synapticeffectiveness of these fibers is barely depressed by (-)-baclofen, a GABAB agonist (Jiménez et al. 1991; Que-vedo et al. 1997).

It thus seems that the intraspinal terminals of de-scending fibers are not subjected to a GABAergic con-trol of their synaptic effectiveness. However, this doesnot mean that these fibers are not subjected to othermechanisms of presynaptic control. Singer and Berger(1996) examined the effects of 5-HT on synaptic trans-mission to hypoglossal motoneurons in brainstem slicesfrom juvenile rats. 5-HT was found to depress the ampli-tude of the monosynaptic postsynaptic current producedin hypoglossal motoneurons by stimulation of the ipsilat-eral lateral reticular formation, but had no effect on thepostsynaptic currents produced by direct application ofL-glutamate. Serotonin inhibition was mimicked by 5-HT1B agonists. These findings suggest that depression ofsynaptic transmission by 5-HT was exerted presynaptic-ally on the axon terminals of reticular neurons. It is,therefore, possible that 5-HT also modulates the synapticeffectiveness of reticular and other descending fibersending in lumbar segments, probably by paracrine inter-actions (see the sections “Presynaptic inhibition ofgroup-II muscle afferents” and “Non-synaptic presynap-tic modulation of transmitter release”).

Group-II premotor interneurons

As discussed below, Alford et al. (1991) have providedconvincing evidence that the synaptic effectiveness ofaxons of spinal premotor interneurons in the lamprey ismodulated both by GABAA and GABAB receptors. Thishas encouraged further investigations aimed at examin-ing possible presynaptic modulation of axonal terminals

21

Fig. 11A, B Spinalization reverses the differential inhibition ofprimary-afferent depolarization (PAD) in segmental and ascendingcollaterals of a single muscle spindle afferent. A, B Inhibition ofthe posterior-biceps-semitendinosus-nerve (PBSt)-induced PAD byconditioning stimulation of sural (SU), superficial peroneus (SP),and posterior articular (PAN) nerves, before (A) and during (B)spinal block. Before the spinal block, conditioning stimulation ofthe SU and SP nerves produced a stronger inhibition of PAD in theL6 collateral ending within the intermediate nucleus than in the L3collateral ending in Clarke’s columns. During spinal block, the in-hibition of PAD produced by these same stimuli was stronger inthe L3 collateral than in the L6 collateral. Peripheral threshold ofthe afferent fiber: 1.14×T; mean conduction velocity: 84.3 m.s–1;separation between microelectrode tips: 2.5 cm. Reproduced withkind permission from Lomelí et al. (1998)

of spinal interneurons in the cat. Aggelopoulos et al.(1997) have examined the intraspinal threshold changesof axonal terminals of interneurons located in the mid-lumbar segments of the spinal cord with projections tothe lower lumbar motor nuclei. These interneurons arepreferentially activated by group-II muscle afferents.Thresholds for antidromic activation of a substantialnumber of interneurons were reduced after electricalstimulation of group-II muscle afferents. According tothese investigators, it is unlikely that the excitability in-crease resulted from electrotonic spread of depolariza-tion from the interneuron soma to its terminals, or by en-vironmental changes in the vicinity of the terminals re-lated to neuronal activity. They suggested further that theexcitability of the interneuronal axon terminals would beincreased because of a depolarization generated by amechanism similar to that acting at the terminals of pri-mary sensory afferents. The mechanisms involved in theexcitability increase of interneuronal axon terminalshave not been disclosed because it has not been possibleto determine whether or not this effect is mediated by ac-tivation of GABAergic mechanisms and if it is associat-ed with reduction of the synaptic effectiveness (Edgley,personal communication).

In this context, it should be mentioned that Maxwellet al. (1997) studied the synapses formed by dorsal-horngroup interneurons in lamina VI that are monosynaptic-ally activated by group-II afferents. They found that themajority of boutons formed by axon collaterals of theseinterneurons made synaptic arrangements with eitherdendrites or cell bodies of spinal neurons. A small num-ber of boutons of one cell made synapses with other ax-on terminals, and these may also have a presynaptic in-fluence on transmission. Since none of the boutons ex-amined were postsynaptic to other axons, they concludedthat it is unlikely that the output of this group of inter-neurons is modified presynaptically. Bannatyne et al.(1998) and Maxwell and Riddell (1999) have recentlymade a systematic EM study of axon terminals of the in-termediate interneurons projecting to motor nuclei, i.e.,those studied by Aggelopoulos et al. (1997), and foundno axo-axonic contacts on some 100 terminals. It, thus,seems unlikely that the depolarization of the terminal ax-ons of the group-II premotor interneurons reported byAggelopoulos et al. (1997) is mediated by axo-axonicGABAergic synapses.

Ia inhibitory interneurons

Enríquez et al. (1996c) have analyzed possible presynap-tic inhibition of synaptic actions of Ia inhibitory spinalinterneurons. They found that conditioning stimulationof extensor nerves with stimuli 1.65–5.0×T, which wouldnot be expected to evoke PAD in Ia afferents, depressedthe disynaptic IPSPs produced in motoneurons by pe-ripheral nerve stimulation. The depression lasted about200 ms and maximal effects were seen at 20–50 ms.Since these same conditioning stimuli had no effect on

monosynaptic EPSPs evoked by Ia volleys in the inhibi-tory interneurons, it was proposed that the inhibition wasexerted on the Ia inhibitory interneurons (Enríquez andPerreault, personal communication). Evidence for chang-es occurring in the soma of the Ia inhibitory interneuronswas that conditioning stimulation increased (first 50 ms)and then decreased (up to 200 ms) the interneuronal dis-charges produced by glutamate iontophoresis and alsothat the probability of inducing an orthodromic spike ofIa interneurons in response to stimulation of the ventro-lateral fasciculus (VLF) was significantly reduced byconditioning stimulation. On the other hand, the proba-bility of producing antidromic spikes of Ia inhibitory in-terneurons by stimulation of their axon terminals in themotor nucleus was increased following conditioningstimulation of extensor nerves, suggesting axon terminaldepolarization. There was also depression of the unitaryIPSPs evoked in motoneurons by the Ia inhibitory inter-neurons. However, these changes were difficult to as-cribe to a presynaptic mechanism because the condition-ing stimulus also induced IPSPs in the motoneurons.

PAD-mediating interneurons

Intraspinal microstimulation can produce a short latency,picrotoxin-sensitive, PAD in the intraspinal terminals ofsingle Ia or Ib afferent fibers, probably because of directactivation of the last-order GABAergic interneurons (seethe section “Selective presynaptic control of group-I af-ferents”). In some Ib intraspinal terminations, the mono-synaptic PAD produced by intraspinal microstimulationwas depressed by (-)-baclofen, while in other termina-tions, even of the same afferent fiber, was baclofen-resis-tant (Quevedo et al. 1997). This has suggested that thePAD produced by stimulation of sensory nerves, whichis baclofen-sensitive, is mostly mediated by the baclo-fen-sensitive, last-order GABAergic interneurons, whilethe PAD produced by reticulospinal fibers, which is ba-clofen-resistant, is mediated by the baclofen-resistant,last-order GABAergic interneurons. In other words, thatsensory and descending pathways may produce PAD ofmuscle afferents by means of different sets of last-orderinterneurons (Quevedo et al. 1992), a feature that may berelevant for motor control and sensory discrimination.

In this context, the presence of GABAB autoreceptorsin the interneurons mediating the sensory-induced PADis envisaged as a part of a self-limiting mechanism thatregulates the synaptic efficacy of the last-order interneu-rons mediating the PAD generated by activation of mus-cle afferents (see also Quevedo et al. 1992; Zytnicki etal. 1990; Zytnicki and L’Hôte 1993). The lack of such alimiting mechanism in the PAD elicited by reticulospi-nal, and possibly also by other descending fibers, alsosuggests a different functional role of the PAD initiatedby these pathways.

22

Lamprey spinal interneurons

In contrast with what has been found in higher verte-brates, spinal interneurons in the lamprey appear to besubjected to presynaptic GABAergic control mecha-nisms. The lamprey spinal cord contains excitatory glu-tamatergic interneurons that project to ipsilateral moto-neurons and interneurons (Buchanan and Grillner 1987)and inhibitory glycinergic interneurons with projectingaxons to contralateral motoneurons and interneurons(Buchanan 1982). During ongoing fictive locomotion,axons of inhibitory and excitatory interneurons displaymembrane-potential oscillations in phase with ipsilateralventral root activity (Alford et al. 1991), which are re-duced following the administration of GABAA- andGABAB-receptor antagonists to the perfusion bath. Inthe presence of tetrodotoxin, the interneuronal axonswere depolarized following the administration ofGABAA and GABAB agonists (Alford and Grillner1991). GABAA agonists open Cl– channels and depolar-ize the interneuronal axons. The conductance changemay shorten the amplitude and duration of the action po-tential and probably reduce the Ca2+ entry occurring dur-ing the action potential. The depolarization induced byGABAB agonists is probably due to an inhibition of volt-age-dependent K+ channels. Further investigations haveshown that activation of presynaptic GABAB receptorscauses a G-protein-mediated depression of transmitterrelease from glutamatergic and glycinergic premotor in-terneurons (Alford and Grillner 1991). The phasic-de-pendent presynaptic inhibition of excitatory interneuro-nal synaptic transmission could allow regulation of thesynaptic efficacy by local segmental circuits.

Reticulospinal axons provide glutamatergic EPSPsutilizing both NMDA and AMPA/kainate receptors onpostsynaptic motoneurons and excitatory and inhibitorypremotor interneurons (Buchanan and Grillner 1987;Ohta and Grillner 1989). In contrast to both interneuronsand sensory afferents, synaptic transmission from reti-culospinal axons is not affected by GABA agonists (Al-ford and Grillner 1991). There is, however, a presynapticmodulation due to activation of two different types ofmetabotropic glutamate receptors, which are activated byL-AP4 and (1S,3R)-ACPD (Krieger et al. 1986). It thusseems likely that a high level of synaptic activity, andthereby release of glutamate, may lead to subsequent de-pression of synaptic transmission from the active axon.Bath application of 5-HT and dopamine also cause a de-pression of synaptic transmission from reticulospinal ax-ons to spinal interneurons and motoneurons (Buchananand Grillner 1991; Shupliakov et al. 1995; Wikstrom etal. 1995). In both cases, there is no postsynaptic conduc-tance change, suggesting a presynaptic mechanism,probably mediated by paracrinic actions.

Presynaptic inhibition and behavior

PAD and fictive locomotion

Functional studies on presynaptic inhibition made duringthe past few years have been concerned mostly with cy-clical changes in PAD of muscle and cutaneous afferentsoccurring during fictive locomotion (Dubuc et al. 1988;Dueñas and Rudomin 1988; Rossignol et al. 1998). Due-ñas and Rudomin (1988) measured the intraspinal excit-ability changes of Ia and Ib afferents form extensor mus-cles during fictive locomotion and found that PAD inboth types of fibers was maximal during the flexor phaseof the locomotor cycle. It was suggested that the cyclicPAD had a GABAergic component, although contribu-tion of increased potassium in the extracellular spacecould not be excluded. In the absence of fictive locomo-tion, stimulation of cutaneous nerves increased the PADof Ib fibers (Rudomin et al. 1983). During fictive loco-motion, these effects were reversed and stimulation ofcutaneous nerves inhibited the PAD elicited in Ib affer-ents (Dueñas and Rudomin 1993). In other words, duringfictive locomotion, Ib fibers changed their initial type-APAD pattern to a type-C PAD pattern. This has providedadditional evidence that the PAD patterns of muscle af-ferents are not invariant, but change according to thetask to be performed (see the section “Selective presyn-aptic control of group-I afferents”).

Disynaptic and oligosynaptic reflex activity elicited inmotoneurons by stimulation of cutaneous nerves is alsomodulated during fictive locomotion (Labella et al.1992; Floeter et al. 1993). Most of these changes appearto be due to postsynaptic effects exerted on interneuronsand motoneurons, but presynaptic inhibition may alsocontribute to the cyclic changes of the EPSPs producedby stimulation of cutaneous afferents.

Gossard (1996) has reported that the monosynapticEPSPs evoked by Ia fibers in motoneurons are depressedwhen conditioned with trains of pulses to the PBStnerve, but not during the cyclic PAD generated in thesesame fibers during fictive locomotion. Based on the re-sults of Eguibar et al. (1994), Gossard explained hisfindings assuming a separate control of the synaptic ef-fectiveness of the Ia collaterals connected with motoneu-rons and those collaterals subjected to the cyclic PAD,which are probably connected to other spinal neurons.Alternatively, the cyclic PAD could arise from a differentmechanism than that producing the PAD by stimulationof muscle afferents, namely extracellular accumulationof potassium ions (see the section “Presynaptic depolar-ization and inhibition by potassium ions”).

Involvement of potassium accumulation in the cyclicPAD produced during fictive locomotion was suggestedby Kremer and Lev-Tov (1997, 1998). They showed that,in the isolated spinal cord of the neonatal rat, bicucullinehad little effect on the antidromic discharges produced inthe dorsal roots by PAD (dorsal root reflexes; see thesection “The generation of PAD by GABAergic axo-axonic synapses”), or on the increased excitability of

23

afferent terminals produced by stimulation of other dor-sal roots, as well as of the ventrolateral funiculus and ofthe white ventral commissural axons. Nevertheless, theactivity-dependent potassium accumulation could con-tribute to transmitter-mediated modulation of the afferentinput in the mammalian spinal cord, including unmyeli-nated afferents that do not seem to be subjected to a con-trol via GABAergic axo-axonic synapses (see the section“Rareness of axo-axonic synapses at central terminals offine afferent fibers”).

Sleep

Electrophysiological studies have provided evidence forPAD and presynaptic inhibition of trigeminal afferent fi-bers (Baldissera et al. 1965, 1966;) and of transmissionfrom cutaneous afferents through the cuneate nucleus(Carli et al. 1966) during the rapid eye movements of de-synchronized sleep in the behaving cat. More recently,Cairns et al. (1996) found that during active sleep thereis also depression of synaptic transmission through therostral trigeminal sensory nuclear complex and depolar-ization of tooth afferent terminals, suggesting that de-pression of synaptic transmission is due, at least in part,to a presynaptic inhibitory mechanism. Intracellularanalysis of membrane potential activity from jaw-opener(Pedroarena et al. 1994) jaw-closer (Chandler et al.1980; Chase et al. 1980; Chirwa et al. 1991) and lumbarmotoneurons (Chase et al. 1989; Soja et al. 1991) clearlyindicates that postsynaptic inhibitory drives are engagedduring the atonia of active sleep. Additional electrophys-iological and pharmacological observations on individu-al tooth-pulp afferents and their target cells are requiredto assess the relative contribution of pre- and postsynap-tic inhibitory mechanisms in mediating the suppressionof sensory transmission through the trigeminal sensorynucleus during the behavioral state of active sleep.

Presynaptic inhibition in humans during voluntarymovements

The exclusion of the extracellular accumulation of potas-sium ions as the main cause of PAD of group-I muscleafferents (see the section “Presynaptic depolarization andinhibition by potassium ions”), together with the findingthat group-Ia and -Ib afferents have distinct PAD pat-terns (see the section “Selective presynaptic control ofgroup-I afferents”), strengthened the view that PAD andpresynaptic inhibition are rather selective, but it was notuntil the work of Hultborn et al. (1987) that this viewwas more firmly established. Using the monosynaptic Iafacilitation of the H-reflex, these investigators devised anon-invasive experimental paradigm that allows estima-tion of changes in the synaptic effectiveness of group-Iaafferents synapsing with motoneurons.

Experiments in cats have indicated that, under thesame experimental conditions, the amount of reflex facil-

itation depends only on the size of the conditioning IaEPSP. A constant conditioning stimulation is expected toelicit an EPSP of constant size in motoneurons and, thus,a constant reflex facilitation, unless presynaptic inhibi-tion of Ia afferents mediating the conditioning volley ischanging. The amount of monosynaptic Ia facilitationcan, therefore, be used to assess ongoing presynaptic in-hibition on these Ia fibers: the larger the reflex facilita-tion, the smaller the presynaptic inhibition (see Hultbornet al. 1987; Pierrot-Deseilligny and Meunier 1998). Thevalidity of the method was further assessed by showingthat conditioning volleys producing presynaptic inhibi-tion of actions of muscle spindle afferents, or counteract-ing facilitation evoked by muscle spindle afferents, donot reduce the facilitation produced by monosynaptic ac-tions of descending fibers, which are known not to besubjected to presynaptic inhibition, as shown in Fig. 1(see the section “Supraspinal control of the spinal pre-synaptic inhibitory pathways” and Rudomin et al. 1991).

With this approach, Hultborn et al. (1987) found inhumans that, at the onset of a voluntary contraction,there is a reduction of the tonic presynaptic inhibition ofthe muscle-spindle afferents arising from the muscle tobe activated and that, at the same time, there is an in-crease of presynaptic inhibition of those group-Ia fibersinnervating the non-contracting muscles. The decreasedpresynaptic inhibition of the Ia input to motoneurons ofthe contracting muscle increases the gain of the mono-synaptic stretch reflex, which, at the beginning of themovement, could be functionally important in rapidlycompensating for the actual load (Meunier and Pierrot-Desseilligny 1989). Subsequent studies in humans madeby Hultborn and collaborators and by other groups of in-vestigators have provided additional evidence for the se-lectivity of presynaptic inhibition of muscle-spindle af-ferents during a variety of motor behaviors (Burke et al.1992; Iles and Pisini 1992a, 1992b; Nielsen and Kag-amihara 1993), including passive pedaling (Misiazek etal. 1995) and gait (Faist et al. 1996).

Modification in the level of presynaptic inhibition hasbeen also reported during changes in posture (Hayashi etal. 1992; Koceja et al. 1993) and during locomotion(Llewellyn et al. 1990; Yang and Whelan 1993; see alsoStein 1995). According to Iles (1996), these changescould be mediated by cutaneous and descending path-ways, as well as by muscle group-I afferents activatedduring muscle contraction (Devanandan et al. 1966).Stimulation of afferents from the intrinsic foot musclesproduces a powerful presynaptic inhibition of soleus Iaafferents. If afferents from intrinsic muscles are activatedduring rapid locomotion, they may be responsible forsome of the differences in level of presynaptic inhibitionbetween walking and running (Edamura et al. 1991).

Capaday et al. (1995) examined changes in the de-pression of soleus H reflexes produced by conditioningstimulation to group-I afferents from the common pero-neal nerve during standing and walking. They found that,during standing, the soleus H-reflex was strongly inhibit-ed, despite little or no change in the background level of

24

EMG activity, and suggested that the depression of theH-reflex was due to presynaptic inhibition. In contrast tostanding, during the early part of the stance phase ofwalking, the conditioning stimulus produced little or noinhibition. The decreased effectiveness of the condition-ing stimulus during walking appeared not to depend onthe amplitude of the test H-reflex and could not be offsetby increasing (or decreasing) the strength of the condi-tioning stimulus. The reduced presynaptic inhibition dur-ing walking can be explained assuming that the condi-tioning input is decreased during walking because of pre-synaptic inhibition of the afferent fibers. It is also possi-ble that postsynaptic inhibition increases during walkingalong the pathway mediating the PAD of the tested affer-ents. Activation of cutaneous afferents during walkingcould provide this inhibitory input, but other (descend-ing) actions could also play some role. A third possibili-ty is that presynaptic inhibition of the test fibers is intro-duced via other pathways.

In humans (as in cats, see the section “Selective pre-synaptic control of group-I afferents”), cutaneous andcorticospinal axons converge on spinal interneurons thatinhibit presynaptic inhibition of group-Ia afferents.These actions may be responsible for the modulation ofpresynaptic inhibition, which has been observed to pre-cede and accompany a wide range of human movements.Iles (1996) analyzed in normal subjects the effects ofstimulation of cutaneous nerves and of mechanical stim-ulation of the skin on group-I-induced presynaptic inhi-bition of soleus-muscle Ia afferent fibers. He found thatstimulation of the ipsilateral sural nerve decreased thepresynaptic inhibition and that this effect was strongestduring voluntary plantar flexion and weaker during dor-siflexion or at rest. Stimulation of other cutaneous nervebranches serving the dorsum of the ipsilateral foot aswell as light brushing of both distal dorsal and plantarsurfaces of the ipsilateral foot decreased presynaptic in-hibition. However, brushing of the proximal dorsal partof the foot and pretibial area was ineffective. Stimulationof ipsilateral plantar nerves increased presynaptic inhibi-tion, but this action was attributed to activation of group-I afferents from the intrinsic muscles of the foot.

Similar results have been obtained by Aimonetti et al.(1997), who investigated changes in H-reflex presynap-tic inhibition during wrist extension while the skin of theinternal side of the finger tips was concurrently mechani-cally stimulated. Under these conditions, both the recip-rocal and the presynaptic inhibition evoked at 20 ms in-tervals were significantly reduced. These investigatorssuggest that, by modulating the Ia presynaptic inhibition,the cutaneous afferents play a major role in the control ofthe reflex loops involved in hand motor activities.

In this regard, it is interesting to mention that Collinset al. (1998) reported that the sensation produced by anelectrically induced twitch of the extensor carpi ulnarismuscle is attenuated prior and during voluntary wristmovements. According to these investigators, this is con-sistent with the finding of a general attenuation of senso-ry feedback during movement. The attenuation of the

perception of muscle twitch could occur at several places(dorsal column nuclei, thalamus, sensorimotor cortex, oreven in pathways involving the cerebellum) and be anadditional example of dissociation between the modula-tion in the Ia-motoneuron pathways and transmission ofIa input to the brain (see also Llewellyn et al. 1990;Prochazka 1996).

In some experimental paradigms, depression of H-re-flexes may not involve presynaptic inhibition. Wood etal. (1996) investigated the effect of muscle-spindle rest-ing discharge on the size of monosynaptic reflexes in thecat and on the H-reflex in humans. Resting dischargewas altered by contracting the triceps surae muscle atlonger (hold-long) or shorter (hold-short) lengths thanthat at which the reflex was tested. Immediately after thehold-long conditioning, there was depression of themonosynaptic reflexes, both in cats and humans. Thisdepression was attributed to the high level of muscle-spindle discharge during the immediately precedinghold-long period and could not be accounted for by pre-synaptic inhibition, measured as PAD or as changes inheteronymous monosynaptic facilitation. It is suggestedthat the depression was due to postactivation transmitterdepletion due to the high frequency of muscle spindleactivation (see also Hultborn and Nielsen 1998).

Transcranial stimulation

Transcranial magnetic stimulation (TMS) of the lower-limb area of the contralateral motor cortex has beenfound to decrease presynaptic inhibition of muscle spin-dles in humans. According to Iles (1996), this effect isstrongest during voluntary plantar flexion and weakerduring dorsiflexion or at rest. Zanette et al. (1995) usedTMS stimulation to study changes in motor-cortex excit-ability after rapid repetitive movements. They found thatexercise induced a reversible, long standing depressionof cortical excitability, probably related to intracorticalpresynaptic modulation, which transiently reduces themotor representation area. This functional adaptive pro-cess is attractive in terms of the economy of neuronal-network mechanisms that control skilled motor move-ments.

Mercuri et al. (1997) found that conditioning stimula-tion of the radial nerve produces disynaptic reciprocalinhibition followed by presynaptic inhibition of the H-reflexes elicited in the carpi radialis muscle. Transcorti-cal electrical stimulation (TES) facilitated the H-reflexand reduced the disynaptic phase of reciprocal inhibitionas well as the phase of presynaptic inhibition. The reduc-tion of presynaptic inhibition by TES is in agreementwith observations in cat showing that stimulation of themotor cortex and of the pyramidal tract inhibits PAD ofIa afferents (see Rudomin et al. 1986; Eguibar et al.1994).

25

Presynaptic inhibition under pathophysiologicalconditions

There have been several investigations of changes in pre-synaptic inhibition during pathophysiological conditions.Only the most recent ones are reviewed here. Nielsen etal. (1995) examined transmission across synapses of Iaafferents on spinal motoneurons of healthy subjects andof spastic multiple sclerosis patients. Slow passivestretch of the soleus muscle evoked a pronounced de-pression of soleus H-reflex lasting for more than 10 s inhealthy subjects. This depression was less pronouncedand had a shorter duration in spastic patients. A tap ap-plied to the biceps femoris tendon also produced an inhi-bition of the soleus H-reflex, which was larger in thehealthy subjects than in the spastic patients. Stimulationof the femoral nerve (FN) facilitated the soleus H-reflex,which was larger in the spastic than in the healthy sub-jects. The authors suggested that the inhibition of the H-reflex evoked by the biceps femoris tendon tap wascaused by presynaptic inhibition of the Ia afferentswhich mediate the reflex. The facilitation of the soleusH-reflex produced by FN stimulation was interpreted asbeing influenced by changes in presynaptic inhibition.The increased facilitation of the FN and the decreasedinhibition from the biceps femoris tendon tap onto thesoleus H-reflex in spastic patients are, thus, both com-patible with a deficient presynaptic inhibition. An alter-native explanation would be higher excitability of moto-neurons, secondary to stronger excitatory actions ofgroup-II afferents via both intermediate group-II inter-neurons and gamma motoneurons when the descendingnoradrenergic pathways are injured (Jankowska, person-al communication).

Morita et al. (1995) examined heteronymous Ia facili-tation from quadriceps to soleus muscle in normal volun-teers and found it decreases linearly with age. This de-crease in facilitation could reflect a decrease in the num-ber of Ia fibers and their conduction velocities as well asan increase in presynaptic inhibition on Ia terminals.They suggest that the increase in presynaptic inhibitionis an adaptive phenomenon in the aging of the neuro-muscular system or, alternatively, a deteriorating processwith decreasing flexible supraspinal modulation. The in-creased presynaptic inhibition appears to be a compensa-tory reaction that changes the gain of the stretch reflexduring movement, without changing the excitability ofthe motoneurons.

Modeling of presynaptic inhibitory networks

Presynaptic inhibition can act as a filter of input fromtendon organs during muscle contraction (Zytnicki andJami 1998). Intracellular recordings from triceps-suraemotoneurons during contraction of the homonymousmuscle show inhibitory potentials at the onset of the con-traction that quickly decrease and disappear. These in-hibitory potentials are ascribed to a sustained activation

of Ib afferents during muscle contraction (Zytnicki et al.1990; Zytnicki and Jami 1998). To explain the decline ofthe contraction-induced inhibition, Zytnicki et al. (1990)proposed that presynaptic inhibition of Ib terminals oc-curred during the sustained contractions, resulting in adecrease of synaptic transmission between these termi-nals and first-order interneurons. Intraaxonal recordingsfrom Ib afferents during muscle contraction showed asustained PAD that closely followed the time course ofcontraction and possibly reduced the synaptic efficacy ofIb fibers (Lafleur et al. 1992; Zytnicki and Jami 1998).

In a relatively simple model of the disynaptic Ib affer-ent pathway to motoneurons, which incorporated theknown functional properties of the natural network com-ponents, Zytnicki and L’Hôte (1993) found declining in-hibitory potentials in the motoneuron-like output stagewhen functions simulating PAD in Ib afferents, similar tothose recorded experimentally, were introduced in thenetwork. These findings support the view that Ib impuls-es arising during sustained contractions are filtered outwhen they enter the spinal cord by a self-limiting presyn-aptic inhibitory mechanism. This filtering mechanism al-lows transient inhibitory potentials to appear in moto-neurons, signaling onset of contraction and subsequentrapid increases in muscle force. These inhibitory poten-tials might help to limit the firing frequency of motoneu-rons and/or the recruitment of new motoneurons in orderto keep a smooth profile of force development and avoidjerky movements.

The contraction-induced presynaptic inhibition ap-pears to spare a significant amount of Ib terminalssynapsing with Clarke column neurons at L3-L4 seg-ments (Zytnicki and Jami 1998). This would imply anon-uniform distribution of presynaptic inhibitionamong all the terminal branches of an individual groupIb afferent, in keeping with observations showing a dif-ferential presynaptic control of corticospinal and cutane-ous inputs on two branches of the same group-I fiber(Eguibar et al. 1994, 1997; Quevedo et al. 1997; Lomelíet al. 1998; Rudomin et al. 1998).

Non-synaptic presynaptic modulation of transmitter release

Rareness of axo-axonic synapses at central terminals of fine afferent fibers

The bottom line of the impressive body of experimentalwork summarized in the section “Functional organizationof presynaptic inhibition” is that all types of somatosen-sory primary afferents from skin, skeletal muscle, anddeep tissue (such as joints) readily and in a well-orga-nized manner induce PAD in other primary afferents andthat this PAD is strongly induced in large myelinated pri-mary afferents (group-I and -II fibers), but not much ornot at all in finely myelinated (group-III or Aδ-fibers)and unmyelinated fibers (group-IV or C-fibers). Never-theless, particularly in view of the postulate by Melzack

26

and Wall (1965) that the terminals of cutaneous C fiberswould be depolarized by cutaneous A-fiber input, a num-ber of (technically very tricky) studies have been carriedout to look for evidence supporting PAD mediated pre-synaptic inhibition of (mainly cutaneous) C fibers (Calv-illo 1978; Carstens et al. 1979, 1981; Hentall and Fields1979; Calvillo et al. 1982). As recently summarized else-where, Schmidt and Schaible (1998), Willis (1999), theseresults indicate that, indeed, a number of C fibers can bedepolarized by activation of skin afferents, although thesefibers may only be a small proportion of all fine afferents.Calvillo et al. (1982) pointed out, however, that it re-mained an open question whether the observed PAD wasdue to the activation of axo-axonic synapses on C fibersor – as frequently discussed at that time – to potassiumaccumulation in the extracellular space (see above). An-other, and eventually more likely mode of action, as willbe proposed here, is that volume-transmitted (autocrineand paracrine transmitted) substances reduce the trans-mitter output of fine afferent terminals upon activation.

The electrophysiological findings summarized aboveare well in line with the fact that there is barely any evi-dence for axo-axonic synapses on fine afferent-fiber ter-minals in the spinal cord. In particular, axo-axonic syn-apses of any type have been reported as relatively un-common in the superficial dorsal horn of the species ex-amined, with estimates of 0.8% of all synapses in this ar-ea in the rat (Zhu et al. 1981) and 1.7% in the cat (Dun-can and Morales 1978). For instance, axo-axonic con-nections with enkephalin immunoreactivity onto primaryafferent terminals have been looked for, but not found(Glazer and Basbaum 1984). Similarly, fine-structurestudies failed to find significant numbers of axo-axonicsynapses in the superficial dorsal horn with immunoreac-tive-enkephalin in the presynaptic element (Glazer andBasbaum 1981; Ruda 1982; Sumal et al. 1982; see alsothe section “The histological basis for presynaptic inhibi-tion: axo-axonic synapses”).

Spontaneous and evoked release of peptides from thecentral terminals of fine afferent fibers and the locationof their receptors

Over the years, considerable evidence has been accumu-lated that activation, in particular antidromic activationof the peripheral branches of fine primary afferents(group III and IV), releases peptides from their terminals(for a review, see Schaible and Grubb 1993). Some ofthese peptides participate in inducing neurogenic inflam-mation, e.g., calcitonin gene-related peptide (CGRP, astrong vasodilator) or substance P (SP, most likely induc-ing plasma extravasation in addition to vasodilation).The role of others, such as somatostatin (SOM), galanin(GAL), or neurokinin A (NKA), is still less obvious, butmost, if not all of them, seem to have receptors not onlyin the surrounding tissue, but also on the fine afferentterminals themselves; and it is likely that, by binding onthese receptors, the peptides modulate the properties of

these endings (see below the remarks on the presynapticaction of neuropeptide Y, NPY).

In conformity with the principle first enunciated byDale (1935) that the same chemical transmitter is re-leased from all the synaptic terminals of a neuron, it maybe postulated that the spinal endings of primary afferentsupon activation release their colocalized peptides in asimilar way as in the periphery, and that these peptidesnot only exert their action on postsynaptic sites, but, inaddition, have autocrine and paracrine action on presyn-aptic structures. Evidence to support this assumption isstill rare, but as recently reviewed by Levine et al.(1993), Schaible and Grubb (1993), Duggan (1995),Schaible (1996), and Schmidt and Schaible (1998), sev-eral neuropeptides are being released in the spinal cordfrom primary afferent terminals (and possibly other neu-rons) either spontaneously or in the course of afferent ac-tivity entering the dorsal horn. These neuropeptides mayact in a neuromodulatory way on secondary neurons inthe dorsal horn.

To what extent the presynaptically released peptidesact on autoreceptors at neuropeptide-containing primaryafferent terminals remains to be established. NeitherCGRP nor neurokinin binding sites have been demon-strated to any appreciable amount on primary afferent fi-bers (Charlton and Helke 1985), although – at least inthe primate – many CGRP-immunoreactive terminalscontact one another (Carlton et al. 1988). These pre-sumed axo-axonic contacts may provide the anatomicalsubstrate for a regulation by CGRP of amino-acid releasefrom fine primary afferent terminals. Indeed, there issome physiological evidence that this release is underthe control of both SP and CGRP (Kangra and Randic1990; Smullin et al. 1990).

Functional role of opioid receptors on fine afferentterminals

Another important aspect of the non-synaptic modulationof fine afferents, particularly of nociceptive afferents,may be that µ-, δ-, and κ-opioid receptors are present atthe spinal terminals of fine afferent fibers projecting intothe superficial laminae of the dorsal horn (e.g., LaMotteet al. 1976; Fields et al. 1980; Zajac et al. 1989; Gou-ardères et al. 1991). The presence of these receptorspoints to a role for endogenous opioids in presynapticcontrol of the transmission of nociceptive and other af-ferent messages to second order neurons. During inflam-mation, the µ-receptors are upregulated, whereas the δ-and κ-receptors are downregulated (Ji et al. 1995).

The possibility that a substantial portion of the anti-nociception exerted by endogenous opioids or by opioidsapplied directly to the spinal cord is exerted via a pre-synaptic reduction of the transmitter release from noci-ceptive afferent terminals is further substantiated by thefinding obtained from spinal-cord slices that met-enkephalin reduces the release of glutamate from prima-ry afferents (Hori et al. 1992), and, as reviewed in detail

27

by Levine et al. (1993), there is considerable (althoughnot unequivocal) evidence that opioids can modulate therelease of peptides, such as SP and CGRP, from the spi-nal terminals of primary afferents. The possible signifi-cance of the opioidergic control of the spinal release ofneuropeptides for the analgesic effects of opioids hasbeen discussed by Bourgoin et al. (1994).

In regard to the mechanism of the opioid action, it isworth noting that, in cultured DRG neurons, opioidsshorten the plateau of the action potential either by di-rectly (κ-opioids) or indirectly (via an increase in K+-conductance, µ- and δ-ligands) shortening the voltage-dependent Ca2+-current responsible for the plateau phase(Mudge et al. 1979; Werz and MacDonald 1983; Mac-Donald and Werz 1986). If this shortening of the actionpotential also occurs at the spinal terminals of nocicep-tive fine afferents, then transmitter release would be re-duced (see the section “The generation of PAD byGABAergic axo-axonic synapses” for the presynapticaction of baclofen).

Volume-transmitted presynaptic inhibition of centraltransmitter release

In the preceding section, evidence from various laborato-ries has been presented showing the existence of opioidreceptors at the central terminals of fine primary affer-ents, and it has been discussed that these receptors par-ticipate in the control – more specifically in the reduc-tion or complete inhibition – of the transmitter releasefrom these (presumed nociceptive) terminals. In view ofthe lack of any substantial evidence for an axo-axonicmode of action (see above), it has to be assumed that thisattractive mechanism of antinociception is operating viavolume transmission.

Evidence has been quoted above that SP and CGRPmay participate in the presynaptic control of transmitterrelease from fine primary afferent terminals in the spinalcord. Vice versa, there is also evidence that their releasefrom such terminals may be under presynaptic control ofother neuropeptides or other endogenous substances.There are several reports that neuropeptide Y (NPY),which extensively coexists with noradrenaline in brainstem neurons of the rat and the cat (Everitt et al. 1984;Holets et al. 1988), has effects on nerve terminals consis-tent with an impairment of transmitter release (Haas etal. 1987; Colmers et al. 1991). For instance, in the cat,the release of SP induced by a noxious stimulus or elec-trical activation of group-IV fibers could be effectivelyreduced by microinjection of NPY into the superficialdorsal horn. This effect lasted for up to 40 min (Dugganet al. 1991). The inhibitory action of NPY presumably isexerted via presynaptic binding sites for NPY, which arepresent at high densities in the superficial dorsal horn ofthe rat and which are significantly reduced in numberfollowing dorsal rhizotomy, peripheral nerve section, andneonatal capsaicin administration (Kar and Quirion1992). These effects may even be (partly) exerted byaxo-axonic synapses, since Doyle and Maxwell (1993)

reported NPY-containing boutons making axo-axonicsynaptic arrangements in the substantia gelatinosa (lami-na II) of the cat dorsal horn.

Regarding the presynaptic control of the release ofCGRP from fine central terminals, the situation is evenmore favorable because of the anatomical evidence thatCGRP within the spinal cord is almost exclusively ofprimary afferent origin. Thus, even when using a slicepreparation and non-specific stimuli to release CGRP, itcan be assumed that any such release is related to activi-ty in the central terminals of primary afferents. Whenadding various compounds to the perfusate, it has beenshown that CGRP release is reduced by adenosine, medi-ated by adenosine A1 receptors (Santicioli et al. 1993),α2-adrenoceptor agonists (Takano et al. 1993; but seeBourgoin et al. 1993), GABAA agonists (Bourgoin et al.1992), and 5-HT acting on 5-HT3 receptors (Barnes1991; but see Bourgoin et al. 1993). In all these cases,the inhibiting compound could act along the entirelength of the fiber as well as at the terminals, the lattersite of action being favored by the finding that 5-HT re-ceptors seem to be located on the peripheral branches ofunmyelinated primary afferents (Neto 1978).

Opioids seem to be involved in the control of CGRPrelease too, because intravenous naloxone (Collin et al.1993a) or low-intensity electrical stimulation of the muz-zle of the rat (Pohl et al. 1992) reduced the spontaneousrelease of CGRP, as did the simultaneous activation of µ-and κ-opiate receptors in normal rats and activation of µ-receptors in rats with an experimental polyarthritis (Col-lin et al. 1993a, 1993b).

As reviewed by Duggan (1995) there is evidence thatsome fine afferents, presumably from thermal nocicep-tors, contain somatostatin (co-localized with gluta-mate?). However, as yet, nothing is known about theeventual control of its release. As discussed in the samereview, not much more is known about the eventual pres-ence of galanin and opioids in fine primary afferents.

Summary and Conclusions

A hundred years ago, Ramón y Cajal (1899) showed thatthe sensory fibers entering the spinal cord have ascend-ing and descending branches, and that both of them sendcollaterals to the gray matter, where they have profuseramifications. To him, the ascending and descendingbranches were “centripetal conductors by which sensoryexcitation is propagated to the various neurons in thegray matter”.

The data presently reviewed provide convincing evi-dence that these intraspinal arborizations of afferent fi-bers are not obligatory routes for the conduction of ac-tion potentials, but are instead dynamic systems that canbe utilized by central mechanisms to convey informationto selected neuronal targets. The available evidence alsosuggests that this differential control could be a meansby which different spinal postsynaptic targets coupled bysensory input from a common source can be uncoupled.

28

The extensive use of non-invasive techniques to mea-sure changes in presynaptic inhibition of muscle spindleafferents has provided additional evidence that supportsthe existence of selective mechanisms modulating infor-mation transmission in humans during the execution of avariety of motor tasks. It seems very likely that themechanisms involved are similar to those documented inother vertebrates, and we may anticipate that the selec-tive control of information flow in the intraspinal termi-nals during the execution of motor tasks is not restrictedto muscle spindle afferents, but may also operate in ten-don organs and group-II muscle and large cutaneous af-ferents.

Of particular interest is the reduction of tonic presyn-aptic inhibition of muscle-spindle afferents conveyinginformation from muscles to be activated in a voluntarymovement in humans. This inhibition appears to have adescending origin, and probably involves corticospinalas well as rubrospinal and reticulospinal pathways. Thereduction of a tonic presynaptic inhibition in particularsets of intraspinal terminals would “open the gate” andallow information to flow in preferential directions. Me-chanical stimulation of skin afferents (light brushing) ap-pears to reduce the presynaptic inhibition of monosynap-tic reflexes in healthy subjects. Quite interestingly, themost effective inhibitory regions in the skin that reducepresynaptic inhibition are those in the distal dorsal andplantar surfaces of the ipsilateral foot (Iles 1996). Theseregions are precisely those that would be activated dur-ing the termination of a programmed movement or whenthe limb finds an unexpected obstacle. Mechanical stim-ulation of these skin areas would remove the presynapticinhibition of muscle-spindle afferents and allow the in-formation on the state of the muscles to reach the spinalcord to be utilized in the processing of compensatorymovements.

Although the mechanisms involved in this presynap-tic control have not been fully elucidated, it is clear thatthey involve, at least in part, local modulation of trans-mitter release by means of activation of GABAA recep-tors via axo-axonic synapses made by GABAergic inter-neurons with the intraspinal terminals of the afferent fi-bers. In the primary afferent terminals, the steady-stateintracellular Cl– concentration is higher than that predict-ed from a passive distribution due to an active transportmaintaining the outwardly directed Cl–-electrical gradi-ent. The underlying mechanism is an inwardly directedCl–- transport system coupled to Na+ and K+, i.e., not achloride pump driven directly by energy derived fromATP splitting, but a secondary active transport system.

Activation of GABAA receptors in the afferent termi-nals increases the efflux of Cl– ions and produces PADand presynaptic inhibition. The inhibition is most likelyachieved by a reduction of the amplitude of the propa-gated action potential in the intraspinal afferent termi-nals, which in turn blocks or reduces Ca2+ influx. Therole of GABAB receptors in the presynaptic modulationof transmitter release by afferent fibers is still unclear,but it is not unreasonable to assume that activation of

these receptors is involved in long-term changes of syn-aptic effectiveness, an issue that remains open for futureinvestigation.

One unsolved issue in the control of informationtransmission by afferent fibers is whether or not all theirintraspinal arborizations are invaded by action potentials,and if PAD affects this invasion. A related question is theextent to which the action potentials reaching the termi-nals activate the mechanisms of transmitter release. Theexistence of silent synapses has been documented in sev-eral systems, including synapses of nociceptive afferentswith dorsal-horn neurons. In the dorsal horn, silent syn-apses become active following the application of seroto-nin, and it has been proposed that this could be a mecha-nism activated by descending noradrenergic and seroto-nergic pathways involved in pain processing (Li andZhuo 1998).

One may, therefore, ask if there is also a central con-trol of silent synapses in the case of group-Ia, -Ib, and -IIafferents. This is an attractive possibility, particularly forgroup-I and group-II afferents, whose synaptic effective-ness is also modulated by serotonin and noradrenaline(see the section “Presynaptic inhibition of group-II mus-cle afferents”). It will be very interesting to examine howthe differential control of the synaptic effectiveness ofindividual collaterals of sensory fibers, including silentsynapses, operates during development and learning ofnew motor skills.

In contrast to their action on muscle-spindle afferents,primary afferents from skin, skeletal muscle, and deeptissue (such as joints) produce PAD in other primary af-ferents. This PAD is strongly induced in primary affer-ents with large diameter (group-Ib and -II fibers), but notmuch or not at all in finely myelinated (group-III or A-fibers) and in unmyelinated fibers (group-IV or C-fi-bers). There is, however, growing evidence that volume-transmitted (autocrine and paracrine transmitted) sub-stances reduce the transmitter output of fine afferent ter-minals. Paracrine and autocrine interactions may alsoregulate the synaptic effectiveness of axon terminals ofdescending fibers, which are not subjected to a presynap-tic axo-axonic modulatory control, and also of axon ter-minals of spinal interneurons, where evidence on presyn-aptic GABAergic, axo-axonic, modulatory control of thetype of PAD is not conclusive.

An important issue is to determine if the intraspinalendings of primary afferents also release their colocal-ized peptides in a similar way as in the periphery. Thesepeptides could not only exert their action on postsynapticsites, but, in addition, have autocrine and paracrine ac-tion on presynaptic structures. To what extent the pre-synaptically released peptides act on autoreceptors atneuropeptide-containing primary afferent terminals re-mains to be established. Also, the role of peptides re-leased by descending pathways in the modulation of thesynaptic effectiveness of muscle and cutaneous afferentshas not been completely disclosed.

In summary, the evidence reviewed at this time sug-gests that intraspinal terminals of sensory fibers are not

29

passive conductors of information originated in their pe-ripheral sensory receptors. They are complex systemsthat convey information in a rather dynamic manner tofulfill the multiple needs associated with integratedmovements and processing of sensory information, in-cluding nociceptive information.

Acknowledgements We wish to thank Prof. Elzbieta Jankowskafor her critical comments on the manuscript. Partly supported bygrant NS 09196 from the National Institutes of Health, Bethesda,Md., USA, by grant 26402-N from the Council of Science andTechnology, México, to PR, by a grant from the Volkswagenstif-tung to PR and RFS, and by grants of the Deutsche For-schungsgemeinschaft to RFS.

References

Aggelopoulos NC, Chakrabarty S, Edgley SA (1997) Evoked ex-citability changes at the terminals of midlumbar premotor in-terneurons in the cat spinal cord. J Neurosci 17:1512–1518

Aimonetti J-M, Schmied A, Vedel J-P, Pagni S (1997) Presynapticinhibition in human wrist extensor muscles: dependence onmotor task. Soc Neurosci Abstr 23:200

Alford S, Grillner S (1991) The involvement of GABAB receptorsand coupled G-proteins in spinal GABAergic presynaptic inhi-bition. J Neurosci 11:3718–3726

Alford S, Christenson J, Grillner S (1991) Presynaptic GABAAand GABAB receptor mediated phasic modulation in axons ofspinal motor interneurons. Eur J Neurosci 3:107–117

Alvarez FJ (1998) Anatomical basis for presynaptic inhibition ofprimary sensory fibers. In: Rudomin P, Romo R, Mendell L(eds) Presynaptic inhibition and neural control. Oxford Uni-versity Press, New York, pp 13–49

Alvarez-Leefmans FJ (1990) Intracellular Cl– regulation and syn-aptic inhibition in vertebrate and invertebrate neurons. In:Alvarez-Leefmans FJ, Russell JM (eds) Chloride channels andcarriers in nerve, muscle, and glial cells. Plenum Press, NewYork, pp 109–158

Alvarez-Leefmans F J, Gamiño S, Giraldez F, Noguerón I (1988)Intracellular chloride regulation in amphibian dorsal root gan-glion neurones studied with ion-selective microelectrodes. JPhysiol 406:225–246

Alvarez-Leefmans FJ, Nani A, Marquez S (1998) Chloride trans-port, osmotic balance, and presynaptic inhibition. In: RudominP, Romo R, Mendell L (eds) Presynaptic inhibition and neuralcontrol. Oxford University Press, New York, pp 50–79

Andersen P, Eccles JC, Oshima T, Schmidt RF (1964a) Mecha-nisms of synaptic transmission in the cuneate nucleus. J Neu-rophysiol 27:1096–1116

Andersen P, Eccles JC, Schmidt RF, Yokota T (1964b) Depolariza-tion of presynaptic fibres in the cuneate nucleus. J Neuro-physiol 27:92–106

Andersen P, Eccles JC, Schmidt RF, Yokota T (1964c) Identifica-tion of relay cells and interneurones in the cuneate nucleus. JNeurophysiol 27:1080–1095

Andersen P, Eccles JC, Schmidt RF, Yokota T (1964d) Slow po-tential waves produced in the cuneate nucleus by cutaneousvolleys and by cortical stimulation. J Neurophysiol 27:78–91

Andersen P, Eccles JC, Sears TA (1964e) Cortically-evoked depo-larization of primary afferent fibers in the spinal cord. J Neu-rophysiol 27:63–77

Angel MJ, Fyda D, McCrea DA, Shefchyk SJ (1994) Primary af-ferent depolarization of cat pudendal afferents during micturi-tion and segmental afferent stimulation. J Physiol479:451–461

Baldissera F, Broggi G, Mancia M (1965) Presynaptic inhibitionof trigeminal afferent fibres during the rapid eye movementsof desynchronized sleep. Experientia 22:754–755

Baldissera F, Broggi G, Mancia M (1966) Primary afferent depo-larization of trigeminal fibres induced by stimulation of brain-stem and peripheral nerves. Experientia 23:398–400

Bannatyne BA, Riddell JS, Maxwell DJ, Scott DT (1998) A con-focal and electron microscopic examination of the terminals ofmidlumbar premotor interneurons in the feline spinal cord. JPhysiol 509:171P

Barbut D, Polak JM, Wall PD (1981) Substance P in spinal corddorsal horn decreases following peripheral nerve injury. BrainRes 205:289–298

Barillot JC (1970) Depolarization presynaptic des fibres sensitivesvagales et laryngees. J Physiol (Paris) 62:273–294

Barker JL, Nicoll RA (1973) The pharmacology and ionic depen-dency of amino acid responses in the frog spinal cord. JPhysiol 228:259–277

Barnes PJ (1991) Neurogenic inflammation in airways. Int AllergyAppl Immunol 94:303–309

Barron DH, Matthews BHC (1935) Intermittent conduction in thespinal cord. J Physiol 85:73–103

Barron DH, Matthews BHC (1938a) Dorsal root reflexes. JPhysiol 94:26P-27P

Barron DH, Matthews BHC (1938b) The interpretation of poten-tial changes in the spinal cord. J Physiol 92:276–321

Bourgoin S, Pohl M, Benoliel JJ, Mauborgne A, Collin E, HamonM, Cesselin F (1992) Gamma-aminobutyric acid, throughGABAA receptors, inhibits the potassium-stimulated release ofcalcitonin gene-related peptide, but not that of substance P-like material from rat spinal cord slices. Brain Res 583:344–348

Bourgoin S, Pohl M, Mauborgne A, Benoliel JJ, Collin E, HamonM, Cesselin F (1993) Monoaminergic control of the release ofcalcitonin gene-related peptide- and substance P-like materialsfrom rat spinal cord slices. Neuropharmacology 32:633–640

Bourgoin S, Benoliel JJ, Collin E, Mauborgne A, Pohl M, HamonM, Cesselin F (1994) Opioidergic control of the spinal releaseof neuropeptides. Possible significance for the analgesic ef-fects of opioids. Fundam Clin Pharmacol 8:307–321

Bowery NG (1993) GABAB receptor pharmacology. Annu RevPharmacol Toxicol 33:109–147

Bras H, Cavallari P, Jankowska E, McCrea D (1989) Comparisonof effects of monoamines on transmission in spinal pathwaysfrom group I and II muscle afferents in the cat. Exp Brain Res76:27–37

Bras H, Jankowska E, Noga B, Skoog B (1990) Comparison of ef-fects of various types of NA and 5-HT agonists on transmis-sion from group II muscle afferents in the cat. Eur J Neurosci2:1029–1039

Brink E, Jankowska E, McCrea DA, Skoog B (1983) Inhibitoryinteractions between interneurones in reflex pathways fromgroup Ia and group Ib afferents in the cat. J Physiol 343:361–373

Brink E, Jankowska E, Skoog B (1984) Convergence onto inter-neurones subserving primary afferent depolarization of group Iafferents. J Neurophysiol 51:432–449

Brooks C, Koizumi K (1956) Origin of the dorsal root reflex. JNeurophysiol 19:61–74

Brown AG, Hayden RE (1972) Presynaptic depolarization pro-duced by and in identified cutaneous afferent fibers in the rab-bit. Brain Res 38:187–192

Bruggencate G, Lux HD, Liebl L (1974) Possible relationship be-tween extracellular potassium activity and presynaptic inhibi-tion in the spinal cord of the cat. Eur J Physiol 349:301–317

Buchanan JT (1982) Identification of interneurons with contralat-eral, caudal axons in lamprey spinal cord: synaptic interactionsand morphology. J Neurophysiol 47:961–975

Buchanan JT, Grillner S (1987) Newly identified “glutamate inter-neurons” and their role in locomotion ih the lamprey spinalcord. Science 236:312–314

Buchanan JT, Grillner S (1991) 5-hydroxytryptamine depresses re-ticulo-spinal excitatory postsynaptic potentials in motoneuronsof the lamprey. Neurosci Lett 122:71–74

Burke D, Gracies JM, Meunier S, Pierrot-Deseilligny E (1992)Changes in presynaptic inhibition of afferents to propriospi-nal-like neurons in man during voluntary contractions. JPhysiol 449:673–687

30

Burke RE, Rudomin P (1977) Spinal neurons and synapses. In: ERKandel (ed) Handbook of physiology. The nervous system.Sect I, vol I, part 2. American Physiological Society, Beth-esda, pp 877–944

Buss RR, Shefchyk SJ (1999) Excitability changes in sacral affer-ents innervating the urethra, perineum and hindlimb skin ofthe cat during micturition. J Physiol 514:593–607

Cairns BE, Fragoso MC, Soja P (1996) Active-sleep related sup-pression of feline trigeminal sensory neurons: evidence impli-cating presynaptic inhibition via a process of primary afferentdepolarization. J Neurophysiol 75:1152–1162

Calvillo O (1978) Primary afferent depolarization of C fibres inthe spinal cord of the cat. Can J Physiol Pharmacol 56:154–157

Calvillo O, Madrid J, Rudomin P (1982) Presynaptic depolariza-tion of unmyelinated primary afferent fibers in the spinal cordof the cat. Neuroscience 7:1389–1400

Canedo A (1997) Primary motor cortex influences on the descend-ing and ascending systems. Prog Neurobiol 51:287–335

Capaday C (1995) The effects of baclofen on the stretch reflex pa-rameters of the cat. Exp Brain Res 104:287–296

Capaday C, Lavoie BA, Comeau F (1995) Differential effects of aflexor nerve input on the human soleus H-reflex during stand-ing versus walking. Can J Physiol Pharmacol 73:436–449

Carlen PL, Yaari Y, Werman R (1980) Postsynaptic conductanceincrease associated with presynaptic inhibition in cat lumbarmotoneurones. J Physiol 298:539–556

Carli G, Diete-Spiff K, Pompeiano O (1966) Presynaptic and post-synaptic inhibition on transmission of cutaneous afferent vol-leys through the cuneate nucleus during sleep. Experientia 22:239–240

Carlton SM (1994) Presynaptic control of thin primary afferents:an ultrastructural analysis. In: Besson JM, Guilbaud G, OllatH (eds) Peripheral neurons in nociception: physio-pharmaco-logical aspects. John Libbey Eurotext, Paris, pp 185–200

Carlton SM, McNeill DL, Chung K, Coggeshall RE (1988) Orga-nization of calcitonin gene-related peptide-immunoreactiveterminals in the primate dorsal horn. J Comp Neurol 276:527–536

Carpenter D, Engberg I, Funkenstein H, Lundberg A (1963a) De-cerebrate control of reflexes to primary afferents. Acta PhysiolScand 59:424–437

Carpenter D, Engberg I, Lundberg A (1963b) Primary afferent de-polarization evoked from the sensorimotor cortex. ActaPhysiol Scand 59:126–142

Carstens E, Tulloch I, Zieglgänsberger W, Zimmermann M (1979)Presynaptic excitability changes induced by morphine in sin-gle cutaneous afferent C and A fibres. Pflügers Arch 37:143–147

Carstens E, Klumpp D, Randic M, Zimmermann M (1981) Effectof iontophoretically applied 5-hydroxytryptamine on the excit-ability of single primary afferent C- and A-fibers in the catspinal cord. Brain Res 220:151–158

Casey KL, Oakley B (1972) Intraspinal latency, cutaneous fibercomposition and afferent control of the dorsal root reflex inthe cat. Brain Res 47:353–369

Castillo L, Lomelí J, Linares P, Rudomin P (1998) Does PADblock spike invasion to axonal branches of individual group Iafferents? Soc Neurosci Abstr 24:1152

Cervero F, Plenderleith MB (1984) Dorsal root potentials are un-changed in adult rats treated at birth with capsaicin. J Physiol357:357–368

Cervero F, Iggo A, Molony V (1978) The tract of Lissauer and thedorsal root potential. J Physiol 282:295–305

Chandler SH, Chase MH, Nakamura Y (1980) Intracellular analy-sis of synaptic mechanisms controlling trigeminal motoneuronactivity during sleep and wakefulness. J Neurophysiol 44:349–357

Charlton CG, Helke CJ (1985) Autoradiographic localization andcharacterization of spinal cord substance P binding sites: highdensities in sensory, autonomic, phrenic, and Onuf’s motor nu-clei. J Neurosci 5:1653–1661

Chase MH, Chandler SH, Nakamura Y (1980) Intracellular deter-mination of membrane potential of trigeminal motoneuronsduring sleep and wakefulness. J Neurophysiol 349–357

Chase MH, Soja PJ, Morales FR (1989) Evidence that glycine me-diates the postsynaptic potentials that inhibit lumbar motoneu-rons during the atonia of active sleep. J Neurosci 9:743–751

Chirwa SS, Stafford-Segert I, Soja PJ, Chase MH (1991) Strych-nine antagonizes jaw-closer motoneuron IPSPs induced by re-ticular stimulation during active sleep. Brain Res 547:323–326

Clamann HP, Lüscher H-R (1985) Quantal analysis of Ia-moto-neurone junction. Experientia 41:827

Clarac F, Cattaert D (1996) Invertebrate presynaptic inhibition andmotor control. Exp Brain Res 112:163–180

Collin E, Frechilla D, Pohl M, Bourgoin S, Bars D le, Hamon M,Cesselin F (1993a) Opioid control of the release of calcitoningene-related peptide-like material from the rat spinal cord invivo. Brain Res 609:211–222

Collin E, Mantelet S, Frechilla D, Pohl M, Bourgoin S, Hamon M,Cesselin F (1993b) Increased in vivo release of calcitoningene-related peptide-like material from the spinal cord in ar-thritic rats. Pain 54:203–211

Collins DF, Cameron T, Gillard DM, Prochazka A (1998) Muscu-lar sense is attenuated when humans move. J Physiol 508:635–643

Colmers WF, Klapstein GJ, Fournier A, St-Pierre S, Treherne KA(1991) Presynaptic inhibition by neuropeptide Y in rat hippo-campal slice in vitro is mediated by a Y2 receptor. Br J Phar-macol 102:41–44

Conradi S (1969) Ultrastructure of dorsal root boutons on lumbo-sacral motoneurons of the adult cat as revealed by dorsal rootsection. Acta Physiol Scand Suppl 332:85–117

Cook WA, Cangiano A (1972) Presynaptic and postsynaptic inhi-bition of spinal neurons. J Neurophysiol 35:389–403

Curtis DR (1998) Two types of inhibition in the spinal cord. In:Rudomin P, Romo R, Mendell L (eds) Presynaptic inhibitionand neural control. Oxford University Press, New York, pp150–161

Curtis DR, Lacey G (1998) Prolonged GABAB receptor-mediatedsynaptic inhibition in the cat spinal cord: an in vivo study. ExpBrain Res 121:319–333

Curtis DR, Malik R (1984) The effect of GABA on lumbar termi-nations of rubrospinal neurons in the cat spinal cord. Proc RSoc Lond B Biol Sci 223:25–33

Curtis DR, Lodge D, Bornstein JC, Peet MJ (1981) Selective ef-fects of (-)-baclofen on spinal synaptic transmission in the cat.Exp Brain Res 42:158–170

Curtis DR, Lodge D, Bornstein JC, Peet MJ, Leah JU (1982) Thedual effects of GABA and related aminoacids on the electricalthreshold of ventral horn group Ia afferent terminations in thecat. Exp Brain Res 48:387–400

Curtis DR, Wilson VJ, Malik R (1984) The effect of GABA on theterminations of vestibulospinal neurons in the cat spinal cord.Brain Res 295:372–375

Curtis DR, Gynther BD, Beattie DT, Lacey G (1995) An electro-physiological investigation of group Ia afferent fibres and ter-minations inthe cat spinal. Exp Brain Res 106:403–417

Curtis DR, Gynther BD, Lacey G, Beattie DT (1997) Baclofen: re-duction of presynaptic calcium influx in the cat spinal cord invivo. Exp Brain Res 113:520–523

Dale HH (1935) Pharmacology and nerve endings. Proc R SocMed 28:319–332

Davidoff RA (1972) Gamma-aminobutyric acid antagonism andpresynaptic inhibition in the frog spinal cord. Science 175:331–333

Davidoff RA, Hackman JC (1983) Drugs, chemicals, and toxins:their effects on the spinal cord. In: Davidoff RA (ed) Hand-book of the spinal cord, vol 1. Pharmacology. Dekker, NewYork Basel, pp 409–478

Davidoff RA, Hackman JC (1984) Spinal Inhibition. In: DavidoffRA (ed) Handbook of the spinal cord, vol 2 and 3. Anatomyand physiology. Dekker, New York Basel, pp 385–459

Destombes I, Horcholle-Bossavit G, Jami L, Thiesson D (1992)Alpha and gamma motoneurones in the peroneal nuclei of the

31

cat spinal cord: an ultrastructural study. J Comp Neurol317:79–90

Destombes J, Horcholle-Bossavit G, Simon M, Thiesson D (1996)Gaba-like immunoreactive terminals on lumbar motoneuronsof the adult cat. A quantitative ultrastructural study. NeurosciRes 24:123–130

Devanandan MS, Eccles RM, Stenhouse D (1966) Presynaptic in-hibition evoked by muscle contraction. J Physiol 185:471–485

Devor M (1983) Plasticity of spinal cord somatotopy in adultmammals: involvement of relatively ineffective synapses.Birth Defects 19:287–314

Dolphin AC, Huston E, Scott RJ (1990) GABAB-mediated inhibi-tion of calcium-currents: a possible role in presynaptic inhibi-tion. In: Bowery NG, Bittinger H, Olpe H-R (eds) GABAB re-ceptors in mammalian function. Wiley, New York, pp 259–271

Doyle CA, Maxwell DJ (1993) Neuropeptide Y immunoreactiveterminals form axo-axonic synaptic arrangements in the sub-stantia gelatinosa (lamina II) of the cat spinal cord. Brain Res603:157–161

Dubuc R, Cabelguen J-M, Rossignol S (1988) Rhythmic fluctua-tions of dorsal root potentials and antidromic discharges ofsingle primary afferents during fictive locomotion in the cat. JNeurophysiol 60:2014–2036

Dueñas S, Rudomin P (1988) Excitability changes of ankle exten-sor group Ia and Ib fibers during fictive locomotion in the cat.Exp Brain Res 70:15–25

Dueñas S, Rudomin P (1993) Reversal of the effects of cutaneousafferents on PAD of Ib fibers during fictive locomotion. SocNeurosci Abstr 19:225

Duggan AW (1995) Release of neuropeptides in the spinal cord.In: Nyberg F, Sharma HS, Wiesenfeld-Halllin Z (eds) Neuro-peptides in the spinal cord. Prog Brain Res 104: 197–223

Duggan AW, Hope PJ, Lang CW (1991) Microinjection of neuro-peptide Y into the superficial dorsal horn reduces stimulus-evoked release of immunoreactive substance P in the anaesthe-tized cat. Neuroscience 44:733–740

Duncan D, Morales R (1978) Relative numbers of several types ofsynaptic connections in the substantia gelatinosa of the cat spi-nal cord. J Comp Neurol 182:601–610

Eccles JC (1957) The physiology of nerve cells. Johns HopkinsPress, Baltimore

Eccles JC (1964) The physiology of synapses. Springer, BerlinHeidelberg New York

Eccles JC, Krnjevic K (1959) Potential changes recorded insideprimary afferent fibres within the spinal cord. J Physiol 149:250–273

Eccles JC, Eccles RM, Magni F (1961) Central inhibitory actionattributable to presynaptic depolarization produced by muscleafferent volleys. J Physiol 159:147–166

Eccles JC, Kostyuk PG, Schmidt RF (1962a) Central pathways re-sponsible for depolarization of primary afferent fibres. JPhysiol 161:237–257

Eccles JC, Magni F, Willis WD (1962b) Depolarization of centralterminals of group I afferent fibres from muscle. J Physiol160:62–93

Eccles JC, Schmidt RF, Willis WD (1962c) Presynaptic inhibitionof the spinal monosynaptic reflex pathway. J Physiol161:282–297

Eccles JC, Schmidt RF, Willis WD (1962d) Primary afferent depo-larization of group Ib afferent fibres of muscle. (XXII IntlCongr Physiol Sci). Excerpt Med 48:919

Eccles JC, Schmidt RF, Willis WD (1963a) Depolarization of cen-tral terminals of group Ib afferent fibres from muscle. J Neuro-physiol 26:1–27

Eccles JC, Schmidt RF, Willis WD (1963b) Depolarization of thecentral terminals of cutaneous afferent fibres. J Neurophysiol26:646–661

Eccles JC, Schmidt RF, Willis WD (1963c) Pharmacological stud-ies on presynaptic inhibition. J Physiol 168:500–530

Edamura M, Yang JF, Stein RB (1991) Factors that determine themagnitude and time course of human H-reflexes in locomo-tion. J Neurosci 11:420–427

Eguibar JR, Quevedo J, Jiménez I, Rudomin P (1994) Selectivecortical control of information flow through different intraspi-nal collaterals of the same muscle afferent fiber. Brain Res643:328–333

Eguibar JR, Quevedo J, Rudomin P (1997) Selective cortical andsegmental control of primary afferent depolarization of singlemuscle afferents in the cat spinal cord. Exp Brain Res 113:411–430

Eide E, Jurna I, Lundberg A (1968) Conductance measurementsfrom motoneurons during presynaptic inhibition. In: Euler Cvon, Skoglund S, Söderberg U (eds) Structure and function ofinhibitory neuronal mechanisms. Pergamon Press, OxfordNew York, pp 215–219

Ellis LC, Rustioni A (1981) A correlative HRP, Golgi and EMstudy of the intrinsic organization of the feline dorsal columnnuclei. J Comp Neurol 197:341–367

Enríquez M, Jiménez F, Rudomin P (1996a) Changes in PAD pat-terns of group I muscle afferents after peripheral nerve crush.Exp Brain Res 107:405–421

Enríquez M, Jiménez I, Rudomin P (1996b) Segmental and supra-spinal control of synaptic effectiveness of functionally identi-fied muscle afferents in the cat. Exp Brain Res 107:391–404

Enríquez M, Nielsen J, Perreault M-C, Morita H, Petersen N,Hultborn H (1996c) Changes in excitability of the terminals ofIa inhibitory interneurons in the cat spinal cord. J Physiol501:40P

Everitt BJ, Hökfelt T, Terenius L, Tatemoto K, Mutt V, GoldsteinM (1984) Differential co-existence of neuropeptide Y (NPY)-like immunoreactivity with catecholamine in the central ner-vous system of the rat. Neuroscience 11:443–462

Faist M, Dietz V, Pierrot-Deseilligny E (1996) Modulation, proba-bly presynaptic in origin, of monosynaptic Ia excitation duringhuman gait. Exp Brain Res 109:441–449

Fedirchuk B, Shefchyk S (1993) Membrane potential changes insphincter motoneurons during micturition in the decerebratecat. J Neurosci 13:3090–3094

Fedirchuk B, Downie JW, Shefchyk S (1994) Reduction of perine-al evoked excitatory postsynaptic potentials in cat lumbar andsacral motoneurons during micturition. J Neurosci 14:6153–6159

Fields HL, Emson PC, Leigh BK, Gilbert RFT, Iversen LL (1980)Multiple opiate receptor sites on primary afferent fibres. Na-ture 284:351–353

Fitzgerald M, Woolf CJ (1981) Effects of cutaneous nerve and in-traspinal conditioning on C fibre afferent terminal excitabilityin decerebrated spinal rats. J Physiol 318:25–39

Fitzgerald M, Wall PD, Goedert M, Emson PC (1985) Nervegrowth factor counteracts the neurophysiological and neuro-chemical effects of chronic sciatic nerve section. Brain Res232:131–141

Floeter MK, Sholomenko GN, Gossard JP, Burke RE (1993) Di-synaptic excitation from the medial longitudinal fasciculus tolumbosacral motoneurons: modulation by receptive activation,descending pathways, and locomotion. Exp Brain Res92:407–419

Forsythe ID (1994) Direct patch recording from identified presyn-aptic terminals mediating glutamatergic EPSCs in the rat CNS,in vitro. J Physiol 479:381–387

Franciolini F (1987) Spontaneous firing and myelination of verysmall axons. J Theor Biol 128:127–134

Frank K (1959) Basic mechanisms of synaptic transmission in thecentral nervous system. IEEE Trans Med Electr 6:85–88

Frank K, Fuortes MGF (1957) Presynaptic and postsynaptic inhi-bition of monosynaptic reflexes. Fed Proc 16:39–40

Fyffe REW, Light AR (1984) The ultrastructure of group Ia affer-ent fiber synapses in the lumbosacral spinal cord of the cat.Brain Res 300:201–209

Fyffe REW, Cheema SS, Rustioni A (1986) Intracellular stainingstudy of the feline cuneate nucleus. Terminal patterns of pri-mary afferent fibers. J Neurophysiol 56:1268–1283

Gage PW (1992) Activation and modulation of neuronal K+ chan-nels by GABA. Trends Neurosci 15:46–51

32

Gasser HS, Graham HT (1933) Potentials produced in the spinalcord by stimulation of the dorsal roots. Am J Physiol103:303–320

Glazer EJ, Basbaum AI (1981) Immunohistochemical localizationof leucine-enkephalin in the spinal cord of the cat: enkephalin-containing marginal neurons and pain modulation. J CompNeurol 196:377–390

Glazer EJ, Basbaum AI (1984) Axons which take up (3H)-seroto-nin are presynaptic to enkephalin immunoreactive neurons incat dorsal horn. Brain Res 298:386–391

Goldstein SS, Rall W (1974) Changes in action potential shapeand velocity for changing core conductor geometry. BiopsyJ14:731–757

Gossard JP (1996) Control of transmission in muscle group Ia af-ferents during fictive locomotion in the cat. J Neurophysiol76:4104–4112

Gouardères C, Beaudet A, Zajac J-M, Cros J, Quirion R (1991)High resolution radioautographic localization of [125I]-K-33-824-labelled mu-opioid receptors in the spinal cord of normaland deafferented rats. Neuroscience 43:197–209

Graham B, Redman SJ (1994) A simulation of action potentials insynaptic boutons during presynaptic inhibition. J Neurophysiol71:538–549

Granit R, Kellerth JO, Williams TD (1964) “Adjacent” and “re-mote” postsynaptic inhibition in motoneurones stimulated bymuscle stretch. J Physiol 174:453–472

Gray EG (1962) A morphological basis for presynaptic inhibition.Nature 193:82–83

Haas HL, Hermann A, Greene RW, Chan-Palay V (1987) Actionand location of neuropeptide tyrosine (Y) on hippocampalneurons of the rat in slice preparations. Comp Neurol257:208–215

Harrison PJ, Jankowska E (1984) Do interneurones in lower lum-bar segments contribute to the presynaptic depolarization ofgroup I muscle afferents in Clarke’s column? Brain Res295:203–221

Harrison PJ, Jankowska E (1989) Primary afferent depolarizationof central terminals of group II muscle afferents in the cat spi-nal cord. J Physiol 411:71–83

Hayashi R, Tako K, Tokuda T, Yanagisawa N (1992) Comparisonof amplitude of human soleus H-reflex during sitting andstanding. Neurosci Res 12:227–233

Henneman E, Lüscher H-R, Mathis J (1984) Simultaneoulsy ac-tive and inactive synapses of single Ia fibres on cat spinalmotoneurones. J Physiol 352:147–161

Hentall ID, Fields H (1979) Segmental and descending influenceson intraspinal thresholds of single C fibres. J Neurophysiol42:1527–1537

Holets VR, Hökfelt T, Rokaeus T, Terenius L, Goldstein M (1988)Locus coeruleus neurons in the rat containing neuropeptide Y,tyrosine hydroxylase or galanin and their efferent projectionsto the spinal cord, cerebral cortex and hypothalamus. Neuro-science 24:893–906

Hongo T, Jankowska E, Ohno T, Sasaki K, Yamashita H, YoshidaK (1983) The same interneurones mediate inhibition of dorsalspinocerebellar tract cells and lumbar motoneurones in the cat.J Physiol 342:160–180

Horch K (1978) Central responses of cutaneous sensory neuronsto peripheral nerve crush in the cat. Brain Res 151:581–586

Horch KW, Lisney SJW (1981) Changes in primary afferent depo-larization of sensory neurones during peripheral nerve regen-eration in the cat. J Physiol 313:287–299

Hori Y, Endo K, Takahashi T (1992) Presynaptic inhibitory actionof enkephalin on excitatory transmission in superficial dorsalhorn of rat spinal cord. J Physiol 450:673–685

Howland B, Lettwin JT, McCulloch WS, Pitts W, Wall PD (1955)Reflex inhibitions by dorsal root interaction. J Neurophysiol18:1–17

Hultborn H, Nielsen JB (1998) Modulation of transmitter releasefrom Ia afferents by their preceding activity – a “postactiva-tion depression”. In: Rudomin P, Romo R, Mendell L (eds)Presynaptic inhibition and neural control. Oxford UniversityPress, New York, pp 178–191

Hultborn H, Meunier S, Pierrot-Deseilligny E, Shindo M (1987)Changes in presynaptic inhibition of Ia fibres at the onset ofvoluntary contraction in man. J Physiol 389:757–772

Iles JF (1996) Evidence for cutaneous and corticospinal modula-tion of presynaptic inhibition of Ia afferents from the humanlower limb. J Physiol 491:197–207

Iles JF, Pisini JV (1992a) Cortical modulation of transmission inspinal reflex pathways of man. J Physiol 455:425–446

Iles JF, Pisini JV (1992b) Vestibular-evoked postural reactions inman and modulation of transmission in spinal reflex pathways.J Physiol 455:407–424

Jackson MB, Zhang SJ (1995) Action potential propagation andpropagation block by GABA in rat posterior pituitary nerveterminals. J Physiol 483:597–611

Jankowska E, Padel Y (1984) On the origin of presynaptic depo-larization of group I muscle afferents in Clarke’s column in thecat. Brain Res 295:195–201

Jankowska E, Riddell JS (1995) Interneurones mediating presyn-aptic inhibition of group II muscle afferents in the cat spinalcord. J Physiol 483:461–471

Jankowska E, Riddell JS (1998) Neuronal systems involved inmodulating synaptic transmission from group II muscle affer-ents. In: Rudomin P, Romo R, Mendell L (eds) Presynaptic in-hibition and neural control. Oxford University Press, NewYork, pp 315–328

Jankowska E, Roberts WJ (1972) Synaptic actions of single inter-neurones mediating reciprocal Ia inhibition of motoneurones. JPhysiol 222:623–642

Jankowska E, McCrea D, Rudomin P, Sykova E (1981a) Observa-tions on neuronal pathways subserving primary afferent depo-larization. J Neurophysiol 46:506–516

Jankowska E, McCrea D, Rudomin P, Sykova E (1981b) Observa-tions on neuronal pathways subserving primary afferent depo-larization. J Neurophysiol 46:506–516

Jankowska E, Riddell JS, McCrea DA (1993) Primary afferent de-polarization of myelinated fibres in the joint and interosseusnerves of the cat. J Physiol 466:115–131

Jankowska E, Szabo Läckberg Z, Dyrehag LE (1994) Effects ofmonoamines on transmission from group II muscle afferents insacral segments in the cat. Eur J Neurosci 6:1058–1061

Jankowska E, Krutki P, Szabo Läckberg Z, Hammar L (1995) Ef-fects of serotonin on dorsal horn dorsal spinocerebellar tractneurones. Neuroscience 67:489–495

Jankowska E, Hammar L, Djouhri L, Hedén CH, Szabo LäckbergZ, Yin X-K (1997) Modulation of responses of four types offeline ascending tract neurones by serotonin and noradrena-line. Eur J Neurosci 9:1375–1387

Jänig W, Schmidt RF, Zimmermann M (1967) Presynaptic depo-larization during activation of tonic mechanoreceptors. BrainRes 5:514–516

Jänig W, Schmidt RF, Zimmermann M (1968a) Single unit re-sponses and the total afferent outflow from the cat’s food padupon mechanical stimulation. Exp Brain Res 6:100–115

Jänig W, Schmidt RF, Zimmermann M (1968b) Two specific feed-back pathways to the central afferent terminal of phasic andtonic mechanoreceptors. Exp Brain Res 6:116–129

Jessell T, Tsunoo A, Kanazawa I, Otsuka A (1979) Substance P:deletion in the dorsal horn of rat spinal cord after section ofthe peripheral processes of primary sensory neurons. BrainRes 168:247–259

Ji RR, Zhang Q, Law P-Y, Low HH, Elde R, Hökfelt T (1995) Ex-pression of µ-, δ-, k-opioid receptor-like immunoreactivities inrat dorsal root ganglia after carrageenan-induced inflamma-tion. J Neurosci 15:8156–8166

Jiménez I, Rudomin P, Solodkin M, Vyklicky L (1983) Specificand potassium components in the depolarization of the Ia af-ferents in the spinal cord of the cat. Brain Res 272:179–184

Jiménez I, Rudomin P, Solodkin M, Vyklicky L (1984) Specificand nonspecific mechanisms involved in generation of PAD ofgroup Ia afferents in cat spinal cord. J Neurophysiol52:921–939

Jiménez I, Rudomin P, Solodkin M (1987) Mechanisms involvedin the depolarization of cutaneous afferents produced by seg-

33

mental and descending inputs in the cat spinal cord. Exp BrainRes 69:195–207

Jiménez F, Rudomin P, Solodkin M (1988) PAD patterns of physi-ologically identified afferent fibres from the medial gastrocne-mius muscle. Exp Brain Res 71:643–657

Jiménez I, Rudomin P, Enríquez M (1991) Differential effects of (-)-baclofen on Ia and descending monosynaptic EPSPs. ExpBrain Res 85:103–113

Jonas P, Bischofberger J, Sandkühler J (1998) Corelease of twofast neurotransmitters at a central synapse. Science 281:419–424

Kangra I, Randic M (1990) Tachykinins and calcitonin gene-relat-ed peptide enhance release of endogenous glutamate and as-partate from the rat spinal dorsal horn slice. J Neurosci10:2026–2038

Kar S, Quirion R (1992) Quantitative autoradiographic localiza-tion of [125I]-neuropeptide Y receptor binding sites in rat spi-nal cord and the effects of neonatal capsaicin, dorsal rhizoto-my and peripheral axotomy. Brain Res 574:333–337

Kellerth JO, Szumski AJ (1966a) Effects of picrotoxin on stretchactivated postsynaptic inhibition in spinal motoneurons. ActaPhysiol Scand 66:146–156

Kellerth JO, Szumski AJ (1966b) Two types of stretch activatedpostsynaptic inhibition s in spinal motoneurons as differentiat-ed by strychnine. Acta Physiol Scand 66:133–145

Koceja DM, Trimble MH, Earles DR (1993) Inhibition of the so-leus H-reflex in standing man. Brain Res 629:155–158

Kremer E, Lev-Tov A (1997) Antidromic discharge and intraspi-nal excitability of rat dorsal root afferents in the presence ofbicuculline. Soc Neurosci Abstr 23:2370

Kremer E, Lev-Tov A (1998) GABA-receptor-independent dorsalroots afferent depolarization in the neonatal rat spinal cord. JNeurophysiol 79:2581–2592

Krieger P, El Manira A, Grillner S (1986) Activation of pharmaco-logically distinct metabotropic glutamate receptors depressesreticulo-spinal evoked EPSPs in lamprey spinal cord. J Neuro-physiol 76:3834–3841

Kriz N, Syková E, Ujec E, Vyklicky L (1974) Changes of extracel-lular potassium concentration induced by neuronal activity inthe spinal cord of the cat. J Physiol 238:1–15

Kriz N, Syková E, Vyklicky L (1975) Extracellular potassiumchanges in the spinal cord of the cat and their relation to slowpotentials, active transport and impulse transmission. J Physiol249:167–182

Labella LA, Niechaj A, Rossignol S (1992) Low-threshold, short-latency cutaneous reflexes during locomotion in semi-chronicspinal cat. Exp Brain Res 91:236–248

Lafleur J, Zytnicki D, Horcholle-Bossavit G, Jami L (1992) Depo-larization of Ib afferent axons in the cat spinal cord during ho-monymous muscle contraction. J Physiol 445:345–354

LaMotte CC, Pert CB, Snyder SH (1976) Opiate receptor bindingin primate spinal cord: distribution and changes after dorsalroot section. Brain Res 112:407–412

Lamotte d’Incamps B, Destombes J, Thiesson D, Hellio R, Las-erre X, Kouchtir-Devanne N, Jami L, Zytnicki D (1997) Dem-onstration of axo-axonic contacts on the intraspinal arboriza-tion of Ib afferent fibers. Soc Neurosci Abstr 23:200

Lamotte d’Incamps B, Destombes J, Thiesson D, Hellio R, Lass-erre X, Kouchtir-Devanne N, Jami L, Zytnicki D (1998) Indi-cations for GABA-immunoreactive axo-axonic contacts on theintraspinal arborization of a Ib fiber in the cat: a confocal mi-croscope study. J Neurosci 18:10030–10036

Lev-Tov A, Meyers DER, Burke RE (1988) Activation of type Bγ-aminobutyric acid receptors in the intact mammalian spinalcord mimics the effects of reduced presynaptic Ca2+influx.Proc Natl Acad Sci USA 85:5330–5334

Levine JD, Fields HL, Basbaum AI (1993) Peptides and the pri-mary afferent nociceptor. J Neurosci 13:2273–2286

Li P, Zhuo M (1998) Silent glutamatergic synapses and nocicept-ion in mammalian spinal cord. Nature 393:695–698

Lisney SJW (1979) Evidence for primary afferent depolarizationof single tooth-pulp afferents in the cat. J Physiol 288: 437–447

Llewellyn M, Yang J, Prochazka A (1990) Human H-reflexes aresmaller in difficult beam walking than in normal treadmillwalking. Exp Brain Res 187:321–333

Lomelí J, Quevedo J, Linares P, Rudomin P (1998) Local controlof information flow in segmental and ascending collaterals ofsingle afferents. Nature 395:600–604

Lopez-Garcia JA, King AE (1996) Pre- and postsynaptic actionsof 5-hydroxytriptamine in the rat lumbar dorsal horn in vitro:implications for somatosensory transmission. Eur J Neurosci8:2188–2197

Lundberg A (1964) Supraspinal control of transmission in reflexpaths to motonerurons and primary afferents. In: Eccles J,Schadé JP (eds) Physiology of the spinal neurons. Elsevier,Amsterdam, pp 197–221

Lundberg A (1982) Inhibitory control from the brain stem oftransmission from primary afferents to motoneurons, primaryafferent terminals and ascending pathways. In: Sjölund B,Björklund A (eds) Brain stem control of spinal mechanisms.Elsevier Biomed Press, Amsterdam, pp 179–224

Lundberg A, Vyklicky L (1966) Inhibition of transmission to pri-mary afferents by electrical stimulation of the brain stem. ArchItal Biol 104:86–97

Lüscher C, Streit J, Lipp P, Lüscher H-R (1994a) Action potentialpropagation through embryonic dorsal root ganglion cells inculture. II. Decrease of conduction reliability during repetitivestimulation. J Neurophysiol 72:634–643

Lüscher C, Streit J, Quadroni R, Lüscher H-R (1994b) Action po-tential propagation through embryonic dorsal root ganglioncells in culture. I. Influence of cell morphology on propagationproperties. J Neurophysiol 72:622–633

Lüscher H-R (1990) Transmission failure and its relief in the spi-nal monosynaptic reflex arc. In: Binder MD, Mendel LM (eds)The segmental motor system. Oxford University Press, Ox-ford, pp 328–348

Lüscher H-R (1998) Control of action potential invasion into ter-minal arborizations. In: Rudomin P, Romo R, Mendell L (eds)Presynaptic inhibition and neural control. Oxford UniversityPress, New York, pp 126–137

Lüscher H-R, Shiner JS (1990a) Computation of action potentialpropagation and presynaptic bouton activation in terminal ar-borizations of different geometry. Biophys J 58:1377–1388

Lüscher H-R, Shiner JS (1990b) Simulation of action potentialpropagation in complex terminal arborization. Biophys J 58:1389–1399

Lüscher H-R, Mathis J, Schaffner H (1983) A dual-time voltagewindow discriminator for multinunit nerve spike decomposi-tion. J Neurosci Methods 7:99–105

MacDonald RL, Werz MA (1986) Dynorphin A decreases voltage-dependent calcium conductance of mouse dorsal root ganglionneurones. J Physiol 377:237–249

Magni F, Melzack R, Moruzzi G, Smith THJ (1959) Direct pyra-midal influences on the dorsal-column nuclei. Arch Ital Biol97:357–377

Manor Y, Gonczarowski J, Segev I (1991a) Propagation of actionpotentials along complex axonal trees. Model and implementa-tion. Biophys J 60:1411–1423

Manor Y, Koch C, Segev I (1991b) Effect of geometrical irregu-larities on propagation delay in axonal trees. Biophys J60:1424–1437

Martin RF, Haber LH, Willis WD (1979) Primary afferent depolar-ization of identified cutaneous fibers following stimulation ofthe medial brainstem. J Neurophysiol 42:779–790

Maxwell DJ, Riddell JS (1999) Axo-axonic synapses on terminalsof identified group II muscle spindle afferent axons in the catspinal cord. Eur J Neurosci 11:2151–2159

Maxwell DJ, Christie WM, Short AD, Brown AG (1990) Directobservations of synapses between GABA-immunoreactiveboutons and muscle afferent terminals in lamina VI of the cat’sspinal cord. Brain Res 53:215–222

Maxwell DJ, Ottersen OP, Storm-Mathisen J (1995) Synaptic or-ganization of excitatory and inhibitory boutons associated withspinal neurons which project through the dorsal columns in thecat. Brain Res 676:103–112

34

Maxwell DJ, Kerr R, Jankowska E, Riddell JS (1997) Synapticconnections of dorsal horn group II spinal interneurons: syn-apses formed with the interneurons and by their axon collater-als. J Comp Neurol 380:51–69

McCrea DA, Sefchyk SJ, Carlen PL (1990) Large reductions incomposite EPSP amplitude following conditioning stimulationare not accounted for by increased postsynaptic conductancesin motoneurons. Neurosci Lett 109:117–122

Melzack R, Wall PD (1965) Pain mechanisms: a new theory. Sci-ence 150:971–979

Mercuri B, Wassermann EM, Ikoma K, Samii A, Hallett M (1997)Effects of transcranial electrical and magnetic stimulation onreciprocal inhibition in the human arm. ElectroencephalogrClin Neurophysiol 105:87–93

Meunier S, Pierrot-Deseilligny E (1989) Gating of the afferentvolley of the monosynaptic stretch reflex during movement inman. J Physiol 419:753–763

Miller RJ (1998) Presynaptic receptors. Annu Rev PharmacolToxicol 38:201–227

Misiaszek JE, Brooke JD, Lafferty KB, Cheng J, Staines WR(1995) Long-lasting inhibition of the human soleus H reflexpathway after passive movement. Brain Res 677:69–81

Morita H, Shindo M, Yanagawa S, Yoshida T, Momoi H, Yanagi-sawa N (1995) Progressive decrease in heteronymous mono-synaptic Ia facilitation with human ageing. Exp Brain Res104:167–170

Mott DD, Lewis DV (1994) The pharmacology and function ofcentral GABAB receptors. Int Rev Neurobiol 36:97–223 1994

Mudge AW, Leeman SE, Fischbach GD (1979) Enkephalin inhib-its the release of substance P from sensory neurons in cultureand decreases action potential duration. Proc Natl Acad SciUSA 76:5487–5491

Neto FR (1978) The depolarizing action of 5-HT on mammaliannon-myelinated nerve fibres. Eur J Pharmacol 49:351–352

Nicol MJ, Walmsley B (1991) A serial section electron micro-scope study of an identified Ia afferent collateral in the cat spi-nal cord. J Comp Neurol 314:257–277

Nicoll RA, Alger BE (1979) Presynaptic inhibition: transmitterand ionic mechanisms. Int Rev Neurobiol 21:217–258

Nicolls J, Wallace BG (1978) Modulation of transmission at an in-hibitory synapse in the central nervous system of the leech. JPhysiol 281:157–170

Nielsen J, Kagamihara Y (1993) The regulation of presynaptic in-hibition during co-contraction of antagonistic muscles in man.J Physiol 464:575–593

Nielsen J, Petersen N, Crone C (1995) Changes in transmissionacross synapses of Ia afferents in spastic patients. Brain 118:995–1104

Nistri A (1983) Spinal cord pharmacology of GABA and chemi-cally related amino acids. In: Davidoff RA (ed) Handbook ofthe spinal cord, vol 1. Pharmacology. Dekker, New York Ba-sel, pp 45–104

Noga BR, Bras H, Jankowska E (1992) Transmission from groupII muscle afferents is depressed by stimulation of locus coeru-leus/subcoeruleus, Kölliker-Fuse and raphe nuclei in the cat.Exp Brain Res 88:502–516

Nusbaum MP, El Manira A, Gossard JP, Rossignol S (1997) Pre-synaptic mechanisms during rhythmic activity in vertebratesand invertebrates. In: Stein PSG, Grillner S, Selverston AI,Stuart DG (eds) Neurons, networks and motor behaviour. MITPress, Cambridge, pp 237–253

Ohta Y, Grillner S (1989) Monosynaptic excitatory aminoacidtransmission from the posterior rhombencephalic reticular nu-cleus to spinal neurons involved in the control of locomotionin the lamprey. J Neurophysiol 62:1079–1089

Parnas I (1972) Differential block at high frequency of branches ofa single axon innervating two muscles. J Neurophysiol 35:903–914

Parnas I, Segev I (1979) A mathematical model for conduction ofaction potentials along bifurcating axons. J Physiol 295:323–343

Pedroarena C, Castillo P, Chase MH, Morales FR (1994) The con-trol of jaw-opener motoneurons during active sleep. Brain Res653:31–38

Peng Y, Frank E (1989a) Activation of GABAA receptors causespresynaptic and postsynaptic inhibition at synapses betweenmuscle spindle afferents and motoneurons in the spinal cord ofbullfrogs. J Neurosci 9:1516–1522

Peng Y, Frank E (1989b) Activation of GABAB receptors causespresynaptic inhibition at synapses between muscle spindle af-ferents and motoneurons in the spinal cord of bullfrogs. J Neu-rosci 9:1502–1515

Peroutka SJ (1988) 5-Hydroytryptamin receptor subtypes. AnnuRev Neurosci 11:45–60

Pierce JP, Mendell LM (1993) Quantitative ultrastructure of Iaboutons in the ventral horn: scaling and positional relation-ships. J Neurosci 13:4748–4763

Pierrot-Deseilligny E, Meunier S (1998) Differential control ofpresynaptic inhibition of Ia terminals during voluntary move-ment in humans. In: Rudomin P, Romo R, Mendell L (eds)Presynaptic inhibition and neural control. Oxford UniversityPress, New York, pp 351–365

Plotkin MD, Kaplan MR, Peterson LN, Gullans SR, Herbert SC,Deplire E (1997) Expression of the Na+-K+-2Cl– cotransporterBSC2 in the nervous system. Am J Physiol 267:C173-C183

Pohl M, Collin E, Bourgoin S, Clot AM, Hamon M, Cesselin F,Bars D le (1992) In vivo release of calcitonin gene-relatedpeptide-like material from the cervicotrigeminal area in therat. Effects of electrical and noxious stimulations of the muz-zle. Neuroscience 50:697–706

Prochazka A (1996) Proprioceptive feedback and movement regu-lation. In: Rowell L, Shepard JT (eds) Handbook of physiolo-gy. Section 12. Exercise: regulation and integration of multiplesystems. American Physiological Society, New York, pp89–127

Quevedo J, Eguibar JR, Jiménez I, Rudomin P (1992) Differentialaction of (-)baclofen on primary afferent depolarization pro-duced by segmental and descending inputs. Exp Brain Res91:29–45

Quevedo J, Eguibar JR, Jiménez I, Schmidt RF, Rudomin P (1993)Primary afferent depolarization of muscle afferents elicited bystimulation of joint afferents in cats with intact neuraxis andduring reversible spinalization. J Neurophysiol 70:1899–1910

Quevedo J, Equibar, JR, Jiménez I, Rudomin P (1995) Raphemagnus and reticulopsinal actions on primary afferent depolar-ization of group I muscle afferents in the cat. J Physiol482:623–640

Quevedo J, Eguibar JR, Lomelí J, Rudomin P (1997) Patterns ofconnectivity of spinal interneurons with single muscle affer-ents. Exp Brain Res 115:387–402

Ramón y Cajal R (1899) Textura del Sistema Nervioso del Hom-bre y de los Vertebrados. Nicolás Moya, Madrid

Randic M, Semba K, Murase K (1980) Presynaptic effects of γ-aminobutyric acid on intraspinal single cutaneous afferent C-and Aδ-fibers. Proc Soc Neurosci 10:571

Redman S (1990) Quantal analysis of synaptic potentials in neu-rons of the central nervous system. Physiol Rev 70:165–198

Ribeiro-da-Silva A (1995) Substantia gelatinosa of the spinal cord.In: Paxinos G (ed) The Rat nervous system. Academic Press,London, pp 47–59

Richter DW, Jordan D, Ballantyne D, Meesmann M, Spyer KM(1986) Presynaptic depolarization in myelinated vagal afferentfibres terminating in the nucleus of the tractus solitarius in thecat. Pflügers Arch 406:12–19

Riddell JS, Jankowska E, Eide E (1993) Depolarization of groupII muscle afferents by stimuli applied in the locus coeruleusand raphe nuclei of the cat. J Physiol 461:723–741

Riddell JS, Jankowska E, Huber J (1995) Organization of neuronalsystems mediating presynaptic inhibition of group II muscleafferents in the cat. J Physiol 483:443–460

Rossignol S, Beloozerova IGJP, Dubuc R (1998) Presynapticmechanisms during locomotion. In: Rudomin P, Romo R,Mendell L (eds) Presynaptic inhibition and neural control. Ox-ford University Press, New York, pp 385–397

Ruda MA (1982) Opiates and pain pathways: demonstration of en-kephalin synapses on dorsal horn projection neurons. Science215:1523–1525

35

Rudomin P (1967) Presynaptic inhibition induced by vagal affer-ent volleys. J Neurophysiol 30:964–981

Rudomin P (1968) Excitability changes of superior laryngeal, va-gal and depressor afferent terminals produced by stimulationof the solitary tract nucleus. Exp Brain Res 6:156–170

Rudomin P (1990a) Presynaptic control of synaptic effectivenessof muscle spindle and tendon organ afferents in the mammali-an spinal cord. In: Binder MD, Mendell LM (eds) The seg-mental motor system. Oxford University Press, New York, pp349–380

Rudomin P (1990b) Presynaptic inhibition of muscle spindle andtendon organ afferents in mammalian spinal cord. Trends Neu-rosci 13:499–505

Rudomin P (1992) Validation of the changes in heterosynaptic fa-cilitation of monosynaptic responses of spinal motoneurons asa test for presynaptic inhibition. In: Jami L, Pierrot-Des-eilligny E, Zytnicki D (eds) Muscle afferents and spinal con-trol of movement. Pergamon Press, Oxford New York, pp 439

Rudomin P, Romo R, Mendell LM (eds) (1998) Presynapticinhibition and neural control. Oxford University Press, NewYork

Rudomin P, Madrid J (1972) Changes in correlation betweenmonosynaptic reflexes and in information transmission pro-duced by conditioning volleys to cutaneous nerves. J Neuro-physiol 35:44–64

Rudomin P, Muñoz-Martínez J (1969) A tetrodotoxin-resistant pri-mary afferent depolarization. Exp Neurol 25:106–115

Rudomin P, Núñez R, Madrid J (1975) Modulation of synaptic ef-fectiveness of Ia and descending fibers in cat spinal cord. JNeurophysiol 38:1181–1195

Rudomin P, Engberg I, Jiménez I (1981) Mechanisms involved inpresynaptic depolarization of group I and rubrospinal fibers incat spinal cord. J Neurophysiol 46:532–548

Rudomin P, Jiménez I, Solodkin M, Dueñas S (1983) Sites of ac-tion of segmental and descending control of transmission onpathways mediating PAD of Ia- and Ib-afferent fibers in catspinal cord. J Neurophysiol 50:743–769

Rudomin P, Solodkin M, Jiménez I (1986) PAD and PAH responsepatterns of group Ia- and Ib-fibers to cutaneous and descend-ing inputs in the cat spinal cord. J Neurophysiol 56:987–1006

Rudomin P, Solodkin M, Jiménez I (1987) Synaptic potentials ofprimary afferent fibers and motoneurons evoked by single in-termediate nucleus interneurons in the cat spinal cord. J Neu-rophysiol 57:1288–1313

Rudomin P, Jiménez I, Quevedo J, Solodkin P (1990) Pharmaco-logical analysis of inhibition produced by last-order intermedi-ate nucleus interneurons mediating non-reciprocal inhibitionof motoneurons in cat spinal cord. J Neurophysiol 63:147–160

Rudomin P, Jiménez I, Enríquez M (1991) Effects of stimulationof group I afferents from flexor muscles on heterosynaptic fa-cilitation of monosynaptic reflexes produced by Ia and de-scending inputs: a test for presynaptic inhibition. Exp BrainRes 85:93–102

Rudomin P, Jiménez F, Quevedo J (1998) Selectivity of the pre-synaptic control of synaptic effectiveness of group I afferentsin the mammalian spinal cord. In: Rudomin P, Romo R, Men-dell L (eds) Presynaptic inhibition and neural control. OxfordUniversity Press, New York, pp 282–302

Santicioli P, Del Bianco E, Maggi CA (1993) Adenosine A1 recep-tors mediate the presynaptic inhibition of calcitonin gene-re-lated peptide release by adenosine in the rat spinal cord. Eur JPharmacol 231:139–142

Sastry ER (1978) Morphine and met-enkephalin effects on suralAδ-afferent terminal excitability. Eur J Pharmacol 50:269–273

Schaible H-G (1996) On the role of tachykinins and calcitoningene-related peptide in the spinal mechanisms of nociceptionand in the induction and maintenance of inflammation-evokedhyperexcitability in spinal cord neurons (with special refer-ence to nociception in joints). In: Kumazawa T, Kruger L, Mi-zumura K (eds) The polymodal receptor – a gateway to patho-logical pain. Elsevier, Amsterdam New York Tokyo, pp423–441

Schaible H-G, Grubb BD (1993) Afferent and spinal mechanismsof joint pain. Pain 55:5–54

Schmidt RF (1963) Pharmacological studies on the primary affer-ent depolarization of the toad spinal cord. Pflügers Arch277:325–346

Schmidt RF (1971) Presynaptic inhibition in the vertebrate centralnervous system. Ergeb Physiol 63:20–101

Schmidt RF, Schaible H-G (1998) Modulation of nociceptive in-formation at the presynaptic terminals of primary afferent fi-bers. In: Rudomin P, Romo R, Mendell L (eds) Presynaptic in-hibition and neural control. Oxford University Press, NewYork, pp 276–285

Segev I (1990) Computer study of presynaptic inhibition control-ling spread of action potentials into axonal terminals. J Neuro-physiol 63:987–998

Shefchyk S (1998) Modulation of excitatory perineal reflexes andsacral striated sphincter motoneurons during micturition in thecat. In: Rudomin P, Romo R, Mendell L (eds) Presynaptic in-hibition and neural control. Oxford University Press, NewYork, pp 398–406

Shefner JM, Buchthal F, Krarup C (1992) Recurrent potentials inhuman peripheral sensory nerve: possible evidence of primaryafferent depolarization of the spinal cord. Muscle Nerve15:1354–1363

Shimoda N, Takakusaki K, Nishizawa O, Tsuchida S, Mori S(1992) The changes in the activity of pudendal motoneurons inrelation to reflex micturition evoked in decerebrate cats. Neu-rosci Lett 135:175–178

Shupliakov O, Pieribone VA, Gad H, Brodin L (1995) Synapticvescicle depletion in reticulospinal axons is reduced by 5-hy-droxytryptamine-direct evidence for presynaptic modulationof glutamatergic transmission. Eur J Neurosci 7:1111–1116

Singer JH, Berger AJ (1996) Presynaptic inhibition by serotonin: apossible mechanism for switching motor output of the hypo-glossal nucleus. Sleep 19:146–149

Smith DO (1980) Mechanisms of action potential propagation fail-ure at sites of axon branching in the crayfish. J Physiol301:243–259

Smullin DH, Skilling SR, Larson AA (1990) Interactions betweensubstance P, calcitonin gene-related peptide, taurine and exci-tatory amino acids in the spinal cord. Pain 42:93–101

Soja PJ, López-Rodriguez F, Morales FR (1991) The postsynapticinhibitory control of lumbar motoneurons during the atonia ofactive sleep: effect of strychnine on motoneuron properties. JNeurosci 11:2804–2811

Solodkin M, Jiménez I, Rudomin P (1984) Identification of com-mon interneurons mediating pre- and postsynaptic inhibitionin the cat spinal cord. Science 224:1453–1436

Solodkin M, Jiménez I, Collins III WF, Mendel LM, Rudomin P(1991) Interaction of baseline synaptic noise and Ia EPSPs:evidence for appreciable negative correlation under physiolog-ical conditions. J Neurophysiol 65:927–945

Somjen GG, Lothman EW (1974) Potassium, sustained focal po-tentials shifts and dorsal root potentials of the mammalian spi-nal cord. Brain Res 69:153–157

Stein RB (1995) Presynaptic inhibition in humans. Prog Neurobiol47:533–544

Stockbridge N (1988) Differential conduction at axon bifurcation.II. Theoretical basis. J Neurophysiol 59:1286–1295

Stockbridge N, Stockbridge LL (1988) Differential conduction ataxon bifurcation. I. Effects of electrotonic length. J Neuro-physiol 59:1277–1285

Stuart GJ, Redman SJ (1992) The role of GABAA and GABAB re-ceptors in presynaptic inhibition of Ia EPSPs in cat spinalmotoneurones. J Physiol 447:675–692

Sumal KK, Pickel VM, Miller RJ, Reis DJ (1982) Enkephalincontaining neurons in substantia gelatinosa of spinal trigemi-nal complex: ultra structure and synaptic interaction with pri-mary sensory afferents. Brain Res 248:223–236

Swadlow HA, Kocsis JD, Waxman SG (1980) Modulation of im-pulse conduction along the axonal tree. Annu Rev BiophysBioeng 9:143–179

36

Takahashi A, Otsuka M (1975) Regional distribution of substanceP in the spinal cord and nerve roots of the cat and the effect ofdorsal root section. Brain Res 87:1–11

Takano M, Takano Y, Yaksh TL (1993) Release of calcitoningene-related peptide (CGRP), substance P (SP), and vasoac-tive intestinal polypeptide (VIP) from rat spinal cord: modula-tion by α2 agonists. Peptides 14:371–378

Tessler A, Himes BT, Soper K, Murray M, Goldberger ME, Reich-lin S (1984) Recovery of substance P but not somatostatin inthe cat spinal cord after unilateral lumbosacral dorsal rhizoto-my: a quantitative study. Brain Res 305:95–102

Thompson SWN, Wall PD (1996) The effect of GABA and 5-HTreceptor antagonists on rat dorsal root potentials. NeurosciLett 217:153–156

Todd AJ (1996) GABA and glycine in synaptic glomeruli of therat spinal dorsal horn. Eur J Neurosci 8:2492–2498

Toennies JF (1938) Reflex discharge from the spinal cord over thedorsal roots. J Neurophysiol 1:378–390

Uchizono K (1975) Excitation and inhibition. Synaptic morpholo-gy. Igaku Shoin, Tokyo

Walberg F (1965) Axo-axonic contacts in the cuneate nucleus,probable basis for presynaptic inhibition. Exp Neurol 13:218–231

Walker JL (1971) Ion specific liquid ion exchanger microelec-trodes. Anal Chem 39:89–92

Wall PD (1958) Excitabilitiy changes in afferent fibre terminationsand their relation to slow potentials. J Physiol 142:1–21

Wall PD (1962) The origin of a spinal cord slow potential. JPhysiol 164:508–526

Wall PD (1994) Control of impulse conduction in long range af-ferents. Eur J Neurosci 6:1136–1142

Wall PD (1995) Do nerve impulses penetrate terminal arboriza-tions? A presynaptic control mechanism. Trends Neurosci18:99–103

Wall PD, Devor M (1984) The effect of peripheral nerve injury ondorsal root potentials and on transmission of afferent signalsinto the spinal cord. Brain Res 209:95–111

Wall PD, Lidierth M (1997) Five sources of a dorsal root poten-tial: their interactions and origins in the superficial dorsalhorn. J Neurophysiol 78:860–871

Wall PD, McMahon SB (1994) Long range afferents in rat spinalcord. III. Failure of impulse transmission in axons and relief ofthe failure after rhizotomy of dorsal roots. Philos Trans R SocLond (Biol) 343:211–223

Wall PD, Fitzgerald M, Gibson SJ (1981) The response of rat spi-nal cord cells to unmyelinated afferents after peripheral nervesection and after changes in substance P levels. Neuroscience11:2205–2215

Walmsley B, Nicol MJ (1998) Action potential propagation alongprimary afferents and presynaptic inhibition in Clarke’s col-umn of the spinal cord. In: Rudomin P, Romo R, Mendell L(eds) Presynaptic inhibition and neural control. Oxford Uni-versity Press, New York, pp 138–149

Walmsley B, Wieniawa-Narkiewicz E, Nicol MJ (1987) Ultra-structural evidence related to presynaptic inhibition of primaryafferents in Clarke’s column of the cat. J Neurosci 7:236–243

Walmsley B, Graham B, Nicol MJ (1995) Serial E-M and simula-tion study of presynaptic inhibition along a group of Ia collat-eral in the spinal cord. J Neurophysiol 74:616–623

Waxman SG, Wood SL (1984) Axon conduction in inhomoge-neous axons: effects of variation in voltage sensitive ionicconductances on invasion of demyelinated axon segments andpreterminal fibers. Brain Res 294:111–122

Werz MA, MacDonald RL (1983) Opioid peptides selective formu- and delta-opiate receptors reduce calcium-dependent ac-tion potential duration by increasing potassium conductance.Neurosci Lett 42:173–178

Whitehorn D, Burgess PR (1973) Changes in the polarization ofcentral branches of myelinated mechanoreceptor and nocicep-tor fibers during noxious and innocuous stimulation of theskin. J Neurophysiol 36:226–237

Wikstrom M, El Manira A, Zhang W, Grillner S (1995) Dopamineand 5HT-modulation of synaptic transmission in the lampreyspinal cord. Soc Neurosci Abstr 20:1145

Willis WD (1999) Dorsal root potentials and dorsal root reflexes:a double-edged sword. Exp Brain Res 124:395–421

Willis WD, Sluka KA, Rees H, Westlund KN (1998) A contribu-tion of dorsal root reflexes to peripheral inflammation. In: Ru-domin P, Romo R, Mendell L (eds) Presynaptic inhibition andneural control. Oxford University Press, New York, pp407–423

Wood SA, Gregory JE, Proske U (1996) The influence of musclespindle discharge on the human H reflex and the monosynap-tic reflex in the cat. J Physiol 497:179–200

Wu W, Ziskind-Conheim L, Sweet MA (1992) Early developmentof glycine- and GABA-mediated synapses in rat spinal cord. JNeurosci 12:3935–3945

Yang JF, Whelan PJ (1993) Neural mechanisms that contribute tocyclical modulation of the soleus H-reflex in walking in hu-mans. Exp Brain Res 95:547–556

Zajac J-M, Lombard M-C, Peschanski M, Besson J-M, Roques BP(1989) Autoradiographic study of mu and delta opioid bindingsites and neutral endopeptidase-24,11 in rat after dorsal rootrhizotomy. Brain Res 477:400–403

Zanette G, Bonato C, Polo A, Tinazzi M, Manganotti P, Fiaschi A(1995) Long-lasting depression of motor-evoked potentials totranscranial magnetic stimulation following exercise. ExpBrain Res 107:80–86

Zhang SJ, Jackson MB (1995a) Properties of the GABAA receptorof rat posterior pituitary nerve terminals. J Neurophysiol73:1135–1144

Zhang SJ, Jackson MB (1995b) GABAA receptor activation andthe excitability of nerve terminals in the rat posterior pituitary.J Physiol 483:583–595

Zhu CG, Sandri C, Akert K (1981) Morphological identificationof axo-axonic and dendro-dendritic synapses in the rat sub-stantia gelatinosa. Brain Res 230:25–40

Zytnicki D, Jami L (1998) Presynaptic inhibition can act as a filterof input from tendon organs during muscle contraction. In:Rudomin P, Romo R, Mendell L (eds) Presynaptic inhibitionand neural control. Oxford University Press, New York, pp303–314

Zytnicki D, L’Hôte G (1993) Neuromimetic model of a neuronalfilter. Biol Cybern 70:115–121

Zytnicki D, Lafleur J, Horcholle-Bossavit G, Lamy F, Jami L(1990) Reduction of Ib autogenetic inhibition in motoneuronsduring contractions of an ankle extensor muscle in the cat. JNeurophysiol 64:1380–1389

37