Article history:Accepted 5 February 2008Available online 21 February 2008
Contrary to the widespread assumption, the filum terminale in the rat possesses a precise glialandneuronal organization. Theprocesses of glial fibrillary acidic protein-stained astrocytes forma rich, three dimensional array. The crescent shaped white matter could be outlined withantibody detecting oligodendrocytes. The neurons in the filum terminale, labeled with neuron-specific nuclear protein, are distributed in a smallmidline group (dorsal nucleus) dorsal to and intwo symmetrical clusters at both sides of the central canal (lateral nuclei). Nitric oxide synthase-,calretinin-, choline acetyltransferase-, substance P- and neurokinin receptor-1-immunoreactiveneurons were detected in the lateral nuclei. Axons were classified based on their course andtermination. Small number of calcitonin gene-related peptide-immunoreactive fiberswas foundexclusively in the dorsal nucleus. Nitric oxide synthase-, substance P-, and neurokinin receptor-1-stained axon arborizations were detected mainly in the lateral nucleus. A dense array ofextremely fine vesicular glutamate transporter 2- and fine, synaptophysin-immunoreactivevaricosities covered densely the lateral nuclei. Fine glycine-transporter 2-immunoreactive axonarborization like structureswere seenalso in the lateral nucleus. Vesicular glutamate transporter1- and choline acetyltransferase-immunoreactive axons arborized in the entire gray matter.Serotonin- and enkephalin-immunoreactive fibers congregated in the dorsolateral portion of thewhitematter, called “shoulder region”, while calretinin- and thick, varicose neurokinin receptor-1-stained axons were also seen in the same area of the white matter. Synaptophysin-immunoreactive fine varicosities colocalized only with vesicular glutamate transporter 2immunoreaction. Substance P and glycine-transporter 2-immunoreactive puncta were foundin close contact with neurokinin receptor-1-immunostained perikarya and dendrites.
Filum terminale is known as the rudimentary terminal portionof the spinal cord. It develops as the consequence of theunequal growth rate of the vertebral canal versus spinal cord(Streeter, 1919; Pinto et al., 2002). The distal end of the filum
terminale is attached to the terminal portion of the vertebralcanal, while its proximal end is the direct continuation of theconusmedullaris. The dorsal and ventral roots of the ultimatespinal nerve originate from the conus medullaris, thereforethe filum terminale has no direct connection with the peri-phery via spinal nerves.
1. Total view of a formalin fixed conus medullaris(short, wider, upper portion) and filum terminale.Arrow points to a transected dorsal root attached tothe side of and running parallel with the filumterminale. Gradation: 1 mm.
2. Cross section of filum terminale, single opticalsection. GFAP-stained astrocyte processes form aradial arrangement starting from the unstainedependyma of the central canal (CC). Scale: 100 μm.
3. Cross section of filum terminale, single opticalsection. RIP-stained oligodendrocytes are clusteringin the crescent shape peripheral white matter.Strong immunoreaction can be seen in the dorsalpeaks of the white matter (arrows). The gray matterconsists of a midline dorsal nucleus (DN) and twosymmetrical lateral nuclei (LN) at both sides of thecentral canal (CC). Scale: 100 μm.
106 B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
Text books describe the human filum terminale as “avestige of the spinal cord of the embryonic tail”, it consists ofpia mater as well as neuroglia, and it has no functionalsignificance in the adult (Barr and Kiernan, 1993). Thisgenerally accepted opinion is echoed by several recent clinicalpapers (Yamada et al., 2001; Selcuki et al., 2003). In an earlierpublication (Réthelyi et al., 2004) we described the gradualchanges of the gray matter at the junction between conusmedullaris and the filum terminale. It was emphasized thatthe gross anatomical structure of the filum terminale wasidentical with that of the spinal cord. The central canal thatcontinues in the filum terminale is surrounded by neuronalperikarya forming the classical gray matter. Size of theperikarya was 8 to 15 µm. The ependymal cells lining thecentral canal carried numerous cilia. Axodendritic and axo-somatic synapses were found in the gray matter where thepresynaptic axons contained various kinds of synaptic vesi-cles. Myelinated and unmyelinated fibers coursed in islandsbetween the perikarya as well as were the obligatory compo-nents in the white matter located at the periphery of the filumterminale.
Degenerated and liquor contact neurons were found in thefilum terminale of the frog (González-Robles and Glusman,1979; Chesler and Nicholson, 1985). More recently, also thephysiological character of the neurons in the frog filumterminale were described (Chvátal et al., 2001). Axons withnormal and abnormal ultrastructure (Miller, 1968), and liquor
contact neurons were shown in the filum terminale of the cat(Rascher et al., 1988), while degenerated neurons and nervefibers were detected in the human filum terminale (Choi et al.,1992; Gamble, 1971). The authors of the above mentionedmorphological studies all agreed in the occurrence of glial cellsin the filum terminale.
In the present paper using multiple immunohistochemistryand confocal microscopy we will demonstrate for the first time1) the structure of the glial components, 2) the number, distri-bution and chemical character of the neuronal perikarya and 3)the architecture and chemical character of the nerve fibers inthe filum terminale of the rat. With this diverse approach wewould like to show that in the rat the neuronal and glial orga-nization of the spinal cord does not terminate with the conusmedullaris, but continues in the filum terminale.
Preliminary results have been published in the form of con-ference abstract (Boros et al., 2005).
2. Results
2.1. Gross structure of the filum terminale
The filum terminale is 2.4 to 2.8 cm long in the rat (Plate I, Fig. 1).At the junctionwith the conusmedullaris thewidth of the filumterminale measured 0.5 to 0.6 mm, which declined to 0.2 to0.3 mm 5 mm more distally. The final dorsal roots were oftenattached sideways to the filum terminale, this prevented theexact widthmeasurements in fixed spinal cord preparations.—The first 1.0 cmlongportionof the filumterminalewasobservedin this study.
2.2. Glial structure of the filum terminale
Since the filum terminale is known, incorrectly, as a thin con-tinuation and prolongation of the spinal cord that contains onlyglial elements, first the glial architecture will be described. Glialfibrillary acidic protein (GFAP)-immunoreactive (IR) astrocytesabound in the filum terminale. Their processes showed a pre-vailing radial arrangement in cross sections (Plate 1, Fig. 2) and awidespread longitudinal orientation in the longitudinally sec-tioned filum terminale (Plate II, Fig. 6). Medium power picturedisclosed GFAP-stained rings of 4–6 μmdiameter indicating thenarrow cytoplasm of the perikarya of the astrocytes. GFAP-stained processes did not enclose neuronal perikarya (Plate II,Fig. 7). GFAP immunoreaction does not differentiate betweengray andwhitematters. In contrast, receptor interacting protein(RIP) immunoreaction, detecting oligodendroglia cells, wasweek in the gray matter, while the peripheral white matterwashighlightedby stronger RIP staining (Plate I, Fig. 3). Unlike inthe spinal cord, the white matter is absent at the dorsalcircumference of the filum terminale, the RIP immunoreactionshowed a dorsally open crescent shape. Especially strongRIP reaction was found in the dorsal tips of the white matter.Fig. 3 in Plate I indicates that the gray matter of the filumterminale can be subdivided into a midline dorsal nucleus andtwo symmetrical lateral nuclei located at both sides of thevertically elongated central canal. The sites of the strong RIPstaining will be called – on anatomical ground – as “shoulderregion”.
107B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
2.3. Distribution and number of the neurons in the filumterminale
Neuronal perikarya, as labeled by antibody detecting neuron-specific nuclear protein (NeuN),were distributed in threegroupsin the gray matter in the cranial portion of the filum terminale(Plate II, Figs. 4 and 5). As it was mentioned earlier, a smallermidline group of neurons is capping the dorsal edge of thecentral canal (dorsal nucleus), while two larger groups ofneurons are located symmetrically, lateral to the central canal(lateral nuclei). The neurons in the dorsal nucleus reach the freedorsal surface of the filum terminale, because of the absence ofthe white matter. The neurons of the dorsal nucleus areseparated by the neuron-poor shoulder region bilaterally fromthe lateral nuclei. Proceeding caudally, the number of neuronsdiminished in all three nuclei, it was not rare to find section inthe more caudal filum terminale in which the dorsal nucleuscompletely disappeared. It is a general finding in the lateralnuclei to see perikarya immediately attached to the ependymallining of the central canal.
In a 1 μm thick transverse optical section generated by theconfocalmicroscopeandprepared fromthecranialportionof thefilum terminale 75 to 85 neuronal perikarya could be counted. Inthe more caudal portion of the filum terminale this figure dec-lined to 20 to 25. Approximately 10% of the neuronal perikaryacould be detected in nomore than two adjacent optical sectionsat 4 μm interval. Fifty-three percent of the perikarya could befound inat least three, another 26% in fourandasmanyas11%ofthe perikarya could detected inall five optical sections. Using thetechnique of approximation as outlined in the Experimentalprocedures, 80 counted neuron in an optical section means37,500 neurons in a 5mmportion of the cranial filum terminale.More caudally thenumberofneuronswoulddrop to 15,000 in thesame volume of the filum terminale.
In addition to the approximation of the numbers of theneurons, the 5-section reconstruction indicated that more thanhalf of the neuronal perikarya has to be longitudinally oriented.Indeed the NeuN immunoreaction showed that the majority ofthe nuclei were ovoid and longitudinally oriented in the longi-tudinally cut specimens (Plate II, Fig. 6 and Plate III, Fig. 11).
2.4. Chemical character of the neurons in the filum terminale
Since the graymatter of the filum terminale appeared to be thecaudal extensions of the intermediate zone of the spinal graymatter including the area around the central canal, antibodiesstaining perikarya residing in the intermediate zonewere usedto study the chemical character of the neurons.
In triple-label experimentsNeuNantibody and those againstnitric oxide synthase (NOS) and calretinin were used. NOS-IRneuronal perikaryaweredistributedsingle ornotmore than twoin one Vibratome section. The immunoreaction surrounded theNeuN-labeled nucleus, occasionally a short portion of thedendrites could also be seen (Plate II, Fig. 4). NOS-labeledperikarya were found often adjacent to the ependymal cells ofthe central canal. They gathered in the ventromedial portion ofthe lateral nucleus, no NOS-stained perikaryon could be seen inthe dorsal nucleus. The average diameter of the NOS-IRperikarya in the transverse sections was 7.96 μm (max:11.96 μm, min: 5.20 μm, n=25). In addition to the perikaryal
reaction, fine, varicose NOS-IR axons could be seen also in thegrey matter.
Calretinin immunoreaction labeled various numbers of neu-rons in filum terminale. As an extreme, four calretinin-stainedneurons were found in one specimen. In rare cases a shortportion of one of the dendrites was also seen. Similarly to theNOS positive neurons calretinin-IR neurons were found adja-cent to the ependymal cells (Plate II, Fig. 5). Coarse calretinin-IRnerve fibers were seen in thewhitematter, often symmetricallyin the dorsal portion of the white matter, i.e. in the shoulderregion.Thisbundleof calretinin-IRaxonsseparated theneuronsin the dorsal nucleus from those of the lateral nucleus. Theaverage size of the calretinin-IR perikarya in the transversesections was 7.43 μm (max: 12.87 μm, min: 5,50 μm, n=42).Calretinin-stained neurons formed a broad group in the ventralhalf of the lateral nucleus, while in the dorsal half they aredistributed along the ependyma, they occurred only exception-ally at the ventral border of the dorsal nucleus. Calretinin-IRneurons outnumbered NOS-IR neuron almost by 2:1. Round,choline acetyltransferase (ChAT)-IR neurons were frequentlyseen in the lateral nucleus (Plate II, Fig. 8). Sporadic measure-ments showed that their average diameter was 7.0 to 9.0 μm.Short initial dendritic processes were occasionally seen. Var-icose ChAT-IR, fine axonal arborizations were seen in the graymatter, both in the dorsal and lateral nuclei. Neurokininreceptor-1 (NK-1r)-IR perikarya are among the largest sizeneurons in the filum terminale, their diameter measures from12 to 18μmin transverse section. The immunoreaction typicallyoutlines the surfacemembranes of the perikarya and dendrites(Plate II, Figs. 9 and 10). A rich systemofNK-1r-staineddendritescould be seen around the ependymal cells of the central canal(Plate II, Fig. 9), long sections of the NK-1r-stained dendriteswere coursing in the longitudinally cut sections of the filumterminale (Plate III, Figs. 11 and 12). NK-1r-IR varicose axonscoursed also in the gray matter (Plate III, Fig. 12) and formedbundles in the shoulder region (Plate II, Fig. 9). Small, round toovoid substance P-IR neuronal perikarya were found, whileelaborate dendritic arborizations and fine, richly arborizingvaricose axons could be followed in the longitudinal sections(Plate III, Fig. 12). Only the dot-like cross sections of thesubstance P-IR dendrites and axons could be seen on thetransverse sections of the filum terminale (Plate II, Fig. 10).
Proteinkinase gamma (PCKgamma)-IR neurons, a dom-inating neuron population of the dorsal horn in the spinalcord (Polgár et al., 1999), and also in the conus medullaris(Réthelyi, unpublished), could not be found in the filumterminale. No calbindin D28K-IR neuronal perikarya could befound, either.
2.5. The nerve fiber architecture
The diverse axon components of the filum terminale could besubdivided tentatively into four groups: 1) axons arborizing ex-clusively in the dorsal nucleus, 2) axons arborizing predomi-nantly in the ventral nucleus, 3) axons arborizing in the entiregray matter and 4) axons concentrating in the shoulder region.
1. Axons arborizing exclusively in the dorsal nucleus.Small number of calcitonin gene-related peptide(CGRP)-IR fibers coursing in longitudinal direction was
Plate II
4 and 5. Triple labelingof filumterminale, non-identical opticalsections. Fig. 4. Neuronal perikarya (red) are labeledwith antibody against NeuN, one NOS-IR neuron isseen (green-yellow) with dorsally and ventrally direc-ted dendrites (arrows). Fine, NOS-IR varicosities (green)are seen among the neurons and in the white matter.Three optical sections. - Fig. 5. Neuronal perikarya (red)are labeled with antibody against NeuN, one calreti-nin-IR neuron is seen (green-yellow) with a ventrallydirected dendrite (arrow). Calretinin-stained straightaxons (green) are coursing in the white matter. Theyform dense bundles symmetrically in the dorsolateralportion of the white matter (shoulder region; aster-isks). Four optical sections; DN = dorsal nucleus, LN =lateral nucleus, CC = central canal. Scale: 50 μm.
6. Longitudinal section of the filum terminale, double-label immunohistochemistry, single optical section.Neuronal perikarya (red) are labeled with antibodyagainst NeuN, GFAP-stained astrocytes (green) form adominant longitudinal fiber system. Scale: 50 μm.
108 B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
found in the dorsal nucleus (Plate III, Fig. 13). Noisolectin B4 from Bandeiraea simplicifolia (IB4)-positiveaxons could be found in the filum terminale, althoughboth CGRP- and IB4-labeled primary afferent fibersoccur in the superficial laminae of the spinal cord.
2. Axons arborizing predominantly in the ventral nucleus.As it was described above, NOS-, substance P-, and NK-1r-stained axonal arborizations were detected in thegray matter of the filum terminale, mainly in the lateralnucleus. It seems to indicate that the neurons of thelateral nuclei contribute richly to the axon arborizationsof the parent nuclei. A dense array of extremely finevesicular glutamate transporter 2 (VGLUT2)-IR varicos-ities were found restricted to the gray matter in thelateral nuclei (Plate III, Fig. 15). The dorsal nucleusharbored only a limited number of VGLUT2-IR varicos-ities. No VGLUT2-stained perikaryon could be found,however. Synaptophysin immunoreaction wasused todetect synapsing terminals and varicosities. Fine,synaptophysin-IR varicosities covered densely the lat-eral nuclei, they occurred only scattered in the dorsalnucleus (Plate III, Fig. 16). Fine glycine-transporter 2-IRaxon arborization like structures were seen also in thelateral nucleus (Plate III, Fig. 10, insert).
3. Axons arborizing in the entire gray matter. Cranio-caudally oriented vesicular glutamate transporter 1(VGLUT1)-stained fibers and large, occasionally unrea-listically large varicosities were seen in both dorsal and
lateral nuclei (Plate III, Fig. 14). In addition, ChAT-IRaxons arborize in the entire gray matter (Plate II, Fig. 8and Plate IV, Fig. 18).
7. Cross section of the filum terminale, double-labelimmunohistochemistry, single optical section. GFAPpositive filaments (green) can be seen in both long-itudinal and cross sections. Rings, indicating theperikaryal cytoplasm of the astrocytes, are labeledwith arrows. Neuronal perikarya (red) are labeled withantibodyagainstNeuN.CC=central canal. Scale: 10μm.
8. Cross section of the filum terminale, five opticalsections. ChAT-IR perikarya (arrows) and a dense,varicoseaxonarborizationare seen.CC=central canal.Scale: 50 μm.
9. Cross sectionof the filumterminale, 17optical sections.Large size, NK-1r-IR perikaryon (red) with a shortportion of a ventrally directed dendrite (arrow) areseen at the junction between dorsal and lateral nuclei.NK-1r-IR dendrites, mainly in cross section, surroundthe central canal (CC). Varicose, NK-1r-IR axons coursein the dorsolateral portion of the white matter(shoulder region, between small arrows). Scale: 50 μm.
10. Cross section of the filum terminale, triple-labelimmunohistochemistry, 3 optical sections. Two NK-1r-immunoreactedperikarya (P1andP2) anddendrites(red), as well as substance P-stained axons anddendrites (green) are seen. Some of the substance P-stained puncta are intimately attached to theperikarya and to the dendrites (arrows). Scale: 10μm. – Insert lower right: Double immunolabelingof P1 neuron with antibodies against NK-1r (red)and glycine transporter 2 (green), 3 optical sections.Glycine transporter 2-IR puncta are intimatelyattached to the NK-1r-labeled perikaryon (arrows).Scale: 10 μm.
Plate III
11 and 12. Longitudinal sections of the filum terminale, triple-label immunohistochemistry, Neuronal perikaryalabeled with antibody against NeuN (red), substance
109B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
4. Axons concentrating in the shoulder region. Serotoninimmunoreaction labeled relatively large size vari-cosities in the shoulder region, while serotonin-stainedvaricosities were also seen scattered in the lateralnucleus (Plate IV, Fig. 17). Also enkephalin-IR fibersconverge in the shoulder region (Plate IV, Fig. 18),enkephalin-IR fibers arborized mainly in the lateralnucleus. In addition to the axons mentioned above,calretinin-stained, non-varicose nerve fibers formed adense bundle in the shoulder region (Plate II, Fig. 5),whereas thick, varicose NK-1r-stained axons were alsoseen in the same area of the white matter (Plate II,Fig. 9).
2.6. Colocalizations, close attachments
Synaptophysin and VGLUT2 immunoreaction showed analmost perfect colocalization in the lateral nucleus (Plate III,Fig. 16, insert), while double immunoreaction with antibodiesfor synaptophysin as well as serotonin and ChAT showedpractically no colocalization (Plate IV, Fig. 17). However, someof the serotonin varicosities in the shoulder region showedcolocalization with ChAT (Plate IV, Fig. 19). Double-labeledvaricosities were found regularly in the shoulder region andless frequently in the lateral nuclei when antibodies againstsubstance-P and enkephalin were used (Plate IV, Fig. 20). Subs-tance P-IR varicosities were seen immediately attached to the
perikarya and dendrites of NK-1r-stained neurons (Plate II,Figs. 10 and Plate III, Figs. 12). Also glycine transporter 2-IRpuncta were found closely attached to NK-1r-IR perikarya(Plate II, Fig. 10, insert).
3. Discussion
3.1. Variety of neurons and axon arborizations
It hasbeendemonstrated in thepresentpaper thatdue tovariousgroups of neurons, fiber systems and glial cells, the filum ter-minale is an orderly part of the nervous system in the rat. Like inthe spinal cord, thewhite and graymatters are divided, and bothterritories have specific features. The gray matter consists ofthree groups of neurons. A midline nucleus is located dorsal tothe central canal (dorsal nucleus) and two concentrations of
P-IR perikarya, dendrites and varicose axons (green;the perikarya are labaled with arrows) as well as NK-1r-IR dendrites and varicose axons (blue; the axon islabeled with short arrows) are seen. Fourteen opticalsections. Scale: 50 μm – Fig. 12. Substance P-IRperikarya, dendrites and varicose axons (green; theperikarya are labeled with arrows) as well as NK-1r-IRdendrites and varicose axons (blue; the axon is labeledwith small arrows) are seen. Fine substanceP-IR axonsrun parallel with and surround the thick, NK-1r-stained dendrites. Higher power detail of Fig. 11;fourteen optical sections. Scale: 50 μm.
13 and 14. Cross section of the filum terminale, double-labelimmunohistochemistry. Fig. 13. CGRP-IR fibers (red)course in the dorsal nucleus. CGRP immunoreactionoutside the filum terminale (arrows) labels fibers inthe dorsal roots adjacent to and running parallel withthe filum terminale. CC = central canal. - Fig. 14. Inaddition to CGRP-IR fibers (red) a dense plexus ofVGLUT1-IR fibers (green) course both in the dorsal andventral nuclei. VGLUT1 immunoreaction outside thefilum terminale (arrows) labels fibers in the dorsal rootadjacent to and running parallel with the filumterminale. Eight optical sections in both figures.Scale: 50 μm.
15 and 16. Double labeling of filum terminale, single opticalsection. Fine, varicose axon terminals labeled withantibody against VGLUT2 (Fig. 15; red) and antibodyagainst synaptophysin (Fig. 16; green) densely fill up thelateral nuclei (LN). CC = central canal. Scale: 50 μm. -Insert in Fig. 16: Higher power detail showing colocali-zation of VGLUT2 and synaptophysin immunoreaction(yellow) in themajority of the varicosities. Scale: 10 μm.
Plate IV
17. Double labeling of filum terminale, single opticalsection. Serotonin-labeled nerve fibers and varicos-ities (green) are concentrating in the shoulderregions (asterisk). Fine synaptophysin-IR varicosities(red) are densely distributed in the lateral nuclei (LN)and scattered in the dorsal nucleus (DN). No coloca-lization was found. CC = central canal. Scale: 50 μm.
18. Double labeling of filum terminale, single opticalsection. Enkephalin-IR fibers (red) course symmetri-cally in the dorsal part of the filum, in the shoulderregion (asterisk).ChAT-IRnerve fibers (green) course inthe entire cross section of the gray matter. Nocolocalization was found. CC = central canal. ChATimmunoreaction outside the filum terminale (arrows)labels axons in the ventral root adjacent to andrunning parallel with the filum terminale. Scale: 50μm.
19. Cross section of the filum terminale, double-labelimmunohistochemistry. (The same preparation asFig. 17). Serotonin-IR puncta (red) form twodense fieldsin the dorsolateral portion of the white matter(shoulder region). ChAT-IR neuronal processes (green)are found in the dorsal and lateral nuclei. Some of theserotonin-labeled larger varicosities are double labeledwithChAT immunoreaction (yellow; arrows). Scale=10μm.
20. Double-label immunohistochemistry of the filumterminale, single optical section. Dense distributionof both enkephalin- (red) and SP-IR (green) nervefibers and varicosities are seen in the dorsolateralpart of the filum terminale (shoulder region; aster-isk). Varicosities occur also in the lateral nucleus.Numerous varicosities in the shoulder region and inthe lateral nucleus show colocalization of the twopeptides (yellow). CC = central canal. Scale: 10 μm.
110 B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
neurons form the lateral nuclei at both sides of the central canal.The crescent-shaped white matter – as it is shown with the RIPantibody detecting oligodendrocytes – is interrupted dorsally.Thedorsal portionof thewhitematter ismissing,mainlybecauseof the lack of dorsal root afferents and dorsal column fibers.
The dorsal peaks of thewhitematter –named as the shoulderregion – seems to be a concentration of axons of various chemicalcharacters. Some of the axons coursing in the shoulder region,like those labeled with serotonin and enkephalin are mostprobably descending fibers. We presume, however, that thecalretinin-andNK-1r-immunolabeledaxons in thesamepositionrepresent ascending tract fibers, similarly to the ascendingprojection of the NK-1r neurons in the dorsal horn of the spinalcord. (Todd et al., 2000). Further tract tracing studies will beneeded to find out that the NK-1r and calretinin neuronsrepresent the ascending connections of the filum terminale.Parallel to the concentration of axons, the shoulder region isrich inRIP immunoreaction (Friedmanetal., 1989), indicating thata significant portion of the axons are myelinated nerve fibers.
The number of neurons in the gray matter of the filum ter-minale was estimated by counting the NeuN-immunostainednuclei of the neurons. There are approximately 37,500 neurons inthemostproximal5mmlongportionof the filumterminale.Morecaudally the number of neurons drops to 15,000 neurons/5 mmfilum terminale. These figures, whatever crude estimates theyare, underscore thevalidityof theefforts to study thepositionandrank of theses neurons in the spinal cord neuronal networks, andultimately to reveal their functional significance. The dorsal nu-cleus of the gray matter is reduced in size caudally. No neuronalperikarya could be labeledwith theantibodiesused in thepresentstudy in the dorsal nucleus. VGLUT1-, ChAT-, low density ofVGLUT2- and synaptophysin-IR varicosities were seen in thedorsal nucleus, while CGRP-IR axons were found exclusively inthe dorsal nucleus. Since CGRP-labeled axons of peripheral originare obligate components of the dorsal horn in the spinal cord, wesuggest that the laminae of the spinal dorsal horns reduce in sizeinto a small, midline group of neurons in the filum terminale.
Both CGRP- and IB4-IR fibers were richly arborizing in thedorsal horn in the most caudal portion of the spinal cord (conusmedullaris; Réthelyi, unbuplished). It was surprising, however,that no IB4 immunoreaction could be seen at all in the filumterminale. It is equally striking that one of the most numeroustypes of dorsal horn neurons (PKC-gamma immunostained neu-rons in laminae II and III; Polgáret al., 1999)hasnocontinuation inthe filum terminale.
A large variety of chemically different neurons wereencountered in the lateral nuclei. NOS-, calretinin-, ChAT-,SP- and NK-1r-IR neurons can be found also in the interme-diate zone of the spinal cord.
NOS-stained neurons are widely distributed not only in thedorsal horn and in the spinal autonomic nuclei, but also aroundthe central canal (Dun et al., 1993; Reuss and Reuss, 2001). Cal-retinin-stainedneuronspredominate in laminaeVandVI and inthe sacral dorsal gray commissure (Ren and Ruda, 1994). Cal-retinin labelingwas detected in small size neurons in laminae I,II, III and VIII, in medium to large size neurons in laminae I andIII–VII, and in small to medium size neurons in lamina X in thecat (Anelli andHeckman, 2005). ChAT-IRspinalneuronsoccur invariousportionsof thegreymatter, they concentrate around thecentral canal (Houser et al., 1983; Barber et al., 1984;). NK-1r-IR
Table 1 – Antibodies, dilutions and sources
Antibody Species Dilution Source
Calbindin D28K Mouse 1:3000 Sigma-Aldrich, St. Luise, Mo. USACalcitonin gene-related peptide (CGRP) Goat 1:1000 Santa Cruz Biotechnology,Heidelberg, GermanyCalretinin Rabbit 1:2000 Millipore (Chemicon), Budapest, HungaryCholin acetyltransferase (ChAT) Goat 1:100 Millipore (Chemicon), Budapest, HungaryEnkephalin Mouse 1:1000 Millipore (Chemicon), Budapest, HungaryGlial fibrillary acidic protein (GFAP) Rabbit 1:5000 Sigma-Aldrich, St. Luise, Mo. USAGlycin transporter 2 Sheep 1:2000 Millipore (Chemicon), Budapest, HungaryNeurokinin A (NK-1r) Rabbit 1:2000 Santa Cruz Biotechnology, Heidelberg, GermanyNeuron-specific nuclear protein (NeuN) Mouse 1:1000 Millipore (Chemicon), Budapest, HungaryNitric oxide synthase (NOS) Goat 1:2000 Gift from Dr. P.C. Emson, Herbison et al. (1966)Oligodendrocytes (NS-1 RIP) Mouse 1:30,000 Millipore (Chemicon), Budapest, HungaryProtein kinase (PKC gamma) Rabbit 1:1000 Santa Cruz Biotechnology, Heidelberg, GermanySerotonin Rabbit 1: 5000 DiaSorin, Stillwater, MN, USASubstance P Rat 1:200 AbD Serotec (Oxford Biotechnology Ltd.)Synaptophysin Mouse 1:20,000 Novocastra (Biomarker Ltd), Budapest, HungaryVesicular glutamate transporter 1 (VGLUT1) Guinea pig 1:1000 Millipore (Chemicon), Budapest, HungaryVesicular glutamate transporter 2 (VGLUT2) Guinea pig 1:10,000 Millipore (Chemicon), Budapest, Hungary
111B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
neurons are widely distributed in the spinal grey matter (Little-wood et al., 1995); and a group of small NK-1r-IR neurons sur-round the central canal (Brown et al., 1995; Benoliel et al., 2000).
Scattered substance P-IR neurons were found in the spinalgray matter in colchicine treated rats (Ljungdhal et al., 1978),including the area around the central canal (Nahin, 1987). Thedorsal central gray area is especially rich in substance P con-taining neurons (Sasek et al., 1984).
ChAT- and NOS-stained cells are considered as inhibitoryinterneurons in the dorsal horn (Laing et al., 1994), while calre-tinin-stained neurons are also partly inhibitory (Résobois andRogers, 1992). Gylcine containingneuronsare the inhibitory inter-neurons both in the dorsal and ventral horn of the spinal graymatter. Dense accumulation of glycine transporter 2-IR neuronsoccur in the intermediate zone (Hossaini et al., 2007). The inhi-bitorycomponentsof the filumterminaleseemstobebalancedbytheexcitatoryNK-1r-IRneurons (Littlewoodetal., 1995) andby therich array of VGLUT2-IR varicosities representing the axons ofexcitatory neurons (Todd et al., 2003). As it wasmentioned abovewe presume that the ascending tract neurons of the filum ter-minale reside among the NK-1r- and calretinin-IR neurons.
The varicosities of at least seven various axon systems couldbe detected in the ventral nuclei of the gray matter. NOS-, andChAT-stained varicosities seem to belong to the axon arboriza-
tions of the local neurons. The VGLUT1-IR axons in the filumterminale my represent the long descending branches of thedorsal root fibers. Substance P-, VGLUT2- and glycine transporter2-IR axons seem to represent local interneurons, while enkepha-lin- and serotonin-IR axons belong most probably to descendingtractneurons.AxonscontainingbothsubstancePandenkephalinmay represent axons of the substantia gelatinosa neurons in theconus medullaris (Ribeiro-da-Silva et al., 1991). The chemicalcharacter of both the neurons and the terminal axon arboriza-tions suggests that the intermediate zoneof the conusmedullarisand the gray matter of the filum terminale posses a thoroughlyinterconnected shared neuron network.
3.2. Rich systemofGFAP-IRastrocytes in the filum terminale
Unique featureof the filumterminale is the richnetworkofGFAP-IR astrocyte processes whose orientation is both radial andlongitudinal. Small, roundareas surrounded byGFAP-IR rings canbe interpreted as the perikarya of the astrocytes, their processeswith radial orientation often reach the surface of the filumterminale. In longitudinally cut filum terminale the astrocyteprocesses course long distances producing a three-dimensionallattice. GAFP-stained astrocytes were described in the rat spinalcord (Hajós andKálmán, 1989),mostly at the periphery and along
Figures
Figs. 2, 13 and 16nti-rabbit-PKCgamma Fig. 20use-NeuN Figs. 10 and 11
Figs. 8 and 9Figs. 6 and 7Fig. 5Fig. 12Figs. 17 and 19Fig. 18Figs. 3 and 4Figs. 14 and 15
112 B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
bloodvessels.Comparedwith the thoracicand lumbarspinal cordsegments (Réthelyi, unpublished observations), the filum termi-nale is exceptionally rich in GFAP-stained astrocyte processes.
4. Conclusion
Even though the filum terminale has been described as a com-plex system of neurons and glial cells of various chemical cha-racters, and as an elongation of the spinal cord, its functionalsignificance is still enigmatic.We concluded in our earlier paper(Réthelyi et al., 2004) that the filumterminaleconsists ofneuronsin an early phase of commitment and differentiation. This in-terpretation is hardly tenable, because of the rich variety ofchemically labeled neurons, as it has been shown in the presentstudy. The classical neuronal networks of the spinal cord, likeneurons in the dorsal horn that receive and forward the sensoryimpulses to higher centers and the premotor neurons in theintermediate zone that forward the descending impulses to themotoneurons arepointless in the filumterminale because of thelack of peripheral connections. On the same token, one cannotthink about neurons as elements of segmental reflex arches.What remains after removing the abovementioned parts of thegray matter is the region of the central canal. We think that thekey to the functional significance of the filum terminale is hid-den in the spinal gray matter surrounding the central canal allalong its length down to the filum terminale.
The gray matter of the spinal cord around the central canalwas described as lamina X by Rexed (1952). The richness invarious peptidergic fiber components of laminaXwas shown byGibson et al. (1981). They hypothesized that the neurons oflamina X represent a “continuous structure running the entirelength of the brain stem and spinal cord in close approximationto the lining of the neural tube”. Apparently, the pericanal net-work does not stop at the coccygeal segments, as Gibson et al.(1981) concluded but continues in the filum terminals.
Further speculations and efforts are rather discouraged, how-ever, by the findings thatneurons seemtooccuronly exceptionallyin the human filum terminale (Cumings and George, 2003; Fonteset al., 2006). In spite of this tract tracing and ultrastructural studiesare going on in our laboratory to learn more about the neuronalstructure and spinal cord connections of the filum terminale.
5. Experimental procedures
All animal experiments were performed in accordance withEuropean Communities Council Directive of 24 November 1986(86/609/ECC) and were approved by the Committee on AnimalExperiments, Semmelweis University, Budapest. All animalswere housed in a room with a 12 h light/dark cycle, and theyhad access to food and water ad libitum. Ten adult SpragueDawley rats of both sexes (260–290 g), were included in thestudy. The animals were deeply anaesthetized with a mixtureof ketamine (Calypsol) and xylazyne (Primazin) then perfusedthrough the left ventricle with a brief rinse of saline followed bya freshly prepared 4% paraformaldehyde solution in 0.1 Mphosphate buffer (PB). The lumbosacral and coccygeal seg-ments of the spinal cord and the filum terminalewere dissectedand postfixed overnight in the same fixative at 4 °C.
One centimeter long proximal portion of the filum termi-nale was embedded in 10% agar and cut into 50 to 60 μm thicktransverse sections with a Vibratome. The sections were keptin groups of five. The filum terminale of two rats were cutlongitudinally. Selected transverse and longitudinal sectionsfrom the filum terminale were processed for double, triple andquadruple immunolabeling. The specimenswere immersed in50% ethanol for 30min to enhance antibody penetration. Next,the sections were treated for an hour in phosphate buffersaline (PBS) containing 2% normal horse serum (NHS) and 0.3%Triton X. Sections were incubated in primary antibodies or incocktails of primary antibodies for 48 h at 4 °C. All antibodiesused in this study were diluted in PBS with 0.1% Triton X.Sections were rinsed in PBS and incubated in solutionscontaining the appropriate species-specific secondary anti-bodies (all raised in donkey and diluted 1:500) coupled tofluorophores (Alexa, Jackson Biotecnology) for 24 h.
Sections were mounted in an antifade medium (Vectashield,Vector Laboratories) and examined with a Bio-Rad Radience 2100Rainbow confocal laser scanning system equipped with a NiconEclipse E800 microscope. Three groups of transverse sections 2 to3 mm apart from each animal and all longitudinal sections werestudied. The confocal imageswere analyzed by Confocal Assistantprogram.
Biotinylated isolectin IB4 from Banderia simplicifolia (Sigma)was also used andwas detectedwith a fluorophor conjugated toStreptavidin (Streptavidin-Alexa 488). Control experimentsincluded the omission and stepwise dilution of the primaryantibodies.
To examine the neurochemical features of the filumterminale in the rat spinal cord several markers were applied.The primary antibodies and the cocktails of antibodies usedfor the identification of neurons and glial cells are listed inTables 1 and 2.
Sections from two animals stained with triple-immunoreac-tion for NeuN, NOS and calretinin were used for quantitativestudies.Twodiametersof theNOSandcalretinin-IRneurons– thelongest and the longest perpendicular to the former diameter –weremeasured in transverse sectionsat 40 to60×magnification.The size of the neuronswas approximated as the average of thetwo measurements.
The same sectionswereused to obtain figures for the absolutenumberofneurons in theproximalportionof the filumterminale.The Vibratome sections were scanned by making single opticalsections at 4 μm interval at 40–60× magnification with theconfocal microscope. The position of the NeuN-IR profiles werefollowed through five consecutive sections starting from themiddle one (reference section). The neurons were grouped basedon their incidence in the sections. A small number of neuronscould be discovered in all 5 optical sections. Approximately thesame number of neurons could be found in 4 and 3 opticalsections, respectively, againa small numberofneurons turneduponly in twoadjacentoptical sections. Theapproximatenumberofneurons was calculated by dividing the number of neurons ineach group by the number of sections through which the neuroncould be recovered. Example: 75 NeuN immunreactive profileswere counted in the reference middle section. 10, 26, 34 and 5neurons were in the four interval groups (5, 4, 3 and 2 sectionincidence, respectively). Thus the approximate (reduced) numberof neurons: 10/5+26/4+34/3+5/2=22 which could be interpreted
113B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
as the number of neurons in a 4 μm thick slab of the filumterminale.
Acknowledgments
We thank Dr. Zita Puskár and Dr. Márk Kozsurek for theparticipation in the initial phase of this study and for advice.
R E F E R E N C E S
Anelli, R., Heckman, C.J., 2005. The calcium binding proteinscalbindin, parvalbumin, and calretinin have specific patterns ofexpression in the gray matter of cat spinal cord. J. Neurocytol.34, 369–385.
Barber, R.P., Phelps, P.E., Houser, C.R., Crawford, D., 1984.The morphology and distribution of neurons contrainingcholine acetyltransferase in the adult rat spinal cord:an immunocytochemical study. J. Comp. Neurol. 229,329–346.
Barr, M.L., Kiernan, J.A., 1993. The human nervous system.Ananatomical viewpoint, 6th ed. J.B. Lippincott, Philadelphia, p. 70.
Benoliel, R., Tanaka, M., Caudle, R.M., Ladarola, M.J., 2000.Co-localization of N-methyl-daspartate receptors andsubstance P (neurokinijn-1) receptors in rat spinal cord.Neurosci. Lett. 291, 61–64.
Boros, C.S., Lukácsi, E.H., Oszwald, E., Réthelyi, M., 2005.Immunohistochemical characterization of the neurons in thefilum terminale. Clin. Neurosci. 58, 19.
Brown, J.L., Liu, H., Maggio, J.E., Vigna, S.R., Mantyh, P.W.,Basbaum, A.I., 1995. Morphological characterization ofsubstance P receptor-immunoreactive neurons in the ratspinal cord and trigeminal nucleus caudalis. J. Comp. Neurol.356, 327–344.
Chesler, M., Nicholson, C., 1985. Organization of the filumterminale in the frog. J. Comp. Neurol. 239, 431–444.
Choi, G.H., Kim, R.C., Suzuki, M., Choe, W., 1992. The ventriculusterminalis and filum terminale of the human spinal cord. Hum.Pathol. 23, 916–920.
Chvátal, A., Anderová, M., Ziak, D., Orkand, R.K., Syková, E., 2001.Membrane currents and morphological properties of neuronsand glial cells in the spinal cord and filum terminale of the frog.Neurosci. Res. 40, 23–35.
Cummings, T.J., George, T.M., 2003. The immunohistochemicalprofile of the normal conus medullaris and filum terminale.Neuroembryology 2, 43–49.
Fontes, R.B.V., Saad, F., Soares, M.S., De Olivera, F., Pinto, F.C.G.,Liberti, E.A., 2006. Ultrastructural study of the filum terminaleand its elastic fibers. Neurosurgery 58, 978–983.
Friedman, B., Hockfield, S., Black, J.A.,Woodruff, K.A.,Waxman, S.G.,1989. In situ demonstration of mature oligodendrocytes andtheir processes: an immunocytochemical study with a newmonoclonal antibody, rip. Glia 2, 380–390.
Gamble, H.J., 1971. Electron microscope observations upon theconus medullaris and filum terminale of human fetuses.J. Anat. 110, 173–179.
Gibson, S.J., Polak, J.M., Bloom, S.R.,Wall, P.D., 1981. The distributionof nine peptides in rat spinal cord with special emphasis on thesubstantia gelatinosa and on the area around the central canal(lamina X). J. Comp. Neurol. 201, 65–79.
González-Robles, A., Glusman, S., 1979. The filum terminale ofthe frog spinal cord. Cell Tissue Res. 199, 519–528.
Hajós, F., Kálmán, M., 1989. Distribution of glial fibrillary acidiprotein (GFAP)-immunoreactive astrocytes in the rat brain. II.Mesencephalon, rhombencephalon and spinal cord. Exp. BrainRes. 78, 164–173.
Hossaini, M., French, P.J., Holstege, J.C., 2007. Distribution ofglycinergic neuronal somata in the spinal cord. Brain Res. 1142,61–69.
Houser, C.R., Crawford, G.D., Barber, R.P., Salvaterra, P.M., Vaughn, J.E.,1983. Organization and morphological characteristics ofcholingergic neurons: an immunocytochemical study with amonoclonal antibody to choline acetyltransferase. Brain Res. 266,97–119.
Laing, I., Todd, A.J., Heizmann, C.W., Schmidt, H.H.H.W., 1994.Subpopulations of GABAergic neurons in laminae I–III of ratspinal dorsal horn defined by coexistence with classicaltransmitters, peptides, nitric oxide synthase or parvalbumin.Neuroscience 61, 123–132.
Littlewood, N.K., Todd, A.J., Spike, R.C., Watt, C., Shehab, S.A.S.,1995. The types of neuron in spinal dorsal horn which possessneurokinin-1 receptors. Neuroscience, 66, 597–608.
Ljungdahl, A., Hökfelt, T., Nilsson, G., 1978. Distribution ofsubstance P-like immunoreactivity in the central nervoussystem of the rat I. Cell bodies and nerve terminals.Neuroscience 3, 861–943.
Miller, C., 1968. The ultrastructure of the conus medullaris andfilum terminale. J. Comp. Neurol. 132, 547–566.
Nahin, R.L., 1987. Immunocytochemical identification of longascending peptidergic neurons contributing to thespinoreticular tract in the rat. Neuroscience 23, 859–869.
Pinto, F.C.G., deVasconcellos Fontes, R.F., deCamargo Leonhardt,M., Amodio, D.T., Porro, F.F., Machado, J., 2002. Anatomic studyof the filum terminale and its correlations with the tetheredcord syndrome. Neurosurgery 51, 725–730.
Polgár, E., Fowler, J.H., McGill, M.M., Todd, A.J., 1999. The types ofneuron which contain protein kinase C gamma in rat spinalcord. Brain Res. 833, 71–80.
Rascher, K., Booz, K.H., Donauer, G., Donauer, E., 1988. The filumterminal. A morphological study in the cat. Z. mikrosk.-anatForsch 102, 1–17.
Ren, K., Ruda, M.A., 1994. A comparative study of thecalcium-binding proteins calbindin-D28K, calretinin,calmodulin and parvalbumin int he rat spinal cord. Brain Res.Rev. 19, 163–179.
Résobois, A., Rogers, J.H., 1992. Calretinin in rat brain: animmunohistochemical study. Neuroscience 46, 101–134.
Réthelyi, M., Lukácsi, E., Boros, C.S., 2004. The caudal end of the ratspinal cord: transformation to and ultrastructure of the filumterminale. Brain Res. 1028, 133–139.
Reuss, M.H., Reuss, S., 2001. Nitric oxide synthase neurons in therodent spinal cord: distribution, relation to Substance P fibers,and effects of dorsal rhizotomy. J. Chemical Neuroanat. 21,181–196.
Rexed, B., 1952. The cytoarchitectonic organization of the spinalcord of the cat. J. Comp. Neurol. 96, 415–495.
Ribeiro-da-Silva, A., Pioro, E.P., Cuello, A.C., 1991. SubstabónceP- and enkephalin-like immunoreactivities are colocalized incertain neurons of the substantia gelatinosa of the rat spinalcord An ultrastructural double-labeling study. J. Neurosci. 11,1068–1080.
Sasek, C.A., Seybold, V.S., Elde, R.P., 1984. Theimmunohistochemical localization of nine peptidesin the sacral parasympathetic nucleus and the dorsalgray commissure in rat spinal cord. Neuroscience, 12,855–873.
Selcuki, M., Vatanserer, S., Inan, S., Erdemli, E., Bagdotoglu, C.,Polat, A., 2003. Is a filum terminale with a normal appearancereally normal? Childs Nerv. Syst. 19, 3–10.
Streeter, G.L., 1919. Factors involved in the formation of the filumterminale. Am. J. Anat., 25, 1–11.
114 B R A I N R E S E A R C H 1 2 0 9 ( 2 0 0 8 ) 1 0 5 – 1 1 4
Todd, A.J., McGill, M.M., Shehab, S.A., 2000. Neurokinin 1 receptorexpression by neurons in laminae I, III and IV of the rat spinaldorsal horn that project to the brainstem. Eur. J. Neurosci. 12,689–700.
Todd, A.J., Hughes, D., Polgár, E., Nagy, G.G., Mackie, M.,Ottersen, O.P., Maxwell, D.J., 2003. The expression ofvesicular glutamate transporters VGLUT1 and VGLUT2 in
neurochemically defined axonal populations in the ratspinal cord with emphasis on the dorsal horn. Eur. J.Neurosci. 17, 13–27.
Yamada, S., Won, D., Kido, D.K., 2001. Adult tethered cordsyndrome: new classification correlated with symptomatol-ogy, imaging and pathophysiology. Neurosurg. Quart. 11,260–275.
