Journal ofh'eurogenetics, 2 (1985) 1-30 Elsevier 1 JNG 00030 - Research Papers Drosophila Mushroom Body Mutants are Deficient in Olfactory Learning Martin Heisenberg Alexander Borst*, Sibylle Wagner** and Duncan B yer s *** Insiiiut fur Geneiik und Mikrobiologie, Wurzburg (F.R. G.) (Received August 14th, 1984) (Revised October 30th, 1984) (Accepted October 30th. 1984) Key words: brain mutants - sexual dimorphisms - avoidance conditioning - reward learning SUMMARY Two Drosophilu mutants are described in which the connections between the input to and the output from the mushroom bodies is largely interrupted. In all forms of the flies (larva, imago, male, female) showing the structural defect, olfactory conditioning is impaired. Learning is completely abolished when electroshock is used as reinforcement and partially suppressed in reward learning with sucrose. No influence of the mushroom body defect on the perception of the conditioning stimuli or on spontaneous olfactory behavior is observed. The defect seems not to impair learning of color discrimination tasks or operant learning involving visual cues. INTRODUCTION During the first half of this century functional maps have been established for the vertebrate brain at a coarse anatomical level. Such information, however, is still scarce for insects. Even the function of the mushroom bodies (corpora pedunculata") which may be considered the simplest part of the insect central brain, is not at all clear. The mushroom bodies (m.b.s) are two large bundles of very thin, parallel fibers * Present address: Max Planck Institut fiir biologische Kybernetik, Spemannstr. 38, 7400 Tiibingen, ** Present address: Institut fur Virologie und Immunbiologie, Versbacherstr. 7, 8700 Wiirzburg, F.R.G. *** Present address: Dept. of Psychology, State University ofNew York, Binghampton, NY 13901, U.S.A. Correspondence: M. Heisenberg, Institut fur Genetik und Mikrobiologie, RBntgenring 11, 8700 Wtirzburg, F.R.G. F.R.G. 0167-7063/85/$03.30 Q 1985 Elsevier Science Publishers B.V. (Biomedical Division) J Neurogenet Downloaded from informahealthcare.com by University of Alberta on 09/02/13 For personal use only.
Transcript
Journal ofh'eurogenetics, 2 (1985) 1-30 Elsevier
1
JNG 00030
- Research Papers
Drosophila Mushroom Body Mutants are Deficient in Olfactory Learning
Martin Heisenberg Alexander Borst*, Sibylle Wagner** and Duncan B yer s ***
Insiiiut fur Geneiik und Mikrobiologie, Wurzburg (F.R. G.)
(Received August 14th, 1984) (Revised October 30th, 1984)
Two Drosophilu mutants are described in which the connections between the input to and the output from the mushroom bodies is largely interrupted. In all forms of the flies (larva, imago, male, female) showing the structural defect, olfactory conditioning is impaired. Learning is completely abolished when electroshock is used as reinforcement and partially suppressed in reward learning with sucrose. No influence of the mushroom body defect on the perception of the conditioning stimuli or on spontaneous olfactory behavior is observed. The defect seems not to impair learning of color discrimination tasks or operant learning involving visual cues.
INTRODUCTION
During the first half of this century functional maps have been established for the vertebrate brain at a coarse anatomical level. Such information, however, is still scarce for insects. Even the function of the mushroom bodies (corpora pedunculata") which may be considered the simplest part of the insect central brain, is not at all clear.
The mushroom bodies (m.b.s) are two large bundles of very thin, parallel fibers
** Present address: Institut fur Virologie und Immunbiologie, Versbacherstr. 7, 8700 Wiirzburg, F.R.G. *** Present address: Dept. of Psychology, State University ofNew York, Binghampton, NY 13901, U.S.A. Correspondence: M. Heisenberg, Institut fur Genetik und Mikrobiologie, RBntgenring 11, 8700 Wtirzburg, F.R.G.
(Kenyon cells, intrinsic neurons3’; 340,000 in the bee36, 2500 in Drosophi l~~’) . They extend as the so-called pedunculi on either side of the central complex from the dorso-caudal cell body layer through the central neuropil to the ventro-rostra1 margin of the protocerebrum (Fig. 2). There the fibers divide, one branch growing straight into the dorso-frontal brain forming the a-lobe, the other extending medially to the sagittal midplane as the /?-lobe. The 3 arms of the mushroom bodies (peduncle, a-lobe, p-lobe) are oriented roughly at right angles to each other. Before entering the central neuropil the Kenyon cell fibers extend a few sparsely branching dendrites that contact “extrinsic” fibers in the so-called calyx. The latter fibers which to a large part come from the antennal ganglion (relay neurons14) constitute the main input to the m.b.s. Most relay neurons are connected to just one glomerulus in the antennal but to many Kenyon fibers in the ~ a l y x ~ ~ , ~ ~ . The output from the m.b.s. consists of extrinsic neurons in the lobes and the peduncle; there fibers form contacts with many Kenyon cells at distinct positions along the track and link the m.b.s. to most other parts of the brain43.44. The branching patterns of some of the output elements may form sharp layers oriented perpendicularly to the Kenyon fibers. This arrangement is reminiscent of that observed for Purkinje cells of the vertebrate c e r e b e l l ~ m ~ ~ . In contrast to Purkinje cells, most m.b. output neurons (1) differ from each other in their branching patterns, (2) are not arranged in a regular array and (3) project diffusely into most parts of the brain.
The function of the m.b.s has been investigated in several insect species using various experimental techniques and looking at different aspects of behavior. A clear picture has not yet emerged. Menzel et aL35,13, tested olfactory conditioning of the proboscis reflex in bees. Local cooling of parts of the m.b.s partially blocked short term memory as did cooling of the whole animal. After cooling was turned off learning capacity recovered. Cooling the lobulas did not interfer with the consolidation process. Puncturing the m.b.s. with a needle had the same suppressive effect as ~ool ing’~ . Elimination of mushroom body function by surgical lesions in the silkworm Cecropia caused the larvae to spin a two-dimensional sheath instead of a cocoon31. Ablation experiments with adult male crickets led to elevated general motor activity and to a complete suppression of courtship singing27. Howse26 studying mushroomectomized locusts and bees, con- firmed the increased activity and noticed in addition simultaneous display of rivalling motor patterns. Unfortunately, learning and memory were not tested in any of these animals. In electrical stimulation experiments on freely moving crickets and grass- hoppers, Huber”, Otto39 and Wadepuhl and H ~ b e r ~ ~ elicited courtship behavior by the application of current pulses in brain areas close to the m.b.s. The significance of the m.b.s. for the generation of male courtship behavior has also been suggested by a very different experiment in Drosophilu: using gynandromorphs and a histochemical stain for male vs female neural cell bodies, Hall” showed that the m.b.s. (or tissue in the vicinity of the m.b.s) had to be male for flies to display male courtship behavior. All these results suggest the m.b. system to play a role in the generation or coordination of certain motor patterns. This speculation gains support from the m.b. fine structure which, as described above, has some characteristic features in common with the c e r e b e l l ~ m ~ ~ .
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
Sensory processing at the level of the m.b.s. has been studied using electrophysio- logical recording techniques. As expected from comparative anatomical considera- tions" the m.b.s. are closely associated with the chemosensory system. Field potentials recorded in the m . b . ~ ~ ~ and recordings from extrinsic neurons'2,25,44*48 , as well as from groups of Kenyon cells (J. Erber, unpublished) showed that the m.b.s. are sensitive to olfactory stimuli. Additional sensory modalities feed into the m.b.s, such as mechano- sensation43 and vision36. Many neurons respond to more than one r n ~ d a l i t y ' ~ * * ~ * ~ ~ . A characteristic feature of Kenyon cells and output neurons of the m.b.s. are long after- effects of sensory stimuli; these may last seconds or even minute^'^,^^. In general, such aftereffects are not detected in the relay neuronsz5. With a few notable exceptions12, it is not yet apparent from the properties of m.b. neurons how they might participate in short-term memory or the generation of motor patterns.
In recent years, Drosophila mutants with structural defects in the m.b .~* ' .~ ' have been isolated. We report here behavioral and anatomical data resulting from the analysis of two of these mutants; the experiments were aimed at clarifying the role of the m.b.s in behavior. In comparison to ablation and cooling experiments in large insects genetically induced brain lesions offer several advantages. First, experimental animals need not to be injured. Second, a nearly unlimited number of animals with the same primary defect can be provided. Third, genetic defects can be highly specific for certain cell types. In the visual system, genetically induced brain lesions have already proven to be useful in establishing structure-function relationships (for reviews see refs. 16 and 24). It will be shown in this paper that central brain lesions can be interpreted behaviorally as well.
MATERIALS AND METHODS
(a) Flies
As wild-type stocks of Drosophila Canton-S, Berlin and Kapelle22 were used. The mutant mushroom bodies derangedKs65 (mbd) was derived from a cross of ethyl methane sulfonate (EMS)-treated WT Berlin males and attached-)< females of unknown origin32. The gene is located on the X chromosome at approximately map position 57 (to the right of the marker forked); mbd is uncovered by the deletion Df(1) N19 ( 1 l A ; 18A2) of G. Lefevre. The mutation is maintained over an attached X% chromosome (In(l)EN, In(l)dl-49, y v f car YjO kindly provided by M.M. Green. The mutant mushroom body miniatureN337 (mbm) was discovered in a collection of 1400 stocks of unknown genetic origin, carrying EMS-treated 2nd chromosomes (courtesy of J. NUsslein-Volhard) marked with the genes cn, bw, and sp and balanced with the multiply inverted chromo- some CyO. The mutant gene is located at the left end of the 2nd chromosome close to the marker aristaless; since mbm is uncovered by Df(2L)al but not by Df(2L)S2, its position is in the region 21B8; 21C6-Dl. For behavioral experiments it was separated from the marker genes by a cross with wild-type Berlin; mbm is kept as a homozygous
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
4
stock. All other X and 2nd chromosome mutants mentioned in Table I1 were obtained from the same mutant screens that gave rise to mbd and mbm with the exception of the mutant minibrain, which was derived from a non-phototactic, multiply mutant stock provided to us by J.A. Memam. The mutants of Table I1 have the following brain phenotypes: In mushroom body miniature BN806 (mbmB) the calyces have only about 30% the normal volume. Pedunculi and lobes are accordingly small. The structural defect of the mutant mushroom body miniature CNz8 (mbmC) is similar to that of mbm (see below) except that in mbmC also the antennal lobes are reduced in size. In mbmB and mbmC both genders are affected alike. In the mutant cabx bulgingN7' calyces are enlarged and have an abnormal shape; pedunculi and lobes are small or missing; the ellipsoid body is also affected. Expressivity is much more stable and uniform than in mbm and mbd. The mutants minibrain (mnb), optomotor-blindH3', small optic lobesKS5* and l o b ~ l a p l a t e - l e s s ~ ~ ~ ~ have been described e l s e ~ h e r e ' ~ , ~ ~ . In all of them including mnb the m.b.s have about normal size and shape. In the mutant central body defectKS96 the whole central complex with the exception of the protocerebral bridge is dissociated into two fiber masses of variable shape. The /? y-lobes of the m.b.s are reduced. In contrast, the salient feature of the mutant no bridgeKS49 is its disorganized protocerebral bridge, whereas the central body and the ellipsoid body have retained their normal shape. In the mutants ellipsoid body and central complex primarily the ellipsoid body, in central complex both the ellipsoid body and the central body are flat and broadened.
(b) Learning tests
( 1 ) Electroshock learning. Learning tests involving odor discrimination and electric shocks followed the original procedure of Quinn et aL4'. Briefly, the apparatus consists of plastic tubes containing grids that carry the odor and can be electrified. Adult flies were induced to run into these tubes, such that they were alternatingly exposed to the two odors and were shocked in the presence of one of them. Two training cycles were followed by a test in which fresh tubes and no electric shocks were presented. Flies which had not entered the tubes after 30 s were counted. Learning indices (A) were calculated from two separate groups of flies which had been trained to avoid one or the other odor:
N, and N, designate the number of flies not having entered the tube containing the odors A or B respectively. N is the total number of flies in the test. A and B refer to tests after electric shocks had been paired with odor A or B41.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
5
(2) Sucrose reward learning. Sucrose reward learning has been described by Tempe1 et al.52. We used the slightly modified version of B o d . Instead of being shocked, the flies encountered 2 stripes of 1 M sucrose in one of the odoriferous tubes during training. In the immediately following test trial flies had to choose between two tubes, each of which was scented with one of the odors. The distributions of flies previously fed in the presence of either of the two odors were then compared. Learning indices A were calculated as d e ~ c r i b e d ~ . ~ ~ :
In this case Na and Nb designate the number of flies counted in the tube containing the odors A or B, respectively. Again A and B refer to tests after sugar had been paired with odors A or B. Flies which during the test did not enter either of the tubes were disregarded. An average odor bias P was calculated from the same data using the equation:
P > 0 indicates a bias towards A, P < 0 indicates a bias towards B.
(3) Larval learning. For larval olfactory conditioning, the procedure of Aceves-Pifia and Quinn' was changed in the following details. The electrically conductive gel was 1 .I5 % agar containing 0.02 m LiC1. As electroshocks, 200 V 50 Hz pulses lasting 100 ms were applied at 0.5 s intervals. The training sequence was fully automatic. Larvae received electric shocks in the presence of one odor and alternately were exposed to a second odor without shock. In the subsequent test the two odors were presented simultaneously at two locations and their attractiveness was compared. Learning scores were calculated as above (section bl).
(4) Arena paradigm. 3-6-day-old flies had their wings cut under anesthesia with nitrogen 24-30 h before the experiment. They were placed in a clean culture vial containing a wet paper towel instead of food. Just before the experiment flies were separated without anesthesia and placed singly in empty vials. Experiments were conducted in dark-red light. Under these conditions, the flies' vision is poor (see e.g. ref. 24). The training arena was a circular platform (@ = 110 mm) of tissue paper spread over a dish containing the odorant ( 5 0 ~ 1 of the indicated concentration in pa ra in oil in each of two small vials). The arena was surrounded by a water ditch in order to keep the flies on the platform (Fig. 1). During the training individual flies were allowed to feed on 1 M sucrose for 10 s in the presence of one of two odors. This was repeated 3, or in a few experiments, 2 times.
Immediately after training, flies were tested for 2 min on a platform covering 8 radial compartments which alternately contained the two odors (20 p1 of the indicated
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
6
ARENA PARADIGM
ODOR A
O D O R B
Fig. 1. Experimental set-up for olfactory conditioning of individual flies on a circular arena surrounded by a water ditch. Flies are fed with sugar on either the leR or the right arena and are tested afterwards on the eight-sector arena shown in the middle
concentration in each compartment). The time the flies spent over each of the odors was recorded. At le?vt 10 flies were trained at each of the two odors. The odors were chosen in an alternating sequence in order to avoid experimental bias. It was crucial for successful conditioning to gently drop the flies onto the arena and take them up into the vial without too much agitation after training.
The learning index was calculated as follows. For each fly the “odor bias”, pa, (with respect to odors A and B but irrespective of previous training) is the time the fly spends over A divided by the total test time for this fly. Pa, is the mean of the odor biases of all flies irrespective of whether they had been trained on odor A or B (see Figs. 7,9). Par* (ParB) is the average of the values for all flies trained on odor A (B). The learning index is given as Aa, = Par* - ParB. (Note that the average change of odor bias per fly due to the training is only half this value.) For all learning scores the level of significance for being different from zero was calculated using Student’s t-test. *** indicates P < 0.001; ** P < 0.01; * P < 0.05.
(c) Test for odor bias of naive flies
The apparatus developed for the sucrose reward paradigm of Tempe1 et al.52 was used. Odor was applied to only one of the tubes; the other tube contained a grid treated with pure solvent (ether) instead of the odor solution. No sucrose was offered. Flies in the tubes were counted after 15 s. The response R was calculated as R = (No - N)/No + N), No being the number of flies in the tube containing the odor, N the number of flies in the tube without an odorant.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
7
(d) Osmotropotaxis
Single flies were made to walk in a locomotion recorder’ which measured the turning tendency of the stationary animal. The antennae were stimulated with two jets of air scented with the odor of fermenting banana. The odor concentration on one side was always constant, on the other side it was varied as indicated. For details see ref. 5.
Drosophila mushroom bodies resemble the general scheme outlined in the intro- duction. Closer inspection, however, reveals that they consist of two parts differing in fine structure2’. Both originate as Kenyon fibers from the cell body layer above the calyx and grow in parallel to the front using the same track. One substructure consisting of about 65 % of the fibers constitutes the so-called median peduncle, the ci- and P-lobes. It has a constant cross-section throughout since each of its Kenyon fibers extends over the whole length of the three arms and the bundle is widened only a little by extrinsic fibers. The other substructure consists of the lateral peduncle, the “knee”, the y-lobe and a wedge-like fiber mass in front of the a-lobe. It has a very small cross-section in the caudal peduncle where the Kenyon fibers grow in parallel. The cross-section is much larger in the knee, wedge and y-lobe where the Kenyon fibers form a loose meshwork in which extrinsic fibers are incorporated abundantly.
The calyx receives input through about 200 fibers of the antenno-glomerular tract (AGT), many of which come from the antennal ganglion (relay neurons), and through roughly 80 fibers from the dorso-lateral p ro to~erebrum~~. These fibers possibly originate in visual centers; according to Strausfeld et al.47, the calyx can be stained in large flies by cobalt injection into the optic lobes.
(B) Structural defects of mutants and their phenotypic variability
( I ) Mushroom bodies derangedKS65 (mbd). The phenotype of mbd has been briefly described before2’. Instead of growing from the dorso-caudal margin of the brain through the central neuropil towards the front, the Kenyon fibers form a large “turban”
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
8
Fig. 2. Schematic drawing of an opened head capsule of Drosophilu showing the most prominent parts (enlarged) of its central brain. Special attention should be paid to the mushroom bodies and their connection with the antenna1 lobes via the antenno-glomerular tract (AGT).
at the site of the calyx. No peduncle, no a-, 8- or y-lobes can be detected. This phenotype is the most extreme of a continuum which, at least at the level of the light microscope, overlaps the wild-type. In most mutant animals a small peduncle can be found, containing a few large and a few hundred small fibers. A small fraction of animals have
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
9
Fig. 3. Golgi stained mushroom bodies of a wild-type DrosophiZu (fronto-horizontal section, dorsal on top). The Kenyon cells run from the calyx to the ventro-rostra1 part of the brain and there bend medially forming the py-lobe (cl-lobe not visible here). Courtesy of K.-F. Fischbach.
intermediate-sized pedunculi, and some have normal-looking ones. The size of the turban is inversely related to the size of the peduncle and lobes: the smaller the peduncle and lobes, the larger the turban.
The defect in the mutant develops during metamorphosis. The larval m.b. has normal shape and size. As in wild-type, the bulk of Kenyon fibers degenerate during the first hours after puparium formation5’. In the wild-type flies, these fibers regrow along the old track. In the mutant, however, they wind up in the calyx area. In some cases, the bundle of Kenyon fibers grows along the AGT, reaching the region where,’in the wild-type, the &lobe is located (‘‘false peduncle”). The reason why the fibers are misrouted is not known. A possible explanation comes from the following observation: a bundle of about 500 very thin Kenyon fibers which is found in the peduncle of the late 3rd instar wild-type larva, appears to be spared from degeneration in metamor- phosis. This fiber mass is frequently missing in the mutant larvae; we infer that, in normal development, it may guide the regenerating fibers through the neuropi15’. No structural defects other than the misrouting of Kenyon fibers have been found. At the level of the light microscope the antennal lobes, AGTs, optic lobes and central complex appear to develop normally. No excess degeneration of cell bodies beyond the level found in the wild-type is observed during late larval and pupal development.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
10
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
11
Without artificial selection for high expressivity of the defect, the mbd phenotype gradually reverts towards normal m.b.s. Alternatively, we have balanced the mbd- carrying X chromosome over a X̂ Y compound chromosome (containing a recombi- nation-suppressing inversion), in a stock lacking a free Y chromosome. In this way, selection pressure is largely eliminated since no homozygous or hemizygous mbd flies contribute to off-spring. The balanced stock has kept high expressivity over the last 5 years without further selection. However, the homozygous cultures, which occasional- ly had to be re-established from the balanced stock, were slightly different from each other. In some cultures false pedunculi were rather frequent, expressivity was higher in males than in females or asymmetry between the two m.b.s was significantly more pronounced. Expressivity of the mutant phenotype was routinely measured for all cultures used in behavioral experiments. Pa ra in sections of mass histology prepara- t i o n ~ ~ ~ were inspected, and the derangement of each mushroom body was graded between 1 and 5 : 1, normal looking m.b.s; 2, only slightly enlarged calyx, good sized peduncle, no a-lobe, normal- By-lobes; 3, significantly enlarged calyx, medium sized peduncle, no a-lobe, reduced By-lobes; 4, large turban, thin peduncle, no a-lobe, small py-lobes; 5, large turban, no or very thin peduncle (less than 400 fibers), no lobes. The average grade of a culture (expressivity) is given in % of the highest possible grade. For all behavioral experiments the expressivity was between 80 and 95 % . (False pedunculi, which occurred rarely, were grade No. 5, although it is not clear whether Kenyon fibers are able to establish proper function in this condition.)
Mushroom bodyminiat~rP~’~(mbm). The most striking feature of this mutant is the sexual dimorphism of the m.b.s. In females most Kenyon cells are missing. No calyx, peduncle or lobes are detected under the light microscope in some females. In many female fiies, however, a rudimentary calyx and a small bundle of thin fibers presumably belonging to Kenyon cells can be found. In mbm males, m.b.s are well developed and may be even slightly larger than normal. As in the case of mbd, the phenotype is variable and overlaps wildtype.
The AGT is well developed and contains a full complement of relay neurons from the antennal ganglion. Fiber degeneration and increased size of glia have been observed in the female AGT, but the degree of this deterioration seems not to be related to the extent of the m.b. defect. In the rudiment of the female calyx, the ratio between Kenyon cell dendrites and club-like side branches of relay neurons is much smaller than in the wild-type. Also, the branches of the relay neurons appear to be substantially reduced55.
Variability is nearly as great in mbm as in mbd. High expressivity of the mbm defect has been maintained by single-pair matings. The graded scale used to estimate expres- sivity in mbm is similar to that used for mbd. However, in mbd the first sign of a defect appears in the a-lobe whereas in mbm the a-lobe is always visible if a peduncle can be detected. Thus, the gradients of expressivity seem to affect the two m.b. substructures in mbd and mbm in a different order (see above and ref. 21). In mbm, the defect can be judged in most cases from the size of the calyx. However, a ball-shaped medium-sized
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
12
calyx without any peduncle is observed occasionally. Possibly, a tiny bundle of Kenyon fibers has not found its way through the central neuropil during regeneration. For behavioral experiments mbm stocks were used in which expressivity was above 80%. Recently we observed that in flies hemizygous for the mbm mutant allele (mbmlDff2L)al) both m.b.s are uniformly missing. These animals have not been tested behaviorally as yet. We also constructed the double mutant mbd; mbm. Males of this stock have the structural brain phenotype of mbd flies, females of the double mutant resemble rnbm females.
The m.b. defect in mbm is clearly visible in female larvae of the late 3rd instar, whereas in the late 2nd instar both males and females still have normal looking m.b.s. The number of Kenyon fibers appears to decrease between the late 2nd and late 3rd instars, but en mass Kenyon cell degeneration has not been observed yet.
A further defect in the present mbm stock is increased cell degeneration in the optic anlage during the 3rd larval instar. The defect occurs in males and females. It has no obvious effect on the final size of the visual neuropil but it may be related to the slight forward rotation of the optic lobes and may give rise to a vertical band of vacuoles in the distal lobula observed in the imago. Whether this phenotype is caused by the mbm mutation has not been investigated.
(C) General behavior of mutants
Flies of the two mutant stocks were observed individually or in male-female pairs. Expression of anatomical defects was determined histologically after behavioral obser- vations. We observed flight and walking, start and landing, feeding, grooming, general activity and posture. As a measure of general fitness negative geotaxis was recorded (for data see ref. 7). Both sexes of mbd and mbm moved only slightly more slowly than Canton-S flies. Thus, taking into account the mixed genetic background of the flies, their climbing behavior can be considered normal. Even with the most severe expression of the m.b. defect, mutant flies behaved in an ostensibly normal manner. Simultaneous display of rivalling motor patterns, as in the case of mushroom body lesioned bees and locusts26, was never observed.
Special attention was payed to courtship behavior. Males without any visible pedunculi and lobes (mbd) performed the whole courtship sequence (following, wing extension, pulse and sine song (F.v. Schilcher, personal communication), attempted copulation and copulation) without any obvious disadvantage in competition with wildtype males. Recently R. Bader (in prep.) tested mbm tru-2 flies, which from their X/autosome ratio were female, but were transformed to male morphology by the tra-2 mutation. These mbm tru2 pseudomales express the m.b. defect and show normal male courtship behavior. Elements of conditioned male courtship behavior 18*46 were not studied in these doubly mutant flies.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
13
.6-
0
3
c rn
n I- . 5 - .- .- c . 4 - .- O
f .3-
u) Q
LL .- - 6 .2- c 0 .- c 0 . I - LL z
0 -
(0) Olfactory conditioning
A
( 1 ) Learningpeformance of mbd. In the original olfactory learning paradigm of Quinn et aL41, flies are conditioned by electroshock to refrain from positive phototaxis if exposed to one of two odors. Our Canton4 stock (Wiirzburg) achieves a learning index A = 0.31 2 0.03 (S.E.M.). Comparison of Fig. 5 (left) with the published datalo shows that fast phototaxis (measured as the fraction of flies entering the rest tube (1-R)) of these particular wild-type flies is less pronounced than that of the Canton-S stock (Pasadena) used in the original work. The fraction of flies entering the tube of the control odor, however, is about the same as that reported". Deterrence by electroshock for Canton-S (Wurzburg) is also very close to that reported for Canton-S (Pasadena) flies.
Fast phototaxis ofrnbdis even less well developed than that of Canton-S (Wiirzburg) (Fig. 5, right). This presumably is not an effect of the mutation: the mutant X chromo-
ELECTROSHOCK LEARNING
T
I I 1 1. 2. Test
A - + 0.31 +- 0.03
n - 6
C
R
I I I 1. 2. Test
- - 0.01 +- 0.03 A n - 7
Fig. 5. Behavior of wild-type Canton-S and mbd flies in the electroshock paradigm during training and testing. Fraction of flies in the start tube after 30 s is shown for the two training trials and the test. The learning index A, is the fraction of flies avoiding the shock-associated odor during the test, minus the fraction avoiding the control odor during the test. Odors used were 0.1 % 3-OCT and 0.2% 4-MCH (concentration refers to dilution in ether). Number of tests (n) as indicated; error bars, S.E.M. S, shock odor; C, control odor; R, rest tube.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
14
some was obtained from the Berlin wild-type strain, and the autosomes stem from various stocks of unknown origin. Preliminary experiments suggest that fast phototaxis of rnbd is about the same as that of Berlin wild-type strain. Taking into account the reduced phototaxis, the fraction of flies avoiding electroshock during training is similar to that observed with Canton-S. Given these properties of the mbd stock the largest possible learning that should be observable is A = 0.2. The learning index of mbd is A = - 0.01 -1- 0.03; thus mbd is deficient in conditioning behavior, assuming it can distinguish the two test odors (see below and section E).
In the following experiment odor preference was modified by exposure to sucrose in the presence of one of the odors (sucrose reward learning, Fig. 6). The odors were the same as used in the electroshock experiment but at an arbitrarily chosen 5-fold higher concentration. Since the genetic background of mbd is not well defined, we measured 3 different wild-type strains. Their learning indices ranged from A = 0.21 (Canton-S) to A = 0.56 (Berlin)6. The small learning index for mbd, A = 0.08 ~f: 0.05 was not significantly different from zero. As in the case of electroshock, the mutant fies expressed the full response to the unconditioned stimulus (sucrose) but afterwards showed only a very small modification of their odor bias in the test trial. At the odor concentrations used, 3-OCT (odor A) was much more repellent for mbd than 4-MCH [see values for average odor bias (P) in Fig. 6 and Material and Methods for definition of this parameter]. Thus the mutant is able to distinguish the two odors. The 4 columns in the left part of Fig. 6 suggest that the average odor preference is inversely related to the learning index. Although the average odor preference of mbd is very close to that of Canton-S, we considered the possibility that the learning deficit in the mutant might be due to the large odor bias (P = 0.37 -1- 0.10). We reduced the concentration of 3-OCT 5-fold for Canton-S and mbd flies. In this experiment (Fig. 6, right) the average odor bias was much smaller, and learning scores were indeed larger. Learning of mbd was still worse than normal but with 13 independent experiments it now was highly significant.
The sucrose-reward and shock-avoidance olfactory learning paradigms resemble each other in that flies are allowed to walk in narrow tubes over grids covered with the odorants, and the animals are repeatedly agitated during the experiment. We wondered how much the learning defect of mbd depended upon the special procedure of these experiments. We therefore developed a new paradigm in which the flies are gently deposited individually on a platform (arena paradigm; see Material and Methods). Using flies from the Berlin and Canton-S strains, 20 individuals (from each) gave highly significant learning indices between Aar = 0.2 and 0.3 (Fig. 7 and Table 11). Using geraniol and geranylformiate as the two scents, mbd showed very little learning (Fig. 7, left). With 3-OCT and 4-MCH (the two substances used above) the learning index again was still smaller than that of Berlin flies but now was significantly larger than zero (Fig. 7, middle).
For each mbd fly tested behaviorally the degree of expression of the m.b. defect was determined histologically. Learning scores were calculated separately for the group of
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
. 6 A I
< 5 . 4 U S
m Is S . 2 L 0 QI
U .3
.r(
-1 . 1
0 ,
P
n
* *
15
SUGAR REWARD L E A R N I N G
0.5% 3-OCT <-> 1% 4-MCH 0.1% 3-OCT <-> 1% 4-MCH
* * * * - KaP
T
cs mbd
T
cs mbd
. I I + - . 0 4 .20+- .03 .30+-.09 .3 7+- .10 -.08+-.09 .02+- . 0 6
3 3 3 4 7 13
Fig. 6. Sucrose reward learning of 3 different wild-type strains and rnbd (left). The P-values below the columns indicate the balance of unconditioned responses between both odors (note the negative correlation between P and A). The test was repeated with Canton-S and mbd using a 5-fold lower concentration of 3-OCT (right). Number of tests (n) as indicated; error bars, S.E.M. Average odor bias P is calculated according to the equation given in Materials and Methods.
flies which had grade 4 or 5 m.b. defects in both hemispheres and for all other flies. A difference in learning performance between the two groups was not apparent. Flies with even the strongest expression of the m.b. defect (i.e. with less than 400 Kenyon fibers on each side) are able to learn certain odor discrimination tasks using sucrose as reward. However, only stocks with high degree of expression of the defect were tested in this manner (Table I).
As mentioned above, the mbd stocks are maintained by selecting lines of high expressivity of the structural defect in order to reduce the accumulation of genetic modifiers. Since such lines might by chance carry other mutations affecting learning behavior we applied the arena test to two further stocks carrying the mbd mutant allele (Fig. 7, right). They both have small, non-significant learning scores. The attached X̂ Y stock is particularly useful in this respect since it has two types of males differing only in their mbd allele and the X and Y chromosomes. The attached X?/O (mbd+) males gave a normal learning score, whereas the X/O (mbd) males of the same culture learned poorly if at all.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
GL <-> GF
f
1
ARENA PARADIGM
3-OCT <-> 4-MCH
WTB m b d
f 4 4 -
WTB
GL <-> GF
r n b d w - m b d / Y mbd/Q X Y / O
Par 0 *- .03 -. 1S+- .05 -.Is- .m -.w.- .os - .00+- .07 - . I I * - . O 4 - . I Z . - . O S
n 00 30 M to 20 30 30
Fig. 7. Learning performance of Berlin and mbd flies in the arena paradigm (left). To test for effects of genetic background (right), the experiment was also performed with the double mutant w mbd and with a stock in which halfthe males carry the mbd mutant allele (mbd/O), the other half an attached x^Y compound chromosome (XY/O) with the mbd+ allele. Odor concentrations were 0.05% (GF, GL) and 1 % (3-0CT, 4-MCH). Number of tests (n) as indicated; error bars, S.E.M.; Pa,, average odor bias.
TABLE I
LEARNING PERFORMANCE OF mbd FLIES WITH DIFFERENT DEGREE O F EXPRESSIVITY OF m.b. DEFECT
Grade* Mean Learning Odor bias, Number expressivity ( % ) index, A,, p*r of flies
1-3 45 0 12 k 0.11 -0.16 5 0.07 25
4 + 5 > 95 0.11 2 0.05 - 0.08 0.03 82
* For definition of grades see Results B1. The arena paradigm (see Material and Methods and Fig. 1) was used to measure learning performance. Flies were males (mbd/O) of the stock In(l)EN, In(l)dl-49, y v f car Y/mbd/O. Odors: 1 % 3-OCT vs 1 % 4-MCH; errors, S.E.M.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
17
(2) Learning of mbm. The same learning paradigms were applied to mbm. Because of the sexual dimorphism of m.b. structure in this mutant, the behavior of males and females was compared. For wild-type, the learning indices of the two genders are not significantly different (see, for instance, ref. 7 and Table 11). In the first experiment using the electroshock procedure we had scored males and females together, since the sexual dimorphism was not known at that time. As a preliminary result from 3 measurements a learning index of A = 0.07
In the modified version with sucrose as a reward, the two sexes were tested together in the same experiments and then scored separately. Their learning performance in the choice between 3-OCT and 4-MCH was dramatically different. In males, odor bias could be modified as much as in wild-type, whereas no significant conditioning effect was observed in females (Fig. 8, left). The average odor bias (P) of the females was
0.04 was obtained.
LEARNING PERFORMANCE OF mbrn SUGAR REWARD LEARNING
Fig. 8. Performance of mbm flies in two different learning paradigms. The learning index of mbm females is reduced in correlation with the structural defect throughout.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
significantly different from zero (P = 0.08 0.03) showing that the flies could dis- tinguish the two odors (see also Section E). In both genders the response to sucrose was less pronounced than in the case of Berlin wild-type. We compensated for this effect, however, by increasing the starvation period applied to the mutant before the training. Only those experiments were considered where the flies showed a strong response to the sucrose.
We compared the expressivity of the m.b. defect between the female flies which in the test trial had chosen the odor previously associated with sucrose and those which had entered the chamber of the control odor. The two groups expressed the m.b. defect equally strongly. Thus in mbm females, as in mbd flies, the degree of morphological defect is not correlated with learning performance.
In the arena paradigm, a female-specific learning defect also was observed (Fig. 8, right). We tested 3-OCTvs 4-MCH and geraniol vs geranylformiate as odor pairs. Males learned well, whereas the learning scores of females were non-significant or small. In both cases average odor bias was significantly different from zero and was sirmlar for males and females.
As mentioned above (Section B), the m.b. defects arise at different stages in the life cycle of mbd and mbm. Reorganization in the early pupal period is impaired in mbd, whereas degeneration of the mushroom bodies occurs between the late 2nd and late 3rd larval instars in mbm females. These develop- mental differences are reflected in larval learning tasks.
Overt larval behaviors, such as feeding and locomotion, appear to be entirely normal in the two mutants. The two genders of mbm larvae cannot be told apart by general behavioral observations. A learning paradigm for larvae has been developed : wild-type Canton-S larvae learn to avoid one odor in a simultaneous choice test between two odors, if shock is paired with the odor during training. We trained and tested mixed populations of male and female larvae. Sex was not determined until after the test, when the larvae were already separated according to their odor preferences. Fig. 9 indicates that wild-type Berlin and mbd larvae readily change their odor preferences in response to electroshock previously paired with one of the odors. Although data are not shown, it is clear from part of the experiments that male and female larvae contribute about equally to the learning score. After analogous experiments with mbm larvae, sex was determined in all experiments, and we found that 3rd instar larvae of mbm have a strong female-specific learning defect in this paradigm( Fig. 9). Control experiments indicated that, without electroshocks, mbm larvae of either sex are attracted by both odors at the concentrations used here (data not shown).
(3) Learning ofmbd and mbm larvae.
(E) Responses to odors of naive flies
It has already been shown in the previous section that rnbd flies and mbm females are not generally anosmic. Average odor bias in the learning tests in most cases was significantly different from zero (see also ref. 21). The results in this section concern the characterization of responses to odors of naive flies.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
19
* * *
a T
* * *
L A R V A L LEARNING
0. l m A A <-=. 0. lm 3-OCT
W TB mbd
n 33 39
* * t t -
mbm ITI
mbm males f emalee
18 10
Fig. 9. Learning performance of late third instar larvae of wild-type Berlin, mbd and mbm. During training, 0.1 M n-amylacetate and 0.1 M 3-OCT were presented alternatingly, one combined with electroshock pulses. The distribution of larvae between the two odorants was determined afterwards. Number of tests (n) as indicated; error bars, S.E.M. Note that the mushroom body defect ofmbdflies does not develop until metamorphosis, whereas in mbm females the defect develops during the third larval instar.
(1) Osmotropotaxis. As reported earlier5, Drosophila can perform turning re- sponses, when an attractive odor arrives at the two antennae in slightly different concentrations. Responses are recorded from tethered flies walking on a locomotion recorder. We recorded osmotropotaxis with mbd and mbm in order to test whether the m.b.s are involved in the detection of spatial odor gradients. In responses to the odor of fermenting banana, female flies of the two mutant strains, like wild-type, show a turning tendency towards the higher of the two concentrations (Fig. 10). In this particular recording the response amplitudes of mbm were significantly smaller than those of Berlin and mbd. This was apparently caused by the high degree of steady walking activity exhibited by mbrn flies, which also produced smaller error bars (Fig. 10). Increased motor activity is not, however, typical of mbm flies in general. The response threshold for the right: left ratio was about the same in the 3 strains. The lowest ratio eliciting a significant turning reaction is 5: 1 in mbd and 10: 1 in the wild-type and mbm. (A significantly lower discrimination threshold for Drosophila has been reported5, which probably is due to the much smaller range of odor concentrations used previ-
Using the test apparatus and procedure of the sucrose reward paradigm, responses of naive flies to 3-OCT and 4-MCH were investigated. One
ously.) (2) Odor bias of naiveflies.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
20
TROPOTAXIS Discrimination Sensitivity
mbd
mbm
Concentration r a t i o ->
0 0 . 01 . I .P .5 1
Concentration r a t i o ->
. 01 . 1 .2 .5 1 -+ 0 +
Concentration r a t i o ->
Fig. 10. Osmotropotactical response, to fermenting banana, of stationarily walking female flies from wild-type and mutant strains. Concentration difference between the left and right antennae decreases to zero from left to right (values on the X-axis indicate the concentrations in one capillary relative to the other). Note that the response value R (turning counts to the side of higher concentration, divided by the forward counts) is activity dependend (increased walking activity of mbm females causes reduced response ampli- tude). See ref. 5 for additional details; 10 flies of each strain were tested for 1 or 2 h; error bars, S.E.M.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
21
SPONTANEOUS ODOR PREFERENCES
ma 1 es fernal es
L ::$ ., A ,
h . 4 ::$
I I I I : I I D .m, ,012 .o .I .I a2
Concnntrntlon [X I ->
I I I I : I i 0 .w ,012 .o .2 ., a2
Conssntr.atlon t11 ->
Fig. 11. Spontaneous response to 4-MCH and 3-OCT (both versus air) of flies from the wild-type Berlin stock (closed circles, solid line), from mbd (open circles, dashed line, symbols displaced to the left), and from mbm (open diamonds, dotted line, symbols displaced to the right). Tests were repeated 10 times with about 50 flies per run; error bars, S.E.M. Negative response values indicate a repellent effect of the odorant. Odor concentrations refer to dilution in ether.
choice tube was not scented; the other one contained one of the odors at concentrations indicated in Fig. 11.
For Berlin, mbd and mbm flies, 3-OCT was repellent in the whole effective con- centration range. No pronounced differences between the strains were detected in the avoidance threshold for 3-OCT. The response amplitude of mbm males and females was significantly smaller than that of WT Berlin and mbd. However, no differences were detected between the two genders of mbm. Incidentally, it should be noted, that the threshold for detectability is below the concentrations used here as shown in con- ditioning experiments6.
For wild-type flies and mbd males, 4-MCH is attractive in the low concentration range, becoming repellent at concentrations above 0.2%. mbd females are not attracted by 4-MCH at low concentrations, and 4-MCH is weakly repellent over the whole concentration range for mbm flies. No sexual dimorphism, which might be related to the m.b. defect, was observed. Whether the small response amplitude at high con- centrations is caused by the mbm mutant gene has not been assessed.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
22
(F) Visual learning
So far only the mutant mbd has been tested for visual conditioning. W. Roos and C.H. Spatz (personal communication) kindly measured color and intensity discrimi- nation learning in the paradigm developed by Menne and spat^^^. The mutant learning scores ranged between those of the two wildtype strains Achkarren Stid and Canton-S (Fig. 12). The expressivity of the m.b. defect in this sample of flies was above 95%. Apparently intact m.b.s are not required for this task.
Recently, Heisenberg and W 0 P 4 developed an operant learning paradigm, in which the flies at a torquemeter have to invert the sign of their body saccades in order to establish proper visuo-motor coordination. Mutant mbd flies, with fully deficient m.b.s, performed as well as wildtype flies (data not shown).
(C) Oyactory learning in other structural brain mutants
In order to test whether the impairment of olfactory learning is specific to m.b. mutants we tested a variety of other structural brain mutants in the arena paradigm using
COLOR DI SCR I M I N AT1 ON LEARN I NG
T
L
Ml,.D,
l T I Conla'"er
1 r' I Shahar
cv c V CV c
AS+ cs GV c V
m b d t strain mbd in the color Fig. 12. Learning performance of two different wild-type strains and the muta
discrimination paradigm of Menne and Spatz (see ref. 34 for explanation of experimental procedure). Flies were shaken heavily for 30 s while being illuminated with one light and let undisturbed in the other. After 9 training runs, distribution of flies between two differently illuminated tubes was determined. GV, wavelength contrast of green vs violet; V, intensity contrast of violet vs dark; G , intensity contrast of green vs dark. Tests were repeated 8 times with about 100 flies per test; error bars, S.E.M. Courtesy W. Roos and H.-C. Spatz.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
23
geraniol vs geranylformiate as odors (Table 11). Three mutants with pronounced defects in the m.b.s all have reduced learning scores. The mutant mbmB, which showed weak but positive learning, has partially formed m.b.s. In the mutant mnb, which did not learn, the whole brain is reduced in volume. We do not know how much the m.b.s are affected. The 3 optic lobe mutants omb, sol and lop gave normal learning scores. This finding is in agreement with earlier results indicating that the structural defects in these mutants are restricted to the visual system (ref. 16 and K.F. Fischbach, unpublished).
Surprisingly 4 of the 5 mutants (cbd, nob, ccb, ebo, ccd) with structural defects in the central complex were deficient in olfactory learning. Of all central brain mutants tested so far, only the mutant ebo gave a normal learning index. However the overall behavioral
TABLE I1
OLFACTORY LEARNING OF STRUCTURAL BRAIN MUTANTS IN THE ARENA PARADIGM
Strain Name Sex Map Aar S.E.M. P,, n position
WTB Berlin WTB Berlin cs Canton-S
Mushroom body mutants: mbmB mushroom body miniature B mbmB mushroom body miniature B rnbmC mushroom body miniature C mbmC/CyO of same stock cxb calyx bulging
General brain hypoplasia mutant: mnb minibrain
Optic lobes mutants: omb optomotor-blind sol small optic lobes lop lobulaplate-less
Central complex mutants: cbd central body defect nob no bridge ccb central complex broad ebo ellipsoid body open ccd central complex deranged
f m 0
m f f f m
m
m m m
rn m m m m
2-3 1 2-3 1 2-35
2-
X-58.1
x-7.5 X-68 2-
X-4 1 x-12 X-56 X-0.6 X-15
+ 0.30 + 0.24 + 0.21
+ 0.13 + 0.15
+ 0.25 - 0.07
- 0.05
- 0.08
+ 0.22 + 0.30 + 0.27
+ 0.04 - 0.01 - 0.03 +0.21 + 0.10
50.06 - 0.14 k 0.03 + 0.00 kO.06 - 0.08
k0.05 -0.11 +0.07 - 0.09 k 0.08 - 0.05 k 0.04 - 0.04
0.08 -0.05
+ 0.05 -0.05
k0.05 -0.10 k 0.05 -0.06 +0.05 - 0.10
k 0.05 -0.08 0.06 -0.04
t 0 . 0 6 -0.05 - +0.03 t 0.02 k 0.05 + 0.02
20 80 20
28 30 40 20 20
40
20 20 20
40 60 50 60 40
Odors: 0.05% geraniol vs 0.05% geranylformiate; P,, odor bias; n, number of flies tested; Aar, learning score (for description of mutants and definition of terms, see Material and Methods); with regard to the map positions of the mutations listed, “2-” means that two of the mutations here have been localized to chromosome 2, but not yet mapped to specific loci.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
24
properties of these mutants are not well investigated. Small differences in arousal, sensitivity to sugar and odor would not have been detected. It remains to be investigated whether the poor olfactory conditioning in the 4 central complex mutants (cbd, nob, ccb, ccd) reflects a special role of the central complex in learning, a general interdependence of the various parts of the brain, or an additional impairment of the olfactory system.
DISCUSSION
The results have shown that in the mutants mbd and rnbm, olfactory conditioning is disturbed in all forms (larva, imago, female, male) with reduced m.b.s and is normal in all those with well-developed ones. Thus our data from Drosophila corroborate the hypothesis of Menzel et al.'3*35 proposing that m.b.s are involved in olfactory con- ditioning.
The apparently normal general behavior of mbd and rnbm flies is at variance with the possible expectation (see Introduction) that the m.b. mutants would be defective in motor coordination and the control of certain motor programs, in particular sexual behavior. It is, however, still possible that m.b.s are involved in motor programming. Insects might have the means to partially compensate for the loss of m.b. function if it occurs during development but not if caused by surgery. Such a hypothesis would suggest that m.b. mutants may in fact show impairment of motor coordination under more extreme conditions of testing. More likely, however, the surgeries or stimulations performed on large species affected other parts of the brain possibly in the immediate vicinity of the m.b.s. Also in Hall'sI9 experiment with Drosophilu gynandromorphs the tissue which had to be male for male courtship behavior to occur could well have been only near to, but not including, the m.b.s.
The remainder of the Discussion will evaluate the new evidence on olfactory learning in m.b. mutants at genetic, developmental and behavioral levels.
(I) Does the same mutation cause the structural and behavioral phenotype?
The genetic experiments indicate that in each of the two mutant strains a small region of the genome has the decisive effect on the structural phenotype but many other parts of the genome influence the m.b. structure as well. This situation being characteristic of a large class of developmental mutants impedes genetic analysis. At present, mbd and mbm are mapped only according to their structural phenotype. If the behavioral phenotype maps to the same position, one can conclude that these phenotypes are two expressions of the same mutation.
Our assertion that a given mutation causes both the structural and the behavioral phenotype is based on indirect arguments. The 5 m.b. mutants, which were selected in two independent mutant searches from two different wild-type strains and which have their map position at different positions on two different chromosomes all have been
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
25
selected according to their structural defects only. It is unlikely that with each of them an independent allele associated with learning deficits would have been selected. Furthermore, for mbm and mbd these (hypothetical) genetic variants would be required to have the same specificity as the mutations causing the structural phenotype, i.e. affect only females in mbm and only adults in mbd.
(2) Are the reduced m.b.s the cause for poor olfactory conditioning?
In none of the structural brain mutants so far isolated is the chain of causations from the primary gene product to the brain defect understood. It can be argued that the structural and the behavioral defect may be independent consequences of the same mutation rather than the abnormal behavior being the functional expression of the structural defect. Parallel investigation of the two mutants mbd and mbm helps to distinguish these alternatives. The development of m.b. defects in the two mutants is entirely different. In mbd, the Kenyon fibers are misrouted while regenerating during metamorphosis; whereas, in mbm, the fibers degenerate (or fail to be formed) during the 3rd larval instar. We infer that these very different mutation-induced defects are a reflection of quite different modes of action of these two loci during CNS development. It seems likely, therefore, that both mutations interfere with olfactory conditioning because of their common effects on the m.b.s. Since only a few Kenyon fibers arrive in the region of the “knee” and the a-, p-and y-lobes in both mutants, m.b. function should be affected alike.
Even though the m.b. defects cause impairment of learning, it does not necessarily follow that the m.b.s themselves mediate learning. The two mutant strains may have common defects “downstream” of the m.b.s. The extrinsic neurons in the region of the “knee” and the lobes most likely are deprived of their proper synapses. As a result they may be badly tuned and, hence, impinge upon the function of other parts of the central neuropil. A similar argument applies to the experiments using surgical lesions and local cooling.
The fact that the degree of structural impairment is not correlated with the degree of learning impairment for individual mutant flies of a given strain is still dificult to reconcile with the hypothesis that the m.b. defects cause the behavioral defects. However, this discrepancy is found at a gross anatomical level. A correlation may well exist at the level of synaptic connections between Kenyon fibers and extrinsic output neurons if, for instance, sets of extrinsic neurons were competing for permanent and temporal synapses. But, so far, no data are available to back up this idea. The situation is reminiscent of the relation between the song repertoire and the “vocal control center” in canaries, for which the over-all size of this brain nucleus in many respects does not represent its functional capacity in b e h a ~ i o r ~ ~ , ~ ~ .
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
26
(3) Does the m.b. defect interfere with olfaction?
Our data indicate that mbd and mbm qualify as true learning mutants. They sense the unconditioned stimuli sucrose and electroshock and they perceive the two odors as different. Thus they should be impaired in the association process, storage or retrieval.
However, these criteria may be questioned. Odor quality and concentration have an influence on learning performance (e.g. Figs. 6 and 7, and ref. 6). Responses to certain odors may be genetically fixed, others may be modifiable by experience. It is not enough to show that the fly can distinguish between the two odors presented. Accordingly, we studied the naive odor responses of the mutants.
Fermenting fruit is highly attractive for both mutants. 3-OCT and benzaldehyde2' are repellent as they are for the wild-type. 4-MCH which is attractive at low concen- trations for Berlin, is repellent over the whole concentration range for mbm and females of mbd. Also, at high concentrations the responses of mbrn are smaller than those of' Berlin and mbd.
The significance of these differences between wild-type and mutant strains is questionable. First of all, different wild-type strains may differ considerably in their naive odor responses (E. Marshall and A. Borst, unpublished). Second, these differences cannot be attributed to the m.b. defects in the mutants. M.b.s are impaired in males and females of mbd and only in females of mbm, whereas the attractiveness of 4-MCH is missing in females of mbd and both genders of mbm. Also, the small responses to 4-MCH and 3-OCT at high concentrations are observed in males and females of mbm. With the tests used so far abnormalities in naive odor responses attributable to the m.b. defect have not been detected. It should be emphasized, however, that the tests are very crude and may not tell much about recognition and evaluation of odors in Drosophilu.
(4) What is the role of the m.b.s in behavior?
A salient feature of the insect olfactory system is that the olfactory pathway splits into two at the level of the so-called relay neurons. These fibers connect the glomeruli of the antennal lobe to the dorso-lateral protocerebrum and also send prominent side branches into the calyx of the m.b.8*16*21. In the mbd and mbm mutants, the m.b. pathway (MBP) is largely blocked whereas the pathway via the dorso-lateral proto- cerebrum (DLPP) may be functioning normally (for a discussion of pathways in the Drosophilu brain see ref. 16).
What are the special functions of these two pathways? At present, we can make two assignments: (a) The independence of naive odor responses and the state of the m.b.s in the two mutants, suggests that naive odor responses are mediated by the DLPP. Thus this pathway must provide the information about odor concentration (osmotropotaxis) and, at least to some extent, about odor quality6. (b)The dependence of olfactory conditioning on the state of the m.b.s shows that plasticity of odor evaluation is provided by the MBP, although it cannot be excluded that some olfactory learning can be mediated by the DLPP.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
21
One can argue that the DLPP carries all the information about odor concentration and quality and thus no other basic olfactory function is left for the MBP except plasticity. Furthermore, the long lasting aftereffects in the responses of Kenyon cells and extrinsic output neurons to sensory s t i m ~ l i ’ ~ , ~ ~ , ~ ~ would be suited to “store” single events for the establishment of short term and long term memory. Many of the extrinsic output neurons receive information from various sensory m ~ d a l i t i e s ’ ~ , ~ ~ , ~ ~ , and at least some of them join the DLPP in the dorso-lateral p r~ tocerebrum~~. Thus they may serve to modify the DLPP at the synapses between relay and descending neurons. Alterna- tively, one can argue that the MBP might provide the fine evaluation of odor quality and that olfactory conditioning, then, reflects the plasticity of this function. Several observa- tions suggest a role of the m.b.s in the evaluation of odor quality.
As mentioned above, most Kenyon fibers degenerate and regrow during the first 20 h of the pupal period5*. A plausible explanation would be that the significance of some odors are different for larvae and imagos. In this context, the sexual dimorphism of the m.b.s in mbm is illuminating. It suggests that wild-type m.b.s may be sexually dimorphic not only in fiber number5() but possibly also in the connectivity of Kenyon fibers and extrinsic neurons. Significances of some odors in fact are different for the two genders, as recent osmotropotaxis experiments have shown (E. Marshall and A. Borst, unpub- lished). Although direct evidence for the involvement of the MBP in odor evaluation is still missing, this second hypothesis cannot be dismissed.
Following this argument one has to explain the apparent independence of naive odor responses and the state of the m.b.s in the mutants. Since methods now available for assessing the fly’s performance in the evaluation of odor quality are very crude, a partial replacement, in the mutants, of the MBP’s function by the DLPP might go undetected. In the visual system, for instance, the mutant optomotor-blind ’6*24 displays no large field course control response, but the mutant fly can use the object response instead for stabilizing its course.
Recently, Technauso discovered that the size of the m.b.s in Drosophila depends upon experience. Flies kept in isolation in small vials with only the standard food medium have significantly fewer Kenyon fibers than flies living in a population cage with various odor sources, plants and other flies. Olfactory deprivation seems to have the strongest influence. The rate of these structural changes presumably is much slower than the short term memory studied in the present series of experiments. Nevertheless, it is attractive to speculate about a possible relation between the role of the m.b.s in short term memory and their structural plasticity. It is conceivable that certain conditioning events leading to long lasting memory traces require an irreversible commitment of synapses between Kenyon fibers and extrinsic neurons. If so, a supply of new fibers for the regeneration of learning capacity would be required. Such ideas, however, go far beyond the presently available experimental evidence.
Both proposals for the function of the MBP give no explanation for the peculiar shape of the m.b.s, in particular of the branching of the Kenyon fibers to form a- and P-lobes and of the spatial orientation of the lobes relative to each other. They also
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
28
disregard the possible involvement of the central complex in learning, indicated by the low learning scorces of 4 of the 5 mutants with structural defects in the central complex. The DLPP may be the “thru way” to fast olfactory behavior. In contrast, the apparent short term memory mediated by the MBP, may be the main pathway for processing the olfactory information involved in the higher functions of the central brain.
ACKNOWLEDGEMENTS
We thank Tim Tully and Jeffrey C. Hall for extensive revision of the manuscript, Erich Buchner, Jochen Erber and Karl-Friedrich Fischbach for comments, Hanna Karwath, Martina Weltner, Gerlinde Neckermann and Gertrud Sch’dflein for skillful technical assistance and Wolfgang Roos and Hans-Christoph Spatz for their data on color discrimination learning. The work was supported by a grant of the Deutsche Forschungsgemeinschaft to M.H.
REFERENCES
1 Aceves-Pifia, E.O. and Quinn, W.G., Learning in normal and mutant Drosophila larvae, Science, 206
2 Boeckh, J., Ernst, K.-D., Sass, H. and Waldow U., Zur nervijsen Organisation antennaler Sinnesein- gange bei Insekten unter besonderer Beriicksichtigung der Riechbahn. In Verh. Dtsch. Zool. Ges., Gustav Fischer, Stuttgart, 1976, pp. 323-139.
3 Bausenwein, B., Eigenschaften der Objektreaktion yon Drosophila melanogaster Diploma Thesis, University of Wiirzburg, 1984.
4 Belote, J.M. and Baker, B.S., Sex determination in Drosophila melanogaster: analysis of transformer-2, a sex-transforming locus, Proc. Natl. Acad. Sci. U.S.A. , 79 (1982) 1568-1572.
5 Borst, A. and Heisenberg, M., Osmotropotaxis in Drosophila melanogaster. J. Comp. Physiol., 147 (1982)
6 Borst, A,, Computation of olfactory signals in Drosophila melanogaster, J . Comp. Physiol., 152 (1983)
7 Borst, A,, Untersuchungen zur zentralnervosen Verarbeitung olfaktorischer Reize bei Drosophila melanogaster,
8 Borst, A. and Fischbach, K.-F., Neurons in the antennal lobes ofDrosophila melanogaster. Cell Tiss. Res.,
9 Buchner, E., Elementary Movement Detectors in an insect visual system, Biol. Cybern, 24 (1976) 85-101. 10 Dudai, Y., Jan, Y.-N., Byers, D., Quinn, W.G. and Benzer, S . , dunce, a mutant of Drosophila deficient
11 Dujardin, F., Memoire sur le systkme nerveux des insectes, Ann. Sci. Nut. Zool., 3 Vol. 14 (1850)
12 Erber, J., Response characteristics and after effects of multimodal neurons in the mushroom body area of the honey bee, Physiol. Entomol., 3 (1978) 77-89.
13 Erber, J., Masuhr, T. and Menzel, R., Localization of short-term memory in the brain of the bee, Apis mellifra. Physiol. Entomol., 5 (1980) 343-358.
14 Ernst, K.-D., Boeckh, J. and Boeckh, V., A neuroanatomical study on the organization of the central antennal pathways in insects. 2) Deutocerebral connections in Locusta migratoria and Pefiplane/a americana, Cell Tiss. Res., 176 (1977) 285-308.
(1979) 93-96.
479-484.
3 7 3 - 3 8 3.
Doctoral thesis, University of WBrzburg, 1984.
in prep.
in learning, Proc. Natl. Acad. Sci. U .S .A . , 73 (1976) 1684-1688.
195-206.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
29
15 Ernst, K.-D. and Boeckh, J., A neuroanatomical study on the organization of the central anatomical pathways in insects. (2) Neuroanatomical characterization of physiologically defined response types of deutocerebral neurons in Periplaneta americana, Cell Tiss. Res., 229 (1983) 1-22.
16 Fischbach, K.-F. and Heisenberg, M., Neurogenetics and behavior in insects, In M. Burrows, and J.E. Treherne, (Eds.), Mechanisms of integration in the Nervous System, Cambridge, 1984.
17 Fischbach, K.-F. and Technau, G., Cell degeneration in the developing optic lobes of the sine oculis and small optic lobes mutants of Drosophila melanogaster, Dev. Biol. 104 (1984) 219-239.
18 Gailey, D.A., Jackson, F.R. and Siegel, R.W., Male courtship in Drosophila: the conditioned response to immature males and its genetic control, Genetics, 102 (1982) 771-782.
19 Hall, J.C., Control of male reproductive behavior by the central nervous system of Drosophila: dissection of a courtship pathway by genetic mosaics, Genetics, 92 (1979) 437-457.
20 Hanstrom, B., Vergleichende Anatornie des Nervensystems der wirbellosen Tiere, Springer, Berlin, 1982. 21 Heisenberg, M., Mutants of brain structure and function: what is the significance of the mushroom
bodies for behaviour? In 0. Siddiqi, P. Babu, L.M. Hall and J.C. Hall (Eds.), Development and Biology qfDrosophila, Plenum, New York, 1980, pp. 337-390.
22 Heisenberg, M. and Buchner, E., The role of retinula cell types in visual behavior of Drosophila melanogaster, J. Comp. Physiol., 117 (1977) 127-162.
23 Heisenberg, M. and Boehl, K., Isolation of anatomical brain mutants ofDrosophila by histological means, Z . NaturjCorsch., 34 (1979) 143-147.
24 Heisenberg, M. and Wolf, R., Vision in Drosophila, Springer, Berlin 1984. 25 Homberg, U., Ableitungen and Lucifer Yellow Markierungen von Neuronen des Tractus olfacto-
globularis im Bienengehirn. In Verh. Dtsch. 2001. Ges. (1981), Gustav Fischer, Stuttgart, p. 176. 26 Howse, P.E., Design and function in the insect brain. In L.B. Brown (Ed.),ExperimentalAnalysis oflnsect
Behavior, Springer, Berlin, 1974, pp. 180-194. 27 Huber, F., Sitz und Bedeutung nervoser Zentren fir Instinkthandlungen beim Mannchen von Gryllus
campestris L., Z . Tierpsychol., 12 (1955) 12-48. 28 Huber, F., Untersuchungen uber die Funktion des Zentralnervensystems und insbesondere des Gehirns
bei der Fortbewegung und der Lauterzeugung der Grillen, 2. vergl. Physiol., 44 (1960) 60-132. 29 Kaulen, P., Feldpoientiale und eine Analyse der Stromquellendichte in den Corpora pedunculaia von Apis
mellifera, Diploma Thesis, Freie Universitat Berlin, 1982. 30 Kenyon, C.F., The brain ofthe bee. A preliminary contribution to the morphology of the nervous system
of arthropoda, J . Cornp. Neurol., 6 (1896) 133-210. 3 1 Kloot, W.G. van der and Williams, C.M., Cocoon construction by the Cecropia silkworm, Behavior, 5
32 Lewis, E.B. and Baker, F., Method of feeding ethyl methane sulfonate (EMS) to Drosophila males, Drosophila InJ Serv., 43 (1968) 193.
33 Matsumoto, S.G. and Hildebrand, J.G., Olfactory mechanisms in the moth Manduca sexta: response characteristics and morphology ofcentral neurons in the antennal lobes, Proc. R. SOC. Lond., 213 (1981)
34 Menne, D. and Spatz, H.-C., Colour vision in Drosophila rnelanogasier, J . Cornp. Physiof., 1 I4 (1977)
35 Menzel, R., Erber, J. and Masuhr, T., Learning and memory in the honeybee, In L.B. Browne (ED.),
36 Mobbs, P.G., The brain of the honeybee Apis rnellfera. I. The connections and spatial organization of
37 Nottebohm, F., Testosteron triggers growth ofbrain vocal control nuclei in adult female canaries, Brain
38 Nottebohm, F., Kasparian, S. and Pandazis, C., Brain space for a learned task, Brain Res., 213 (1981)
39 Otto, D., Untersuchungen der zentralnervosen Kontrolle der Lauterzeugung der Grillen, 2. vergl.
(1953) 141-174.
249-277.
30 1-3 12.
Experimental Analysis of Insect Behavior, Springer, Berlin, 1974, pp. 195-217.
the mushroom bodies, Philos. Trans. R. SOC. Lond, Ser. B, 298 (1982) 309-354.
Res., 189 (1980) 429-436.
99-109.
Physiol., 44 (1971) 60-132.
J N
euro
gene
t Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
Uni
vers
ity o
f A
lber
ta o
n 09
/02/
13Fo
r pe
rson
al u
se o
nly.
40 Pearson, L., The corpora pedunculata of Sphinx ligustri and other lepidoptera: an anatomical study, Biol.
41 Quinn, W.G., Harris, W.A. and Benzer, S., Conditioned behavior in Drosophila melanogaster. Proc. Natl.
42 Quinn, W.G., Work in invertebrates on the mechanisms underlying learning. In P. Mahler and H.S.
43 Schildberger, K., Local interneurons associated with the mushroom bodies and the central body in the
44 Schildberger, K., Multimodal interneurons in the cricket brain: properties of identified extrinsic
45 Schiirmann, F.-W., Bemerkungen zur Funktion der Corpora pedunculata im Gehirn der Insekten aus
46 Siegel, R.W. and Hall, J.C., Conditioned responses in courtship behavior of normal and mutant
47 Strausfeld, N.J., Bassemir, U., Singh, N. and Bacon, J.P., Organizational principles of outputs from
48 Suzuki, H. and Tateda, H., An electrophysiological study of olfactory interneurons in the brain of the
49 Technau, G., Die Entwicklung der Corpora pedunculata von Drosophila melanogaster, Doctoral Thesis,
50 Technau, G., Fiber number in the mushroom bodies of adult Drosophila melanogaster depends on age,
51 Technau, G. and Heisenberg, M., Neural reorganization during metamorphosis of the corpora peduncu-
52 Tempel, B.L., Bonini, N., Dawson, D.R. and Quinn, W.G., Reward learning in normal and mutant
53 Vowles, D.M., The structure and connections of the corpora pedunculata in bees and ants, Quun. J.
54 Wadepuhl, M. and Huber, F., Elicitation of singing and courtship movements by electrical stimulation
55 Wagner, S., Struktur und Funktion der corpora pedunculata einer strukturellen Hirnmutante von Drosophila
56 Weiss, M.J., Neuronal connections and the function of the corpora pedunculata in the brain of the
Sci., 259 (1971) 477-516.
Acad. Sci. U.S.A. , 71 (1974) 708-712.
Terrace (Eds.), The Biology of Learning, Dahlem Conferences, Springer, Berlin, 1984.
brain of Achaeta domesticus, Cell Tim. Res., 230 (1983) 573-586.
mushroom body cells, J . Comp. Physiol. A , 154 (1984) 71-79.
