- Email: [email protected]
Chapter 24 The development of topographically-aligned maps of visual and auditory space in the superior colliculus
M. Norita, T. Bando and B. Stein (Eds.)
Progress in Brain Researrh. Vol 112 Q 1996 Elsevier Science BV. All rights reserved.
CHAPTER 24
The development of topographically-aligned maps of visual and auditory space in the superior colliculus Andrew J. King*, Jan W.H. Schnupp, Simon Carlile, Adam L. Smith and Ian D. Thompson University Labomtory of Physiolw, Parks Rood, Ox$ord, OX1 3m, UK
The role of the superior colliculus in attending and orienting to sensory stimuli is facilitated by the presence within this midbrain nucleus of superimposed maps of different sensory modalities. We have studied the steps involved in the develop ment of topographically-aligned maps of visual and auditory space in the ferret superior colliculus. Injections of fluorescent beads into the superficial layers showed that the projection from the contralateral retina displays topographic order on the day of birth (PO). Recordings made from these layers at the time of eye opening, approximately 1 month later, revealed the presence of an adult-like map of visual space. In contrast, the auditory space map in the deeper layers emerged gradually over a much longer period of postnatal life. In adult ferrets in which one eye had been deviated laterally just before eye opening, the auditory spatial tuning of single units recorded in the contralateral superior colliculus was shifted by
a corresponding amount, so that the registration of the visual and auditory maps was maintained. Chronic application of the NMDA-receptor antagonist MK801 disrupted the normal development of the auditoly space map, but had no effect on the visual map in either juvenile or adult animals, or on the auditory map once it had matured. These findings indicate that visual cues may play an instructive role, possibly via a Hebbian mechanism of synaptic plasticity, in the development of appropriately tuned auditory responses, thereby ensuring that the neural representations of both modalities share the same coordinates. Changes observed in the auditory representation following partial lesions of the superficial layers at PO suggest that these layers may provide the source of the visual signals responsible for experience-induced plasticity in auditory spatial tuning.
Introduction
activate a common set of premotor neurons, leading to an appropriate orientation response that is directed toward the location of the sensory target. The sensory receptive fields of neurons recorded in the intermediate and deep layers of the SC can occupy a substantial fraction of a hemifield, and the size and position of the different modality receptive fields of multisensory neurons are rarely precisely the same. However, within these large receptive fields, sensory neurons in the SC tend to respond most strongly to stimuli presented at much more restricted regions of space. When these so-called best positions or best areas are used to define the neurons’ spatial preferences, the close correspondence between the different sensory representations, and particu-
The ability to re-direct attention and orientate to novel sensory stimuli requires that those stimuli are first localized. To achieve its sensorimotor function, the superior colliculus (SC) contains multiple, topographically-aligned maps of sensory space, as well as motor maps that mediate movements of the peripheral sense organs (for review see Stein and Meredith, 1993). This arrangement would appear to provide an efficient means by which any unimodal or multimodal stimulus can
*Corresponding author. Tel.: +44 1865 272523; Fax: +44 1865 272469; email:[email protected]
336
larly between the maps of visual and auditory space (Knudsen, 1982; King and Palmer, 1983; Middlebrooks and Knudsen, 1984; King and Hutchings, 1987), becomes apparent. Although they appear to share a common set of coordinates within the SC, the various sensory maps found there are constructed in different ways. The visual and somatosensory maps are derived from point-to-point projections from the retina and body surface respectively, whereas the map of auditory space is first generated within the brain by neurons that exhibit systematic variations in sensitivity to combinations of sound localization cues that result from the acoustical properties of the ears and head (King and Carlile, 1995). Because the sensory systems of different modalities initialiy represent spatial information in different coordinate frames, establishing and maintaining the alignment of the different sensory maps in the SC is not Straightforward. For example, the registration of the maps should be degraded if an animal moves its eyes or pinnae relative to the head. However, recordings in awake primates (Jay and Sparks, 1987) and cats (Hartline et al., 1995; Peck et al., 1995) have shown that the responses of many auditory neurons in the SC are altered as the direction of gaze changes, in a manner that may at least partially transform the auditory space representation from ear and head-centred coordinates into retinocentric coordinates. Setting up and maintaining the alignment of the maps during development is also problematic because of growth-related changes in the relative geometry of different sense organs. Numerous studies have shown that achieving a common topographic organization for the representation of visual and auditory space in the SC results from the developmental plasticity of auditory spatial tuning (reviewed by King and Carlile, 1995; Knudsen and Brainard, 1995). In this paper, we investigate the role of visual experience in this process. In particular, we examine the possibility that the visual map in the superficial layers provides the instructive signals for the development
of deeper layer auditory responses that are tuned to corresponding regions of space. Methods
AU procedures were carried out on ferrets that were born and reared in the laboratory animal house. Anatomical erperiments On the day of birth (PO), ferrets were anaesthetized with Saffan (Alphaxalone/Alphadolone acetate i.m.). A craniotomy was made to expose one SC, which, at birth, is not yet overlaid by the cortex, and the dura removed. Calibrated borosilicate tubing pipettes were used to make pressure injections of green or rhodamine (red) latex microspheres (hmafluor, Ny; diluted 1:lO in sterile saline) into the SC. One tracer was injected into the rostra1 part of the nucleus and the other into the caudal region. The space above the midbrain was filled with absorbable gelatine sponge (Sterispon) and the wound margins were sutured together. A local anaesthetic was applied to the scalp and, following recovery from anaesthesia, the animals were returned to their mother. Following a survival period of 24 h, the animals were re-anaesthetized and perfused transcardially with phosphate buffered saline (PBS) followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). A small incision was made into the dorsal pole of the cornea to indicate the orientation of the eye. The eyes were removed, opened up, and the retina carefully dissected away from surrounding structures. The brain was also removed, sunk in 30% sucrose in 0.1 M phosphate buffer, and sectioned parasagitally at 50 pm on a freezing microtome.
A simple outward deviation (exotropic strabismus) of the left eye was surgically induced by removal of the medial rectus muscle under halothane and nitrous oxide anaesthesia on P27-28. The eyelids were temporarily re-closed
337
were anaesthetized with either Saffan (2 ml/kg) or a mixture of Domitor (medetomidine hydrochloride, 250 pm/kg) and ketamine (60 mg/kg). Anaesthesia was then maintained by continuous i.v. infusion of these agents. All the adult ferrets used in this study were first anaesthetized with E l m implants Saffan and then paralysed with Flaxedil (gallamine triethiodide). Anaesthesia and paralysis were Full details are given in Schnupp et al. (1995). maintained with sodium pentobarbital (1 mg/kg Briefly, Elvax 40P pellets were impregnated per h) and Flaxedil(20 mg/kg per h) respectively. with tritiated( + )-10,11-dihydro-5-methyl-5H-diIn all ferrets, the trachea was cannulated, body benzo(a,d)-cyclohepten-5,10-imine maleate (3Htemperature was maintained at 39"C, and the MK801) to a final concentration of either 1 mM electrocardiogram was monitored. The electroenor 10 mM. P25-27 ferrets were anaesthetized cephalogram and end-tidal CO, were also moniwith Saffan (2 ml/kg). After making a craniotomy tored continuously where paralysis was used. A and aspirating the cortex above the midbrain, craniotomy was made above the cortex overlying 400-pm thick Elvax slices were placed on the the SC and a minimal metal headholder, which dorsal surface of the SC. The cavity was filled supported the animal from behind, was attached with Sterispon, the bone flap replaced, and the to the skull. The juvenile ferrets were not paralwound margins sutured together. The animals ysed, so eye position was stabilized with fine suwere allowed to survive until P61-70 when they tures that connected the conjunctiva to the surwere used for electrophysiological recording. Elrounding skin. vax slices containing MK801 were also implanted All recordings were carried out in an anechoic in three adult ferrets at P108-115. These animals chamber. The cortex above the midbrain was left were allowed to recover and returned to the intact, except where Elvax sheets had previously animal house for 5-6 weeks until the terminal been implanted. A flashing LED or filament bulb recording experiment was carried out on was used as the visual stimulus. The auditory P136-158. stimulus consisted of 100-ms broadband noise Superior collicuh lesions bursts delivered from a KEF "27 loudspeaker. Both visual and auditory stimuli could be preA partial lesion of the superficial layers of the SC sented from any direction with respect to the was made in Saffan-anaesthetized ferrets on the animal's head by means of a robotic hoop that day of birth. Following exposure of the midbrain, was controlled from outside the chamber. Neural the superficial layers were sectioned medio-lateractivity was recorded with a tungsten microelecally and the caudal region removed by aspiration. trode that was lowered into the brain using a The extent of the lesion varied between animals remotely-controlled, motorized microdrive. The from about 40-90% of the total area of the neural signals were filtered, amplified and digisuperficial layers. At the conclusion of the surgery tized. In most cases, single units were discrimithe animals were recovered and returned to their nated on the basis of spike amplitude alone or mothers. When they reached adulthood, the anispike shape (see Schnupp et al., 1995). In one set mals were prepared for recording. of experiments, stimulus-evoked responses were quantified by applying a fast Fourier transform to Electrophysiological recording the amplified neural signal, in order to estimate Full details are given in King and Hutchings the power spectral density in multi-unit clusters (1987) and Schnupp et al. (1995). Juvenile ferrets (see King and Carlile, 1994 for details). The magby suturing with absorbable chromic collagen, as they normally remain closed until about P32. The right eye was enucleated at the same time. The animals were returned to their home cages until they were fully grown.
338
nitude of the stimulus-evoked response was estimated for both single-unit and power spectral density measurements by subtracting the values in a spontaneous, control window from those in the response window. Electrolytic lesions were made in each electrode track to allow subsequent reconstruction of the recording sites. At the end of the recording session,the animal was deeply anaesthetized and perfused transcardialfy with PBS followed by 10% formaldehyde in PBS. The brainstem was removed, cryoprotected and sectioned on a freezing microtome. Results
Development of the visual map in the mpmjiciul layers We have examined the evolution of the visual map in the superficial layers of the SC by using fluorescent tracers to define the topographic order in the retinocollicular projection and by mapping the visual receptive fields of the neurons found there. Fig. 1 shows the pattern of retrograde labelling in a flattened whole-mount of the retina following injections of rhodamine and green latex beads into the contralateral SC at Po. The rhodamine beads, which were injected into rostro-lateral SC, were concentrated in cell bodies in the inferior temporal region of the retina. Multiple injections of green beads into caudo-medial regions of the SC resulted in labelling of ganglion cells in the superior retina, particularly on the nasal side. Although labelled neurons were found in large regions of the retina, there was very little overlap in the distribution of ganglion cells containing each tracer. This shows that the retinocollicular projection already possesses topographic order on the day of birth. Eye opening in ferrets occurs naturally at around P32. Although the optics were cloudy at tbis time, we had no difficulty in mapping the visual receptive fields of neurons in the superficial layers. The variation in the azimuthal centre of the receptive fields with recording site is shown in
Fig. 2 for data pooled from three ferrets at P33-37. The visual azimuths varied systematically from the anterior midline near the rostra1 end of the SC to 130" into the contralateral hemifield at the caudal end. The map of visual space remained unchanged in animals recorded in the second and third postnatal months and closely resembled that found in normal, adult ferrets (see Fig. 4).
Development of the auditory map in the deeper layers The auditory responses recorded in ferrets at just over one month of age were mostly rather poorly tuned for sound location and there was little indication of any topographic order in the representation (King, 1993; King and Carlile, 1995). Consequently, the close alignment of the visual and auditory maps, which is characteristic of adult animals (see Fig. 51, was not apparent in the juvenile ferrets. This is illustrated in Fig. 3 in which the preferred sound azimuths of multi-unit responses recorded in the deeper layers are plotted against the visual azimuths of superficial layer units recorded in the same vertical electrode tracks. During the second postnatal month, the auditory responses became more sharply tuned for sound location and the registration with the overlying visual map gradually emerged.
Effect of ear& eye deviation on the devebpment of the visuul and auditory map A comparison of the map of visual azimuth in the supeficial layers of the SC in normal adult ferrets and in animals reared with a lateral deviation of the contralateral eye is shown in Fig. 4. The relative magnification of this dimension of the visual field is not uniform as a slightly greater area of the SC is used for representing anterior visual space than for more peripheral receptive fields. Consequently, more of the variance in the relationship between visual receptive field position and SC recording site was accounted for by a second-order polynomial function. In the strabismic ferrets, measurements with a reversible
339
ophthalmoscope revealed that the left eye was deviated laterally by 15-20". At each recording site in the SC, the head-centred visual receptive field positions in these animals tended to be more peripheral than those mapped in the normals. The polynomial function that provided the best fit to these data was virtually identical to that fitted to the normal data, except that it was shifted laterally at its midpoint by 15", which is very close
to the mean difference in eye position between the two groups. We have previously reported that, compared to the responses in normal ferrets, the preferred sound directions of auditory units recorded in these animals were shifted laterally by a corresponding amount (King et al., 1988). We have now extended these observations by concentrating on the difference between the azimuthal coor-
-
TEMPORAL
..
1 mm
H
SUPERIOR
INFERIOR
NASAL
0
0
0
0
0
4
a
12
16
a
0
0
0
6
9
12
15 GREEN LABELLED CELLS
3
20 RHODAMINE LABELLED CELLS
Fig. 1. Flattened whole-mount of the retina showing the distribution of retrogradely-labelled ganglion cells following a single 50-111 injection of rhodamine beads into the rostro-lateral region and multiple injections of green beads (total of 75 nl) into the caudo-medial region of the contralateral SC in a PO ferret. The retina was sampled at 200-pm intervals with 75pm X 75 p m grid boxes. The open circles represent cells containing rhodamine beads, while the filled circles represent cells labelled by green beads. The number of labelled cells in the grid boxes sampled in each part of the retina is shown by the size of the circles, as indicated below the figure.
340
VISUAL MAP AT EYE OPENING 0
0.0
0.5
0
1.0
1.5
2.0
2.5
3.0
Distance from mstmlabral border (mm) Fig. 2. Representation of visual azimuth in the superficial layers of the SC in ferrets recorded at P33-37. The best azimuths of multi-unit responses are plotted against the distance of each recording site from the rostro-lateral border of the nucleus. These data are expressed in this and subsequent figures in head-centred co-ordinates. 0" lies directly in front of the animal and negative numbers denote the hemifield contralateral to the recording site.
dinates of visual and auditory units recorded in the same electrode penetrations. We have considered only those auditory single units that were tuned to single positions in space at sound levels of around 25 dB above threshold. In doing so, we have excluded a small proportion of the units recorded in both the normal and strabismic animals that were classified as either being too broadly tuned to be attributed a single best position or as having bilobed azimuth response profiles (see Schnupp et al., 1995 for details). A comparison of the visual and auditory best azimuths is illustrated in Fig. 5. The spatial representations of the two modalities were closely aligned in both normal and strabismic animals. The superficial layer visual azimuth was subtracted from the auditory azimuth of each unit recorded in the deeper layers to yield auditoryvisual misalignment values. These are plotted as histograms in Fig. 5. For both groups, the histograms appeared to be. normally distributed and clustered around 0. Following the chronic change
in eye position in the strabismic ferrets, the visual receptive fields were centred at more contralateral positions than in normal animals. Had the auditory map remained uncompensated, we would have expected to see a systematic shift in auditory-visual misalignment of about 15" toward positive values in the strabismic ferrets. Instead, we observed a mean misalignment of nearly -6", although the mean values were not significantly different in the two groups of animals (t9, = 1.51; P = 0.14). By comparing the variance of the misalignments, we found that the widths of the histograms were not significantly different either (two-tailed F test; P = 0.08). These results suggest that the map of auditory space had shifted rostrally within the SC to compensate for the displacement in the visual field representation caused by the chronic change in eye position.
Eflect of chronic MDA-receptor blockade on the visual and auditory maps Electrophysiological recordings were made from animals that had received MKBO1-Elvax implants only if it was established during the surgical preparation on the day of recording that the implant still covered the whole of at least one SC. Recordings were commenced unilaterally within an hour of removing the Elvax. Compared to control animals, we noted no discernible change in the visual multi-unit activity recorded in the superficial layers of the SC. The topographic organization of the representation of visual azimuth is shown for adult and juvenile ferrets in Fig. 6. The best azimuths of the visual receptive fields varied systematically between the rostral and caudal ends of the nucleus. The visual maps in both adult and juvenile ferrets that had received MK801-Eh implants closely resembled those found in unoperated, normal control animals recorded at corresponding ages. In contrast to the normal visual map in the superficial layers, we found that chronic release of the NMDA receptor antagonist MK801 did
341
affect the auditory responses in the deeper layers. In all animals treated with Elvax containing MK801, the proportion of auditory units tuned to single regions of space was significantly reduced compared to unoperated, age-matched animals or to animals that had received Elvax implants containing dimethyl sulfoxide (DMSO) as drug-free controls (Schnupp et al., 1995). We observed an age-dependent effect on the topography of the auditory representation. In the adult M a 0 1 group, plotting the azimuthal best positions of these tuned units against the visual best azimuths of the superficial layer responses revealed a close alignment between the two maps, as further illustrated by a narrow auditory-visual misalignment histogram (Fig. 7). The spread (variance) in misalignment values was not significantly different from that found in either normal, unoperated adults (two-tailed F test, P = 0.48) or drugfree, adult controls (P = 0.84). In contrast, the best azimuths of the tuned units recorded in the SC of juvenile ferrets reared with MK801-Elvax implants were poorly correlated with the histolog-
-
P33 P37 180
140
0
ical coordinates of their recording sites. Because the visual map in these animals was essentially normal (Fig. 61, a comparison of visual and auditory best azimuths showed that there was considerable scatter in the relationship between the two representations (Fig. 7). The variance in the auditory-visual misalignment values for these animals was significantly greater than that found in both juvenile, unoperated (P< 0.02) and drugfree (P< 0.01) control groups. Effect of partial lesions of the superficial layers on the development of the ukmal and auditory maps
Visual responses recorded from the remaining rostra1 region of the superficial layers in the SC of adult ferrets in which the caudal pole had been aspirated on PO appeared to have normal receptive fields. These units were tuned to azimuthal locations that covered a restricted region of the anterior hemifield and fell either within or very close to the normal range of visual azimuths represented in this part of the nucleus in adult
I
-
-
P41 - P a
P53 P62
P48 P52
180
1 0
180
140
140
140
100
100
60
60
100 60
0
s__tt
0
20 I
:
,c-:
:
:
t
20
-20 -20 so
60
so
-im -140
-140
-180
-10
-1m
0
0
0
0
% S 8 8 8 W ,
I
,
Visual best 8zimuth [dog.]
Visual best azimuth [deg.]
Fig. 3. Development of topographically-aligned visual and auditory representations in the SC. These data were obtained from juvenile ferrets at the postnatal ages indicated at the top of the figure. For each vertical electrode penetration, the visual best azimuth of multi-unit activity recorded in the superficial layers is plotted against the auditory best azimuth of multi-unit responses recorded in the deeper layers. These estimates of spatial selectivity are based on power spectral density measurements, whereas the number of sound-evoked spikes from single units is used in the remainder of the figures. We have previously shown that the two recording techniques produce equivalent results (King and Carlile, 1994).
342 VISUAL TOPOORAPHY IN THE SUPERRclAL LAYERS W l
that measured for the equivalent rostral region of the SC in normal animals (P= 0.13). We determined the extent of the lesioned area both histologically and by the absence of characteristically strong visual drive in the caudal region of the SC. Most of the response properties of the auditory units recorded beneath this lesioned area appeared to be normal and the majority were tuned to single spatial regions. Despite the absence of visual activity in the superficial layers, approximately one third of these deep layer auditory units were also visually responsive (presumably as a result of descending cortical inputs). The azimuthal best positions of the units are plotted against the rostro-caudal coordinates of their recording sites in Fig. 9. The great majority were tuned to sound locations within the contralateral hemifield. But whereas many had best azimuths that fell within the normal range, there were also numerous units with best azimuths not normally Seen in the corresponding regions of the SC. Because of the lack of a visual map in the superficial layers against which to compare these preferred sound directions, we calculated the angular difference between the observed auditory best azimuths and the value predicted for each recording site from the polynomial function that provided the best fit to the representation of sound azimuth in normal, adult ferrets. The variance of the differences between observed and predicted best azimuths for the auditory units for which the superficial layers were missing was significantly greater than the variance associated with either the rostral region with intact superficial layers in the same animals (P4 0.01) or the normal control group (P4c 0.01). Interestingly, the auditory best azimuths appeared to be equally scattered in both unimodal and bimodal (visual-auditory)neurons.
1
-la -160
Fig. 4. Effect of lateral deviation of the eye on the representation of visual azimuth in the superficial layers of the SC. The filled circles represent data from normal, adult ferrets, while the open circles indicate the responses recorded in adult ferrets in which an exotropic strabismus had been surgically induced in the contralateral eye on €27-28. The lines correspond to the second-order polynomial functions fitted to each set of data. The visual receptive fields of the SC neurons remained aligned with the retina; when expressed in headcentred co-ordinates they became displaced relative to the normal group by an amount equivalent to the lateral shift in eye position.
animals (Fig. 8A). The extent of the visual field representation was therefore determined by the size of the lesion. The electrode penetrations from which these units were recorded were then extended into the deeper layers. The auditory units we encountered there were tuned to very similar anterior locations. Again their best azimuths fell within the normal range of values associated with the rostral part of the SC (Fig. 8B). Some of these auditory units were also visually responsive. The auditory best positions of both unimodal and bimodal neurons were in close correspondence with the receptive fields of the visual units recorded in the overlying superficial layers, and the variance of the auditory-visual misalignments did not differ significantly from
Discussion
Maps of visual and auditory space in the SC share common co-ordinates. As well as providing a ba-
sis for delivering modality-independent spatial
343 STRABISMUS
NORMAL 40 20
I
40
n
20
-20 -40
'8; *
15 10 5 0
Fig. 5. Effect of lateral deviation of the eye on the registration of the visual and auditory representations in the SC. The data on the left are from normal, adult ferrets and those on the right are from adult ferrets in which an exotropic strabismus had been surgically induced in the contralateral eye on P27-28. The histograms show the angular difference between the best azimuths of tuned auditory units recorded in the deeper layers and those of visual multi-unit responses recorded in the superficial layers of the same vertical electrode penetrations. Neither the mean nor the variance of the auditory-visual mis-alignments differ significantly between the two groups. Adapted from King et al. (1988).
signals to the neural pathways that control orientation behaviour, the alignment of the visual and auditory representations allows multisensory cues to be integrated in ways that may improve the accuracy of orienting behaviour (Stein and Meredith, 1993). In this report, we have examined the developmental steps that lead to the formation of spatially matched visual and auditory maps in the ferret SC. We have shown that the visual map in the superficial layers matures some weeks before the higher-order map of auditory space in the deeper layers, and that the activity-dependent process of auditory map development is regulated by visual signals that may arise from the superficial layers.
Development of the visual map We found that the retinocollicular projection has topographic order at PO. Changes in the number
of retinal ganglion cells (Henderson et al., 1988; Cucchiaro, 1991; Thompson and Morgan, 1993) and in the specificity of this projection (Snider and Chalupa, 1993) continue to occur until the end of the first postnatal week. Nevertheless, by the time of eye opening at around P32, the retinocollicular pathway should be topographically mature. This is supported by the results of our recordings from juvenile ferrets within a few days of the eyes opening, which revealed the presence of an adult-like representation of visual space in the superficial layers. In adult animals in which an exotropic strabismus of the contralateral eye had been induced just before normal eye opening, we found that the visual receptive fields were centred on locations that were, on average, displaced laterally by 15" compared to the normal map. We can therefore conclude that the retinocollicular projection was unaltered by this
344
Chronic MK-801 in adulthood 40 T
P
E
%
.E
I
20
-40
;-60 Q
-
0 0
-20
-100-120 -140
F
P
'0
Y
.
,
.
,
.
,
.
0 -
-
I
-60-
e?
-40-
6% 'ed
3
5 I
Distance from rostro-lateral border [mm]
0
ee8 8
! m -80-
0
I
-
-20
-
-
-
%
.g
-
3
(b
20 0
-
-80-
5
40
-
f (0
Chronic MK-801 in infancy
-
-1401
e
e o
-100-120
b
.
I
.
,
.
,
,
,
,
1
Distance from rostro-lateral border [mm]
Fig. 6. The representation of visual azimuth in the superficial layers of adult (left) and juvenile (P61-70, right) ferrets. The azimuthal best positions of multi-unit responses are plotted against the distance of each recording site from the rostro-lateral border of the SC. The filled circles represent data from animals in which MK801-Ehrax implants had been placed on the dorsal surface of the SC 5-6 weeks before recording. The open circles represent the data obtained from normal, unoperated animals. Adapted from Schnupp et al. (1995).
procedure, as the lateral shift in the visual field representation matched the change in eye position. We were also unable to detect any developmental change in the visual map in the superficial layers following chronic treatment with the NMDA receptor antagonist -1. Other reports have indicated a role for NMDA receptors in the refinement of retinocollicular topography (Cline and Constantine-Paton, 1989; Simon et al., 1992). However, in contrast to these studies, we implanted the Elvax on the SC after the early postnatal changes in the projection had been completed. The early maturation of the retinocollicular pathway in the ferret may also explain why the visual map in the SC did not appear to become reorganized following a partial ablation of the superficial layers on the day of birth. In neonatal hamsters, partial SC lesions can result in compression of the entire contralateral visual field onto the remaining portion of the superficial lay-
ers (see Finlay, this volume). However, we found that, in ferrets, visual units recorded in this region were tuned to the same azimuthal locations as in n o d animals. The visual map was therefore abruptly truncated at the caudal end of the intact superficial layers. Whereas the ferret retinocollicular projection is topographically organized at PO, the hamster pathway is still disordered at this age (Thompson and Cordery, 1994). The capacity for reorganization of the pattern of connections from the retina to the SC following manipulation of neural activity or reduction in target volume therefore appears to be limited by the developmental maturity of the projection at the time of the surgery.
V%ualmperience induedphticity in the developing m a p of auditory space
In contrast to visual responses in the superficial layers, the auditory responses recorded in the
345
Chronic MK-801 treatment in adulthood
20
Chronic MK-801 treatment in infancy
+
Visual best azimuth [deg.]
8
-s 6 5
4
1 2 0
auditory-visual misalignment [deg.]
audiIaryvisual mi8alignmenl [deg.]
Fig. 7. Relationship between the representations of visual azimuth in the superficial layers and auditory azimuth in the deeper layers of the SC of adult (left) and juvenile (P61-70, right) ferrets that had received MK801-Elvax implants 5-6 weeks before recording. The histograms plot the angular difference between the best azimuths of auditory and visual units recorded in each electrode track. The histogram is significantly wider in the juvenile animals. Adapted from Schnupp et al. (1995).
intermediate and deep layers were initially poorly tuned for sound location and exhibited very little order in the distribution of azimuthal locations to which they responded most strongly. The topographic order in the auditory representation, and the alignment with the overlying visual map, gradually emerged over a course of several weeks following the onset of hearing. A similar result, albeit with a different time course, has been reported for the guinea pig (Withington-Wray et al., 1990). This delayed appearance of the auditory space map can be understood in terms of the way in which it is constructed. Whereas the visual map results from a topographic projection from the retina, an equivalent projection from the
cochlea produces a neural map of sound frequency, not sound source position. The topographic representation of auditory space must therefore be computed within the brain by tuning the neurons to different localization cue values. In mammals, SC neurons are sensitive to monaural spectral cues and interaural level differences (Hirsch et al., 1985; Wise and Irvine, 1985; Middlebrooks, 1987; Carlile and King, 1994; King et al., 1994). We have shown that, in the ferret, the spatial pattern of these cues continues to change for a period of several weeks after the onset of hearing (King and Carlile, 1995). Thus, the acoustical basis for the map of auditory space is continually changing during much of the period over
346
Fig. 8. (A) Visual representation in the remaining superficial layers of the SC in adult ferrets following aspiration of the caudal portion of these layers at PO. (B) Auditory representation in the deeper layers of the same vertical electrode tracks. The azimuthal best positions are plotted against the distance of their recording sites from the rostro-lateral border of the SC. The open triangles in (B) represent the auditory best azimuths of units that were also visually responsive, whereas the filled circles indicate the values obtained from unimodal auditory neurons. The two lines in each panel represent two standard deviations on either side of the polynomial function that provides the best fit to the visual and auditory maps in normal ferrets. The hatched area of the SC in the parasagittal view on the right indicates the aspirated region of the superficial layers.
which the topographic order in the representation emerges. The correlation between the developmental time course for the localization cues and the map of auditory space suggests that the establishment of the topographic alignment of the visual and auditory maps may, at least in part, be limited by the maturation of these cues, which, in turn, depends on growth of the head and outer ears. However, sensory experience also plays a crucial role in refining the auditory map, so that a close correspondence with the visual map is achieved in spite of individual variations in the auditory localization cue values and in the relative position of
the eyes and em. For example, distorting the binaural localization cues by plugging one ear in infancy leads to a compensatory adjustment in the auditory map so that it remains precisely aligned with the visual map (budsen, 1985; King et al., 1988). That visual cues may play a dominant role in aligning the two maps is suggested by our finding that displacement of the visual receptive fields relative to the head, caused by lateral deviation of the contralateral eye in infancy, is followed by an equivalent change in the auditory representation. Rotation of the eye disrupts the development of the auditory map, but to a much lesser extent if the animals are visually deprived
347
degraded to some extent (Knudsen et al., 1991; Withington, 1992; King and Carlile, 1993). Taken together, these observations suggest that a crude map of auditory space can develop in the absence of visual cues. However, when available, visual information can influence the development of the auditory map, and adjust its topographic order to match that of the visual map in the SC.
A A
Basisfor visual instruction of the developing auditory space map
U
8 e
n
.-8
'2!
-20
-60
-140 -180, 0
0 v
s
0.5
I
1
.
I
1.5
,
I
2
.
1
2.5
Distance from rostro-lateralborder [mm] Fig. 9. Auditory representation in the adult ferret SC following aspiration of the caudal region of the superficial layers at PO. The auditory best positions of single units recorded in the intermediate and deep layers are plotted against the distance of their recording sites from the rostro-lateral border of the SC for those electrode tracks where the superficial layers were missing. The open triangles represent the auditory best azimuths of units that were also visually responsive, whereas the filled circles indicate the values obtained from unimodal auditory neurons. The two lines each represent two standard deviations on either side of the polynomial function that provides the best fit to the map of sound azimuth in normal ferrets.
at the same time, suggesting that the shift in auditory spatial tuning results from the displacement of the visual receptive fields rather than the abnormal eye position (King and Carlile, 1995). Optical displacement of the visual field in barn owls reared with prisms mounted in front of their eyes also leads to a corresponding change in the auditory space map (Knudsen and Brainard, 1991). On the other hand, if the visual image is blurred (by eyelid suture), the auditory representation is
A Hebbian mechanism of synaptic plasticity could provide the basis for establishing the alignment of the visual and auditory maps in the SC. Support for this possibility is provided by our fjnding that chronic application of MK801 during the period when the topographic order in the auditory representation normally emerges, results in a degraded map and, consequently, a much poorer alignment with the unchanged visual map. On the other hand, the registration of the maps was unaltered by treatment of adult animals with MK801-Elvax implants. The activation of NMDA-type glutamate receptors may provide a means of detecting temporal correlations among simultaneously active synaptic inputs (Fox and Daw, 19931, and thereby act as a trigger for the activity-dependent steps that lead to the emergence of a precise and unambiguous representation of auditory space in the SC. Feldman et al. (1995) have also reported that NMDA-receptor mediated currents may be particularly involved in expressing the visuallyguided changes in the auditory receptive fields of neurons in the barn owl optic tectum. The involvement of these receptors in the SC is consistent with a Hebbian mechanism in which more accurate visual signals could lead to the selective strengthening of auditory inputs that convey information from the same region of space. Source of instructive visual s i p &
Most auditory neurons in the barn owl optic tectum are also visually responsive (Knudsen,
348
1982). Optical displacement of the visual field has been reported to shift the auditory receptive fields of not only these units but also those of apparently unimodal units recorded in the external nucleus of the inferior colliculus, which projects topographically to the tectum (Brainard and Knudsen, 1993). The only equivalent study in mammals has shown that dark rearing disrupts the development of auditory spatial selectivity in the guinea pig SC, but has no effect on the ICX (Binns et al., 1995). By using [3H]MK801,we were able to determine the area over which this NMDA-receptor antagonist is released (Schnupp et al., 1995). Our results suggest that MK801 released from the Elvax penetrated at least as far as 800 p m below the SC surface and therefore reached both the superficial and intermediate layers of the nucleus. Although further experiments will be required to verify this, we currently believe that the changes that we have observed in the developing auditory map are most likely to be localized to the SC itself, rather than the sources of afferent input. Many auditory neurons in the SC also receive visual inputs, which most likely originate from extrastriate visual areas of the cortex (reviewed in Stein and Meredith, 1993). However, their visual receptive fields tend to be much larger than those of neurons recorded in the superficial layers. Moreover, the proportion of deep SC neurons that receive multisensory inputs is much smaller in early postnatal life (Wallace et al., 1993). If the site of developmental visual-auditory interaction does reside within the SC, an alternative source of guiding visual input would appear to be needed. Although the superficial and deeper layers of the SC have been regarded as anatomically and physiologically distinct regions (Casagrande et al., 1972; Edwards et al., 19791, connections are now known to exist between them (e.g. Mooney et al. 1984, 1992; Behan and Appell, 1992). Our finding that the auditory representation was disrupted only in the region beneath the ablated superficial layers is consistent with the possibility that those layers may provide local signals that calibrate the development of the underlying audi-
tory neurons. That bimodal neurons in the region of the SC where the superficial layers were missing also tended to exhibit aberrant auditory best positions would seem to indicate that other visual afferents to the SC were unable to rescue the spatial tuning of the auditory responses. At first sight, the lack of a measurable visual input to all acoustically-responsive neurons, particularly during early life, would appear to question the significance of synchronized activity in the superficial layers for the development of auditory spatial tuning. However, visual stimuli can alter the responses of these neurons to sound, even if, by themselves, they do not evoke a change in firing rate (King and Palmer, 1985). It may therefore be necessary to employ more sensitive cross-correlation techniques to reveal the influence of activity in the superficial layers on the responses of neurons in the deeper layers. Acknowledgements
We are grateful to Pat Cordery for excellent technical assistance and to the Wellcome Trust for financial support. Andrew King is a Wellcome Senior Research Fellow (grant number 031316 9O/Z) and Jan Schnupp is a Wellcome Prize Student (039456 93/21. Simon Carlile was supported by a Beit Memorial Fellowship and Adam Smith by a Wellcome Vision Research Training fellowship (034242 .91/Z).
-
-
References Behan, M. and Appell, P.P. (1992) Intrinsic circuitry in the cat superior colliculus: projections from the superficial layers. J. C o w . NeuroL, 315: 230-243. Binns, ICE., Withington, D.J. and Keating, M.J. (1995) The developmental emergence of the representation of auditory azimuth in the external nucleus of the inferior colliculw. the effects of visual and auditory deprivation. Deu. Bmin Rs., 85: 14-24. Brainard, M.S. and Knudsen, E.I. (1993) Experience-dependent plasticity in the inferior colliculus: a site for visual calibration of the neural representation of auditory space in the barn owl. J. N e m c i , 13: 4589-4608. Carlile, S. and King, A.J. (1994) Monaural and binaural spectrum level cues in the ferret: acoustics and the neural
349
representation of auditory space. J. Neurophyswl., 71: 785-801. Casagrande, V.A., Harting, J.K, Hall, W.C., Diamond, I.T. and Martin, G.F. (1972) Superior colliculus of the tree shrew: a structural and functional subdivision into superficial and deep layers. Science, 177: 444-447. Cline, H.T. and Constantine-Paton, M. (1989) NMDA receptor antagonists disrupt the retinotectal topographic map. Neuron, 3: 413-426. Cucchiaro, J.B. (1991) Early development of the retinal line of decussation in normal and albino ferrets. J. Comp. Neurol., 312: 193-206. Edwards, S.B., Ginsburgh, C.L., Henkel, C.K. and Stein, B.E. (1979) Sources of subcortical projections to the superior colliculus in the cat. J. Comp. Neurol., 184: 309-30. Feldman, D.E., Brainard, M.S. and Knudsen, E.I. (1996) Newly learned auditory responses mediated by NMDA receptors in the owl inferior colliculus. Science, 270 525-528. Fox, K. and Daw, N.W. (1993) Do NMDA receptors have a critical function in visual cortical plasticity? Trends Neurosci.,1 6 116-122. Hartline, P.H., Pandey Vimal, R.L., King, A.J., Kurylo, D.D. and Northmore, D.P.M. (1995) Effects of eye position on auditory localization and neural representation of space in superior colliculus of cats. @. Brain Res., 104: 402-408. Henderson, Z., Finlay, B.L. and Wikler, K.C. (1988) Develop ment of ganglion cell topography in ferret retina. J. Neurosci, 8: 1194-1205. Hirsch, J.A., Chan, J.C.K. and Yin, T.C.T. (1985) Responses of neurons in the cat’s superior colliculus to acoustic stimuli. I. Monaural and binaural response properties. J. Neurophysiol., 5 3 726-745. Jay, M.F. and Sparks, D.L. (1987) Sensorimotor integration in the primate superior colliculus. 11. Coordinates of auditory signals. J. NeurophysioL, 5 7 35-55. King, A.J. (1993) The Wellcome Prize Lecture. A map of auditory space in the mammalian brain: neural computation and development. Ekp. Physiol., 7 8 559-590. King, A.J. and Carlile, S. (1993) Changes induced in the representation of auditory space in the superior colliculus by rearing ferrets with binocular eyelid suture. @. Brain Res., 94: 444-455. King, A.J. and Carlile, S. (1994) Responses of neurons in the ferret superior colliculus to the spatial location of tonal stimuli. Hear. Res., 81: 137-149. King, A.J. and Carlile, S. (1995) Neural coding for auditory space. In: M. S. Gazzaniga (Ed.), nte Cognitiue Neurosciences, MIT Press, pp. 279-293. King, A.J. and Hutchings, M.E. (1987) Spatial response properties of acoustically responsive neurons in the superior colliculus of the ferret: a map of auditory space. J. Neurophysiol.,57: 596-624. King, A.J. and Palmer, A.R. (1983) Cells responsive to freefield auditory stimuli in guinea-pig superior colliculus: dis-
tribution and response properties. J. Physiol. Lond., 342: 361-381. King, A.J. and Palmer, A.R. (1985) Integration of visual and auditory information in bimodal neurones in the guinea-pig superior colliculus. Exp. Brain Res., 60: 492-500. King, A.J., Hutchings, M.E., Moore, D.R. and Blakemore, C. (1988) Developmental plasticity in the visual and auditory representations in the mammalian superior colliculus. Nature, 332:73-76. King, A.J., Moore, D.R. and Hutchings, M.E. (1994) T o p e graphic representation of auditory space in the superior colliculus of adult ferrets after monaural deafening in infancy. J. Neurophysiol., 71: 182-194. Knudsen, E.I. (1982) Auditory and visual maps of space in the optic tectum of the owl. J. Neurosci., 2 1177-1194. Knudsen, E.I. (1985) Experience alters the spatial tuning of auditory units in the optic tectum during a sensitive period in the barn owl. J. Neurosci., 5: 3094-3109. Knudsen, E.I. and Brainard, M.S. (1991) Visual instruction of the neural map of auditory space in the developing optic tectum. Science, 253: 85-87. Knudsen, E.I. and Brainard, M.S. (1995) Creating a unified representation of visual and auditory space in the brain. Annu. Rev. Neurosci, 18: 19-43. Knudsen, E.I., Esterly, S.D. and du Lac, S. (1991) Stretched and upside-down maps of auditory space in the optic tecturn of blind-reared owls; acoustic basis and behavioral correlates. J. Neurosci., 11: 1727-1747. Middlebrooks, J.C. (1987) Binaural mechanisms of spatial tuning in the cat’s superior colliculus distinguished using monaural occlusion. J. Neurophysiol.,57: 688-701. Middlebrooks, J.C. and Knudsen, E.I. (1984) A neural code for auditory space in the cat’s superior colliculus. J. Neurosci., 4: 2621-2634. Mooney, R.D., Bradley, G.K., Jacquin, M.F. and Rhoades, R.W. (1984) Dendrites of the deep layer, somatosensory superior colliculus neurons extend into the superficial laminae. Brain Res., 324: 361-365. Mooney, R.D., Huang, X. and Rhoades, R.W. (1992) Functional influence of interlaminar connections in the hamster’s superior colliculus. J. Neurosci., 1 2 2417-2432. Peck, C.K, Baro, J.A. and Warder, S.M. (1995) Effects of eye position on saccadic eye movements and on the neuronal response to auditory and visual stimuli in cat superior colliculus. Exp. Brain Res., 103: 227-242. Schnupp, J.W.H., King, A.J., Smith, A.L. and Thompson, I.D. (1995) NMDA-receptor antagonists disrupt the formation of the auditory space map in the mammalian superior colliculus. J. Neurosci., 15: 1516-1531. Simon, D.K., Prusky, G.T., O’Leary, D.D.M. and Constantine-Paton, M. (1992) N-methyl-D-aspartate receptor antagonists disrupt the formation of a mammalian NatL Acad. Sci. USA, 89: 10593-10597. neural map. h.
350
Snider, C.J. and Chalupa, L.M. (1993) Specificity of the retinocollicular pathway in developing cat and ferret. Soc. Neurosci. Abstr., 1 9 454. Stein, B.E. and Meredith, M.A. (1993) The Merging of the Senses, MIT Press. Cambridge, MA. Thompson, 1.D. and Cordery, P.M. (1994) The development of the retino-collicular map in the Syrian hamster. Soc. Neurosci. Abstr., 2 0 1704. Thompson, I.D. and Morgan, J.E. (1993) The development of retinal ganglion cell decussation patterns in postnatal pigmented and albino ferrets. Eur. J. Neurosci., 5: 341-356. Wallace, M.T., Meredith, M.A. and Stein, B.E. (1993) Development of multisensory integration in cat superior colliculus. Soc. Neurosci. Abstr., 1 9 240.
Wise, L.Z. and Irvine, D.R.F. (1985) Topographic organization of interaural intensity difference sensitivity in deep layers of cat superior colliculus: implications for auditory spatial representation. J. NeumphysioL, 5 4 185-211. Withington, D.J. (1992) The effect of binocular lid suture on auditory responses in the guinea-pig superior colliculus. Nemsci. Lett., 136 153-6. Withington-Wray, D.J., Binns, KE. and Keating, MJ. (1990) The developmental emergence of a map of auditory space in the superior colliculus of the guinea pig. Deu. Bruin Res., 51: 225-36.
Progress in Brain Researrh. Vol 112 Q 1996 Elsevier Science BV. All rights reserved.
CHAPTER 24
The development of topographically-aligned maps of visual and auditory space in the superior colliculus Andrew J. King*, Jan W.H. Schnupp, Simon Carlile, Adam L. Smith and Ian D. Thompson University Labomtory of Physiolw, Parks Rood, Ox$ord, OX1 3m, UK
The role of the superior colliculus in attending and orienting to sensory stimuli is facilitated by the presence within this midbrain nucleus of superimposed maps of different sensory modalities. We have studied the steps involved in the develop ment of topographically-aligned maps of visual and auditory space in the ferret superior colliculus. Injections of fluorescent beads into the superficial layers showed that the projection from the contralateral retina displays topographic order on the day of birth (PO). Recordings made from these layers at the time of eye opening, approximately 1 month later, revealed the presence of an adult-like map of visual space. In contrast, the auditory space map in the deeper layers emerged gradually over a much longer period of postnatal life. In adult ferrets in which one eye had been deviated laterally just before eye opening, the auditory spatial tuning of single units recorded in the contralateral superior colliculus was shifted by
a corresponding amount, so that the registration of the visual and auditory maps was maintained. Chronic application of the NMDA-receptor antagonist MK801 disrupted the normal development of the auditoly space map, but had no effect on the visual map in either juvenile or adult animals, or on the auditory map once it had matured. These findings indicate that visual cues may play an instructive role, possibly via a Hebbian mechanism of synaptic plasticity, in the development of appropriately tuned auditory responses, thereby ensuring that the neural representations of both modalities share the same coordinates. Changes observed in the auditory representation following partial lesions of the superficial layers at PO suggest that these layers may provide the source of the visual signals responsible for experience-induced plasticity in auditory spatial tuning.
Introduction
activate a common set of premotor neurons, leading to an appropriate orientation response that is directed toward the location of the sensory target. The sensory receptive fields of neurons recorded in the intermediate and deep layers of the SC can occupy a substantial fraction of a hemifield, and the size and position of the different modality receptive fields of multisensory neurons are rarely precisely the same. However, within these large receptive fields, sensory neurons in the SC tend to respond most strongly to stimuli presented at much more restricted regions of space. When these so-called best positions or best areas are used to define the neurons’ spatial preferences, the close correspondence between the different sensory representations, and particu-
The ability to re-direct attention and orientate to novel sensory stimuli requires that those stimuli are first localized. To achieve its sensorimotor function, the superior colliculus (SC) contains multiple, topographically-aligned maps of sensory space, as well as motor maps that mediate movements of the peripheral sense organs (for review see Stein and Meredith, 1993). This arrangement would appear to provide an efficient means by which any unimodal or multimodal stimulus can
*Corresponding author. Tel.: +44 1865 272523; Fax: +44 1865 272469; email:[email protected]
336
larly between the maps of visual and auditory space (Knudsen, 1982; King and Palmer, 1983; Middlebrooks and Knudsen, 1984; King and Hutchings, 1987), becomes apparent. Although they appear to share a common set of coordinates within the SC, the various sensory maps found there are constructed in different ways. The visual and somatosensory maps are derived from point-to-point projections from the retina and body surface respectively, whereas the map of auditory space is first generated within the brain by neurons that exhibit systematic variations in sensitivity to combinations of sound localization cues that result from the acoustical properties of the ears and head (King and Carlile, 1995). Because the sensory systems of different modalities initialiy represent spatial information in different coordinate frames, establishing and maintaining the alignment of the different sensory maps in the SC is not Straightforward. For example, the registration of the maps should be degraded if an animal moves its eyes or pinnae relative to the head. However, recordings in awake primates (Jay and Sparks, 1987) and cats (Hartline et al., 1995; Peck et al., 1995) have shown that the responses of many auditory neurons in the SC are altered as the direction of gaze changes, in a manner that may at least partially transform the auditory space representation from ear and head-centred coordinates into retinocentric coordinates. Setting up and maintaining the alignment of the maps during development is also problematic because of growth-related changes in the relative geometry of different sense organs. Numerous studies have shown that achieving a common topographic organization for the representation of visual and auditory space in the SC results from the developmental plasticity of auditory spatial tuning (reviewed by King and Carlile, 1995; Knudsen and Brainard, 1995). In this paper, we investigate the role of visual experience in this process. In particular, we examine the possibility that the visual map in the superficial layers provides the instructive signals for the development
of deeper layer auditory responses that are tuned to corresponding regions of space. Methods
AU procedures were carried out on ferrets that were born and reared in the laboratory animal house. Anatomical erperiments On the day of birth (PO), ferrets were anaesthetized with Saffan (Alphaxalone/Alphadolone acetate i.m.). A craniotomy was made to expose one SC, which, at birth, is not yet overlaid by the cortex, and the dura removed. Calibrated borosilicate tubing pipettes were used to make pressure injections of green or rhodamine (red) latex microspheres (hmafluor, Ny; diluted 1:lO in sterile saline) into the SC. One tracer was injected into the rostra1 part of the nucleus and the other into the caudal region. The space above the midbrain was filled with absorbable gelatine sponge (Sterispon) and the wound margins were sutured together. A local anaesthetic was applied to the scalp and, following recovery from anaesthesia, the animals were returned to their mother. Following a survival period of 24 h, the animals were re-anaesthetized and perfused transcardially with phosphate buffered saline (PBS) followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). A small incision was made into the dorsal pole of the cornea to indicate the orientation of the eye. The eyes were removed, opened up, and the retina carefully dissected away from surrounding structures. The brain was also removed, sunk in 30% sucrose in 0.1 M phosphate buffer, and sectioned parasagitally at 50 pm on a freezing microtome.
A simple outward deviation (exotropic strabismus) of the left eye was surgically induced by removal of the medial rectus muscle under halothane and nitrous oxide anaesthesia on P27-28. The eyelids were temporarily re-closed
337
were anaesthetized with either Saffan (2 ml/kg) or a mixture of Domitor (medetomidine hydrochloride, 250 pm/kg) and ketamine (60 mg/kg). Anaesthesia was then maintained by continuous i.v. infusion of these agents. All the adult ferrets used in this study were first anaesthetized with E l m implants Saffan and then paralysed with Flaxedil (gallamine triethiodide). Anaesthesia and paralysis were Full details are given in Schnupp et al. (1995). maintained with sodium pentobarbital (1 mg/kg Briefly, Elvax 40P pellets were impregnated per h) and Flaxedil(20 mg/kg per h) respectively. with tritiated( + )-10,11-dihydro-5-methyl-5H-diIn all ferrets, the trachea was cannulated, body benzo(a,d)-cyclohepten-5,10-imine maleate (3Htemperature was maintained at 39"C, and the MK801) to a final concentration of either 1 mM electrocardiogram was monitored. The electroenor 10 mM. P25-27 ferrets were anaesthetized cephalogram and end-tidal CO, were also moniwith Saffan (2 ml/kg). After making a craniotomy tored continuously where paralysis was used. A and aspirating the cortex above the midbrain, craniotomy was made above the cortex overlying 400-pm thick Elvax slices were placed on the the SC and a minimal metal headholder, which dorsal surface of the SC. The cavity was filled supported the animal from behind, was attached with Sterispon, the bone flap replaced, and the to the skull. The juvenile ferrets were not paralwound margins sutured together. The animals ysed, so eye position was stabilized with fine suwere allowed to survive until P61-70 when they tures that connected the conjunctiva to the surwere used for electrophysiological recording. Elrounding skin. vax slices containing MK801 were also implanted All recordings were carried out in an anechoic in three adult ferrets at P108-115. These animals chamber. The cortex above the midbrain was left were allowed to recover and returned to the intact, except where Elvax sheets had previously animal house for 5-6 weeks until the terminal been implanted. A flashing LED or filament bulb recording experiment was carried out on was used as the visual stimulus. The auditory P136-158. stimulus consisted of 100-ms broadband noise Superior collicuh lesions bursts delivered from a KEF "27 loudspeaker. Both visual and auditory stimuli could be preA partial lesion of the superficial layers of the SC sented from any direction with respect to the was made in Saffan-anaesthetized ferrets on the animal's head by means of a robotic hoop that day of birth. Following exposure of the midbrain, was controlled from outside the chamber. Neural the superficial layers were sectioned medio-lateractivity was recorded with a tungsten microelecally and the caudal region removed by aspiration. trode that was lowered into the brain using a The extent of the lesion varied between animals remotely-controlled, motorized microdrive. The from about 40-90% of the total area of the neural signals were filtered, amplified and digisuperficial layers. At the conclusion of the surgery tized. In most cases, single units were discrimithe animals were recovered and returned to their nated on the basis of spike amplitude alone or mothers. When they reached adulthood, the anispike shape (see Schnupp et al., 1995). In one set mals were prepared for recording. of experiments, stimulus-evoked responses were quantified by applying a fast Fourier transform to Electrophysiological recording the amplified neural signal, in order to estimate Full details are given in King and Hutchings the power spectral density in multi-unit clusters (1987) and Schnupp et al. (1995). Juvenile ferrets (see King and Carlile, 1994 for details). The magby suturing with absorbable chromic collagen, as they normally remain closed until about P32. The right eye was enucleated at the same time. The animals were returned to their home cages until they were fully grown.
338
nitude of the stimulus-evoked response was estimated for both single-unit and power spectral density measurements by subtracting the values in a spontaneous, control window from those in the response window. Electrolytic lesions were made in each electrode track to allow subsequent reconstruction of the recording sites. At the end of the recording session,the animal was deeply anaesthetized and perfused transcardialfy with PBS followed by 10% formaldehyde in PBS. The brainstem was removed, cryoprotected and sectioned on a freezing microtome. Results
Development of the visual map in the mpmjiciul layers We have examined the evolution of the visual map in the superficial layers of the SC by using fluorescent tracers to define the topographic order in the retinocollicular projection and by mapping the visual receptive fields of the neurons found there. Fig. 1 shows the pattern of retrograde labelling in a flattened whole-mount of the retina following injections of rhodamine and green latex beads into the contralateral SC at Po. The rhodamine beads, which were injected into rostro-lateral SC, were concentrated in cell bodies in the inferior temporal region of the retina. Multiple injections of green beads into caudo-medial regions of the SC resulted in labelling of ganglion cells in the superior retina, particularly on the nasal side. Although labelled neurons were found in large regions of the retina, there was very little overlap in the distribution of ganglion cells containing each tracer. This shows that the retinocollicular projection already possesses topographic order on the day of birth. Eye opening in ferrets occurs naturally at around P32. Although the optics were cloudy at tbis time, we had no difficulty in mapping the visual receptive fields of neurons in the superficial layers. The variation in the azimuthal centre of the receptive fields with recording site is shown in
Fig. 2 for data pooled from three ferrets at P33-37. The visual azimuths varied systematically from the anterior midline near the rostra1 end of the SC to 130" into the contralateral hemifield at the caudal end. The map of visual space remained unchanged in animals recorded in the second and third postnatal months and closely resembled that found in normal, adult ferrets (see Fig. 4).
Development of the auditory map in the deeper layers The auditory responses recorded in ferrets at just over one month of age were mostly rather poorly tuned for sound location and there was little indication of any topographic order in the representation (King, 1993; King and Carlile, 1995). Consequently, the close alignment of the visual and auditory maps, which is characteristic of adult animals (see Fig. 51, was not apparent in the juvenile ferrets. This is illustrated in Fig. 3 in which the preferred sound azimuths of multi-unit responses recorded in the deeper layers are plotted against the visual azimuths of superficial layer units recorded in the same vertical electrode tracks. During the second postnatal month, the auditory responses became more sharply tuned for sound location and the registration with the overlying visual map gradually emerged.
Effect of ear& eye deviation on the devebpment of the visuul and auditory map A comparison of the map of visual azimuth in the supeficial layers of the SC in normal adult ferrets and in animals reared with a lateral deviation of the contralateral eye is shown in Fig. 4. The relative magnification of this dimension of the visual field is not uniform as a slightly greater area of the SC is used for representing anterior visual space than for more peripheral receptive fields. Consequently, more of the variance in the relationship between visual receptive field position and SC recording site was accounted for by a second-order polynomial function. In the strabismic ferrets, measurements with a reversible
339
ophthalmoscope revealed that the left eye was deviated laterally by 15-20". At each recording site in the SC, the head-centred visual receptive field positions in these animals tended to be more peripheral than those mapped in the normals. The polynomial function that provided the best fit to these data was virtually identical to that fitted to the normal data, except that it was shifted laterally at its midpoint by 15", which is very close
to the mean difference in eye position between the two groups. We have previously reported that, compared to the responses in normal ferrets, the preferred sound directions of auditory units recorded in these animals were shifted laterally by a corresponding amount (King et al., 1988). We have now extended these observations by concentrating on the difference between the azimuthal coor-
-
TEMPORAL
..
1 mm
H
SUPERIOR
INFERIOR
NASAL
0
0
0
0
0
4
a
12
16
a
0
0
0
6
9
12
15 GREEN LABELLED CELLS
3
20 RHODAMINE LABELLED CELLS
Fig. 1. Flattened whole-mount of the retina showing the distribution of retrogradely-labelled ganglion cells following a single 50-111 injection of rhodamine beads into the rostro-lateral region and multiple injections of green beads (total of 75 nl) into the caudo-medial region of the contralateral SC in a PO ferret. The retina was sampled at 200-pm intervals with 75pm X 75 p m grid boxes. The open circles represent cells containing rhodamine beads, while the filled circles represent cells labelled by green beads. The number of labelled cells in the grid boxes sampled in each part of the retina is shown by the size of the circles, as indicated below the figure.
340
VISUAL MAP AT EYE OPENING 0
0.0
0.5
0
1.0
1.5
2.0
2.5
3.0
Distance from mstmlabral border (mm) Fig. 2. Representation of visual azimuth in the superficial layers of the SC in ferrets recorded at P33-37. The best azimuths of multi-unit responses are plotted against the distance of each recording site from the rostro-lateral border of the nucleus. These data are expressed in this and subsequent figures in head-centred co-ordinates. 0" lies directly in front of the animal and negative numbers denote the hemifield contralateral to the recording site.
dinates of visual and auditory units recorded in the same electrode penetrations. We have considered only those auditory single units that were tuned to single positions in space at sound levels of around 25 dB above threshold. In doing so, we have excluded a small proportion of the units recorded in both the normal and strabismic animals that were classified as either being too broadly tuned to be attributed a single best position or as having bilobed azimuth response profiles (see Schnupp et al., 1995 for details). A comparison of the visual and auditory best azimuths is illustrated in Fig. 5. The spatial representations of the two modalities were closely aligned in both normal and strabismic animals. The superficial layer visual azimuth was subtracted from the auditory azimuth of each unit recorded in the deeper layers to yield auditoryvisual misalignment values. These are plotted as histograms in Fig. 5. For both groups, the histograms appeared to be. normally distributed and clustered around 0. Following the chronic change
in eye position in the strabismic ferrets, the visual receptive fields were centred at more contralateral positions than in normal animals. Had the auditory map remained uncompensated, we would have expected to see a systematic shift in auditory-visual misalignment of about 15" toward positive values in the strabismic ferrets. Instead, we observed a mean misalignment of nearly -6", although the mean values were not significantly different in the two groups of animals (t9, = 1.51; P = 0.14). By comparing the variance of the misalignments, we found that the widths of the histograms were not significantly different either (two-tailed F test; P = 0.08). These results suggest that the map of auditory space had shifted rostrally within the SC to compensate for the displacement in the visual field representation caused by the chronic change in eye position.
Eflect of chronic MDA-receptor blockade on the visual and auditory maps Electrophysiological recordings were made from animals that had received MKBO1-Elvax implants only if it was established during the surgical preparation on the day of recording that the implant still covered the whole of at least one SC. Recordings were commenced unilaterally within an hour of removing the Elvax. Compared to control animals, we noted no discernible change in the visual multi-unit activity recorded in the superficial layers of the SC. The topographic organization of the representation of visual azimuth is shown for adult and juvenile ferrets in Fig. 6. The best azimuths of the visual receptive fields varied systematically between the rostral and caudal ends of the nucleus. The visual maps in both adult and juvenile ferrets that had received MK801-Eh implants closely resembled those found in unoperated, normal control animals recorded at corresponding ages. In contrast to the normal visual map in the superficial layers, we found that chronic release of the NMDA receptor antagonist MK801 did
341
affect the auditory responses in the deeper layers. In all animals treated with Elvax containing MK801, the proportion of auditory units tuned to single regions of space was significantly reduced compared to unoperated, age-matched animals or to animals that had received Elvax implants containing dimethyl sulfoxide (DMSO) as drug-free controls (Schnupp et al., 1995). We observed an age-dependent effect on the topography of the auditory representation. In the adult M a 0 1 group, plotting the azimuthal best positions of these tuned units against the visual best azimuths of the superficial layer responses revealed a close alignment between the two maps, as further illustrated by a narrow auditory-visual misalignment histogram (Fig. 7). The spread (variance) in misalignment values was not significantly different from that found in either normal, unoperated adults (two-tailed F test, P = 0.48) or drugfree, adult controls (P = 0.84). In contrast, the best azimuths of the tuned units recorded in the SC of juvenile ferrets reared with MK801-Elvax implants were poorly correlated with the histolog-
-
P33 P37 180
140
0
ical coordinates of their recording sites. Because the visual map in these animals was essentially normal (Fig. 61, a comparison of visual and auditory best azimuths showed that there was considerable scatter in the relationship between the two representations (Fig. 7). The variance in the auditory-visual misalignment values for these animals was significantly greater than that found in both juvenile, unoperated (P< 0.02) and drugfree (P< 0.01) control groups. Effect of partial lesions of the superficial layers on the development of the ukmal and auditory maps
Visual responses recorded from the remaining rostra1 region of the superficial layers in the SC of adult ferrets in which the caudal pole had been aspirated on PO appeared to have normal receptive fields. These units were tuned to azimuthal locations that covered a restricted region of the anterior hemifield and fell either within or very close to the normal range of visual azimuths represented in this part of the nucleus in adult
I
-
-
P41 - P a
P53 P62
P48 P52
180
1 0
180
140
140
140
100
100
60
60
100 60
0
s__tt
0
20 I
:
,c-:
:
:
t
20
-20 -20 so
60
so
-im -140
-140
-180
-10
-1m
0
0
0
0
% S 8 8 8 W ,
I
,
Visual best 8zimuth [dog.]
Visual best azimuth [deg.]
Fig. 3. Development of topographically-aligned visual and auditory representations in the SC. These data were obtained from juvenile ferrets at the postnatal ages indicated at the top of the figure. For each vertical electrode penetration, the visual best azimuth of multi-unit activity recorded in the superficial layers is plotted against the auditory best azimuth of multi-unit responses recorded in the deeper layers. These estimates of spatial selectivity are based on power spectral density measurements, whereas the number of sound-evoked spikes from single units is used in the remainder of the figures. We have previously shown that the two recording techniques produce equivalent results (King and Carlile, 1994).
342 VISUAL TOPOORAPHY IN THE SUPERRclAL LAYERS W l
that measured for the equivalent rostral region of the SC in normal animals (P= 0.13). We determined the extent of the lesioned area both histologically and by the absence of characteristically strong visual drive in the caudal region of the SC. Most of the response properties of the auditory units recorded beneath this lesioned area appeared to be normal and the majority were tuned to single spatial regions. Despite the absence of visual activity in the superficial layers, approximately one third of these deep layer auditory units were also visually responsive (presumably as a result of descending cortical inputs). The azimuthal best positions of the units are plotted against the rostro-caudal coordinates of their recording sites in Fig. 9. The great majority were tuned to sound locations within the contralateral hemifield. But whereas many had best azimuths that fell within the normal range, there were also numerous units with best azimuths not normally Seen in the corresponding regions of the SC. Because of the lack of a visual map in the superficial layers against which to compare these preferred sound directions, we calculated the angular difference between the observed auditory best azimuths and the value predicted for each recording site from the polynomial function that provided the best fit to the representation of sound azimuth in normal, adult ferrets. The variance of the differences between observed and predicted best azimuths for the auditory units for which the superficial layers were missing was significantly greater than the variance associated with either the rostral region with intact superficial layers in the same animals (P4 0.01) or the normal control group (P4c 0.01). Interestingly, the auditory best azimuths appeared to be equally scattered in both unimodal and bimodal (visual-auditory)neurons.
1
-la -160
Fig. 4. Effect of lateral deviation of the eye on the representation of visual azimuth in the superficial layers of the SC. The filled circles represent data from normal, adult ferrets, while the open circles indicate the responses recorded in adult ferrets in which an exotropic strabismus had been surgically induced in the contralateral eye on €27-28. The lines correspond to the second-order polynomial functions fitted to each set of data. The visual receptive fields of the SC neurons remained aligned with the retina; when expressed in headcentred co-ordinates they became displaced relative to the normal group by an amount equivalent to the lateral shift in eye position.
animals (Fig. 8A). The extent of the visual field representation was therefore determined by the size of the lesion. The electrode penetrations from which these units were recorded were then extended into the deeper layers. The auditory units we encountered there were tuned to very similar anterior locations. Again their best azimuths fell within the normal range of values associated with the rostral part of the SC (Fig. 8B). Some of these auditory units were also visually responsive. The auditory best positions of both unimodal and bimodal neurons were in close correspondence with the receptive fields of the visual units recorded in the overlying superficial layers, and the variance of the auditory-visual misalignments did not differ significantly from
Discussion
Maps of visual and auditory space in the SC share common co-ordinates. As well as providing a ba-
sis for delivering modality-independent spatial
343 STRABISMUS
NORMAL 40 20
I
40
n
20
-20 -40
'8; *
15 10 5 0
Fig. 5. Effect of lateral deviation of the eye on the registration of the visual and auditory representations in the SC. The data on the left are from normal, adult ferrets and those on the right are from adult ferrets in which an exotropic strabismus had been surgically induced in the contralateral eye on P27-28. The histograms show the angular difference between the best azimuths of tuned auditory units recorded in the deeper layers and those of visual multi-unit responses recorded in the superficial layers of the same vertical electrode penetrations. Neither the mean nor the variance of the auditory-visual mis-alignments differ significantly between the two groups. Adapted from King et al. (1988).
signals to the neural pathways that control orientation behaviour, the alignment of the visual and auditory representations allows multisensory cues to be integrated in ways that may improve the accuracy of orienting behaviour (Stein and Meredith, 1993). In this report, we have examined the developmental steps that lead to the formation of spatially matched visual and auditory maps in the ferret SC. We have shown that the visual map in the superficial layers matures some weeks before the higher-order map of auditory space in the deeper layers, and that the activity-dependent process of auditory map development is regulated by visual signals that may arise from the superficial layers.
Development of the visual map We found that the retinocollicular projection has topographic order at PO. Changes in the number
of retinal ganglion cells (Henderson et al., 1988; Cucchiaro, 1991; Thompson and Morgan, 1993) and in the specificity of this projection (Snider and Chalupa, 1993) continue to occur until the end of the first postnatal week. Nevertheless, by the time of eye opening at around P32, the retinocollicular pathway should be topographically mature. This is supported by the results of our recordings from juvenile ferrets within a few days of the eyes opening, which revealed the presence of an adult-like representation of visual space in the superficial layers. In adult animals in which an exotropic strabismus of the contralateral eye had been induced just before normal eye opening, we found that the visual receptive fields were centred on locations that were, on average, displaced laterally by 15" compared to the normal map. We can therefore conclude that the retinocollicular projection was unaltered by this
344
Chronic MK-801 in adulthood 40 T
P
E
%
.E
I
20
-40
;-60 Q
-
0 0
-20
-100-120 -140
F
P
'0
Y
.
,
.
,
.
,
.
0 -
-
I
-60-
e?
-40-
6% 'ed
3
5 I
Distance from rostro-lateral border [mm]
0
ee8 8
! m -80-
0
I
-
-20
-
-
-
%
.g
-
3
(b
20 0
-
-80-
5
40
-
f (0
Chronic MK-801 in infancy
-
-1401
e
e o
-100-120
b
.
I
.
,
.
,
,
,
,
1
Distance from rostro-lateral border [mm]
Fig. 6. The representation of visual azimuth in the superficial layers of adult (left) and juvenile (P61-70, right) ferrets. The azimuthal best positions of multi-unit responses are plotted against the distance of each recording site from the rostro-lateral border of the SC. The filled circles represent data from animals in which MK801-Ehrax implants had been placed on the dorsal surface of the SC 5-6 weeks before recording. The open circles represent the data obtained from normal, unoperated animals. Adapted from Schnupp et al. (1995).
procedure, as the lateral shift in the visual field representation matched the change in eye position. We were also unable to detect any developmental change in the visual map in the superficial layers following chronic treatment with the NMDA receptor antagonist -1. Other reports have indicated a role for NMDA receptors in the refinement of retinocollicular topography (Cline and Constantine-Paton, 1989; Simon et al., 1992). However, in contrast to these studies, we implanted the Elvax on the SC after the early postnatal changes in the projection had been completed. The early maturation of the retinocollicular pathway in the ferret may also explain why the visual map in the SC did not appear to become reorganized following a partial ablation of the superficial layers on the day of birth. In neonatal hamsters, partial SC lesions can result in compression of the entire contralateral visual field onto the remaining portion of the superficial lay-
ers (see Finlay, this volume). However, we found that, in ferrets, visual units recorded in this region were tuned to the same azimuthal locations as in n o d animals. The visual map was therefore abruptly truncated at the caudal end of the intact superficial layers. Whereas the ferret retinocollicular projection is topographically organized at PO, the hamster pathway is still disordered at this age (Thompson and Cordery, 1994). The capacity for reorganization of the pattern of connections from the retina to the SC following manipulation of neural activity or reduction in target volume therefore appears to be limited by the developmental maturity of the projection at the time of the surgery.
V%ualmperience induedphticity in the developing m a p of auditory space
In contrast to visual responses in the superficial layers, the auditory responses recorded in the
345
Chronic MK-801 treatment in adulthood
20
Chronic MK-801 treatment in infancy
+
Visual best azimuth [deg.]
8
-s 6 5
4
1 2 0
auditory-visual misalignment [deg.]
audiIaryvisual mi8alignmenl [deg.]
Fig. 7. Relationship between the representations of visual azimuth in the superficial layers and auditory azimuth in the deeper layers of the SC of adult (left) and juvenile (P61-70, right) ferrets that had received MK801-Elvax implants 5-6 weeks before recording. The histograms plot the angular difference between the best azimuths of auditory and visual units recorded in each electrode track. The histogram is significantly wider in the juvenile animals. Adapted from Schnupp et al. (1995).
intermediate and deep layers were initially poorly tuned for sound location and exhibited very little order in the distribution of azimuthal locations to which they responded most strongly. The topographic order in the auditory representation, and the alignment with the overlying visual map, gradually emerged over a course of several weeks following the onset of hearing. A similar result, albeit with a different time course, has been reported for the guinea pig (Withington-Wray et al., 1990). This delayed appearance of the auditory space map can be understood in terms of the way in which it is constructed. Whereas the visual map results from a topographic projection from the retina, an equivalent projection from the
cochlea produces a neural map of sound frequency, not sound source position. The topographic representation of auditory space must therefore be computed within the brain by tuning the neurons to different localization cue values. In mammals, SC neurons are sensitive to monaural spectral cues and interaural level differences (Hirsch et al., 1985; Wise and Irvine, 1985; Middlebrooks, 1987; Carlile and King, 1994; King et al., 1994). We have shown that, in the ferret, the spatial pattern of these cues continues to change for a period of several weeks after the onset of hearing (King and Carlile, 1995). Thus, the acoustical basis for the map of auditory space is continually changing during much of the period over
346
Fig. 8. (A) Visual representation in the remaining superficial layers of the SC in adult ferrets following aspiration of the caudal portion of these layers at PO. (B) Auditory representation in the deeper layers of the same vertical electrode tracks. The azimuthal best positions are plotted against the distance of their recording sites from the rostro-lateral border of the SC. The open triangles in (B) represent the auditory best azimuths of units that were also visually responsive, whereas the filled circles indicate the values obtained from unimodal auditory neurons. The two lines in each panel represent two standard deviations on either side of the polynomial function that provides the best fit to the visual and auditory maps in normal ferrets. The hatched area of the SC in the parasagittal view on the right indicates the aspirated region of the superficial layers.
which the topographic order in the representation emerges. The correlation between the developmental time course for the localization cues and the map of auditory space suggests that the establishment of the topographic alignment of the visual and auditory maps may, at least in part, be limited by the maturation of these cues, which, in turn, depends on growth of the head and outer ears. However, sensory experience also plays a crucial role in refining the auditory map, so that a close correspondence with the visual map is achieved in spite of individual variations in the auditory localization cue values and in the relative position of
the eyes and em. For example, distorting the binaural localization cues by plugging one ear in infancy leads to a compensatory adjustment in the auditory map so that it remains precisely aligned with the visual map (budsen, 1985; King et al., 1988). That visual cues may play a dominant role in aligning the two maps is suggested by our finding that displacement of the visual receptive fields relative to the head, caused by lateral deviation of the contralateral eye in infancy, is followed by an equivalent change in the auditory representation. Rotation of the eye disrupts the development of the auditory map, but to a much lesser extent if the animals are visually deprived
347
degraded to some extent (Knudsen et al., 1991; Withington, 1992; King and Carlile, 1993). Taken together, these observations suggest that a crude map of auditory space can develop in the absence of visual cues. However, when available, visual information can influence the development of the auditory map, and adjust its topographic order to match that of the visual map in the SC.
A A
Basisfor visual instruction of the developing auditory space map
U
8 e
n
.-8
'2!
-20
-60
-140 -180, 0
0 v
s
0.5
I
1
.
I
1.5
,
I
2
.
1
2.5
Distance from rostro-lateralborder [mm] Fig. 9. Auditory representation in the adult ferret SC following aspiration of the caudal region of the superficial layers at PO. The auditory best positions of single units recorded in the intermediate and deep layers are plotted against the distance of their recording sites from the rostro-lateral border of the SC for those electrode tracks where the superficial layers were missing. The open triangles represent the auditory best azimuths of units that were also visually responsive, whereas the filled circles indicate the values obtained from unimodal auditory neurons. The two lines each represent two standard deviations on either side of the polynomial function that provides the best fit to the map of sound azimuth in normal ferrets.
at the same time, suggesting that the shift in auditory spatial tuning results from the displacement of the visual receptive fields rather than the abnormal eye position (King and Carlile, 1995). Optical displacement of the visual field in barn owls reared with prisms mounted in front of their eyes also leads to a corresponding change in the auditory space map (Knudsen and Brainard, 1991). On the other hand, if the visual image is blurred (by eyelid suture), the auditory representation is
A Hebbian mechanism of synaptic plasticity could provide the basis for establishing the alignment of the visual and auditory maps in the SC. Support for this possibility is provided by our fjnding that chronic application of MK801 during the period when the topographic order in the auditory representation normally emerges, results in a degraded map and, consequently, a much poorer alignment with the unchanged visual map. On the other hand, the registration of the maps was unaltered by treatment of adult animals with MK801-Elvax implants. The activation of NMDA-type glutamate receptors may provide a means of detecting temporal correlations among simultaneously active synaptic inputs (Fox and Daw, 19931, and thereby act as a trigger for the activity-dependent steps that lead to the emergence of a precise and unambiguous representation of auditory space in the SC. Feldman et al. (1995) have also reported that NMDA-receptor mediated currents may be particularly involved in expressing the visuallyguided changes in the auditory receptive fields of neurons in the barn owl optic tectum. The involvement of these receptors in the SC is consistent with a Hebbian mechanism in which more accurate visual signals could lead to the selective strengthening of auditory inputs that convey information from the same region of space. Source of instructive visual s i p &
Most auditory neurons in the barn owl optic tectum are also visually responsive (Knudsen,
348
1982). Optical displacement of the visual field has been reported to shift the auditory receptive fields of not only these units but also those of apparently unimodal units recorded in the external nucleus of the inferior colliculus, which projects topographically to the tectum (Brainard and Knudsen, 1993). The only equivalent study in mammals has shown that dark rearing disrupts the development of auditory spatial selectivity in the guinea pig SC, but has no effect on the ICX (Binns et al., 1995). By using [3H]MK801,we were able to determine the area over which this NMDA-receptor antagonist is released (Schnupp et al., 1995). Our results suggest that MK801 released from the Elvax penetrated at least as far as 800 p m below the SC surface and therefore reached both the superficial and intermediate layers of the nucleus. Although further experiments will be required to verify this, we currently believe that the changes that we have observed in the developing auditory map are most likely to be localized to the SC itself, rather than the sources of afferent input. Many auditory neurons in the SC also receive visual inputs, which most likely originate from extrastriate visual areas of the cortex (reviewed in Stein and Meredith, 1993). However, their visual receptive fields tend to be much larger than those of neurons recorded in the superficial layers. Moreover, the proportion of deep SC neurons that receive multisensory inputs is much smaller in early postnatal life (Wallace et al., 1993). If the site of developmental visual-auditory interaction does reside within the SC, an alternative source of guiding visual input would appear to be needed. Although the superficial and deeper layers of the SC have been regarded as anatomically and physiologically distinct regions (Casagrande et al., 1972; Edwards et al., 19791, connections are now known to exist between them (e.g. Mooney et al. 1984, 1992; Behan and Appell, 1992). Our finding that the auditory representation was disrupted only in the region beneath the ablated superficial layers is consistent with the possibility that those layers may provide local signals that calibrate the development of the underlying audi-
tory neurons. That bimodal neurons in the region of the SC where the superficial layers were missing also tended to exhibit aberrant auditory best positions would seem to indicate that other visual afferents to the SC were unable to rescue the spatial tuning of the auditory responses. At first sight, the lack of a measurable visual input to all acoustically-responsive neurons, particularly during early life, would appear to question the significance of synchronized activity in the superficial layers for the development of auditory spatial tuning. However, visual stimuli can alter the responses of these neurons to sound, even if, by themselves, they do not evoke a change in firing rate (King and Palmer, 1985). It may therefore be necessary to employ more sensitive cross-correlation techniques to reveal the influence of activity in the superficial layers on the responses of neurons in the deeper layers. Acknowledgements
We are grateful to Pat Cordery for excellent technical assistance and to the Wellcome Trust for financial support. Andrew King is a Wellcome Senior Research Fellow (grant number 031316 9O/Z) and Jan Schnupp is a Wellcome Prize Student (039456 93/21. Simon Carlile was supported by a Beit Memorial Fellowship and Adam Smith by a Wellcome Vision Research Training fellowship (034242 .91/Z).
-
-
References Behan, M. and Appell, P.P. (1992) Intrinsic circuitry in the cat superior colliculus: projections from the superficial layers. J. C o w . NeuroL, 315: 230-243. Binns, ICE., Withington, D.J. and Keating, M.J. (1995) The developmental emergence of the representation of auditory azimuth in the external nucleus of the inferior colliculw. the effects of visual and auditory deprivation. Deu. Bmin Rs., 85: 14-24. Brainard, M.S. and Knudsen, E.I. (1993) Experience-dependent plasticity in the inferior colliculus: a site for visual calibration of the neural representation of auditory space in the barn owl. J. N e m c i , 13: 4589-4608. Carlile, S. and King, A.J. (1994) Monaural and binaural spectrum level cues in the ferret: acoustics and the neural
349
representation of auditory space. J. Neurophyswl., 71: 785-801. Casagrande, V.A., Harting, J.K, Hall, W.C., Diamond, I.T. and Martin, G.F. (1972) Superior colliculus of the tree shrew: a structural and functional subdivision into superficial and deep layers. Science, 177: 444-447. Cline, H.T. and Constantine-Paton, M. (1989) NMDA receptor antagonists disrupt the retinotectal topographic map. Neuron, 3: 413-426. Cucchiaro, J.B. (1991) Early development of the retinal line of decussation in normal and albino ferrets. J. Comp. Neurol., 312: 193-206. Edwards, S.B., Ginsburgh, C.L., Henkel, C.K. and Stein, B.E. (1979) Sources of subcortical projections to the superior colliculus in the cat. J. Comp. Neurol., 184: 309-30. Feldman, D.E., Brainard, M.S. and Knudsen, E.I. (1996) Newly learned auditory responses mediated by NMDA receptors in the owl inferior colliculus. Science, 270 525-528. Fox, K. and Daw, N.W. (1993) Do NMDA receptors have a critical function in visual cortical plasticity? Trends Neurosci.,1 6 116-122. Hartline, P.H., Pandey Vimal, R.L., King, A.J., Kurylo, D.D. and Northmore, D.P.M. (1995) Effects of eye position on auditory localization and neural representation of space in superior colliculus of cats. @. Brain Res., 104: 402-408. Henderson, Z., Finlay, B.L. and Wikler, K.C. (1988) Develop ment of ganglion cell topography in ferret retina. J. Neurosci, 8: 1194-1205. Hirsch, J.A., Chan, J.C.K. and Yin, T.C.T. (1985) Responses of neurons in the cat’s superior colliculus to acoustic stimuli. I. Monaural and binaural response properties. J. Neurophysiol., 5 3 726-745. Jay, M.F. and Sparks, D.L. (1987) Sensorimotor integration in the primate superior colliculus. 11. Coordinates of auditory signals. J. NeurophysioL, 5 7 35-55. King, A.J. (1993) The Wellcome Prize Lecture. A map of auditory space in the mammalian brain: neural computation and development. Ekp. Physiol., 7 8 559-590. King, A.J. and Carlile, S. (1993) Changes induced in the representation of auditory space in the superior colliculus by rearing ferrets with binocular eyelid suture. @. Brain Res., 94: 444-455. King, A.J. and Carlile, S. (1994) Responses of neurons in the ferret superior colliculus to the spatial location of tonal stimuli. Hear. Res., 81: 137-149. King, A.J. and Carlile, S. (1995) Neural coding for auditory space. In: M. S. Gazzaniga (Ed.), nte Cognitiue Neurosciences, MIT Press, pp. 279-293. King, A.J. and Hutchings, M.E. (1987) Spatial response properties of acoustically responsive neurons in the superior colliculus of the ferret: a map of auditory space. J. Neurophysiol.,57: 596-624. King, A.J. and Palmer, A.R. (1983) Cells responsive to freefield auditory stimuli in guinea-pig superior colliculus: dis-
tribution and response properties. J. Physiol. Lond., 342: 361-381. King, A.J. and Palmer, A.R. (1985) Integration of visual and auditory information in bimodal neurones in the guinea-pig superior colliculus. Exp. Brain Res., 60: 492-500. King, A.J., Hutchings, M.E., Moore, D.R. and Blakemore, C. (1988) Developmental plasticity in the visual and auditory representations in the mammalian superior colliculus. Nature, 332:73-76. King, A.J., Moore, D.R. and Hutchings, M.E. (1994) T o p e graphic representation of auditory space in the superior colliculus of adult ferrets after monaural deafening in infancy. J. Neurophysiol., 71: 182-194. Knudsen, E.I. (1982) Auditory and visual maps of space in the optic tectum of the owl. J. Neurosci., 2 1177-1194. Knudsen, E.I. (1985) Experience alters the spatial tuning of auditory units in the optic tectum during a sensitive period in the barn owl. J. Neurosci., 5: 3094-3109. Knudsen, E.I. and Brainard, M.S. (1991) Visual instruction of the neural map of auditory space in the developing optic tectum. Science, 253: 85-87. Knudsen, E.I. and Brainard, M.S. (1995) Creating a unified representation of visual and auditory space in the brain. Annu. Rev. Neurosci, 18: 19-43. Knudsen, E.I., Esterly, S.D. and du Lac, S. (1991) Stretched and upside-down maps of auditory space in the optic tecturn of blind-reared owls; acoustic basis and behavioral correlates. J. Neurosci., 11: 1727-1747. Middlebrooks, J.C. (1987) Binaural mechanisms of spatial tuning in the cat’s superior colliculus distinguished using monaural occlusion. J. Neurophysiol.,57: 688-701. Middlebrooks, J.C. and Knudsen, E.I. (1984) A neural code for auditory space in the cat’s superior colliculus. J. Neurosci., 4: 2621-2634. Mooney, R.D., Bradley, G.K., Jacquin, M.F. and Rhoades, R.W. (1984) Dendrites of the deep layer, somatosensory superior colliculus neurons extend into the superficial laminae. Brain Res., 324: 361-365. Mooney, R.D., Huang, X. and Rhoades, R.W. (1992) Functional influence of interlaminar connections in the hamster’s superior colliculus. J. Neurosci., 1 2 2417-2432. Peck, C.K, Baro, J.A. and Warder, S.M. (1995) Effects of eye position on saccadic eye movements and on the neuronal response to auditory and visual stimuli in cat superior colliculus. Exp. Brain Res., 103: 227-242. Schnupp, J.W.H., King, A.J., Smith, A.L. and Thompson, I.D. (1995) NMDA-receptor antagonists disrupt the formation of the auditory space map in the mammalian superior colliculus. J. Neurosci., 15: 1516-1531. Simon, D.K., Prusky, G.T., O’Leary, D.D.M. and Constantine-Paton, M. (1992) N-methyl-D-aspartate receptor antagonists disrupt the formation of a mammalian NatL Acad. Sci. USA, 89: 10593-10597. neural map. h.
350
Snider, C.J. and Chalupa, L.M. (1993) Specificity of the retinocollicular pathway in developing cat and ferret. Soc. Neurosci. Abstr., 1 9 454. Stein, B.E. and Meredith, M.A. (1993) The Merging of the Senses, MIT Press. Cambridge, MA. Thompson, 1.D. and Cordery, P.M. (1994) The development of the retino-collicular map in the Syrian hamster. Soc. Neurosci. Abstr., 2 0 1704. Thompson, I.D. and Morgan, J.E. (1993) The development of retinal ganglion cell decussation patterns in postnatal pigmented and albino ferrets. Eur. J. Neurosci., 5: 341-356. Wallace, M.T., Meredith, M.A. and Stein, B.E. (1993) Development of multisensory integration in cat superior colliculus. Soc. Neurosci. Abstr., 1 9 240.
Wise, L.Z. and Irvine, D.R.F. (1985) Topographic organization of interaural intensity difference sensitivity in deep layers of cat superior colliculus: implications for auditory spatial representation. J. NeumphysioL, 5 4 185-211. Withington, D.J. (1992) The effect of binocular lid suture on auditory responses in the guinea-pig superior colliculus. Nemsci. Lett., 136 153-6. Withington-Wray, D.J., Binns, KE. and Keating, MJ. (1990) The developmental emergence of a map of auditory space in the superior colliculus of the guinea pig. Deu. Bruin Res., 51: 225-36.