Visual cortex controls retinal output in the rat

Emin

Rr.\c~r&~

Bu//rti,z.

Vol. 17, pp. 21-32.

1986. ” Ankho

International

Inc. Printed

0361-9230186

in the U.S.A

$3.00

+ .OO

Visual Cortex Controls Retinal Output in the Rat STBPHANE DGpoI-ternrnt

MOLOTCHNIKOFF

de Scic>ncc>s Biologiyues,

Univrrsitc;

AND FRANCOIS

dc Montkrl,

TREMBLAY

C. P. 6128, SUCT. A, Montrt%rl, P. Q.) H3C 357

Received 4 February 1986 MOLOTCHNIKOFF, S. AND F. TREMBLAY. Vi.s/rtr/ car-tc’.r c.c~nfr.ols wtiutrl o/rrp/r~ ;,I rhr rvrt. BRAIN RES BULL 17(I) 21-32. 1986.-The first objective of the present investigation was to shed more light on corticofugal influences on the retina by providing an analysis of the type and proportion of retinal ganglion cells that are affected by cooling the visual cortex in rats. The second question was to determine if the pretectum participates in functional cortico-retinal relationships. In urethane-anesthetized and paralyzed hooded rats, axonal activity of retinal ganglion cells was recorded with glass micropipettes at optic chiasm level. Units were classitied as ON, OFF. suppressed-by-light and concentric. The visual cortex was inactivated by cooling its surface with a 4 mm! steel probe using the Peltier effect. The pretectum was blocked with microinjections of 50 to 100 nanoliters of cobalt ions, lidocaine hydrochloride or KCI. The inactivations and recoveries at both sites were monitored by simultaneously recording evoked field potentials. Interrupting corticofugal impulses caused modifications of the evoked discharge pattern in all types of cells. The concentric type was the group least affected by cortical cooling. A common trend emerged suggesting that cooling of the visual cortex led to an enhancement of the initial evoked excitation. This was often followed by an enhanced post-excitatory inhibition. The Pearson coefficient allowed us to measure the degree of similarity between two histograms. When all data were pooled, a weak correlation between control and test histograms (r=0.29, N=56) was found, while the control and recovery patterns averaged a correlation of more than twice that size (r=0.68). In a second series of experiments. the pretectum and visual cortex (VC) were simultaneously inactivated. It is shown that both sites summed their influence and acted synergistically upon the pattern of ganglion cell responses. The results strongly suggest that the visual cortex exerts a major control over the response pattern of thirty percent of retinal ganglion cells, and that the pretectum participates in the functional relationships between visual cortex and retina in rats. Visual

cortex

Retina

Rat

Corticofugal

influences

THROUGH its centrifugal innervation, the brain is capable of controlling the activity of the peripheral sensory apparatus. The organs of olfaction and audition and the muscle spindles are the most widely recognized structures to be innervated by axons whose cell bodies are located in the brain stem. In several vertebrate classes. retinopetal neurons have been successfully demonstrated by means of histological tract-tracing methods. For instance, the lamprey reticular mesencephalic area is a site of origin for the centrifugal visual system [57]. In fishes, the retina receives centrifugal fibers emerging from several brain stem nuclei. Thus, retinopetal fibers have been found to have multiple sources in teleost fishes [IO]. Pretectal and tectal areas as well as the corpus geniculatum laterale ipsum and the dorsomedial optic nucleus send axons to the eye [lo, 35, 551. In various amphibians [3,1 l] and reptiles [29,45], the intraocular injection of HRP has revealed peroxidase positive cells in the isthmic region of the mesencephalic tegmenturn, the ventral thalamus and the optic tectum. However, the retinopetal system has been identified most clearly in birds [8,9, 3 1,441. The isthmo-optic nucleus is considered to be the most important avian nucleus that projects fibers to the retina [3 I]. In rats, Itaya [20, 21, 401 showed that medial pretectal area (PRT) neurons are labelled with horseradish peroxidase if the vitreous body of the eye is injected with the

enzyme, while hypothalamic neurons of the dog seem to project axons to the retina [7,53]. Finally, electron microscopic observations following lesions of the central visual pathway revealed the presence of centrifugal fibers to the retina of cats and monkeys [5]. Despite these results the physiological role of the retinopetal axons has been critically questioned (see [56]) and even denied [4,48]. This scepticism stems from the fact that the investigations carried out to demonstrate a corticofugal control over the retina left themselves open to criticism [56]. Two types of difficulties were encountered: (a) an efferent sympathetic influence on retinal activity is difficult to rule out completely and (b) the observed modifications of retinal activity following an electrical excitation of the optic nerve may be attributed to an antidromic invasion rather than to an orthodromical excitation of efferent fibers [38]. Nevertheless, intriguing observations do suggest a centrifugal influence on the retina. For instance, Spinelli and Weingarten [.50] showed in cats that combining auditory and somatic stimulations with visual stimuli resulted in the following effects: a modification of latencies and patterns of light-evoked responses of retinal ganglion cells and altered dimensions of their receptive fields. The authors suggested that these interactions between different sensory modalities at the peripheral level were

MO~(~TCHNIKOFF

1’ /

i r--

.-. ‘--

--

-

lNJECYlONS

J [ CRY0] BkOCKyE

LGN,--_.-ivc

/

FIG. 1. Schematic diagram of the experiments. Unitary activity was recorded from axons of retinal ganglion cells at the optic chiasm Ievel (OC). Microinjections were carried out at the pretectum levei (PRT). Field potentials were recorded through the injecting pipette. The visual cortex (VC) was cooled, and recorded cortical 6eld potentials allowed us to monitor cortical activity. LGN: lateral geniculate nucleus.

brought about by a retinopetal system. Recently, we have demonstrated that a local cryoblockade of the rat’s visual cortex produced similar changes in latencies and bursting patterns of flash-evoked responses of retinal ganglion cells 1331. The objective of the present study was to investigate in more detail this cent~ugal influence by providing an analysis of the types of retinal ganglion cells which are most affected, and a description of the nature of the modifications of the responses produced by cooling the visual cortex. Since a direct projection from the cortex to retina has not yet been demonstrated, one has to postulate a relay between both structures. This relay should contain the retinopetal cells and receive an input from the visual cortex. One area which seems to fulfill the two requirements is the pretectum, since it receives an input from the cortex and is the source of centrifugal axons in rats [20,21]. Thus, the question posed in the second part of the paper was whether the pretectum (PRT) participates in corticoretinal functions. If, indeed, the influence of the visual cortex on the retina passes through the pretectum, the inactivation of the latter should mimic cryoblockade of the cortex. In addition, pretectal depression should prevent a modification of ganglion cell firing if the cortex is cooled afterward. Preliminary reports of these findings have been published [34].

METHOD Animal Preparation

Hooded rats of both sexes (2.50 g) were prepared for acute experiments. The animals were anesthetized with urethane (1.2-I .5 gskg-* IP), and all surgical and pressure sites were injected with lidocaine hydrochlo~de (%). The EKG was continuousIy monitored as was the rectal temperature which was mainlined at about 37°C using a heating plate. The rat’s head was mounted in a modified stereotaxic apparatus with FIG. 2. A: Photomicrograph of a section through the rostra1 part of the superior colliculus of the rat that received a 75 nanoliter injection of KC1 (3 M) with neutral red (0.1 mg/ml). In the described experiments, the volume injected amounted to approximately 0.40 pm3. B: collicular responses to ON- and OFF-steps of a flash were abolished by the injection. Horizontal calibration: 100 microns. C: photomicrograph of a section through the optic chiasm. The electrolytic lesion (arrow head) conlkms that the tip of the electrode was positioned within the chiasm.

AND TREMBLAY

CORTICO-RETINAL

23

INFLUENCE TABLE INFLUENCE

OF CORTICAL

1

CRYOBLOCKADE

ON 188 RETINAL

Cell Type

OFF

SBL

ON

INFLiTOT %

12129 41

5117 29

19157 33

GANGLION

CONC

PTC

lW.53 19

515 100

N.ID 5127 19

0.24 0.66

0.36 0.72

0.27 0.71

r-Test

t P

4.28 <0.005

5.57 <0.005

0.47 0.65

Total 561188 29 Mean

Pearson Coefficient CONTKRYO CONT/REC

CELLS

-

0. I4 0.59

0.29 0.68

CONTKRYO

4.67 <0.005

1.86 NS

-

-

-

Cell types: OFF: off-cells; SBL: suppressed-by-light; ON: on-cells; CONC: concentric cells; PTC: presumed pretectal cells; NID: non-identified cells. INFL/TOT: Proportion of influenced cells of the total number of units recorded in each class; 70: percentages of the total in a given cell category; CONTICRYO: Ratio of Pearson coefftcients of control PSTH over cryoblockade PSTH; CONT/REC: Ratio of Pearson coefficients of control over recovery PSTHs.

an unobstructed visual field. The contralateral eye was kept closed throughout the experiment. Visual cortex areas V, and VZ were made directly accessible by cutting away a section of the overlying skull. A second craniotomy above the optic chiasm allowed the penetration of microelectrodes toward the retinal ganglion cell axons, which were reached stereotaxically [43]. The stimulated eye was firmly tied to the stereotaxic frame. The stable spatial coordinates of the receptive fields confirmed that the eye maintained its position. A contact lens (O.D.) covered the cornea to keep it humid and clear. Finally, the pupil was dilated with atropine sulfate (0.0%).

A steel probe (surface area=4 mm2) was gently applied against the cortical surface. Cooling was achieved by the Peltier effect (Cambion). It allowed us to cool and rewarm the cortical surface in a gradual fashion. The temperature of the cortical surface was measured through the thermoprobe attached to the cooling plate. A fine stainless steel wire was glued to the plate in order to record visually evoked responses and the EEG. The disappearance of the evoked responses assured us that the visual cortex was inactivated. In addition, control recordings at increasing distances from the probe indicated that the spread of cooling did not extend more than 2.0 mm beyond the plate, since undist~bed evoked potentials were obtained at these distances. In each animal, it was possible to cool and rewarm the cortex several times without any obvious deleterious influence on the EEG or evoked potentials.

Inac.tivation. Pretectal inactivation was carried out in a second series of experiments whose design is schematized in Fig. 1. The visual cortex was inactivated by cryoblockade in a manner similar to that described above. To block the pretectal area (PRT), a micropipette tilled with an inactivating drug was lowered toward the pretectum, The three drugs used were potassium chloride (KC1 0.3 M, N=l2), cobalt chloride (CoCl, 4 nM, N=lO) and lidocaine hydrochloride

296 (xylocaine, N=l8). The pipette was connected to a nanoliter hydraulic pump whose head holder was modified to allow the introduction of a fine silver wire up to the tip of the pipette. This modification permitted us to record pretectal evoked field potentials as they were detected by the pipette’s aperture. Consequently, abolition of the visually evoked potential confirmed that the drug was effectively ejected. This device also permitted us to monitor the recovery. In two separate experiments, neutral red was added to the KC1 in order to histologically assess the volume of tissue infiltrated by the drug. Figure 2 illustrates diffusion when 75 nanoliters (Fig. 2A) were injected into the anterior part of the superior colliculus. The dye diffused in a spherical volume of about 40 pmJ. Figure 2B shows that collicular visually evoked responses (on- and off-stimuli) were totally abolished. It is important to stress that in all our experiments, 40 to 100 nanoliters were applied in 3 to 5 minutes, that is until the evoked responses deciined by at least 80%. The injections of the three drugs in the PRT area produced very simiiar effects on retinal ganglion cell responses. Consequently, all data were grouped together in this paper. KC1 exerts its action rapidly, and recovery is obtained within an hour. This procedure has a drawback, however, because it may produce spreading depression [23,24], which extends far beyond the injection site. In particular, it may depolarize the ganglion cell axon terminals and backfire the unit. Cobalt ions antagonize synaptic transmission [28] and are very effective in modifying ganglion cell discharges. The recovery, however, is very long in spite of the small injected quantity, and often the unit was lost before the pretectal potential could recover. It was found that xylocaine was the most convenient agent, because the PRT area recuperated rapidly and also because it allowed us to verify that the effects were reversible. Light Stimulation

In this paper, most tests were performed with diffuse light stimulation of the eye. A translucent screen, positioned in front (20 cm) of the eye was illuminated by short flashes (GRASS PS 33) at a rate of one per second. The intensity

MOLOTCHNIKOFF

AND ‘PREMBLAY

-

-CONT.

-CRYO.

FIG. 3. Effects of cooling of the visual cortex on responses of an ON-type retinal ganglion cell (ON). A: post-stimulus time histograms based on 64 flash presentations (F). Bin width: 2 msec. B: averaged cortical light-evoked responses, same stimulus presentation as in A. CONT: control prior to cooling. CRYO: cryoblockade of the visual cortex. REC: recovery. Calibrations: horizontal: 100 msec in A and B, vertical: 4 action potentials in A and 20 WV in B.

A

B -CONT.

-

I

-

CRYO.

-

2

“_

-

REC.

-

3

F

7*

F

FIG. 4. Influence of cooling of the visual cortex on responses of an OFF-type retinal ganglion cell (OFF). A: post-stimulus time histograms based on 64 flash presentations (F). Bin width: 2 msec. B: averaged cortical light-evoked responses, same stimulus presentation as in A. CONT: control prior to cooling. CRYO: cryoblockade of the visual cortex. REC: recovery. Calibrations: horizontal: 100 msec in A and B, vertical: 4 action potentials in A and 20 PV in B.

CORTICO-RETINAL

INFLUENCE

?5

-

CRY0 -

-

REC. -

-

REC. -

4

t F FIG. 5. Effects of cooling of the visual cortex on responses of a suppressed-by-light type retinal ganglion cell (SBL). Post-stimulus time histograms based on 64 flash presentations (F). Bin width: 2 msec. CONT: controi prior cooling. CRYO: cryoblockade of the visual cortex. REC: recovery. Calibrations: horizontal: 100 msec. vertical: 4 action potentials.

FIG. 6. Influence of cooling of the visual cortex on responses of a presumed pretectal cell (PTC). Post-stimulus time histograms based on 64 Bash presentations (F). Bin width: 2 msec. CONT: control prior to cooling. CRYO: cryoblockade of the visual cortex. REC: recovery. calibrations: horizontal: IO0 msec. vertical: 4 action potentials.

(350 mW/m’) was approximately 3 log units above response threshold. Particular care was taken to keep light conditions (ambient intensity and flash frequency) constant when a cell was under study. Although all cells were tested with the stroboscopic, unstructured stimulus, additional tests were carried out for some units (N= 18) with localized stimuli positioned within their receptive fields [32]. Spots or slits of appropriate dimensions were displayed on an oscilloscope screen (Wavetek phosphor P 33) which was put in place of the translucent screen. This analysis was carried out to study the effect of cooling on responses evoked from circumscribed stimuli evoked at various light intensities (range 60 to 250 mW/m?.

stainless steel wire was introduced into the pipette up to its tip. The micropipette was then relowered to exactly the same depth. A current (50 microamp, 8-10 set) passing through the steel wire produced lesions at the recording site. Each animal was perfused transcardially with saline (O.%) followed by a 1% formaldehyde solution. Histological examinations of 40 pm thick, cresyl violet stained sections allowed us to localize the recording sites (Fig. 2C).

Single unit activity was recorded from axons of retinal ganglion cells at the level of the optic chiasm with a glass micropipette filled with I M NaCl (IO MfI DC). Cortical and axonal responses were amplified and fed to a tape recorder for later analysis. Unless otherwise specified, computer averaging of the cortical and pretectal evoked potentials and post-stimulus time histograms (PSTH) of single unit activity were based on 64 successive presentations. The bin width was set at 2 msec and the time of analysis was 500 msec post-stimulus. Immediately after the recording of the last unit, the micropipette was withdrawn and a fine varnished

RESULTS PART I

Recordings were made from one hundred and eighty-eight retinal ganglion cells. Receptive fields of single neurons were determined with discrete stimuli (spots, light or dark slits) which were projected onto the tangent screen (see [32]). Then. cells were grouped into four categories according to their responses evoked by localized stimuli positioned within their receptive field. Concentric cells (N=53) had bipartite receptive fields, with each area giving rise to on- or offresponses, depending upon whether the location of the stimulating spot was in the center or in the periphery. ONcells (N=57f responded to the on-stimulus (brightening of the flash) throughout their receptive field, and showed no excitation to the off-stimulus (dimming of the flash). The OFF-units (N=29) discharged to the off-stimulus over the entire activating area, and were not excited by the on-

MOLOTCWNIKOFF

AND TREMBLAY

c

CaNT

cffm.

REC.

6

m

_I

FIG. 7. Intluence of cortical cryoblockade on retinal responses for two iight intensities. A: a spot (2%X”) of weak intensity (t .Olog unit above thwshootd) positioned in the receptive field. B: the same stimulating spot, but with its intensity raised to 2.0 log units above threshold. CONT: control prior to cortical blockade. CRY0 cryoblockade of the visual cortex. REC: recovery. A and JB: PSTHs based on 64 flash presentations (flashes shown on the bottom line). Bin width: 2 msec. C: total number of spikes per histogram. Empt.y squares: weak intensity. Full squares: high intensity. Calibration: horizontal: 100 msec, vertical: 4 action potentials.

stimulus, Cells of a fourth group displayed continuously sustained firing in the dark, which was abruptly suppressed by light. No excitation could be revealed with any of the available stimuli. These retinal ganglion cells were called suppressed-by-light cells (SBL) (N=l’?). Table 1 summarizes the number of cells recorded in each class and the proportion of units influenced by cortical cryoblockade- Although further analysis of Table I wili foflow, it is clear at this point that the retinal ganglion cells with concentric receptive fields were least influenced by cortical blockade. Only 19 percent (N=53) of encountered units with concentric receptive fields had their responses altered by visual cortex cooling. Typical Examples ON-cells. Interrupting corticofugal impulses causes modifications of the light-evoked responses ofsome retinal ganglion cells. Typical examples of the influence of visual cortex on retinofugal activity are provided by the next three figures. An ON-unit is presented in Fig_ 3. Following Bash presentation, the unit responded with an oscillatory pattern containing four periods of burst activity at 25, SO, 250 and 375 msec, interrupted by silent pauses (Fig. 3, trace 1). Cooling of the visual cortex produced a completely different firing profile. The most salient features were an increase in magnitude and duratiQn of the first excitatory period which was broken up into five brief bursts, and the occurence of a second weak firing period at 125 msec, The Iatter coincided with the silent period which separated the second and third volleys of the control trace (Fig. 3, trace 2). As the cortex recovered from cryoblockade, the cell adopted its initial pattern of discharge (Fig. 3, trace 3). OFF-crli?r. Figure 4 illustrates the aReration of the lightevoked response by cortical cooling of an OFF-retinal gangf-

ion cell. A flash elicited a series of successive brief bursts with a latency of 125 msec (Fig. 4, trace 1). Following cortical cooling, the post-stimulus histogram presented a new pattern. The latency of the first burst shortened to 100 msec. This burst was followed by an inhibitory period of over 70 msec and a robust excitatory rebound at 270 msec. Cortical rew~ming restored the control fatency and histogram pattern (Fig. 4, trace 3). Suppressed-by-light wlls, A unit belonging to this class is presented in Fig, 5. It showed a high rate of sustained activity in the dark. Light impinging on the retina caused the cell to stop its firing rate abruptly for 100 to 2% msee (latency 40 msec, trace I). High frequency spike activity resumed after this period of diminished excitation+ Profound alterations in this firing pattern occurred when the visual cortex was rendered inoperative. It is shown in Fig. 5, trace 2, that the initial reaction of the unit to a flash was a robust excitatory response which contained three brief bursts. Therefore, it appears that the visual cortex prevented the unit from responding with short latency excitation. Although the above three ceIIs provide typical examples of the various modifications of firing patterns pertaining to the three different types of units, a common trend seems to emerge from the analysis. Cooling of the visual cortex produced qualitatively similar effects in all three cases, namely an enhancement of the initia1 evoked excitation. For instance, in the ON-unit (Fig. 3), the first per&d of excitation contained more bursts with more spikes, while the OFF-cell responded with a discharge of shorter Iatertcy {Fig. 4) and also with a strong rebound excitation. The suppressed-bylight cell produced an initial excitation instead of an inhibition (Fig. 5). It is then suggested that the cortex dampens out the excitatory-inhibitory oscillations of retinal ganglion cell responses. A minority ofcelIs (N=S) deserves some attention. These

CORTICO-RETINAL

21

INFLUENCE

A

f3

FIG.

c

X. Comparative

effects

of cooling of the visual cortex

tcryo)

and microinjections

of cobalt,

KC1 and xylocaine

(xylo) into the pretectal area on ganglion cell responses to a flash. A: all PSTHs from the same cell. Trace I: control prior to cooling; Trace 2: cortex cooled; Trace 3: recovery from the first cryoblockade: ‘Trace 4: cobalt injection. B and C: two different cells. Traces BI and Cl: controls: Trace 2B: KCI injection: Trace 2C: xylocaine injection. Note that in both units the microinjections produced similar effects, that is: shorter latency for the first burst and a more pronounced post-excitatory inhibition. Calibration: horizontal: 100 msec, vertical: 4 action potentials. In all traces, the flashes were applied at the beginning of each trace,

cells (PTC) had their evoked responses completely eliminated by cortical blockade. One such example is shown in Fig. 6. Shortly after the cortical blockade, the cell’s responses started to decline (Fig. 6, trace 2) and minutes later, the unit failed altogether to respond to flashes (Fig. 6, trace 3). After rewarming of the cortical surface, the unit recovered its ability to respond to light (Fig. 6, trace 4). Since only a few cells exhibited such reactions to cryoblockade, it would be unwise to draw general conclusions. Nevertheless, it is our assumption that these units were retinopetal fibers. This assumption is based on results of our current investigations. In these experiments, the methodology of the present paper is applied to the pretectal area. Thus, light responses of single pretectal ceils are analyzed while the visual cortex is cooled. Most pretectal units seem to react to cortical blockade in a manner similar to that illustrated in Fig. 6, i.e., they lose their light-evoked responses. This similarity leads us to suggest that the axon from which recordings are carried out in this study, belongs to a pretectal cell projecting toward the retina. Finally, in 27 cells the receptive field boundaries could not be determined, and their discharges were erratic. These cells are labelled “not identified” (NID) units, Within this group, five cells (19%) showed a modification of their firing pattern following cortical blockade.

For the purposes of this study, a cell was considered to be influenced by cryoblockade if the pattern of impulses was altered reliably within a few minutes after the cortical temperature started to decline, if the unit recovered its discharge profile with cortical rewarming and if the Pearson coefficient, which measures the degree of similarity of two histograms, varied more than 10%. The Pearson coefficient (PC) counts the number of spikes bin by bin (2 msec in all tests) in each histogram. On a celf by cell basis two coefficients were computed: the first compared control (prior to cooling) to test (during cooling) PSTHs, the second measured the difference between the control and the recovery PSTHs. Obviously, strong effects lead to a coefficient which approaches 0, meaning that the similarity between both histograms is weak, whereas a coefficient approaching the vatue of 1 indicates that the two patterns resemble each other. The latter should be the case when a comparison between the control and the recovery histograms is made. Pooling the entire population of affected cells of Table 1 gave a mean Pearson coefficient for control vs. test PSTHs of 0.29 (N=56), while the control vs. recovery correlation was more than twice as strong (0.68). For comparison, in 21 randomly selected, uninfluenced cells, Pearson coefficients

MOLO’I’C’HNIKOFF

A

GANGL.

B vc

G

AND ‘I‘REMBI,AY

PRT

FIG. 9. Effects of dual inactivations of the pretectal area (PRT) and the visual cortex (VC) on responses of a retinal ganglion cell (gangl) to a flash. A: PST&. Bin width: 2 msec. B and C: averaged cortical and pretectal field potentials, respectively. All responses based on 64 flash presentations (F). Traces i, 8 and 13: control prior to cooling; Traces 2 and 9: first cortical cryoblockade; Traces 3 and 10: first recovery; Traces 4 and 14: pretectal blockade with KCI; Traces 5 and I I: a second cryoblockade of the cortex is added to the pretectal inactivation; Traces 6 and 15: only the pretectum recovers; Traces 7 and 12: full recovery (control). Curved arrow indicates increased excitation: shorter latency or stronger response. Vertical arrow indicates increased inhibition. Calibration: horizontal: 100 msec. vertical: 4 action potentials in A and 100 PV in B and C.

0.76 and 0.77, respectively. Clearly, therefore, disrupting cortical physiology resulted in a pattern of the evoked response which lacked any strong similarity with the profile of the discharges recorded prior to cooling. Additional observations can be drawn from Table 1. Cortical blockade influenced mostly OFF-, suppressed-by-light and ON-cells. Indeed, the mean coefficients of control vs. test distributions were 0.24,0.27 and 0.36 respectively. Concentric units were the least affected with a coefftcient of 0.47 for the control vs. test comparison. Out of 10 concentric cells which were influenced by cortical blockade, only two exhibited a robust reaction. Furthermore, the difference in PC was signi~cantly smaller (f-test: p
The influence of cortical blockade on responses of retinal gan~ion cells was also studied at more than one stimulus intensity (N= 18). The investigations revealed that responses

to high intensity flashes were more profoundly modified than discharges to weak flashes. In a typical example an ON-cell responded to a small flash (intensity: I L.U. above threshold) with one robust short latency burst (Fig. 7A, trace I). Interrupting cortical functions produced a slight enhancement of the evoked excitation with brief spikeless periods between the bursts (Fig. 7A, trace 2). However, high intensity flashes (3 L.U. above threshold) caused the cell to fire with two bursts (Fig. 78, trace 4), and lowering the temperature of the cortical surface reduced the evoked excitation noticeably. This reduction was accompanied by prolonged inhibitory periods (Pig. 7B, trace 5). The total number of spikes per histogram remained roughly constant when a small light intensity was used (Fig. 7C empty squares), whereas it decreased by approximately 30% from the control level with cryoblockade for a large light intensity (Fig. 7C, filled squares). The decrease was essentially attributable to a prolongation of inhibitory periods that followed excitatory bursts. This might be due to a facilitation of a recurrent inhibitory mechanism at the inner plexiform layer of the retina (See the Discussion section).

CORTICO-RETINAL

INFLUENCE PART II

Thirty-seven ganglion cells were studied in this series of experiments. Twenty-four exhibited modifications in their response discharges to light following an injection of potassium chloride (KC]), xylocaine or cobalt ions into the pretectal area: all of these units were also sensitive to cryoblockade of the visual cortex. In thirteen additional cells. cooling of the visual cortex failed to change the evoked responses, and none of these cells could be influenced by pretectal injections. Figure 8 serves a dual purpose. First, part A demonstrates that cooling the cortex or injecting 2 M CoCl, into the PRT produced virtually identical changes in the firing pattern of the same retinal ganglion cell. To the presentation of a flash, the cell responded with two periods of excitation: a brief latency burst (110 msec) which was followed by a second, more robust discharge (190 msec). As the cortical temperature was lowered, the cell discharged with a succession of short bursts, and the first inter-burst inhibitory episode disappeared (Fig. SA, trace 2). The unit’s response recovered its initial profile with cortical rewaking (Fig. 8A, trace 3). Injecting CoCl, into the pretectal area produced a response pattern that was very similar to that observed when the cortex was inactivated. That is, the unit responded again with a series of brief bursts (Fig. 8A. trace 4). Clearly then, inactivating the visual cortex or the PRT yielded very comparable profiles of activity in retinal ganglion cells. Since it is believed that cobalt ions compete with calcium ions at synaptic sites, it may then be proposed that corticofugal axons establish contacts at the PRT level. The second objective of Fig. 8 is to illustrate that similar changes in the ganglion cell’s responses occurred when the PRT area was in~ltrated by KC1 or xylocaine. Following the inactiv~~tion of PRT by KCI, two modifications in the light response emerged: the latency of the response became shorter (curved arrow, Fig. 8B, trace 2), and a postexcitatory inhibition appeared (vertical arrow, Fig. 8B, trace 2). The same two types of modifications occurred when the pretectal area was inactivated by an injection of xylocaine (Fig. 8C, trace 2).

If one assumes that the visual cortex influences retinal output via the pretectal area, then it is reasonable to postulate that both sites sum their impact and act synergistically. The results displayed in the next figure (Fig. 9) suggest such a cooperation. In response to light the unit fired several bursts at I50 msec (Fig. 9A. trace I). When the temperature of the visual cortex was lowered, the unit reacted with a decline of the second burst (vertical arrow, Fig. 9A, trace 2). This decline may be attributed to an enhancement of postexcitatory inhibitory processes. As the cortex regained its control temperature, the second burst reappeared (Fig. 9A, trace 3). At this stage of the experimental run, the pretectal area was injected with KC]. This injection resulted in a new profile of the light response of the ganglion cell. The latency shortened by SO msec (curved arrow) and, similarly to the previous observation, the first burst was followed by an inhibitory period (Fig. 9A, trace 4). Histograms of tracings 2 and 4 exhibit similar characteristics as both contain a short post-excitatory inhibition. The next step of the experimental

?9 run consisted of blocking the visual cortex while the pretectal area was still inactivated. This dual inactivation resulted in an accentuation of the modifications observed earlier, namely, the latency of the response shortened even further (curved arrow) (20 msec) and the post-excitatory inhibition was more pronounced (vertical arrow, Fig. 9A. trace 5). A post-inhibitory excitatory rebound emerged at 380 msec (second curved arrow). As the pretectal area was allowed to recover while the cortex remained cool, the response pattern resembled the profile obtained earlier for cryoblockade alone (compare traces 2 and 6, Fig. 9A). Finally. the cell resumed its control pattern with complete recovery (Fig. 9. trace 7). This example illustrates that interrupting the activity of both PRT and VC produced additive effects on retinal ganglion cells. Of 19 units which were sensitive to cryoblockade and pretectal depression, fourteen behaved similarly to the typical example presented in Fig. 9. It should be emphasized that in this series of experiments the cortex was cooled twice and the observed effects were repeatable in the same cell. Thus the changes were not due to some random variations in excitability. DISCUSSION The physiological findings emerging from the present investigations may be summarized as follows. The visual cortex controls the response pattern of about 30% of retinal ganglion cells. The cells which were most sensitive to cortical inactivation were the ON-, OFF- and the suppressedby-light cells. while the units with concentric receptive fields were relatively unaffected. This is in contrast to results in the cortico-geniculate system [32,33]. The cortico-retinal influence seems to be exerted via specific efferent fibers whose network originates in the visual cortex. We have already shown previously that moving the cooling probe outside the visual cortex and closer to the recording site failed to modify ganglion cell responses of a unit which was sensitive to cryoblockade of the cortex 1331. Since the effect is one of enhanced excitation its explanation is incompatible with a spread of coolness toward the optic chiasm, which is located quite remotely. Had the cryoblockade produced a decline of blood or oxygen supply at the retinal level via a sympathetic pathway, one would have observed a general decrease of all neuronal activity, certainly not an enhancement of excitatory responses. All cells in part two of this study were tested at least twice with cortical blockade, and each cryoblockade produced identical results. This assured us that the observed effects were not due to some random changes in excitability. One may argue that we have recorded from nearby visual areas and particularly from the SCN, since fine glass micropipettes are prone to bending, although our electrode’s shank never exceeded 3 mm. Several facts permit us to rule out such a possibility with confidence. (I) In a previous study. light responses were recorded through steel electrodes which do not bend. The tip was localised in the optic chiasm or tract [33]. (2) Histological lesions were located in the optic chiasm. (3) Since we recorded from the optic chiasm the electrode picked signals very frequently from axons belonging to the ipsilateral eye (relative to the cooled side). This occurrence would have been impossible if the electrode’s tip had been placed in the SCN because in rats almost all ganglion cell axons cross the midline and project to the contralateral side. (4) Visual responses from SCN differ considerably from those of ganglion cells [55]. (5) The temporal characteristics of flash evoked responses are very

comparable to these reported for retinal discharges [ 11,331. Retinal ganglion cell responses were modified in a similar manner by either cortical cryoblockade or microinjections of inactivating drugs into the pretectal area. Since cobalt ions. which are believed to antagonize synaptic transmission, mimic cortical blockade, it may be proposed that corticofugal axons establish synaptic contacts at the level of the pretectum. This claim is supported by histological observations that cobalt destroys cell somata but spares fibers of passage f%]. In addition, histological investigations showed that the pretectal area receives efferents of cortical origin (see below). The modifications of light-evoked responses of retinal ganglion cells, produced by the interruption of corticofugal impulses, were very similar to those reported by Spinelli and Weingarten [50] These authors combined clicks or tactile stimuh with flashes. This pairing resulted in an alteration of the light-evoked responses. What is particularly relevant to the present investigation is the fact that the addition of a non-visual stimulus shortened the latency of the flash responses. Hence, in this regard results in cat and rat seem to be comparable. Spine& and Weinga~en 1501 hypothesized that combining auditory or somesthetic stimuli with light stimuli increased the animal’s awareness and directed its attention to the visual stimuli. They suggested that changes in the flash-evoked responses were brought about by retinopetal fibers.

lt appears, however, that retinopetal cells are not abundant. We have already mentioned that Itaya [20] counted about 7 to 8 cells per brain. Our experiments (current investigations in our laboratory) seem to indicate that indeed they constitute a very small ovulation of the optic nerve fibers (less than 2%). Estimates of the same order of magnitude have been reported in lampreys (2%) [.57] and monkeys (1%:) tf,37]. Reperant et nf. [44,46] counted no more than 10 retinopetai fibers per retina in both of these species. In this regard, it is pertinent to ask how such a small population of cells can influence about one third of the retinal ganglion cells. A solution to this apparent paradox may be the suggestion that the retinal endings of centrifugal axons arborize extensively in such a way that one axon establishes direct or indirect contacts with a large number of ganglion units. In birds, Maturana and Frenk [29] have shown that centrifugal axons collateralize extensively, expanding several hundreds of microns and synapsing with both ganglion and amacrine ceils. Since the processes of the tatter type of cells extend laterally for considerable distances and thro~ghou~ the inner plexiform layer, it can be expected that the cent~fugal impact reaches all areas of the retina. In addition, the ~n~uence of centrifugal endings of the “divergent type” [29] may be further amplified by amacrine cells, which in turn contact numerous ganglion units. To our knowledge, no such studies are available in mammals. Brooke c,r al. [S], however. observed degenerating presynaptic endings of presumably centrifugal origin throughout the inner plexiform layer. In monkeys, Reperant and Gallego [46] reported simiiar data and suggested that one retinopetal axon through its collaterals may cover one quadrant of the retina. Thus, all these investigations point toward rather diffuse and extensive arborizations of centrifugat axons within the inner plexiform layer, Interestingly, Matu~na and Frenk 1301 demonstrated in pigeons that retinopetal axons ~ontactedganglion cells in the

vicinity of the axon hillock. This is a strategic site. since it is the trigger zone from which action potentials propagate. Such an organization provides an advantage because very few synaptic contacts may suffice to effectively modulate the firing sequence of spikes. Since it would be unreasonable to assume the existence of direct projections from the cortex to the eye, one has to postulate a reiay between both structures. From our results and from data available in the literature, the pretectal area [20,21] and/or the area around the nucleus oculomotorius [ 181may form the link between the cortex and the retina. The cortex can reach the pretectum directly or via the superior cotlicuius, Both types of pathways have been disclosed in various mammals, including rats [36,39,41, 491, rabbits [12, 13, 521, hamsters [19], and cats [22]. interestingly, in the opossum corticofugal and retinofugal terminals project to the same pretectal area 1271. Finally, although a detailed anafysis of pretectal-tectal connections is lacking for the rat. such a pathway has been disclosed in several species (2, 15. 171. in conclusion, the visual cortex is likely to influence the functioning of retinal ganglion cells via a multisynaptic pathway involving the pretestal complex.

What would be the purpose of modulating retinal output by a co&o-pretecto-retinal pathway? As mentioned previously, the ON-. OFFand suppressed-by-light cells are most affected by cryoblockade. To our knowledge, no systematic information concerning retinal ganglion cell receptive field properties is available for the rat [6, 16, 421, and it is beyond the scope of the present paper to deal with this issue. In cats and monkeys, however, this heterogenous group of cells has a few common response properties. These cells often Iack an excitatory surround and respond with excitation or inhibition to any invasion of their activating area by a visual stimulus (diffuse or localized) 1511. They seem to be p~~icularIy suited to signal unspecific modi~~ations occurring in the visual field, such as diffuse brightening or dimming. Thus it is possible that the retinopetal system is associated with attention mechanism, which arouse the animal’s awareness to any disruption of the visual background 1141. This suggestion is compatible with Ebbesson and Meyer’s claim [lo] that the development of the retinopetal system is correlated with the ability of fishes to perform eye movements. A growing number of reports of comparative neurobiology point to a close association between the retinopetal system and nuclei involved in oculo-motor functions 12,251. For instance, retinopetal cells have been identified in the tectaf and pretectal complex of monkeys 1377, rats [20,21], various fishes FlO, 35, 551 and amphibians ]I I]. The retinopetal system has been described best [31] in birds, and centrifugal neurons have been revealed in the isthmo-optic nucleus. This nucleus, however, receives an input from tectal areas. Our proposal is in line with Miles’ [31] suggestion that in birds the retinopetal system operates in such a fashion as to permitting rapid and provide “a dynamic adaptation,” phasic Inodi~cations of sensitivity depending on eye/head movements and/or background-target differences in light intensities. Other alternatives, however, cannot be ruled out. Recently, Trejo and Cicerone fS4f showed that cells of the pretectal olivary nucleus are associated with pupif reflexes to light. With this in mind. a fascinating hypothesis emerges.

CORTICO-RETINAL

INFLUENCE

31

Since the pupil reflex changes the amount of light that strikes the retina in a rather rapid fashion, it follows then that the retina must be capable of an immediate adaptation to meet the new

light

conditions.

Since

chemical

processes

of light

or

dark adaptation are too slow to respond to these rapid changes, the visual cortex may set new threshold sensitivities

using

neuronal

mechanisms

via the

ACKNOWLEDGEMENTS

The authors are indebted to professors S. K. Itaya, F. Lepore and M. Anctil for their helpful criticism and suggestions, and to P. Heinerman and M. Von Griinau for improving the English. This research was supported by NSERC (Canada) and FCAC (Qutbec) grants to SM.

pretectum.

REFERENCES 1. Benevento, L. A. and J. H. Fallon. The ascending projections of the superior colliculus in the rhesus monkey (Mrrctrc,tr mullrrtftr). I Coni!, Nc,uro/ 160: 339-362, 1975. 2. Berman, N. Connections of the pretectum in the cat. J Cor?rp Ncirrol 174: 227-254, 1977. 3. Branston, N. M. and D. G. Fleming. Efferent fibers in the frog optic nerve. E.rp Nrurol 20: 61 l-623, 1977. 4. Brindley, G. S. and D. I. Hamasaki. Histological evidence against the view that cat’s optic nerve contains centrifugal fibers. .I Physiol (Lmd) 184: 444-449, 1966. 5. Brooke, R. N. L.. D. J. de Downer and T. P. S. Powel. Centrifugal fibers to the retina in the monkey and cat. Ntrruw 207: 1365-1367. 1965. 6. Brown, J. E. and J. A. Rojas. Rat retinal ganglion ceils. Receptive organization and maintained activity. J Nc~roplr~.sio/ 28: 1073-1090. 1965. Cellule 9: 119-229. 1892. 7. Cajal. J. La retine des vertebres. 8. Cowan. W. M. Centrifugal fibers to the avian retina. Br Mccl B/r// 26: 112-118, 1970. fibers in the 9. Cowan. W. M. and T. P. S. Dowell. Centrifueal avian retina. Proc, R Sot, Lo& [Bid] 158: 232-252. 1963. S. 0. E. and D. L. Meyer. Efferents to the retina have 10. Ebbesson, multiple sources in teleost fish. .Scic,/icc, 214: 924-926, 1981. B. and W. Himstedt. Pretectal neurons project to the 11. Fritzsch, salamander retina. Nertrosc,i Lclt 24: 13-17, 1981. 12. Fukuda. Y., M. Sugitani and K. Iwama. Flash-evoked responses of two types of principal cells of the rat lateral geniculate body. Brtritr Rcs 57: 20%212, 1973. 13. Giolli. R. A., L. C. Towns. T. T. Takahashi. A. N. Karamanlidis and D. D. Williams. An autoradiographic study of the projections of visual cortical area I to the thalamus. pretectum and superior colliculus of the rabbit. J C’ofrip Nrrrrol 180: 743-752. 1978. 14. Goldberg, M. E. and R. H. Wurtz. Activity of superior colliculus in behaving monkey. II. Effect of attention on neuronal responses. J Ncrrrophy.tiol 35: 56&574. 1972. 15. Graham, J. An autoradiographic study of the afferent connections of the superior colliculus in the cat. J Camp Ncr~rol 173: 629-654. 1977. 16. Hale, P. T., A. J. Sefton and B. Breher. A correlation of receptive field properties with conduction velocity of cells in the rat’s retino-geniculo-cortical pathway. Erp Brtrit7 Rrs 35: 425-442, 1979. 17. Holstege. G. and H. Collewijn. The efferent connections of the nucleus of the optic tract and the superior colliculus in the rabbit. .I C’r~r~p Ncrrrol 209: 139-175. 1982. 18. Hoogland. P. V.. A. Vanderkrans. F. D. Koole and H. J. Groenewegen. A direct projection from the nucleus oculomotorius to the retina in rats. Nerrrosc,i Lctt 56: 323-328. 1985. 19. Huerta, M. F. and J. Harting. The Mammalian superior colliculus: studies of its morphology and connections. In: C‘CUUprrrtrtir’c, Nrr,ro/o~~ (!f/hc, Optic, Tectrrnr, edited by H. Vanegas. New York: Plenum Press. 1984. DD. 687-773. 20. Itaya. S. K. Retinal efferents fromthe pretectal area in the rat. Brtrin Rrs 201: 436441, 1980. 21. Itaya. S. K. and P. W. Itaya. Centrifugal fibers to the rat retina from medial pretectal area and the periaqueductal grey matter. Brtri~ Rr.,. in press.

S., J. M. Sprague and K. Niimi. Corticofugal pro22. Kawamura, jections from the visual cortices to the thalamus, pretectum and superior colliculus in the cat. .I Con~p Neural 158: 33%362,

1974. 23. Koroleva, V. I. and J. Bures. Blockade of cortical spreading depression in electrically and chemically stimulated areas of cerebral cortex in rats. E/c,c,troc,n(.c,phrr/oRr C/in Ncurophysiol 48: l-15, 1980. 24. Leao, A. A. P. Spreading depression of activity in the cerebral cortex. .I Nrurophysiol 7: 359-390, 1944. 25. Leichnetz. G. R. Connections between the frontal eye field and pretectum in the monkey. An anterogradehetrograde study using HRP gel and TBM neurochemistry. .I C‘ornp Nrurd 206: 227-242. 1982. 26. Lent, R. The organization of subcortical projections of the hamsetter’s visual cortex. .I C’omp Ncurol 206: 227-242. 1982. 27. Linden, R. and C. E. Rocha-Miranda. The pretectal complex in the opossum: Projections from the striate cortex and correlation with retinal terminal fields. &win Rcs 207: 267-278, 1981. 28. Malpeli, J. G. and B. D. Burch. Cobalt destroys neurons without destroying fibers of passage on the lateral geniculate nucleus of the rat. Nerrrosc,i Lctt 32: 29-34. 1982. P. L. Centrifugal actions on amacrine and gangl29 Marchiafava. ion cells in the retina of the turtle. J Phv.rio/ ILondJ 255: l37155, 1976. H. R. and S. Frenk. On the synaptic connections of 30 Maturana. centrifugal fibers to the retina in the pigeon. .Sc,ic,nc,c150: 359361, 1965. 31 Miles, F. A. Centrifugal control of the avian retina. Brtriu Rrs 48: 65-156, 1972. 32 Molotchnikoff, S., F. Tremblay and F. Lepore. The role of the visual cortex in response properties of lateral geniculate cells in rats. Exp Bruit1 Rc.7 53: 223-232, 1984. 33. Molotchnikoff. S. and F. Tremblay. Influence of the visual cortex on responses of retinal ganglion cells in the rat. J Nrrtrosci Kc.\ 10: 397-409, 1983. 34. Molotchnikoff, S. and F. Tremblay. The cortico-retinal path: the implication of the pretectum. f~r,c,st Ophthdmol Vi.r .Ci (Suppl) 24: 224, 1983. 35. Munz, H. and B. Claas. Centrifugal innervation of the retina in cichlid and poecilid fishes: an HRP study. Nrurosci Left 22: 223-226, 1981. 36. Nauta. W. J. H. and V. M. Bucher. Efferent connections of the striate cortex in the albino rat. .I Cc~rnp Ncord 100: 257-286, 1954. 37. Noback, C. R. and F. Mettler. Centrifugal fibers to the retina in the Rhesus monkey. Brtrin Behtrl, EIYI/ 7: 382-399, 1973. 38. Ogden, T. E. and K. T. Brown. Intraretinal responses of the Cynamolgus monkeys to electrical stimulation of the optic nerve and retina. .I Neurophysiol 27: 682-705. 1964. 39. Olavarria, I. and R. Van Sluyters. The projection from striate and extra-striate cortical areas to the superior colliculus in the rat. Et-p Brcrin Rrs 41: 184187. 1981. 40. Palkama. A. and H. Sipponen. Light and electron histochemical observations on cholinesterase-containing centrifugal fibers in the optic nerve of the rat. Am Esp Bid Fenn 46: 564567, 1968. 41. Parnavelas, J. G.. A. Chatzissavidov and R. A. Burne. Subcortical projections to layer I of the visual cortex area 17 of the rat. Ev/> Brrritr Rcs 41: 184187, 1981,

~OLOTCHNIKOFF

32

42. Partridge, L. D. and I. E. Brown. Receptive fields of rat retinal ganglion cells. Vision Rcs 10: 455-460, 1970. 43. Pelligrino, L., A. S. Pelligrino and A. J. Cushman. A S’tc,r<,
5 1. Stone. c ‘/us.sJykrJrictrl

53.

54.

55.

56.

57. 58.

~rrrrJ/~ri of

~rwr~ssin~g

Kcfifd

&JtJ~hJ

in

riJc C’ciis

E’;.si~u crrrd

i il.5

.s,iwv1i. Inipwr

I Ire 01,

r/J<,

New York: Plenum Press, 1983. Swadlow, H. A. Efferent systems of primary visual cortex: a review of structure and function. BrrrirrRrs Rev 6: l-74, 1983. Terubayashi. H., H. Fujisawa and Y. Ibata. Hypothalamoretinal centri-fugal projection in the dog. lL’c,r,n~.~(~i i.c,fl 40: 1-6. 1983. Trejo, L. J. and C. M. Cicerone. Cells in the pretectal olivary nucleus in the pathway for the direct light reflex of the pupil in the rat. Brtritt Rr.5 300: 4%62. 1984. Tremblay. F. Influences comparCes des voies corticog&ticulCes et co~ico-r~tiniennes chez le rat. Ph. dissertation. Universitk de Montrkaf. 1985. Uchiyama, H. and H. Ito. Fiber connections and synaptic organization of the preoptic retinopetal nucleus in the filefish (B~ll;.sti&<, ~c+ct.s~<+).&TJi~t Rcs 298: 1l-24, 1984. Van Hasselt, P. The centrifugal control of retinal function. Onhthnmot Rc.r 4: 291320. 1972-73. Vksselkin, N. P., J. Reperant, N. R. Kenigfest, D. Miceli. T. V. Ermakova and J. P. Rio. An anatomical and electrophysiological study of the centrifugal visual system in the lamprey (Llcr/?pctrtr ./lu\~itr/itis). Hrrrin Rm 292: 4 I-56. 1984. Nrurohiokqp

52.

J.

AND TR~~BLAY

of‘

Vi.sioJt.