- Email: [email protected]
The conduction velocity and central projections of retinofugal fibers in the rabbit
Brain Research
Bulletin,
Vol. 5, pp.
687-692.
Printed in the U.S.A.
The Conduction Velocity and Central Projections of Retinofugal Fibers in the Rabbit S. MOLOTCHNIKOFF’
AND I. LESSARD
Dkpartement de Sciences biologiques, Universite’ de Montreal C.P. 6128, Succursale A, Mont&al H3C 3J7 Received
21 June 1980
MOLOTCHNIKOFF, S. AND I. LESSARD. The conduction velociry and central projections of retinofugalfibers in the BRAIN RES. BULL. 5(6)687-692, 1980.-In anesthetized and paralyzed rabbits, action potentials were elicited antidromically following electrical stimulation of optic tract terminals at the geniculate level (LGN). The conduction velocity spectrum extended from 7 m.s-t to 34 m.s-I. The distribution of conduction velocities indicated four major modes at 10, 18, 22 and 26 m.s+. Antidromic compound action potentials exhibited good correlation between the conduction latency and the major modes of the distribution histogram. These results suggest that the rabbit optic tract is composed of four classes of fibers varying in their conduction velocities. The central projections of retinofugal axons were studied with electrical stimulations of the Superior Collicuhts (CS) and the LGN while recording from the same optic tract fiber. Antidromic spikes could be elicited from all conduction velocity groups and 71% of axons responded to both sites of stimulation. This finding indicates that most retinofugal fibers branch to and innervate both CS and LGN. Further, there is a tendency for fast-conducting axons to have their receptive field located eccentrically relative to the optic axis of the eye. rabbit.
Retinofugal axons
Conduction velocity
Superior colliculus
THE visual neuronal message leaves the retina and is conducted to brain centers via the axons of retinal ganglion cells. These axons form the optic nerves and optic tracts. In most mammals the retinofugal fibers may be grouped into several (three in most species) functionally distinct classes [4, 5, 7, 13, 24, 271. Several physiological criteria are adopted to establish the classification: e.g. receptive field properties, retinal distribution, central projections and firing patterns. However, the most easily comparable criterion of classification from species to species is conduction velocity since it depends on few factors. Essentially, these are the diameter of the fiber and the response latency, whereas the other physiological criteria depend upon retinal complexity, ambient light conditions, stimulus parameters, etc . . . although in many investigations, good correlations between receptive field properties and conduction velocity have been found. The only mammal in which these classes of retinal ganglion cells have not yet been demonstrated is the rabbit. This stems from several difficulties specific to this animal. First, its retina is sufficiently complex to yield up to fifteen physiologically different ganglion cells [l]. Second, in most studies recordings are made from ganglion cell somata [ 1,241 and because axons are randomly myelinated in their intraretinal portion [8], measurements of conduction velocities do not allow a clear separation of different classes. To shed some light on the organization of the rabbit optic nerve, we recorded antidromic action potentials generated in the optic tract axon by applying electrical pulses in the vic-
Lateral geniculate nucleus
inity of the lateral geniculate colliculus.
Vision
nucleus
o 1980 ANKHO
and of the superior
METHOD
Rabbits were anesthetized with urethane and paralysed with Gallamine triethiodide. Local anesthetic (Xylocaine) was applied at pressure and surgical sites. Animals were fixed in a stereotaxic frame which was modified to give a clear visual field. Temperature and EKG were continuously monitored. Electrical stimuli were delivered to the lateral geniculate (LGN) in most animals. In addition in four animals, the superior colliculus (CS) was also stimulated. Electrical shocks were delivered through three stainless steel wire electrodes (dia.: 100 CL)insulated to their tips, and inserted in a vertical plane 1 mm apart at Sawyer et al. coordinates (LGN; P: 4-5 mm, L: 6-8 and H: 8-10 mm, close to the optic tract; CS. P: 9-10 mm, L: l-3 mm, H: 4-5 mm) [21]. Receptive fields were hand-plotted on a tangent screen located 12 cm from the eye. The screen was fixed on a perimeter centered in the optic axis of the rabbit’s eye. A +14 to + 15 D lens covered the cornea to correct the eye optics and prevent the cornea from drying. Spike activity from optic tract axons were recorded with tungsten microelectrodes varnished with Insl-X and lowered just behind the optic chiasma. Unit activity was amplified and displayed in the usual way. Stimulating sites were marked with electrolytic lesions (electrode +) for histological examination.
‘Send reprint requests to: S. Molotchnikoff.
Copyright
Rabbit
International Inc.-0361-9230/80/060687-05$01.10/O
688
MOLOTCHNIKOFF
AND LESSARD
FIG. 1-A. Frequency histogram of optic tract conduction velocities. Fig. 1-B: antidromic compound action potential. Superimposed traces following three shocks to the optic tract axon terminals at the geniculate level. The negative waves are identified as N,, N,, N:, and N,. The conduction latencies of these negative waves are indicated with black dots above the histogram.
RESULTS Conduction latencies of antidromic spikes evoked by stimulating the optic tract in the vicinity of the LGN were from 0.5 to 2.7 m.s. Since the conduction distance varied slightly from animal to animal, the latencies were converted into conduction velocities. Figure 1-A shows the distribution of conduction velocities for axons excited by the electrical pulses. The calculation of conduction velocities was calculated by dividing the distance between the stimulating and recording sites as measured with a thread aligned along the optic tract. It was estimated to be 16 mm, which is in agreement with a previous report [24]. It is apparent from Fig. 1-A that the distribution is multimodal with two major peaks centered at 18 and 22 m.s-‘. In addition smaller peaks emerge at 10, 26 and 32 m.s-‘. These modes suggest that optic tract fibers may be divided into several classes. Antidromic compound action potentials evoked by stimulating axon terminals at the geniculate level were recorded at the same site and exhibited four main negative components: N1, NP, N3 and N,. The mean conduction latencies as measured at the peak of negative waves were 0.5 f 0.06; 0.86 ? 0.09; 1.19 -+ 0.06 and 2.6 2 0.28 msec (N= 10). The respective velocities were 28.5; 18.6; 13.4 and 7 m.s-‘. A typical example of an antidromically-evoked compound action potential is shown in Fig. 1-B. These data are compared with the major modes of the distribution of conduction velocities of unit
responses (black dots, Fig. 1-A). In spite of the different nature of the responses, it can be seen that cells from each conduction velocity group are well correlated with conduction velocities as derived from compound responses. The slight shift in Nz and N3 components may be attributed to the fact that latencies from compound action potentials were measured at the peak of several superimposed waves with similar time courses, whereas the distribution of the histogram was based on peak latency of a unit response. Examples of action potentials belonging to each conduction velocity class are illustrated in Fig. 2. These fibers were recorded in a single penetration and, therefore, with the same electrode. These results indicate that the rabbit optic nerve contains at least four classes of axons which can be dissociated according to their conductance velocities and, consequently, their diameter. Collateralization In a recent report [ 161, it was suggested that all classes of retinofugal axons project to both CS and LGN. This assumption was based on the similarity of histograms of latency distribution of spikes evoked at the CS and the LGN following optic nerve stimulation. An unambiguous demonstration that the same retinofugal fibers indeed innervate the CS and the LGN is to stimulate these two structures and
RETINOFUGAL
2
1
FIBERS:
+
CONDUCTION
AND PROJECTIONS
IL--
689
The mean difference in latency of spikes evoked by collicular and geniculate stimulation was 0.5 msec. This value reflects the average conductance time between the axonal terminal at the CS and the bifurcation loci(CT,,), since the retinofugal axons bifurcate just prior to entering the LGN. Then, CTQb = CD,dCVQb (1) where CDcb is the conductance distance from the colliculus to the bifurcation site. This distance is estimated to be approximately 5 mm according to stereo&&c coordinates [21]. CV,, is the conductance velocity from the CS to the bifurcation. Resolving equation (1) for CVcb gives: CVci, = CD&T,, = 10 m.s-l, which is the average conductance velocity for the retino-collicular collateral. Since the average conductance velocity (CV,,) from the bifurcation to the recording site was 20 rt: 6.4 m.s-’ as derived from the values of Fig. I-A, the ratio of conductance velocities between retin~collicul~ collaterals (CV,,) and optic tract fibers gives the approximate ratio of diameter of each branch. This ratio is equal to 0.5. We assume that the stimulus utilization times are about equal in both sites. In D of Fig. 3 is shown an example of the compound action potential evoked at the optic tract level following electrical stimulation of the CS. The morphology is very similar to that obtained when the geniculate is stimulated (compare with Fig. 1-B). In both cases, four negative waves (N, to N,) are clearly present. Additional information about retinofugal projections are provided by Fig. 4. This figure is a plot of the latency of antidromic spikes evoked from optic tract terminals at the geniculate level against the latency of the antidromic spike in the same axon evoked by st~ulating optic tract terminals at the collicular level. Although the correlation coefficient is very high (r=0.94), there is a considerable scatter around the best fitting line (Y = 1.10 X + 4.8). This scatter suggests that the ratio of diameters between the retino-geniculate and retino-collicular axons is not very constant. For instance, as shown in tracings 1,2 of Fig. 3-B, in a few Ebers, the difference in latencies between the antidromic spike evoked from collicular and geniculate stimulation was less than 0.1 msec despite the greater distance travelled by the spike originating from the collicular terminal. Thus, the diameter of the collicular branch seems to be of comparable size or even larger than the diameter of the geniculate branch. C~~~ffctio~ and Ecce~t~jc~ty
FIG. 2. Examples of antidromic action potentials elicited in optic tract fibers following electrical stimulation of their terminals at the geniculate level. (OTS, arrow: moment of stimulus application.) Trace 1: fast conducting; Trace 2: intermediate-fast conducting; Trace 3: inte~edia~-slow conducting; Trace 4: slow conducting. All traces are from the same penetration.
record the elicited antidromic spikes from the same axon. Figure 3 shows examples of antidromic action potentials evoked by electrical pulses applied at the levels of the superior colliculus and the lateral geniculate nucleus [211. The examples given are chosen from each conduction velocity class. From a total of 31 tested cells, 22 (71%) responded to stimulation of both sites. It seems that the vast majority of retinofugal fibers bifurcate and provide collaterals to the CS and the LGN. Only three cells responded exclusively to stimulation of the colliculus and six cells were activated from the geniculate level, while no responses were elicited from the collicular level.
In several species, including rabbits [12], it has been demonstrated that axons from peripheral retina appear to belong to the fastest velocity group. It follows that the receptive fields of the fastest conducting cells should tend to be located eccentrically, relative to the optical axis that is the center of the retina. The conductance velocity is plotted against receptive field eccentricity in Fig. 5. The best fitting line, Y = 1.01 X + 30.5, clearly indicates that there is a tendency for the slowest conducting fibers to have their receptive field positioned in the center of the visual tield. However, the very weak correlation coefficient (r=0.24) also indicates that deviation from the regression line is considerable. Thus one cannot predict with high probability the relative field location from the conductance velocity. This weak correlation might be due to the large diversity of trigger features of retinal ganglion cells. By contrast, cats and monkeys seem to have a relatively homogeneous retina since the great majority of their ganglion cells have the simplest concentric receptive fields. Since diameters, and thus conductances, are related to the soma size the data of Fig. 5 complement recent histological reports [17,28], indicating that in rabbits’ retina
690
MOLOTCHNIKOFF
-
AND LESSARD
1Ms
FIG. 3-A. Examples of the antidromic spikes elicited at the same optic tract fibers following electrical stimulation of their terminals at the geniculate level (LGN) (Traces 1,3,5,7) and of the superior colliculus (CS) (Traces 2,4,6,8). Note in B: The peak latency of the spike from collicular stimulation (Trace 2) is about equal to the latency of the spike evoked from geniculate stimulation (Trace 1). C: Antidromic compound action potentials evoked from geniculate stimulation. Trace 1: from collicular stimulation. Trace 2: Note three negative waves. D: Antidromic compound action potentials from collicular stimulation at slower sweep speed (3 superimposed sweeps). Note four negative waves: compare with Fig. 1-B.
small and large ganglion cells are uniformly distributed across the retina. In addition, the distribution of soma diameters from various retinal locations is unimodal. DISCUSSION
From antidromic compound potentials (Figs. 1-B and 3-C) of the rabbit optic tract, it is possible to distinguish four negative peaks which reflect four conductance velocity classes. Granit and Marg [8] reported that they occasionally recorded live peaks. In our experiments, no additional peak was found. Thus, if a fifth population exists, it does not contribute significantly to the antidromic compound action potential. Semm [24] reported three latency groups; however, since in this latter investigation recordings were from
the intraretinal segment, it is likely that a random myelination of ganglion cell axons obscured a clear subdivision, particularly since there is considerable overlap in latencies of response for fibers belonging to each class [ 1, 16, 181and this study (Fig. 1-A). In this investigation, antidromic compound action potentials indicated that the three fastest groups (N,, N2 and NJ have average conductance velocities ranging from 25 to 15 m.s-I. The fourth group (NJ is relatively slow, 7 m.s-‘, and clearly segregates. The fourth component (NJ has a slower time course if it is compared to that of the first three waves. It is suggested that action potentials travelling along the slowest fibers are not synchronous. Two explanations may be offered. Axon terminals are located remotely from the tips of the stimulating electrodes. It could be
RETINOFUGAL
FIBERS:
CONDUCTION
691
AND PROJECTIONS .
. .
10’
. 0
/,,,,,,,,,,,,,,,,,/,.~.ll~ 2
.1
.6
8
.
.
.
.
Pi= 43
. . :
. 12
16
. 20 m.s-’
24
28
32
FIG. 5. Latency response of spikes evoked from geniculate stimulation plotted as a function of eccentricity.
.24 0
4
.
.
.e
1.0
1.2 Ill”<
1.4
1.6
I.(1
*.o
2.2 2.4 LAT. O.l.(Gl
FIG. 4. Latency response of spikes evoked from geniculate (G)
stimulation plotted as a function of latency response of spikes evoked from collicular stimulation (CS). Spikes evoked from the same fiber.
suggested that N, originates from spikes travelling in the orthodromic direction, and thus providing an input, for the pupilloconstrictor reflex. This hypothesis is at odds with findings reported for cats [26,27], where it has been shown that the slowest conducting group (W fibers) project exclusively to the midbrain. If it is so, the N4 waves in the rabbit could not be evoked by stimulating the LGN. Rabbit conductances are comparable to those obtained for other mammals although most of them have three groups of fibers, based upon conductances. In rats, conductances range from ~15 to 2.5 m.s-1 [5,9, 231 and three groups were identified in these animals [23]. Golden hamsters also exhibit similar values: >15 to 5.4 m.s-’ [193. Monkey conductances have been estimated to be >14 to 7 m.s-’ [13]. In cats three classes of ganglion cells have been dissociated [27] with a wide range of extra-retinal velocity: 30-40 to 2.2 m.s-’ [20]. In a recent report, Freeman [2] has suggested four conduction latency groups with a conduction velocity of 46 m.s-* for the fastest group. The optic nerve of the Bush-tailed possum has four conducting latency groups (>13 to 3 m.s-‘) [3,4] similar to rabbits. A second conclusion emerging from this investigation is that it is very likely that the majority of fibers that form these four conduction velocity groups branch off and innervate both the CS and the LGN as is suggested by Fig. 3. This hypothesis is supported by several findings. Stimulation of the LGN and CS evokes antidromic spikes in the same fibers of the optic tract. In addition, stimulating the superior colliculus or the geniculate evokes antidromic compound potentials of similar waveform. Further, the latency distribution of post-synaptic spikes evoked at the LGN and CS levels following optic nerve stimulation revealed similar histograms [13]. However, as shown by traces of Fig. 3-B, it seems that in some fibers the diameter of the retino-collicular collateral is larger than the diameter of the retino-geniculate branch after the bifurcation. One assumes that the stimulus utilization time is about equal in both sites as well. However, some methodological criticisms must be considered. For example, it is possible that collicular stimulus excited optic tract fibers directly at a point before their entry to the LGN. This is unlikely because latencies of antidromic
compound potentials from the CS to the optic tract were longer than from the LGN to the optic tract. In addition, applied shock to LGN may stimulate optic fibers passing to midbrain structures since bifurcation of optic tract fibers occurs close to the LGN. This eventuality cannot be excluded but it is thought to be unlikely [ 151. If there were no branching of optic tract axons and the LGN stimulus spread to collicular fibers directly, then the response in the optic tract to an LGN stimulus should be stronger than to a collicular stimulus [23] but in fact it has a smaller amplitude (Fig. 3-C). Further, because it is impossible to be sure that all axons entering the CS were stimulated, the real number of fibers that bifurcate may be higher than our estimate of 71%. The functional implication of this dual projection of most retinofugal fibers is that the majority of ganglion cells supply both the CS and the LGN with same neuronal messages. Hughes [ 1l] has suggested that numerous retinal inputs converge on a collicular unit. However, previous studies on receptive field characteristics of the superior colliculus [14] and the lateral geniculate [25] have suggested that the rabbit retina projects unequally to each structure. It may be that each class of ganglion cells has at least a few units which project to both sites but the proportion of fibers within each class that bifurcate may vary from class to class. The answer to this question deserves further investigation. It is of interest to point out that Sefton [23] in a similar study conducted on rats concluded that in these animals most if not all optic tract fibers bifurcate. By contrast, in cats [10,26] and primates [22], there is a clear segregation of retinofugal fibers. Fast fibers (Y) project to both sites, slow fibers (W) project to the colliculus and to the C lamina of the geniculate, while fibers whose conduction velocity is intermediate (X) seem to belong exclusively to the retino-geniculate system. In our study, axons with intermediate conductances are those which are both the most numerous and show antidromic spikes from collicular and geniculate stimulation. This is one of the fundamental differences in the organization of the visual systems between rabbits and rats [23] on the one hand and cats and monkeys on the other. ACKNOWLEDGMENTS
We thank J. L. Verville and H. Farid for their valuable assistance in the nrenaration of the figures and L. Pelletier for histology. This study was supported by financial contributions from the Ranting Research Foundation. Universite de Montreal and NSERC Canada: A 6943, granted to S.M.
692
MOLOTCHNIKOFF AND LESSARD REFERENCES
I. Caldwell, J. H. and N. W. Daw. New properties of rabbit ganglion cells. 1. Physiol., Lond. 276: 257-276, 1978. 2. Freeman, B. Myelin sheath thickness and conduction latency groups in the cat optic nerve. J. camp. Neurol. 181: 183-l%, 1978. 3. Freeman, B. The retinal origins of the optic nerve conduction latency groups in the Bush-tailed possum. Trichosurus vulpecula. J. camp. Neurol. 179: 753-760, 1978. 4. Freeman, B. and C. R. R. Watson. The optic nerve of the Bushtailed possum, Trichosurus vulpecula: fiber diameter spectrum and conduction latency groups. J. camp. Neurol. 179: 739-752, 1978. 5. Fukuda, Y. A three-group classification of rat retinal ganglion cells: Histological and physiological studies. Brain Res. 119: 327-344, 1977. 6. Fukuda, Y. and J. Stone. Retinal distribution and central projections of Y-, X- and W-cells of the cat’s retina. J. Neurophysiol. 37: 749-772, 1974. 7. Giolli, R. H. and M. D. Guthrie. The primary optic projections in the rabbit. An experimental degeneration study. J. camp. Neurol. 136: 9Fl26, 1969.
8. Granit, R. A. and E. Marg. Conduction velocities in rabbit’s optic nerve and some observations on antidromic retinal spikes. A. J. Ophrhal. 46: 223-231, 1958. 9. Hale, P. T., A. J. Sefton and B. Dreher. A correlation of recep-
tive field properties with conduction velocity of cells in the rat’s retina-geniculo-cortical pathway. Expl Brain Res. 35: 425-442, 1979.
10. Hayashi, Y., 1. Sumitomo and K. Iwama. Activation of lateral geniculate neurons by electrical stimulation of superior colliculus in cats. Jap. J. Physiol. 17: 638-651, 1967. 11. Hughes, A. Topographical relationships between the anatomy and physiology of the rabbit visual system. Docums Ophthal. 30: 33-159, 1971. 12. Lederman, R. J. and W. K. Noel]. Fast-fiber optic nerve. Vision Res. 8: 1385-1398, 1968.
system of rabbit
13. Marocco, R. T. Sustained and transient cells in monkey lateral geniculate nucleus: conduction velocities and response properties. J. Neurophysiol. 39: 340-353, 1976. 14. Masland, R. H., K. L. Chow and D. L. Stewart. Receptive field characteristics of superior colliculus neurons in the rabbit. J. Neurophysiol.
34: 141156,
1971.
15. Molotchnikoff, S. and P. Lachapelle. Evidence of a collicular input to the dorsal lateral geniculate nucleus in rabbitselectrophysiology. Expl Brain Res., in press. 16. Molotchnikoff, S., P. Lachapelle, D. Richard, P. L’Archeveque and I. Lessard. Latency distribution from orthodromic stimulation at the optic nerve in the lateral geniculate nucleus and superior colliculus of rabbits. Brain Res. Bull. 4: 57F581, 1979. 17. Provis, J. M. The distribution and size of ganglion cells in the retina of the pigmented rabbit: a quantitative analysis. J. camp. Neurol. 185: 121-138, 1979. 18. Reuter, J. H. and K.-P. Hoffman. The conduction velocity in retinogeniculate afferent fibers in the rabbit. Neurosci. Letr. 16: 175-179, 1980. 19. Rhoades, R. W. and L. M. Chalupa. Conduction velocity distribution of the retinocollicular pathway in the golden hamster. Brain Res. 159: 396-401, 1978. 20. Rowe, M. H. and J. Stone. Conduction velocity groupings among axons of cat retinal ganglion cells and their relationship to retinal topography. Expl Brain Res. 25: 339-357, 1978. 21. Sawyer, CH., J. W. Everette and J. D. Green. The rabbit diencephalon in stereotaxic coordinates. J. camp. Neurol. 101: 801-824, 1954. 22. Schiller, P. H. and J. G. Malpeli. Properties and tectal projections of monkey retinal ganglion cells. J. Neurophysiol. 40: 428-445, 1977. 23. Sefton, A. J. The innervation
of the lateral geniculate nucleus and anterior colliculus in the rat. Vision Res. 8: 867-881, 1%8. 24. Semm, P. Antidromically activated direction selective ganglion cells of the rabbit. Neurosci. Lert. 9: 207-211, 1978. 25. Steward, D. L., K. L. Chow and R. H. Masland. Receptivefield characteristics of lateral geniculate neurons in the rabbit. J. Neurophysiol.
jq: 139-147,
1971.
26. Stone, J. and Y. Fukuda. Properties of cat retinal ganglion cells. A comDarison of W-cells with X- and Y-cells. J. Neurophvsiol. _
37: 722-748, 1974. 27. Stone, J. and R. B. Freeman. Conduction velocity groups in the cat’s optic nerve classified according to their retinal origin. Expl Brain Res. 13: 489-497, 1971. 28. Vaney, D. I. A quantitative comparison between the ganglion
cell populations and axonal outflows of the visual streak and periphery of the rabbit retina. J. camp. Neural. 189: 215-233, 1980.
Bulletin,
Vol. 5, pp.
687-692.
Printed in the U.S.A.
The Conduction Velocity and Central Projections of Retinofugal Fibers in the Rabbit S. MOLOTCHNIKOFF’
AND I. LESSARD
Dkpartement de Sciences biologiques, Universite’ de Montreal C.P. 6128, Succursale A, Mont&al H3C 3J7 Received
21 June 1980
MOLOTCHNIKOFF, S. AND I. LESSARD. The conduction velociry and central projections of retinofugalfibers in the BRAIN RES. BULL. 5(6)687-692, 1980.-In anesthetized and paralyzed rabbits, action potentials were elicited antidromically following electrical stimulation of optic tract terminals at the geniculate level (LGN). The conduction velocity spectrum extended from 7 m.s-t to 34 m.s-I. The distribution of conduction velocities indicated four major modes at 10, 18, 22 and 26 m.s+. Antidromic compound action potentials exhibited good correlation between the conduction latency and the major modes of the distribution histogram. These results suggest that the rabbit optic tract is composed of four classes of fibers varying in their conduction velocities. The central projections of retinofugal axons were studied with electrical stimulations of the Superior Collicuhts (CS) and the LGN while recording from the same optic tract fiber. Antidromic spikes could be elicited from all conduction velocity groups and 71% of axons responded to both sites of stimulation. This finding indicates that most retinofugal fibers branch to and innervate both CS and LGN. Further, there is a tendency for fast-conducting axons to have their receptive field located eccentrically relative to the optic axis of the eye. rabbit.
Retinofugal axons
Conduction velocity
Superior colliculus
THE visual neuronal message leaves the retina and is conducted to brain centers via the axons of retinal ganglion cells. These axons form the optic nerves and optic tracts. In most mammals the retinofugal fibers may be grouped into several (three in most species) functionally distinct classes [4, 5, 7, 13, 24, 271. Several physiological criteria are adopted to establish the classification: e.g. receptive field properties, retinal distribution, central projections and firing patterns. However, the most easily comparable criterion of classification from species to species is conduction velocity since it depends on few factors. Essentially, these are the diameter of the fiber and the response latency, whereas the other physiological criteria depend upon retinal complexity, ambient light conditions, stimulus parameters, etc . . . although in many investigations, good correlations between receptive field properties and conduction velocity have been found. The only mammal in which these classes of retinal ganglion cells have not yet been demonstrated is the rabbit. This stems from several difficulties specific to this animal. First, its retina is sufficiently complex to yield up to fifteen physiologically different ganglion cells [l]. Second, in most studies recordings are made from ganglion cell somata [ 1,241 and because axons are randomly myelinated in their intraretinal portion [8], measurements of conduction velocities do not allow a clear separation of different classes. To shed some light on the organization of the rabbit optic nerve, we recorded antidromic action potentials generated in the optic tract axon by applying electrical pulses in the vic-
Lateral geniculate nucleus
inity of the lateral geniculate colliculus.
Vision
nucleus
o 1980 ANKHO
and of the superior
METHOD
Rabbits were anesthetized with urethane and paralysed with Gallamine triethiodide. Local anesthetic (Xylocaine) was applied at pressure and surgical sites. Animals were fixed in a stereotaxic frame which was modified to give a clear visual field. Temperature and EKG were continuously monitored. Electrical stimuli were delivered to the lateral geniculate (LGN) in most animals. In addition in four animals, the superior colliculus (CS) was also stimulated. Electrical shocks were delivered through three stainless steel wire electrodes (dia.: 100 CL)insulated to their tips, and inserted in a vertical plane 1 mm apart at Sawyer et al. coordinates (LGN; P: 4-5 mm, L: 6-8 and H: 8-10 mm, close to the optic tract; CS. P: 9-10 mm, L: l-3 mm, H: 4-5 mm) [21]. Receptive fields were hand-plotted on a tangent screen located 12 cm from the eye. The screen was fixed on a perimeter centered in the optic axis of the rabbit’s eye. A +14 to + 15 D lens covered the cornea to correct the eye optics and prevent the cornea from drying. Spike activity from optic tract axons were recorded with tungsten microelectrodes varnished with Insl-X and lowered just behind the optic chiasma. Unit activity was amplified and displayed in the usual way. Stimulating sites were marked with electrolytic lesions (electrode +) for histological examination.
‘Send reprint requests to: S. Molotchnikoff.
Copyright
Rabbit
International Inc.-0361-9230/80/060687-05$01.10/O
688
MOLOTCHNIKOFF
AND LESSARD
FIG. 1-A. Frequency histogram of optic tract conduction velocities. Fig. 1-B: antidromic compound action potential. Superimposed traces following three shocks to the optic tract axon terminals at the geniculate level. The negative waves are identified as N,, N,, N:, and N,. The conduction latencies of these negative waves are indicated with black dots above the histogram.
RESULTS Conduction latencies of antidromic spikes evoked by stimulating the optic tract in the vicinity of the LGN were from 0.5 to 2.7 m.s. Since the conduction distance varied slightly from animal to animal, the latencies were converted into conduction velocities. Figure 1-A shows the distribution of conduction velocities for axons excited by the electrical pulses. The calculation of conduction velocities was calculated by dividing the distance between the stimulating and recording sites as measured with a thread aligned along the optic tract. It was estimated to be 16 mm, which is in agreement with a previous report [24]. It is apparent from Fig. 1-A that the distribution is multimodal with two major peaks centered at 18 and 22 m.s-‘. In addition smaller peaks emerge at 10, 26 and 32 m.s-‘. These modes suggest that optic tract fibers may be divided into several classes. Antidromic compound action potentials evoked by stimulating axon terminals at the geniculate level were recorded at the same site and exhibited four main negative components: N1, NP, N3 and N,. The mean conduction latencies as measured at the peak of negative waves were 0.5 f 0.06; 0.86 ? 0.09; 1.19 -+ 0.06 and 2.6 2 0.28 msec (N= 10). The respective velocities were 28.5; 18.6; 13.4 and 7 m.s-‘. A typical example of an antidromically-evoked compound action potential is shown in Fig. 1-B. These data are compared with the major modes of the distribution of conduction velocities of unit
responses (black dots, Fig. 1-A). In spite of the different nature of the responses, it can be seen that cells from each conduction velocity group are well correlated with conduction velocities as derived from compound responses. The slight shift in Nz and N3 components may be attributed to the fact that latencies from compound action potentials were measured at the peak of several superimposed waves with similar time courses, whereas the distribution of the histogram was based on peak latency of a unit response. Examples of action potentials belonging to each conduction velocity class are illustrated in Fig. 2. These fibers were recorded in a single penetration and, therefore, with the same electrode. These results indicate that the rabbit optic nerve contains at least four classes of axons which can be dissociated according to their conductance velocities and, consequently, their diameter. Collateralization In a recent report [ 161, it was suggested that all classes of retinofugal axons project to both CS and LGN. This assumption was based on the similarity of histograms of latency distribution of spikes evoked at the CS and the LGN following optic nerve stimulation. An unambiguous demonstration that the same retinofugal fibers indeed innervate the CS and the LGN is to stimulate these two structures and
RETINOFUGAL
2
1
FIBERS:
+
CONDUCTION
AND PROJECTIONS
IL--
689
The mean difference in latency of spikes evoked by collicular and geniculate stimulation was 0.5 msec. This value reflects the average conductance time between the axonal terminal at the CS and the bifurcation loci(CT,,), since the retinofugal axons bifurcate just prior to entering the LGN. Then, CTQb = CD,dCVQb (1) where CDcb is the conductance distance from the colliculus to the bifurcation site. This distance is estimated to be approximately 5 mm according to stereo&&c coordinates [21]. CV,, is the conductance velocity from the CS to the bifurcation. Resolving equation (1) for CVcb gives: CVci, = CD&T,, = 10 m.s-l, which is the average conductance velocity for the retino-collicular collateral. Since the average conductance velocity (CV,,) from the bifurcation to the recording site was 20 rt: 6.4 m.s-’ as derived from the values of Fig. I-A, the ratio of conductance velocities between retin~collicul~ collaterals (CV,,) and optic tract fibers gives the approximate ratio of diameter of each branch. This ratio is equal to 0.5. We assume that the stimulus utilization times are about equal in both sites. In D of Fig. 3 is shown an example of the compound action potential evoked at the optic tract level following electrical stimulation of the CS. The morphology is very similar to that obtained when the geniculate is stimulated (compare with Fig. 1-B). In both cases, four negative waves (N, to N,) are clearly present. Additional information about retinofugal projections are provided by Fig. 4. This figure is a plot of the latency of antidromic spikes evoked from optic tract terminals at the geniculate level against the latency of the antidromic spike in the same axon evoked by st~ulating optic tract terminals at the collicular level. Although the correlation coefficient is very high (r=0.94), there is a considerable scatter around the best fitting line (Y = 1.10 X + 4.8). This scatter suggests that the ratio of diameters between the retino-geniculate and retino-collicular axons is not very constant. For instance, as shown in tracings 1,2 of Fig. 3-B, in a few Ebers, the difference in latencies between the antidromic spike evoked from collicular and geniculate stimulation was less than 0.1 msec despite the greater distance travelled by the spike originating from the collicular terminal. Thus, the diameter of the collicular branch seems to be of comparable size or even larger than the diameter of the geniculate branch. C~~~ffctio~ and Ecce~t~jc~ty
FIG. 2. Examples of antidromic action potentials elicited in optic tract fibers following electrical stimulation of their terminals at the geniculate level. (OTS, arrow: moment of stimulus application.) Trace 1: fast conducting; Trace 2: intermediate-fast conducting; Trace 3: inte~edia~-slow conducting; Trace 4: slow conducting. All traces are from the same penetration.
record the elicited antidromic spikes from the same axon. Figure 3 shows examples of antidromic action potentials evoked by electrical pulses applied at the levels of the superior colliculus and the lateral geniculate nucleus [211. The examples given are chosen from each conduction velocity class. From a total of 31 tested cells, 22 (71%) responded to stimulation of both sites. It seems that the vast majority of retinofugal fibers bifurcate and provide collaterals to the CS and the LGN. Only three cells responded exclusively to stimulation of the colliculus and six cells were activated from the geniculate level, while no responses were elicited from the collicular level.
In several species, including rabbits [12], it has been demonstrated that axons from peripheral retina appear to belong to the fastest velocity group. It follows that the receptive fields of the fastest conducting cells should tend to be located eccentrically, relative to the optical axis that is the center of the retina. The conductance velocity is plotted against receptive field eccentricity in Fig. 5. The best fitting line, Y = 1.01 X + 30.5, clearly indicates that there is a tendency for the slowest conducting fibers to have their receptive field positioned in the center of the visual tield. However, the very weak correlation coefficient (r=0.24) also indicates that deviation from the regression line is considerable. Thus one cannot predict with high probability the relative field location from the conductance velocity. This weak correlation might be due to the large diversity of trigger features of retinal ganglion cells. By contrast, cats and monkeys seem to have a relatively homogeneous retina since the great majority of their ganglion cells have the simplest concentric receptive fields. Since diameters, and thus conductances, are related to the soma size the data of Fig. 5 complement recent histological reports [17,28], indicating that in rabbits’ retina
690
MOLOTCHNIKOFF
-
AND LESSARD
1Ms
FIG. 3-A. Examples of the antidromic spikes elicited at the same optic tract fibers following electrical stimulation of their terminals at the geniculate level (LGN) (Traces 1,3,5,7) and of the superior colliculus (CS) (Traces 2,4,6,8). Note in B: The peak latency of the spike from collicular stimulation (Trace 2) is about equal to the latency of the spike evoked from geniculate stimulation (Trace 1). C: Antidromic compound action potentials evoked from geniculate stimulation. Trace 1: from collicular stimulation. Trace 2: Note three negative waves. D: Antidromic compound action potentials from collicular stimulation at slower sweep speed (3 superimposed sweeps). Note four negative waves: compare with Fig. 1-B.
small and large ganglion cells are uniformly distributed across the retina. In addition, the distribution of soma diameters from various retinal locations is unimodal. DISCUSSION
From antidromic compound potentials (Figs. 1-B and 3-C) of the rabbit optic tract, it is possible to distinguish four negative peaks which reflect four conductance velocity classes. Granit and Marg [8] reported that they occasionally recorded live peaks. In our experiments, no additional peak was found. Thus, if a fifth population exists, it does not contribute significantly to the antidromic compound action potential. Semm [24] reported three latency groups; however, since in this latter investigation recordings were from
the intraretinal segment, it is likely that a random myelination of ganglion cell axons obscured a clear subdivision, particularly since there is considerable overlap in latencies of response for fibers belonging to each class [ 1, 16, 181and this study (Fig. 1-A). In this investigation, antidromic compound action potentials indicated that the three fastest groups (N,, N2 and NJ have average conductance velocities ranging from 25 to 15 m.s-I. The fourth group (NJ is relatively slow, 7 m.s-‘, and clearly segregates. The fourth component (NJ has a slower time course if it is compared to that of the first three waves. It is suggested that action potentials travelling along the slowest fibers are not synchronous. Two explanations may be offered. Axon terminals are located remotely from the tips of the stimulating electrodes. It could be
RETINOFUGAL
FIBERS:
CONDUCTION
691
AND PROJECTIONS .
. .
10’
. 0
/,,,,,,,,,,,,,,,,,/,.~.ll~ 2
.1
.6
8
.
.
.
.
Pi= 43
. . :
. 12
16
. 20 m.s-’
24
28
32
FIG. 5. Latency response of spikes evoked from geniculate stimulation plotted as a function of eccentricity.
.24 0
4
.
.
.e
1.0
1.2 Ill”<
1.4
1.6
I.(1
*.o
2.2 2.4 LAT. O.l.(Gl
FIG. 4. Latency response of spikes evoked from geniculate (G)
stimulation plotted as a function of latency response of spikes evoked from collicular stimulation (CS). Spikes evoked from the same fiber.
suggested that N, originates from spikes travelling in the orthodromic direction, and thus providing an input, for the pupilloconstrictor reflex. This hypothesis is at odds with findings reported for cats [26,27], where it has been shown that the slowest conducting group (W fibers) project exclusively to the midbrain. If it is so, the N4 waves in the rabbit could not be evoked by stimulating the LGN. Rabbit conductances are comparable to those obtained for other mammals although most of them have three groups of fibers, based upon conductances. In rats, conductances range from ~15 to 2.5 m.s-1 [5,9, 231 and three groups were identified in these animals [23]. Golden hamsters also exhibit similar values: >15 to 5.4 m.s-’ [193. Monkey conductances have been estimated to be >14 to 7 m.s-’ [13]. In cats three classes of ganglion cells have been dissociated [27] with a wide range of extra-retinal velocity: 30-40 to 2.2 m.s-’ [20]. In a recent report, Freeman [2] has suggested four conduction latency groups with a conduction velocity of 46 m.s-* for the fastest group. The optic nerve of the Bush-tailed possum has four conducting latency groups (>13 to 3 m.s-‘) [3,4] similar to rabbits. A second conclusion emerging from this investigation is that it is very likely that the majority of fibers that form these four conduction velocity groups branch off and innervate both the CS and the LGN as is suggested by Fig. 3. This hypothesis is supported by several findings. Stimulation of the LGN and CS evokes antidromic spikes in the same fibers of the optic tract. In addition, stimulating the superior colliculus or the geniculate evokes antidromic compound potentials of similar waveform. Further, the latency distribution of post-synaptic spikes evoked at the LGN and CS levels following optic nerve stimulation revealed similar histograms [13]. However, as shown by traces of Fig. 3-B, it seems that in some fibers the diameter of the retino-collicular collateral is larger than the diameter of the retino-geniculate branch after the bifurcation. One assumes that the stimulus utilization time is about equal in both sites as well. However, some methodological criticisms must be considered. For example, it is possible that collicular stimulus excited optic tract fibers directly at a point before their entry to the LGN. This is unlikely because latencies of antidromic
compound potentials from the CS to the optic tract were longer than from the LGN to the optic tract. In addition, applied shock to LGN may stimulate optic fibers passing to midbrain structures since bifurcation of optic tract fibers occurs close to the LGN. This eventuality cannot be excluded but it is thought to be unlikely [ 151. If there were no branching of optic tract axons and the LGN stimulus spread to collicular fibers directly, then the response in the optic tract to an LGN stimulus should be stronger than to a collicular stimulus [23] but in fact it has a smaller amplitude (Fig. 3-C). Further, because it is impossible to be sure that all axons entering the CS were stimulated, the real number of fibers that bifurcate may be higher than our estimate of 71%. The functional implication of this dual projection of most retinofugal fibers is that the majority of ganglion cells supply both the CS and the LGN with same neuronal messages. Hughes [ 1l] has suggested that numerous retinal inputs converge on a collicular unit. However, previous studies on receptive field characteristics of the superior colliculus [14] and the lateral geniculate [25] have suggested that the rabbit retina projects unequally to each structure. It may be that each class of ganglion cells has at least a few units which project to both sites but the proportion of fibers within each class that bifurcate may vary from class to class. The answer to this question deserves further investigation. It is of interest to point out that Sefton [23] in a similar study conducted on rats concluded that in these animals most if not all optic tract fibers bifurcate. By contrast, in cats [10,26] and primates [22], there is a clear segregation of retinofugal fibers. Fast fibers (Y) project to both sites, slow fibers (W) project to the colliculus and to the C lamina of the geniculate, while fibers whose conduction velocity is intermediate (X) seem to belong exclusively to the retino-geniculate system. In our study, axons with intermediate conductances are those which are both the most numerous and show antidromic spikes from collicular and geniculate stimulation. This is one of the fundamental differences in the organization of the visual systems between rabbits and rats [23] on the one hand and cats and monkeys on the other. ACKNOWLEDGMENTS
We thank J. L. Verville and H. Farid for their valuable assistance in the nrenaration of the figures and L. Pelletier for histology. This study was supported by financial contributions from the Ranting Research Foundation. Universite de Montreal and NSERC Canada: A 6943, granted to S.M.
692
MOLOTCHNIKOFF AND LESSARD REFERENCES
I. Caldwell, J. H. and N. W. Daw. New properties of rabbit ganglion cells. 1. Physiol., Lond. 276: 257-276, 1978. 2. Freeman, B. Myelin sheath thickness and conduction latency groups in the cat optic nerve. J. camp. Neurol. 181: 183-l%, 1978. 3. Freeman, B. The retinal origins of the optic nerve conduction latency groups in the Bush-tailed possum. Trichosurus vulpecula. J. camp. Neurol. 179: 753-760, 1978. 4. Freeman, B. and C. R. R. Watson. The optic nerve of the Bushtailed possum, Trichosurus vulpecula: fiber diameter spectrum and conduction latency groups. J. camp. Neurol. 179: 739-752, 1978. 5. Fukuda, Y. A three-group classification of rat retinal ganglion cells: Histological and physiological studies. Brain Res. 119: 327-344, 1977. 6. Fukuda, Y. and J. Stone. Retinal distribution and central projections of Y-, X- and W-cells of the cat’s retina. J. Neurophysiol. 37: 749-772, 1974. 7. Giolli, R. H. and M. D. Guthrie. The primary optic projections in the rabbit. An experimental degeneration study. J. camp. Neurol. 136: 9Fl26, 1969.
8. Granit, R. A. and E. Marg. Conduction velocities in rabbit’s optic nerve and some observations on antidromic retinal spikes. A. J. Ophrhal. 46: 223-231, 1958. 9. Hale, P. T., A. J. Sefton and B. Dreher. A correlation of recep-
tive field properties with conduction velocity of cells in the rat’s retina-geniculo-cortical pathway. Expl Brain Res. 35: 425-442, 1979.
10. Hayashi, Y., 1. Sumitomo and K. Iwama. Activation of lateral geniculate neurons by electrical stimulation of superior colliculus in cats. Jap. J. Physiol. 17: 638-651, 1967. 11. Hughes, A. Topographical relationships between the anatomy and physiology of the rabbit visual system. Docums Ophthal. 30: 33-159, 1971. 12. Lederman, R. J. and W. K. Noel]. Fast-fiber optic nerve. Vision Res. 8: 1385-1398, 1968.
system of rabbit
13. Marocco, R. T. Sustained and transient cells in monkey lateral geniculate nucleus: conduction velocities and response properties. J. Neurophysiol. 39: 340-353, 1976. 14. Masland, R. H., K. L. Chow and D. L. Stewart. Receptive field characteristics of superior colliculus neurons in the rabbit. J. Neurophysiol.
34: 141156,
1971.
15. Molotchnikoff, S. and P. Lachapelle. Evidence of a collicular input to the dorsal lateral geniculate nucleus in rabbitselectrophysiology. Expl Brain Res., in press. 16. Molotchnikoff, S., P. Lachapelle, D. Richard, P. L’Archeveque and I. Lessard. Latency distribution from orthodromic stimulation at the optic nerve in the lateral geniculate nucleus and superior colliculus of rabbits. Brain Res. Bull. 4: 57F581, 1979. 17. Provis, J. M. The distribution and size of ganglion cells in the retina of the pigmented rabbit: a quantitative analysis. J. camp. Neurol. 185: 121-138, 1979. 18. Reuter, J. H. and K.-P. Hoffman. The conduction velocity in retinogeniculate afferent fibers in the rabbit. Neurosci. Letr. 16: 175-179, 1980. 19. Rhoades, R. W. and L. M. Chalupa. Conduction velocity distribution of the retinocollicular pathway in the golden hamster. Brain Res. 159: 396-401, 1978. 20. Rowe, M. H. and J. Stone. Conduction velocity groupings among axons of cat retinal ganglion cells and their relationship to retinal topography. Expl Brain Res. 25: 339-357, 1978. 21. Sawyer, CH., J. W. Everette and J. D. Green. The rabbit diencephalon in stereotaxic coordinates. J. camp. Neurol. 101: 801-824, 1954. 22. Schiller, P. H. and J. G. Malpeli. Properties and tectal projections of monkey retinal ganglion cells. J. Neurophysiol. 40: 428-445, 1977. 23. Sefton, A. J. The innervation
of the lateral geniculate nucleus and anterior colliculus in the rat. Vision Res. 8: 867-881, 1%8. 24. Semm, P. Antidromically activated direction selective ganglion cells of the rabbit. Neurosci. Lert. 9: 207-211, 1978. 25. Steward, D. L., K. L. Chow and R. H. Masland. Receptivefield characteristics of lateral geniculate neurons in the rabbit. J. Neurophysiol.
jq: 139-147,
1971.
26. Stone, J. and Y. Fukuda. Properties of cat retinal ganglion cells. A comDarison of W-cells with X- and Y-cells. J. Neurophvsiol. _
37: 722-748, 1974. 27. Stone, J. and R. B. Freeman. Conduction velocity groups in the cat’s optic nerve classified according to their retinal origin. Expl Brain Res. 13: 489-497, 1971. 28. Vaney, D. I. A quantitative comparison between the ganglion
cell populations and axonal outflows of the visual streak and periphery of the rabbit retina. J. camp. Neural. 189: 215-233, 1980.