Cell Dlfferentiatton. 14 (1984) 33 45 33 Elsevier Scientific Publishers Ireland, Ltd.

C D F 00210

Demonstration of a polarizing signal that reverses future retinotectal patterns across Nuclepore filter barriers, in Xenopus

embryonic eye

Kath l een Sull ivan 1, Kev in M. C o n w a y 2 and R. Kev in H u n t 2

Jenkins Biophysical Laboratories. The Johns Hopkins Uniuersi(v, Baltimore, MD 21218, U.S.A.

(Accepted 1 December 1983)

We have studied the interactions which occur in surgically disarranged eye rudiments by recombining a left anterior half-eye graft from a donor Xenopus embryo with a right host posterior half eye, across a varie~ of physical barriers, at embryonic stages 31 or 32. The anterior half-eye graft and barrier were removed 18 h later at stage 38, and the host posterior half-eye was allowed to reconstitute a whole eye whose visuotectal projection could be mapped electrophysiologically after metamorphosis. Anteroposteriorly reversed maps and double-anterior twinned maps that are characteristic of anterior half-eyes, were found in 50-65% of cases in each of the experimental series using no barrier (N= 16), or using Nuclepore filter barriers (N= 47), including 5 of 8 cases when a filter of 0.015 micrometer pore diameter was used. The latter cases are especially interesting, because the filter pores were much smaller than the minimum size known to permit cell-cell contact through the pores. No animals showed AP-reversed retinotectal maps or double-anterior twinned maps when the graft and host half-eyes were separated by a tantalum or plastic barrier ( N = 21). Only a single case of AP-reversed mapping was found in 115 control animals including simple posterior half-eye preparations at stage 32 or 38 ( N = 13), sham fusions (30 min) across Nuclepore filters ( N = 35), or chronic application of a filter (or plastic or tantalum) barrier from stages 3 2 - 3 8 ( N = 55) without a left anterior half-eye graft. We conclude that signals from an anterior half-eye can act to repolarize a posterior half-eye in the absence of cell transfer and under conditions which permit little or no direct cell-cell contact.

neuronal specificity; embryonic pattern regulation

Introduction

During development, the frog eye forms a topo- graphic pattern of synaptic connections with the midbrain optic tectum (Gaze et al., 1974). The assembly of retinotectal connections is a complex

1Present address." U.C.S.F. Medical Center, Department of Microbiology/Immunology, San Francisco, CA 0000, U.S.A. 2 Present address and address for all correspondence: The Salk Institute for Biological Studies, P.O. Box 85800, San Diego, CA 92138-9216, U.S.A.

and protracted process (Sperry, 1963: Fraser and Hunt, 1980; Schmidt, 1982), as the retina con- tinues to grow throughout larval and juvenile life, adding new cells in rings to its rim (Straznicky and Gaze, 1971; Jacobson, 1976; Tompkins et al., 1983). However, the pattern of its connections with the optic tectum is determined at an early stage, when the eye-rudiment contains but a few hundred cells (Jacobson, 1966: Jacobson and Hunt, 1973).

Embryonic determination of retinotectal pat-

0045-6039/84/$03.00 ~,5 1984 Elsevier Scientific Publishers Ireland, Ltd.

34

terns in the embryonic eye-rudiment is similar in many respects to pattern determination in a num- ber of other organ-rudiments (Harrison, 1921, 1945; Nicholas, 1955: Yntema, 1955). A tentative spatial plan or prepattern is present in Xenopus from the earliest optic vesicle stages (Hunt and Jacobson, 1973a) but only becomes fixed as the result of a succession of covert differentiation steps as the anlage matures (Hunt and Jacobson, 1974). Thus, it has been found that the eye-rudiment loses its ability to respond to pattern determining signals from the trunk at stages 28-32 (Hunt, 1975): after stage 32 it is no longer perturbed by the local implantation of 'organizer' tissue next to it (Cooke, 1977); and neither whole eye-rudiments nor half-rudiment fragments can produce a nor- mally oriented map following surgical rotation oy left-right misalignment (Hunt and Jacobson, 1972a, b; Hunt and Berman, 1975). Yet for a day or more after stage 32, embryonic half-eyes, prepared by partial ablation, seem to retain considerable inter- nal plasticity. For example, solitary anterior or posterior half-eye can still reconstitute a whole eye and form a normal retinotectal map or, in some cases, a twinned retinotectal map whose mirror- symmetric topography accurately duplicates the polarity of the half-eye in the sense that anterior half-eyes make double-anterior maps, posterior half-eyes make double-posterior maps (Feldman and Gaze, 1975; Berman and Hunt, 1975; Hunt, 1975). Moreover, mismatched pairs of embryonic half-eyes, when grafted together to form a recom- binant eye, can modify their developmental pre- pattern and produce a coherent normal or twinned retinotectal pattern (Hunt and Jacobson, 1973b; Hunt, 1975; Ide et al., 1979). Thus, after replacing the posterior half of a right embryonic eye with the posterior half of a left embryonic eye, the resultant retinotectal map from posterior retina is not re- versed in the dorsoventral dimension (as might be expected from the original polarity of the retinal tissue) but rather shows a 'double-anterior ' twinned pattern, in which the map from posterior retina is reversed in the anteroposterior dimension but nor,- mal in the dorsoventral (Hunt and Jacobson. 1973b; Hunt, 1975).

The experiments of Hunt and Frank (1975), which represent the starting point for the present

study, suggested that only a brief interaction be- tween the two half-eye fragments is necessary to redirect their future retinotectal patterns. They made left-posterior/right-anterior half-eye recom- binants at stage 32 and allowed the two pieces to remain fused for varying intervals. The posterior (graft) fragment was then separated and allowed to reconstitute a whole eye, and some time after metamorphosis the resulting retinotectal map was analyzed electrophysiologically. Following very ~hort-term fusions (30 min), the adult visual map ':eflected the initial polarities of the grafted poste- rior piece - that is, only normal, or a double-post- erior twinned, maps were formed. Following 12 16 h of fusion, however, the entire map was reversed in the anteroposterior dimensions in more than half the cases. After fusions of 30 32 h, recon- stituted whole eyeballs were found in a smaller percentage of cases, but all of these showed ante- roposterior-reversed maps, and in many of them the dorsoventral-polarity was also adjusted to match that seen in the caudal half of a 'double-an- terior' twinned pattern.

In the present study, we have attempted to elucidate some of the factors that may be involved in the determination of the retinotectal patterns in such recombinant eye fragments. Gaze et al. (1979), Cooke (1980), and Gaze and Straznicky (1980) have suggested that grafting experiments of this kind appear to produce 'regulated' retinotectal patterns because the graft dies and the host eye simply replaces it with new tissue, and that the saline solution used in earlier grafting experiments (Jacobson and Hunt, 1973) selectively damages the graft tissue, while sparing host eye tissue (Gaze et al., 1979; Gaze and Straznicky, 1980). Gaze and Straznicky (1980) have, in fact, confirmed that left-posterior/right-anterior recombinants gener- ate double-anterior retinotectal patterns. But they have also pointed out that solitary anterior frag- ments (upon reconstituting a whole eyeball) may, in some cases, form characteristic 'double-anterior ' twinned retinotectal patterns. For this reason they were inclined to the view that the recombinant may simply be equivalent to an anterior-half alone. Although Gaze and Straznicky (1980) did not con- sider the evidence for the transient fusion/separa- tion experiments of Hunt and Frank (1975), the

~eye autonomy' model which they have put for- ward is quite explicit: The anteroposterior-re- versed patterns must reflect: (1) generalized damage to the grafted posterior-half eyes, and (2) its covert takeover by cells from the host anterior half-eye. The final step of salvaging the 'grafted P-half' for whole eye reconstitution would, in truth, be salvaging host eye tissue from an earlier, covert reconstitution by the host (A-half) piece.

To clarify this issue we have repeated the tran- sient fusion/separat ion experiments of Hunt and Frank (1975) but have modified the original ex- perimental Iparadigm in two ways. First, by left- to-right replacement of the anterior half-eye, and later examining the retinotectal map formed by the posterior half-eye; in this situation the graft constitutes the putative 'signaling fragment', and the ' responding fragment' is the ungrafted host half eye. The 'eye autonomy' model would predict that under these conditions no anteroposterior re- versals would occur and that one would see the normal patterns, or more rarely, double posterior twinning (Feldman and Gaze, 1975; McDonald, 1975; Berman and Hunt, 1975; Hunt and Berman, 1975; Hunt, 1975; Ide.et al., 1979). The second modification is that we have fused the two half-eyes across differentially porous physical barriers. The barrier was removed after various intervals to- gether with the grafted anterior half-eyes, and the resulting retinotectal map formed by the host half eye fragment was mapped in the usual way after metamorphosis.

Materials and Methods

General

All experiments were carried out on wild-type Xenopus laeuis embryos, following our usual meth- ods (see Conway et al., 1980, for details). The animals were maintained throughout in rearing solution (15% Holtfreter's; 5% Steinberg's, pH 7.4), at 20 24°C. The surgical procedures used to make embryonic half-eyes, and to graft two such half- eyes together to make a surgically recombinant eye, are described elsewhere (Hunt, 1975; Hunt and Frank, 1975; Berman and Hunt, 1975). Seven

35

different experimental series and 15 control series were performed as detailed below. All the animals were reared to metamorphosis, and the visuotectal projection from the experimental right eye to the left optic tectum was analyzed by standard electro- physiologic methods (see Conway et al., 1980).

The heads from representative animals were fixed in Bouin's solution, after recording, and em- bedded in paraffin, serial 7-/xm sections were stained with Panceau S-Hematoxylin, and ex- amined microscopically to confirm that the eyes were grossly normal at the time of recording. Histologic observations on healed embryos were made in four experimental groups (Series E-2, E-3, E-4, E-5), in which half-eye-rudiments had been fused across Nuclepore filter barriers. Ten animals in each series were fixed in glutaraldehyde at stage 38 (with the healed recombinant eye and filter-bar- rier undisturbed since the operations at stage 31/32), embedded in Eponaraldite, and sectioned roughly perpendicular to the filter barrier. Sections 1 2/~m thick were stained with Toluidine Blue for light microscopy; and ultrathin sections were stained in uranyl acetate, lead citrate and lead tartarate and examined and photographed in a Philips electron microscope.

Experimental series

The anterior half of the right eye-rudiment was surgically excised at stage 31/32, leaving the post- erior half-rudiment in situ. In Series E-2 through E-7, one of several varieties of physical barrier was applied flat against the cut edge of the half eye as schematized in Fig. la. The anterior half of a left eye-rudiment was then excised from a stage 31/32 donor embryo, and grafted into the orbit of the host with its cut surface against the filter, and aligned with the host piece (Fig. la). In Series E-l, in which no barrier was used, the cut-faces of graft and host half-rudiment fragments were directly apposed to each other. The seven experimental series differed only in the barrier material that was used. These were: no barrier (Series E-l); tantalum foil (Series E-2); plastic membrane made up of Nuclepore material but lacking pores (Series E-3); and Nuclepore filters of graded pore diameter: 1.0-/xm pores, flow rate 300 m l / m i n per cm 2 (Series

36

TRANSFILTER PREPARATION

l a . PreDarat ion of host P - h a l f . . .

C)onor Hosf

l c . . . . a n d lef t A - h a l f eye g r a f t .

Hos t Host

l b . . . . and in t roduct ion ot b a r r i e r . . .

18 hours

Hos t

ld Remova b a r r i e r and A - h a l f .

2 - 5 d a y s

Hos t

l e . Reconst i tu t ion

of who le eye

from P - h a l f .

Host

Fig. 1. To show steps in the surgical protocols used on stage to surgically recombine a (host) right posterior half-eye with a (grafted) left anterior half-eye, with a physical barrier interposed between the two half-eyes (a-c), followed 18 h later by removal of the barrier and the grafted A-half (d), allowing the isolated host P-half to reconstitute a whole eyeball and innervate the midbrain optic rectum. In the seven experimental series described in the text, a variety of physical barriers were used including none (Series E-l), tantalum (E-2), plastic (E-3), and Nuclepore filters with pore diameters of 1.0 ~m (E-4), 0.6 /Lm (E-5), 0.08 t~m (E-6), and 0.015 t~m (E-7). Control series included unoperated sibling embryos (C-l), simple preparation of a posterior half-eye at embryonic stage 31/32 (C-2) or stage 38 (C-3): seven variants of sham-fusion (SF-1 through SF-7), identical to the experimental series save that the interval between step 'c" and step 'd" was only 30 rain; and seven variants of no-graft control (NG-I through NG-7), identical to the experimental series save that the grafted A-half was omitted in step 'c', and additional barrier material was inserted into the anterior orbit to hold the primary barrier flat against the host posterior half-eye for the 18-h interval, prior to removal of all barrier material at stage 38.

E-4); 0.6-#m pores, flow rate 200 ml /min per c m 2

(Series E-5); 0.08-#m pores, flow rate 2 ml /min per cm 2 (Series E-6); and 0.015-/~m pores, flow rate 0.02-0.08 ml /min per cm 2 (Series E-7).

The host embryos were gently removed from anesthesia and reared for 18 h to stage 38 where they were reanesthetized, and both the barrier and

the anterior half-eye graft were completely excised. Any embryo in which the surgery was not obvi- ously successful as evidenced by a "barbell' ap- pearance with two half eye-cups, apposed, prop- erly aligned, and centered on the barrier, was discarded immediately. The anterior half eye graft was excised completely (usually in one piece), the

37

an te r ior orb i ta l bed was careful ly c leaned (by stage 38, the eye is encapsu la ted in a lea thery black choro ida l capsule, which adheres min imal ly to the orb i ta l bed and te legraphs (rare) failures of com- plete excision by a res iduum of black cells on the bed of white orbi ta l mesenchyme), and the barr ier was then removed in one piece, leaving the host pos te r ior half-eye und is tu rbed to reconst i tu te a whole eye by itself.

Control series

Three control series were p repared : normal , unope ra t ed animals (Series C- l ) , s imple p repara - t ion of a pos te r io r -ha l f rud iment at stage 3 1 / 3 2 (Series C-2), or a s imilar p repa ra t ion at stage 38 (Series C-3). Seven series of shamfus ion cont ro ls were pe r fo rmed (Series SF-1 through SF-7) para l - leling the seven exper imenta l series; in these cases the graft and bar r ie r were removed after only 30 min. Final ly , seven series of no-graf t controls were p repa red (NG-1 through NG-7) , each identical to the cor respond ing ly numbered exper imenta l series, except that no anter ior half-eye graft was intro- duced; the bar r ie r was app l ied as usual, and an add i t iona l wedge of bar r ie r mater ia l was inser ted into the anter ior par t of the orb i t to keep the bar r ie r closely app l ied to the cut-face of the host half eye- rudiment .

Results

Healing and reconstitution

The eyes which held their r ecombinan t config- u ra t ion dur ing the immedia t e opera t ive pe r iod showed the expected ' b a r b e l l ' form of appos i t ion across the filter at s tage 38 (Fig. 2a). The filter was roughly twice the dorsovent ra l d imens ion of the eye-cup, and on each side of it the half-eye cups had ma tu red but remained encapsu la ted half eyes with their cut faces t ightly apposed to the filter. The barr ier was observed to extend deep into the embryo , well media l to the choro ida i coa t of each half eye. There was no evidence that eye cells had d ispersed or grown through or a round the barriers .

The second opera t ion left the pos ter ior host

Fig. 2. To show the appearance of the eye, with the filter barrier in place, at stage 38 (18 h after the initial surgery and just prior to the scheduled removal of filter and grafted A-half) in a Series E-7 frog. (b) The adult eye at the time of electro- physiologic recording in the same Series E-7 frog whose em- bryonic photograph was presented in (a). Note that the eye is normally oriented and is grossly normal in appearance, and several choroidal markers indicate that the eye is normally oriented. These include the choroidal coat which is black dorsally and silver ventrally- reflecting a higher density of iridophores ventrally- and the iris cleft, ventral fissure scar and associated blood vessels at the ventral pole of the eye.

38

-ofero I

rSO~rn

-" / \ "- v~ / ,\ 7 /

/ " \ . /

I I

D ~ ~ \ C ~ % ~\~.,

2524~3i222~', ~1.2°~9 B 7 ~ '

~, 20 19h81 7 6 '/ ~ lS|F4|13~12| l lK/Latera l -~~15II4~13~12|ll=~LaterGI 5--~[~10 9 ' 8 [ 7 6 )[ESO~m Q--|IO 9 1 8 / 7 6 ~r

Rosfrol v ~ Rostrol ~ ~ S I I 150~m

/ \ /?oj>,0 \ 1 / ~ ~ - 4 _ ~ \

/ '~ l 11~ ,) , \ ~ 1 1 ~ 6 ~ 1 //' '\\ 6 //' \x /

\ / \ / / \ , \ \ / / ,

, \ / \ \ . .

/ ~ . / /

Fig. 3. (A) To show a grossly normal visuotectal projection in a Series C-3 frog, assayed electrophysiologically after metamorphic transformation. Each number in the visual field shows the position at which a spot of light (10' to 2 °) optimally evoked action potentials, when a platinum-tipped platinum-iridium microelectrode was allowed to penetrate the superficial neurophil of the rectum at the correspondingly numbered position on the tectal surface. Responses consisted of action potentials, primarily from terminal arborizations of optic nerve fibers in the rectum, and were monitored through a notch-filter, amplifier, oscilloscope, and loudspeaker. Distance between electrode recording positions is given by the bars. The visual field of the right eye (circle) extends 100 ° from center to periphery. Conventions are the same throughout Figs. 3 and 4. Note that in the normal projection, more nasal visual field projects more rostrally on the rectum, while more temporal visual field projects more caudally: more inferior visual field projections laterally on the rectum, while more superior field projects toward the ventral midline. It is also worth recalling that the eye inverts the visual field image, so that posterior retina subtends nasal visual field, and ventral retina subtends superior visual field. Finally, the projection extends round the lateral edge of the tectum, not accessible to the recording electrode. Thus the extreme lateral representation of the visual map on the tectum and its visual field representation (far inferior visual field in the normal projection) are not represented in the visual projection diagrams.

(B) To show a normal visuotectal projection from a Series C-2 control frog. (C) To show the "double-posterior'-twinned visuotectal projection from another Series E-3 control frog, so named because the

topography normal to the posterior retina (and nasal visual field) is mirror-twinned in the complementary nasal retina, and because recombinant eyes prepared by fusing two P-halves (one left and one right) give this 'double posterior' pattern. The pattern is rare but absolutely characteristic in topography, following simple preparation of a P-half fragment and its subsequent reconstitution of a whole eyeball (19 22).

(D) To show a normal visuotectal projection in a Series E-2 frog. No reversal has occurred in this case, in which the two half eye-rudiments were fused across a non-porous barrier of tantalum. No AP-reversed patterns or double-anterior twins were obtained m 21 cases in Series E-2 and E-3, the latter series using non-porous barriers of plastic.

fragment in its half-cup shape. Over the next two days, the posterior eye fragments usually ' round- ed-up' to form a lopsided small eyeball (Berman and Hunt, 1975; Conway et al., 1980). By the end of the first week of larval life, most of the eyes were again at normal size and were more or less normal in shape. In a minority of both the experi- mental and control animals the eye had been resorbed, and in a small number of cases ocular malformations (e.g., no lens, two lenses) had oc- curred. A total of 183 (88.4%) animals, whose right eye showed grossly normal morphology (Fig. 2b) were available for recording. In all of these the eyes were correctly oriented (Fig. 2b) as judged by the position of the choroidal fissure and other landmarks (Hunt and Jacobson, 1973b; Berman and Hunt, 1975; Hunt, 1975).

Controls

The normal visuotectal projection, obtained from a Series C-3 control, is shown in Fig. 3A. It is smooth and continuous, with the superior visual field (i.e., ventral retina) mapping onto the medial part of the tectum and the nasal field (posterior retina) mapping rostrally. The visuotectal mapping data from all 15 control series, comprising some 115 animals, produced only a single case of ante- roposterior reversal (a double-anterior twinned pattern): the reason for this one exception is not known. Major abnormalities of any kind were rare - one Series NG-2 frog showed a few dupli- cated points in the visual field, two frogs from the SF Series showed double-posterior twin patterns, and one Series NG-5 frog showed a disorganized projection. It is important to emphasize, however, that rare abnormalities of this kind, including the double-posterior-twinned pattern (see Fig. 3C), have been found on occasion after simple recon- stitution of a whole eye by a single posterior half eye-rudiment (Feldman and Gaze, 1975; Berman and Hunt, 1975). Fully 111 of the 115 control frogs showed grossly normal visuotectal projec- tions.

Experimental series

Fig. 3C shows the double-posterior-twinned visuotectal pattern, obtained from a Series E-3

39

frog. In the two experimental series (Series E-2 and Series E-3) in which non-porous barriers were interposed between the host posterior-half and a grafted left-anterior-half (for 18 h, followed by removal of the barrier and the graft) there were no cases in which the visuotectal projections AP was reversed ( N = 2 1 : Fig. 5). In these cases normal maps (Fig. 3D) predominated, but one map was disorganized and three others showed the 'double-posterior ' twinned pattern (Fig. 3C).

Fig. 4 and 5 summarize the visuotectal projec- tion data from the seven experimental series. In Series E-l, in which no barrier was used, the two fragments being directly fused for 18 h before removing the 'graft ' , results were mixed (see also Hunt and Frank, 1975). Nine of the 16 frogs in this series showed frank evidence of anteroposte- rior reversal: the map from the posterior retina (nasal visual field) was rostrocaudally inverted, and had duplicated to form a double-anterior twinned pattern (Fig. 4A). Significantly, in seven of these cases the pattern was also reversed in the dorsoventral-axis, as in the case shown in Fig. 4A. It is worth stressing that double-anterior twinned p a t t e r n s - a characteristic minority result when a solitary anterior half-eye is challenged to recon- stitute a whole eyeball have never been observed following reconstitution by a posterior half-eye: nor have the latter, in the rare cases which do form twinned patterns, ever been observed to form a twinned pattern other than double-posterior (Feld- man and Gaze, 1975; McDonald, 1975; Berman and Hunt, 1975; Hunt and Berman, 1975: Hunt, 1975; Hunt and Frank, 1975; Ide et al., 1979).

The majority of the 47 frogs in Series E-4, E-5, E-6 and E-7, in which Nuclepore filter barriers were used, also showed anteroposterior-reversed patterns. A few showed a simple anteroposterior- reversed ordering from the entire visual field (Fig. 4B); the rest showed double-anterior-twinned pat- terns (Fig. 4C F) similar to those seen in the control animals without a barrier (Series E-l). The dorsoventral topography of these maps varied con- siderably; sometimes it was reversed (Fig. 4C, D), occasionally it was normal (Fig. 4B, E, F) and in some cases it could not be identified (Fig. 4G). While dorsoventral-reversals were common in the larger-pore Series E-4 and E-6, no cases of dorso-

40

z

I--

z

x / I-

z

~. . , . I o ." ~. jJol

. . . . . 0"> ,,~ , ~ I

,Vea,a, o leo: " , ' \ ,

z

z -

o o Lo /

/~ , _~ ~. ,, - \ l o / <,,= i ~ " - - ~ " - < " \ - i ~ ~

- \ P ,'-

z

i

I I .

/; ~ll P , i - + - [ ,

41

SERIES

BARRIER

NUMBER AP-REVENSED

NUMBER AP-NORMAL

C-1,2,3 E-I E-4 E-5 E-6 E-7 E-3 E-2 S ~ I - 7

NONE

NG I-7 PORE SIZE

1.0um 0.6um O.08um O.015um NUCLEOPORE FILTER BARRIERS PLASTIC TA CONTROLS

9 7 4 3 5 O 0 1

5 5 4 2 3 I0 11 115

Fig. 5. To show the incidence of AP-reversal in the seven experimental series, shown in the histogram as percentages and below the histogram in actual case numbers from each series. Note that 50 65% of cases in Series E-I, where no barrier was used, and in Series E-4 through E-7, where Nuclepore filter barriers were used, showed AP-reversed patterns (including AP-reversal with twinning to a double-anterior pattern). No reversals occurred in Series E-2 and E-3 when non-porous barriers of tantalum and plastic were used. Only one case among 115 controls showed an AP-reversed pattern. These controls spanned 17 series, including unoperated siblings (N = 12), simple preparation of P-half eye fragments at stage 32 or 38 ( N = 6, N = 7), sham-fusions across each class of barrier for 30 min (N = 35), and no-graft controls in which each class of barrier was applied for 18 h to the P-half eye-rudiment without adding a 'signalling' A-half eye graft (N = 55). The one excep- tional case occurred in a tantalum sham-fusion control, which made a double-anterior-twinned pattern and is currently inex- plicablc.

ventral-reversed patterns were observed in Series E-7, in which the filter-barrier contained the smal- lest pores examined. But most importantly, the majority of cases in all four Nuclepore series showed reversed patterns and their characteristic double-anterior-twinned maps (Fig. 5).

Discussion

Our observations show that a grafted anterior left half-eye rudiment can interact with a posterior right half-eye, and bring about an anteroposterior reversal in the later retinotectal pattern formed by the host fragment. And furthermore, that this in- teraction can occur across a Nuclepore filter with pores as small as 0.015 /~m. After 18 h of fusion, the interaction that leads to anteroposterior-rever- sal is completed by between two-thirds and three- quarters of the recombinant eyes; moreover, the interaction is 'remembered', as the posterior frag- ment later reconstitutes a complete eyeball, and as the eyeball grows to adulthood, increasing its gan- glion cell population from a few hundred to about 70,000 (Straznicky and Gaze, 1971; Jacobson, 1976). The resulting adult visual projections were thus either reversed across the entire adult retina, or showed the double-anterior twinned patterns that is characteristic of whole eyes reconstituted from an anterior embryonic half-eye, but is never observed in eyes that are reconstituted only from a posterior half eye (Feidman and Gaze, 1975; Ber- man and Hunt, 1975; Hunt and Berman, 1975; McDonald, 1975; Hunt, 1975; Hunt and Frank, 1975; Ide et al., 1979).

Fig. 4. (A) To show the "double-anterior'-twinned visuotectal projection from a Series E-1 frog. This pattern reflects AP-reversal in the P-half fragment, with propagation of the twinned pattern characteristic of anterior half-anlagen reconstituting a whole eye. The pattern is also dorsoventrally inverted, indicating that both axes of the P-half pattern were reversed, and that the fragment went on to form an upside-down double-anterior-twinned pattern. (B) To show an AP-reversed visuotectal projection in a Series E-4 frog. This projection is continuous, and following AP-reversal of the pattern in the P-half eye-rudiment, the pattern was not duplicated or ' twinned' as the fragment reconstituted a whole eyeball and innervated the midbrain. (C) To show AP-reversal, with twinning to the double-anterior pattern, in a Series E-4 frog; the visuotectal pattern is also DV-inverted. (D) To show AP-reversal with twinning to a double-anterior pattern in another Series E-4 frog; this pattern is also DV-inverted. (E) To show AP-reversal with twinning to a double-anterior pattern in another Series E-4 frog; in this case the pattern is not DV-inverted, indicating that polarity reversal had occurred in only the AP-axis, before twinning and reconstitution of a whole eye by the P-half eye-rudiment. (F) To show AP-reversal with twinning to a double-anterior pattern in a Series E-7 frog; the pattern is normal in the DV-axis. (G) To show AP-reversal with twinning to a double-anterior pattern in another Series E-7 frog; this pattern is disordered in the DV-axis.

42

Although the fusion interval of 18 h was insuffi- cient to reverse the retinotectal pattern in 100~: of the cases (Hunt and Frank, 1975), the control series were exceedingly uniform: 15 separate con- trol protocols yielded 111 normal maps, and only three double-posterior-twinned maps in 115 cases. Posterior half eyes prepared at embryonic stage 32 or 38, sham-fusion preparations in which a left- anterior half-eye (was apposed to the host poste- rior half eye, with or without a barrier for 30 rain), posterior half-eyes in contact for 18 h with the barrier alone (in the absence of an anterior half-eye graft) all failed to show reversal of the final retinotectal pattern. The experimental series also provided a number of internal controls. No re- versed pattern was found when the half-eye rudi- ments were fused for 18 h across plastic or tanta- lum barriers. Thus, the reversed patterns seen in the other experimental series cannot be due to some mechanical effect of chronic recombination with a barrier on the development of the eye, for example by influencing the way fibers grow out of the host posterior fragment. Finally, many cases from the trans-filter Series E-4 through E-6 showed maps that were also reversed in the dorsoventral axis, but Series E-7 (with the smallest pores) showed reversal only along the anteroposterior- axis. One-axis reversed pa t t e rns - which were shown to decline in incidence with increasing fu- sion intervals in an earlier time series (Hunt and Frank, 1975) are significant in two respects. First, they suggest that dorsoventral-reversal re- quires either larger pores, or if small pores with very slow flow rates are used, much longer period of interaction. Secondly, because ' turning' cannot transform an eye's pattern from righthanded to lefthanded (as occurred in the observed one axis reversals), they add further evidence that the pat- tern reversals we have observed are not achieved by 'postural turning' of the host fragment or the eye reconstituted from it.

Implications for 'polarizing signals'

In addition to strengthening the evidence for polarizing signals that act between one eye frag- ment and another, we had hoped to devise a paradigm that would enable us to more critically

distinguish 'signal' from 'response' in this type of study. The recombination of a right-posterior and left-anterior embryonic half-eye seemed an espe- cially promising preparation, for the following rea- sons: (1) The final pattern is clearly reversed, relative to the starting polarities. (2) By stage 32 the eye has outgrown responsiveness to body cues and the position and orientation of the 're- sponding' posterior-half remain normal with re- spect to the rest of the body. (3) The tissues that serve as ' transmitter ' and 'receiver/responder' are unambiguously identifiable, and since both are embryonic half-eyes they are similar in size, shape, and cellular composition, and each has a cut-face apposed to the barrier during fusion. (4) Specificity of both elements can be studied by substituting other fragments (right-posterior/ right-anterior or right-posterior/left-posterior re- combinations, for example, do not reverse the pattern in the host (right-posterior) half-eye). And (5) the interaction is fast enough so that the com- ponent fragments, when isolated, are still able to reconstitute a whole eyeball, yet slow enough to allow intermediate forms (such as one-axis rever- sals) to be detected. We as yet have no clues to the chemical basis of polarizing signals. We are still analyzing ultrastructural data, but preliminary analysis indicates that the filters of the smallest- pore size used (0.015/,m) much smaller than the threshold pore size for cell contact in many trans- filter-induction systems (Saxbn et al., 1977: rev. Toivonen, 1979; Sax6n, 1980) - show no cell con- tact, and no penetrating cell processes, in elec- tronmicrographs of the pores in situ, in the stage 38 recombinant.

Implications for 'recombination' t~s. 'reconstitution"

Harrison (1918) visually distinguished the re- constitution of a whole organ from a single half- anlage (the organ develops with much delay, pass- ing through intermediate forms of a truncated half-bud, through a small and lopsided early larval organ, to a small but internally perfect organ that slowly regains normal size) from the recombina- tion of two half-anlagen to form a single in- tegrated organ (the organ matures on normal schedule, with no size lag, after passing through a °barbell' form that catches the two half-anlagen

merging to form a coherent single organ). A nearly identical distinction was drawn between the evolv- ing form of solitary eye fragments (Berman and Hunt, 1975) and the 'barbell ' integration of two half-eyes grafted together to make a recombinant eye (Hunt, 1975; Hunt and Frank, 1975). Only recently have cell marking studies confirmed Har- rison's (1918) suspicions and shown that territorial cell fates are smoothly shifted and expanded throughout a half-eye fragment, when a single fragment labors to reconstitute a whole full-size eye (Conway et al., 1980). The present study has exploited the ability of a solitary embryonic eye fragment to reconstitute a whole eyeball, but neither depends upon nor adds much to our un- derstanding of the reconstitution process per se. Both the simple AP-reversed (and thus left-handed) maps (Fig. 4B) and the more common double- anterior-twinned maps are characteristic and relia- ble markers of AP-polarity reversal in the poste- rior half-eyes of our experimental series. Yet it remains to be explained (1) how individual cells within our repolarized posterior half-eyes may ad- just their growth and territorial fate in reforming a whole eyeball, and (2) why they formed double- anterior-twinned maps more often than simple AP-reversed maps (cf. Berman and Hunt, 1975).

Nevertheless, we hoped the transfilter experi- ments would either validate, or reveal the limita- tions of, more classical grafting methods used in earlier studies. We found no evidence for graft death or damage under our routine operating con- ditions (see also Hunt et al., 1983a). Significantly, half-eye recombinations without a filter - in which later retrieval of the 'graft' or 'host ' piece is based on identifying the two by their (distinct and robust) silhouettes in the 'barbell '-shaped recombinant - gave the same results as fusion across a filter, in which identification and retrieval of the original 'graft ' and 'host ' pieces is unambiguous. The former method, which reflects the essence of classical 'scoring' of successfully healing grafts (Harrison, 1918, 1921), received considerable validation by the new method.

Irnphcations for "eve autonomy"

The reversed retinotectal patterns obtained in this study do not reflect the 'autonomous expres-

43

sion' of how the component pieces would have mapped to the tectum in situ. Recent studies (Tompkins et al., 1982; Hunt et al., 1982a, b) used a tetraploid cell marker to show that repolarized retinotectal patterns from grafted eyes are in fact mediated by graft-derived retinal tissue. Thus, fol- lowing left-right replacement of a 60 ° wedge of eye tissue at stage 32, the 4n graft contributed a sector of final retina and subtended a sector share of a smooth seamless visuotectal projection; stage 32 dorsal-and-anterior half-eye recombinants, which formed regulated normal visuotectal projec- tions, showed complementary hemiretinae of 4n and 2n cells in the adult eye, and early left-right transplantation of (stage 26) whole eye rudiments gave rise to regulated (righthanded, normal) visual projections, and the neural retina and pigment retina were entirely of donor genotope.

The present study represents a second, indepen- dent strategy for analysis of internal pattern regu- lation in embryonic half-eye recombinants - tran- sient recombination of half-eyes across physical barriers (which keep the graft and host half-anla- gen physically separated during their period of interaction), followed by complete removal of the 'signalling' fragment (and barrier). Pattern reversal is then assayed in the eye reconstituted by the 'responding' half-eye alone. The present experi- mental design cast the 'signalling' anterior half-eye as the graft, and challenged the host posterior half-eye to show reversal of pattern. Unlike earlier assays, which retrieved (and assayed reversal in) the graft, in this paradigm the results cannot be explained by graft death or by its covert replace- ment by reconstitution from the host fragment. 'Autonomous' retinotectal patterns, in which the prepattern is autonomously expressed in the final eye and map, are important because they show that a prepattern exists as inferred when 'auton- omous' maps were first reported in early whole eye grafts (Hunt and Jacobson, 1973a) and in stage 32 recombinant eyes (Hunt and lde, 1977). We also note that no case of a full pattern reversal, achieved by death of the graft and complete takeover of the whole retina by host tissue, was ever obtained in any of the studies which engendered the 'eye au- tonomy' model (Gaze et al., 1979; Cooke, 1980; Gaze and Straznicky, 1980). We believe, as Harri-

44

son (1933) did, tha t ' a u t o n o m y ' m a y be be t t e r

v iewed , no t as a mode l , bu t s imp ly as a c a t e g o r y of

resu l t o n e sees, in e m b r y o n i c t i ssues w h i c h have

a c q u i r e d p a t t e r n i n s t r u c t i o n s bu t have no t yet set

t h e m in to i r revers ib le m o t i o n .

Acknowledgements

W e t h a n k Drs . M. D u d a a n d M. Beer for

a s s i s t a n c e wi th the e l e c t r o n m i c r o s c o p y , Drs. M a x

C o w a n , Bren t S tanf ie ld , a n d F r a n c i s Cr ick a n d the

r e v i e w e r s fo r v a l u a b l e s u g g e s t i o n s on the

m a n u s c r i p t , a n d Je r ry C o h e n , Pat . T h o m a s , a n d

Kr i s T ru lock for t imely a s s i s t ance in the p r e p a r a -

t ion of the m a n u s c r i p t . Specia l t h a n k s are ex-

t e n d e d to D i a n a R o s s a n d to J e n n i f e r F o r r e n c e for

the i r a s s i s t ance in m a i n t a i n i n g a n d rea r ing the

e x p e r i m e n t a l an ima l s . Th i s w o r k was s u p p o r t e d by

g r a n t s f r o m the N I N C D S a n d the N S F to Dr.

H u n t , an N 1 H t r a i n e e s h i p to Dr . C o n w a y , a n d a

f e l l owsh ip to Dr. H u n t f rom the A l f r e d P. S loan

F o u n d a t i o n .

References

Berman, N. and R.K. Hunt: Visual projections to the optic tecta in Xenopus after partial extirpation of the embryonic eye. J. Comp. Neurol. 162, 23-42 (1975).

Conway, K., K. Feiock and R.K. Hunt: Polyclones and pat- terns in growing Xenopus eye. Curr. Tops. Dev. Biol. 15, 216 317 (1980).

Cooke, J.: Specification in the developing eye for orientation of mapping to the brain- cellular autonomy or intercellular communication. Trends Neurosci. 3, 45 48 (1980).

Cooke, J.: Organizing factors for anatomical patterns and for selective nerve connections in developing amphibian brain. In: Cell Interactions in Differentiation, eds. L. Saxbn et al. (Academic Press, London) pp. 111-120 (1977).

Feldman, J.D. and R.M. Gaze: The retinotectal projection from embryonic half eyes in Xenopus. J. Comp. Neurol. 162, 13 22 (1975).

Fraser, S.E. and R.K. Hunt: Retinotectal specificity: models and experiments in search of a mapping function. Ann. Rev. Neurosci. 3, 319 352 (1980).

Gaze, R.M., J.D. Feldman, J. Cooke and S.-H. Chung: The orientation of the visuotectal map in Xenopus: Developmen- tal aspects. J. Embryol. Exp. Morphol. 53, 39-66.

Gaze, R.M., M.J. Keating and S.-H. Chung: The evolution of the retinotectal map during development. Proc. Roy. Soc, London Ser. B 185, 301-330 (1974).

Gaze, R.M. and K. Straznicky: Stable programming for map orientation in disarranged embryonic eyes in Xenopus. J. Embryol. Exp. Morphol. 55, 143-165 (1980).

Harrison, R.G.: On relations of symmetry in transplanted limbs. J. Exp. Zool. 32, 1-136 (1921).

Harrison, R.G.: Some difficulties of the determination prob- lem, Am. Nat. 67, 306 321 (1933).

Harrison, R.G.: Relations of symmetry in the developing em- bryo. Trans. Conn. Acad. Arts. Sci. 36. 277-330 (1945).

Hunt, R.K.: Developmental programming for retinotectal pat- terns. Ciba Found. Symp. 29, 131 159 (1975).

Hunt, R.K. and N. Berman: Patterning of neuronal locus specificities in retinal ganglion cells after partial extirpation of the embryonic eye. J. Comp. Neurol. 162, 43-62 (1975).

Hunt, R.K. and E. Frank: Neuronal locus specificity: transre- polarization of Xenopus after the time of specification. Science 189, 563-565 (1975).

Hunt, R.K. and C.F. lde: Radial propagation of positional signals for retinotectal patterns in Xenopus. Biol. Bull. (Woods Hole, Mass.) 153, 430 (1977).

Hunt, R.K. and M. Jacobson: Development and stability of positional information in Xenopus retinal ganglion cells. Proc. Natl. Acad. Sci. USA 69, 780-783 (1972a).

Hunt, R.K. and M. Jacobson: Specification of positional infor- mation in retinal ganglion cells of Xenopus: stability of the specified state. Proc. Natl. Acad. Sci. USA 69, 2860.-2864 (1972b),

Hunt, R.K. and M. Jacobson: Specification of positional infor- mation in retinal ganglion cells of Xenopus: assays for analysis of the unspecified state. Proc. Natl. Acad. Sci. LISA 70, 507-511 (1973a).

Hunt, R.K. and M. Jacobson: Neuronal locus specificity: Altered pattern of spatial deployment in fused fragments of embryonic Xenopus eye. Science 180, 509-511 (1973b).

Hunt, R.K. and M. Jacobson: Specification of positional infor- mation in retinal ganglion cells of Xenopus: intraocular control of the time of specification. Proc. Natl. Acad. Sci. USA 71, 3616-3620 (1974).

Hunt, R.K.. R. Tompkins, D. Reinschmidt, E. Bodenstein and R.K. Murphey: Starting points for a developmental genetics of nerve patterns. Am. Zool. 22, 185-207 (1982a).

Hunt, R.K., R. Tompkins, B. Szaro, and K. Conway: Repolarized retinotectal patterns in genetically marked eyes of Xenopus. Biophys. J. 37, 59a (1982b).

lde, C.F., B.E. Kosofsky and R.K. Hunt: Control of pattern duplication in the retinotectal system of Xenopus: suppres- sion of duplication by eye-fragment interactions. Dev. Biol. 69, 337-360 (1979).

Jacobson, M.: Starting points for research. Growth Symp. 25, 341 (1966).

Jacobson, M.: Histogenesis of the retina in the clawed frog with implications for the pattern of development of retinotectal connections. Brain Res. 103, 541 547 (1976).

Jacobson, M. and R.K. Hunt: The origins of nerve cell specific- ity. Sci. Am. 228(2), 26 (1973).

McDonald, N.: Appendix. J. Comp. Neurol. 162, 21-22 (1975). Nicholas, J.S.: Limb and girdle. In: Analysis of Development,

eds. B.H. Willier, P. Weiss and V. Hamburger (Saunders. Philadelphia), pp. 429-439 (1955).

Sax6n, L.: Neural induction: past, present and future. Curr. Tops. Dev. Biol. 15, 409-418 (1980).

Sax+n, L., M. Karkinen-Jaaskelaiven, and L. Weiss: Cell Inter- actions in Differentiation (Academic Press, London) 000 pp. (1977).

Schmidt, J.H.: The formation of retinotecta[ projections. Trends Neurosci. 5. 111 115 (1982).

Sperry, R.W.: Chemoaffinity in the orderly growth of nerve fiber patterns and connections. Proc. Natl. Acad. Sci. USA 50, 703-710 (1963).

45

Straznicky, K. and R.M. Gaze: The growth of the retina in Xenopus: An autoradiographic study. J. Embryol. Exp. Morphol. 26, 67-79 (1971).

Toivonen, S,: Transmission in primary induction. Differentia- tion 15, 177 182 (1979).

Tompkins, R., D. Krotoski-Gwozdqiowski, and R.K. Hunt: Mutants and cell markers in Xenopus lae~,is: applications to studies of eye development, Neurosci. Abstr. 8, 514 (1982).

Yntema, C.L.: Ear and nose. In: Analysis of Development, eds. B.H. Willier, P. Weiss and V. Hamburger (Saunders, Philadelphia), pp. 415 428 (1955).