pr0~. Natl. Acad. Sci. USA \.ol. 78, No. 4, pp. 2145-2149. April 1981 Biochemistry A topographic gradient of molecules. in retina can be used to identify neuron position (cell membrane/antigen/embryo/synapses/hybrid cells) G. DAVIDTRISLER, MICHAEL D. SCHNEIDER, AND MARSHALLNIRENBERC &oratory of Biochemical Genetics. National Heart, Lung. and Blood Institute, National Institutes of Health, Bethesda, Ma&nd ~0205 contributed by Marshall Warren Nirenberg, December 24, 1980 ABSTRACT A monoclonal antibody was obtained that hinds to cell membrane molecules distributed in a topographic gradient in avian retina. Thirty-five-fold more antigen was detected in dor- soposterior retina than in ventroanterior retina. Most of the an- tigen was associated with the synaptic layers of the retina. Less antigen was detected in cerebrum, thalamus, cerebellum, and optic tectum, but little or none was found in non-neural tissues tested. The antigen was found on most or all cell types in retina, and the concentration of antigen found is a function of the square of the circumferential distance from the ventroanterior pole of the gradient toward the dorsoposterior pole. Thus, the antigen can be used as a marker of cell position along the ventroanter- ior-dorsoposterior axis of the retina. Topographic relationships between retina ganglion neurons are conserved, forming point-to-point representations ofthe retina, when ganglion neurons synapse in tectum or certain other re- gions of brain. Thus, the retina is a favorable model system for studying the formation of synaptic circuits. However, the mo- lecular basis for spatial order has not been defined. Sperry (1) hypothesized that two orthogonal gradients of molecules on ret- ina ganglion neurons and corresponding gradients of compli- mentary molecules in the optic tectum might determine the specificity of synaptic connections between retina and tectum neurons. Other mechanisms proposed include adhesive inter- actions between migrating neurites, myelination of bundles of axons, and formation of extracellular channels by glia to guide axons in appropriate directions (2). Antibodies provide a means of identifying surface molecules and reactions required for neural function (3-Y). Rabbit anti- serum to clonal retina hybrid cells was used to detect an antigen with a restricted domain in retina (8). Our objective was to detect cell surface molecules with topo- graphic specificity in the retina, as candidates for neuronal rec- ognition molecules. Monoclonal antibodies were obtained by fusing P3X63 Ag8 mouse myeloma cells (9) with spleen cells from mice immunized with small portions of dorsoposterior or ventral chicken embryo retina. A hybridoma antibody was ob- tained that binds to cell membrane molecules that are distrib- uted in a dorsoposterior + ventroanterior gradient in retina. RESULTS The rationale for the experiments is shown in Fig. 1. Spleen cells from mice immunized with small portions ofdorsoposterior or ventral neural retina from the left eyes of Id-day chicken embryos were fused with P3X63 Ag8 mouse myeloma cells, and the binding of hybridoma antibody to cultured cells from the sector of the retina used for immunization was compared with The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "adcertise- nzent" in accordance \vith 18 LT. S. C. $1731 solely to indicate this fact. MOUSE IMMUNIZATION WITH CHICK RETINA SEGMENTS [ A(;;) A$P HYBRIDOMA ANTIBODY SPECIFICITY A<--& 6, v NUMBER (;, A&PNuMBER v OF OF HYBRIDOMA HYORlDOMl COLONIES COLONIES ++++ f 1 + f 5 + + 13 + + 3 - - 54 - 101 FIG. 1. (Upper) Hybridoma cell lines were derived from mice im- munized with cells mechanically dissociated from dorsoposterior (Left) or ventral (Right) retina from the left eyes of 14-day White Leghorn chicken embryos (Gallusgallus) (sections labeled 4 and 1, respectively, in Fig. 2). Symbols, A, D, P, and V correspond to anterior, dorsal, pos- terior, and ventral, respectively. The choroid fissure, through which axons exit or enter the retina, shown extending from the ventroanteri- or margin of the retina, was used as a landmark for dissection. Female BALB/c mice were injected intraperitoneally at 0,7, and 14 days with 8 x lo6 retina cells in 0.4 ml of Dulbecco's phosphate-buffered saline and intravenously with 2 x lo6 cells in 0.1 ml of the saline. On day 17, 10' dissociated spleen cells from each mouse were fused (10) with 2 x lo7 P3X63 Ag8 mouse myeloma cells. After fusion, the cells were suspended with 5 x lo6 spleen cells from a mouse that had not been immunized in 50 ml of medium A. [Medium A is selective medium DZOSHAT of Berzofsky et al. (11) except that it contains 10% fetal bo- vine serum, 1 @M aminopterin, 2 milliunits of insulin per ml, and each nonessential amino acid at 0.1 mM.1 Cells were added to 96-well plates (Falcon) (186,000 cells per 74 ~1 of medium A per well), incubated at 37oC in a humidified atmosphere of 10% CO@@% air, and fed addi- tional medium A (0.1 ml per well) on the 5th and 10th days of incu- bation. (Lower) To identify hybridomas synthesizing antibodies to re- gional antigens in retina, ceils from the portion of la-day chicken embryo retina (left eye) used for mouse immunization, or the rest of the retina, were dissociated with trypsin and collagenase, and transfer plates (12) were inoculated with 1.2 x lo6 of the cells per well. Cells were cultured for 2 days in 90% Eagle's minimal essential medium/ 10% fetal bovine serum; the medium was replaced with 50 ~1 of me- dium conditioned by hybridoma cells, and plates were incubated for 30 min at 37oC. Each well was washed three times with 150 ~1 of solution B (1 mg of pigskin gelatin per ml of phosphate-buffered saline) at 4oC. Each cell monolayer was incubated for 30 min at 37oC with 50 ~1 of solution B containing 1.6 nM rabbit 1251-labeled F(ab'jP antibody frag- ment [`251-F(ab')z] directed against mouse IgG heavy and light chains (9799 cpm) and 500 H of bovine serum albumin, Cells were washed as above, and bound radioactivity was determined. F(ab'jz (Cappel) was purified on a mouse IgG-Sepharose 4B column and iodinated with mono-`25I Bolton-Hunter reagent (Amersham) (13). binding to cells from the remainder of the retina. Cultured ret- ina cells were used to increase the probability of detecting an- tibodies directed against cell surface molecules. Sixty-eight of 2145 2146 Biochemistry: Trisler et al. 672 wells inoculated contained hybridoma colonies derived from a mouse immunized with dorsoposterior retina cells. One cell line (14H3) synthesized antibody that bound more to dor- soposterior retina cells than to cells from the rest of the retina, whereas I3 hybridoma antibodies bound equally to cells from both portions of retina. Antibodies to retina were not detected with 54 additional cell lines. Hybridoma colonies derived from a mouse immunized with ventral retina cells were detected in 109 of the 672 wells plated. Five antibodies of the 87 tested bound somewhat more to cells from ventral retina than to cells from the rest of the retina, and 3 antibodies bound equally. Thus, 14% of the 155 hybridoma cell lines examined synthesize antibodies to retina; however, only one antigen was detected, termed TOP for toponymic (i.e., a marker of position), with an asymmetric distribution, more abundant in dorsoposterior ret- ina than in the remainder of the retina. Distribution of TOP Molecules in Retina. Retinas from right or left eyes of chicken embryos were cut into eight sections as shown in Fig. 2, and cells from each section were assayed for [`%I-F(ab'),.anti-TOP antibody*TOP antigen] complexes. Bi- laterally symmetrical gradients of TOP were found in retina from right and left eyes. The highest concentrations of antigen detected were in dorsoposterior or dorsal retina and the lowest, in ventroanterior or ventral retina. Little lasI-F(ab'), bound to I / I I I I 1 / TOPOGRAPHIC GRADIENT OF ANTIGEN hh LEFT RIGHT RETINA LEFT RETINA TOP P3X63 Age BUFFER 12345678 RETINA SECTION FIG. 2. Gradients of TOP molecules in retina from right (01 and left (A) eyes of 1Cday chicken embryos. M and A-A, anti-TOP antibody; e----e and A----A, P3X63 Ag8 antibody; +O and A-A; buffer B without antibody to TOP. Values shown within aoorooriate sections ofretina are omol of "?-F(ab'L soecificallv bound p;tr m'g protein Le., pm01 of `251-F(ab')2 bound in debresence'bf &ti- TOP antibody minus pmol of iz61-F(ab'1, bound in the presence of P3X63 Ag8 antibody). The assay conditions for TOP antigen used in this and subsequent experiments, except where stated, were as follows. Each retina was cut into eight sections as shown. Retina cells were mechanically dissociated in phosphate-buffered saline at 4oC by trit- uration (10 times) with a 200 ~1 micropipette tip (Medical Laboratory Automation), and 100 ~1 of a cell suspension (usually 160 pg ofprotein; range. 5&250 me) was added to each well of wlvvinvl chloride 96-well V&&m plat&-(Dynatech) treated with solution B. Cells were cen- trifuged at 1300 x g for 5 min at 4oC and supernatant solutions were decanted. Each pellet was washed three times in solution B at 4oC (150 ~1 each wash), suspended in 50 ~1 of an antibody solution containing 1.0 ~1 of ascites fluid in solution B (except where specified), and in- cubated for 30 min at 4oC. Retina cells were washed three times as above, suspended in 50 ~1 of solution B containing 440 nM rabbit `*`I- F(ab'), (5-10 x lo4 cpm) and 500 pg of bovine serum albumin, and incubated at 4oC for 30 min, unless otherwise specified. Pellets were washed three times as described above, wells were separated, and ra- dioactivity was determined. Each value shown is the mean of two or three determinatione. Protein was determined by a modification of the method of Lowry et al. (14). Proc. .Vatl.Acad. Sci. USA 78 (19Sl) 0 ;00.0010.01 0.1 ,q RETINA PROTEIN 9 pl ASCITES FLUID ; PM 12%Flab')2 P E FIG. 3. (A) Effect of retina protein concentration on rz51-F(ab'), binding in the Qresence of anti-TOP antibody (A) or P3X63 Ag8 anti- body (01. o , I2 I-F(ab'), bound specifically. (B) Effect on hybridoma antibody (ascites fluid? concentration on- 1251-F(ab'12 binding. Solid svmbols. soecific bindine of `251-F(ab'L due to anti-TOP antibodv: ooen symbols; nonspecific biiding with P3X63 Ag8 antibody. o , o,-Dorsal retina, sections 4 and 5; A, A, ventral retina, sections 1 and 8. Reaction mixtures contained 220 nM lz5 IFtab'), (3.68 x 1O-3 &i/pmoll. (C) Effect of concentration of rabbit `251-F(ab')2 anti-mouse JgG. Solid symbols, specific `251-F(ab')2 binding; open symbols, nonspecific bind- ing to cells from dorsal retina (0, c) or ventral retina (B, O , A, A). retina cells when antibody to TOP was omitted or was replaced with P3X63 Ag8 antibody. Other hybridoma antibodies includ- ing A2B5 (6) bound equally to cells from different regions of retina (not shown). Assay Conditions. The effects of varying retina protein, hy- bridoma antibody, or `%I-F(ab'), anti-mouse IgG are shown in Fig. 3. Specific binding of `=I-F(ab'), was proportional to retina protein in the range 65-300 pg of protein (Fig. 3A). Half-max- imal and maximal specific binding of `=I-F(ab'), to dorsopos- terior retinacells were obtained with 1:1350 and 1:lOO dilutions of ascites antibody to TOP, respectivelv; higher concentrations of antibody to TOP reduced specific `*51-F(ab'), binding (Fig. 3B). Specific isI-F(ab'), binding to ventral retina was low; how- ever, the concentrations of anti-TOP antibody required for half- maximal and maximal specific binding did not differ greatly from those found with dorsal retina. Thus, no obvious difference was detected in the affinity of the antibody for antigen in dorsal and ventral retina. Nonspecific binding of `=I-F(ab'), to retina cells was low at all concentrations of P3X63 Ag8 antibody tested. The antibody to TOP was identified as an IgGl with K light chains (not shown). Half-maximal and maximal specific binding of `a51- F(ab'), were obtained with approximately 0.5 and 2 PM `%I- F(ab'&, respectively, both to dorsal and ventral retina cells (Fig. 3C). Nonspecific `%I-F(ab'), binding in the resence of P3X63 Ag8 ascites antibody was proportional to 18 I-F(ab'), concen- tration and was not saturating at 3.52 PM, the highest `%I- F(ab'), concentration tested. Geometry of the Gradient. The retina grows by accretion of concentric rings ofneurons at the periphery; thus, central retina is the oldest portion of the retina and peripheral retina is the youngest. To determine whether the antigen gradient is a polar gradient that rotates around the center of the retina with uni- form antigen concentration along any arc from center to pe- riphery or is a circumferential gradient extending from dorso- posterior to ventroanterior retina, left retinas of I4-day chicken embryos were cut into eight central and eight peripheral sec- tions (Fig. 4A) which were assayed for TOP. A 35-fold gradient of antigen was found extending from dorsoposterior to ventroan- terior margins of the retina aligned parallel to the long axis of the choroid fissure. As shown in Fig. 48, strips of retina extending from the dor- soposterior to ventroanterior margins, or perpendicular to this axis from anterior to posterior margins, were removed and each Biochemistry: Trisler et al. FIG. 4. Geometry of the TOP gradient in 14-day chicken embryo retina. Specifically bound `251-F(ab')z (pmol/mg of protein) is shown on the ordinate in A and B and within the appropriate segment of ret- ina tested in A. (A) Each retina (1eR eye) was cut into eight 45" sections ~7.25 mm in length from the center to periphery ofretina) and each was divided into central (4.9 mm) and outer (2.35 mm) segments. (B) Dem- onstration that TOP concentration detected depends on the square of distance from the ventroanterior margin of the retina. Percentage of maximal circumferential distance is shown on the abscissa; 100% cor- responds to 14.5 mm. A, Strips of retina from ventroanterior (0%) to dorsoposterior (100%) retina margins, 14.5 x 2.5 mm, were removed from eight retinas (left eyes), and each was cut into nine segments (1.6 x 2.5 mm) as shown; each segment was assayed for TOP antigen. O , Strips of retina from dorsoanterior (0%) to posterior (100%) margins of the retina perpendicular to the choroid fissure were prepared and assayed as above. 0, Data from A. The length ofthe arc from the ventral pole of the gradient to the center of each segment was calculated by assuming the retina to be a hemisphere and using equations for spher- ical triangles. was cut into nine pieces which were assayed for antigen. The concentration of TOP molecules detected varied continuously and logarithmically with the logarithm of distance along the circumference of the retina from the ventroanterior pole of the gradient to the dorsoposterior pole, with a slope of 2. In con- trast, little or no change was detected in retina cells along a perpendicular axis from anterior to posterior margins of the ret- ina. The data are described somewhat better by a power func- tion than by a logarithmic function, but a logarithmic function has not been ruled out. Proc. Natl. Acad. Sci. USA 78 (1981) 2147 The concentration of [`251-F(ab'),.anti-TOP antibodyTOP antigen] complex detected is described by the relationship: [TOP complex] = distance". Thus, cell position along a ventroanterior-dorsoposterior axis ofretinacan be identified by the concentration ofTOP detected: D, = Qnax (~JFmax)"~5 in which [TOP complex] is the fraction of maximal pmol of `=I- F(ab'), specifically bound per mg of protein, FJF,,, and dis- tance is the fraction of maximal circumferential distance from the ventroanterior to the dorsoposterior poles of the gradient, DJDmw Under the conditions used, F,, is 20 pmol of `=I- F(ab'), bound specifically per mg of protein and D,, is 14.5 mm. Thus, the calculated mean position in retina of cells that bind 5 pmol `=I-F(ab'), specifically per mg protein is 7.25 mm from the ventroanterior pole of the gradient, which agrees well with the experimental values. Autoradiography and Immunofluorescence. Autoradiogra- phy revealed most antigen in dorsoposterior chicken embryo retina in the inner and outer synaptic layers of the retina (Fig. 5 A and B). The antigen was detected in lesser amounts on the soma of most, or all, cell types in dorsoposterior retina. Little antigen was detected in ventroanterior retina (Fig. 5 C and D). Similar results were obtained by immunofluorescence (not shown). All cells mechanically dissociated from S-day chicken embryo dorsoposterior retina exhibited punctate ring fluores- cence. All cells from middle retina also were fluorescent, but less intensely than dorsoposterior retina cells. At each location, no obvious heterogeneity in cell population was seen. No flu- orescent cells were detected in ventroanterior retina, although low levels ofTOP were detected in ventroanterior retina by'?- F(ab'), binding. Expression of Antigen During Development. The concen- tration of TOP antigen detected was higher in dorsoposterior retina than in ventroanterior retina at every age tested from the 4-day embryo through the adult (Fig. 6A Inset), and the axis and polarity of the gradient did not change during development (Fig. 6A). The concentration of antigen detected in the dorsal halfof the retina increased S-fold between the 4th and 12th days of embryonic development and then decreased slightly the adult. Antigen concentration detected in ventral retina did not vary with development. The amount of protein per retina and FIG. 5. Autoradiographs of 14-day chicken embryo retina. (A and Bl Dark-field (A) and phase-contrast (Bl views of dorsal retina. In A, some silver grains over cell soma in the inner nuclear layer appear dim due to staining of cells by toluidine blue. (C and D) Dark-field (C) and phase- contrast 0) views of ventral retina. R, photoreceptor layer; OS, outer synaptic layer; IN, inner nuclear layer; IS, inner synaptic layer: G, ganglion cell layer; A, ganglion cell axon layer. (x630.1 Outer halves of 14-day chicken embryo dorsal or ventral retina were immersed in liquid Freon and then in liquid nitrogen and were cut in sections 16 Frn thick. Each section was incubated with antibody to TOP or P3X63 Ag8 antibody (ascites fluid diluted 1:50 with solution Bl for 45 min at 4oC washed six times (15 min) with solution B (50 ~1 per wash), incubated with 50 ~1 of solution B containing 20 pg of fluorescein conjugate of rabbit IgG anti-mouse IgG (Cappell for 45 min at 4oC and washed as above. For autoradiography, each section was incubated with 50 ~1 of solution B containing 440 mM `251-F(ab')2 (1 x lo5 cpm) and 500 clg of bovine serum albumin for 30 min at 4"C, washed as above, and coated with NTB-2 nuclear track emulsion (Kodak). Slides were exposed in the dark (22 days) at 4oC in the presence of a dessicant and stained with 0.02% toluidine blue. 2148 Biochemist?: Trisler et al. Proc. Mad. licad. Sci. USA 78 (1981) RETINA SECTION PROTEIN/RETINA 4 8 12 16 20 ADULT DAYS AFTER FERTILIZATION FIG. 6. TOP antigen in chick retina as a function of developmental age. (A) Each retina (left eye) was cut into eight sections as shown in Fig. 2. Symbols and days in ouo are: o, 8; A, 10; O , 12; , 14; v, 16; Q, 18; and * , adult. (Inset) `251-F(ab'), bound specifically is shown on the ordinate; days in ouo and adult (AD) are shown on the abscissa. o, dorsal half of retina (sections 3-6); A, ventral half of retina (sections 1, 2, 7, and 8). Data for retina from 4- and g-day embryos are shown only in Inset. (B) 0, 1251-F(ab')z bound specifically per retina; A, protein per retina; m, ordinate represents pmol 1251-F(ab')2 bound specifically per mg of protein. TOP antigen detected per retina increased 470- and 620-fold, respectively, between the 4-day embryo and the adult; the amount of antigen detected per mg of protein remained rela- tively constant (Fig. 6B). These results show that a gradient of TOP molecules is formed early in retina development, during the period ofactive neuroblast proliferation and neuron genesis, and that the gradient is maintained after neuron genesis ceases. Tissue Specificity. Highest concentrations of TOP antigen detected were in regions of the nervous system derived from prosencephalon (forebrain): retina > cerebrum > thalamus (Table 1). Low levels ofantigen were found in dorsal and ventral retina pigment epithelium, optic nerve, optic tectum, and cer- ebellum; little or no antigen was detected in heart, liver, kid- ney, or cells from blood. Chicken Embryo with an Ectopic Eye. During the course of these studies, a chicken embryo with three eyes was found (Fig. 7). The third eye was situated in the middle of the fore- head, facing in a dorsoanterior direction. Retinas from right, middle, and left eyes contained gradients of TOP molecules with normal polarity and alignment with the choroid fissure. Table 1. Distribution of TOP antigen in chicken tissues `251-F(ab')z specifically bound, pmol/mg protein Tissue Dorsal neural retina Ventral neural retina Cerebrum Thalamus Optic nerve Optic tectum Cerebellum Dorsal retina pigment epithelium Ventral retina pigment epithelium Heart, liver, kidney, or blood cells 1Cday embryo Adult 12.0 7.70 1.08 2.70 3.30 3.12 2.13 - - 0.58 0.15 - 0.23 0.46 0.26 - 0.25 - 0.002-0.055 KIlNA SECTION FIG. 7. (A) TOP antigen gradients in retinas from the right co), middle 0, and left (A) eyes of a ll-day chicken embryo with three eyes. Total `251-F(ab')2 bound per retina section is shown. Reaction mixtures contained 2.38 nM `251-F(ab')2. (B) Frontal view of head of embryo. The third eye is situated on the forehead facing in a dorsoanterior direction. The embryo had two pairs of beaks, two brains in one head, and one MY. Thus, a gradient of TOP antigen was generated with normal orientation in the supernumerary eye despite the abnormal orientation of the eye in the embryo. Species Specificity. Gradients ofTOP molecules with similar opientation and symmetry were detected in turkey, quail, and duck embryo retina, 17, 15, and 16 days after fertilization, re- spectively (Fig. 8). `=I-F(ab'), concentrations in reaction mix- tures were low (0.074-0.26 nM); thus, bound `=I-F(ab'), also was low. With 440 n?vl `=I-F(ab'),, 7.5 pmol of 12'1-F(ab'), bound specifically per mg ofdorsoposterior quail retina protein, comparable to chicken retina. The antigen was not detected in retina of goldfish, Xenopus laevis, Rana pipiens, or Fisher rats (not shown). Properties of TOP Antigen. A dorsal+ventral gradient of GM, ganglioside molecules (I13KeuAc-G,Ose,Cer) in chicken embryo retina has been postulated but not detected (15). Bovine brain gangliosides (1~10.000 $vI) did not inhibit the binding of anti-TOP antibody to retina (not shown). Rabbit antisera to mono-, di-, and trisialogangliosides (gift of C. Alving) bound to retina; however, a ganglioside gradient was not detected. 2346678 RETINA SECTION FIG. 8. TOP gradients in retina from Japanese quail (o), Coturnh coturnti japonica (15-day embryo); White Pekin duck (A), Anas pb tyrhynchos U6-day embryo); and turkey (II), Meleogris gollopovo (17s day embryo). The eggs hatch 17, 28, and 28 days after fertilization. respectively. 1261-F(ab')2 concentrations and &i/pmol were as follows: quail, 0.26 nM, 0.90 pCi/pmol; duck, 0.074 nM, 2.92 /.&i/pmol; and turkey, 0.078 nM, 2.38 &i/pmol. Biochemistry: Trisler et al. Table 2. Effect of trypsin or heat on TOP antigenicity `z51-F(ab')2 bound specifically to retina Treatment of retina cells cells Exp. O-30 min 30-40 min cpm % 1 Control + Trypsin inhibitor 1700 100 *sin + Trypsin inhibitor 93 6 Tqpsin + trypsin inhibitor - 1803 106 2 4"C, 30 min 1607 100 lOo"C, 30 min 132 8 Inactivation of TOP retina molecules in 14day chicken embryo dor- sal retina by trypsin or heat. In Exp. 1, retina cells were incubated for 30 min at 37"C, in phosphate-buffered saline alone or with 11 WM tryp- sin (crystallized three times, Worthington) or with 11 PM trypsin in- activated with 12 PM soybean trypsin inhibitor (Worthington), and then for 10 min in soybean trypsin inhibitor. TOP aacites fluid was diluted 1:lOOO; 1.86 nM `251-F(ab')z (90.5 nCi/pmol) was used. In Exp. 2, TOP and P3X63 Ag8 antibodies were diluted 1:lOO; 1.82 nM 1251- F(ab'Jz (96.9 nCi/pmol) was used. TOP antigenicity was upon incubation at 1OO'C or by incu- bation with trypsin (Table 2). All TOP antigenicity in retina cell homogenates was recovered from the 100,000 X g particulate fraction; soluble antigen was not detected (not shown). DISCUSSION Fusion of spleen cells from a mouse immunized with dorso- posterior chicken embryo retina with P3X63 Ag8 mouse my- eloma cells yielded a line of hybridoma cells that synthesizes antibody to molecules that are distributed in a topographic gra- dient in the retina. At least 35-fold more antigen was detected in dorsoposterior than in ventroanterior retina. The antigen was found by autoradiography in highest concentration in 14-day chicken embryo retina on neurites in the inner and outer syn- aptic layers. The antigen was found by immunofluorescence on most, or all, cells from dorsoposterior and middle portions of B- and 14-day chicken embryo retina. These results suggest that antigen molecules are distributed in the retina on the basis of cell position, rather than cell type. Neurons in dorsal and ventral retina differ in several ways. For example, X'enopus retina is composed ofat least three clonal domains (16). One cell on each side of the 16-cell embryo gives rise to dorsal retina, another cell gives rise to middle retina, and a third cell gives rise to cells that migrate across the midline of the embryo and form ventral retina on the opposite side. In addition, dorsal retina ganglion neurons synapse in ventral tec- turn, whereas ventral retina ganglion neurons synapse in dorsal tectum (1, 17, 18), establishing a continuous, point-to-point re- tino-tectal map. Neurons are generated in chicken embryo retinabetween the 2nd and 12th days after fertilization; central retina is the oldest portion of the retina, and peripheral retina is the youngest. The concentration of TOP molecules detected in dorsoposterior ret- ina increased as the diameter of the retina increased, and it varied with cell position at every stage tested, from the 4-day embryo through the adult, suggesting that the gradient ofTOP is established as neurons and glia (and possibly their precursors) are generated in retina. Proliferation of clonal populations of cells with different genotypes in mosaic mice results in radial patterns of cells in neural retina and retina pigment epithelium (19); whether the gradient of TOP molecules is due to clonal inheritance remains to be determined. Proc. Natl. Acad. Sci. USA 78 (1981) 2149 Dorsal and ventral retina cells also differ in adhesive speci- ficity-i.e., cells from dorsal retina adhere preferentially to cells from ventral retina or tectum and vice versa (Z&22). However, the properties of TOP molecules differ from those reported for preferential adhesion of dorsal retina cells and for retina adhe- sion factors such as cognin (4), CAM (5), and ligand and agglu- tinin (23). The dorsoposterior portion of chicken or pigeon retina con- tains a fovea, 3- to 5-fold more amacrine synapses than ventral retina (24), and a relatively high concentration of red droplets in photoreceptor cells and functions as a binocular visual field for pecking (25). Thus, dorsoposterior retina differs from other portions of retina in embryologic development, migration of cells and axons across the midline of the embryo, synapse spec- ificity, adhesive specificity, and function. The results suggest that a gradient of TOP molecules is formed by a gradient of cells which have different numbers of antigenic TOP molecules depending on cell position in the ret- ina. No evidence was found for topographically distributed dif- ferences in antigen affinity for antibody or in the proportion of cells expressing antigen. Variation in antigen accessibility has not been excluded. The mechanism of generating and main- taining the gradient of TOP, highly ordered with respect to the axis of the retina, remains to be determined. The function of TOP molecules has not been determined. However, since TOP antigen concentration detected is contin- uously graded and distributed on the basis of cell position not cell type, our working hypothesis is that TOP molecules play a role in the coding of positional information in the retina. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22 23 24 2.5 Sperry, R. W. (1963) Proc. Natf. Acad. Sci. USA 50, 703-710. Silver; J. & Sidman, R. L. (198O)j. Camp. Net&. 189, 101-111. Goldschneider. I. & Moscona. A. A. 11972) 1. Cell Biol. 53. 435449. x 3, Hausman, R. E. & Moscona, A. A. (1979) Exp. Cell Res. 119, 191-204. Thiery, J.-P., Brackenbury, R., Rutishauser, U. & Edelman, G. M. (19771 1. Bio[. Chem. 252. 6841-6845. Ei&nba&, G. S., Walsh, F: S. & Nirenberg, M. (1979) Proc. Natl. Acad. Sci. USA 76, 4913-4917. Barnstable, C. J. (1980) Nature (London) 286, 231-235. Trisler. G. D.. Donlon. MM. A.. Shain. W. G. & Coon. H. G. (1979) bed. Proc. Fed. Am. Sot: Exp. Viol. 38, 2368-2373. Kijhler, G. & Milstein, C. (1975) Nature (London) 256,495-497. Galfre, G., Howe, S. C., Milstein, C., Butcher, G. W. & How- ard, J. C. (1977) Nature (London) 266, 550-552. Berzofsky, J. A., Hicks, G., Fedorko, J. & Minna, J. (1980) 1. Biol. Chem. 255, 11188-11191. Schneider, M. D.. & Eisenbarth, G. S. (1979) 1. Immunol. Meth- ads 29,331-342. Bolton, A. E. & Hunter, W. M. (1973) Biochem.]. 133,529-539. Lowrv. 0. H.. Rosebroueh. N. 1.. Farr. A. L. & Randall. R. 1. (195lj;. Biol. Chem. 193: 26!&`5. Marchase, R. B. (1977) I. Cell Biol. 75, 237-257. Hirose. G. & Jacobson, h4. (1979) Dev. Biol. 71, 191-202. Hamdi, F. A. & Whitteridge, D. (1954) Quunt. 1. hp. Physiol. 39, 111-119. McGill. 1. I., Powell, T. P. S. & Cowan, W. M. (1966)/. Anat. 100, ti. Sanyal, S. & Zeilmaker, G. M. (1977) Nature (London) 265, 731-733. Rarbera, A. J., Marchase, R. B. & Roth, S. (1973) Proc. N&j. Acad. Sci. USA 70, 2482c2486. Gottlieb, D. I., Rock, K. & Glaser, L. (1976) Proc. Natf. Acad. Sci. USA 73, 410-414. Cafferata. R.. Panosian. 1. & Bordlev. G. (1979) Deu. Biol. 69, 108-117. Rutz, R. & Lilien, J. (1979) J, Cell Sci. 36, 323-342. Yazulla, S. (1974)]. Camp, Neural. 153, 309-324. Galifret, Y. (1968) Zeitschriftfiir Zellforschung 86, 535545.