Characterisation of novel uveal melanoma cell lines under serum-free conditions.

BACKGROUND: The establishment of long-term uveal melanoma (UM) cell lines is difficult. However, studying living cells and their behaviour in the presence of other cells and the extracellular matrix is important in terms of understanding tumour biology and malignant behaviour. We have established three UM cell lines and report a first characterisation of these cell lines. METHODS: Three established UM cell lines (UMT2, UMT26 and UMT33) were analysed according to their morphologic characteristics, melanocytic differentiation, adhesion on different extracellular matrices and proliferative activity. Copy number changes of chromosomes 1, 3, 6 and 8 were studied by multiplex ligation-dependent probe amplification (MLPA). Oncogenic mutations in UM involving exons 4 and 5 of GNAQ and GNA11, respectively, were analysed by sequencing. RESULTS: All cell lines grew in suspension. UMT2 cells were homogeneous, UMT26 and UMT33 cells heterogeneous with regard to cell size and pigmentation. All UM cell lines revealed a melanocytic differentiation. UMT2 and 33 adhered on various extracellular matrices, while UMT26 only adhered to basal membrane extract (BME). This difference corresponded to the different expression of various integrins. Ki67 was expressed by 89% of UMT2 and 95% of UMT33 cells, which thus were in a proliferative stage, while only 2% of UMT26 cells revealed immunostaining for this proliferation marker. The doubling time of UMT2 was 3 days, 12 days for UMT33, and circa 3-4 months for UMT26. MLPA revealed disomy 3 in UMT2 and monosomy 3 in UMT33. The same point mutation was found in UMT2, 26 and 33, in exon 5 of GNA11 at codon 209 (p.Q209L). CONCLUSIONS: The establishment of UM cell lines under serum-free conditions is possible. Characterisation of UMT2, 26, and 33 revealed obvious differences in cytomorphology, melanocytic differentiation, adhesion on extracellular matrices, and proliferative activity. UMT2, 26 and 33 showed the same oncogenic mutation in exon 5 of GNA11.

Antiproliferative and cytotoxic properties of bevacizumab on different ocular cells.

AIM: To evaluate the antiproliferative and cytotoxic properties of bevacizumab, a monoclonal antibody against vascular endothelial growth factor (VEGF), on human retinal pigment epithelium (ARPE19) cells, rat retinal ganglion cells (RGC5), and pig choroidal endothelial cells (CEC). METHODS: Monolayer cultures of ARPE19, RGC5, and CEC were used. Bevacizumab (0.008-2.5 mg/ml), diluted in culture medium, was added to cells that were growing on cell culture dishes. Cellular proliferative activity was monitored by 5'-bromo-2'-deoxyuridine (BrdU) incorporation into cellular DNA and the morphology assessed microscopically. For cytotoxicity assays ARPE19, RGC5, and CEC cells were grown to confluence and then cultured in a serum depleted medium to ensure a static milieu. The MTT test was performed after 1 day. The "Live/Dead" viability/cytotoxicity assay was performed and analysed by fluorescence microscopy after 6, 12, 18, 24, 30, 36, and 48 hours of incubation. Expression of VEGF, VEGF receptors (VEGFR1 and VEGFR2) and von Willebrand factor was analysed by immunohistochemistry. RESULTS: No cytotoxicity of bevacizumab on RGC5, CEC, and ARPE19 cells could be observed after 1 day. However, after 2 days at a bevacizumab concentration of 2.5 mg/ml a moderate decrease in ARPE19 cell numbers and cell viability was observed. Bevacizumab caused a dose dependent suppression of DNA synthesis in CEC as a result of a moderate antiproliferative activity (maximum reduction 36.8%). No relevant antiproliferative effect of bevacizumab on RGC5 and ARPE19 cells could be observed when used at a concentration of 0.8 mg/ml or lower. CEC and ARPE 19 cells stained positively for VEGF, VEGFR1, and VEGFR2. More than 95% of the CEC were positive for von Willebrand factor. CONCLUSIONS: These experimental findings support the safety of intravitreal bevacizumab when used at the currently applied concentration of about 0.25 mg/ml. Bevacizumab exerts a moderate growth inhibition on CEC when used in concentrations of at least 0.025 mg/ml. However, at higher doses (2.5 mg/ml) bevacizumab may be harmful to the retinal pigment epithelium.

Verteporfin photodynamic therapy induced apoptosis in choroidal neovascular membranes.

AIM: To evaluate the impact of verteporfin photodynamic therapy (PDT) on the induction of apoptosis in choroidal neovascular membranes (CNV) secondary to age related macular degeneration. METHODS: Retrospective review of 22 surgically excised CNV. 12 of these patients had been treated with PDT 3-146 days previously. Apoptotic cells were detected with the TUNEL technique and compared to the expression of CD34 (endothelial cells, EC), CD105 (activated endothelial cells), Ki-67 (proliferation marker), and cytokeratin18 (retinal pigment epithelial cells, RPE). RESULTS: CNV excised 3 days after PDT were characterised both by collapsed and patent vessels. The EC displayed a statistical significant positive TUNEL reaction when compared to the remaining treated CNV (p < 0.001) and untreated CNV (P = 0.002). The proliferative activity was reduced. CNV excised 1-5 months after PDT displayed a patent vascularisation and high proliferative activity. All membranes either treated or untreated disclosed only sporadic TUNEL positive cells within the stroma and the RPE. CONCLUSIONS: Verteporfin PDT leads to selective and effective damage of EC within CNV. Both patent and occluded vessels were lined by apoptotic EC. This finding and the increased expression of proliferation marker at later time points suggest that revascularisation after PDT is caused by angiogenesis rather than recanalisation.

Differential responsiveness to the chemorepellent Semaphorin 3A distinguishes Ipsi- and contralaterally projecting axons in the chick midbrain.

In the chick dorsal mesencephalon, the optic tectum, the developing axons must choose between remaining on the same side of the midline or growing across it. The ipsilaterally projecting axons, forming the tectobulbar tract, course circumferentially toward the ventrally situated floor plate but before reaching the basal mesencephalon, the tegmentum, gradually turn caudally. Here, they follow the course of the medial longitudinal fasciculus (MLF), located parallel to the floor plate. By in vivo labeling of tectal axons, we could demonstrate that these axons arise primarily in the dorsal tectum. To test the idea that chemorepellent molecules are involved in guidance of the nondecussating axons, we performed coculture experiments employing tectal explants from various positions along the dorso-ventral axis. Axons emanating from dorsal tectal explants were strongly repelled by diencephalic tissue containing the neurons that give rise to the MLF whereas ventral tectal axons showed only a moderate response. This inhibitory effect was substantially neutralized by the addition of anti-neuropilin-1 antibodies. A similar differential response of axons was observed when tectal explants were cocultured with cell aggregates secreting the chemorepellent Semaphorin 3A (Sema3A). Sema3B and Sema3C, respectively, did not inhibit growth of tectal axons. In addition, neither the floor plate nor Slit2-secreting cell aggregates influenced outgrowth of dorsal fibers. In Sema3A-deficient mice, DiI-labeling revealed that dorsal mesencephalic axons cross the MLF instead of turning posteriorly upon reaching the fiber tract, thus behaving like the ventrally originating contralaterally projecting axons. A differential responsiveness of tectal axons to Sema3A most likely released by the MLF thus contributes to pathfinding in the ventral mesencephalon.

The polysialic acid moiety of the neural cell adhesion molecule is involved in intraretinal guidance of retinal ganglion cell axons.

We have characterized the antigen recognized by mab10, a monoclonal antibody that has been shown to modify outgrowth of thalamic and cortical axons in vitro, and investigated the influence of this antibody on axonal growth in the chicken retina in vivo. Immunopurification, peptide sequencing, and biochemical characterization proved the epitope recognized by mab10 to be polysialic acid (PSA), associated with the neural cell adhesion molecule (NCAM). Intravitreal injections of antibody-secreting hybridoma cells were combined with whole-mount studies using the fluorescent tracer 1,1'-dioctadecyl-3,3,3', 3'-tetramethylindocarbocyanine perchlorate (DiI). Pathfinding at the optic fissure was affected, resulting in a failure of axons to exit into the nerve. Misprojections also occurred in more peripheral areas of the retina; however, axons eventually oriented toward the center. Similar projection errors were observed after enzymatic removal of PSA by injecting endoneuraminidase N (endo N). Quantitative measurements of the optic nerve diameter as well as the width of the optic fiber layer confirmed that many axons failed to leave the retina and grew back in the optic fiber layer of the retina. Our findings suggest that NCAM-linked PSA is involved in guiding ganglion cell axons in the retina and at the optic fissure.

Characterization of a new brain-derived proteoglycan inhibiting retinal ganglion cell axon outgrowth.

A proteoglycan was identified and isolated from physiological saline extracts of chick embryo brains by using a new monoclonal antibody (hybridoma clone mab Te38). The purified proteoglycan displayed an apparent molecular mass of 2500-3500 kDa, which became reduced to 370 and 600 kDa after digestion with chondroitinase ABC or chondroitinase AC. After additional treatment with keratanase the 600-kDa band was no longer detectable in Western blots. The specific epitope recognized by mab Te38 is an O-linked carbohydrate associated with the core protein. Tenascin-C, an extracellular matrix protein known to associate with several proteoglycans, copurified with the mab Te38 proteoglycan on the immunoaffinity column. Mab Te38 binds to the surface of nonneuronal cells; in sections from the primary visual system, expression is restricted to cells in the optic fissure, the dorsal optic nerve, and the chiasm. No retinal cells were found to express the mab Te38 epitope. The isolated molecule inhibited axon outgrowth from retinal explants when offered bound to a substrate consisting of either matrigel or collagen, chondroitinase treatment did not alter the inhibitory properties. The distribution and in vitro function of the Te38 proteoglycan indicate that it may serve a role in guidance of retinal ganglion cell axons.

Dual action of a carbohydrate epitope on afferent and efferent axons in cortical development.

During development of the mammalian cerebral cortex, ingrowing afferents from the thalamus take a path that is different from that of axons leaving the cortical plate. Thalamic axons arrive at the cortex at the time before their target cells of layer 4 are generated in the ventricular zone, but they invade the cortex only shortly before these cells have migrated to their final position in the cortex. Growth-promoting molecules are up-regulated in the developing cortical plate during this period. To identify such molecules, we have generated monoclonal antibodies against membrane preparations from rat postnatal cortex. In Western blots, one antibody (mAb 10) recognized a carbohydrate epitope of a glycoprotein with an apparent molecular weight extending from 180 to 370 kDa. Immunohistochemical staining revealed that the staining pattern of mAb 10 at embryonic stages delineates the pathway of thalamocortical axons, with only very faint labeling of the corticofugal pathway. In vitro assays in combination with time-lapse imaging indicated that mAb 10 has opposite effects on the growth of thalamic and cortical axons. The growth speed and axonal elongation of thalamic fibers on postnatal cortical membranes preincubated with mAb 10 was reduced compared with untreated cortical membranes. In contrast, cortical axons grew faster and stopped their growth less frequently after addition of mAb 10 to a cortical membrane substrate. Taken together, these results suggest that a carbohydrate moiety of a membrane-associated glycoprotein plays a role in the segregation of afferent and efferent cortical axons in the white matter. Moreover, the epitope recognized by mAb 10 might also contribute to regulation of the timing of the thalamocortical innervation at later developmental stages.

Interobserver variance in perceptual performance and learning.

PURPOSE: Normal observers and patients with apparent disease usually are tacitly expected to yield homogeneous thresholds in clinical tests of visual perception. The authors tested this assumption. METHODS: Through training of 70 observers, performance and improvement of performance were tested for different hyperacuity tasks using psychophysical tests. RESULTS: Although the assumption of homogeneous results might be true for many tasks limited by the physical properties of the eye, such as two-point resolution, the authors find a relatively wide variation of performance, especially in untrained observers, for tasks that require more elaborate processing in the visual cortex. Observers vary widely both in their baseline performance and in the extent and speed of learning for tasks such as vernier discrimination and stereoscopic depth perception. On average, speed of learning is inversely correlated to baseline performance, that is better initial performance usually is associated with slower improvement. CONCLUSIONS: This finding indicates that retesting of unusually high (pathologic) thresholds in clinical tests of visual perception might improve discrimination between patients whose performance is poor because of lack of familiarity with the task and who improve with training and patients who improve far less if their poor performance results from pathologic conditions.

Specification of layer-specific connections in the developing cortex.

One of the basic tasks of neurobiology is to understand how the precision and specificity of neuronal connections is achieved during development. In this paper we reviewed some recent in vitro studies on the developing mammalian cerebral cortex that have been made towards this end. The results of these experiments provided evidence that membrane-associated molecules are instrumental for the formation of specific afferent and efferent cortical projections. Substrate-bound molecules guide growing axons towards their target, regulate the timing of thalamocortical innervation and mediate target cell recognition. Moreover, a newly described glycoprotein, defined by a monoclonal antibody, revealed a molecular heterogeneity in the developing white matter. Since this molecule has opposite effects on thalamic and cortical axons, it might play a role in the segregation of axons running to and from the cortex. Substrate-bound cues are important during the formation of local cortical circuits. In vitro assays demonstrated that molecular components confined to individual cortical layers control the laminar specificity of cortical axon branching. This suggests that similar developmental strategies contribute to the laminar specification of extrinsic and intrinsic cortical circuits. Thus substrate-bound molecules might provide the framework for subsequent activity-dependent mechanisms that control the elaboration of precise connections between the cortical columns. A major challenge ahead is to identify the factors that mediate these processes and to determine their mode of action. Recently, two families of proteins, the netrins and the semaphorins/collapsins, have been identified as growth cone signals in the developing spinal cord (reviewed in Goodman, 1994; Colamarino and Tessier-Lavigne, 1995a; Dodd and Schuchardt, 1995; Kennedy and Tessier-Lavigne, 1995). Semaphorins/collapsins appear to regulate axonal guidance by repelling growth cones and by inhibiting axonal branching and synapse formation. Originally, netrins have been purified as diffusible chemoattractants for commissural axons of the dorsal spinal cord, but it is now well established that they can also function as chemorepellent factors for other classes of neurons. Since netrins are related to extracellular matrix components and since they can bind to the cell surface, they might also act as local guidance cues. A possible role of netrins and semaphorins/collapsins in the development of cortical connections is likely to be resolved in the near future. The identification of the factors that regulate specific branching patterns of cortical neurons might provide a better understanding of cortical development, but it might also be relevant to some aspects of plasticity and repair in the adult cortex.

Antibodies against the T61 antigen inhibit neuronal migration in the chick optic tectum.

Cell migration in the central nervous system depends, in part, on receptors and extracellular matrix molecules that likewise support axonal outgrowth. We have investigated the influence of T61, a monoclonal antibody that has been shown to inhibit growth cone motility in vitro, on neuronal migration in the developing optic tectum. Intraventricular injections of antibody-producing hybridoma cells or ascites fluid were used to determine the action of this antibody in an in vivo environment. To document alterations in tectal layer formation, a combination of cell-nuclei staining and axonal immunolabeling methods was employed. In the presence of T61 antibody, cells normally destined for superficial layers accumulated in the ventricular zone instead, leading to a reduction of the cell-dense layer in the tectal plate. Experiments with 5-bromo-2'-deoxyuridine labeling followed by antibody staining confirmed that the nonmigrating cells remaining in the ventricular zone were postmitotic and had differentiated. The structure of radial glial cells, as judged by staining with a glia-specific antibody and the fluorescent tracer 1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate (DiI), remained intact in these embryos. Our findings suggest that the T61 epitope is involved in a mechanism underlying axonal extension and neuronal migration, possibly by influencing the motility of the leading process.

Definition of thresholds for stereoscopic depth.

In the laboratory, thresholds for stereoscopic depth perception are usually determined by asking observers to discriminate between a stimulus with a given depth offset and its mirror image. Threshold is most often defined as the disparity difference that yields 75% or 83% correct responses. Disparities used for clinical tests of stereopsis are much higher. Here it is argued that, among other factors, this is because of the fact that clinical tests usually require the detection of a depth difference (offset versus no offset), rather than the discrimination between two directions of depth difference (in front versus behind). From a formal comparison of the two tasks, the data show that discrimination, or classification is easier by at least a factor of 2 than detection. The contribution of variations of the threshold criterion and learning to the differences between stereoacuity as measured in laboratory and clinic is also discussed. These differences are relevant to the design of tests for clinical use.

[Assessment of visual acuity in simulation and aggravation]. / Zur Einschätzung der Sehschärfe bei Simulation und Aggravation.

Following a review of various tricks and techniques described in the literature for measuring the visual acuity of less cooperative patients, three new methods are presented. The first is based on the so-called "preferential-looking" method, which has been in use for several years to determine the visual acuity of infants and small children. The second method uses polarizing filters to separate the beam paths of the two eyes and a flashlamp to project the test symbols, to prevent the patient from finding out by blinking which eye is being tested. In the third method the stimuli are presented on a computer monitor; the beam paths of the two eyes are separated by special spectacles which can briefly cover either eye while a test symbol is being shown. The results so far are encouraging.

Embryonic neurons as in vitro inducers of differentiation of nephrogenic mesenchyme.

Nephrogenic mesenchyme differentiates into epithelium as a result of morphogenetic tissue interactions. In vivo, the ureter bud is thought to induce tubular differentiation of the mesenchyme. In vitro recombination experiments have shown that various embryonic tissues can act as inducers when put in close proximity to nephrogenic mesenchyme. Induction also occurs across a porous filter. In the present study we show that only a few embryonic tissues are potent inducers in transfilter cultures in which mesenchyme and inducing tissue are separated by a membrane filter. Of the tissues tested, only embryonic spinal cord and brain were effective, whereas the ureter bud did not induce. All tissues tested sent processes through the filter. Weak inducing capacity of embryonic tissues is thus not due to a failure of the cells to make contact with the mesenchyme. To analyze which cell type within the embryonic brain possesses inducing capacity, neurons were selectively removed from primary cultures of chick tectal cells by antibody and complement-mediated cell lysis. These cultures, consisting of glial and undifferentiated cells, were then recombined with nephrogenic mesenchyme. They proved to be ineffective in inducing tubulogenesis, whereas cell populations containing neurons retained their inducing capacity. In transfilter cultures, ingrowth of neuronal processes deep into the mesenchyme, as assayed by anti-neurofilament staining, occurred within the first 24 hr of culture. Thus, it is not the time needed for processes to grow through the filter, but the time needed to grow into the mesenchyme that corresponds to the minimal induction time. These studies suggest that embryonic neurons are the most effective inducers of nephrogenic mesenchyme in vitro. Differentiation may be triggered by neuronal processes that establish cell contacts deep within the mesenchyme. Neurons might be important for nephrogenesis in vivo as well, although we can present no direct evidence to support this idea, since we failed to detect neurons at early stages of kidney development when the first tubules are induced.

The effect of neuronal cells on kidney differentiation.

During embryonic growth, tissue interactions between dissimilar cells are the driving forces of morphogenesis. Although their importance has been well known for over the past 50 years, the molecular background of these interactions has remained unelucidated. The unrecognized heterogeneity of those mesenchymal cells that are involved in the epithelio-mesenchymal tissue interactions may be one reason for this. For example, studies of kidney differentiation show that the metanephric organ rudiment contains more cell-lines than previously thought. Identification of both neural crest- and mesoderm-derived cells in the nephrogenic mesenchyme helps in re-evaluating the biology of the tubule induction. The neural crest-derived cells of the nephric rudiment differentiate into neuronal cells, and later during differentiation some of them are found in the stroma. There is also experimental evidence for the role of these neuronal cells in the morphogenetic tissue interaction.

Spatial arrangement of radial glia and ingrowing retinal axons in the chick optic tectum during development.

Neuroanatomical tracing of retinal axons and axonal terminals with the fluorescent dye, DiI, was combined with immunohistochemical characterization of radial glial cells in the developing chick retinotectal system. Emphasis was placed on the mode of the tectal innervation by individual retinal axons and on the distribution and fate of the tectal radial glial cells and their spatial relation to retinal axons. It was obvious from fluorescent images obtained from anterogradely filled axons that these axons deserted the superficial stratum opticum (SO) to penetrate the stratum griseum et fibrosum superficiale (SGFS) by making right-angled turns within the SO. Frequently, axons which had invaded the SGFS were bifurcated and had a superficial branch which remained within the SO. Terminal axonal arborization occurred at various depths within the SGFS. Characterization of the tectal glial cells and their radial fibers by means of the anti-filament antibody, R5, and post-mortem staining with the fluorescent dye, DiI, revealed the following. (a) At least from day E8 to P1, tectal glial fibers traversed all tectal layers from the periventricular location of their somata to the superficial interface between SO and pia mater. In this interface they enlarged and formed characteristic endfeet. (b) Glial endfeet covered the whole tectal surface. They showed at early ages anterior-posterior differences having a higher density in the posterior tectum. These differences disappeared at embryonic day E13. (c) After innervation, glial endfeet of the anterior tectal third were arranged in rows parallel to the retinal fibers within the SO. This arrangement was not observed in eyeless embryos. (d) Radial glial fibers could be stained with R5 from day E8 to late embryonic stages throughout their entire length. (e) At the first posthatching days, only the segments of the radial glial fibers restricted to the thickness of the SO were R5-positive, although the fibers still traversed throughout the depth of the tectum. The results are discussed in context to the genesis of the retinotectal projection.

Early innervation of the metanephric kidney.

During kidney differentiation, the nephrogenic mesenchyme converts into renal tubules and the ureter bud branches to form the collecting system. Here we show that in the early undifferentiated kidney rudiment there is a third cell type present. In whole-mount preparations of cultured undifferentiated metanephric kidneys, neurones can be detected by immunohistochemical means with antibodies against the neurofilament triplet, 13AA8, and against neuronal cell surface gangliosides, Q211. Clusters of neuronal cell bodies can be seen in the mesenchyme close to the ureter bud. The terminal endings of neurites are found around the mesenchymal condensates that later become kidney tubules. A similar distribution of neurites can be revealed in tissue sections of kidney grafts growing in the chicken chorioallantoic membranes. In primary cultures of the ureter bud cells, neurones are constantly present. In another report, we have shown that, in experimental conditions, neurones are involved in regulation of kidney morphogenesis. The present results raise the possibility that neurones of the metanephric kidney may have this function in vivo as well.

Developmental expression in embryonic rat and chicken brain of a polysialoganglioside-antigen reacting with the monoclonal antibody Q 211.

The monoclonal mouse antibody Q 211 binds to an antigen, which is expressed by postmitotic growing neurons of embryonic chicken and rat brain. In chicken, thin layer chromatography (TLC) immunostaining confirms the presence of the Q 211 antigen in at least 3 different polysialoganglioside fractions. One comigrates on TLC plates with GP1c and the others with gangliosides, which have been previously preliminary characterized as GQ1c, and as a hexasialoganglioside. Thus, 3 sialic acid residues linked to the inner galactose of a complete tetraose moiety is suggested as the common epitope of the different Q 211-antigen-active gangliosides. Also in the embryonic rat brain, immunohistochemistry reveals a transient expression of the Q 211 antigen in areas containing growing nerve fibres. Unlike chicken, however, in the rat the staining is restricted to early thalamocortical innervations and to a fibre system (probably long distance projections) connecting the mamillary body with the hippocampus formation. In ganglioside extracts from rat forebrain 2 polysialogangliosides are shown by immuno-TLC to bind Q 211. One of these fractions, occurring transiently and in parallel with histochemical staining, comigrates on TLC plates with chicken GP1c. The other comigrates with the second main Q 211 antigen-containing band of chicken, which was preliminary identified as GQ1c.

Antibodies to cell surface ganglioside GD3 perturb inductive epithelial-mesenchymal interactions.

Most epithelial sheets emerge during embryogenesis by a branching and growth of the epithelium. The surrounding mesenchyme is crucial for this process. We report that branching morphogenesis and the formation of a new epithelium from the mesenchyme in the embryonic kidney can be blocked by a monoclonal antibody reacting with a surface glycolipid, disialoganglioside GD3. In contrast, a more than 10-fold excess of antibodies to adhesive glycoproteins (N-CAM, L-CAM, fibronectin) fails to inhibit morphogenesis. Although the anti-GD3 antibody affected epithelial development, the disialoganglioside GD3 was expressed not in the epithelium, but in the mesenchyme surrounding the developing epithelia. The data raise the intriguing possibility that the anti-GD3 antibody inhibits epithelial development by interfering with epithelial-mesenchymal interactions.

Avoidance of posterior tectal membranes by temporal retinal axons.

Membrane carpets consisting of alternating membrane stripes were prepared from plasma membranes of anterior and posterior chick optic tectum. Axons from retinal explants extend neurites on these carpets. Axons of the nasal retina do not distinguish between the stripes. Axons of the temporal retina prefer to extend neurites on anterior tectal membranes. Treatment of the membrane fragments with high temperature interferes with the pattern of neurite outgrowth from temporal axons. When growing on carpets consisting of treated anterior and posterior tectal membranes, temporal retinal axons no longer distinguish between the stripes. Treatment of posterior membranes alone is sufficient to abolish the preference of temporal axons to extend neurites on anterior tectal membranes. Treatment of the anterior membranes alone has no effect. This result is best explained by a repulsive component in the posterior tectal membranes. Temporal, but not nasal, axons specifically recognize and avoid that component, with the result that they do not extend neurites on posterior tectal membrane stripes. Once the repulsive component is destroyed, temporal axons are able to extend neurites on posterior tectal membranes.