Defective lens fiber differentiation and pancreatic tumorigenesis caused by ectopic expression of the cellular retinoic acid-binding protein I

Retinoic acid plays an important role during embryonic development of vertebrates (Tickle et al., 1982; Durston et al., 1989). Vitamin A is also of prime importance for the maintenance of vision. Congenital blindness and eye malformations due to maternal vitamin A deficiency have been described in the literature (Thompson et al., 1964; Warkany and Schraffenberger, 1946). Administration of retinol or retinoic acid (RA) prior to organogenesis of the eye prevents the appearance of eye abnormalities associated with vitamin A-deficiency in rats (Dowling and Wald, 1960; Wilson et al., 1953). However, retinoic acid alone can not prevent the degeneration of the retina because it cannot be converted to retinal, which is required for the visual metabolic pathway.

Maternal hypervitaminosis A has also been shown to induce ocular abnormalities, such as anophthalmia, microphthlamia, exophthalmos and cataract (Murakami and Kameyama, 1965; Geelen, 1979). The nature of ocular abnormalities depends on the timing of hypervitaminosis A treatment. In rats, anophthalmia is induced by maternal hypervitaminosis A when the optic evagination just starts to develop. Teratogenic treatment with vitamin A, immediately after the period of optic evagination and optic cup formation, induces microphthalmia. Exophthalmos is produced at the stage when the eye has already developed, but before the orbit has reached its final shape. Congenital cataract is induced by hypervitaminosis A treatment late in the fetal life (Murakami and Kameyama, 1965; Geelen, 1979; Shenefelt, 1972). Thus, this complex teratogenic effects in embryos may suggest that retinol and retinoic acid are important in eye development.

The molecular basis for the teratogenic action of retinoic acid has not been fully elucidated. Two different types of cellular proteins are thought to mediate retinoic acid action: the nuclear receptors, which consist of the retinoic acid receptors (RARs α, β, γ) (Petkovich et al, 1987; Giguere et al., 1987; Zelent et al., 1989), and the cytoplasmic binding proteins, known as cellular retinoic acid-binding proteins (CRABP I and II) (Ong and Chytil, 1975; Giguere et al.,1990). The function of the RARs is to act as RA-inducible transcriptional factors, which interact with specific responsive elements of different genes (Glass et al., 1989; De The et al., 1990; Vasios et al., 1989). The RARs interact for efficient DNA binding with the retinoid X receptors (RXRs α, β, γ) which in turn have 9 cis-retinoic acid as ligand (Zhang et al., 1992; Kliewer et al., 1992; Heyman et al., 1992; Leid et al., 1992). Both of the CRABPs bind retinoic acid, but the affinity of CRABP II for retinoic acid seems to be lower than that of CRABP I (Chytil and Stump, 1991). Based on the finding of two opposing anteroposterior gradients in the chicken limb, one of CRABP I and the other of retinoic acid, Maden et al. (1988) suggested that the function of the CRABPs is to modulate the action of endogenous retinoids by sequestering retinoic acid and thus preventing it from activating the RARs. Recent evidence suggests that this may be the case. Boylan and Gudas (1991) reported that overexpression of CRABP I in F9 cells reduces retinoic acid activation of transcription by the RARs, suggesting that less retinoic acid reaches the nucleus. In a parallel study, Fiorella and Napoli (1991) presented evidence that the complex retinoic acid-CRABP is a substrate in retinoic acid metabolism, therefore sequestering retinoic acid and serving as an intermediate for its efficient catabolism. In both instances, CRABP would be modulating the concentration of free retinoic acid in the nucleus, hence supporting the initial hypothesis.

CRABP I is expressed in the retina but not in the lens of normal mice (Perez-Castro et al., 1989; Dolle et al., 1990, Ruberte et al., 1992). To investigate the role of CRABP I during development, we ectopically expressed it in the mouse lens using the αA-crystallin promoter. We show that, in the resulting transgenic mice, the ectopic CRABP I interferes with lens fiber differentiation, causing the formation of cataract. In addition, these mice present expression of the transgene in the pancreas and develop pancreatic carcinomas. Together these results suggest a role of CRABP I during development and cell differentiation.

The paAp(BS) plasmid, a bluescript KS(−) derivative vector containing the 412 bp BglII-BamHI αA-crystallin promoter (D. Silversides, personal communication) was cut at the HincII site located immediately downstream of the promoter. A 493 bp PvuII-StyI fragment of the bovine CRABP I cDNA (Wei et al., 1987) was ligated to the HincII site. The StyI enzyme was chosen to remove the endogenous cDNA polyadenylation site. This whole construct was cut at the SmaI site downstream of the cDNA. A 2.5 kbp BglII-BamHI fragment, containing the SV40 intervening sequence and polyadenylation site was ligated to the SmaI site. The whole construct was released from the plasmid as a NotI fragment. This DNA was microinjected into (B6CBA F₂ hybrid) fertilized eggs from which 17 F_o transgenic mice were obtained. Seven transgenic lines were generated, out of which lines α10, α37 and α38 expressed the transgene.

Total RNA from embryos, microdissected embryo parts and adult tissues were isolated as reported by Chomczynski and Sacchi (1987). The isolated RNAs purified by the above procedure were quantitated by spectrophotometric determination. To isolate RNA from the eyes, excised eyes were flash frozen in liquid nitrogen and pulverized with a Bessman Tissue pulverizer chilled with liquid nitrogen. The pulverized sample was treated in the same manner as all other tissues.

10 μg of total RNA were separated in a 1% formaldehyde-agarose gel and blotted onto nylon. The filter was prehybridized and then hybridized to a cRNA ³²P-labeled probe. Prehybridization, hybridization and washing conditions were as recommended by Strata-gene.

Embryos or tissues were isolated and fixed for 24 hours in 4% paraformaldehyde in phosphate-buffered saline (PBS) at 4°C. Following fixation, the embryos were placed in 0.5 M sucrose in PBS at 4°C for 24 hours, and embedded in OTC (Miles Laboratories). Sections were cut (8-10 μm thickness), placed on poly-L-lysine-coated slides, and stored at −70°C. Slides bearing cryostat sections were treated exactly as in the procedure described by Perez-Castro et al. (1989). Briefly, slides were postfixed with 4% paraformaldehyde, dehydrated in a series of ethanol, acetylated in 0.1 M triethanolamine (pH 8.0) and 0.25% acetic anhydride for 10 minutes, and washed in 30 mM NaCl and 3 mM sodium citrate for 10 minutes. After dehydration in a series of ethanol, the sections were prehybridized at room temperature for 2 hours in hybridization solution (50% formamide, 0.08% bovine serum albumin, 0.6 M NaCl, 0.01 M Tris-HCl pH 7.5, Denhardt’s solution, 0.12 mM EDTA, 0.1 μg of salmon sperm DNA/ml, and 0.05 μg of total yeast RNA/ml). Sections were hybridized at 45°C overnight in 20 μl of hybridization solution containing 10% dextran sulfate, 0.1% SDS, 10 mM dithiothreitol, and ³⁵S-labeled RNA at a concentration of 2.5 ng/ml. Slides were washed in 50% formamide, 0.15 M NaCl, 15 mM sodium citrate, 10 mM dithiothreitol at 50°C for 30 minutes and 75 mM NaCl, 7.5 mM sodium citrate for 30 minutes at room temperature. Remaining unhybridized probe was digested with RNase A at 20 μg/ml in RNase buffer (0.5 M NaCl, 0.1 mM Tris pH 8.0, and 1 mM EDTA for 30 minutes at room temperature). Slides were washed in RNase buffer for 10 minutes, and in 30 mM NaCl and 3 mM sodium citrate at 50°C for 2 hours, dehydrated in a series of ethanol containing 300 mM ammonium acetate and coated with a 1:1 dilution of photographic emulsion NTB-2 in water. Slides were developed after 4 days of exposure.

³⁵S-labeled RNA probes were prepared as described by Perez-Castro et al. (1989).

For ³²P-labeled RNA probes, the procedure was the same as that for 35S probes, except that 100 μCi of ³²P-αUTP (800 Ci/mmol) were dried down and 40 μM of cold UTP was added.

Isolated total RNA (50-75 μg) was dried and dissolved in 30 μl of hybridization buffer (80% formamide/40 mM Pipes pH 6.4/400 mM NaCl, and 1 mM EDTA) containing a 20-fold excess of probe (2-7×10⁵cts/minute), heated at 85°C for 10 minutes, and incubated at 45°C for at least 8 hours. To the mixture, 350 μl of 10 mM Tris-HCl pH 7.5, 5 mM EDTA, 300 mM NaCl containing 40 μg/ml of RNAse A and 2 μg/ml of RNAse T1 were added, and incubated for 30 minutes at 30°C. 50 μg of proteinase K and 20 μl of 10% SDS were added to the reactions, and incubated at 37°C for 15 minutes. Each sample was phenol extracted and ethanol precipitated with tRNA as carrier. The RNA samples were then dissolved, denatured and fractionated on a 6% sequencing gel (Zinn et al., 1983).

For gross examination of the lens, animals were anaesthetized with Avertin and their eyes dilated with neosynaphrine (2.5%). Animals were then examined under a dissecting microscope.

Eyes were dissected and immediately frozen in OTC. 8 μm sections were placed on poly-L-lysine-coated slides and stored at −70°C. Sections were fixed in 1:1 acetone:chloroform and blocked with 10% goat serum and 1% bovine serum albumin (BSA) for 30 minutes, incubated with primary turkey antibody against CRABP (Kato et al., 1985) for 2 hours and washed extensively with phosphate-buffered saline (PBS). Immediately afterwards, sections were blocked once more with 10% goat serum and 1% BSA for 15 minutes, and then rinsed several times with PBS. FITC-conjugated goat antiserum to turkey immunoglobulin G (1:500) was applied for 30 minutes at room temperature. Slides were washed extensively in PBS, mounted with cover slips and immediately examined under a fluorescence microscope.

Eyes and tissues from other organs were dissected and fixed in 4% paraformaldehyde overnight at 4°C. A window of sclera was removed from the eye just behind the cornea in order to permit better penetration of fixatives to the posterior segment. Tissues were dehydrated with three changes of 95% ethanol for one hour each, followed by the same regimen with a 100% ethanol, embedded in paraffin and sectioned. Sections were stained with hematoxylin and eosin, and examined.

To express CRABP I ectopically in the lens, a linear DNA construct consisting of the mouse αA-crystallin promoter (Chepelinsky et al., 1985) and the coding sequence of the bovine CRABP I (Fig. 1A) (Wei et al., 1987) was microinjected into fertilized mouse eggs. Seven transgenic lines were produced, and their progeny analyzed for transgene expression. Three lines, α10, α37 and α38, expressing the transgene were obtained. They contained 2, 10 and 5 copies of the transgene, respectively. All the F_o mice appeared phe-notypically normal, although they eventually died of pancreatic endocrine tumors between 7 to 9 months.

Starting with the F₁ generation, all subsequent generations of these transgenic lines developed cataracts that were characterized by a central lenticular opacity (Fig. 2A,B). Analysis indicated that cataracts were more severe in homozygous compared with heterozygous transgenic mice of line α10. However, only homozygous mice from the lines α37 and α38 had cataracts.

A variety of different tissues from progeny of transgenic animals were tested for expression of the transgene by northern blots using an SV40 probe. No hybridization was detected in any organ tested except the eye (Fig. 1B). This shows that the transgene transcript is expressed exclusively in the eye until 4.5 months of age at which time the transgene can also be detected in the pancretic tumors (see ‘transgene expression in the pancreas’).

In order to detect the expression of the transgene in the lens, normal and transgenic whole eyes were analyzed by in situ hybridization. The probes used to detect the trans gene were 240 bp of SV40 and 493 bp of the CRABP I bovine cDNA.

Whole-eye sections hybridized with SV40 antisense probe show very strong hybridization only in lenses of transgenic animals, indicating the expression of the transgene (Fig. 3B,D). Controls were done with the SV40 sense probe, and no hybridization grains were observed in transgenic and non-transgenic eye sections. The retina did not show hybridization above the level of background. These results are consistent with results from experiments using the bovine cDNA probe: hybridization grains are seen only in lenses of transgenic eye sections, showing the presence of CRABP I transcripts (data not shown). In contrast, hybridization with non-transgenic eye sections showed the absence of CRABP I transcripts. This indicates that the CRABP I expression in the lens is due to the transgene.

Transgenic transcripts are detected in the lens of embryos as early as 12.5 days post coitum (p.c.). This is in agreement with reports showing developmental regulation of the αA-crystallin promoter (Overbeek et al., 1985).

We ascertained further that CRABP I is expressed in the lenses of the transgenic animals by immunohistochemistry using a polyclonal antibody against CRABP I (Kato et al., 1985). As shown in Fig. 3E,F, in the non-transgenic lens, no immunostaining can be detected, indicating the total absence of endogenous CRABP I (Fig. 3E). Conversely, in the trans-genic lens, all the lens cells are immunostained (Fig. 3F).

Because animals were dying prematurely at 7-9 months, we investigated their cause of death. At autopsy, these animals had multinodular tumor of the pancreas. Systematic investigation showed that these tumors became visible at around 4.5 months of age. The tumors were characterized histologically and were found to derive from the islets of Langerhans (Fig. 4C,D).

Expression of the transgene in the tumors was analyzed. Total RNA from tumors and pancreas of the different trans-genic lines at different ages was isolated. In order to detect the transgenic transcript, RNase protection analysis was performed using the bovine CRABP cDNA as a probe. A 493 bp protected fragment is expected for the transgenic CRABP transcript, and a 235 bp protected fragment for the endogenous CRABP (Fig. 5). The 493 bp protected band was detected in all the RNA tumor samples, but it was not detected either in the apparently normal pancreatic tissue samples of the same transgenic mice of 4.5 months of age, or in pancreatic tissue samples of younger transgenic mice. It was not until 5.5 months of age that the transgene transcript was detected in these apparently normal pancreatic tissues.

We studied histologically the lens of animals from line α10. In the wild-type lens, a single layer of mitotically active epithelial cells lines the anterior part of the lens. At the equator, these cells cease to divide and begin to differentiate into elongated lens fibers which ultimately lose their nuclei (Fig. 2C) (McAvoy, 1980). Interestingly, the trans-genic lens is cataractous when the mice opened their eyes (day 14 after birth). We studied histologically the development of the abnormal phenotype. Accumulation of nuclei seems to become apparent between day 2 and day 3 after birth, when most of the primary and some secondary fibers in normal lenses have finished their process of denucleation. When eyes from e14, e16 and e18 day embryos, and newborns were examined, no appreciable difference from lenses of their normal littermates is found. At day 4 and day 8 after birth, lenses are characterized by the retention of the nucleus in many of the secondary lens fibers (Fig. 6B,D). The normal elongation of the lens fibers is also impaired, so that the anterior lens surface becomes flattened (Fig. 6B,D). By 4 weeks after birth, the transgenic lenses are grossly cataractous (Fig. 2D). Histologically, the lens nucleus is more eosinophilic. There are many retained nuclei in the lens cells anteriorly and posteriorly (Fig. 2E). The lens equator consists of many clefts and disorganized lens fibers (Fig. 2E,F). Morgagnian globules are noted occasionally in the cytoplasm of lens cells (Fig. 2E).

Histological comparison of heterozygous and homozygous transgenic lenses of line α10 is shown in Fig. 8. Lenses from both, heterozygous and homozygous transgenic animals show histologic abnormalities. However, lenses from heterozygous transgenic animals present a milder cataract than homozygous transgenic ones. This can be clearly distinguished by comparing the amount of tissue that presents clefts in each case (Fig. 7A,B), and by comparing the severity of the disorganization of the lens fibers at the equator (Fig. 7C,D). The alteration in shape of the heterozygous transgenic lens is not as severe as in the homozygous transgenic lens (Fig. 7A,B). This indicates that the elongation process of the homozygous transgenic lens fiber cells is more severely impaired than in the heterozygous transgenic lens. The retention of nuclei in the anterior part of the lenses seems to be similar in both cases (Fig. 7E,F), and the lens nucleus is more eosinophilic also in both cases (Fig. 7A,B). Fig. 9 shows the same histological comparison for transgenic lenses of line α37. Lenses from homozygous transgenic animals of line α37 show severe histologic abnormalities, such as the accumulation of highly eosinophilic material in the lens nucleus (Fig. 8C), disorganization of the lens fibers (Fig. 9F) and retention of pyknotic nuclei (Fig. 8I). The severity of the cataract in the lens of α37 homozygous mice is only as severe as the cataract of an heterozygous lens from line α10. The lens of heterozygous transgenic animals of line α37 is almost completely normal, with the exception of very few lens fiber cells that have retention of nuclei in the anterior part of the lens (Fig. 8H). In all cases, all other tissues from these eyes are normal.

The role of CRABP on retinoid metabolism during development has not been elucidated. It has been postulated that CRABP may sequester free retinoic acid, thus reducing the availability of retinoic acid reaching the nucleus at a sufficient level for the differential regulation of gene transcription (Maden et al., 1988; Boylan and Gudas, 1991, Fiorella and Napoli, 1991). The effects of the transgenic CRABP I on the expression of intrinsic lens proteins such as the γF-crystallin gene was examined (Maurer-Orlando et al., 1987).

We decided to look at the γF-crystallin gene, because it has three half-site repeats related to the consensus sequence (A/G)GGTCA in its promoter region, two of which are responsible of conferring ligand-dependent activation by retinoic acid (Tini et al., 1993). A downregulation of the γF-crystallin gene was expected in the transgenic lens if CRABP served to sequester free retinoic acid. With this purpose, a γF-crystallin-specific probe in northern blot analysis was used (data not shown). Interestingly, no difference in the expression of the γF-crystallin transcript from normal and transgenic lenses was detected.

Our results suggest that ectopic expression of CRABP I in the lens alters the normal differentiation of secondary lens fibers by interfering with the process of denucleation.

During normal lens development, at day 15 of embryogenesis, the posterior epithelial cells have fully occupied the lens vesicle lumen and their nuclei have migrated anteriorly. These constitute the primary lens fibers that will be forming the lens nucleus. From this point onwards, secondary lens fibers will derive from lens epithelial cells at the equator (Vrensen et al., 1991). The fact that transgenic lenses show histologic abnormalities on and after day 2 after birth indicates that the secondary lens fibers are affected. Interestingly, the ectopic expression of CRABP begins at around day 12 in embryogenesis. These facts taken together indicate that the expression and accumulation of CRABP I preceded the appearance of the phenotype in these transgenic mice. This temporal relationship further supports the hypothesis that ectopic expression of CRABP I in the lens is causally related to the formation of cataracts, and that there may be a threshold requirement for the amount of the ectopic CRABP I to induce this abnormal phenotype. These conclusions are further strengthened by the observations that in two of the lines (α37, α38), cataracts are observed only in homozygous animals and that, in line α10, more severe cataracts were observed in homozygous compared with heterozygous mice. We presume in these cases that homozygosity confers a higher level of CRABP I expression than heterozygosity because of the gene dosage effect. Similar gene dosage effects have been reported for several other transgenes expressed in the lens of transgenic mice (Breitman et al., 1989; Griep et al., 1989; Capetanaki et al., 1989; Harrington et al., 1991).

We would like to emphasize that many different proteins, such as chloramphenicol transferase and β-galactosidase, have been expressed in the lens of transgenic mice without generating a phenotype (Goring et al., 1987; Wawrousek et al., 1990). Thus, the expression of a protein in the lens is not enough per se to produce an abnormal phenotype. In contrast, when an oncogene or a viral gene has been expressed in the lens, tumor formation or microphthalmia have been reported (Mahon et al., 1987; Khillan et al., 1987; Griep et al., 1989). An interesting report by Capetanaki et al. (1989) showed that overexpression of the vimentin gene in the lens of transgenic mice inhibits lens cell differentiation and leads to cataract formation. However, formation of cataract in these mice does not start until they are at least five months old.

A very interesting and unexpected result in all our transgenic lines is the expression of the transgene in the pancreas. These observations exclude the possibility of the existence of genomic positional effects on the expression of the transgene. Expression of the transgene in the pancreas was not expected, since detailed studies on the regulation of the murine αA-crystallin promoter have shown lens-specific expression (Wawrousek et al., 1990). However, a recent study by Srinivasan et al. (1992) shows by PCR analysis that the αA-crystallin is also expressed in non-ocular tissues, raising the possibility of pancreatic expression due to the promoter itself. Another possible explanation is that the combination of the promoter and the CRABP I cDNA may be promoting expression outside the lens. This seems likely, since other regulatory sequences could exist in the downstream region of the gene. The possibility that a coding region may serve as a transcriptional enhancer element has been previously shown for the murine histone H3 gene (Hurt et al., 1989). However, what seems peculiar about our αA/CRABP I transgenic mice is the time of appearance of the transgenic transcript in the pancreas at around 5.5 months of age. This could be due to the fact that, under our assay conditions, we are not able to detect the transgenic transcript until it is very abundant. This would seem to be the case since our observations suggest that the tumors arise from the β Langerhans islets which, under normal conditions, only represent 1% of the whole pancreatic mass.

An interesting question, but one for which we do not have yet a conclusive explanation, is the cause of pancreatic carcinomas. Kato et al. (1985b) by radioimmunoassay experiments, reported the presence of both the cellular retinolbinding protein I (CRBP I) and CRABP I in the Langerhans islets of the rat. They suggested that the retinoids and their binding proteins may play important metabolic roles within the islet cells, and hence that they may be involved in some way in the endocrine function of the islets. If this is also the case in the mouse, then overexpression of CRABP I in the islets would lead to the deregulation of the normal homeostasis of the retinoids in mouse pancreatic islet cells. Experiments are under way to address this issue.

What is the function of CRABP I in normal development? It is likely that CRABP I is involved in regulating retinoid homeostasis in cells that are the targets for retinoid. For example, in the mouse embryo, CRABP I has been found in all of the tissues that are known to be teratogenic targets of retinoic acid (Perez-Castro et al., 1989; Maden et al., 1990; Ruberte et al., 1992). Thus, CRABP I could be controlling the actual concentration of free retinoic acid in these tissues, and when retinoic acid exceeds certain saturation levels (i.e. maternal hypervitaminosis A), then CRABP I would no longer be able to regulate the concentration of retinoic acid reaching the nucleus, and consequently teratogenesis occurs. Conversely, at a higher concentration of CRABP I, it could exert its effect by preventing retinoic acid from reaching the nucleus to a level sufficient for the differential regulation of gene transcription. This hypothesis has been supported by three different sets of experiments. First, two opposing gradients have been observed in the anteroposterior axis of the chicken limb, one of retinoic acid and another reciprocal gradient of CRABP. It was proposed that these two gradients would result in a flatter distribution of retinoic acid-CRABP complex, but in a steeper gradient of free retinoic acid available for binding to the Retinoic acid receptors (RARs) (Maden et al., 1988). Second, experiments with F9 EC cells in vitro showed that increased expression of CRABP I reduces the expression of retinoic acid induced genes (Boylan and Gudas, 1991), supporting the idea that CRABP I sequesters retinoic acid within the cell, thereby preventing it from reaching its nuclear targets. Third, Fiorella and Napoli (1991) have recently shown in an in vitro system that CRABP I modulates the concentration of free retinoic acid not only by sequestering retinoic acid but also by serving as a conduit for its efficient catabolism.

Two recent transgenic experiments in which the reporter gene lacZ was expressed under the control of the retinoic acid responsive element (RARE) from the βRAR, confirms the existence of retinoic acid in both the lens and retina (Rossant et al., 1991; Balkan et al 1992). In our case, ectopic expression of CRABP I in the lens interferes with the normal program of differentiation from epithelial to lens fiber cells. The effect of the ectopic CRABP I on lens fiber differentiation could be explained as a change in the concentration of free retinoic acid, either because this ectopic CRABP I binds retinoic acid, or because the ectopic CRABP I increases the catabolism of retinoic acid. In both instances, the concentration of the normally available free retinoic acid would have been altered, thus affecting differential regulation of gene transcription, which is ultimately reflected in the alteration of the program of differentiation from epithelial to lens fiber cells.

To see if we could determine the mode of action of the ectopic CRABP I in our transgenic mice, we examined the effects of the transgenic CRABP I on the expression of the intrinsic lens protein gene, γF-crystallin. The γF-crystallin gene has a RARE in its promoter region (Tini et al., 1993). A downregulation of the γF-crystallin gene was expected in the transgenic lens if CRABP sequesters free retinoic acid. By northern blot analysis of total RNA using a γF-crystallinspecific probe (Maurer-Orlando et al., 1987), we could not detect any difference in the expression levels of the γF-crystallin transcripts in normal and transgenic lenses. One possible explanation for this result is that the changes in transcription rates of the γF-crystallin gene are not detected by examination of steady state levels of transcripts in whole lenses because of the relatively large amounts of γF-crystallin transcripts that are present in the lens. The potential effects of the transgene on the expression of other lens proteins are the subject of future studies.

We believe that transgenic mice bearing the αA-crystallin/CRABP transgene provide us with the opportunity to elucidate the interactions between retinoids, nuclear receptors and retinoid-binding proteins. It will be interesting to cross some of the βRAR-lacZ transgenic mice (Rossant et al., 1991; Balkan et al 1992) with our transgenic αA-crystallin/CRABP mice and look for changes in the pattern of lacZ expression which might result from changes in intracellular concentrations of retinoic acid in transgenic mice.

The results of our transgenic experiments suggest that retinoic acid is involved in the development of the lens and in lens fiber differentiation. We have presented the first evidence that CRABP I interferes with normal development and cell differentiation when ectopically expressed, suggesting an important role for this protein in embryonic development, probably by regulating retinoid homeostasis in the cells where it is normally present.

We thank Drs Paul Overbeek and David Silversides for providing us with the αA-crystallin promoter; Drs DeWitt Goodman, David Talmage and Tom Wright for kindly sharing with us their CRABP I antibody and giving us advice in the immunohistochemistry experiments; Drs Brigid Hogan, Christopher Tuggle, Lucie Jeanotte and Robert Maxson, for helpful discussions and critically reading the manuscript. We also thank David Stanford for technical assistance with histologic sections.