Molecular Vision 2001; 7:253-260 <http://www.molvis.org/molvis/v7/a36/> Received 2 July 2001 | Accepted 18 November 2001 | Published 19 November 2001

Download Reprint

Modulation of retinal cell populations and eye size in retinoic acid receptor knockout mice

Guomin Zhou,^1,2 Richelle C. Strom,¹ Vincent Giguere,³ Robert W. Williams¹
 
 

¹Center for Neuroscience, University of Tennessee, Memphis, TN; ²Department of Histology and Embryology, Fudan University, Shanghai, China; ³Molecular Oncology Group, Royal Victoria Hospital, McGill University, Montreal, Canada

Correspondence to: Robert W. Williams, Department of Anatomy and Neurobiology, University of Tennessee, 855 Monroe Avenue, Memphis, TN, 38163; Phone: (901) 448-7018; FAX: (901) 448-7193; email: rwilliam@nb.utmem.edu

Abstract

Purpose: The retinoic acid receptors are expressed from early stages of development in the diverse tissues that make up the vertebrate eye. Their loss has subtle effects on eye development. We adapted sensitive quantitative trait locus (QTL) mapping methods to assess consequences of inactivating alleles of the alpha and beta receptors, Rara and Rarb, on eye and retinal development. Rara is of particular interest because this gene is a candidate for Nnc1, a QTL that controls retinal ganglion cell proliferation.

Methods: We studied lines of mice in which expression of the a1 isoform of Rara or all isoforms of Rarb had been disrupted by gene targeting. We measured eye weight, lens weight, retinal area, and retinal ganglion cell number in each of six genotypes (Rara and Rarb -/-, +/-, +/+; 10-25 cases/genotype).

Results: Loss of either protein is associated with a small but significant loss of eye weight and retinal area. However, only the Rarb knockout has a significant effect on the ganglion cell population and the loss of both wildtype alleles leads to an 8,000 cell deficit. Surprisingly, loss of the Rara a1 isoform that is expressed in this cell population from early stages has no effect on number. Null alleles of both genes have little if any effect on lens growth.

Conclusions: Despite its expression in embryonic retina, Rara is unlikely to be the Nnc1 QTL. In contrast, Rarb, a gene that maps to Chr 14 and which is not an Nnc1 candidate gene, has a significant effect on cell number and is therefore a QTL controlling this key population. This raises the intriguing possibility that normal allelic variants of Rarb modulate the ganglion cell population in other vertebrates, including humans.

Introduction

Retinoic acid is a potent morphogen and teratogen with widespread effects on limb, eye, and brain development [1-9]. The actions of retinoic acid are mediated by two small families of DNA-binding receptors: the conventional RA receptors (RARs) and the retinoid X receptors (RXRs). These receptors bind to promoter elements of target genes, often as RAR-RXR heterodimers [10,11].

The RAR family consists of three genes: Rara on mouse Chr 11 at 58 cM in the Hoxb-Krt1 interval (17q12 in human), Rarb on Chr 14 at 1.5 cM (3p24.3), and Rarg on Chr 15 at 58 cM in the Hoxc-Krt2 interval (12q13) [1,12-16]. All three genes and their proteins are expressed in the embryonic mouse eye, although only Rara is expressed heavily in retina [17]. In contrast, Rarb and Rarg are preferentially expressed in mesenchymal and neural crest derivatives, including the ciliary body, choroid, sclera, and cornea.

Knockouts of these genes and their isoforms, singly and in combination, have been reported previously by Giguere, Chambon and colleagues [10,18,19], and what is perhaps most striking is the relatively mild, even undetectable phenotypes associated with single knockouts [10,18-20]. In contrast, double knockouts usually have strong or lethal phenotypes. Long ago Sewall Wright made an insightful suggestion, namely that sets of related genes (so-called paralogs), such as the retinoic acid receptors, may be "maintained by selection for quantitative effects, while protected from drastic effects of inactivation or loss" [21,22]. Functional overlap and pleiotropy may be associated with increased trait variability, viability, and plasticity. Normal allelic variants in the retinoic acid receptors may behave like quantitative trait loci (QTLs).

We are interested in determining whether inactivation of these receptors have subtle and unsuspected quantitative effects on eye development for the following reason. Rara maps in the same location as Nnc1, a QTL that we have shown has a prominent effect on the proliferation of retinal ganglion cells in many common strains of mice [23-25]. The common inbred strain C57BL/6J has the low Nnc1 variant whereas DBA/2J and C3H/HeJ have the high variant. The difference between these strains is close to 10,000 cells (15-20% of the total ganglion cell population) and Nnc1 accounts for as much as half the genetic variance in this cell population. The RARA a1 receptor is expressed at an early stage of retinal development in mouse (E10.5), at the initiation of ganglion cell generation [7]. The association between Rara gene and Nnc1 is thus supported by time and place of expression, as well as by chromosomal location. As we show in this paper, however, the effects of inactivation of Rara and Rarb on the retinal ganglion cell population run precisely counter to expectation; only the inactivation of Rarb (a gene that maps to proximal Chr 14) markedly reduced the retinal ganglion cell population.

Methods

Mice

Animals were treated in accordance with the ARVO Statement for the Use of Animals. The Rara line of mice that we used (known formally as the Rara^tm2Ipc line) have been targeted with a 9 kb insert that disrupts exon 3 of Rara. Homozygotes do not express an a1 isoform [26]. The Rarb line (the Rara^tm1Ipc line) contains an inset that disrupts the transcription of all isoforms of the beta receptor [27]. In both knockouts the mutation was introduced into embryonic stem cells derived from 129S2/SvPas that were subsequently injected into C57BL/6 blastocysts. Carriers were backcrossed to C57BL/6 stock for several generations and the C57BL/6-enriched KO stock were subsequently intercrossed to maintain the disrupted alleles. All of the animals that we have studied had the black coat color typical of strain C57BL/6. All animals were bred, raised, and genotyped at McGill University and were shipped as adults to the University of Tennessee for quantitative analysis.

Eyes and optic nerves were taken from 103 mice belonging to each of the three genotypes of Rara and Rarb. The mix of sexes among the genotypes varied, but the sex ratio across all genotypes was close to 1:1 (49 males, 54 females). The age of mice ranged from 35 to 218 days. The average age of all Rara stock was approximately 85 days (range from 50 to 130 days), whereas that of Rarb mice was 117 days (range from 35 to 220 days). We also collected data on body weight and fixed brain weight.

Genotyping

DNA was isolated from tails and genotyped by PCR or by Southern blotting. Rara: Mice were genotyped using the method of Li et al. [28]. Genomic DNA was digested with BglII and probed using the pSX. Rarb: Tail DNAs were genotyped by PCR using conventional methods. Reaction product (10 ml) was run on 1.4% agarose, stained with ethidium bromide, and photographed. Three primers were used together to identify Rarb wildtype and mutant alleles. Primer 1: 5'-GTAGCCATCGAGACACAGAGT-3' anneals to the 5' region of exon 6 was used as a common primer; primer 2: 5'-TGGTAGCCCCGATGACTTGTCC-3' anneals to the 3' region of exon 6 that is deleted in the knockout allele; and primer 3: 5'-CCTCTGAGCCCAGAAAGCGAA-3' anneals to the promoter region of the Pgk-neo cassette. Primers 1 and 2 amplify a 121 bp wildtype product whereas primers 1 and 3 amplify a 385 bp knockout product.

Eye weight and lens weight

Animals were deeply anesthetized with Avertin (about 1.0 ml IP). Right eyes were enucleated and weighed immediately. Within one minute the animals were perfused with 0.1 M phosphate buffered saline followed by 1.25% glutaraldehyde and 1.0% paraformaldehyde in 0.1 M phosphate buffer, and then by 2.5% glutaradehyde and 2.0% paraformaldehyde in 0.1 M buffer for electron microscopy [29,30]. Freshly dissected eyes were subsequently placed in the same aldehyde fixative and weighed again 1 h to 5 days later. The weights of fixed eyes, whether perfused (left eye) or immersion fixed (right eye), were generally 6.1% less than those of fresh eyes. The correlation between fresh and fixed eye weights is high (r=0.96). Lens weights were also obtained for these mice.

Retinal wholemounts and sections

The cornea was removed and retina was gently separated from the choroid plexus and the ciliary body. Five radial cuts extending 1-2 mm into the retina toward the optic disc were made, and the retina was flattened onto a slide and coverslipped in Gelvatol [31]. Wholemounts were prepared from the left eyes of a subset of 32 Rara cases (about 10 per genotype) and 9 Rarb cases (3 per genotype). Low power drawings of retinal wholemounts were traced on a digitizing tablet and retinal areas were determined with a precision of better than 0.5 mm². A 1 mm wide strip of retina and pigment epithelium, extending from the head of the optic nerve to the inferior ora serrata was cut and embedded flat in Spurr's resin. A set of 1 mm thick sections were cut along the radial axis, mounted, and stained with hematoxylin. Complete central to peripheral cross-sections of the ventral retina were drawn at low power. The radial depth of cells in the inner and outer nuclear layers was subsequently determined at 7 to 11 evenly spaced sites along sections from three animals belonging to each of the six genotypes at 400x using differential interference contrast optics.

Estimation of retinal ganglion cell number

We used a quantitative electron microscopic assay to estimate the total population of retinal ganglion cells in 47 mice (28 Rara and 19 Rarb animals). A single nerve was counted in each case. The methods we have used have been described fully [29]. In brief, a series of 22 high magnification electron micrographs were taken using a systematic random protocol across a transverse section of each nerve. A counting frame was traced on the negatives and all axons within the frame and intersecting the upper and right edges were counted using conventional counting rules. To ensure that unmyelinated fibers were recognized, negatives were counted while wearing magnifying glasses. The effective magnification was above 25000x. The average density of axons was multiplied by the area of the nerve cross-section to estimate the total axon population. Data were entered into a relational database (FileMaker Pro, FileMaker, Inc., Santa Clara, CA).

Regression analysis

Eye weight, lens weight and retinal area vary with age, sex, and body weight. This presents a problem in comparing these traits from different genotypes since each genotype may include different numbers of males and females, animals of different ages, etc. Multiple linear regression was performed to investigate relations among these variables, and to reduce unwanted effects on these factors when comparing genotypes. Data on litter size, age, sex, and body weight were available for all mice. Eye weight increases linearly as a function of the logarithm of age and for this reason we used log-transformed age data in the regression analysis. Exploratory analysis and multiple linear and non-linear regressions were carried out using the program DataDesk 6 (Data Description Inc., Ithaca NY). As in our previous studies [30,32] when necessary we have normalized data on eye weight, lens weight, and retina area to those expected of 22 g females at an age of 75 days. Body weights of the three genotypes of each of the two knockout groups did not differ in general. However, for unknown reasons, one set of Rarb heterozygotes (three female littermates, 103 day old) was extraordinarily obese (50-60 g). Their eye weighs were normal.

Results

Both eyes from a sample of 103 mice were weighed and averaged. Our main purpose was to quantify effects of inactivation of the Rara gene product (a candidate for the Nnc1 QTL) on retinal development. To ensure adequate power we examined eyes from as many as 25 cases of each of the three genotypes (74 total). In comparison, our analysis of the Rarb knockout involved a smaller sample size; approximately 10 cases of each of the three genotypes (29 total, 10 -/-, 7 +/-, and 12 +/+).

Eye weights

The average eye weights (mean±SE in mg) of three Rara genotypes (-/-, +/-, and +/+) were 18.3±0.2, 18.7±0.2, and 18.8±0.2, respectively. After adjusting for differences in age, sex, and body weight by multiple linear regressions, we estimate that the loss of each wildtype allele is associated with a drop in eye weight of 0.3 mg (p=0.02, Figure 1). The loss of wildtype Rara alleles also has a small effect on body weight of about 0.6 g per allele (p=0.03 when treated as an additive effect and p=0.015 when the wildtype allele is treated as a dominant allele; 1 tailed tests). The same is true of brain weight; the loss of a single wildtype allele is associated with a reduction in brain weight of about 7 mg (mean brain weight is about 450 mg). The effects of the wildtype Rara allele on eye weight is dominant and there is no significant difference between heterozygotes and wildtype (p=0.7). This is true for the other major weight traits, body and brain. The Rara effect on eye weight, while statistically significant, is modest in size and amounts to only a 0.6 mg reduction from a base of about 19 to 20 mg. Rara gene dosage only accounts for 3% of the total phenotypic variance.

Rarb effects on eye size

Eyes of Rarb knockout mice are significantly smaller than those of heterozygotes and wildtype mice. The averages of the three genotypes without correction are 17.28±0.32, 18.47±0.36, and 18.83±0.25 mg. Our sample of Rarb mice had a wide age range, but even with correction for this factor, the effect of genotype is still highly significant (p=0.002). In Figure 2 we illustrate our best estimate of the pure gene dosage effect of Rarb on eye size after correction for differences in age, sex, and body weight. The loss is about 1.2 mg, which is twice that in the Rara knockouts. Approximately 3.5% of the total variance in eye weight is accounted for by dosage of the Rarb wildtype allele.

Lens weight

Both fixed lenses from a subset of 57 mice were weighed and averaged (38 Rara, 19 Rarb). The average lens weights (mean±SE in mg) of three Rara genotypes (-/-, +/-, and +/+) were 5.84±0.13, 5.56±0.15, and 5.93±0.12, respectively. The difference of lens weight among three genotypes of Rara is very small and insignificant (p=0.6). Rarb had a very slight effect on lens weight. However, the significance of the effect is entirely dependent on a single knockout outlier with an unusually small lens. If this case is excluded, then Rarb also has no appreciable effect on lens weight (Figure 3). There is however a dramatic qualitative effect of the Rarb knockout. In 7 of 9 Rarb knockouts, the posterior surface of the lens is darkly pigmented as a result of the retention of a postlenticular membrane as described in the initial analysis of the Rarb knockout [33]. Both Rara and Rarb have an effect on total eye weight, but not on the weight of the lens. Lens makes up 25-30% of total eye weight and the dosage effect discussed in the preceding section is restricted to the non-lenticular part of the eye.

Retinal area

Wholemounts were prepared from eyes of 32 Rara animals and 9 Rarb animals. Retinal area is also reduced slightly in Rara knockouts (Rara areas: 19.2 mm² (-/-), 19.6 mm² (+/-), and 20.1 mm² (+/+); p=0.035). The difference between wildtype and knockout groups amounts to about 0.7 mm² after correction for sex, age, and body weight. Just over 10% of the total variance in retinal area is accounted for by gene dosage. Unlike eye weight, an additive model fits the data better than a dominance model; heterozygotes tend to have intermediate values. Retinal area is also reduced in the Rarb line for both the absolute values (19.4 mm², 20.5 mm², and 21.0 mm² for the three genotypes) and for the residuals (Figure 4, blue symbols).

Retinal histology

Cell and plexiform layers of retinas of Rara and Rarb knockout mice have apparently normal structure (Figure 5). Inner and outer nuclear layers are of the appropriate thickness. Cell size and vascular patterns appear normal. Given the generally mild quantitative changes detected in eye weight and retinal area, there may be numerous subtle quantitative distortions in the numbers or characteristics of retinal cells in knockouts and heterozygotes. One example, and a focus of this study, is illustrated by the loss of retinal ganglion cells in the Rarb knockout mice, described in the next paragraph.

Retinal ganglion cell population

The ganglion cell population is unaffected by the loss of the Rara a1 isoform. Cell counts are 63,000±2,000 (+/+, n=9), 64,000±2,300 (+/-, n=9), and 62,500±1,200 (-/-, n=10), which are not significantly different (p=0.8, ANOVA; Figure 5). Multiple regression analysis that controls for differences in body and brain weight differences among groups does not change the neutrality of the Rara knockout. In contrast, Rarb has a significant effect on RGC number: 64,800±2,300 SE (+/+, n=8), 60,900±4,000 (+/-, n=3), 57,400±2,600 (-/-, n=8; p<0.05; Figure 6). The differences among the groups are consistent with an additive model of gene dosage with each wildtype allele contributing approximately 3,700 ganglion cells. The effect of Rarb is robust, and a multiple regression analysis that takes into account pleiotropic effects that this gene may have on body weight or brain weight does not materially alter the significance of the effect (p<0.05 with control for body and brain weight, 3,500 ganglion cells/allele).

Discussion

Synopsis

Our aim in this report has been to critically assess the effects of two key retinoic acid receptors on the size and structure of the mouse eye, lens, retina, and on the population of the retinal ganglion cells. This work was motivated primarily to test whether Rara is a viable candidate gene for the Nnc1 QTL. Previous studies have looked at qualitative effects of inactivating the major isoforms of Rara and Rarb on ocular development, but with the exception of an aberration in the development of the hyaloid vessel in the Rarb knockouts (persistent hyperplastic primary vitreous) [33,34], no phenotypes have yet been described for knockouts of single receptors in this family. In our work, we have used biometric analysis of the type used in complex trait analysis with the hope of detecting subtle quantitative deviations in knockouts. We have succeeded in detecting several statistically significant and novel changes in both Rara and Rarb knockouts.

Expression of Rara and Rarb in relation to effects of the eye

Mori and colleagues [17] have recently described spatial and temporal patterns of retinoic acid receptor expression in eyes of mice from E10.5 through maturity using immunohistochemical techniques. Rara and Rarb and their proteins are expressed with the characteristics noted below.

Retina and PE: Rara is expressed intensely and persistently in the mouse retina starting before the generation of any postmitotic cells (E10.5). Expression is initially widespread but eventually is most intense in the middle of the INL (possibly bipolar cells), and finally in a subset of cells in the ganglion cell layer and the INL. In contrast, Rarb expression is weak and is restricted to the INL and the outer neuroblast (ventricular) layer during the period of most intense neurogenesis (E14 to P7). Rara is expressed at all ages in the pigment epithelium. In contrast, Rarb is expressed in the pigment epithelium only transiently (E14 to P7) and at much lower levels.

Lens: Rara, but not Rarb is expressed at all stages in the lens.

Neural crest/mesenchymal derivatives: Rarb is expressed at high levels in sclera and choroid from E10.5 through to about P4. Both Rara and Rarb are expressed in the iris and ciliary body from E14 through to maturity.

Our finding that Rarb has a more pronounced effect on total eye weight than does Rara may be consistent with the intense and long lasting expression of Rarb and the meager expression of Rara in periocular mesenchyme that gives rise to sclera and choroid. Our assumption is that the activation of gene transcription by Rarb in a normal molecular context has a cumulative positive trophic effect on eye development. This gene's actual role in transcriptional networks is certainly more complex and at a minimum involves interactions with its main dimerization partners. The weight of the lens is not materially affected by loss of either Rara or Rarb in isolation, a finding that is surprising given the persistent expression of Rara in lens throughout life. The Rarb knockout does have a severe and presumably indirect effect on lens development by disrupting the normal course of development of the vitreous and the transient blood supply to the lens.

Nnc1 and Rara

The early and intense expression of Rara in mouse retina, particularly in the ganglion cell layer and INL, makes it a strong candidate regulator of the development of ganglion cells. When combined with its location in the same 0.2% of the mouse genome as the Nnc1 locus [24] and in vitro results showing that RA application alters the differentiation of retinal cells [35], the potential of allelic variants of Rara to modulate ganglion cell proliferation appears compelling. In contrast, the transient and weak expression of Rarb makes it a less compelling candidate for a gene influencing retinal cell populations. However, our results contradict these expectations: the Rara a1 knockout mice have normal numbers of ganglion cells whereas the Rarb knockout mice have significantly fewer ganglion cells than either wildtype or heterozygous littermates. There are possible explanations for this unexpected finding, but any that tries to preserve Rara as a candidate of Nnc1 will be somewhat less parsimonious than the alternative hypothesis: that Nnc1 is not an allele of Rara. One important caveat is that the a2 isoform of Rara may provide a fully functional backup allele in the Rara a1 knockout. The a2 is expressed in some CNS regions, particularly the developing spinal cord and striatum [36] and we should not discount the possibility that there is compensatory expression in retina.

Given the expression patterns of Rarb, this gene's effects on ganglion cell number is likely to be indirect and could be associated with the persistence of hyaloid vessels and altered vitreal composition. The vitreous normally contains a high concentration of fibroblast growth factor [37], and if the availability of FGF is disturbed this could alter ganglion cell survival. [38]. It would be interesting to determine whether the Rarb knockout has an effect on ganglion cell proliferation or the severity of naturally-occurring cell death in mice.

The hitchhiker effect

The interpretation of the effects of gene inactivation in knockouts mice relies on a comparison with the phenotypes of wildtype and heterozygous animals. The use of littermate controls would appear to be an almost ideal control because parental effects and environmental factors will be shared. However, there is an important caveat [39], namely that the segment of the chromosome that carries the knockout gene is derived from strain 129 and this chromosomal interval will carry large numbers of adjacent 129 alleles that may affect the phenotype. For example, the entire Hoxb-Krt1 interval in the Rara a1 knockout mice has a homozygous 129 haplotype, whereas that of the wildtype littermates has a pure C57BL/6 haplotype. This region also happens to have very high gene density; in other words there are hundreds of hitchhiking genes inherited from strain 129. When dealing with complex traits such eye size and neuron number, it is possible that any one of these alleles will exacerbate or neutralize effects of the knockout. In our study there is no exacerbation of a Rara effect on retinal ganglion cell number: all three genotypes have the same population. But it is possible that a hitchhiking 129 allele masks a Rara effect. To make this possibility more concrete consider the following model: 129X1/SvJ has a larger population of retinal ganglion cells than does C57BL/6J [29] and this difference is probably generated by the Nnc1, a locus that overlaps Rara. When we transfer high type Nnc1 alleles from 129 onto the C57BL/6 background, we expect an increase in ganglion cell number. This increase could amount to 5,000 cells per allele. If at the same time the loss of the functional alpha 1 isoform has an independent effect that decreases the number of cells by 5,000 per allele, then Rara and Nnc1 will counterbalance and the phenotypes of all three genotypes will be similar. Given this problem, the appropriate control is a B6.129 congenic strain in which the congenic interval covers the particular knockout gene that is been studied. These control lines will be easy to make from the B6.129<Chr 11> and B6.129<Chr 14> that are now being constructed by Nadeau and colleagues [40]. As an interim solution to this problem, we are now studying effects of a second knockout located in the same interval (Thra) in which the same 129 alleles have been introgressed onto C57BL/6J. This knockout does appear to have a significant effect on ganglion cell number. Our analysis of single nucleotide polymorphisms (SNPs) in 129X1/SvJ and C57BL/6J indicates that most of the distal half of Chr 11 (from 48 cM to the telomere) has a close common descent. This reduces the risk of a hitchhiker effect associated with the introgressed 129 chromosomal segment.

The QTL approach to the analysis of knockouts

Complex trait analysis is a genetic technique that makes it possible to map gene loci that are responsible for heritable quantitative differences between individuals. These heritable differences can often be relatively small, and sensitive methods and large sample sizes are often used to detect QTLs. In this study of Rara and Rarb knockouts, we provide an example of how biometric analysis of the type used in QTL mapping [23,30,32] can reveal effects that would be hard to detect using standard qualitative histological methods. Eye weight has proved to be both an easy and remarkably sensitive way to detect differences generated by normal and abnormal alleles. The analysis of the Rara and Rarb mice provides an interesting reverse genetic approach to complex trait analysis. Rarb allelic variants could now be considered "candidate" QTL. It is possible that natural polymorphisms of this gene contribute to variation in ganglion cell number in other populations of mice and possibly even in humans. The combination of standard forward genetic approaches (from phenotype to gene) with this reversed approach could increase the efficiency of complex trait analysis.

Acknowledgements

This work was funded by the National Eye Institute (EY12991 and EY13070) to RW, and the Canadian Institute for Health Research (CIHR) and the National Cancer Institute of Canada (VG). We thank Jiangming Luo for helping generating Rarb knockout mice and Céline Champigny for technical assistance.

References

1. Giguere V, Ong ES, Segui P, Evans RM. Identification of a receptor for the morphogen retinoic acid. Nature 1987; 330:624-9.

2. Sulik KK, Dehart DB, Rogers JM, Chernoff N. Teratogenicity of low doses of all-trans retinoic acid in presomite mouse embryos. Teratology 1995; 51:398-403.

3. Tini M, Otulakowski G, Breitman ML, Tsui LC, Giguere V. An everted repeat mediates retinoic acid induction of the gamma F-crystallin gene: evidence of a direct role for retinoids in lens development. Genes Dev 1993; 7:295-307.

4. LaMantia AS, Bhasin N, Rhodes K, Heemskerk J. Mesenchymal/epithelial induction mediates olfactory pathway formation. Neuron 2000; 28:411-25.

5. LaMantia AS, Colbert MC, Linney E. Retinoic acid induction and regional differentiation prefigure olfactory pathway formation in the mammalian forebrain. Neuron 1993; 10:1035-48.

6. Rubin WW, LaMantia AS. Age-dependent retinoic acid regulation of gene expression distinguishes the cervical, thoracic, lumbar, and sacral spinal cord regions during development. Dev Neurosci 1999; 21:113-25.

7. Dräger UC, McCaffery P. Retinoic acid and development of the retina. Prog Retin Eye Res 1997; 16:323-46.

8. Duncan T, Swint C, Smith SB, Wiggert BN. Levels of retinoic acid and retinaldehyde dehydrogenase expression in eyes of the Mitf-vit mouse model of retinal degeneration. Mol Vis 1999; 5:9 <http://www.molvis.org/molvis/v5/a9/>.

9. Stull DL, Wikler KC. Retinoid-dependent gene expression regulates early morphological events in the development of the murine retina. J Comp Neurol 2000; 417:289-98.

10. Giguere V. Retinoic acid receptors and cellular retinoid binding proteins: complex interplay in retinoid signaling. Endocr Rev 1994; 15:61-79.

11. Mangelsdorf DJ, Umesono K, Evans RM. The retinoid receptors. In: Sporn MB, Roberts AB, Goodman DS, editors. The Retinoids. New York: Raven Press; 1994. p. 319-49.

12. Benbrook D, Lernhardt E, Pfahl M. A new retinoic acid receptor identified from a hepatocellular carcinoma. Nature 1988; 333:669-72.

13. Brand N, Petkovich M, Krust A, Chambon P, de The H, Marchio A, Tiollais P, Dejean A. Identification of a second human retinoic acid receptor. Nature 1988; 332:850-3.

14. Petkovich M, Brand NJ, Krust A, Chambon P. A human retinoic acid receptor which belongs to the family of nuclear receptors. Nature 1987; 330:444-50.

15. Siracusa LD, Rosner MH, Vigano MA, Gilbert DJ, Staudt LM, Copeland NG, Jenkins NA. Chromosomal location of the octamer transcription factors, Otf-1, Otf-2, and Otf-3, defines multiple Otf-3-related sequences dispersed in the mouse genome. Genomics 1991; 10:313-26.

16. Nadeau JH, Compton JG, Giguere V, Rossant J, Varmuza S. Close linkage of retinoic acid receptor genes with homeobox- and keratin-encoding genes on paralogous segments of mouse chromosomes 11 and 15. Mamm Genome 1992; 3:202-8.

17. Mori M, Ghyselinck NB, Chambon P, Mark M. Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes. Invest Ophthalmol Vis Sci 2001; 42:1312-8.

18. Chambon P. The retinoid signaling pathway: molecular and genetic analyses. Semin Cell Biol 1994; 5:115-25.

19. Kastner P, Grondona JM, Mark M, Gansmuller A, LeMeur M, Decimo D, Vonesch JL, Dolle P, Chambon P. Genetic analysis of RXR alpha developmental function: convergence of RXR and RAR signaling pathways in heart and eye morphogenesis. Cell 1994; 78:987-1003.

20. Krezel W, Dupe V, Mark M, Dierich A, Kastner P, Chambon P. RXR gamma null mice are apparently normal and compound RXR alpha +/-/RXR beta -/-/RXR gamma -/- mutant mice are viable. Proc Natl Acad Sci U S A 1996; 93:9010-4.

21. Wright S. Evolution and genetics of populations. Vol. 3. Chicago: University of Chicago Press; 1977.

22. Hartman JL 4th, Garvik B, Hartwell L. Principles for the buffering of genetic variation. Science 2001; 291:1001-4.

23. Williams RW, Strom RC, Goldowitz D. Natural variation in neuron number in mice is linked to a major quantitative trait locus on Chr 11. J Neurosci 1998; 18:138-46.

24. Williams RW. Neuroscience meets quantitative genetics: using morphometric data to map genes that modulate CNS architecture. In: Morrison J, Hof P, editors. Quantitative Neuroanatomy: A picture is worth a thousand words, a number is worth a thousand pictures. Washington: Society for Neuroscience 1998; p. 66-78. [Updated online version]

25. Strom RC, Williams RW. Cell production and cell death in the generation of variation in neuron number. J Neurosci 1998; 18:9948-53.

26. Lufkin T, Lohnes D, Mark M, Dierich A, Gorry P, Gaub MP, LeMeur M, Chambon P. High postnatal lethality and testis degeneration in retinoic acid receptor alpha mutant mice. Proc Natl Acad Sci U S A 1993; 90:7225-9.

27. Luo J, Pasceri P, Conlon RA, Rossant J, Giguere V. Mice lacking all isoforms of retinoic acid receptor beta develop normally and are susceptible to the teratogenic effects of retinoic acid. Mech Dev 1995; 53:61-71.

28. Li E, Sucov HM, Lee KF, Evans RM, Jaenisch R. Normal development and growth of mice carrying a targeted disruption of the alpha 1 retinoic acid receptor gene. Proc Natl Acad Sci U S A 1993; 90:1590-4.

29. Williams RW, Strom RC, Rice DS, Goldowitz D. Genetic and environmental control of variation in retinal ganglion cell number in mice. J Neurosci 1996; 16:7193-205.

30. Zhou G, Williams RW. Eye1 and Eye2: gene loci that modulate eye size, lens weight, and retinal area in the mouse. Invest Ophthalmol Vis Sci 1999; 40:817-25.

31. Heimer GV, Taylor CE. Improved mountant for immunofluorescence preparations. J Clin Pathol 1974; 27:254-6.

32. Zhou G, Williams RW. Mouse models for the analysis of myopia: an analysis of variation in eye size of adult mice. Optom Vis Sci 1999; 76:408-18.

33. Grondona JM, Kastner P, Gansmuller A, Decimo D, Chambon P, Mark M. Retinal dysplasia and degeneration in RARbeta2/RARgamma2 compound mutant mice. Development 1996; 122:2173-88.

34. Ghyselinck NB, Dupe V, Dierich A, Messaddeq N, Garnier JM, Rochette-Egly C, Chambon P, Mark M. Role of the retinoic acid receptor beta (RARbeta) during mouse development. Int J Dev Biol 1997; 41:425-47.

35. Kelley MW, Turner JK, Reh TA. Retinoic acid promotes differentiation of photoreceptors in vitro. Development 1994; 120:2091-102.

36. Mollard R, Viville S, Ward SJ, Decimo D, Chambon P, Dolle P. Tissue-specific expression of retinoic acid receptor isoform transcripts in the mouse embryo. Mech Dev 2000; 94:223-32.

37. Schulz MW, Chamberlain CG, de Iongh RU, McAvoy JW. Acidic and basic FGF in ocular media and lens: implications for lens polarity and growth patterns. Development 1993; 118:117-26.

38. Guillemot F, Cepko CL. Retinal fate and ganglion cell differentiation are potentiated by acidic FGF in an in vitro assay of early retinal development. Development 1992; 114:743-54.

39. Gerlai R. Gene-targeting studies of mammalian behavior: is it the mutation or the background genotype? Trends Neurosci 1996; 19:177-81.

40. Nadeau JH, Singer JB, Matin A, Lander ES. Analysing complex genetic traits with chromosome substitution strains. Nat Genet 2000; 24:221-5.