|Molecular Vision 2004;
Received 15 January 2004 | Accepted 22 June 2004 | Published 6 July 2004
EST analysis of mouse retina and RPE/choroid cDNA libraries
Hisashi Ida,1,2 Sharon A. Boylan,1,2 Andrea L.
Weigel,1,2 Zeljka Smit-McBride,1,2
Leonard M. Hjelmeland1,2
Departments of 1Biological Chemistry and 2Ophthalmology, University of California, Davis, CA; 3Institute of Statistics, National Tsing Hua University, Hsin-Chu, Taiwan; 4Section on Molecular Structure and Function, National Eye Institute, National Institutes of Health, Bethesda, MD
Correspondence to: Leonard M. Hjelmeland, Vitreoretinal Research Laboratory, School of Medicine, University of California, One Shields Avenue, Davis, CA, 95616-8794; Phone: (530) 752-2250; FAX: (530) 752-2270; email: firstname.lastname@example.org
Purpose: cDNA libraries from the mouse retina have recently been reported, but no well characterized library from the retinal pigment epithelium (RPE) or choroid of the mouse has yet appeared in the literature. To complement these libraries and to provide the first mouse RPE/choroid library, we used freshly dissected tissue from adult C57BL/6J mice to construct new retina and RPE/choroid libraries.
Methods: Eyes from 100 six to eight week old C57BL/6J mice were dissected in groups of 10. The whole retina and RPE/choroid were isolated individually and then homogenized before RNA isolation. Over 5000 clones each were sequenced from the unamplified and un-normalized retina and RPE/choroid libraries. All sequences were analyzed using GRIST (GRouping and Identification of Sequence Tags), a bioinformatics program for gene identification and clustering.
Results: The RPE/choroid library contained 3145 clusters with 76% of the clusters representing single clones. Nearly 87% of the clusters corresponded to named genes in GenBank, and 8% of the RPE clusters remain unidentified. The retina library contained 3190 clusters of which 78% represented only one clone. Approximately 85% of the clusters matched sequences in GenBank, and 9% of the clusters remain unidentified. The clones most abundant in each library were all well-known sequences and both libraries contained a number of tissue specific or tissue-enhanced genes.
Conclusions: These new libraries should provide a valuable resource for gene discovery and cDNAs for expression analysis and functional studies.
The mouse serves as an excellent model to further define genes important to the health of the posterior pole of the eye. Invaluable information about the molecular and pathological basis of several retinal diseases has come from elucidating the gene mutations in mouse retinal degeneration models [1-5]. Sixteen mutations causing retinal degeneration in the mouse have been identified, and several studies have appeared on retinal degeneration associated with age and environmental factors such as light and oxidative stress [1,6-11]. Other advantages to using the mouse include the ability to isolate fresh RNA preparations, the ability to control environmental factors, and the ability to experimentally manipulate the genome using techniques such as transgenics and the construction of knock-out or knock-in lines.
cDNA libraries from mouse retina and retinal pigment epithelium (RPE) should provide valuable tools for identifying new disease loci and studying gene expression during disease development. Most notably, cDNA libraries of high quality from the developing mouse eye and adult retina have recently been developed using freshly dissected tissue [12,13]. To complement these libraries, and to provide the first mouse RPE/choroid specific library, we undertook the construction of new retina and RPE/choroid libraries using freshly dissected tissue from adult C57BL/6J mice. Expressed sequence tag (EST) analysis of over 5000 clones from each library identified a large dataset of non-redundant gene clusters. The data provided good estimates for the most highly expressed sequences in each tissue and revealed potentially novel or previously unidentified ESTs.
RNA isolation from RPE/choroid and retina
Approximately 200 eyes from six to eight week old male C57BL/6J mice were dissected in groups of 10. Animal care guidelines comparable to those published by the US Public Health Service (Public Health Service Policy on Humane Care and Use of Laboratory Animals) were followed. Globes were removed and placed immediately into a petri dish kept on ice and containing 0.5 M EDTA/PBS (without calcium chloride and magnesium chloride, pH 7.4). Eyes were dissected in an RNase free environment using a stereo zoom microscope (Nikon SMZ800, Tokyo, Japan). The optic nerve head was cut and the anterior segments were removed with a circumferential incision of the eye. The whole retina was peeled away, and then the pigmented layer including RPE/choroid was peeled gently from the sclera as a sheet in PBS/EDTA (Figure 1). The RPE/choroid was processed in batches of 5 eyes and the retina in batches of 2 to 4 eyes. Each batch was placed in a tube containing 600 μl RLT lysis buffer (QIAgen, Valencia, CA). Before RNA isolation, tissues were first homogenized with a 22 gauge needle and a QIAshredder spin column. Total RNA was isolated using the QIAgen RNeasy kit (QIAgen) following the manufacturer's protocol. The purity and yield of RNA were determined by measuring the optical density at 260 and 280 nm. The ratio of OD260 to OD280 measured consistently between 1.8 and 2.0. The typical yield of total RNA per mouse was approximately 1 μg from RPE/choroid and 5-10 μg from retina (both eyes). Total RNA preparations were pooled into groups of 10, and their integrity was determined by SYBR Gold stained agarose gels (Molecular Probes, Inc., Eugene, OR). RNA was visualized by scanning the gels in a phosphorimaging instrument (Storm 860, Molecular Dynamics, Sunnyvale, CA; image not shown).
Preparation of RPE/choroid/sclera for histology
In order to visually verify the specificity of dissection, a representative eye was prepared for examination by light microscopy. Just as the RPE/choroid sheet was being peeled away, the eyecup was placed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M cacodylate buffer. The whole tissue was postfixed with 1% osmium tetroxide/cacodylate and dehydrated through a series of graded alcohols. The tissue was embedded in poly/Bed 812 resin (Polysciences, Inc., Warrington, PA) and serial sections were cut 1 μm in thickness. Sections were stained with 2% toluidine-blue, and examined by light microscopy.
cDNA library construction
Complete details of library construction are described elsewhere . Briefly, poly(A) RNA, isolated by oligo(dT) cellulose column chromatography, was used for the synthesis of cDNA. A Not I primer-adaptor [GAC TAG TTC TAG ATC GCG AGC GGC CGC CC(T)15] and SuperScript II reverse transcriptase (Invitrogen Corp., Carlsbad, CA) were used for first strand synthesis. After second strand synthesis with E. coli DNA polymerase, cDNA fragments > 500 bp were fractionated on a Sephacryl S-500 column. Libraries were made from the first two 35 μl fractions containing cDNA. Sal I linkers were ligated onto the blunt ends of the cDNA. The Not I/Sal I fragments were directionally cloned into the Not I/Sal I sites of the pSPORT1 vector (Invitrogen). Plasmids were transferred by electroporation into E. coli DH10B cells.
cDNA sequencing and bioinformatics
High throughput sequencing was performed on more than 5500 individual RPE and more than 5600 individual retina clones at the NIH Intramural Sequencing Center (NISC). Data were analyzed as described elsewhere . EST sequences were identified and clustered using GRIST (GRouping and Identification of Sequence Tags), a bioinformatics program that uses sequence match parameters derived from BLAST program .
Statistical estimation of library depth
Our analysis of library depth makes use of the well-known problem of estimating numbers of species given a fixed area in which sampling is performed. Results from the estimator/model ACE (Abundance-based Coverage Estimator)  are reported. The total number of samples is expressed as the sum of a sequence of observed numbers. Let N1 represent the number of species where only one individual was observed, N2 equals the number of species where 2 individuals were observed, up to Nn equal to the number of species for which n individuals was observed. With this formulation, N0, the number of species that exist for which no individual has yet been found, can be estimated by fitting expressions to the curve for species up to N10. The ACE approach uses the frequency of observed rare species to estimate the number of missing species (N0).
This problem statement is analogous to the problem of sampling a large library of ESTs to find the number of unique sequences that exist in the library. If we let a specie be a unique cluster in the library and the number of individuals in that specie be equal to the number of clones in the resulting cluster, then a series expression can also be written for the abundance of unique clusters in the library. As Chao et al. have shown , very good estimates of N0 can be made by using the data for N1...N10, and the SPADE (Species Prediction And Diversity Estimation) algorithm  provides a simple means for calculating the estimated value of N0 with 95% confidence limits.
During synthesis of the RPE/choroid and retina libraries, cDNA was fractionated into two size ranges and subcloned into libraries designated mi and mj for the RPE/choroid and mk and ml for the retina. The mi library contained 16x106 primary recombinants with an average insert size of 1.8 kbp. Out of over 2870 quality sequence reads with an average size of 573 bp, less than 2% contained no inserts, 2.3% contained mitochondrial genome sequences, and 1% contained rRNA. For mj, there were 10.2x106 primary recombinants with an average insert size of 1.5 kbp. From over 2700 reads with an average size of 554 bp, no inserts were found in 2.2% of the sequences, 3.8% contained mitochondrial genome, and less than 1% had rRNA. The mk library from the mouse retina contained 16x106 primary recombinants with an average insert size of 1.9 kbp. Of more than 2700 quality sequence reads with an average size of 559 bp, only 1.9% contained no insert, 3.6% had mitochondrial sequences, and less than 1% had rRNA. For ml, there were 17.6x106 primary recombinants with an average insert size of 1.2 kbp. From over 2900 quality reads with an average size of 564 bp, no inserts were found in 1.6% of the sequences, 6.1% contained mitochondrial genome sequence, and less than 1% rRNA. In summary, normally >91% of recombinants contained inserts that appeared to be from the source material, and hence contained valuable clones.
High quality reads from both libraries were analyzed using GRIST . GRIST identified and grouped ESTs into clusters that represent transcripts from the same gene. The analysis yielded 3145 gene clusters in the RPE/choroid library, of which 76% contain single clones, and thus potentially unique genes (Table 1). Three percent of the clusters match sequences from UniGene but not GenBank, 1% match ESTs in dbEST, 1% do not match other DNA sequences but do match possible ORFs, and 8% of the RPE clusters remain unidentified (Table 2). The retina library contains 3190 gene clusters, with 78% of the clusters containing single clones (Table 1). Approximately 9% of the retina clusters remain unidentified, 4% match sequences from UniGene but not GenBank, 1% match ESTs from dbEST, and 1% do not match other DNA sequences, but do match possible ORFs (Table 2).
Both libraries show that only 9% of the clusters appear more than twice (Table 1). The estimator/model ACE was used to estimate the number of missing clusters (species) for both libraries . The ACE approach uses the frequency of observed rare species to estimate the number of missing species. For the combined RPE/choroid libraries, the number of missing clusters was calculated to be 7745. For the combined retina libraries, it was 9176. Thus, the total number of clusters in the combined RPE/choroid libraries is estimated to be 10,896 with a 95% confidence interval of (10400 to 11500). In the combined retina libraries, the total number is estimated to be 12,346 with a 95% confidence interval of (11700 to 13000).
Tissue specific clones
The mouse eyes were dissected such that the whole retina could first be peeled away from the posterior pole and then the RPE/choroid peeled away from the sclera (Figure 1). Our data analysis indicated however that there was some contamination of the RPE library with retina cDNA. For the most part, the most abundant photoreceptor transcripts are present in the RPE library at less than 10% of the level that they are seen in the retina (data not shown). No contamination of the retina by the RPE was evident from the library data. Except for CRALBP, none of the abundant RPE transcripts was seen in the retina library (data not shown).
The most highly represented clones in the RPE/choroid and retina libraries are listed in Table 3 and Table 4, respectively. These lists are derived from the contents of NEIBank in August 2003. A complete list of the Mouse Retina and Mouse RPE/Choroid libraries are available at the NEIBank web site. Table 3 and Table 4 also list characteristic RPE and retina markers found in our mouse libraries. These tissue specific markers are represented by both abundant and non-abundant cDNAs.
While cDNA libraries cannot be completely faithful representations of the normal abundance of transcripts, they can give some indications to the level of expression of abundant transcripts. Previously published NEIBank libraries for human lens and retina show that the expected tissue specific and tissue preferred genes are indeed well represented by cDNA. Using the NCBI Homologene database, we identified probable homologous genes represented in the mouse retina and RPE/choroid libraries and in the equivalent NEIBank human libraries made in a similar fashion without amplification. We then compared the frequency of abundant genes in one species with the other.
In Table 3 and Table 4, the M/H column shows the ratio of frequency of occurrence of each gene in mouse and human libraries. In retina, the majority of abundant and tissue specific transcripts in mouse have similar abundance in human (M/H between 0.5 and 2). RPE/choroid, in contrast shows a much greater level of discordance between the two species. For example, retinal G protein coupled receptor (RGR) opsin is extremely abundant in mouse and is found at lower levels in human. Some gene transcripts that are highly abundant or moderately abundant in the mouse RPE/choroid library, such as retinol dehydrogenase 5 and LRAT, are not represented in the (larger) human collection. On the other hand, genes such as glutathione peroxidase 3, opticin, and bestrophin that are abundant in the human library are absent from the mouse collection. Indeed, none of the ESTs for the mouse homolog of bestrophin in dbEST and Unigene are from eye or head libraries, most come from testis .
This set of un-normalized libraries for the retina and RPE/choroid of the C57BL/6J mouse represents part of the ongoing efforts of the NEIBank program. Previous to this work, we could only find a single well characterized cDNA library for the adult C57BL/6 retina. In the works published by Farjo et al. and Yu et al., roughly the same percentage of unique sequences were found, and the library also appears to have the same extent of contamination by mitochondrial and ribosomal sequences [12,13]. It should be noted that the publicly available sequence data for the mouse was generated from the C57BL/6 mouse. The overall profile of the distribution of numbers of sequences found in each cluster is similar for our retina and RPE/choroid libraries and the vast majority (>75%) of sequences were found only once.
We used an approach that is well known in ecological studies for estimating the total number of species in a discrete area that is sampled in a random fashion. The correlate to our approach is to consider a cluster a species, and the number of clones found to be like the number of individuals counted in the sample. Using this approach, we were able to estimate that our set of observed sequences is likely to represent approximately 28% of the total numbers of clusters which could be found in both libraries given very much larger numbers of sequence reads. The numerical approach to this method of estimation utilizes data only from the number of clusters (species) having from 1 to 10 individuals counted. The development of this method is reviewed in . It is interesting to note that a modeling approach has been used to estimate coverage for genomics clones in a library .
The clones most abundant in each library were all known sequences and both libraries contained a number of tissue specific or tissue-enhanced genes. There is good agreement between our list and the list of Yu et al. of the most highly expressed genes in the adult mouse retina . Six of 12 genes in the list of Yu et al. are in our list of the top 10 clones found in our retina library. The mostly highly expressed sequence in the RPE/choroid library was prostaglandin D2 synthase. This gene is also expressed in the choroid plexus and has previously been shown to be expressed only in the RPE when examining the posterior pole of the eye . These data appear to conflict with the data published by Sharon et al. . In their work on SAGE libraries of human retina and RPE for example, the authors stated that of the fifty most common tags observed in the human RPE, 29 were orphan tags (did not correspond to any known gene), and of these 29, 22 appeared to be uniquely expressed in the RPE. This assertion was based on a sequence search of all available SAGE libraries at the time of the work. An examination of the published literature on the SAGE technique reveals several important issues. First, a large number of tags cannot be identified by comparison with cDNA libraries . Second, attempts to validate SAGE tags against model transcriptomes derived from completely sequenced genomes from human, mouse, and Arabidopsis also lead to large numbers of orphan or unidentified tags [23,24]. At the present time, it is not clear whether these tags represent systematic methodological errors in the generation of the tags or incomplete annotation of the genomes with respect to alternative mRNA splicing, antisense gene expression, or problems in the assumptions of the size of 5' and 3' UTRs in generating model transcriptomes.
It is interesting to note that some tissue specific transcripts were observed only once among the total number of sequences in the library. MCT-3, for example, is the monocarboxylate transporter expressed only in the RPE. Some reported tissue specific transcripts such as bestrophin were not observed in our sequence reads. Bestrophin protein was first detected in mouse RPE at postnatal day P10, and its message was detected in whole eye with highest expression in early postnatal development . It is likely that mouse the mouse bestrophin sequence might reside among the roughly 65% of the total number of transcripts in the library which have not yet been sampled. Interestingly, Yu et al. also did not find bestrophin in their library . In addition, none of the ESTs for mouse bestrophin in dbEST or Unigene is from eye or head libraries (most are from testis) while in the NEIBank human RPE/choroid library transcripts for bestrophin are moderately abundant. This certainly suggests that this gene is expressed at lower levels in mouse than in human RPE. Finally, the different observations on the abundance of bestrophin in the mouse RPE may reflect a strain variation. The most recent observation of bestrophin expression was performed with BALB/c mice, while the libraries of Yu et al. and our own were derived from the C57BL/6J strain. A comparison of single nucleotide polymorphisms among several mouse strains indicates that BALB/c is in a cluster of strains including A/J and C3H/He, while the B6 strain is somewhat distant from this cluster in an evolutionary tree . The C57BL/6J strain varies in a number of functional properties from BALB/c, including sensitivity to light-induced photoreceptor degeneration and age-related retinal degeneration.
Indeed, a number of other RPE expressed genes have quite different levels of abundance in mouse and human RPE/choroid libraries, while retina libraries seem to be more similar. This may reflect real functional differences between the rodent and primate RPE that could be relevant to diseases such as AMD that occur in human but not in mouse. Alternatively, this could be related to differences in tissue processing and differences between an inbred (mouse) and outbred (human) population.
Finally, these libraries should provide a valuable resource for continuing work in the mouse for studies ranging from in situ or northern expression analysis to the identification of novel splice forms which may have a tissue specific pattern of expression.
This research was supported by NIH Grant EY06473 (LMH), Foundation for Fighting Blindness Grant (LMH), an unrestricted grant from Research to Prevent Blindness (Department of Ophthalmology, University of California, Davis), NEI Core Grant P30EY12576, and the National Science Council of Taiwan (AC). We thank Peter Dudley and Jack McLaughlin of NEI for help in supporting NEIBank.
1. Danciger M, Lyon J, Worrill D, LaVail MM, Yang H. A strong and highly significant QTL on chromosome 6 that protects the mouse from age-related retinal degeneration. Invest Ophthalmol Vis Sci 2003; 44:2442-9.
2. Pittler SJ, Baehr W. Identification of a nonsense mutation in the rod photoreceptor cGMP phosphodiesterase beta-subunit gene of the rd mouse. Proc Natl Acad Sci U S A 1991; 88:8322-6.
3. Travis GH, Brennan MB, Danielson PE, Kozak CA, Sutcliffe JG. Identification of a photoreceptor-specific mRNA encoded by the gene responsible for retinal degeneration slow (rds). Nature 1989; 338:70-3.
4. Akhmedov NB, Piriev NI, Chang B, Rapoport AL, Hawes NL, Nishina PM, Nusinowitz S, Heckenlively JR, Roderick TH, Kozak CA, Danciger M, Davisson MT, Farber DB. A deletion in a photoreceptor-specific nuclear receptor mRNA causes retinal degeneration in the rd7 mouse. Proc Natl Acad Sci U S A 2000; 97:5551-6.
5. Haider NB, Naggert JK, Nishina PM. Excess cone cell proliferation due to lack of a functional NR2E3 causes retinal dysplasia and degeneration in rd7/rd7 mice. Hum Mol Genet 2001; 10:1619-26.
6. Chang B, Hawes NL, Hurd RE, Davisson MT, Nusinowitz S, Heckenlively JR. Retinal degeneration mutants in the mouse. Vision Res 2002; 42:517-25.
7. Klein JA, Longo-Guess CM, Rossmann MP, Seburn KL, Hurd RE, Frankel WN, Bronson RT, Ackerman SL. The harlequin mouse mutation downregulates apoptosis-inducing factor. Nature 2002; 419:367-74.
8. Choi S, Hao W, Chen CK, Simon MI. Gene expression profiles of light-induced apoptosis in arrestin/rhodopsin kinase-deficient mouse retinas. Proc Natl Acad Sci U S A 2001; 98:13096-101.
9. Hao W, Wenzel A, Obin MS, Chen CK, Brill E, Krasnoperova NV, Eversole-Cire P, Kleyner Y, Taylor A, Simon MI, Grimm C, Reme CE, Lem J. Evidence for two apoptotic pathways in light-induced retinal degeneration. Nat Genet 2002; 32:254-60.
10. Yamada H, Yamada E, Ando A, Esumi N, Bora N, Saikia J, Sung CH, Zack DJ, Campochiaro PA. Fibroblast growth factor-2 decreases hyperoxia-induced photoreceptor cell death in mice. Am J Pathol 2001; 159:1113-20.
11. Mervin K, Stone J. Regulation by oxygen of photoreceptor death in the developing and adult C57BL/6J mouse. Exp Eye Res 2002; 75:715-22.
12. Farjo R, Yu J, Othman MI, Yoshida S, Sheth S, Glaser T, Baehr W, Swaroop A. Mouse eye gene microarrays for investigating ocular development and disease. Vision Res 2002; 42:463-70.
13. Yu J, Farjo R, MacNee SP, Baehr W, Stambolian DE, Swaroop A. Annotation and analysis of 10,000 expressed sequence tags from developing mouse eye and adult retina. Genome Biol 2003; 4:R65.
14. Wistow G, Bernstein SL, Wyatt MK, Behal A, Touchman JW, Bouffard G, Smith D, Peterson K. Expressed sequence tag analysis of adult human lens for the NEIBank Project: over 2000 non-redundant transcripts, novel genes and splice variants. Mol Vis 2002; 8:171-84 <http://www.molvis.org/molvis/v8/a24/>.
15. Wistow G, Bernstein SL, Touchman JW, Bouffard G, Wyatt MK, Peterson K, Behal A, Gao J, Buchoff P, Smith D. Grouping and identification of sequence tags (GRIST): bioinformatics tools for the NEIBank database. Mol Vis 2002; 8:164-70 <http://www.molvis.org/molvis/v8/a23/>.
16. Chao A, Lee SM. Estimating the number of classes via sample coverage. Journal of the American Statistical Association 1992; 87:210-7.
17. Chao A, Ma MC, Yang MCK. Stopping rules and estimation for recapture debugging with unequal failure rates. Biometrika 1993; 80:193-201.
18. Shen T-J, Chao A, Lina C-F. Predicting the number of new species in further taxonomic sampling. Ecology 2003; 84:798-804.
19. Petrukhin K, Koisti MJ, Bakall B, Li W, Xie G, Marknell T, Sandgren O, Forsman K, Holmgren G, Andreasson S, Vujic M, Bergen AA, McGarty-Dugan V, Figueroa D, Austin CP, Metzker ML, Caskey CT, Wadelius C. Identification of the gene responsible for Best macular dystrophy. Nat Genet 1998; 19:241-7.
20. Wendl MC, Marra MA, Hillier LW, Chinwalla AT, Wilson RK, Waterston RH. Theories and applications for sequencing randomly selected clones. Genome Res 2001; 11:274-80.
21. Beuckmann CT, Gordon WC, Kanaoka Y, Eguchi N, Marcheselli VL, Gerashchenko DY, Urade Y, Hayaishi O, Bazan NG. Lipocalin-type prostaglandin D synthase (beta-trace) is located in pigment epithelial cells of rat retina and accumulates within interphotoreceptor matrix. J Neurosci 1996; 16:6119-24.
22. Sharon D, Blackshaw S, Cepko CL, Dryja TP. Profile of the genes expressed in the human peripheral retina, macula, and retinal pigment epithelium determined through serial analysis of gene expression (SAGE). Proc Natl Acad Sci U S A 2002; 99:315-20.
23. Fizames C, Munos S, Cazettes C, Nacry P, Boucherez J, Gaymard F, Piquemal D, Delorme V, Commes T, Doumas P, Cooke R, Marti J, Sentenac H, Gojon A. The Arabidopsis root transcriptome by serial analysis of gene expression. Gene identification using the genome sequence. Plant Physiol 2004; 134:67-80.
24. Chrast R, Scott HS, Papasavvas MP, Rossier C, Antonarakis ES, Barras C, Davisson MT, Schmidt C, Estivill X, Dierssen M, Pritchard M, Antonarakis SE. The mouse brain transcriptome by SAGE: differences in gene expression between P30 brains of the partial trisomy 16 mouse model of Down syndrome (Ts65Dn) and normals. Genome Res 2000; 10:2006-21.
25. Bakall B, Marmorstein LY, Hoppe G, Peachey NS, Wadelius C, Marmorstein AD. Expression and localization of bestrophin during normal mouse development. Invest Ophthalmol Vis Sci 2003; 44:3622-8.
26. Lindblad-Toh K, Winchester E, Daly MJ, Wang DG, Hirschhorn JN, Laviolette JP, Ardlie K, Reich DE, Robinson E, Sklar P, Shah N, Thomas D, Fan JB, Gingeras T, Warrington J, Patil N, Hudson TJ, Lander ES. Large-scale discovery and genotyping of single-nucleotide polymorphisms in the mouse. Nat Genet 2000; 24:381-6.