Opposite Orientations Inverted Duplication and Allelic ... · indicated by dashes. Upper case letters indicate transcribed sequences of exon 1A as determined by cDNA cloning studies

Copyright 0 1996 by the Genetics Society of America

Opposite Orientations of an Inverted Duplication and Allelic Variation at the Mouse agouti Locus

Yanru Chen, David M. J. Duhl’ and Gregory S. Barsh

Departments of Pediatrics and Genetics, and the Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, California 94305-5428

Manuscript received March 19, 1996 Accepted for publication June 3, 1996

ABSTRACT The mouse agouti protein is a paracrine signaling molecule that causes yellow pigment synthesis. A

pale ventral coloration distinguishes the light-bellied agouti (AW) from the agouti ( A ) allele, and is caused by expression of ventral-specific mRNA isoforms with a unique 5’ untranslated exon. Molecular cloning demonstrates this ventral-specific exon lies within a 3.1-kb element that is duplicated in the opposite orientation 15-kb upstream to produce an interrupted palindrome and that similarity between the duplicated elements has been maintained by gene conversion. Orientation of the palindrome is reversed in A compared to AW, which suggests that mutation from one allele to the other is caused by intrachromo- soma1 homologous recombination mediated by sequences within the duplicated elements. Analysis of 15 inbred strains of laboratory and wild-derived mice with Southern hybridization probes and closely linked microsatellite markers suggests six haplotype groups: one typical for most strains that carry AW (129/SvJ, LP/J, CE/J, CAST/Ei), one typical for most strains that carry A (Balb/cJ, CBA/J, FVB/N, PERA/Rk, RBB/Dn); and four that are atypical (MOLC/Rk, MOLG/Dn, PERA/Ei, PERC/Ei, SPRET/Ei, RBA/Dn). Our results suggest a model for molecular evolution of the agouti locus in which homologous recombination can produce a reversible switch in allelic identity.

M UTATIONS that affect mouse coat color have provided powerful tools for the study of mammalian

genetics over the last century (SILVERS 1961; SEARLE 1968; JACKSON 1994). In addition to their utility as s u b strates for studying mutagenesis and recombination, alleles of some coat color loci have received special inter- est because their effects are confined to certain regions of the body.

One of the best examples of this type is agouti, which encodes a novel paracrine signaling molecule that causes hair follicle melanocytes to switch from the synthesis of eumelanin, a brown or black pigment, to pheomelanin, a yellow pigment (BULTMAN et al. 1992; MILLER et al. 1993) (reviewed in SILVERS 1979; SIRACUSA 1994). In mice that carry the A allele, transient expression of agouti RNA during the midportion of the hair growth cycle produces on individual hairs a subapical band of yellow pigment on an otherwise black or brown background that gives an overall brushed or golden appearance to the entire coat. By contrast, in mice that carry the light-bellied agouti (AW) allele, agouti RNA is expressed transiently in dorsal hairs but persists throughout the entire hair growth cycle in ventral hairs (VRIELING et al. 1994). As a result, animals that carry AWhave banded hairs on their dorsal surface but yellow hairs on their ventral surface.

Corresponding author: Greg Barsh, Beckman Center B271A, Stanford University School of Medicine, Stanford, CA 94305-5428. E-mail: [email protected]

94608. ’ Present address: Chiron Corporation, 4560 Horton, Emeryville, CA

Some insight into the underlying mechanisms responsible for agouti-related patterning has come with the realization that there are two groups of agouti RNA isoforms with different 5’ untranslated termini whose expression can be classified as either hair cycle-specific or ventral-specific (BULTMAN et al. 1994; VRIELING et al. 1994). In mice that carry the AW allele, both hair cycle- and ventral-specific isoforms are found, but mice that carry the A allele express only the hair cycle isoforms. Because the 5‘ untranslated termini for hair cycle- and ventral-specific isoforms are located approximately 100 kb apart, we have suggested previously that their expression is controlled by distinct regulatory regions and that mutation of AW to A is caused by a molecular lesion that affects only ventral-specific regulatory elements (VRIELING et al. 1994).

The A allele represents a loss-of-function with regard to expression of ventral-specific isoforms and therefore its evolutionary origin may be more recent than Aw. However, designation of AW us. A as “wild-type” is to some extent arbitrary since both the agouti and light- bellied agouti phenotypes are found in Mus musculus wild mice (M. m. domesticus, M. m. musculus, and M. m. castaneus) (SAGE 1981; BONHOMME and GUENET 1989; FESTING 1994), as well as many other mammals (SEARLE 1968). The AW allele is present at a relatively low frequency in most populations of wild mice, but often predominates in populations that reside in sandy or arid regions (reviewed by SAGE 1981). Thus, unlike agouti alleles such as lethal yellow (A?) and viable yellow

Genetics 144: 265277 (September, 1996)

266 Y . Chen, D. M. Duhl and G. S. Barsh

A I Pl-848(Ar

P1-047 (Aw) I1 * I / P1-25

P1-24 cosD2 I P1-75 IA) cosDl

P1-74 ( A ) cosD3 BamHl

specific specific

C

5 kb

Aw-5:h xhlBKI A C B K B Xh Xh B

cen E E E E E E tel ' 4- 0 , 1kb , 411 (-17 kb)" m

pslA 1A 1A' 2 1 c- cen

B A A- $h x:? cen E E E n +3-

Xh E

u 0 t9l

(-100 kb) 1A' 1A pslA 2

D Agouti genotype A Aw nn

Reversetranscriptase - + + -

-* I 0.1 kb , 128bp -

FIGURE 1 .-Genomic structure and expression of the agouti locus in the A" and A alleles. (A) Location and splicing pattern of alternative untranslated first exons used in ventral-specific and hair cycle-specific agouti isoforms, as taken from VRIELINC el al. (1994). Isoforms that contain exons 1A' and/or 1A are referred to as form I1 in RULTMAN pt dl. (1994). Untranslated and protein-coding sequences are indicated with open and closed boxes, respectively. Differences between the A and A'"alle1es are apparent by comparing the structure of clones P1-74 and PI-75, whose isolation is described in VRIELINC et al. (1994), to the clones PI-847 and P1-848, which were isolated using PCR primers from exon 1A as described in MATERIAIS ANI) METHODS. Sequences similar hut not identical to exon lA, referred to as pseudoexon 1A (pslA), are shown in B and C, which depict Apal (A), BamHl (B), ClaI (C), LcoRI (E), KpnI (K) , and XhoI (Xh) restriction maps from P1-847 (A") and PI-75 ( A ) . The relative positions of exon 1A and pslA were determined hy analysis of a 16kh Xhol fragment suhcloned from P1-847 and were confirmed by examining suhclones, from both PI-75 and P1-847, of RnmHI fragment5 of 7 and 11 kh, and of EcoRI fragments of 2.1 and 4.8 kh. Partial sequence analysis of these suhclones indicated a duplicated element of approximately 3.1 kh, shown as an interrupted grey arrow in l3 and C. The interruption represen& a unique sequence of approximately 0.5 kh that contains exon 1A' and originates from the 7-kh BarnHl fragment in P1-847 and the 11-kh BnmHI fragment in PI-75. Intrachromosomal homologous exchange hehveen distal portions of the duplicated element5 would lead to a 22-kh inversion ( i ) and provides a simple explanation for the difference between the A"'and A alleles. The 5-kb scale bar refers to the restriction maps for P1-847 (A") and P1-75 (A); the 1-kh scale har refers to the size of exons 1A and 1A' and pseudoexonlA. (D) Expression of untranslated first exons 1A and 1A'. Total RNA (5 pg) from the ventral skin of P5.5 A"/A"' or A/A mice was reverse transcribed with an antisense primer in exon 1A' and then was PCR-amplified using the same primer combined with a sense primer from exon 1A.

agouti Inverted Duplication 267

E 1 ggatccggaagggcagataactcagc-ttaaaatcagattctggagtgacagatagatcc 1A ( A W ) .................... t ..... c......................tg.t....... pslA ( A W ) ........ ..............................................

g ........................................ tg.t P P l A ( A ) g..... 1A ( A ) ........ .......

61 aaaaccagccctttgatcctctggtcctctttctactctgccatgggctcccagaaactt 1A (A''') ................. g ................. t........................ p ~ l A ( A W ) ................................... t ........................ 1A ( A ) ................. g ................. t........................ pslA ( A )

121 gagctagaacacacacaaaccaggcacagatctcccaaaaatgtgaacaagttcatggac 1A (Aw) .............---- ..... ............... cgt. 1A ( A )

aa g.ca................. palA ( A W ) .......................................... .............. .............---- ..... aa ............... g.ca................. palA ( A )

181 tttaccagcatgagacaaagccaccacacaacacagaggagcacttccaggtctggagac 1A (Awl ...... t ..................................................... palA (A') ............................................................ 1A ( A ) ...... t ..................................................... palA ( A )

241 acaagtggcaaagagcctccttgctaggtgcccacccagacacaagtgagcatgcttgct lA (Awl ............................. t ..................... t........ palA ( A W ) ............................................................ lA ( A ) ............................................................ pslA ( A )

301 tgacatttgctcccctcccccttagcaactttggtcatggctacagcatcctgacaacct lA (Aar) ...........................g...... c ......................... pslA ( A W ) ............................................................ 1A ( A ) ........................... g ................................ palA ( A )

361 atttAAACCMCATGCAGGAGCTGGCATCAAAGTACCATTTCCCACCAGTCTGAGTCC!l'T 1A (A") .................... A ....................................... pslA ( A W ) ............................................................ 1A ( A ) ......................................... T ......... C A . . . . . . . pslA ( A )

421 GAGCCTCTGCAGCCTCAGMGAGGGAGTCATCAGCTGAAACCTCCAGGAACCACCGGGGG 1A (Aw) ............................................................ pslA ( A W ) ............................................................ lA ( A ) ............................................................ palA ( A )

481 TCCCAGAAGgtatgtgctaaactccatccagatgttgtgtttcgttttgttctttttctt 1A (A") ............................................................ palA ( A W ) ............................................................ 1A ( A ) ............................................................ palA ( A )

541 cttttttcctctctaatcataatcaaagagtgaacaagtggttttcacattatcaagcaa 1A ( A W ) ............................................................ pSlA (Aw) ............................................................ 1A ( A ) ............................................................ pslA ( A )

601 cgcagggaagaaatacagtttccttctttcctgaccaggtctgtaccaaataaaggagca 1A ( A W ) ................................. g .......................... pslA ( A W )

............................................................ palA ( A )

................................. g .......................... 1A ( A )

661 agaaagagacttagggtttttgggtaatacagacagcacctgaattc ............................................... ............................................... ............................................... FIGURE 1.-Continued. The amplified products were transferred to a nylon membrane and the 128-bp cDNA product detected

by hybridization with a radiolabeled internal oligonucleotide as described in MATERIALS AND METHODS. Control lanes did not receive reverse transcriptase. (E) DNA sequence flanking exon 1A and pslA in the AW and A alleles, as determined from analysis of the 7- and 11-kb BamHI fragments shown in B and C. Identity with the 1A (AW) sequence is indicated by dots; deletions are indicated by dashes. Upper case letters indicate transcribed sequences of exon 1A as determined by cDNA cloning studies (VRIELING et al. 1994) and primer extension (unpublished data).

(A") that are not found in wild mice or other rodents (DUHL et al. 1994), A and AW are likely to serve an adaptive function.

Like wild mice, A is generally more common than AW in inbred strains of laboratory mice, but mutations of one allele to the other have not been described. Under- standing the molecular difference between the AW and A alleles might provide insight into the evolution of inbred strains as well as the basis of a polymorphic trait present in other mammals. Here we report that the 5' flanking region of ventral-specific agouti isoforms contains an inverted duplication and that orientation of this duplication is reversed in mice that carry the AW and A alleles. Analysis of this structure and nearby mi-

crosatellite markers among inbred strains of laboratory and wild-derived mice suggests a model for molecular evolution of the agouti locus in which homologous recombination can produce a reversible switch in allelic identity.

MATERIALS AND METHODS

Mouse strains and mutations: The CAST/Ei and SPRET/ Ei strains, and all laboratory mouse strains were obtained from the Jackson Laboratory (Bar Harbor, ME) except for FvB/ N, which was obtained from Taconic (Germantown, NY). Ge- nomic DNA was prepared from spleen by standard procedures techniques (SAMBROOK et al. 1989). Assignment of agouti genotype as Awns. A is based on the presence or absence, respec-

268 Y. Chen, D. M. Duhl and G. S. Barsh

tively, of a ventral pelage that is entirely yellow or cream- colored, and therefore exhibits a clear line of demarcation from banded, 2.e. agouti hairs, in the dorsal skin. (The Balb/ cJ and FVB/N strains carry the c mutation and therefore are albino; their agouti genotype becomes apparent when crossed to a strain such as C57BL/6J-ala; C/C.)

For wild-derived strains other than CAST/Ei and SPRET/ Ei, genomic DNA was purchased from the Mouse DNA Re- source at The Jackson Laboratory, and information regarding agouti genotype was kindly provided by the respective colony managers. These assignments and our findings are generally in accord with published information with two exceptions. The MOLC/Rk strain has been described as carrying the A allele but actually carries A“, similar to the MOI,G/Dn strain (FESTING 1994). The two strains are identical with regard to the structure surrounding exon 1A and microsatellite genotypes (see RESULTS).

The PERA/Ei and PERA/Rk strains are both derived from the same isolate of wild mice and have been inbred separately since 1971 (BONHOMME and GUENET 1989). Both strains carry the A allele, but their molecular structures are completely different (see KESUI.TS). This may represent residual heteroge- neity in the colony at the time the strains were separated, especially since the structure of PERA/Ei is similar (but not identical) to a different isolate of Peruvian wild-derived mice, PERC/Ei.

Cloning and analysis of genomic DNA: The P1 clones P1- 74, P1-75, P1-847, and P1-848, were obtained by screening a commercial bacteriophage P1 library (Genome Systems, St. Louis, MO) with primer pairs from agouti exon lA, 5’-AGT- CTGAGTCCTTGAGCCTG3‘ and 5“TGGGACCCCCGGTGG TTC-3’, that amplify a 78-base pair (bp) fragment. P1-847 and P1-848 were isolated from a library constructed from 129 mice. P1-74 and P1-75 have been described previously (VRIELING et al. 1994) and were isolated from a library constructed from a mouse fibroblast cell line, C127, derived from the RIII mouse strain, which carries the A allele. Single copy probes a, c , and d are derived from genomic DNA and are described in the legend to Figure 3. Southern hybridizations and DNA sequence analysis using modified T7 polymerase and dideoxy chain ter- mination were performed according to standard procedures (SAMBROOK et al. 1989) using radiolabeled nucleotides and hybridization in the presence of 10% dextran sulfate. DNA sequence alignments were obtained using the DNASTAR (Madison, WI) Align program or the GCG (Madison, WI) Gap program.

Expression analysis of exon 1A: Ventral skin RNA from P5.5 129/SvJ-AMyAW or FVB/N-A/A mice was reverse transcribed with the primer 5’-CTCTTACAGTACAGGGCATG-3’, which lies at the 3’ end of exon lA’, then PCR-amplified using the same primer in combination with a primer, 5”AGTCTGAGT- CC’ITGAGCCTG3’, that lies in the center of exon 1A. After Southern blotting, the 128-bp PCR product was detected by hybridization to a radiolabeled primer internal to those used for amplification, 5’-TGGGACCCCCGGTGGTTC-3’.

Microsatellite markers and genotypes: Oligonucleotide primer pairs for D2MIT27, D2MIT48, D2MIT55, D2MIT225, D2MIT286, D2MIT347, D2MIT409, and D2MIT496 were obtained from Research Genetics (Huntsville, AL) . Allele sizes were determined as described in DIETIUCH et al. (1992).

RESULTS

Regions that contain the ventral-specific first exons are part of an inverted duplication: Previous studies from our laboratory demonstrated that agouti mRNA isoforms expressed in postnatal skin contain one of

three untranslated first exons, lA, lB, or 1C (Figure 1) (VRIELINC et al. 1994). Some of the isoforms beginning with exon 1A also contain an alternatively spliced 4 6 nt untranslated exon, lA’, between exon 1A and exon 2. The isoforms beginning with exon 1A are expressed in the ventrum but not in the dorsum of Aw/AW mice, and are not expressed in A/A mice. By contrast, the isoforms beginning with exons 1B or 1C are expressed in both dorsum and ventrum but only in the midportion of the hair growth cycle in A/A and AW/AW mice (BULTMAN et al. 1994; VRIELINC et al. 1994).

Our initial characterization of the region surrounding exons 1A and 1A’ was based on two bacteriophage P1 clones, P1-74 and P1-75, that originated from the RIII strain that carries the A allele (VRIEIJNG et al. 1994). When probes that contained exon 1A were hybridized to Southern blots containing genomic DNA from other mouse strains (see below), we observed a pattern of variant restriction fragments that suggested the presence of (1 ) a duplication that included exon lA, but not other portions of the agouti gene and (2) a genomic rearrangement that involved the duplicated sequences and differed in structure between the A“and A alleles. To characterize the duplication and rearrangement in more detail, we obtained bacteriophage P1 clones that originated from the 129/Sv strain that carries the A“ allele, and compared the structure of these clones, P1-847 and P1-848, to the A-derived P1 clones (Figure 1).

Restriction map analysis, subcloning, and Southern hybridization experiments using oligonucleotide probes and cloned DNA revealed that exon 1A was duplicated on 11- and 7-kb BamHI fragments. To help distinguish between the duplicated elements, BamHI and EcolU restriction fragments that contained exon 1A sequences were subcloned from both the A’: and the A-derived P1 clones, and an approximately 700-bp fragment that contained exon 1A was sequenced from each of the fragments (Figure 1E). Pairwise comparisons revealed that the duplication included the entire 700-bp region, but in no case were two regions completely identical, which allowed us to distinguish between different alleles and also to distinguish between duplicated elements from a single allele (Figure 1E). In general, sequence mismatches in the 700-bp region are greater between duplicated elements than between alleles. In addition, sequence mismatches are more frequent in the 5’ flanking region of exon 1A than in the 3‘ region (Figure 1E).

The sequence information shown in Figure 1E was used to determine the location of each region on the restriction map of the P1 clones. For both the AW- and the A-derived clones, exonlA sequences in the 11-kb BamHI fragment lay approximately 15 kb upstream of their counterparts in the 7-kb BamHI fragment and, surprisingly, were oriented in the opposite direction, antiparallel to the rest of the agouti gene as one half of an interrupted palindrome. We describe the region

n,p7c/i Inverted Duplication

A "upstream region"; 471 8 bp "downstream region"; 4195 bp I I I 1

cen "J,L" WFFW+- tel psi A

129 (Aw)

, + t 6 /I-, cen

. . . / / ?

1 kb , . ~ , . ,. . ,

, I

- . . (100 kb) " tel

2

ggagttcagaagcccaccag 1-5 kb ggaaacaaagactattgagc 0.85 kb

tgtgcctcagagcccaccag I I I I I I I I I I I I IIIIII1111l I I I I I I I I I I I ( 1.5 kb )ggaaacaatgtgggtttttt( 0.5 kb

B psl A or

or n

I

I I I

(kilobase pairs) 1 2 3

FIGURE 2.-Sequence similarity hehveen duplicated elements that contain exon IA. (A) A 471Khp "upstream" region ((;en- Bank accession no. 1,76475) and a 4195-hp "downstream" region (GenBank accession no. L76476) that contain approximatelv 3.1 kh of duplicated DNA were sequenced from P1-847 ( A " ) . The tipstream region corresponds t o the 1 I-kh I<mcr,nHI fr:tgmc*nt and 2.1-kh EcoRI fragment shown in Figure 1B and contains 252 antl 526 hp of unique sequence from the lefthantl antl rightlland sides, respectively, ol'the duplication. The downstream region corrcsponds to the 7-kh RnmHl fragment antl 4.8-kh I?oKI fragmt-nt shown in Figure 1B and contains 233 and 381 hp of unique sequence from the lefthand and righthand sides, respcctivcly, of the duplication. The upstream and downstream duplicated elements are each interrupted approximately midway hy unique regions of 0.85 o r 0.5 kh, respectively. (B) Alignment of the duplicated regions according to the GCG Gap prognm reveals that mismatches are clustered at the.junctions of duplicated DNA with uniquc DNA.

contained on the Al'derived 11-kb RnmHI and the A- derived 7-kb fragment as a pseudoexon (pslA) because it lies in a transcriptional orientation opposite that of agouti coding sequences in the A"' allele (Figure 1, B and C), and because it does not give rise to a stable mRNA in the A allele (VRIEI-IN<; ef nl. 1994).

This structure was characterized in further detail by sequencing an approximately 4.5-kb region that contained exon 1A and a corresponding region that contained pslA from Al'derived RnmHI and I+oRI sub- clones. Nearly 3.1 kb of DNA was found to be duplicated but the duplicated elements were each interrupted by a unique sequence of 0.85 or 0.5 kb in the 11- or 7-kb RnmHI fragment, respectively (Figure 2). The overall level of similarity between the duplicated elements was 97% but mismatches between the two elements were most frequent at the junctions of duplicated DNA with

unique DNA (Figures 1E and 2B). Exon 1A' was located within unique sequence that interrupted one of the two duplicated elements and lay on the 7-kb HnmHI in A''' but on the 11-kb RnmHI fragment in A (Figures 1 and 2). Thus, the A"'and the A allele both contain exon 1A sequences in similar locations, but exon 1A' lies approximately 20 kb upstream in A compared to A"'and its orientation is reversed (Figure 1C). Taken together, these observations are explained most simply by postu- lating that the A"' and A alleles differ by virtue of an inversion mediated by homologous exchange between duplicated segments 3.1 kb in length that lie approximately 22 kb apart (Figure 1, B and C).

Expression of exon 1A in mice carrying the A allele: We have shown previously that stable RNA isoforms containing exon 1A are not detectable by North- ern hybridization in A/A mice (VRIEI-INC et nl. 1994).


The structure depicted in Figure 1C helps to explain this observation because, in the A allele, the transcriptional orientation of exon 1A is pointing away from the rest of the agouti gene. To investigate whether the exon 1A promoter was still functional in the A allele, albeit in a nonproductive orientation, we used oligonucleotide primers within exons 1A and 1A’ to examine by RT- PCR whether an RNA that contained exon 1A could be detected in A/A mice. Ventral skin RNA from P5.5 AW/ AW and A/A mice was reverse transcribed with an antisense primer from exon 1A‘ and then with PCR-amplified using the same primer in combination with a sense primer from exon 1A (Figure 1D). An internal oligonucleotide primer from exon 1A detected the 128-bp cDNA product by Southern hybridization in both the AW/AW and A/A samples, but not in control samples that had been treated identically except for the addition of reverse transcriptase (Figure 1D). These results indicate that exon 1A is still functional in A/A mice and suggest that the inversion likely to have caused the altered structure in the A allele is also responsible for the absence of ventral-specific agouti RNA isoforms.

Molecular analysis of the inverted duplication in genomic DNA: To confirm that the inverted duplication was present in genomic DNA and to investigate its structure in other mouse strains, we hybridized probes from the duplicated regions to genomic DNA that had been digested with BamHI from three strains of laboratory mice that carried the A allele (1 29/SvJ, CE/J, and LP/ J) and three strains of laboratory mice that carried the A allele (Balb/cJ, CBA/J, and FVB/J) . We also examined genomic DNA from four strains of wildderived mice that carried the AW allele (CAST/Ei, MOLC/Rk, MOLG/Dn, and SPRET/Ei) and from five strains of wild-derived mice that carried the A allele (PERA/Ei, PERA/Rk, PERC/Ei, RBA/Dn, and RBB/Dn).

A BamHI-PstI 0.45-kb fragment that contains exon lA, designated probe a (Figure 3), detected two BamHI fragments of 11 and 6-7 kb in genomic DNA from most of the strains with the following exceptions. The MOLC/Rk and MOLG/Dn strains carried a 0.6-kb fragment instead of a 7-kb fragment; the PERA/Ei, PERC/ Ei, and CE/J strains carried a 6-kb fragment instead of a 7-kb fragment, SPRET/Ei carried 15- and 4kb fragments instead of 11- and 7-kb fragments; and RBA/Dn apparently carried three fragments of 11, 7, and 6 kb. Thus, with the exception of RBA/Dn, all strains of mice we examined have a duplication of sequences that contain exon 1A.

To determine the relative position of the duplicated sequences, we generated probes specific for the 7- or the 11-kb BamHI fragments. A 0.4kb XbaI fragment designated probe c corresponds to the unique region of the Aw-derived 11-kb BamHI fragment, and a 0.5-kb XbaI fragment designated probe d corresponds to the unique region of the Aw-derived 7-kb fragment (Figure 3B). As predicted by the map of cloned DNA depicted

in Figure 1, probe c detected an 11-kb BamHI fragment in laboratory mice (129/SvJ, CE/J, and LP/J) that carry the AW allele and a 7-kb BamHI fragment in laboratory mice (Balb/cJ, CBA/J, and FVB/J) that carry the A allele (Figure 3 and data not shown). A reciprocal pattern was observed for probe d, although a 6kb instead of a 7-kb BamHI fragment was detected in the CE/J strain (Figures 3 and 4).

When this analysis was extended to wildderived strains, several exceptions were noted which, together with the results described above, can be summarized in terms of six haplotypes, designated groups I-VI in Fig- ure 4. Groups I and I1 comprise the typical orientation for laboratory strains that carry the AW and A alleles, respectively. Group I also includes the wild-derived strain CAST/Ei, while group I1 also includes the wildderived strains PERA/Rk and RBB/Dn.

Group 111 consists of the MOLC/Rk and the MOLG/ Dn strains, which carry AW yet show a Southern blot pattern most similar to the A-carrying group I1 strains. Probes c and d detected 7- and 11-kb BamHI fragments, respectively, in group I1 strains, but detected 3.5- and 1 1-kb BamHI fragments, respectively, in the MOLC/Rk and the MOLG/Dn strains. Group IV consists of the PERA/Ei and PERC/Ei strains, which carry A yet show a Southern blot pattern similar to the Aw-carrying group I strains. Probes c and d detected 11- and 6-7-kb BamHI fragments, respectively, in group I strains, but detected 11- and 6-kb BamHI fragments, respectively, in the PERA/Ei and PERC/Ei strains. This suggests that groups I11 and IV exhibit an “atypical” position of exon 1A’ compared to strains in groups I and 11, respectively.

Group V consists of the Aw-carrying SPRET/Ei strain, in which probes c and d detected 4 and 15-kb fragments, respectively. These same fragments were detected by probe a, which suggests that the overall structure in SPRET/Ei is similar to most other strains. Finally, Group VI consists of the RBA/Dn strain, in which probe c detects two fragments (11 and 7 kb) and probe d detects two fragments (1 1 and 6 kb). Although probe a detects fragments of three different sizes in the RBA/Dn strain (1 1, 7, and 6 kb), this pattern is most easily explained by a combination of four fragments, two of which are 11 kb (see below).

In addition to the strongly hybridizing BamHI fragments of 11- and 6-7 kb detected by probes a, c, and d in the group I and group I1 strains, weakly hybridizing fragments of 7.5, 3, and 2.5 kb were detected in the CE/J, PERA/Ei, RBA/Dn, and LP/J strains (Figure 3) . The 7.5- and 3-kb fragments were detected by probe c, the 2.5-kb fragment was detected by probe d, and all three fragments were detected by probe a. Because these four strains are from three different haplotype groups as defined in Figure 4, the weakly hybridizing fragments may represent a related DNA sequence at another chromosomal location.

Genotype of microsatellite markers in the region of

kb 23 -

9.9 - 6.7 -

4.4

2.3 - 2.1 -

0.5 -

c

L +

probe a probe c probe d

psl A 1A 1A - l 1kb I

FIGURE 3.-Analysis of the inverted duplication containing exon 1A i n genomic DNA. (A) Genomic DNA from the indicated mouse strains was digested with RnmHI and analyzed by Southern hybridization with three different probes. Probe a is a 0.4-kb RnmHI-Psfl fragment within the duplicated region as shown in Figures 1 and 2; probes c and d are 0.4- and 0.5-kb wniqne M n l fragments, respectively, that interrupt the duplicated regions. The same filter \vas used for all three hybridizations; removal of residual probe was checked by autoradiography. Probe d as obtained from the ;\"&rived i-kb fragment is nearly identical to a probe of similar length obtained from the ;-derived 1 I-kh fragment (Figure 1 ) but the A"derivcd probe produces a slightly higher background with genomic DNA than the Aderived probe, and therefore the latter probe was used for the autoradiogram depicted here. Results obtained for genomic DNA of the strains 129/Svj, Balb/cJ, and FVB/N are not shown but arc summarized in Figure 4. (R) Position of probes a, c, and d within cloned DNA from PI-847 ( A " ) as shown in Figure 1. The interrupted grey arrows represent the drrplicated elements which are approximately 3.1 kh in length as determinrtl by DNA sequence analysis (Figure 2 ) .

agouti: Nearly all of the differences between laboratory strains that carry the A"'and A alleles (groups I and 11, respectively, Figure 4) can be accounted for by a single intrachromosomal crossover that reversed the position of duplicated regions containing exon lA, as depicted in Figure 1. However, four of the wildderived strains we examined (groups 111 and IV) appear to have an "atypical" orientation of the inverted duplication. In addition, the Southern blot pattern observed for RBA/ Dn (group VI) can only be explained by rearrange- ments that occured independentlv of the one that distinguishes groups I and 11. To gain further insight into molecular evolution of this region, we determined the genotype of closely linked microsatellite loci for the 15 strains described above.

Eight microsatellite loci were chosen for study because they had been positioned previously within two recombination units of agou/i (SIRACUSA and ARROTT 1994). Primer pairs for two, D2MIT27 and D2MIT48, did not ampli5 any DNA from the majority of strains, bu t amplification reactions for the remainder were suc- cessful in all but two combinations (Table 1). Each locus was found to have five to nine alleles, and no two loci produced the same strain distribution pattern.

In general, the distribution of haplotype groups o h tained by painvise comparison of microsatellite genotypes (Table 1) resembles that obtained bv Southern blot analysis with probes a, c, and d (Figure 4) . For example, among the A-carrying strains that exhibit a "typical" pattern on Southern analvsis (group I1 in Fig-


Group I 129/SvJ 7

CAST/Ei J

11 BalblcJ

FVBlN PERA/Rk R B BlDn

MoLCIRk MOLGlDn I AW

IV PERAIEi PERCEi I A

C a a d

D pslA

I-, ll kb - I- 7kb 1 B B

? ;; ' ;; "0- d a a C

0 ps lA

F~CURE 4.-Orientation of the inverted duplication containing exon 1A in laboratory and wildderived strains of mice. The BamHI (B) Southern blot and inversion results for the mouse strains shown in Fig- ure 3 and the 129/SvJ, Balb/cJ, and FVB/ N strains can be organized into six h a p b type groups. The interrupted gray arrows represent 3.1 kb of duplicated DNA as shown in Figure 2; gray, hatched, and white squares represent probes a, c, and d, respectively, and are not drawn to scale. For groups I-IV, the inverted duplication is oriented with the centromere on the left and the telomere and agouti coding sequences on the right; these orientations are based on the assumption that the results obtained by analysis of cloned DNA as shown in Figure 1 can be applied to the patterns obtained by Southern blot analysis of genomic DNA as shown in Figure 3. Groups I and I1 represent the "typical" orientation observed for most strains that carry A'" or A, respectively. Groups 111 and IV represent an "atypical" orientation; group IV strains carry the A allele but exhibit a Southern blot pattern similar to the A"&nying group I strains, whereas group 111 strains carry the A"' allele but exhibit a Southern blot pattern similar to the A- carrying group I1 strains. An additional difference between the group I11 and group I1 strains is the presence of one or more variant BumHI sites in the "downstream" duplicated element of the group 111 strains such that probe a detects a O.6kb fragment and probe c detects a 3.5kb fragment. The Southern blot patterns for group V and group VI strains are completely different from other strains and therefore no con- clusions can be drawn regarding orientation of the duplicated elements.

ure 4), Balb/cJ, PERA/Rk, and RBB/Dn are identical at all six loci, Balb/cJ and FVB/N are identical at five of six loci, and the most dissimilar member of the group, CBA/J, is still identical to the other four strains at three of six loci. By contrast, the SPRET/Ei and RBA/ Dn strains each comprise their own groups V and VI, respectively, according to Southern blot analysis, and exhibit microsatellite haplotypes that are unique or nearly unique. Similarly, MOLC/Rk and MOLG/Dn which comprise group I11 by Southern analysis, are identical at all six microsatellite loci and quite different from all other strains.

In the AWcarrying strains that exhibit a "typical" Southern blot pattern (group I in Figure 4), 129/SvJ and LP/J are identical at four of six loci, but CE/J and CAST/Ei have unique or nearly unique microsatellite haplotypes, which suggests that the A"' alleles of these latter two strains are more distantly related. Finally, the PERA/Ei and PERC/Ei strains, which carry the A allele and show an "atypical" Southern blot pattern (group V in Figure 4), are identical at three of six loci. There-

fore, the A alleles of these two strains may not be closely related although they are likely to have a common origin independent of the A allele for strains in group 11.

DISCUSSION

Agouti and light-bellied agouti are easily recognized as common coat color patterns in many mammalian species (SEARLE 1968). In mice, the difference between these phenotypes is caused by the presence or absence in AWor A, respectively, of a unique set of RNA isoforms that contain a 5' untranslated exon located approximately 100 kb 5' of protein coding sequences (VRIELING e.! al. 1994). As shown here, this exon lies within a 3.1- kb region that is duplicated in the opposite orientation to produce an interrupted palindrome approximately 22 kb in length. We suggest that inversion of this 22-kb segment mediated by intrachromosomal crossing over is responsible in most cases for the difference between strains that carry A " and those that carry A. Our findings suggest a model for molecular evolution of the


TABLE 1

Genotype of microsatellite markers closely linked to agouti

Haplotype group D2 D2 D2 D2 D2 D2 according to Agouti MIT MIT MIT MIT MIT MIT

Figures 3 and 4 allele Strain 55 225 286 347 409 496

I AW 129/SyJ 172“ 0 141 86 98 132 LP/J 172 144 141 0 98 132 CE/J 170 144 137 110 100 116 CAST/Ei 164 152 111 84 112 130

I1 A Balh/cJ 172 142 145 138 92 118 CBWJ 166 142 143 138 92 116 FVB/N 172 142 129 138 92 118 PERA/Rk 172 142 145 138 92 118 RBB/Dn 172 142 I45 138 92 118

111 A ‘v MOLC/Rk 184 144 103 86 96 164 MOLG/Dn 184 144 103 86 96 164

IV A PERA/Ei 174 142 109 120 96 120 PERC/Ei 174 140 141 120 110 120

V SPRET/Ei 176 140 129 116 120 120 VI A RBA/Dn 182 146 165 110 132 128

A 12

“Size of the major PCR-amplified product. Allele sizes listed for Balb/cJ, CAST/Ei, and SPRET/Ei match those previously determined, with the following exceptions in which the results of the MIT Genome Center (DIETRICH et al. 1992) are listed in parentheses: Balb/cJ and D2MIT347 (no product), CAST/Ei and D2MIT286 (123), CAST/Ei and D2MIT347 (no product), CAST/Ei and D2MIT496 (122), SPRET/Ei and D2MIT55 (152), and SPRET/Ei and D2MIT347 (no product).

agouti locus, confirm earlier predictions regarding the molecular nature of agouti alleles, and have implications for genetic control of ventral pigmentation patterns in other mammalian species.

Molecular evolution of the agouti locus in laboratory mice: Because two or more copies of exon 1A are present in all of the strains we examined, the duplication that gave rise to the structure described here must have occurred prior to the time that M. spretus and other Mus strains diverged approximately 4 million years ago (reviewed by SILVER 1995). The duplicated elements exhibit 97% similarity over a 3.1-kb region, a level that is comparable to the similarity between other noncoding regions in M. spretus and laboratory mice (RIKKE et al. 1991; NACHMAN and AQIJADRO 1994). However, direct comparison of the duplicated elements revealed that mismatches are not distributed randomly but instead are clustered at the ends of each element. This suggests that sequence similarity between the elements has been maintained by intrachromosomal crossing over and/or gene conversion (SHEN et al. 1981; MAEDA and SMITHIES 1986) and implies that the level of similarity between the two elements is likely to underestimate the time at which the duplication occurred.

Each element is interrupted by a unique central fragment 0.5 or 0.85 kb in length that gves rise to two “homology blocks,” labeled X and Y in Figure 5. This type of structure could have arisen from an insertion into one of the duplicated elements, but, if so, must have been accompanied by a deletion of DNA surrounding the insertion site since both elements are interrupted at the same location and neither of the

unique fragments is similar to other portions of the duplication. Because Southern hybridization probes derived from the unique fragments detect specific sequences in all of the strains we examined, a putative simultaneous insertion and deletion must have occurred, like the duplication itself, prior to the time that M. spretus diverged from the progenitors of laboratory mice.

Taken together, these considerations suggest a model (Figure 5) for molecular evolution of this region in which a 3.1-kb fragment containing exon 1A and exon 1A’ was duplicated (Figure 5A) prior to the evolution of MUS species (Figure 5A). Similarity between the duplicated elements was then maintained by homologous exchange and/or gene conversion (Figure 5B), allowing sequence differences to accumulate only at the junctions between unique DNA and duplicated DNA (indicated by X X in Figure 5C). Some time later, insertion of unique DNA into the “upstream” duplicated element was accompanied by deletion of flanking DNA at the insertion site to produce the homology blocks X and Y (Figure 5, C and D), to produce the structure observed in most laboratory mice that carry the AW allele (group I).

The insertion responsible for homology blocks X and Y (Figure 5, C and D) is likely to have decreased the rate of sequence transfer between the duplicated elements, but intrachromosomal crossing over within homology block Y (Figure 5E) occurred at least once during this period to produce the structure observed in most strains that carry the A allele (group 11).

This model does not make explicit predictions re-

Y . Chen, D. M. Duhl and G. S. Barsh 274

A

B

Y X

D d ~ ” . * r J X X ” - - Q . r c a

Group I ( A T pslA

x Y

E

FIGURE 5.-A model for molecular evolution of the 5’ flanking region of ventralspecific agouti isoforms in laboratory strains. This model is based on the results depicted in Figure 4 and Table 1 and is described further in the text. (A) A progen- itor of a ventralspecific q u t i gene with two noncoding exons 1A and 1A‘ (gray and white squares, respectively) and a hTc- thetical positiondependent regulatory element (indicated by the gray circle). Duplication of a 3.1-kb region that contained exons 1A and 1A‘ followed by upstream insertion of the duplicated material would give rise to an interrupted palindrome where the upstream component of the palindrome contained two pseudoexons (pslA’ and pslA). (B) Gene conversion and/ or homologous exchange between the duplicated elements would maintain sequence similarity except at the junctions between duplicated DNA and unique DNA. (C) Simultaneous insertion and deletion of DNA into the midportion of the u p stream duplicated element would account for the absence of pslA’ and flanking sequences as well as the presence of unique DNA at the same site, and would give rise to two “homology blocks” labeled X and Y. (D) In an ancestral A”’ allele similar to those in group I (Figure 4), sequence differences between the upstream and downstream duplicated elements (indicated by XX) would accumulate at the ends of each element. Base- pair changes within the promoter that inhibited transcriptional actimtion would be selected against in the downstream duplicated element, but the same selection would not apply to the corresponding region of the upstream element. (E) Crossing over within homology block Y would reverse the orientation of the duplicated element to give rise to an ancestral A allele similar to those in group I1 (Figure 4). Failure to express ventral- specific agouti isoforms in the group I1 alleles could be caused either by the presence of a nonfunctional promoter (indicated by XX) in the downstream duplicated element, referred to in the text as “promoter switching,” or by displacement of a positiondependent cis-acting regulatory sequence (indicated by the circle), referred to in the text as “enhancer displacement.”

garding the origin of the “atypical” structures observed in the MOLC/Rk and MOLG/Dn strains (group HI), in the PERA/Ei and PERC/Ei strains (group IV), or in the RRA/Dn strain (group VI) although several possi- bilities are discussed further below. Nonetheless, the structures in these groups are likely to have arisen after the events depicted in Figure 5D, because Southern hybridization probes c and d (Figures 3 and 4) that distinguish between the duplicated elements in strains from groups I and I1 also distinguish between the duplicated fragments in strains from groups 111, IV, and VI.

In the CE/J, LP/J, PERA/Ei, and RBA/Dn strains, Southern hybridization to genomic DNA revealed the presence of weakly hybridizing fragments in addition to those predicted by analysis of cloned DNA. These strains are from three different haplotype groups as defined by analysis of the strongly hybridizing fragments and closely linked microsatellite markers, which suggests that the weakly hybridizing fragments lie in another chromosomal location.

The cause or causes of the AW to A mutation: An important implication of the work described here is that intrachromosomal crossing over caused not only an inversion that distinguishes most A”karrying strains (group I ) from most A-carrying strains (group 11), but also caused a change in phenotype from light-bellied agouti to agouti. The model depicted in Figure 5 suggests two possible molecular explanations for the absence of ventral-specific agouti isoforms in the group I1 strains.

One explanation, referred to as “promoter switching,” is based on the hypothesis that the 5‘ flanking region of exon 1A in the A”’allele contains information required for transcriptional initiation of ventral-specific ngouli isoforms and therefore has been constrained by selection. By contrast, the corresponding region in the “upstream” duplicated element does not give rise to a transcript and therefore will have accumulated mutations since the duplication occured. Furthermore, reso- lution of sequence differences between the duplicated elements as a consequence of gene conversion is likely to have skipped over the 5’ flanking regions of exon 1A and pseudoexon 1A because they lie at an end rather than the middle of the duplication (Figures 2B and 5C). Thus, intrachromosomal exchange that converted a group I structure to a group I1 structure (Figure 5E) may have replaced a functional with a nonfunctional promoter.

An alternative explanation, referred to as “enhancer displacement,” posits that position-dependent cis-acting regulatory sequences (depicted as a circle in Figure 5) lie in the 15 kb of DNA that separates the duplicated elements. The distance between this putative enhancer and the transcriptional initiation site for exon 1A will have been altered by the inversion that distinguishes strains in group I1 from those in group I and therefore


may contribute to the absence of ventral-specific agouti isoforms in most strains that carry the A allele.

These explanations are not mutually exclusive and each is consistent with the observation that transcrip- tion between exon 1A and 1A' is detected by RT-PCR from the "upstream" duplicated element in a group I1 strain (Figure 1E). However, the enhancer displacement hypothesis implies that regulatory elements lie between the duplicated regions and could be investi- gated by determining whether or not DNA fragments from the inverted DNA can recapitulate normal expression of agouti when fused to a minimal promoter and reporter gene in transgenic mice.

Microsatellite allele sizes and molecular evolution of the agouti locus in wildderived mice: Among nine wild- derivcd strains we studied, three, CAST/Ei, PERA/Rk, and RBB/Dn, exhibited a structure surrounding exon 1A similar to laboratory strains that carried A"' (group I ) or A (group 11). However, the six remaining strains, MOLC/Rk, MOLC/Dn, PERA/Ei, PERC/Ei, SPRET/ Ei, and RRB/Dn, exhibited unusual or atypical structures that could be subdivided into four different groups. Comparison of allele sizes for six microsatellite markers located within 1-2 cM of the agoutilocus (SIRA- CUSA and ABROTT 1994) reinforced this conclusion. With one exception (D2MIT286 is proximal to the other five markers), the microsatellite loci that we studied have not been ordered with respect to each other, and so it is impossible to say whether the different allele sizes arose by mutation or recombination. Nonetheless, the patterns of allele sizes still provide useful information. For example, of two closely related strains derived from Peruvian mice, PERA/Ei and PERA/Rk, the pattern of microsatellite allele sizes in PERA/Rk mice is most similar to laboratory mice that carry the A allele (group 11), while the pattern of microsatellite allele sizes in PERA/Ei mice is most closely related to PERC/ Ei mice (group IV), which also originate in Peru but at a very different time and location than the PERA strains. For the two strains that represent carriers of Robertson- ian translocations originating in different regions of Switzerland, the RBB/Dn mice exhibit a pattern of microsatellite allele sizes similar to laboratory strains that carry the A allele (group 11), while the RBA/Dn mice exhibit a unique pattern.

With regard to the inverted palindrome that contains exon lA, the structure observed in the MOLC/Rk and MOLG/Dn strains (group 111) indicates that intrachromosomal crossing over is not sufficient to cause a mutation from A'" to A since MOLC/Rk and MOLG/Dn carry A"' yet exhibit a Southern blot pattern similar (but not identical) to most A-carrying strains (group 11). Successive intrachromosomal crossover events from an ancestral A"'chromosome, one in homology block X and one in Y, could explain the configuration observed for MOLC/Rk and MOLG/Dn, and is consistent with the observation that these strains exhibit a unique

6 Y X X Y

--.Xxiic"-. *-. oxx,-@m" Group I (A@') ~ S I A

c a 1A I A I

X Y

G I xxrJ-*"k?zb-

0 u. -. -=*-

- $.F"-"-.xnR.c)-~. d

c a a d Group IV(A) 1A' 1 A p s l A

FIGURE 6.-A model for molecular evolution of the 5' flanking region of ventral-specific agouti isoforms in some wildderived strains. (A) An ancestor of the A allele found in laboratory strains (group 11) could have produced the atypical A" allele found in the MOLC/Rk and MOLG/Dn strains (group 111) by crossing over within homology block. (B) Cross- ing over within homolog block X of an ancestral A" allele would give rise to a nonfunctional gene ( A ) similar to those in group IV. Additional mechanisms that could have given rise to group I11 and group IV alleles are described in the text. Symbols are the same as for Figures 4 and 5.

microsatellite genotype (Figure 6 ) . Alternatively, if promoter switching is responsible for mutation of A" to A in group I and group I1 strains, the MOLC/Rk and MOLG/Dn chromosome may have undergone intrachromosomal exchange within homology blockY at a time when the promoter in the "upstream" duplicated element was still functional.

Because the PERA/Ei and PERC/Ei strains (group IV) carry the A allele but exhibit a Southern blot pattern similar to most strains that carry A''' (group I ) , origin of the A allele in PERA/Ei and PERC/Ei is clearly distinct from most strains that carry A (group 11). The two explanations discussed above as potential causes of the A mutation in group I1 strains, promoter switching and enhancer displacement, could explain the A mutation in group IV strains if the PERA/Ei and/or PERC/Ei structures were produced from an ancestral A''' allele by intrachromosomal crossing over within homology block X (Figure 6). The PERA/Ei and/or PERC/Ei agouti alleles also could have been produced as descen- dant.. of an A" allele by a simple point mutation in


regulatory sequences required for normal expression of ventral-specific agouti isoforms. If so, the ancestral A" allele for PERA/Ei and/or PERC/Ei is not repre- sented in group I or group 111, since the microsatellite genotypes of PERA/Ei and of PERC/Ei are very different from other strains we examined.

In the AW-carrying SPRET/Ei strain (group V), the Southern blot pattern is similar to most other strains we examined, but 4 and 15-kb BamHI fragments are detected instead of the 11- and 6-7-kb fragments typical for most A"-carrying strains (Figure 6). Finally, the pattern observed in the RBA/Dn strain (group V I ) is a combination of the patterns seen individually for groups I and I1 and therefore is most easily explained by a secondary duplication to produce a total of four fragments (11, 11, 7, and 6 kb).

Homologous recombination as a genetic switch for ventral pigmentation? Because ventral-specific agouti isoforms are absent from the A allele it is logically con- sidered as a mutation from A" rather than vice versa. However, unlike most forward and reverse mutations whose likelihood of occurence differs by 100- to 1000- fold, the observations described here suggest that mutation of A to A" might occur at a similar frequency as mutation of A" to A. This frequency is difficult to esti- mate, but a reasonable upper limit is provided by con- sidering the average ratio of physical to genetic distance for crossing over between homologous mouse chromo- somes, 2000 kb/cM. In the structure described here, separation of the duplicated elements by 20 kb predicts a maximal frequency of

An interesting parallel to the difference between AW and A exists in the case of the nonagouti ( a ) mutation, in which hair cycle- and ventral-specific agouti isoforms are expressed at near-zero levels. BULTMAN et al. (1994) have shown that the a allele is caused by an 11-kb insertion of VL-30 retrotransposon sequences into an agouti intron and that excision by homologous but unequal crossing over mediated by long terminal repeat (LTR) elements that lie at both ends of the inserted material leaves behind a single LTR yet results in phenotypic reversion to the light-bellied agouti phenotype (usually designated as the allele A""). Mutation of a to A W j occurs at a rate approximating per generation (SCHLAGER and DICKIE 1967; DICKIE 1969; FAVOR et al. 1987), which is 10-100 times that expected for reversion of a point mutation (SCHLAGER and DICKIE 1971; FAVOR et al. 1991; FAVOR 1994).

Unlike a and AWY', mutations of A to A" or of A" to A have not been described in laboratory mice, but may have escaped detection because the phenotypic difference between AW/A and A/A mice is less striking than between a / a and A"/a mice. Several studies of wild mice have indicated that the relative prevalence of A us. A" fluctuates in response to environmental conditions (reviewed by SAGE 1981), and a molecular mechanism whereby mutation between these two alleles occurred

at a relatively high rate ( would increase their adaptive value in populations. Intrachromosomal exchange between the duplicated elements that contain exon 1A and pseudoexon 1A may also have occurred more frequently in an ancestral A"a1lele that lacked the insertion of unique DNA that interrupts the duplicated elements (Figure 5). In Peromyscus, for example, presence or absence of a pale ventrum has been suggested to provide a selective advantage depending on whether the soil color is grey or dark red (DICE 1933; BLAIR 1947). These considerations raise the intriguing possi- bility that an inverted duplication similar to the one described here may be present in more distantly related mamamls that commonly exhibit both light-bellied agouti and agouti phenotypes and may provide an evolutionary mechanism for subspecies to acquire heredi- tary yet reversible changes in their ventral pigmentation.

Y.C. and D.M.J.D. contributed equally to this work. We are grateful to EVA EICHER and HOPE SWEET for providing information about wildderived strains. This work was supported in part by a grant from the National Institutes of Health to G.S.B. (DK-28506) who is an Assistant Investigator of the Howard Hughes Medical Institute.

LITERATURE CITED

BWR, W. F., 1947 Estimated frequencies of the buff and grey genes (G . g. ) in adjacent populations of deer-mice (Peromyscus maniculatus). Contrib. Lab. Vert. Biol., Univ. Mich. 3 6 1-13.

BONHOMME, F., and J.-L. GUENET, 1989 The wild house mouse and its relatives, pp. 649-662 in Genetic Variants and Strains of the Labwatoly Mouse, edited by M. F. a. S. LYON, A. G. Oxford Univer- sity Press, New York.

BULTMAN, S. J., M. L. KI.ERIG, E. J. MICHAIJD, H. 0. SWEET, M. T. DAVISSON et al., 1994 Molecular analysis of reverse mutations from nonagouti (A) to Black-and-Tan (A(T)) and White-Bellied agouti (A(W)) reveals alternative forms of agouti transcripts. Genes Dev. 8: 481-490.

BULTMAN, S. J., E. J. MICHAIJD, and R. P. WOYCHIK, 1992 Molecular characterization of the mouse agouti locus. Cell 71: 1195-1204.

DICE, L. R., 1933 The inheritance of dichromatism in the deer- mouse, Peromyscus maniculatus blandus. Am. Nat. 62: 571-574.

DIGKIE, M. M., 1969 Mutations at the agouti locus in the mouse. J. Hered. 60: 20-25.

DIETRICH, W., H. KATZ, S. E. LINCOL,N, H. S. SHIN, J. FRIEDMAN et al., 1992 A genetic map of the mouse suitable for typing intraspe- cific crosses. Genetics 131: 423-447.

DUHI., D. M. J., H. VRIELING, K. A. MIILER, G. L. WOLFF and G. S. BARSH, 1994 Neomorphic agouti mutations in obese yellow mice. Nat. Genet. 8: 59-65.

FAVOR, .J., 1994 Spontaneous mutations in germ cells of the mouse-estimates of mutation frequencies and a moleculal- characterization of mutagenic events. M ~ t a t . Res. 304: 107-118.

FAVOR, J., K A. NEUHAUSER and U. H. EHIJNC:, 1991 The induction of forward and reverse specific-locus mutations and dominant cataract mutations in spermatogonia of treated strain DBA/2 mice by ethylnitrosourea. Mutat. Res. 249: 293-300.

FAVOR, J., A. NEUHAUSER-KIAUS and U. H. EHI.ING, 1987 Radiation- induced forward and reverse specific locus mutations and dominant cataract mutations in treated strain BALB/c and DBA/2 male mice. Mutat. Res. 177: 161-169.

FESTING, M. F. W., 1994 Inbred strains of mice. Mouse Genome 92: 373-495.

JACKSON, I. J., 1994 Molecular and developmental genetics of mouse coat color. Annu. Rev. Genet. 28: 189-217.

~MAEDA, N., and 0. SMITHIES, 1986 The evolution of multigene fami- lies: human haptoglobin genes. Annu. Rev. Genet. 20: 81-108.

MIILER, M. W., D. M. J. DUIIL, H. VRIFXING, S. P. CORI)ES, M. M.


OILMANN et al., 1993 Cloning of the mouse agouti gene predicts a secreted protein ubiquitously expressed in mice carrying the Lethal-Yellow mutation. Genes Dev. 7: 454-467.

NACHMAN, M. W., and C. F. AQUADRO, 1994 Polymorphism and divergence at the 5’ flanking region of the sexdetermining locus, Sry, in mice. Mol. Biol. Evol. 11: 539-547.

Rim, B. A,, L. D. GARVIN and S. C. HARDIES, 1991 Systematic identi- fication of LINE-1 repetitive DNA sequence differences having species specificity between Mus spretus and Mus domesticus. J. Mol. Biol. 219: 635-643.

SAGE, R. D., 1981 Wild mice, pp. 40-90 in The Mouse in Biomedical Research, edited by H. L. FOSTER, J. D. SMALL and J. G. FOX. Academic Press, New York.

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Clon- ing: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York.

SCHLAGER, G., and M. M. DICKIE, 1967 Spontaneous mutations and mutation rates in the house mouse. Genetics 57: 319-330.

SCHLAGEK, G., and M. M. DICKIE, 1971 Natural mutation rates in the house mouse. Estimates for five specific loci and dominant mutations. Mutat. Res. 11: 89-96.

SEARLE, A. G., 1968 Comparative Genetics of Coat Color in Mammals. Academic Press, New York.

SHEN, S. H., J. L. SLICHTOM and 0. SMITHIES, 1981 A history of the human fetal globin gene duplication. Cell 26: 191-203.

SILVEK, L. M., 1995 Town mouse, country mouse, pp. 15-31 in Mouse Genetics. Oxford University Press, New York.

SILW,RS, W. R, 1961 Genes and the pigment cells of mammals. Science 134 368-373.

S I L ~ R S , W. R, 1979 The agouti and extension series of alleles, umbrous and sable, pp. 6-44 in Thr Coat Colors of Mice. Springer- Verlag, New York.

SIRACUSA, L. D., 1994 The agouti gene: turned on to yellow. Trends Genet. 10: 423-428.

SIRAC:USA, L. D., and C. M. ABBOM, 1994 Mouse chromosome 2. Mamm. Genome 5 Spec No: S22-39.

VRIELING, H., D. M. DUHI., S. E. MILIAR, K. A. MII.I.ER and G. S. BARSH, 1994 Differences in dorsal and ventral pigmentation result from regional expression of the mouse agouti gene. Proc. Natl. Acad. Sci. USA 91: 5667-5671.

YONEKAWA, H., K. MORIWAKI, 0. GOTOH, N. MIYASHITA, Y. M ~ r s u - SHIMA et al., 1988 Hybrid origin ofJapanese mice “Mus musculus molossinus”: evidence from restriction analysis of mitochon- drial DNA. Mol. Biol. Evol. 5: 63-78.

Communicating editor: N. A. JENKINS

Opposite Orientations Inverted Duplication and Allelic ... · indicated by dashes. Upper case letters indicate transcribed sequences of exon 1A as determined by cDNA cloning studies

Documents