Proc. Natl. Acad. Sci. USA Vol. 82, pp. 2819-2823, May 1985 Evolution

Evolution of mouse major histocompatibility complex borne by t chromosomes (class II genes/t complex/restriction fragments/polymorphism)

FELIPE FIGUEROA, MLADEN GOLUBI(*, DEAN NIETI(*, AND JAN KLEIN Max-Planck-Institut fur Biologie, Abteilung Immungenetik, Corrensstrasse 42, 7400 Tubingen, Federal Republic of Germany Communicated by Ray D. Owen, December 26, 1984

ABSTRACT Virtually all wild mouse populations carry t H-2 genes are highly polymorphic and hence can be used to haplotypes that cause embryonic lethality or semilethality, differentiate chromosomes (8). On the t chromosomes, the distortion of segregation ratios, suppression of crossing-over, H-2 complex is included in the region of suppressed cross- and male sterility. The t complex of genes is located on ing-over and it is therefore ideally suited for tracing the origin , closely linked to the H-2, the major histocom- of t haplotypes. In a previous study, we used serology of the patibility complex of the mouse. The t haplotypes differ from H-2 as a tool to determine the degree of relatedness each other not only in lethal genes they carry but also in their among the different t haplotypes (9, 10). In this report, we linked H-2 haplotypes. In this study, we compared the class II applied the genomic Southern blot hybridization technique to H-2 genes present on 31 t chromosomes extracted from wild the class II genes of the H-2 complex carried by the different populations in different parts ofthe world. The comparison was t chromosomes. The combination of these two approaches based on the analysis of DNA fragments obtained after diges- provides a glimpse into the t complex genealogy. tion with restriction endonucleases. The results reveal the existence of three major groups of class II alleles representing MATERIALS AND METHODS main branches on the evolutionary tree of the t chromosomes. Alleles within each group are similar if not identical, although Materials. Proteinase K, DNA molecular weight markers, they are borne by chromosomes that have been separated in ribonucleases, and restriction endonucleases were purchased time and space. The presence of similar alleles in Mus musculus from Boehringer (Mannheim, F.R.G.), blotting materials and Mus domesticus suggests that some of them may have been were from Bethesda Research Laboratories (Neu Isenburg, separated for more than 1 million years. This must also be the F.R.G.), the nick-translation kit was from Amersham Inter- minimal age of the t chromosomes but, because at least two of national (Braunschweig, F.R.G.), and [a-32P]dCTP was from the three main branches appear to be related in their origin, the New England Nuclear (Dreieich, F.R.G.). actual age of t chromosomes could be much greater. The Mice. All mice came from our animal colony at the Max observations support the proposal that H-2 genes evolve slowly. Planck Institute for Biology. The strains carrying new t haplotypes were established in our laboratory from wild mice Some matings between wild mice and laboratory strains trapped at the locations listed in Table 1 (6, 9, 10). carrying the dominant mutation short tail, T, invariably DNA Isolation. DNA was isolated from kidney cells. Six to produce tailless progeny. The absence of the tail is attributed eight kidneys were cut into small pieces, which were then to the interaction between T and another mutation, t, which pressed through a sieve (no. 1985-00050, Technorame, occurs in some wild mouse populations with a frequency of Zurich, Switzerland). The cells were washed twice, resus- up to 40% (1). The t mutation affects, in addition to tail length, pended in Tris/saline, pH 8.0, containing proteinase K and several other characteristics (2-5). The t/t embryos may all NaDodSO4, and the suspension was incubated overnight at die (lethal t) or some may die while others survive (semilethal 450C. The incubation was followed by phenol/chloroform t). Only rarely does one find fully viable t haplotypes among extraction, RNase treatment, and ethanol precipitation. wild mice (6), although in the laboratory they can arise from Digestion with Restriction Endonucleases. The high molecu- the lethal haplotypes by recombination. In the +/It mice the lar weight DNA was diluted in the appropriate buffer and lethal or semilethal t haplotypes suppress crossing-over over incubated overnight with the selected restriction endonucle- a distance of at least 15 centimorgans, cause male sterility, ase at the required temperature. A proportion of 6 units of and skew segregation ofthe t chromosome transmitted by the enzyme per 1 ,g of DNA was used. male. At least some of these characteristics are controlled by Southern Blot Hybridization. Electrophoresis of the di- different genes that are locked together into a single unit, the gested DNA was done in a 0.8% agarose gel at 30 mA for t the The about 30 hr. The DNA was then transferred to nitrocellulose complex, by crossing-over suppressor. different t filters (11). The blots were baked, prehybridized for 4-5 hr at haplotypes are distinguished mainly by the type of lethal 42°C, and hybridized with 32P-labeled probes in a solution genes they carry. containing 50% formamide, 0.6 M NaCl/0.06 M sodium Among the many unanswered questions about the t com- citrate, lx concentrated Denhardt's solution, sodium phos- plex, the most fascinating is how did the t haplotypes arise. phate buffer (pH 6.5), 10% dextran sulfate, and salmon sperm As the products of most of the t genes have not been DNA. The filters were washed three times (two times for 20 identified, it is not possible to answer this question by the min in 0.3 M NaCl/0.03 M sodium citrate/0.1% NaDodSO4 at direct study of the t haplotypes. However, linked to the t 48°C and one time for 20 min in 15 mM NaCl/1.5 mM sodium complex on chromosome 17 is the H-2 complex, which codes citrate/0.1% NaDodSO4 at 52°C). Autoradiographs of the for plasma membrane glycoproteins involved in the recogni- blots were prepared by using Kodak XS1 film with a Lanex tion of foreign antigens by T (7). Some of the intensifier screen. The films were exposed for 16-48 hr.

The publication costs of this article were defrayed in part by page charge Abbreviation: kb, kilobase(s). payment. This article must therefore be hereby marked "advertisement" *On leave from the Department of Physiology, University of Zagreb in accordance with 18 U.S.C. §1734 solely to indicate this fact. Medical Faculty, Zagreb, Yugoslavia.

2819 Downloaded by guest on September 25, 2021 2820 Evolution: Figueroa et al. Proc. Natl. Acad. Sci. USA 82 (1985) Table 1. Strains carrying t haplotypes tested in the present study each experiment. The DNA was then digested with restric- t H-2 tion nucleases and the fragments were hybridized with Ap- haplo- t haplo- and Ep-specific probes. Six restriction enzymes were used type Strain Origin group type (EcoRI, HindIII, Xba I, Sac I, BamHI, BstEII, Alu I, Hae IIIB, and Taq I) but not all of the samples were digested with t6 Laboratory 0 all of these enzymes. tw1 - New York or Phila- wI The patterns obtained are shown in Figs. 1 and 2. We have delphia, U.S.A. designated them by letters a, b, c, and so on (see Tables 2 and tuwll GPC882 Buin, Chile Tuwll w30 3). An asterisk indicates that an additional band is contrib- tTuw15 MOY336 Moya, Spain Tuwl2 w65 Alpic Drobic, Italy uted by the non-t chromosome and that we could not decide tLubl Lubi w33 whether a similar band is also contributed by the t chromo- tTuwl2 LRA410 La Roca, Spain Tuwl2 w66 some. Based on the different patterns, the t strains can be tLub4 - Cremona, Italy Lubi, Lub4 w64 Calcinato, Italy divided into four groups, which we designate I through IV. tLub9 Lub4 w64 The Ap3 group I (Table 2) consists of 13 strains, 8 of which tTuw32 ISL33 Haifa, Israel V w56 yielded identical restriction patterns, while 5 (tw73, KPB68, tTuw30 B10.KPB68 Ann Arbor, MI, SL w2 EDY589, ISL37, and ISL33) produced variants of the basic U.S.A. pattern. We designate the basic allele Ad30 and differentiate tTuw18 ISL37 Haifa, Israel SL w2 the variants by additional numbers following the decimal tTuwlO EDY589 Eday, Orkney Is- SL w2 point. The An group II consists of 8 strains, of which all, lands except CR0435 (one difference), yielded identical restriction tw73 S. Jutland, Den- w73 w32 patterns (allele Aw36). The Ag3 group III consists of 5 strains, mark 4 of which yielded identical patterns (allele A 31), although tTuw26 PLD826 Bialowieza, Poland w73 w59 the CR0437 and the Lub7 strains could be distinguished tTuw2 BNK266 Wendelsheim, Tuw2, Tuw25 w36 serologically by their loci (10). The BRU382 strain produced F.R.G. a variant of the basic pattern. The An group IV contains 5 tTuw2l BNK761 Wendelsheim, Tuw2, Tuw25 w36 strains that do not fit into any of the previous groups and F.R.G. whose relationships to one another are not obvious. How- tTuw25 OBL984 Oberer Lindenhof, Tuw25 w36 ever, serological analysis suggests that their Aq3 genes (with F.R.G. the exception of BRW942) are related to the A430 allele (10). tTuw27 ROD1455 Dudelhof, F.R.G. Tuw27 w36 The En group I contains the same strains as the Ad group tTuw24 LGN925 Langenargen, Tuw24 w36 I (Table 3). All of these strains, with the exception of ISL33, F.R.G. yielded identical restriction patterns with all of the enzymes tTuw28 ERP1465 Erpenhausen, Tuw28 w60 tested (allele EP2). ISL33 produces a variant of the basic F.R.G. pattern (allele EI2.1). The EO3II group contains the same 8 tTuw7 CR0435 Nahya, Giza Gover- SL w37 strains as the An group II, 7 of which produced identical nate, Egypt restriction patterns (allele EP36), and only 1 (CR0435) yielded tws New York, U.S.A. w5 w31 a variant pattern (the strain may actually belong to group tTuw23 GPC183 Temuco, Chile w5 w31 EOIII). The En group III consists of 6 strains (or possibly 7 tLub7 Tortona, Italy Lubi w63 strains, if CR0435 is included in this group), 5 of which are tTuw8 CR0437 Nahya, Giza Gover- SL w57 also contained in the group AO3III. The 5 strains carry an allele nate, Egypt (E31) indistinguishable by restriction enzyme analysis but tTuw6 BRU382 Brno, Czechoslova- w73 w58 differentiable by serology. Finally, the En group IV, like the kia An group IV, consists of miscellaneous strains that are t12 Paris, France 12 w28 difficult to place into one of the above groups. tTuw29 BRW942 Aulendorf, F.R.G. Lubi w61 Tauw2O MSW251 Ryazan, Astrachan, w73,0 w38 U.S.S.R. DISCUSSION tw2 New York or Phila- SL w29 The most striking finding of this study of the t-associated delphia, U.S.A. major histocompatibility complex genes is the small number to Paris, France 0 w29 of alleles found at the An and Ed loci among the different SL, semilethal; V, viable. Strains tLubl, tLu,4 tLub7, and tLub9 strains. Twenty-six of the 31 tested strains carry one of three were produced by H. Wiking and A. Gropp (Klinikum der Medizin- major Ad or Ed alleles [or the minor variants thereof, as ischen Hochschule Lubeck, F.R.G.). determined by both restriction enzyme analysis and serology (10)]. The remaining 5 strains carry four different alleles at the Ap3 locus or three different alleles at the En locus. This Probes. The Ad probe was a 5.6-kilobase (kb) EcoRI occurrence of the same class II allele among the wild t mice fragment from the 34.2 BALB/c genomic cosmid clone (12). is not the result of repeated sampling of the same chromo- The probe covered the whole Ap3 from the first exon to some because the chromosomes carrying these alleles have the middle of the 3' untranslated region. The En probe was a been extracted from mouse populations in different parts of 2-kb EcoRI fragment from the 24.2 3ALB/c genomic cosmid the world. For example, the t chromosomes in group I were clone (12) and corresponded to the second exon of the gene. isolated from wild mice captured in the United States, Chile, Italy, Spain, Denmark, Scotland, and Israel. This finding contrasts with the observation that virtually every non-t wild RESULTS mouse in a sample of 60 obtained from different localities has We isolated high molecular weight DNA from kidneys of a different An allele, as determined by the same method as each of the t strains in our collection, Mice carrying the that used in this study (unpublished data). That the t chro- semilethal and viable t haplotypes were homozygotes for the mosomes in each group are not the same is indicated also by t and H-2 complexes; mice carrying lethal haplotypes were the fact that they carry different combinations of class I and heterozygotes with H-2b, H_2k, or H-2w17. Control DNA class II genes (10) and different lethality genes (6). For samples from these three haplotypes were always included in example, group I contains at least nine different t chromo- Downloaded by guest on September 25, 2021 Evolution: Figueroa et A Proc. Natl. Acad. Sci. USA 82 (1985) 2821

Eco RI Hind m Xba I 0M Sac I Bam H I Bst ER e C"! n en C" 3 St. 3: -A C! CD! 3 ' CDl _ ,,_ Cli rem A? v7 An ~~~~~m C IM -. C) C$ C -1 C4 C13 - -- Cl) -- C x 3c 3 31 3 3r _r _. 3 3 3 - 3 C-o -~ o" C" C k b 3C 3 3 3: 3 3 3 kb 23.1 -

4UP 0~~~~~~ 0 1. 7, -~~~~- !IS l 6.7-4* I4. o v

2.4 - 9. 0 0 16K : a a.I $ By

2.0

a c a c b c a b d a b c d e a b e c d a b d c e b d FIG. 1. Autoradiogram of Southern blots of genomic DNA hybridized with the A, probe. Top line, restriction endonucleases and AP alleles of strains that donated the DNA; bottom line, letter designation of patterns. Not shown are patterns produced by the following enzymes: Alu I (a, 2.2 kb; b, 2.0 kb; and c, 2.0 and 1.8 kb); Taq I (a, 2.5 kb; b, 2.7 and 2.5 kb; c, 3.0 and 2.8 kb; and d, 2.7 and 4.4 kb); Hae ITIB (a, 2.2 kb; b, 2.4 kb; and c, 1.8 kb).

somes as distinguished by the t lethality genes 0, wi, wil , caused by rare intra-H-2 recombination. The correlation w12, w73, Lubi, Lub4, SL, and V. These observations extends also to the EJ0 and class I K and D loci. At the Ea locus suggest that all of the alleles within each group derive from all strains in group I carry the E28.1 allele (13). Furthermore, a single ancestral gene and hence that they are all closely strains with the E:2 allele express the E molecules on the related in their origin. The minor variants found within the surface, whereas strains with the C28 alleles do not. Simi- groups are attributable to more recent mutations. Additional larly, group I H-2 haplotypes are characterized by the variation will undoubtedly be discovered when the genes are variants of the KY30 and DW2 alleles, group II by variants of sequenced, but this will not change the conclusion that the KW29 and DW29 alleles, and group III by the KW3N allele (10). We alleles have a common origin. conclude, therefore, that the class I and class II genes within The second striking observation is that the groups ofalleles each group of t chromosomes are in strong linkage disequi- at the Ad locus match, with one exception, the groups at the librium. This observation is again in sharp contrast to wild Ed locus. Thus, the Al30 allele always occurs together with non-t mice in which linkage disequilibria within the H-2 the E2 allele, A336 with Ew36, and A3E331.with The one complex are far less striking (14), and it underscores the exception is the CR0435 strain in which an A 36 allele common origin of H-2 haplotypes within each group. occurs together with an EP3I allele. This occurrence may be What, however, is the relationship among the groups?

Eco RI HindM XbaI Sac I Barn HI Bst ElI

c

C3 . c 3. 1 C1 --c - - - __ _,,3 -- 3: - _- 3 -- 3 .m -0 kb en Cn cn cm en co n 33 3 B 3: it 3t 3c 3: 3m 3c 3c 3r -b-jc X C3j] 3 u.4 D

6.7

6 W 4.4 0 4D *b

0-~ 40 0 2.3 lk".H \.. 40~~~~~~~ 2.0 & * a W. * \~~~'

a b c d a b b a c a c b b d c e a f c a b FIG. 2. Autoradiogram of Southern blots of genomic DNA hybridized with the En probe. Top line, restriction endonuclease and E,3 alleles of strains that donated the DNA; bottom line, letter designation of patterns. Downloaded by guest on September 25, 2021 2822 Evolution: Figueroa et al. Proc. Natl. Acad. Sci. USA 82 (1985) Table 2. Restriction enzyme pattern distribution among t strains: Table 3. Restriction enzyme pattern distribution among t strains: the Ap3 gene the Ep gene Restriction enzyme pattern Ap Restriction enzyme pattern E Group Strain Ec Hi Xb Sa Ba Bs Al Ha Ta allele Group Strain Ec Hi Xb Sa Ba Bs allele I twl, t6 a a a a a a a a a w30 I tw1, t6 a a a a a a w2 GPC882 a a a a a w30 GPC882 a a a a a a w2 MOY336 a a a a a a a a a w30 MOY336 a a a a a a w2 tLub1 a a a a a a a a a w30 tLubl a a a a a a w2 tLuM, tLub9 a a a a a a a a a w30 tLub4, tLub9 a a a a a a w2 LRA410 a a a a a a a a w30 LRA410 a a a w2 tw73 a a a b* a a a w30.2 tw73 a a a a a a w2 KPB68, KPB68, EDY589, EDY589, ISL37 a a a a a a w2 ISL37 a a a a b a a a a w30.1 ISL33 d a b a a a w2.1 ISL33 a a a d a a w30.3 II PLD826 b b b b b w36 II PLD826 a c b c a b b b* a w36 BNK266, BNK266, BNK761 b b b* b b b w36 BNK761 a c b c a b b b* a w36 OBL984 b b b* b b b w36 OBL984 a c b c a b b b* a w36 ROD1455 b b b* b b w36 ROD1455 a c b c a b b* a w36 LGN925 b b b* b b b w36 LGN925 a c b c a b b b* a w36 ERP1465 b b b* b b b w36 ERP1465 a c b c a b b b a w36 III CRO435 b b a b f a w31.1 CRO435 a c b b b c a w36.1 tLub7 b b a b c* a w31 III tLub7 b* b* c* b c* c* c b* b* w31.1 GPC183 b b a b c* a w31 GPC183 b* b* b c* c* b* b* w31 tws b b a b c* a w31 tws b* b* c* c* b* b* w31 CRO437 b b a c a w31 CRO437 b b c b c c c b b w31.2 BRU382 b b a b c* a w31.2 BRU382 c b* d b c* c* c b* c w31.3 t12 b b a b c* a w31 IV MSW251 d c e d a b* b c* a w38 IV MSW251 c b c c d a w38 t12 e b d c* c* d w28 to, tw2 b b a b e c w29 to, tw2 b d d c c* c* b* w29 BRW942 b b a b e c w61 BRW942 d e e d a c* b c* a w61 See legend to Table 1 for restriction enzyme designations. Ec, EcoRl; Hi, HindIl; Xb, Xba I; Sa, Sac I; Ba, BamHI; Bs, *In heterozygotes, the extra band is presumed to be contributed by BstEII; Al, Alu I; Ha, Hae IIIB; Ta, Taq I. the non-t chromosome. *In heterozygotes, the extra band is presumed to be contributed by the non-t chromosome. The second explanation is that all of the t-associated H-2 haplotypes derive from a single ancestral haplotype and that There are three possibilities. The first possibility is that each the large intergroup differences reflect great evolutionary group originated independently so that the t chromosomes distances. The haplotypes in group IV might then represent arose de novo at least three times (we ignore the fourth group the "missing links" connecting the three main groups or of miscellaneous haplotypes, which are difficult to classify). additional branches ofthe tree. Additional branches might be We consider this possibility unlikely because of serological, discovered by further sampling ofwild populations. The third DNA, and protein analysis of the 31 t haplotypes. The possibility is that some ofthe class II alleles are derived from serological study (10) has led to the identification of several non-t chromosomes by intra-H-2 recombination. Crossing- class I antigens that are characteristic ofthe t strains and thus over involving t chromosomes might have been more fre- far have not been found in either non-t wild mice or inbred quent in the early stages of t evolution before all of the strains. These antigens do not occur in all of the t strains but suppressors accumulated than it is now they are dispersed in all four groups defined by the analysis We are unable to decide between these last two possibili- of class II genes. This situation can best be explained by ties because the available data do not permit an accurate postulating that the antigens, originally present in a single estimate of evolutionary distances between the groups. Such ancestor of the t haplotypes, are in the process of being lost an estimate might have to await DNA sequence data. How- from some of the haplotypes. Relatedness of class I major ever, it should be pointed out that groups II and III carry the histocompatibility complex genes present in the classical t same deletion in the Ea gene (19) and that group II appears to strains is also suggested by limited restriction enzyme analy- be related to group I in the Ap gene and group III may be sis (14-16). The DNA analysis utilized probes produced by related to group II in the Ep gene. These observations may Lehrach and his colleagues (17) and produced by hybridizing favor the second explanation, although it is very likely that with different segments of the centromeric portion of chro- the nondefective Ea gene has been brought into the t mosome 17. The restriction enzyme pattern of the DNA chromosomes by recombination. fragments identified by these probes is characteristic of the t The data presented in this communication also give an haplotypes and distinct from non-t chromosomes. This pat- estimate of the minimal age of the t chromosomes. In group tern is the same for all of the lethal t chromosomes in our I, the tw73 chromosome very likely derives from Mus collection (unpublished data). Finally, the evidence from the musculus, the eastern form of the house mouse, because it protein analysis is based on the identification of T-complex was extracted from a population in the M. musculus territory proteins (Tcp), which are characteristic of t haplotypes (18). (20) and because we have thus far found tw73-bearing chro- Again, these proteins are shared by all of the t strains thus far mosomes only in this species. All other t chromosomes in this tested (unpublished data). These studies indicate that all of group derive from Mus domesticus, the western form of the the known t haplotypes derive from a single ancestral form house mouse. Similarly, in group II the tTuw26 chromosome of and thus argue against the first explanation. PLD826 is definitely derived from M. musculus, whereas all Downloaded by guest on September 25, 2021 Evolution: Figueroa et al. Proc. Natl. Acad. Sci. USA 82 (1985) 2823 other chromosomes come from M. domesticus. The fact that 1. Klein, J. (1975) Biology of the Mouse Histocompatibility-2 within a single group chromosomes occur with identical or Complex (Springer, New York). nearly identical class II genes in two different mouse species 2. Bennett, D. (1975) Cell 6, 441-454. suggests that an ancestral chromosome rise to all ofthe 3. Klein, J. & Hammerberg, C. (1977) Immunol. Rev. 33, 71-104. giving 4. Lyon, M. F. (1981) Symp. Zool. Soc. 47, 71-104. haplotypes in this group existed already before the species 5. Silver, L. M. (1981) Cell 27, 239-240. separated from a common ancestor. Since it is estimated that 6. Klein, J., Sipos, P. & Figueroa, F. (1984) Genet. Res. 44, M. musculus separated from M. domesticus more than 1 39-46. million years ago (21), the t chromosomes must be at least 7. Klein, J., Figueroa, F. & Nagy, Z. A. (1983) Annu. Rev. that old, and, if the different groups have a common origin, Immunol. 1, 119-142. the large intergroup differences between the class II alleles 8. Klein, J. & Figueroa, F. (1981) Immunol. Rev. 60, 23-57. suggest that the t chromosomes are much older than that. One 9. Sturm, S., Figueroa, F. & Klein, J. (1982) Genet. Res. 40, could, of course, argue that there has been a recent exchange 73-88. of between M. musculus and M. but for 10. Niletid, D., Figueroa, F. & Klein, J. (1984) Immunogenetics genes domesticus, 19, 311-320. this there is no evidence as far as nuclear genes are con- 11. Southern, E. M. (1975) J. Mol. Biol. 98, 503-517. cerned. 12. Steinmetz, M., Minard, K., Horvath, S., McNicholas, J., If our estimate of the age of t chromosomes is correct, it Frelinger, J., Wake, C., Long, E., Mach, B. & Hood, L. (1982) supports our earlier suggestion (22) that the H-2 genes do not Nature (London) 300, 35-42. evolve more rapidly than most other genes. If some of the 13. Dembid, Z., Singer, P. & Klein, J. (1984) EMBO J. 3, genes in each of the three groups have been separated for 1647-1654. some 1 million years, they have changed remarkably little. 14. Silver, L. M. (1982) Cell 29, 961-968. This slow evolution ofthe H-2 genes is not a peculiarity ofthe 15. Shin, H.-S., Stavnezer, J., Artzt, K. & Bennett, D. (1982) Cell t chromosomes: in our earlier we reached same 20, 969-976. study (22) the 16. Rogers, J. H. & Willison, K. R. (1983) Nature (London) 304, conclusion by studying non-t chromosomes. Only now, 549-552. however, can we place the course of the evolution of the H-2 17. Rohme, D., Fox, H., Herrmann, B., Frischauf, A.-M., genes on a rough time scale. Our earlier data (13, 19) suggest Edstrom, J.-E., Mains, P., Silver, L. M. & Lehrach, H. (1984) that the location of H-2 genes on the t chromosome does not Cell 36, 783-788. exempt them from selection. 18. Silver, L. M., Uman, J., Danska, J. & Garrels, J. J. (1983) Cell 35, 35-45. We thank Mr. Peter Sipos for establishing and maintaining of the 19. Dembid, Z., Ayane, M., Klein, J., Steinmetz, M., Benoist, t strains, Dr. Matthias Wabl (Friedrich-Miescher-Laboratorium der C. 0. & Mathis, D. J. (1985) EMBO J. 4, 127-131. Max-Planck-Gesellschaft, Tubingen, F.R.G.) for introducing one of 20. Dunn, L. C., Bennett, D. & Cookingham, J. (1973) J. Mam- us to the blotting techniques, Dr. Lee Hood for the An probe, Dr. mal. 54, 822-830. Michael Steinmetz for the En probe and advice on the experimental 21. Sage, R. D. (1981) in The Mouse in Biomedical Research, eds. protocol, Dr. H. Winking for supplying us with some of the mouse Foster, H. L., Small, I. D., & Fox, I. D. (Academic, New strains, Ms. Martha Kimmerle for editorial assistance, and Ms. York), pp. 39-90. Marion Vogelsang for secretarial help. This research was supported 22. Arden, B. & Klein, J. (1982) Proc. Natl. Acad. Sci. USA 79, in part by Grant CA 34965 from the National Cancer Institute. 2342-2346. Downloaded by guest on September 25, 2021