EVIEWS
I n 1927, a report entitled Sur ia reproduction des
souris anoures (On the reproduction of tailless mice)
by two Russian expatriates working in Paris appeared
in the proceedings of the French Academy of
Science#. Although these researchers did not realize it
at the time, they had stumbled, quite by accident,
upon a fascinating 'organism' within an organism,
known today as a t haplotype 2,3. Even though its own
genome is ten times the size of that of E. coli, a
t haplotype can unobtrusively hide away inside the
genome of a wild house mouse - like a Trojan horse
with genetic soldiers waiting to slay competitor
homologs during the haploid stage of the life cycle,
the only time when such an action would not be
suicidal4.5.
From 1927 until perhaps a dozen years ago,
t haplotypes - originally referred to as 't mutations' or
't alleles' - were viewed as mysterious elements that
existed and persisted in various subspecies populations of the house mouse Mus musculus found all
over the world 6. The bizarre features of t haplotypes
were many and varied. They appeared to be mutant
alleles of the T locus on chromosome 17 that had
pleiotropic effects on embryonic development, male
fertility and genetic recombination; they appeared to
revert to less deleterious alleles at a surprisingly high
frequency (2-5 x 10-3); and they acted in direct violation of Mendel's first law in being transmitted from a
heterozygous +/t male to 99°6 or more of his
offspringz.3. But t haplotypes themselves are invisible:
neither heterozygous +/t nor live-born homozygous t / t
mice show any evidence of gross abnormalities. It is
only in conjunction with a null mutation at the pseudoallelic Tlocus (on the other chromosome 17 homolog)
that a cryptic t haplotype can be uncovered through a
developmentally inconsequential effect on tail length:
T/+ animals have a short tail whereas T/t animals are
born tailless.
The peculiar journey of a
selfish chromosome:
mouse t hapl0types and
meiotic drive
LEE M. SILVER
Mouse t haplotypes are descemleats of a variant form of
chromosome 17 that evolved the ability to propagate itself
at the expense of the wild-type homolo&from heterozygous
+/t male~ A~thoughonce etagnmttc, these widespread
selfish chromosomes have revealed m a y of their secrets
in response to a combttwd assault with molecular, getwtic
aml phylo&ettetic teclmiques. This review summarizes the
current uuderstanding oft haplotypes and their raison
d'etre.
With the many properties of t haplotypes came
many questions: What exactly is the nature of the
't mutation'? How and why does it affect so many
unrelated traits? Where, when and how did it
originate? Why did it arise and why does it persist
throughout most wild populations today? We now
know the answers, at least in general terms, to most of
these questions, and these will be summarized in this
review.
What is the natm'e of the 't mutatiom'?
with the combined use of molecular markers and
formal genetic analysis, much progress has been made
in understanding the structure of t haplotypes, their
relationship to each other, and the genetic basis of
the various phenotypes that they express. All contemporary t haplotypes are descendents of a single
recent ancestor (as discussed
more fully below) and, as
such, are all closely related in
structure and function. A complete t haplotype extends over
the proximal half of chromosome 17 and is defined relative
to the wild-type homolog by a
series of four non-overlapping
inversions (Fig. 1). As would
be expected, these inversions
act to suppress recombination
along this region of the chromo•
• q,'p•
~[UU'F&~
J]ob
some in heterozygous +/t
animals.
The view of a t haplotype
as an extended genetic entity
FIGH
that has altered chromosomal
Schematic maps of t haplotype and wild-typeforms of mouse chromosome 17. The structure was first enunciated
t haplotype is defined by four relative inversionsrepresented by boxes on the top t haplotype
chromosome and numbered 1--4for in(17)1-in(17)4 respectively.Loci that play a role in the over 30 years ago by Mary
TRD phenotype are shown above this line. DI-D5 represent Tcdl-Tcd5; R represents Tcr Lyon7 in a series of papers
(Ret. 30). The relative positions of four marker loci - T, Tcp-1, Hba-ps,t and H-2- are ignored by other investigators
indicated below the wild-typechromosome, and lines show their locationson the in the field. With the dist haplotype. The locations of various t-associated lethal mutations are indicated by asterisks covery of the inversions,
within the t inversionboxes2,13,3°. Lyon's view of t haplotypes was
t haplotyp
Wild-type
T1GJ~'," 1993 VOL.9 NO. 7
C 1993 Elmvier ~'ience Publishers Lid {ilK) 0168 - 9525/93/$06.00
~EVIEWS
validated, and there was no longer any need to
ascribe all other mutant t phenotypes to the
pleiotropic expression of a single locus. In fact, the
various phenotypic effects can be attributed to
mutant alleles at unrelated genes throughout the
20 cM (-40 Mb) t region. As a result of recombination
suppression, these mutant alleles are generally transmitted together as a single genetic entity. This entity
appears, in formal terms, to be an allele of the T
locus (as well as of all other markers that were later
found to lie within the affected region of chromosome 17). However, infrequent crossovers do occur,
most often at sites between adjacent inversions, at a
rate of 2-5 per 1000 gametes s. These crossovers produce 'partial' t haplotypes that retain only a portion
of the original 'complete' t haplotype and express
only a subset of its properties. Thus, what were once
thought to be high-frequency partial genetic revertants are now seen to be low-frequency recombinants, as first suggested by Lyon 7.
carrying the wild-type chromosome are killed, but in
males with two complete t haplotypes, the spermatids
kill each other.
Selfish chromosomes that have many of the t-like
properties just described have been found in other
species; examples are the Segregation distorter system
in Drosophila and the Spore killer system in Neurospora 2,5. Interestingly, although functionally analogous,
all three of these systems appear to operate through
entirely different molecular mechanisms.
How do t haplotypes function?
Complete t haplotypes can express three different
classes of phenotypes, all of which are charac:erized
by adverse effects on cell differentiation: distortion of
male transmission ratio, embryonic lethality and male
sterility. The most important phenotypic property is
the first of these: the ability of a t haplotype to transmit
itself at the expense of the wild-type homolog from
heterozygous +/t males. (This phenomenon has been
observed in a variety of organisms and is commonly
referred to as meiotic drive even though, in this and
other cases, it is entirely due to differential effects on
gamete differentiation that occur postmeiotically.) One
can not imagine a more advantageous mechanism for
ensuring allelic survival than the extermination of
gametes containing the competing allele. To date, six
distinct loci have been identified that interact to mediate the transmission ratio distortion (TRD) phenotype 9a0 (Fig. 1). Since these loci are spread across the
entire length of the t haplotype, all recombination
events yield partial haplotypes that do not distort
transmission ratio to the same degree as a complete
t haplotype. In spite of extensive research in this
area na2, the molecular mechanism of TRD remains
elusive.
Most, but not all, naturally occurring t haplotypes
carry recessive lethal mutations that act prenatally. In
general, lethal phenotypes can be traced to single
locus mutations that map within the t region ~3 (Fig. 1).
Within the t haplotypes examined to date, 16 complementing lethal loci have been identified1"t. The
t-lethal loci appear not to have any relationship to
one another, nor to any of the loci that mediate TRD
(Ref. 2).
Finally, all males that carry two complementing
complete t haplotypes are unconditionally sterile. It
seems likely that the same set of genes that acts in a
dominant fashion to mediate TRD also acts in a recessive manner to produce sterility~5. Although it is clearly
an over-simplification, the relationship between TRD
and sterility can be understood by imagining that
spermatids that bear the t haplotype always 'kill' their
meiotic partners; in heterozygous males, only sperm
The ancestral t haplotype
Although t haplotypes can co-exist within the same
nuclei as their wild-type counterparts, the inversions
that distinguish wild-type and t forms of chromosome
17 limit genetic exchange between the two*. As a
consequence, DNA sequences located within the
boundaries of t will accumulate nucleotide changes
that can be used by phylogenetic sleuths as clues both
to the origin of t haplotypes, and to their relationships
to each other and to the wild-type forms of chromosome 17 present in the house mouse and its relatives.
The eadiest comparative studies of t-associated loci
were based on small numbers of restriction fragment
length polymorphisms or limited sequence data.
Although these studies were of limited resolution, they
did provide evidence for a common origin of all
t haplotypes3,17-19. More recently, a much clearer
picture of the origin and evolution of t haplotypes has
been obtained with high-resolution phylogenetic analyses of DNA sequences associated with two t-associated
loci: an intron from the Tcp-1 gene located in the second inversion, in(17)2 (Ref. 20), and the Hba-ps4
pseudogene located in the fourth inversion, in(17)4
(Refs 16, 21; Fig. 1).
The first critical observation to come out of
sequence analyses of both Tcp-1 and Hba-ps4 is that
t haplotypes are accumulating nucleotide changes at
the same rate as wild-type chromosomes 21,22. This
finding is important for two reasons. First, it demonstrates that the peculiar characteristics of t haplotypes
- particularly meiotic drive and recombination suppression - have no discernible effect on the ticking of
the molecular clock. Second, it validates the use of
sequence comparisons among t haplotypes as a means
for estimating the time at which their most recent common ancestor was alive. The data collected for this last
purpose are remarkable. Four t haplotypes were
sequenced across a 610 bp region in one Tcp-1 intron,
yet no nucleotide differences were observed, and
among 19 haplotypes analysed across a 2.6 kb region
at the Hba-ps4 locus, only a single polymorphic
nucleotide site was uncovered. This very low level
of polymorphism is in contrast to results typically
obtained in comparisons of wild-type chromosomes,
even within a single population of mice. The accumulated data indicate that a mouse carrying the common
*The limitation in genetic exchange is not absolute in that
the products of gene conversion between t and wild-type
alleles that do not recombine flanking markers have been
observed in the inversion region in(17)4 (Ref. 16). Converted
alleles can be identified and treated as such in further
analyses of t haplotype origins and evolution.
"lagJULY1993 VOL.9 NO. 7
B
EVIEWS
provided the evidence for this
conclusion, as well as a means
for estimating both the time at
domesticus
which t lineages became distinct from wild-type lineages,
M. musculus
musculus
and the type of mice in which
these
events occurred. An imcastaneus
portant assumption that is
made in the interpretation of
bacwianus
such data is that the divergence
of
wild-type- and t-alleles was
M. macedonicus
initiated by the appearance of
the inversion polymorphism
M. spicilegus
that encompasses the locus
under analysis. This being the
M. spretus
case, times of divergence calculated for different loci would
3 myr
2 myr
1 myr
0
provide dates for the generation of each of the four t haploFIG~
type inversions.
High quality phylogenetic
Phylogenetic history of the house mouse and its closest relatives, and the evolution of
t haplotypes. To understand the history of t haplotypes, one must first have a clear picture of data are only available for two
the evolutionary relationships among populations within the house mouse M. musculus and loci that map to the two largest
its closest relatives. The consensus picture after a decade of molecular phylogenetic studies inversions 2o.zl. The data indicate
at numerous loci is depicted here in the form of a tree, or cladogram. The approximate times that the in(17)2 polymorphism
at which different evolutionary branchpoints occurred are indicated by the scale (in millions is much older than the in(17)4
of years) at the bottom. There are four distinct lineages - domesticus, musculus, castaneus polymorphism (-3 as opposed
and bactrianus- within the highly commensal house mouse group. Although each of these
lineages has a separate geographical range in nature, individuals can usually interbreed to -2 million years old). This
freely in the laboratory and produce fully viable and fertile offspring. This and the evidence order of evolutionary events is
of recent t haplotype flow among wild populations (indicated by a vertical thick line), consistent with the e#netic data
suggests that the four house mouse lineages are not yet independent species 3~, a!though that map the single most crucial
some have argued otherwise3~, The most distant species with which M. musculus has mated gene responsible for TRD (the
productively in the laboratory is M. spretus. The resulting FI hybrid males, but not the t complex responder locus, Tcr)
females, are sterile in accordance with Haldane's rule, Two other well-defined extant as well as other TRD loci in
relatives of M. musculus are M. macedonicus (formerly called M. abbotti ) and M, spicilegus tight linkage with in(l 7)2 (Ref.
t formerly called M, hopItdanus)3t~,The position of the breakpoint for the lineage containing 24). The implication is that
these two sister species varies depending on the locus analysed: the one shown here is
the evolution of a modern-day
based on data from studies of T~p-l, The lineage of t haplotypes is represented as a thick
grey line. Arrows represent branchpoints induced by each of the four t region inversion t haplotype must have begun
events; the associated number corresponds to the inversion number and the arrow indicates with the in(17)2 inversion,
the lineage on which it occurred, which was later followed by the
appearance of the other three
inversions.
ancestor of all present-day t haplotypes must have
Independent evidence for the timing of tile event
been alive within the past 100 000 years and may have
that led to the in(1 7)2 inversion has come from breedbeen on the planet as recently as 10 000 years ago.
ing studies aimed at determining the form of this polyThe very close relationship among all t haplotypes
morphism present in the three species M. musculus,
is most surprising because these genomic entities
M. spretus and M. macedonicus 25. The results of these
are distributed across four different M. musculus
experiments, in conjunction with other data, led to the
subspecies that have been evolving separately for
remarkable finding that the actual in(17)2 inversion
0. 5- 1 . 0 x 106 years and now occupy distinct
event occurred in a wild-type lineage and not within
geographical rangesZ3 (see legend to Fig. 2). Taken
the nascent lineage of the t haplotype. The demontogether, the data indicate that the ancestral t chromostration that M. musculus and M. macedonicus both
some must have been present in one of these subcarry a similar inverted chromosome, while M. sprettts
species after its divergence from the others. The impliand t haplotypes are both non-inverted, places upper
cation is that the descendants of this t chromosome
and lower limits on the time at which the inversion
were able to move rapidly across the geographical
could have occurred - before the divergence of the
borders that separate all of these subspecies2,19.
M. muscuh~s lineage from the M. macedonicus lineage,
but after the branching of M. spretus (Fig. 2).
When and where did t haplotypes originate?
Although the most recent common ancestor to all
Putting it all together:, the history and future of
present-day t haplotypes was alive within the past
t haplotypes
100 000 years, the evolutionary events that led to this
Origin
final product had accumulated over a much longer
From the data accumulated to date, it is possible
period of time. Once again, phylogenetic analysis has
to speculate on a scenario for the origin and
haplotype
I
I
I
,
I
TIG JULY 1993 VOL.9 NO, 7
m
[~EVIEWS
evolution of mouse t haploSecond variant chromosome
types. First, it appears that 'the
Fir-" "'+-'--" .t. . . . . . . . .
t chromosome' evolved initially
from a wild-type form of
chromosome 17 present in -an
ancestor common to the house
mouse ( M. musculus) and
M. macedonicus, some three
million years ago (Fig. 2). This
evolution was probably initiated by the chance accumulation of alleles at several
2. Dispersion of
3. Present-day M. musculus
1. Ancestral population
closely linked genes [in the
f'wstvariant chromosome
3 cM region now defined by
2.5 myr
the inversion in(17)2], which
1.5 myr
Today
together acted to enhance
2
myr
]
3 myr
[
their
own
transmission
through the male germ line at
pIGNa
the expense of their meiotic
partners. Once such a group A population view of chromosome 17 evolution. The three rounded boxes contain schematic
of 'TRD alleles' came into representations of the forms of chromosome 17 present in populations of mice at different
existence, it would take on a stages of evolution along the lineage that led to the modem-day house mouse. Box 1 shows
"¢e of its o w n in the form of a the ancestral population before the introduction of any inversions. The first inversion,
in(17)2, occurred within one of these ancestral chromosomes approximately 2.5 million
'selfish' chromosomal region.
years ago and rose to a high frequency as shown in Box 2. The success of this first variant
In
this
context,
genetic chromosome could have been due to a positive meiotic drive associated with the isolated
changes that enhanced the
in(17)2 inversion l° or, alternatively, it could have been a consequence of positive selection
survival potential of the vari- caused by the presence of a positive meiotic drive system that had already coalesced on a
ant allele group would show non-inverted form of the chromosome ~. Whatever the case, this first variant failed to reach
positive selection in the com- fixation. However, the original non-inverted chromosome began (or continued) to evolve a
meiotic drive system which accumulated inversions in(l 7)1, in(17)3 and in(17)4.
petition that would ensue
between
different chromo- Eventually, the original ancestral form of chromosome 17 became extinct; this led to
somal descendants z6. Increased present-day house mouse populations in which the original variant chromosome is now
survival of the t-bearing chromo- called wild-type and the second variant chromosome is called a t haplotype (Box 3).
Sc !e represents evolutionary time in millions of years (myr).
some would result from two
types of changes:
first in(17)l and in(17)3, and finally in(17)4, the
(1) further mutations in the original TRD alleles or
culminating event in the structural assembly of a
others nearby that increased relative rates of self-transmodem-day t haplotype.
mission;
Although the structurally complete primordial
(2) inversions that extended over the allele group
t haplotype was present in the population tl.*t
to reduce the frequency at which self-destruction
diverged into the four M. muscuhts subspecies (Fig. 2),
could occur as a result of crossing over.
it appears to have survived only along the line leading
It is important to point out that any allele that can
to one of these - most likely Mus mus domesticus.
transmit itself genetically at the expense of alternative
alleles at the same locus will always reach rapid fix- Why? One possibility is that the primordial t chromosomes did survive along other subspecies lines, but
ation within a population unless counteracting negawere later eliminated by competition with more fit,
tive selective factors come into play. However, if the
fully evolved descendants of the ancestor of all
original TRD alleles had become fixed, there would be
modern-day t haplotypes (Fig. 2).
no t haplotype to study today. The absence of fixation
can be interpreted as evidence that members of the
The most recent addition: lethal alleles
original group of TRD alleles had deleterious effects
The last important event in the past evolution of
on the fertility of homozygous males, as is seen with
t haplotypes occurred after their recent divergence
their modern-day descendants tS.
from a common ancestor. This involved the recruitment of various lethal mutations by t haplotypes
Evohttion
present in different populations TM. The high level of
The data strongly suggest that the t haplotype
polymorphism associated with t-lethal alleles stands in
evolved piecemeal out of 'wild-type' forms of chromostriking contrast to the low levels of polymorphism
some 17 that were present along the lineage leading to
found at other t loci. This empirical finding strongly
the M. muscuhts group of species. The original selfish
suggests
that recessive lethality imparts a selective
entity was composed from variant alleles of genes
advantage to the t chromosome. The rationale for such
within the in(17)2 region that were held together by
an advantage is that it prevents the birth of homothis inversion polymorphism. Later, over a period of
zygous t / t males who, being sterile, could not pass
perhaps one million years, variant alleles at other,
the t chromosome on to future generations, but could
more distant, genes on the chromosome were recruited
still compete with +/t males for parental investment
with the addition of accompanying inversions m
"nt; Jt,L'," 1993 VOL.9 NO. 7
[~qEVIEWS
and material resources. The precise nature of the lethal
effect is not important so long as the mutation occurs
within t haplotype chromatin and acts early during
development; this is, in fact, true of all known
t-associated lethal mutations.
The distant past a n d the fiav~re?
There are two peculiar phylogenetic observations
that deserve closer attention. First, even though the
t haplotype is quite old, all present-day examples derive
from a very recent ancestor. This suggests that as the
t haplotype evolved over millions of years, new
improved versions - with higher effective levels of TRD
- may frequently have swept through the population,
eliminating all previous versions of t. But why would a
genetic system of this type remain in constant turmoil
over such a long period? The answer might be that the t
haplotype is forced to continuously overcome the
effects of suppressors of TRD: indeed, such suppressors
have been found on present-day wild-type homologs of
chromosome 17, as well as on other chromosomes zT.
The term 'chromosome wars' was coined by Mike
Han~ner (University of Arizona) as a metaphor for this
hypothesis, which leads to the suggestion that a current
wild-type homolog could, some day, do more than just
fight for parity, and eventually become the selfish
chromosome of the future.
This last notion would certainly be dismissed by
most as extreme speculation except for the fact that a
much earlier switch in selfish chromosome status is an
intriguing explanation for the peculiar finding that the
original chr, nosome 17 inversion in(17)2 occurred
on what is now the -'ild-type homolog. This might
suggest that tile primordial t haplotype arose in
response to an assault by the original selfish chromosome, which we now see as the modern-day wild-type
form (Fig. 3). However, an alternative possibility that
must be considered is that the first inversion, although
present on the 'wild-type chromosome', was selected
in response to the expression of meiotic drive by a
primordial t haplotype that already existed a~,
Mouse chromosome 17 and sperm function
Why has a t haplotype-like entity evolved only on
chromosome 17 and not on other mouse chromosomes? There are several plausible hypotheses, but
the one I favor is based on an original chance distribution of genes that led to a unique closely linked
set with the potential for evolving altered transmission ratio. Linkage of these genes in the mouse
was almost certainly a random event, since the same
linkage relationship has not been conserved in the
human genome29.30. However, it is peculiar that so
many genes with dramatic effects on fertility map
within or near the t region. These include not only
TRD loci, but also four independent interspecific
hybrid-sterility factors (Refs 31, 32; S. Pilder and
L.M. Silver, submitted), a polypeptide likely to facilitate the species-specific binding of sperm to eggs
(Tsai et al., submitted), and numerous cloned
sequences with sperm-cell-specific expression33. Is
the existence of this conjunction of loci meaningful
in any way? Has the t haplotype revealed all its
secrets? Only time will tell!
Acknowledgements
I thank J. Forejt and K. Ardlie for critical reading of the
manuscript, and also M. Hammer, J. Schimenti, S. Pilder, F.
Bonhomme, K. Willison and B. Charlesworth for further ideas
and critical commentary.
References
1 Dobrovolskaia-Zavadskaia, N. and Kobozieff, N. (1927)
C. R. S~ances Soc. Biol. Ses Fil. 97, 116-119
2 Silver, L.M. (1985) Annu. Rev. Genet. 19, 179-208
3 Klein, J. (1986) NaturaiHistory oftheMajor
Histocompatibility Complex, John Wiley & Sons
4 Silver, L.M. (1988) Curr. Top. MicrobioL Immunol. 137,
64-69
5 Wu, C-I. and Hammer, M. (1990) in Evolution at the
Molecular Level (Selander, R.K., Clark, A.G. and Whittam,
T.S., eds), pp. 177-203, Sinauer Associates
6 Figueroa, F., Neufeld, E., Ritte, U. and Klein, J. (1988)
Genetics 119, 157-160
7 Lyon, M.F. and Phillips, R.J.S. (1959) Heredity 13, 23-32
8 Bennett, D., Dunn, L.C. and Artzt, K. (1976) Genetics83,
361-372
9 Lyon, M.F. (1984) Cell37, 621--628
10 Silver, L.M. (1989) Genet. Res. 54, 221-225
11 Snyder, L.C. and Silver, L.M. (1992) Mamm. Genome 3,
588-596
12 Bullard, D.C., Ticknor, C. and Schimenti, J.C. (1992)
Mature. Genome 3, 579-587
13 Artzt, K., McCor;nick, P. and Bennett, D. (1982) Cell28,
463-470
14 Klein, J., Sipos, P. and Figueroa, F. (1984) Genet. Res. 44,
39--46
15 Lyon, M.F. (1986) Cell 44, 357-363
16 Hammer, M,F., Bliss, S. and Silver, L.M. (1991) Genetics
128, 799--812
17 Delarbre, C. et al. (1988) Mol. Biol. Evoi. 5, 120-133
18 Silver, L.M. et al. (1987) Moi. Biol. EvoL 4, 473--482
19 Willison, K.R., Dudley, K. and Potter, J. (1986) Cell 44,
727-738
20 Mork~. T. etal. (1992~ Proc. NatlAcad. Sci. USA 89,
6851-6855
21 Hammer, M,F. and Silver, L.M. Mol. Biol. Evoi. (in press)
22 Kubota, H. et al. (1992) Mature. Genome 2, 246-251
23 Bonhomme, F. (1986) Curr. Top. Microbiol. lmmunol. 127,
19--34
24 Rosen, L.L., Bullard, D.C., Silver, L.M. and Schimenti, J,C.
(1990) Genomics 8, 134--140
25 Hammer, M.F., Schimenti, J. and Silver, L.M. (1989) Proc.
Naa Acad. Sci. USA 86, 3261-3265
26 Lewontin, R.C. (1962) Am. Nat. 887, 65--78
27 Bennett, D., Alton, A.K. and Artzt, K. (1983) Genet. ICes.41,
29-45
28 Chaflesworth, B. and Hartl, D. (1978) Genetics 89,
171-192
29 Willison, K. et al. (1987) EMBOJ. 6, 1967-1974
30 Silver, L.M. et al. (1992) Mature. Genome 3, $241-$260
31 Forejt,J. and Iv~nyi, P. (1975) Genet. ICes.24, 189-206
32 Pilder, S.H., Hammer, M.F. and Silver, L.M. (1991) Genetics
129, 237-246
33 Ha, H. etal. (1991) Dev. Genet. 12, 318-332
34 Bonhomme, F. and Guenet, J-L. (1989) in Genetic Variants
and Strains of the laboratory Mouse (Lyon, M.F. and
Seade, A.G., eds), pp. 649--662, Oxford University Press
35 Sage, R.D. (1986) Nature 324, 60-63
36 Auffray,J-C., Marshall, J.T., Thaler, L. and Bonhomme, F.
(1991) Mouse Genome 88, 7-8
1
L.M. SILVER IS 1N THE DEP&WFMENT OF MOLECULAR
PRINCETONUNIVERSTFg,PRINCETON,NJ 08544-1014, U&L
TIGJULY1993 VOL.9 NO. 7
/
BIOLOGY, I
I