683
Development 109, 683-690 (1990)
Printed in Great Britain © T h e Company of Biologists Limited 1990
Effects of the lethal yellow (A*) mutation in mouse aggregation chimeras
GREGORY S. BARSH*, MICHAEL LOVETT and CHARLES J. EPSTEIN
Department of Pediatrics and Department of Biochemistry and Biophysics, University of California, San Francisco, California, 94143, USA
•Present address: Department of Pediatrics, Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford,
California 94305, USA
Summary
The Ay allele is a recessive lethal mutation at the mouse
agouti locus, which results in embryonic death around
the time of implantation. In the heterozygous state, Ay
produces several dominant pleiotropic effects, including
an increase in weight gain and body length, a susceptibility to hepatic, pulmonary and mammary tumors, and a
suppression of the agouti phenotype, which results in a
yellow coat color. To investigate the cellular action of Av
with regard to its effects upon embryonic viability and
adult-onset obesity, we generated a series of aggregation
chimeras using embryos that differ in their agouti locus
genotype. Embryos derived from Ay/axAy/a matings
were aggregated with those derived from A/Ax A/A
matings, and genotypic identification of the resultant
chimeras was accomplished using a molecular probe at
the Emv-15 locus that distinguishes among the three
different alleles, Ay, A, and a. Among 50 chimeras, 25
analyzed as liveborns and 25 as 9.5 day embryos, 29
were a/a*-+A/A and 21 were Ay/a+->A/A. The absence of
Ay/Ay<r+A/A chimeras demonstrates that Ay/Ay cells
cannot be rescued in a chimeric environment, and the
relative deficiency of Ay/a*-+A/A chimeras suggests that,
under certain conditions, Ay heterozygosity may partially affect cell viability or proliferation. In the 25
liveborn chimeras, Ayja*^AJA animals became obese as
adults and a/a+-*A/A animals did not. There was no
correlation between genotypic proportions and rate of
weight gain, which shows that, with regard to its effects
on weight gain, Ay heterozygosity is cell non-autonomous.
Introduction
and chemically induced mammary gland, pulmonary,
hepatic and skin tumors (Deringer, 1970; Heston, 1942;
Heston and Deringer, 1947; Heston and Vlahakis, 1961;
Vlahakis and Heston, 1963), premature infertility
(Granholm and Brock, 1981; Granholm et al. 1986) and
glucose intolerance (Carpenter and Mayer, 1958; Hellerstrom and Hellman, 1963). In cases where tested, the
latter effects, like those of Ay on coat color, are
dominant over the effects of all other agouti locus
alleles (Silvers, 1979).
Embryos homozygous for Ay do not survive much
beyond implantation, but the exact time and site of
action of J4y-mediated recessive lethality is controversial. A multitude of morphologic abnormalities have
been described in embryos derived from matings between Ay heterozygotes (Brock and Granholm, 1979;
Calarco and Pederson, 1976; Cizadlo and Granholm,
1978a; Cizadlo and Granholm, 1978b; Eaton and
Green, 1962; Eaton and Green, 1963; Granholm and
Johnson, 1978; Pedersen, 1974), but without an independent marker, genotypic identification of AyJAy
homozygotes has always been presumptive. In addition,
only one study has formally tested if abnormalities in a
particular cell lineage or tissue are responsible for
subsequent embryonic death. Papaioannou and
Mutations that interfere with embryonic viability are
important tools for understanding developmental
mechanisms. In mammals, however, mutagenesis is
relatively inefficient and large numbers of embryos
cannot be screened expeditiously. Instead, the assumption is often made that mutations that affect different
types of tissue identify genes that control fundamental
aspects of embryonic development. Consequently,
much attention has been focused on existing recessive
lethal mutations which also exhibit pleiotropism.
The lethal yellow allele (Ay) at the mouse agouti
locus, one of the longest recognized and most intensively studied recessive lethal mutations (Castle and
Little, 1910; Cuenot, 1908; Silvers, 1979), is associated
with a multitude of pleiotropic effects. In the heterozygous state, Ay alters the normal back and forth switch
between eumelanin and phaeomelanin granule formation by follicular melanocytes so that only phaeomelanin granules are formed, and mice that carry Ay (with
wild-type alleles at the albino and brown loci) are
entirely yellow. In addition, heterozygosity for Ay
produces increased somatic growth and obesity (Castle,
1941; Danforth, 1927), a predisposition to spontaneous
Key words: agouti locus, chimera, mouse, obesity.
684
G. S. Barsh, M. Lovett and C. J. Epstein
Gardner (1979) performed a numerical analysis of
reciprocal blastocyst injection experiments in which
either the inner cell mass component or the host
blastocyst component was expected to contain Ay/Ay
homozygotes 25 % of the time. The results suggested
that an Ay/Ay inner cell mass could be rescued by a
control blastocyst but not vice versa, pointing to a cellautonomous defect of the trophectoderm.
These and similar experiments with recessive lethal
mutations have been complicated by an inability to
distinguish cells derived from homozygous embryos. To
overcome these difficulties and to understand better the
role of recessive lethal agouti mutations in normal
development, we have used molecular markers that
identify different agouti alleles to analyze aggregation
chimeras between embryos of different agouti genotypes. Restriction fragment length polymorphisms
(RFLPs) at the Emv-15 locus, closely linked to agouti
(Copeland et al. 1983; Lovett et at. 1987; Siracusa et al.
19876), distinguish between cells that contain zero, one
or two doses of the Ay mutation and have allowed us to
compare the genotypic proportions in each chimera
with two components of the pleiotropic phenotype embryonic viability and increased weight gain. We now
report that recessive lethal effects of Ay/Ay homozygosity cannot be rescued in a chimeric environment and
that adult-onset obesity in Ay heterozygotes is a non
cell-autonomous process.
Ay, a, or A, after digestion with Hindlll. This probe detects a
2.6 kb Hindlll fragment from the A allele, a 3.8 kb HindUl
fragment from the Ay allele, and a 5.4 kb Hindlll from the a
allele. Therefore, AyJAy<^AlA chimeras will contain the
2.6 kb and 3.8 kb fragments, a/a<-+A/A chimeras will contain
the 2.6kb and 5.4kb fragments, and Ay/a*-*A/A chimeras
will contain all three fragments, 2.6kb, 3.8kb, and 5.4kb,
with the latter two present in equal abundance. The genetic
distance between Emv-15 and Ay is <0.2cM (upper 95%
confidence limit; Siracusa et al. 1987a), and the genetic
distance between Emv-15 and a is 0.6±0.4cM (Siracusa et al.
1989). Therefore, although there is some question whether
the Ay mutation is located at precisely the same genetic map
position as A and a (Siracusa et al. 1987a; Wallace, 1965), the
likelihood of incorrectly identifying a particular agouti allele
due to recombination between the agouti locus and Emv-15 is
extremely small.
DNA from embryos or from tail, brain, liver and kidney
samples of liveborn animals was digested with Hindlll,
subjected to electrophoresis through 1% agarose, and transfered to a nylon membrane by capillary action. The P0.5
probe was radiolabeled by random priming and hybridization
in the presence of 10% dextran sulphate was performed
according to standard procedures. For the 9.5 day embryos,
50% of the entire sample was used per lane. After an 8 day
exposure, 55 of 69 embryos produced a detectable signal, but
in 29 cases, chimeric genotype could not be established
unequivocally (see RESULTS), because of a low recovery of
DNA. For the samples from liveborn chimeras, 5-10 micrograms of DNA was used per lane. Chimeric genotypes were
readily apparent in every case after a 1-2 day autoradiographic exposure, and densitometric measurement of each
allele was used to estimate directly zygotic proportions.
Materials and methods
Mouse strains and mutations
The strains, C57BL/6J-/l>r and BALB/cJ, of agouti genotypes
Ay/a and A/A, respectively, were obtained from The Jackson
Laboratory, Bar Harbor, Maine. Random bred CD-I females
(Charles River Laboratories) were used as uterine foster
mothers.
Embryo manipulation and production of aggregation
chimeras
After superovulation, embryos from Ay/axAy/a crosses were
harvested at 1.5 days post coitum at the 2- to 4-cell stage and
placed in microdrop culture for 24 h, after which most
embryos had undergone a single round of division. 4- to 8-cellstage embryos from A/Ax A/A crosses were then harvested
at 2.5 days post coitum and aggregated with stage-matched
embryos from the Ay/axAy/a cross. (This protocol allowed
the most efficient generation of embryos matched for cleavage
stage, due to interstrain variability in the rate of in vitro
preimplantation development.) Embryo recovery, dissociation, aggregation and culture were all as described
previously (Cox et al. 1984).
Identification of chimera genotypes and determination
of genotypic proportions
Insertion of the Emv-15 provirus is very closely associated
with Ay (Siracusa et al. 1987a). Furthermore, two alternative
haplotypes defined by RFLPs in genomic DNA surrounding
the insertion site are closely associated with the a or A alleles
(Lovett et al. 1987; Siracusa et al. 19876). A probe immediately 5' to the genomic insertion site of Emv-15, P0.5
(Fig. 1A), can therefore distinguish between all three alleles,
Results
Generation and identification of adult chimeras
Animals heterozygous for the lethal yellow and nonagouti alleles were intercrossed to produce a population
expected to contain 25 % a/a homozygotes, 50 % Ay/a
heterozygotes, and 25 % Ay/Ay homozygotes. Individual embryos from this group were aggregated with
developmentally matched BALB/cJ (A/A) embryos.
In the first set of experiments, 25 liveborn animals were
obtained from 89 chimeric embryos transferred into
pseudopregnant mothers. Genomic DNA was recovered from a segment of each animal's tail, and a
Southen blot was prepared after digestion with Hindlll.
Hybridization with the P0.5 probe produced the pattern
of fragments expected for Ay/a++A/A chimeras in 11
animals and the pattern of fragments expected for
a/a*-+A/A chimeras in 14 animals (Table 1). No animals produced the pattern expected for Ay/Ay++A/A
chimeras (X2=16.2, ^=0.003 for difference from a
1:2:1 distribution).
We compared the coat color of each animal to its
genotypic composition as determined by analysis of the
Emv-15 RFLPs. Because the C57BL/6J-/F and
BALB/cJ strain carry different albino alleles (C57BL/
63-Ay is C/C and BALB/cJ is c/c), as well as different
agouti locus alleles, it was necessary to consider the
cellular action of both genes to interpret properly the
resultant coat color patterns. The agouti locus acts
Avla
JI
AVla
4
AIA x AIA
4
AVla -AlA
ala -AIA
A
AYIAY-AIA
i
a
P0.5 probe
*
1 Hhdlll
1 absent in a
Fig.. 1. Generation and identification of lethal yellow and non-agouti chimeras. (A) Breeding strategy, possible outcomes
and map of the Emv-15 locus as described in the text. The asterisk indicates a polymorphic HindIII site present on
chromosomes that carry the AYand A alleles, but not present on chromosomes that carry the a allele. (B) An autoradiogram
of tail DNA from a series of liveborn chimeras after digestion with HindIII and hybridization to the P0.5 probe. The first
two lanes are A/A and a/a control samples as indicated. The 3.8kb fragment, seen exclusively in DNA from the Ahallele,
contains approximately 450bp of cellular DNA joined to 3.35 kb of proviral DNA from Emv-15. DNA from the A and a
alleles does not contain Emv-15, and the 2.6 kb and 5.4 kb fragments result from the presence or absence, respectively, of a
polymorphic HindIII site. (C) Appearance of two chimeric mice. Coat color patches correspond to melanocyte rather than
hair follicle clones.
Cellular action of the lethal yellow (Ay) mutation
Table 1. Generation of lethal yellow and non-agouti
chimeras
Experiment I.
Adult chimeras
No. of aggregates transferred"
No. recovered
No. genotyped
Chimera genotype
Ay/a++A/A
Ay/a
a/a+-*A/A
a/a
Ay/Ay++A/A
Ay/Ay
A A
Experiment II.
9-day embryo
chimeras
89
25
25
157
69
11
0
14
0
0
0
0
9
1
13
2
0
0
29
54*
"Excluding transfers to females that failed to become pregnant.
b
Fifteen of the embryos failed to produce a detectable
hybridization signal, but, as a group, did not exhibit any readily
apparent morphologic differences from the remainder of the 9.5
day embryos.
within cells of the dermal papilla to influence the
behavior of overlying follicular melanocytes, but a
minimum level of tyrosinase activity, controlled by the c
locus, is required to generate phaeomelanin or eumelanin (Silvers, 1961). In addition, the patch size of
melanocyte clones is larger than the patch size of hair
follicle clones. Because the albino (c) mutation abolishes tyrosinase activity, we expected that any patch of
melanocytes that had originated from a BALB/cJ
{A c/A c) progenitor would appear albino, regardless of
the agouti locus genotype (- C/- C) of the underlying
dermal papillae cells. Likewise, some hair follicles that
contained C57BL/6J-/t-v melanocytes (-C/-C) would
be expected to produce agouti hairs if the underlying
dermal papilla had originated from a BALB/cJ (A cj
A c) progenitor.
In accord with these predictions, the observed coat
color patterns included patches that were phenotypically albino, yellow or non-agouti, although, as
expected, the latter two types were never seen in a
single animal (Fig. 1C). In contrast, there were no large
patches of agouti. However, close inspection revealed
frequent but scattered agouti hairs in most animals.
Animals that contained any yellow patches were scored
as Ay/a++A/A, and animals that contained any nonagouti patches were scored as a/a<-*A/A. Six animals
were entirely albino and therefore could not be identified on the basis of coat color alone, but in the
remainder, genotypic identification inferred from Emv15 RFLP analysis agreed with that predicted by coat
color.
Generation and identification of 9.5 day embryo
chimeras
Failure to obtain the Ay/Ay++A/A genotype in any of
the 25 adult animals suggested that this class of chimeras was not surviving to term. To characterize further
the developmental potential of Ay/Ay cells, we generated a second set of chimeras for analysis midway
685
through gestation. As before, individual 4- to 8-cell
embryos from an intercross between Ay/a heterozygotes were aggregated with A/A embryos at a similar
stage. Chimeric embryos were recovered at 9.5 days
gestation, and constituent genotypes were determined
by analysis of the Emv-15 RFLPs. Each embryo and
attached yolk sac were dissected free of maternal
decidua, but, because we wished to recover as much
DNA as possible, no attempt was made to separate
embryonic and extraembryonic components. In the 69
embryos recovered after 157 transfers, there was a
spectrum of developmental stages from early neural
fold with 5-10 visible somites to early limb bud with
20-30 somites. Enough DNA was recovered to generate a hybridization signal in 54 cases, but in 29 of these
the only visible DNA fragment was 2.6 kb, indicative of
the A/A genotype. Of the remaining 25 embryos, 10
were Ay/a*^A/A or Ay/a and 15 were a/a*^A/A or a/a
(Table 1).
Distribution of chimeric genotypes
As with the adult animals, none of the 9.5 day embryos
generated the pattern expected for Ay/Ay<r->A/A chimeras. Furthermore, in both the adults and the embryos,
the observed ratio of Ay/a*+A/A to a/a+^A/A chimeras was distorted relative to the Mendelian expectation
of 2:1 (P=0.0001 assuming non-viability of Ay/Ay++A/
A chimeras) (Table 1). Because the amounts of embryo
DNA are limiting, we estimate that the minimum
genotypic contribution that can be detected in the 9.5
day embryos is 20%. Therefore, the existence of three
groups of embryos with an apparently single genotypic
contribution (1 Ay/a, 2 a/a, and 29 A/A embryos) can
be explained in two ways: either they represent aggregates that failed to form chimeras at the morula stage,
or they are actually chimeras in which the contribution
of one of the genotypes is too small for detection. Both
explanations could contribute to the altered ratio of
Ay/a*r+A/A to a/a<r*A/A chimeras.
To address the possibility that Ay/a cells, in chimeric
combinations with A/A, might exhibit a generalized
disadvantage relative to a/a cells, we examined the
frequency distribution of genotypic proportions in tail
tissue from the two types of adult chimeras. Measuring
the relative intensities of the 2.6 kb, 3.8 kb, and 5.4 kb
Hindlll fragments allowed us to estimate the proportion of Ay/a or a/a cells in the tail of each Ay/a*-+A/
A or a/a<r+A/A chimera. (In the samples from the adult
chimeras, enough DNA could be applied to a single
well to detect a chimeric contribution of 5 % -10 %, but
due to limiting amounts of DNA, a similar analysis
could not be applied to the 9.5 day embryo chimeras.)
Although small differences would not have been
detected without applying more sensitive techniques to
a larger population of animals, the frequency distributions of the proportions of Ay/a or a/a cells do not
show gross differences between the two types of chimera (Fig. 2). Although these results do not exclude the
possibility that Ay/a cells are less effective than a/a cells
in contributing to the inner cell mass following embryo
aggregation, they do suggest that any possible disadvan-
686
G. S. Barsh, M. Lovett and C. J. Epstein
40
60
BO
100
% a/a or AVa cells
A/A
A/A
Fig. 2. Genotypic proportions in a/a<^*A/A and Ay/a<^>A/
A chimeras. In the 25 liveborn chimeras, the proportion of
a/a or Ay/a cells in a tail sample was estimated by
comparing the densitometric signal(s) from the 5.4kb
fragment or the 5.4kb and 3.8kb fragments to the signal
from the 2.6 kb fragment. The minimum genotypic
contribution detectable in these experiments is 5 %, which
represents 0.5 micrograms of genomic DNA. Each symbol
represents a single measurement from one animal.
80
• a/a
•A*/ a
A/A
A/A
Fig. 3. Growth curves of a/a<-+A/A and Ay/a<-+A/A
chimeras. Animals were fed a standard diet with a fat
content of 6.5 % and were weighed weekly. Each curve
represents a single animal.
tage of Ay/a cells relative to a/a cells does not affect cell
proliferation or viability in a generalized fashion, i.e at
all stages of pre- and post-natal life.
Adult-onset obesity in liveborn chimeras
Increased somatic growth and adult onset obesity are
two of the most visible pleiotropic effects associated
with the Ay allele. A weight difference between yellow
and non-yellow littermates first becomes apparent at
4-6 weeks of age and increases, more so in females than
males, over the next 6-12 months. Central to understanding the underlying physiologic mechanism is
whether increased lipid deposition in Ay/a adipocytes is
caused by expression of the /Fgene within the adipocytes or within the environment in which they reside,
i.e. whether /^'-mediated obesity is cell-autonomous.
The populations of a/a<-+A/A and Ay /a<^AJA animals
with different degrees of chimerism provided the opportunity to investigate this question. We expected
that, if Av-mediated obesity were cell-autonomous, (1)
Ay/a*-+A/A animals with a low proportion of Ay/a cells
would exhibit a weight gain similar to a/a<-*A/A
animals; (2) Ay /a^A/ A animals with a high proportion
of Ay/a cells would exhibit a rate of weight gain similar
to Ay/a animals; and (3) the proportion of Ay/a cells in
Ay/a+-*A/A chimeras would be related linearly to rate
of weight gain.
Growth records indicated that, by 6 weeks of age, the
range of weights was significantly different between the
two types of adult chimeras (20.2±1.36g for a/a<r->A/A
animals; 26.2±2.75g for Ay/a<^A/A
animals;
P=0.0001). This difference became increasingly apparent over the next 12-14 weeks (Fig. 3). We calculated
the rate of weight gain in each animal as a linear
regression between 10 and 20 weeks of age and compared the frequency distributions of the two types of
chimeras (Fig. 4). Rates of weight gain were similar
100
Age (days)
1
£.
i
I
I
I
i
I
female
I
I OOO
I
1
i
i
1
i
male
o
0
•3
(
O
• I
I
.05
.10
15
20
25
30
Growth rate (g/day; 10-20 weeks)
# a/a -|—• A/A
QAy/a -*—- A/A
Fig. 4. Rates of weight gain of a/a<-*A/ A and Ay/a++A/A
chimeras. The plotted value represents a linear regression
of weight vs age between 10 and 20 weeks. Each symbol
represents a single animal.
between female and male Ay/a++A/A chimeras, but
the difference relative to a/a+*A/A chimeras was
greater in females then in males. Although the proportion of Ay/a cells in tails from Ay/a<-+A/A animals
varied from 20% to 90% (Fig. 2), the distribution of
rates of weight gain did not show a similar degree of
variation. Instead, the distributions in both types of
chimeras exhibited a similar degree of variation, which
suggested that rates of weight gain in Ay/a<r+A/A
animals with a low proportion of Ay/a cells were
indistinguishable from those with a high proportion of
Ay/a cells.
To address this question more directly, we compared
genotypic proportion to rate of weight gain in both
types of chimeras. Although we could not directly
Cellular action of the lethal yellow (Ay) mutation
20
30
40
60
50
%a/a or A'/a cells
70
80
O«
90
100
tall
D • brain
A A kidney
<>• liver
Fig. 5. Relation of rates of weight gain to genotypic
proportions in a/a*+A/A and Ay/a*-*A/A chimeras. Each
symbol represents a single measurement from tissue, tail,
brain, kidney, or liver DNA from an adult chimera,
estimated densitometrically as in Fig. 2. The symbols are
displaced along the abscissa by 2-5 percentile units in cases
where multiple values overlapped each other.
measure genotypic proportions in the isolated adipose
tissue compartment, we examined liver, kidney and
brain DNA from most of the chimeric animals in
addition to the tail DNA sampled previously. We found
no correlation between genotypic proportion and rate
of weight gain for either Ay/a<r+A/A or aja*^AJA
chimeras, regardless of the tissue examined (Fig. 5).
Assuming that genotypic proportions in fat do not vary
systematically from those in solid organs, a minimal
proportion of Ay/a cells in Ay/a<-*A/A animals is
sufficient to produce a maximal effect on rate of weight
gain, and therefore, adult-onset obesity associated with
Ay appears to be mediated by a cell non-autonomous
mechanism.
Discussion
Genetic mosaics and chimeras are valuable tools for
studying the relationship of cellular genotype to a
particular developmental phenotype. In the mouse, the
timing of X inactivation, primary sex determination,
and the regulation of the switch from alphafetoprotein
to albumin are most directly observed in the whole
animal, and in each case the analysis of aggregation or
blastocyst injection chimeras has played a central role in
understanding the underlying physiologic mechanisms
(Burgoyne et al. 1988; Gardner et al. 1985; Vogt et al.
1987).
We have applied the analysis of aggregation chimeras
to the study of the pleiotropic effects of the lethal
yellow allele. In 25 liveborn chimeras, none contained a
component from Ay/Ay homozygotes, which demonstrates that with regard to fetal viability, the effects of
the Ay gene cannot be rescued in a chimeric environ-
687
ment. Neither was Ay/Ay tissue detectable in a series of
25 embryonic chimeras, which suggests that the effects
of Ay on cell viability are manifest by 9.5 days of
gestation. Finally, measurements of weight gain and
comparison of these measurements to genotypic proportions in Ay/a<r->A/A chimeras suggests that the
effects of Ay on rate of weight gain are mediated by a
cell non-autonomous mechanism.
A variety of morphologic abnormalities have been
described in embryos derived from heterozygous Ay
intercrosses, including an increased frequency of
excluded blastomeres and ultrastructural abnormalities
in cleavage stage embryos (Calarco and Pederson, 1976;
Pedersen, 1974), reduced growth and development of
inner cell mass and trophectoderm cultures in vitro
(Papaioannou, 1988), decreased inner cell mass number
and delayed hatching in peri-implantation blastocysts
(Cizadlo and Granholm, 1978a; Granholm and Johnson, 1978), and reduced trophectoderm differentiation
and cell death in post-implantation blastocysts (Eaton
and Green, 1962; Eaton and Green, 1963). In one of the
few studies that directly addressed tissue specificity of
Ay homozygous lethality, inner cell mass cells derived
from Ay/acxAy/ac matings were injected into normal
(random bred) blastocysts (Papaioannou and Gardner,
1979). The recovery of 10.5 day chimeric embryos was
not significantly different from that observed in a
control experiment in which blastocysts were derived
from Ay/aexac/ac matings. Although these results
suggested that AyJAy cells can survive to at least
midgestation in a normal tissue environment, genotypic
identification of these cells was not then possible, and
we have not detected any Ay/Ay cells in our series of
chimeras. Because Ay/Ay trophectoderm cells are
likely to have been included in Ay/Ay++A/A aggregation chimeras but not in the more defined environment of Ay/Ay++A/A blastocyst injection chimeras,
one explanation for the absence of Ay/Ay cells in our
chimeras is that even a small number of these cells in
the trophectoderm is sufficient to prevent post-implantation development. Alternatively, if Ay/Ay cells were
present but too infrequent to be detected in our 9.5 day
embryos, Ay/Ay++A/A chimeras would have contributed to the apparently non-chimeric A/A group. However, all of the liveborn chimeras contained detectable
components from two zygotes, and therefore we expect
that any 9.5 day Ay/Ay*^A/A chimeras would not have
survived until term. At present, we cannot say whether
homozygosity for Ay in inner cell mass, trophectoderm,
or both tissues is responsible for embryonic death, but
an analysis of blastocyst injection chimeras using
RFLPs at the Emv-15 locus should allow this question
to be addressed definitively.
The number of chimeras that contained a component
from Ay/a embryos was approximately 50 % less than
expected, given that Ay heterozygosity has no effect on
viability in standard breeding experiments. One explanation is that Ay heterozygosity affects cell proliferation
or viability at specific periods of embryogenesis, and
that a proportion of chimeric embryos die due to a
developmental mismatch between the Ay and non-/4r
688
G. S. Barsh, M. Lovett and C. J. Epstein
components of the chimera. This effect might well be
masked in a normal (non-chimeric) embryo, in which
there is no developmental mismatch and a tremendous
regulative capacity in response to cell death during
early development (Snow et al. 1981). Such an effect
may have contributed to the relative excess of apparently non-chimeric A/A 9.5 day embryos, although, as
argued above, a putative class of Ay/Ay*-*A/A and
Ay/a*+A/A embryos in which the Ay/Ay or Ay/a
component was non-detectable is unlikely to have
surived until term. Which chimeras exhibit such a
developmental mismatch could depend on the random
contribution of A/A and Ay/a cells to a particular preor peri-implantation lineage with a small number of
founder cells. For example, in chimeras in which the
Ay/a component constitutes the majority of trophectoderm, a developmental mismatch between trophectoderm and inner cell mass might prevent subsequent
growth and differentiation of extraembryonic tissues,
the chimera would appear non-chimeric at 9.5 days
gestation, and would not survive until term. Alternatively, if the Ay/a component of the chimera is, by
chance, excluded from trophectoderm, the embryo
might not be susceptible to the effects of a developmental mismatch and would not manifest altered genotypic
proportions in most adult tissues. This speculative
series of mechanisms constitutes a single explanation
for both the excess of non-chimeric 9.5 day embryos
and the reduced number of Ay/a*-*A/A chimeras
among both 9.5 day embryos and liveborn animals.
However, a larger series of chimeras along with an
extensive analysis of genotypic proportions in different
tissues will be necessary to allow these issues to be
investigated more thoroughly.
Our analysis of rates of weight gain in liveborn
chimeras shows that Ay/a<r->A/A animals with a small
proportion of Ay/a cells in tail, liver, kidney or brain,
have growth characteristics indistinguishable from animals with a large proportion of Ay/a cells. In earlier
studies directed at understanding the mechanism of Aymediated obesity, means other than aggregation chimeras have been used to alter the environment of
heterozygous Ay cells. Experiments in which parabiotic
connections were established between Ay/A and a/a
animals suggested that a circulating substance from the
former animal could lead to increased body length in
the latter (Wolff, 1963). More recently, reciprocal
adipose tissue transplantation experiments between
Ay/a and A/A animals have suggested that fat cell size
and proliferation are a function of cellular environment
rather than cellular genotype (Meade et al. 1979). Thus,
several lines of evidence now point to a cell nonautonomous mechanism underlying ^-induced obesity, but the primary physiologic changes are still unknown. Obesity associated with Ay is not caused by a
decrease in activity or an increase in nutrient intake
(Carpenter and Mayer, 1958; Cizadlo and Granholm,
1976; Fenton and Chase, 1951; Hollifield and Parson,
1957) and is not prevented by pituitary or adrenal
ablation (Jackson et al. 1976; Plocher and Powley,
1976). Although altered thermoregulation has been
described in fat yellow mice as well as in other genetic
obesity syndromes (Cizadlo et al. 1977; Herberg and
Coleman, 1977), these changes are likely to be secondary effects of increased body fat.
Cell-autonomous behavior of a lethal mutation is
often taken to mean that the normal gene controls an
essential intracellular process. Conversely, rescue of a
mutant phenotype may imply that the normal gene
codes for a diffusible product. These conclusions are
valid, however, only in situations in which the mutation
being tested is amorphic or null. For example, given a
particular gene that codes for a secreted protein,
expression of a mutated coding sequence might block
secretion of that protein, leading to progressive intracellular accumulation and eventual cell lysis. In this
case, the mutant cells could not be rescued and cellautonomous death would indicate erroneously that the
normal gene product is part of an essential intracellular
process. If, on the other hand, a different mutation
entirely prevented expression of the secreted protein,
mutant cells would survive if the normal protein was
provided via a chimeric environment. In this case, cell
non-autonomy would indicate correctly that the normal
gene coded for a diffusible product. It is not known if
the effects of Ay on recessive lethality are due to
inactivation, altered expression, or overexpression of a
normal gene. However, we have previously reported
that a 75 kb deletion is closely associated with the
radiation-induced non-agouti lethal allele, a1 (Barsh
and Epstein, 1989), and a is lethal in combination with
Ay (Lyon et al. 1985). To this extent, ,4-v-mediated
recessive lethality may represent a null mutation, and
the normal agouti gene product (or at least the one
affected by Ay and a1) may indeed be part of an essential
intracellular process. In this case, the effect of Ay/Ay or
clja1 cell loss on chimeric embryo survival would
depend on the time of cell death. Prior to implantation,
extensive cell death may be compensated by the regulative capacities of the mammalian embryo (Snow et al.
1981), but after gastrulation, significant cell death is
likely to be incompatible with embryonic survival.
Analysis of aggregation chimeras with a1 and other
radiation-induced recessive lethal agouti mutations is
now possible and will help to resolve these questions.
Mutations induced by radiation frequently affect
multiple transcriptional units, in which case the pleiotropic effects of a particular mutation are more likely to
signify the proximity of one gene to the next, rather
than a single gene that affects multiple types of tissue
(Russell, 1989). There are two agouti alleles with
pleiotropic effects other than recessive lethality, Ay and
Avy (viable yellow), and both arose spontaneously
(Silvers, 1979). Animals heterozygous for Avy exhibit
variable expression of both yellow coat color and
obesity, and variation in expression of one phenotype is
directly correlated with variation in expression of the
other (Wolff and Pilot, 1973; Wolff et al. 1986). Furthermore, like the effect of Ay on obesity and female
infertility (Granholm and Dickens, 1986), suppression
of the phaeomelanin to eumelanin switch is cell nonautonomous (Silvers, 1961), supporting the possibility
Cellular action of the lethal yellow (Ay) mutation 689
that these phenotypes are mediated by the same or
overlapping transcriptional units. However, although
allelomorphic series for the agouti locus exist in most
mammalian orders, Mus musculus is the only species in
which agouti locus mutations are also known to affect
growth and viability (Searle, 1968). In addition, interaction and recombination between agouti locus alleles
have suggested a complex genetic structure for the locus
(Russell etal. 1963; Wallace, 1954; Wallace, 1965). It is
possible, therefore, that different transcriptional units
mediate one or more of the pleiotropic effects associated with Ay. Resolution of these questions will require
isolation of agouti locus transcripts, a goal now clearly
in sight.
We thank Sandra Smith for preparing the adult chimeras,
Elaine Carlson for preparing the 9.5 day embryo chimeras,
Teodosia Zamora for excellent technical assistance, and Dr
Virginia Papaioannou for helpful discussions. This work was
supported by research grant HD-03132 from the National
Institute of Child Health and Human Development. G.S.B.
was supported by a postdoctoral research training grant from
the National Institute of General Medical Sciences (GM07085).
DANFORTH, C. H. (1927). Hereditary adiposity in mice. J. Hered.
18, 153-162.
DERINGER, M. K. (1970). Influence of the lethal yellow (/*-*) gene
on the development of reticular neoplasms. J. natn. Cancer Inst.
45, 1205-1210.
EATON, G. J. AND GREEN, M. M. (1962). Implantation and lethality
of the yellow mouse. Genetica 33, 106-112.
EATON, G. J. AND GREEN, M. M. (1963). Giant cell differentiation
and lethality of homozygous yellow mouse embryos. Genetica 34,
155-161.
FENTON, P. F. AND CHASE, H. B. (1951). Effect of diet on obesity
of yellow mice in inbred lines. Proc. Soc. exp. Biol. med. 77,
420-422.
GARDNER, R. L., LYON, M. L., EVANS, E. P. AND BURTENSHAW, M.
D. (1985). Clonal analysis of X-chromosome inactivation and the
origin of the germ line in the mouse embryo. J. Embryol. exp.
Morph. 88, 349-363.
GRANHOLM, N. H. AND BROCK, K. T. (1981). Effects of the lethal
yellow gene (Ay) on mating preference in C57BL/6J mice. Proc.
S. D. Acad. Sci. 60, 24-28.
GRANHOLM, N. H. AND DICKENS, G. A. (1986). Effects of
reciprocal ovary transplantation on reproductive performance of
lethal yellow mice (Ay/a; C57BL/6J). J. Reprod. Fen. 78,
749-753.
GRANHOLM, N. H., JEPPESEN, K. W. AND JAPS, R. A. (1986).
Progressive infertility in female lethal yellow mice (Ay/a\ strain
C57BL/6J). J. Reprod. Fen. 76, 279-287.
GRANHOLM, N. H. AND JOHNSON, P. M. (1978). Enhanced
identification of lethal yellow (Ay/Ay) mouse embryos by means
of delayed development of four-cell stages. J. exp. Zool. 205,
327-334.
References
HELLERSTROM, C. AND HELLMAN, B. (1963). The islets of
BARSH, G. S. AND EPSTEIN, C. J. (L989). Physical and genetic
characterization of a 75-kilobase deletion asssociated with d', a
recessive lethal allele at the mouse agouti locus. Genetics 121,
811-818.
BROCK, K. T. AND GRANHOLM, N. H. (1979). Histological analyses
of lethal yellow mouse embryos (Ay/Ay) at 132h post coitum.
Anat. Embryol. 157, 161-166.
BURGOYNE, P . S . , BUEHR, M . , KOOPMAN, P . , ROSSANT, J. AND
MCLAREN, A. (1988). Cell-autonomous action of the testisdetermining gene: Sertoli cells are exclusively XY in XX<-*XY
chimaeric mouse testes. Development 102, 443-450.
CALARCO, P. G. AND PEDERSON, R. A. (1976). Ultrastructural
observations of lethal yellow (Ay/Ay) mouse embryos. J.
Embryol. exp. Morph. 35, 73-80.
CARPENTER, K. J. AND MAYER, J. (1958). Physiologic observations
on yellow obesity in the mouse. Am. J. Physiol. 193, 499-504.
CASTLE, W. E. (1941). Influence of certain coat color mutations on
body size in mice, rats, and rabbits. Genetics 26, 177-191.
CASTLE, W. E. AND LITTLE, C. C. (1910). On a modified Mendelian
ratio among yellow mice. Science 32, 868-870.
CIZADLO, G. R. AND GRANHOLM, N. H. (1976). Relationship of
activity and obesity in genetically obese yellow (Ay/a) mice.
Proc. S. D. Acad. Sci. 55, 97-99.
CIZADLO, G. R. AND GRANHOLM, N. H. (1978a). In vivo
development of the lethal yellow (Ay/Ay)
post coitum. Genetica 89-93,
mouse embryo at 105 h
CIZADLO, G. R. AND GRANHOLM, N. H. (1978ft). Ultrastructural
analysis of preimplantation lethal yellow (Ay/Ay)
embryos. J. Embryol. exp. Morph. 45, 13-24.
mouse
CIZADLO, G. R., GRANHOLM, N. H. AND WAGERS, R. C. (1977).
Aberrant body temperature regulation in obese yellow (Ay/a)
mice. Proc. S. D. Acad. Sci. 56, 159-162.
COPELAND, N. G., JENKINS, N. A. AND LEE, B. K. (1983).
Association of the lethal yellow (Ay) coat color mutation with a
ecotropic murine leukemia virus genome. Proc. natn. Acad. Sci.
U.S.A. 80, 247-249.
Cox, D. R., SMITH, S. A., EPSTEIN, L. B. AND EPSTEIN, C. J.
(1984). Mouse trisomy 16 as an animal model of human trisomy
21 (Down syndrome): production of viable trisomy 16<->diploid
mouse chimeras. Devi Biol. 101, 416-424.
CUENOT, L. (1908). Sur quelques anomalies apparentes des
proportions Mendeliennes. Arch. Zool. exp. Gener. 9, 7-15.
Langerhans in yellow obese mice. Metabolism 12, 527-536.
HERBERG, L. AND COLEMAN, D. L. (1977). Laboratory animals
exhibiting obesity and diabetes syndromes. Metabolism 26,
59-99.
HESTON, W. E. (1942). Relationship between the lethal yellow (Ay)
gene of the mouse and susceptibility to induced pulmonary
tumors. J. natn. Cancer Inst. 3, 303-308.
HESTON, W. E. AND DERINGER, M. K. (1947). Relationship between
the lethal yellow (Ay) gene of the mouse and susceptibility to
spontaneous pulmonary tumors. J. natn. Cancer Inst. 7, 463-465.
HESTON, W. E. AND VLAHAKIS, G. (1961). Influence of the Ay gene
on mammary-gland tumors, hepatomas, and normal growth in
mice. J. natn. Cancer Inst. 26, 969-982.
HOLUFIELD, G. AND PARSON, W. (1957). Food drive and satiety in
yellow mice. Am. J. Physiol. 189, 36-38.
JACKSON, E., STOLZ, D. AND MABTIN, R. (1976). Effect of
adrenalectomy on weight gain and body composition of yellow
obese mice (Ay/a). Horm. Metab. Res. 8, 452-455.
LOVETT, M., CHENG, Z. Y., LAMELA, E. M., YOKOI, T. AND
EPSTEIN, C. J. (1987). Molecular markers for the agouti coat
color locus of the mouse. Genetics 115, 747-754.
LYON, M. F., FISHER, G. AND GLENISTER, P. H. (1985). A recessive
allele of the mouse agouti locus showing lethality with yellow,
Ay. Genet. Res. 46, 95-99.
MEADE, C. J., ASHWELL, M. AND SOWTER, C. (1979). Is genetically
transmitted obesity due to an adipose tissue defect? Proc. R.
Soc. Lond. B. 205, 395-410.
PAPAIOANNOU, V. E. (1988). Investigation of the tissue specificity of
the lethal yellow (Ay) gene in mouse embryos. Dev. Genet. 9,
155-165.
PAPAIOANNOU, V. E. AND GARDNER, R. L. (1979). Investigation of
the lethal yellow (Ay/Ay) embryo using mouse chimaeras. J.
Embryol. exp. Morph. 52, 153-163.
PEDERSEN, R. A. (1974). Development of lethal yellow (Ay/Ay)
mouse embryos in vitro. J. exp. Zool. 188, 307-320.
PLOCHER, T. A. AND POWLEY, T. L. (1976). Effect of
hypophysectomy on weight gain and body composition in the
genetically obese yellow (Ay/a) mouse. Metabolism 25, 593-602.
RUSSELL, L. B. (1989). Functional and structural analyses of mouse
genomic regions screened by the morphological specific-locus
test. Mutat. Res. 212, 23-32.
RUSSELL, L. B., MCDANIEL, M. N. C. AND WOODIEL, F. N. (1963).
690
G. S. Barsh, M. Lovett and C. J. Epstein
Crossing over within a "locus" of the mouse (abstr.). Genetics
48,907.
SEARLE, A. G. (1968). Comparative Genetics of Coat Color in
Mammals. New York: Academic Press.
SILVERS, W. K. (1961). Genes and the pigment cells of mammals.
Science 134, 368-373.
SILVERS, W. K. (1979). The agouti and extension series of alleles,
umbrous and sable (ed. W. K. Silvers), pp. 6-28. New York:
Springer-Verlag.
SlRACUSA, L . D . , BUCHBERG, A . , CoPELAND, N . G . AND JENKINS,
N. A. (1989). Recombinant inbred strain and interspecific
backcross analysis of molecular markers flanking the murine
agouti coat color locus. Genetics 122, 669-679.
SIRACUSA, L. D . , RUSSELL, L. B., EICHER, E. M., CORROW, D. J.,
COPELAND, N. G. AND JENKINS, N. A. (1987a). Genetic
organization of the agouti region of the mouse. Genetics 117,
93-100.
SIRACUSA, L. D., RUSSELL, L. B., JENKINS, N. A. AND COPELAND,
N. G. (19876). Allelic variation within the Emv-15 locus defines
genomic sequences closely linked to the agouti locus on mouse
chromosome 2. Genetics 117, 85-92.
embryos. In S. Subtelny (ed.), Levels of Genetic Control iii
Development. Cambridge: Cambridge University Press.
VLAHAKIS, G. AND HESTON, W. E. (1963). Increase of induced skin
tumors in the mouse by the lethal yellow gene (Ay). J. natn.
Cancer Inst. 31, 189-195.
VOCT, T. F., SOLTER, D. AND TILGHMAN, S. M. (1987). Raf, a trans-
acting locus, regulates the alpha-fetoprotein gene in a cellautonomous manner. Science 236, 301-303.
WALLACE, M. E. (1954). A mutation or crossover in the house
mouse? Heredity 8, 89-105.
WALLACE, M. E. (1965). Pseudoallelism at the agouti locus in the
mouse. J. Hered. 56,267-271.
WOLFF, G. L. (1963). Growth of inbred yellow (Aya) and nonyellow (ofl) mice in parabiosis. Genetics 48, 1041-1058.
WOLFF, G. L. AND PTTOT, H. C. (1973). Influence of background
Avy/-)
genome on enzymatic characteristics of yellow (Ay/-,
mice. Genetics 73, 109-123.
WOLFF, G. L., ROBERTS, D. W. AND GALBRAITH, D. B. (1986).
Prenatal determination of obesity, tumor susceptibility, and coat
color pattern in viable yellow (A'y/a) mice. The yellow mouse
syndrome. J. Heredity 77, 151-158.
SNOW, M. H. L., TAM, P. P. L. AND MCLAREN, A. (1981). On the
control and regulation of size and morphogenesis in mammalian
(Accepted 22 March, 1990)