.I. Genet"
Vot. 70, No. 3, December 1991, pp. 137-146. 9 P,inted in India.
How do heterogametic females survive without gene dosage
9 compensation?
H SHARAT CHANDRA
Department of Microbiology and Celt Biology, Indian Institute of Science,
Bangalore 560 012, and
Centre for Cellular and Molecular Biology, Hyderabad 500 007, India
MS received 22 March 1991
Abstract.
When tbe male is the heterogametic sex (XX~-XYc? or XX~-XOc~), as in
Drosophila,
orthopteran insects, mammals and
Caenorhabditis elegans,
X-linked genes are
subject to dosage compensation: the single X in the mate is functionally equivalent to the
two Xs in the female. However, when the female is heterogametic (ZZd-ZW~?), as in birds,
butterflies and moths, Z-linked genes are apparently not dosage-compensated. This
difference between X-linked and Z-linked genes raises fundamental questions about the
role of dosage compensation. It is argued that (i) genes which require dosage compensation
are primarily those that control morphogenesis and the prospective body plan; (ii) the
products of these genes are required in disomic ~toses especially during oogenesis and early
embryonic developmentl (iii) heterogametic females synthesize and store during oogenesis
itself morphogenetically essential gene products - including those encoded by Z-linked
genes - in large quantities; (iv) the abundance of these gene products in the egg and their
persistence relatively late into embryogenesis enables heterogametic females to overcome
the monosomic state of the Z chromosome in ZW embryos. Female heterogamety is
predominant in birds, reptiles and amphibians, all of which have megalecithal eggs
containing several thousand times more maternal RNA and other maternal messages than
eggs of mammals,
Caenorl~abditis elegans,
or
Drosophila.
This increase in egg size, yolk
content and, concomitantly, the size of the maternal legacy to the embryo, may have
facilitated female heterogamety and the absence of dosage compensation.
Keywords.
Z chromosomes; dominant maternal genes; megalecithal eggs; maternal
mRNA.
1. Introduction
Dosage compensation is a classic example of the precision of genetic adaptation
(Muller 1948). In
Drosophila melanogastei~,
in which this concept was first developed,
it is believed that the evolution of dosa~ge compensation has enabled the male to
survive the deleterious effects of the monosomic (single-copy) condition of the X
chromosome. Monosomy for the two other large ,chromosomes - which are
autosomes, and lack such compensatory mechanisms - are lethak However, analysis
of the dose requirements for normal development of different segments of the
Drosophila
genome suggests that the number of dose-sensitive genes is not large.
Heterozygous deletions or duplications of most single loci have little or no obvious
effect on development. In a study of the effects of segmental aneuploidy, it was
estimated that the number of dose-sensitive genes in the
Drosophila
genome is at
least fifty seven (Lindsley
et al.
1972). One locus causes lethality when present in
triplicate in an otherwise diploid genetic background. Whereas subsequent analyses
have revealed additional dose-sensitive genes, it is unlikely that their mtmber would
go beyond a few hundred. Thus, the imbalance observed in monosomies (and
137
138
H. Sharat Chandra
trisomies) is probably attributable to tile cumulative cffects of a number of dose-
sensitive genes rather than to any unknown property of whole chrosomosomes
(John and Miklos 1988).
One result of dosage compensation is that the sexual dimorphism in phenotype,
which might develop if dosage compensation did not occur, is avoided. Clearly,
however, at least some X-linked genes - those concerned with sex determination
would not be subject to dosage compensation (in the usual phenotype sense),
because what such genes presumably require is a dosage effec| rather than
compensation (see. for example. Chandra 19861. Many other X-linked such as
those coding ['or enzymes are subject to dosage compensation even though they
appear to be 'dose-insensitive'; drastic changes m the levels of their products do not
lead to clear-cut fitness effects (Kacser and Burns 1981).
2. Evolution of dosage compensation
An early step in the evolution of heteromorphic sex chromosomes is believed to be
a gross reduction in or elimination of recombination between the "'primitive" X and
Y chromosomes. Suppression of recombination would lead to the isolation of the
male-determining gene or genes in the primitive Y chromosome. Non-functional
alleles resulting from mutations would accmnulate in the Y because, in the absence
of recombination, a functional allele at the corresponding locus on the X would not
be able to crossover to the Y (Charlesworth 19781. In a population of Y
chromosomes, then, there would be some with zero and others with one. two, or
more non-functional alleles. If population size is small, there is a non-zero probability
with which each of these chromosomes would be lost by chance alone. When a Y
chromosome that is free of non-functional mutations is lost, there would be no
mechanism by which another mutation-h'ee (wild-type) Y can be generated because
of the absence of recombination. This principle, referred to as Muller's ratchet,
would repeatedly eliminate that class of Y's with the lowest number of mutations.
As a consequence. Y chromosomes with non-functional genes would accumulate in
the population. Selection would then favour dosage compensation of those X-linked
genes which now exist in single copy in the male. Enhancement of transcription of
X-linked genes would be expected to be nonspecific to some extent (Charlesworth
1978). However, it is not clear whether dosage compensation evolves at the level of.
large chromosomal domains or on a gene-by-gene basis.
Rice (1987) has suggested genetic "hitchhiking" as another mechanism by which
the reduced genetic activity of the Y chromosome could evolve following supression
of recombination. This mechanism is mediated by tile fixation of Y-linked
mutations that are linked to other, beneficial genes. There would be concurrent
evolution of dosage compensation as in Charlesworth's model. Hitchhiking can
reinforce Muller's ratchet as well as operate in conditions in which tile ratchet is
ineffective. An attractive aspect of this model is that it appears to permit either the
evolution of dosage compensation or its absence ("dose tolerance"). In case of dose
tolerance, the two sexes would have the capacity to accommodate ditTerent
concentrations of X- and Z-linked gene products.
Dosage compensation has also been demonstrated in orthopteran insects (XX.-
XO) (Hebbert 1984), tile nematode
Caenorhabditis ele#ans
(Meyer 1988), and in
Absence of dosacje compensation 139
mammals, in which it is a result of inactivation of one of the two X chromosomes in
the female (Lyon 1961).
Dosage compensation may not be necessary when the number and morphology
of chromosomes is tile same in the two sexes. But it appears that dosage
compensation is not an absolute requirement even for organisms with distinctly
heteromorphic sex chromosomes. For example, in birds and butterflies (ZZ d'-ZWg),
a large body of evidence suggests that there is, instead of compensation,
a dosage effect of Z-linked genes (Stehr 1959; Cock 1964; Johnson and
Turner 1979; Baverstock et al. 1982). Genes coding for the enzyme 6-
phosphogh~conate dehydrogenase in butterflies (Johnson and Turner 1979), and
aconitase I in birds (Baverstock, et al. 1982) are Z-linked, and both show a clear
dosage effect. Data on several sex-linked morphological characteristics in birds are
consistent with a dosage effect of Z-linked genes (Cock 1964). Moreover, the Z
chromosome does not show cytological indications of dosage compensation (such
as enlargement of the single X in male Drosophila or X-chromatin in female
mammals). In the moth Choristoneura, genes controlling hemolymph colour are not
dosage compensated (Stehr 1959). Among butterflies, female-limitation of colour
polymorphism, mimicry and certain non-mimetic polymorphisms are thought to
result from an interaction betweeen autosomal genes and uncompensated Z-linked
genes (Stehr 1959: Sheppard 1961: Cock 1964: Grula and Taylor 1980). A large
proportion of the genes controlling female mate-selection behavlour and male
courtship signals are located on the Z chromosome (Grula and Taylor 1980), and
genetic data on these phenotypes are consistent with the view that most, if not the
whole, of the Z chromosome lacks dosage compensation (Johnson and Turner 1979:
Grula and Taylor 1980). Taken together, these results suggest that many
organisms are ['ully viable in spite of a two-fold disparity in the functional levels of
Z-linked genes.
Among insects, dosage compensation occurs among the Orthoptera and Diptera.
but not among the Lepidoptera, and this has led Hebbert (1984) to suggest that
dosage compensation was present in related archeopteran ancestors of these groups
and that it has been secondarily lost in the Lepidoptera. It has been noted that in
species in which dosage compensation has been shown to occur (Drosophila,
orthopteran insects, mammals and Caenorhabditis elegans), the males are
heterogametic, whereas m groups lacking dosage compensation (birds, moths and
butterflies), females are heterogametic (Cock 1964; JohnsoT,, and Turner 1979:
Baverstock et al. 1982). The Z chromosome, like the X, is invariably one of the
larger chromosomes and constitutes a significant part of the haploid genome (Ohno
[9871. In birds tb.e Z chromosonae is of the same size as the X chromosome of
mammals (Ohno 1987). Therefore, the viability of heterogametic females in spite
of an absence of dosage compensation raises questions about the role of dosage
compensation. This problem has acquired additional interest in the light of recent
results on butterflies which suggest that the absence of dosage compensation also
may be the result of a precise adaptation (Stehr 1959: Grula and Taylor 1980). In
the remainder of this paper I consider a possible explanation for the viability of
heterogametic females in spite of an absence of dosage compensation. Our starting
point is the observation, first made by Lyon (1974, page 259), that there is an
apparent difference m egg size, egg content and timing of zygotic gene activity
between male-heterogametic and felnale-heterogametic organisms.
140 H. Sha','at Chandra
3. Egg size and content in relation to female heterogamety
Fmnale heterogamety occurs in a majority of amphibian species, including Xenopus
laevis (Bull 1983; Schmid 1983; Ohn0 1987) [although heter0morphic sex
chromosomes have apparently been observed in the females of. only one strain of
this species (Ohno 1987)1. X. laevis is the best characterized amphibian in terms ot"
its oogcnesis and early development and can be considered as representative of
amphibians generally. The Xenopus egg is about 4000 times the size of the mouse or
human egg and contains roughly 10,000 times as much stored mRNA (Wilkins
1986; Tara 1986). The same relationship is thought to apply to proteins, lipids and
other egg constituents. The C. elegans egg (40 tml x 50 ~/m) is even smaller than the
mouse egg and it may be the smallest egg so far studied. All birds studied to date
show female heterogamety, and the so-called yolk in the avian egg represents a
giant oocyte whose diameter can reach a few centimetres. However, irrespective of
whether the organism is oviparous or viviparous, vertebrate or invertebrate, the
amount of RNA per unit mass of the mature egg, as well as the complexity of this
RNA, are constant (Wilkins 1986; Tata 1986).
3.1 Sex chromosome monosomy and oogenesis
An unexpected result of molecular studies of embryogenesis in Drosophila using
cDNA libraries is that over 80% of the genes expressed as late as in the gastrula are
also expressed in the oocyte, suggesting that a significant fraction of the maiernal
genome is active in oogenesis (Wilkins 1986; Davidson 1986). Such results are
consistent with evidence obtained from embryos as diverse as those of Drosophila,
Caenorhabditis, sea urchin, Xenopt~s and mannnals, that there is a quantitative
dominance of maternal transcripts in the zygote and during the. immediate
postzygotic development (Wilkins 1986; Davidson 1986). Maternal control of some
early embryonic events may be obligatory because norlnal cmbryogenesis appears
tO require a diploid dosage of a number of maternal genes during the preceding
oogenesis (Wilkins 1986). A function of reactivation of the inacti{,e X in oogonia of
human females is thought to be restoration, in effect, of a disomic dosage of X-
linked genes in the gerln line (Gartler et al. i975). When the XX constitution is
female, then this apparent need for diploid dosage of maternal genes is satisfied. On
the other hand, at the time of evolution of ZW females the seemingly inadequate
gene dosage in the female germline has to be overcome. There are essentially three
ways in which this could have come about: increased synthesis and storage of the
products of maternally acting genes; a shift in the timing of action of zygotic genes
such that they became, in effect, "maternal" genes; and, by a similar shift in the time
of action of genes .a.vhich are both maternal and zygotic, such that they became
exclusively maternal genes. In other words, if the pattern of oogenesis, egg size,
amount of stored maternal mRNA in the egg, and the time of onset of zygotic gene
expression differ sufficiently between representative organisms with female
heterogamety and those with male heterogamety, such differences might account for
the survival without dosage compensatiou of the former. If such increased storage
of maternal products and shifts in timing of action of developmentally significant
genes have in fact occurred, it may be possible to detect such instances by
appropriate comparisons between male-heterogametic and fcmale-heterogametic
systems.
Absence of dosage compensation 141
4. Timing of zygotic gene activity
If the timing of zygotic gene function depends on the extent of stored maternal
messages in the egg, one might expect to see a delayed onset of zygotic activity in
Xenopus relative to that in D. melanocdaster, C. ele,qans or mammals. This appears to
be the case. For instance, in the oocytes of Xenopus and other amphibia, large
quantities of ribosomes are formed by amplification of ribosomal genes and these
maternal ribosomes are stored and used until gastrulation when embryonic cells -
then numbering about 10,000- begin transcription (Tara 1986). In contrast, mouse
and human eggs store very few ribsomes and transcription begins in the mouse
embryo as early as in the 2- to 4- cell stage (Flach et al. 1982) and in the human
embryo, between the 4- to 8- cell stages (Braude et al. 1988). Moreover, in mouse,
maternal mRNA is actively degraded at the 2-cell stage and by early blastocyst
stage the mRNA pool is virtually all zygotic in origin (Davidson 1986), suggesting
that embryogenesis need be under the control of stored maternal messages only
until then. In neither mammalian species is there evidence of amplification of
ribosomal or other genes (Tata 1986). In D. melanogaster zygotic gene expression
begins at the blastoderm stage; in C. elegans the latest period of onset of expression
of the zygotic genome is 90- to 125-ce!1 stage (Wilkins 1986). Mouse einbryos treated
with actinomycin-D do not develop beyond first cleavage whereas Xenopus
embryos similarly treated (or enucleated) continue to develop until the 4000-cell
blastula stage (Tata 1986; Davidson 1986).
Among the approximately 50 genes determining metameric pattern in
Drosophila, 6 function both maternally and zygotically (Akam 1987; Nusslein-
Volhard et al. 1987). For a few others, such as hopscotch and dissheveled, either
maternal or zygotic function is sufficient for normal segmentation to occur
(Nusslein-Volhard et al. 1987). If a Drosophila pattern gene acts zygotically or later
in development, one might expect the Xenopus 'homolog' to be maternally acting.
Such an expectation seems to be met in the case of XIH box 2, a maternally
expressed homeobox-containing locus in X. laevis which shares sequence homology
with caudal, a Drosophila pattern gene (Wright et al. 1987). XIH box 2 and caudal
are among the first such genes for which comparative data are available. Whereas
caudal is expressed maternally as well as during embryogenesis, XIH box 2 is
expressed ahnost entirely in stage 2 and 3 oocytes when many gene products are
stockpiled in the oocyte for utilization during early development (Wright et al.
1987).
5. Dose-sensitivity of genes during early development
Genes performing 'housekeeping' functions would be required to act throughout the
life of an organism, and many such genes appear to be dose-insensitive, or largely
so. For example, the copy number of genes coding for enzymes is not a significant
factor in the control of metabolic flux (Kacser and Burns 1981). Similarly,
heterozygotes for hnman metabolic errors are not readily distinguishable from
normal homozygotes because of a large overlap in enzyme levels. Such experience,
and the results of studies on the evolution Of dominance, have led to the view that
because of the structure of enzyme networks and the interactions among enzymes,
their substrates, and products, "a 50% reduction in activity, a common feature for
many mutants, is not detectable in the phenotype" (Kacser and Burns 198l).