Genetic Testing of the Hypothesis That Hybrid Male Lethality Results From a Failure in Dosage Compensation

Abstract

Several recent studies have suggested that F₁ hybrid male lethality in crosses between Drosophila melanogaster and D. simulans is due to a failure in dosage compensation, caused by incompatibilities between D. simulans dosage compensation proteins and the D. melanogaster X chromosome. Contrary to the predictions of this hypothesis, mutations in four essential D. melanogaster dosage compensation genes are shown here to moderately increase rather than decrease hybrid male viability.

CROSSES between Drosophila melanogaster females and males from its sibling species D. simulans produce semiviable F₁ daughters and lethal F₁ sons. This lethality is specific to F₁ hybrid males carrying the D. melanogaster X chromosome (referred to henceforth as X_mel), because patroclinous exceptional sons inheriting the D. simulans X (X_sim) are viable (Sturtevant 1920). F₁ hybrid males carrying X_mel are delayed in development and die as larvae or pseudopupae (Sturtevant 1929; Bolkan et al. 2007) and their X chromosome has an aberrant condensation state in polytenized salivary glands (Hutter et al. 1990; Chatterjee et al. 2007), but the mechanistic basis of the lethality remains unknown.

Dosage compensation is a regulatory process whereby transcription of X-linked genes is equalized between the two sexes (reviewed in Lucchesi et al. 2005). In Drosophila, dosage compensation is achieved by binding of a multisubunit dosage compensation complex (DCC) to the male X chromosome, and mutations in genes encoding DCC subunits result in male-specific lethality. Dosage compensation is controlled by the master sex determination gene Sex-lethal (Sxl). Sxl is active only in females, where it represses the DCC gene msl-2, while in males the absence of Sxl-mediated repression allows for activation of dosage compensation (reviewed in Cline and Meyer 1996). Males mutant for DCC genes develop slowly and die as larvae or prepupae (Belote and Lucchesi 1980). The similarity of this phenotype to that observed in lethal hybrid males and the apparent male specificity of both lethalities might suggest that hybrid male lethality is caused by a failure in dosage compensation.

However, the proposal that dosage compensation defects are the primary cause of hybrid male lethality seems implausible because hybrid lethality is not in fact male specific but rather X_mel specific. This idea was first suggested by Sturtevant (1920), who generated viable X_sim-carrying exceptional F₁ hybrid sons but no corresponding viable X_mel/X_mel exceptional F₁ daughters. Lethality of this genotype of hybrid daughters was later confirmed using compound-X chromosomes (Biddle 1932). Furthermore, Hutter et al. (1990) demonstrated that the genetic basis of lethality in X_mel/Y_sim male and X_mel/X_mel female hybrids is the same, because both are rescued by mutations in Hybrid male rescue (Hmr). In addition, X_mel/X_sim females are lethal when raised at high temperature, and this lethality also is suppressed by Hmr mutations (Barbash et al. 2000). Since dosage compensation is a male-specific process, it is difficult to envision how dosage compensation defects could be the underlying cause of male and female hybrid lethalities that have the same genetic basis. It is possible, however, that hybrid lethality in both males and females could involve interactions with some of the DCC genes that are expressed in both sexes.

One test of the dosage compensation hypothesis found that Sxl mutants do not rescue hybrid males (Orr 1989). This experiment demonstrated that hybrid males do not die due to inappropriate activation of Sxl, but left open the possibility that hybrid males might die due to a failure in dosage compensation that is downstream of Sxl regulation. A subsequent study found that Sxl mutations increase larval viability of hybrid males but do not suppress adult lethality, even when other DCC components (rox RNAs) are overexpressed (Pal Bhadra et al. 2006).

A genomewide microarray comparison of mRNA abundance between lethal (Hmr⁺) and viable (Hmr⁻) hybrids found that a lower than expected number of genes downregulated in lethal hybrid males are located on the X chromosome (Barbash and Lorigan 2007). If lethal hybrids fail to execute dosage compensation, then one would expect the opposite pattern of increased numbers of X-linked genes having significantly reduced expression. This study therefore indicates that lethal hybrids do not suffer from a general failure in dosage compensation, although it remains possible that a small subset of dosage-compensated X-linked genes are misregulated in hybrids.

A number of recent studies have discovered aberrations in DCC localization and dosage compensation in hybrids, as well as interesting evolutionary patterns of DCC genes. Pal Bhadra et al. (2006) discovered that several DCC proteins fail to localize to the D. melanogaster X chromosome in dying hybrid males, but do localize correctly when lethality is suppressed using the Lethal hybrid rescue¹ (Lhr¹) mutation. Chatterjee et al. (2007) observed reduced transcription on X_mel in lethal hybrid males compared to rescued hybrid males and pure-species controls. In addition, two population genetic studies of DCC genes in D. melanogaster and D. simulans found that four of them (msl-1, msl-2, msl-3, and mof) show evidence of having diverged under positive selection along the lineage leading to D. melanogaster, likely due to excessive nonsynonymous (amino acid changing) substitutions (Levine et al. 2007; Rodriguez et al. 2007). Finally, population genetic analyses also found evidence for positive selection acting in D. melanogaster on several X chromosome entry sites to which the DCC binds (Bachtrog 2008).

Following from these reports several groups have suggested that defects in dosage compensation contribute to or cause hybrid male lethality (Chatterjee et al. 2007; Rodriguez et al. 2007; Bachtrog 2008). The most explicit model was provided by Rodriguez et al. (2007). They proposed that DCC proteins and their corresponding binding sites evolved on the D. melanogaster lineage, such that DCCs composed of either D. simulans proteins or a heterospecific mix of D. melanogaster and D. simulans proteins would be unable to bind to X_mel in F₁ hybrid males. F₁ hybrid males carrying X_sim, however, would survive because this X chromosome corresponds to the ancestral state and could be bound by either homospecific or heterospecific DCCs.

This model makes the clear prediction that X_mel/Y_sim F₁ hybrid males heterozygous for mutations in the D. melanogaster orthologs of DCC genes should be lethal (or perhaps have reduced viability), because they will be unable to assemble a DCC that is composed only of D. melanogaster subunits. Here I tested these predictions for four DCC genes in hybrids that are rescued from lethality by D. simulans stocks homozygous for either the Lhr¹ or the Lhr² mutation (Watanabe 1979; Brideau et al. 2006). All DCC mutants are loss-of-function alleles that cause male lethality (Tweedie et al. 2009).

Not all DCC mutant stocks mated successfully to both D. simulans stocks, and in some crosses few or no hybrid sons were produced. Low and variable penetrance of hybrid male rescue by Lhr¹ has been previously observed (Presgraves 2003). But among informative crosses, all produced results inconsistent with the above hypothesis. F₁ hybrids heterozygous for DCC mutations had either equivalent or higher viability compared to their balancer reference brothers (Table 1). This result was not observed in a previous screen that tested several multilocus deficiencies predicted to remove mle in cytological region 42A6 or msl-1 in cytological region 36F11–37A1 (Presgraves 2003). In none of those crosses were Deficiency/+ F₁ males more viable than Balancer/+ controls. The reason for this difference from the results presented here might be explained if the deleted regions contain recessive hybrid inviability genes that decrease rather than increase viability when hemizygous.

TABLE 1.

Testing DCC mutants for suppression or enhancement of hybrid viability

	D. simulans paternal stock		F1 hybrid females		F1 hybrid males			Relative viability of males/females
Allele tested		Temp.	No. mutant/+	No. balancer/+	No. mutant/+	No. balancer/+	No. unscoreablea	Mutant/+	Balancer/+	FETb
mle1	Lhr1	22°−23°	99	103	0	0	0	0	0	1
	Lhr2	22°−23°	91	83	0	0	0	0	0	1
mle9	Lhr2	22°−23°	68	57	9	0	0	0.13	0.00	0.010
	Lhr2	25°	210	206	3	0	2	0.01	0.00	0.248
msl-1γ216	Lhr2	22°−23°	67	76	50	13	0	0.75	0.17	1.51E-05
	Lhr2	25°	147	153	91	9	13	0.62	0.06	2.81E-14
msl-21	Lhr1	18°	194	187	267	193	17	1.38	1.03	0.0435
	Lhr1	25°	565	535	465	505	16	0.82	0.94	P > 0.131c
	Lhr2	18°	73	77	50	14	30	0.68	0.18	7.31E-05
	Lhr2	25°	151	157	65	12	35	0.43	0.08	4.51E-08
msl-2γ227	Lhr1	18°	50	52	52	47	0	1.04	0.90	0.673
	Lhr2	18°	64	57	26	4	44	0.41	0.07	7.11E-04
	Lhr2	25°	213	278	25	3	27	0.12	0.01	1.11E-06
msl-31	Lhr2	18°	34	31	12	0	0	0.35	0.00	2.42E-03
	Lhr2	22°−23°	91	86	18	0	0	0.20	0.00	6.76E-05
	Lhr2	25°	94	15	1	0	0	0.01	0.00	1

D. melanogaster females heterozygous for the indicated alleles were crossed to D. simulans males homozygous for either Lhr¹ or Lhr². For most crosses two replicates were performed containing ∼15–25 female and ∼20–30 male parents. Eclosion was the criterion for viability. Full genotypes of D. melanogaster mutant stocks were (1) y¹ w¹ N^spl-1/Dp(1;Y)B^S; mle¹/SM1, (2) mle⁹ cn¹ bw¹/CyO, (3) msl-1^γ216 cn¹ bw¹/CyO, (4) w; msl2¹ cn/CyO, (5) w; msl2^γ227 bw¹/CyO, and (6) msl-3¹ red¹/TM3, Sb¹ Ser¹. All mutant alleles, including molecular information when known, are described in FlyBase (Tweedie et al. 2009).

^aUnscoreable males were mostly progeny that had died and become stuck in the media and thus could not be scored for the Cy marker on the balancer chromosome. A negligible number of females were unscoreable (between 0 and 4 in all crosses) and are not shown.

^bInhomogeneity in the numbers of the four classes of scoreable progeny was tested by Fisher's exact test (FET), unless otherwise indicated.

^cSignificance was tested using a chi-square test due to the large sample size.

Surprisingly, a significant viability increase was observed here in at least one cross for all four DCC genes tested (Table 1). Viability of rescued hybrid males is highly sensitive to temperature (Hutter and Ashburner 1987; Aruna et al. 2009). In several crosses, such as with mle¹, mle⁹, and msl-3¹, a significant enhancement was observed at <25°, but few or no hybrid males were produced at 25°. This result suggests that these DCC mutations are acting as minor modifiers of hybrid viability rather than as major-effect hybrid rescue mutations comparable to Hmr and Lhr. Nevertheless, the positive effect on viability among all DCC genes is intriguing.

I suggest that hybrid lethality is caused by a defect in X_mel chromatin state or structure, one consequence of which is that the DCC fails to properly localize in male hybrids as described by Pal Bhadra et al. (2006). This failure of DCC localization certainly must further reduce hybrid male fitness but it does not appear to be the primary cause of the lethality. Several genes encoding heterochromatin proteins such as HP1 and SU(VAR)3–7 are required in both sexes, but also show preferential effects in males on viability and X chromosome structure (Liu et al. 2005; Spierer et al. 2005). Su(var)3–7 shows particularly intriguing interactions, with both reduced and increased dosage leading to defects in DCC localization (Spierer et al. 2008). Increased dosage of Su(var)3–7 also induces changes in X chromosome morphology and enrichment of SU(VAR)3–7, HP1, and the post-translational histone modification H3K9me2 on the X chromosome. Strikingly, these alterations require a functional DCC, even in females where the DCC is not normally assembled.

LHR (also called HP3) interacts with HP1 and its localization to heterochromatin depends on HP1 (Brideau et al. 2006; Greil et al. 2007). One possibility is that Lhr-dependent hybrid lethality is caused by increased localization of HP1 and other heterochromatin proteins to the X chromosome, which induces an improper level of X chromosome heterochromatin formation. By analogy to the Su(var)3–7 overexpression phenotype, mutations in DCC genes may reduce this X chromosome enrichment of heterochromatin proteins in hybrid males and thus enhance their viability.