Balancing sex chromosome expression and satisfying the sexes

Abstract

Equalizing sex chromosome expression between the sexes when they have largely differing gene content appears to be necessary, and across species, is accomplished in a variety of ways. Even in birds, where the process is less than complete,¹ a mechanism to reduce the difference in gene dose between the sexes exists. In early development, while the dosage difference is unregulated and still in flux, it is frequently exploited by sex determination mechanisms. The Drosophila female sex determination process is one clear example, determining the sexes based on X chromosome dose. Recent data show that in Drosophila, the female sex not only reads this gene balance difference, but at the same time usurps the moment. Taking advantage of the transient default state of male dosage compensation, the sex determination master-switch Sex-lethal which resides on the X, has its expression levels enhanced before it works to correct the gene imbalance.² Intriguingly, key developmental genes which could create developmental havoc if their levels were unbalanced show more exquisite regulation,³ suggesting nature distinguishes them and ensures their expression is kept in the desirable range.

Dosage Compensating the Sexes

Whenever the sex chromosomes carry substantially different numbers of genes, a mechanism for equality is required, and some form of dosage compensation occurs (reviewed in ref. 4). In mammals where males are XY and females XX, transcription of one of the two female X chromosomes is generally shut down. By contrast, in Drosophila, males compensate for their single X chromosome by upregulating expression of their X-linked genes by approximately two-fold. The nematode C. elegans solves the imbalance differently; it downregulates transcription of both X chromosomes in the hermaphrodite by approximately half to match the single X of males.

Perhaps the best understood dosage compensation mechanism is that of Drosophila. Early in development, males assemble the dosage compensation complex (DCC), whose key members include several male specific lethal proteins (the MSLs) and two non-coding RNAs (roX1 and roX2) which function in a redundant capacity, onto their single X. The DCC paints the X chromosome, presumably spreading from its high affinity entry sites, to result in the hyperacetylation of lysine 16 on histone H4 by one of the complex proteins, males absent on the first (MOF). H4 lysine 16 acetylation is associated with increased transcriptional activity, and it is only detected at high levels on the male X (reviewed in ref. 5). Convention held that the DCC complex fails to assemble in females, as the sex determination masterswitch protein Sex-lethal (SXL, an RNA-binding protein) prevents its assembly. Full length SXL is only expressed in females in response to the X chromosome to Autosome set counting system [abbreviated X:A ratio (reviewed in ref. 6)]; in females it inhibits the expression of MSL-2, a key and stabilizing component of the DCC that is a histone H2B E3 ubiquitin ligase.⁷ SXL inhibits MSL-2 production at two levels; first, by altering the splicing of msl-2 pre-mRNA so that the first intron is retained in the msl-2 mRNA (males splice out this intron by default).^8-12 This change in splicing leads to instability of the msl-2 mRNA and a significant decrease of the transcript in females.⁸ Second, SXL translationally silences MSL-2 expression, binding to poly(U) stretches in the msl-2 mRNA 5′ and 3′ UTRs to accomplish this.^9,11-13

Drosophila Females Take Advantage of the Ground State

In 2010, we provided evidence that females whose viability was partially compromised by a weakened X:A counting system, required all of the key components of the male DCC to robustly determine their sex.² This is in contrast to when the X:A counting system is intact; unlike males, females homozygous for msl mutations are fully viable, hence the ‘male specific lethals’. As the sex determination decision is made very early in development, we proposed that upregulating the transcription of the sex determination master switch Sxl, an X-linked gene, by the DCC contributes to the process of feminizing the embryo in response to their two X chromosomes. If the ground state very early in development is male, before the female state is set and SXL expression is established, there would be no reason a priori for the DCC not to assemble and elevate female X chromosome expression, provided the components of the DCC are present. Females would thus utilize the male dosage compensation process to amplify the signal which determines their fate, as the DCC complex would upregulate the level of early Sxl transcripts which set in motion the pathway of determining the female state.

Many of the DCC components are maternally deposited. Two key missing ingredients are roX1 RNA and MSL-2; however, roX1 is transcribed at about 2 h of embryogenesis¹⁴ and RNA-seq data show a low level of msl-2 RNA is deposited into the egg (Flybase, www.flybase.org). In 2–4 h embryos, both genes are well expressed at high and moderately high levels, respectively. Taken with the genetic data, which show msl-2 has a maternal effect on the differentiation and viability of daughters with reduced Sxl dose,^2,15 we argued that all of the elements were in place for dosage compensation to initiate by 2–3 h not only in males but also in females. Semi-quantitative RT-PCR of key early sex determination genes during this period, including Sxl, and in situ data of Sxl early transcription provided the evidence for this view. Direct detection of the complex by immunofluorescence with high-affinity anti-MSL-1 antibodies gave the earliest detected embryonic signal in mid cycle 14 (~3 h). We suggested this slightly late window of detection was likely a technical limitation of reagents, as suggested by McDowell et al.,¹⁶ and may also be influenced by the fact that before cycle 14 and the mid-blastula transition, the number of X-linked genes being transcribed is too low to allow the ready detection of the DCC.¹⁷ The complex only binds to genes which are being actively transcribed.^18-20

Recently Lott et al.³ published an extensive analysis of early Drosophila gene expression from single embryos, from cycle 10 to when cellularization of the blastoderm is completed, cycle 14. Cycle 14 was assigned four stages, set by the extent to which the membranes have invaginated from the periphery until cell closure. The developmental stage and gender of each embryo was carefully scored, and the authors took advantage of natural biallelic single nucleotide polymorphisms between strains to identify the maternal vs. paternal genome. Of the high number (10,492) of polymorphisms within annotated genes found, 2,210 were available to score as purely zygotic genes in male and female embryos.

Lott et al.³ report that zygotically derived transcripts from autosomal genes are at the same levels in females and males over all time points. Zygotically derived X chromosome transcripts, however, are consistently higher in females than in males, yet not twice as high. The female to male ratio ranged between 1.0 and 2.0 with a mean of 1.5, median of 1.4. By contrast, for chromosome 2L, the mean and median female to male ratios were 1.1 and 1.0, respectively. Because the abundance of zygotic gene transcripts from the single male X chromosome was consistently higher than from either female X chromosome, the authors conclude “that transcript abundance in the early embryo transcription is subject to some form of dosage compensation.”

I would like to suggest that this compensation is from the activity of the DCC. As noted above, all of the known components are available after 2 h of development, and it is possible to detect the consequence of their absence.

While Lott et al.³ point out that males are showing signs of some form of dosage compensation, they do not consider the possibility that females might be downregulating the expression from each of their X chromosomes. Without other data such as developmental outcomes from specific target genes—in our case that Sxl expression, and thus female sex determination, depends on the msls—there is no reference for whether male X chromosome expression is going up, or if females are lowering the expression of each of their Xs.

A factor which may have contributed is that their RNA-seq data show essentially no msl-2 RNA in embryos prior to cycle 12, and the levels increase in both males and females beyond that stage.³ As noted above, the Flybase RNA-seq data show low levels of msl-2 mRNA in 0–2 h embryos and this time frame would suggest the RNA is maternally deposited. Given the importance of MSL-2 to the DCC and our proposal,² it seems appropriate to consider possible explanations for this discrepancy.

For both scenarios, the contribution of RNA-seq data plays a prominent role. However, the approach has some shortcomings as highlighted recently by McIntyre et al.;²¹ technical as well as biological replicates can have variability which results in inconsistent detection of exons with low levels of coverage. For any given gene, RNA-seq replicates of the same library can be off in their quantitation by one log10 or even two, at all RNA levels. A reasonable explanation they raise is that in a library of 4 pmoles, 30 million reads is still only 0.0013% of the molecules undergoing sequencing.²¹ Given a low sampling frequency, variability will arise, and it is amplified for genes of low abundance. With only one embryo of each sex sequenced for each of the cycles prior to cycle 14 (two for each of the cycle 14 substages), and the low abundance of msl-2 mRNA, it is possible that its signal was not detected or underrepresented in Lott et al.³ Additionally, they suggest they sequenced 40 pmoles, which would lower the mRNA sampling by another 10-fold, to 0.00013%.

It is also possible that msl-2 mRNA is somewhat unstable, and was not recovered as efficiently in their challenging single embryo libraries. The Flybase RNA-seq data, which should not be as constrained given that single embryos were not analyzed, have the signals for Sxl vs. msl-2 in the 2–4 h window with a similar order of magnitude - arbitrary units of 2705 and 1851, respectively. Comparing the Lott et al.³ quantitation of Sxl transcripts against msl-2 in their Figure 3B, the y-axes have the msl-2 levels on a smaller scale than Sxl by almost 4-fold, suggesting the msl-2 signal is underrepresented.

The signal Flybase reports for maternal msl-2 mRNA while low, is at 10% of the peak over all developmental times, which is in 2–4 h embryos. Ten percent of MSL-2 would seem more than adequate to upregulate the handful of X-linked genes being transcribed early in development.¹⁷ The maternal effect we and Uenoyama et al.¹⁵ report of msl-2 mutations on females with reduced Sxl dose, would support this view.

An Episode of Conflict

In the window between 2–3.25 h of development (roughly cycle 12 to the end of cycle 14) a battle between the sexes is being waged in females. On the one hand, fairly early in this time frame, the DCC is being assembled and the limited genes on the X that are being transcribed¹⁷ have their transcription elevated. On the other hand, females are establishing the expression of SXL from the X:A ratio sensitive promoter.

Before and while the DCC is operational in both sexes, X chromosome transcript levels between females and males are expected to remain at or close to two (Fig. 1). Indeed, the data in Figure 4C of Lott et al.³ show that up to cycle 12, each of the X chromosomes, whether in males or females, is expressed at the same level (i.e., 1:1 per X chromosome, or as noted in reference 2, 2:2 per X chromosome). From then onwards, the per X chromosome expression in females begins to drop, decreasing through the completion of cycle 14 which is the end of the assay period.

Figure 1. Status of X chromosome transcription in early Drosophila embryos. Female with two Xs shown at the top, male with single X below, with the chromosome size depicting relative intensity of transcribed genes. As SXL levels rise in females, it begins to reduce X chromosome transcript elevation by repressing MSL-2 expression as well by directly dosage compensating genes, like run. Male state is default and continues uninterrupted.

Cycle 12 is also when the first burst of SXL protein is made. As SXL protein increases, it should decrease the levels of MSL-2 and decrease the level of dosage compensation in females. By the scenario in (2) the female to male output would then be < 2:2 per X chromosome, and this downturn would account for the described female: male ratio of overall X chromosome transcripts being 1.4–1.5, and not 2.

As the levels of SXL rise, this ratio is expected to continue to fall below 2, until the battle between the sexes is resolved and Sxl protein effectively and completely shuts down the DCC. One might even predict that the exact rate of departure from 2 would vary between genes close to the DCC high affinity sites^22,23 from those that are further. Additionally, and inclusively, the ability of Sxl to directly regulate transcript levels of some X chromosome genes, as in the dosage compensation of the segmentation gene runt (run),^24,25 would also contribute to equalizing the overall transcript levels between the sexes. The final outcome would be the interplay between all the opposing activities.

The exact timing for the complete takeover of SXL in females is not known—the early protein which has a different N-terminus from the late maintenance protein is less effective at splicing regulation,²⁶ and although Sxl transcripts can be readily detected by cycle 12, a readily detectable protein signal is not seen until cycle 14. As the levels of the maintenance protein rise, the expectation is that inhibition of assembly of the DCC in females would become more complete.

Finally, the two timings for Sxl transcript expression are noteworthy in view of the two inflection points in overall female X chromosome mRNAs levels in Figure 4B of Lott et al.³ These inflections, at cycles 12 and 14C, show decreases in overall female X chromosome transcript levels, which draw them away from the predicted male 2-fold level, closer toward the levels detected in males. Cycles 12 and 14B are also when the two Sxl promoters which generate the SXL early and maintenance proteins, respectively, turn on. It would be interesting to determine if these dips in female X chromosome mRNAs would disappear in female embryos without a functional Sxl gene. Conversely, a loss of the DCC through mutation of any of the components, particularly msl-2 or the roX genes (the two “missing” maternal components noted above), should show interesting changes in X chromosome expression levels in both sexes.

Key Developmental Regulators Are Kept at Near Equal Levels Between the Sexes

For regulatory genes where protein levels must be maintained within a narrow and given range, an obvious function of transcriptional dosage compensation is to keep the mRNA levels within specification. Alternatively, feedback and/or translational regulatory mechanisms must exist to correct for the differences between the sexes, unless the organism requires the difference for dimorphic development.

The RNA-seq data³ indicate that several X-linked developmental regulator genes, such as giant (gt), brinker, buttonhead, and short gastrulation, show good zygotic dosage compensation with nearly identical transcript levels in male and female embryos. However, several genes were also noted not to show evidence of compensation at the transcript level. If as proposed above, canonical DCC is operational much earlier in development, then the equalized expression of many X-linked genes is accommodated. Indeed to quote Lott et al.³, “The simplest explanation is that the MSL-based dosage compensation system is active before and during cycle 14, leading to hypertranscription of the male X.”

While pre-cycle 14 dosage compensation explains the corrected expression of many genes, two exceptions should be noted. First, SXL performs a dosage compensation function in females and has been shown to compensate the X-linked segmentation gene run in an MSL independent manner.^24,25 The 3′ UTR of run mRNA has several matches to the SXL consensus binding sequence, as do many X-linked (but relatively few autosomal) genes, although a direct role for SXL in run dosage compensation has not been demonstrated. Second, a few of the key X-linked developmental regulators which had mRNAs at roughly equal levels in males and females are expressed very early, gt or nullo for example, starting as early as cycle 10–11. Unlike run, these genes precede the early SXL protein so it is unlikely that they would be regulated (at least initially) by SXL or the DCC. Their close mRNA levels between the sexes suggest an additional/ alternative mechanism for correction, which may adjust their transcription and/or mRNA stability. Being regulatory genes, it may not be too far reaching to postulate some form of feedback mechanism. Understanding the regulatory network of these exceptions would provide insight into how genes maintain a tight control of their given expression levels.

Gladstein N, McKeon MN, Horabin JI. Requirement of male-specific dosage compensation in Drosophila females—implications of early X chromosome gene expression. PLoS Genet. 2010;6:e1001041. doi: 10.1371/journal.pgen.1001041.