Reduced histone gene copy number disrupts Drosophila Polycomb function

Abstract

The chromatin of animal cells contains two types of histones: canonical histones that are expressed during S phase of the cell cycle to package the newly replicated genome, and variant histones with specialized functions that are expressed throughout the cell cycle and in non-proliferating cells. Determining whether and how canonical and variant histones cooperate to regulate genome function is integral to understanding how chromatin-based processes affect normal and pathological development. Here, we demonstrate that variant histone H3.3 is essential for Drosophila development only when canonical histone gene copy number is reduced, suggesting that coordination between canonical H3.2 and variant H3.3 expression is necessary to provide sufficient H3 protein for normal genome function. To identify genes that depend upon, or are involved in, this coordinate regulation we screened for heterozygous chromosome 3 deficiencies that impair development of flies bearing reduced H3.2 and H3.3 gene copy number. We identified two regions of chromosome 3 that conferred this phenotype, one of which contains the Polycomb gene, which is necessary for establishing domains of facultative chromatin that repress master regulator genes during development. We further found that reduction in Polycomb dosage decreases viability of animals with no H3.3 gene copies. Moreover, heterozygous Polycomb mutations result in de-repression of the Polycomb target gene Ubx and cause ectopic sex combs when either canonical or variant H3 gene copy number is reduced. We conclude that Polycomb-mediated facultative heterochromatin function is compromised when canonical and variant H3 gene copy number falls below a critical threshold.

Introduction

To control access to information encoded in the genome, eukaryotes organize their DNA into chromatin, which regulates all DNA-dependent processes including transcription, DNA replication, and DNA damage repair (Allis et al. 2007; Kornberg and Lorch 2020). The fundamental unit of chromatin is a nucleosome composed of approximately 147 bp of DNA wrapped around a histone octamer containing two copies of each of the four core histones: H2A, H2B, H3, and H4 (Luger et al. 1997). Tight control over histone levels is essential for normal genome function. For instance, mutations in abo and mute—which negatively regulate histone mRNA levels—reduce viability in Drosophila melanogaster (Berloco et al. 2001; Bulchand et al. 2010). In budding yeast, mutations that cause an accumulation of excess histone proteins result in impaired growth, DNA damage sensitivity, and chromosome loss (Meeks-Wagner and Hartwell 1986; Gunjan and Verreault 2003). Conversely, conditional repression of histone transcription during S phase impairs DNA replication and causes cell cycle arrest in yeast and fruit flies (Han et al. 1987; Sullivan et al. 2001; Gossett and Lieb 2012). Similarly, deletion of all D. melanogaster canonical histone genes leads to cell cycle arrest and embryonic lethality (Smith et al. 1993; Günesdogan et al. 2010; McKay et al. 2015). Histone chaperone mutations that reduce incorporation of histone proteins into chromatin cause spurious transcription, chromosome segregation defects, chromosomal rearrangements, and enhanced DNA damage (Clark-Adams et al. 1988; Nelson et al. 2002; Myung et al. 2003; Ye et al. 2003; Nashun et al. 2015; Mühlen et al. 2023). For these reasons, precise regulation of histone mRNA and protein levels is critical for normal cell function and development, yet we have an incomplete understanding of the mechanisms involved.

Most research investigating the mechanisms of histone expression has focused on the canonical histone genes, which are synthesized in large amounts during S phase to properly package newly replicated DNA into chromatin. This work provides evidence supporting regulation at both the transcriptional and post-transcriptional levels (Marzluff and Duronio 2002; Duronio and Marzluff 2017). For example, in Chinese hamster ovary cells, canonical histone mRNA levels increase 35-fold as cells enter S phase (Harris et al. 1991). As cells exit S phase, canonical histone transcription is terminated and the corresponding mRNAs are rapidly degraded (Kaygun and Marzluff 2005; Eriksson et al. 2012). Coordinate expression among histone genes to maintain nucleosome subunit stoichiometry is also important; this requirement is reflected in the clustered arrangement and co-regulation of the canonical histone genes in multiple species, including D. melanogaster, yeast, and mammals (Lifton et al. 1977; Smith and Murray 1983; Eriksson et al. 2012). In the D. melanogaster histone gene complex (HisC, Fig. 1a), H2A and H2B share a bidirectional promoter, as do H3 and H4 (Lifton et al. 1977). Histone protein levels are also controlled post-translationally. For example, yeast histones that are not chromatin-bound are rapidly degraded, suggesting that excess histone proteins are deleterious to cell function (Singh et al. 2009). Another example of post-translational control that applies to a variant histone occurs during Drosophila oogenesis when the pool of free H2Av protein is regulated by Jabba, which physically sequesters H2Av to prevent its degradation (Stephenson et al. 2021).

Fig. 1.

H3.3 is required for viability when H3.2 gene copy number is reduced. a) Diagram of the genomic locations of the three H3 genes and the 12xHWT transgene. 100x and 12x indicate gene copy number. b) Schematic model of H3.2 and H3.3 expression during the cell cycle. c) Amino acid sequence differences of H3.2 and H3.3. d) Bar plot of viability for the indicated genotypes. Circles represent the full complement of H3.3 or H3.2 gene copies present in the wild-type genome, and the color filling each circle (H3.3 green, H3.2 blue) indicates the number of gene copies present in each experimental genotype, e.g. A solid blue circle indicates the 200 H3.2 gene copies of the diploid HisC locus. A solid green circle indicates 3 H3.3 gene copies in males and 4 gene copies in females since H3.3B is on the X chromosome. The number of genes is shown below the circles, as well as whether surviving adults are fertile. Percent of expected genotypic frequencies based on Mendelian ratios. Asterisks indicate fewer than expected survive (chi-square test, ** P < 0.01, see Supplementary Table 1 for P-values).

Whereas canonical histones are encoded by multiple genes that are expressed exclusively during S phase of the cell cycle (Fig. 1, a and b), an additional layer of complexity is provided by the expression of cell cycle independent histones (Franklin and Zweidler 1977; Verreault et al. 1996; Marzluff et al. 2002; Tagami et al. 2004). These so-called “variant” histones are typically encoded by one or two genes and are expressed throughout the cell cycle (Fig. 1, a and b) (Urban and Zweidler 1983; Pantazis and Bonner 1984; Zweidler 1984; Brown et al. 1985; Piña and Suau 1987; Wunsch and Lough 1987; McKittrick et al. 2004; Tagami et al. 2004; Mito et al. 2005; Szenker et al. 2011; Maze et al. 2015; Tvardovskiy et al. 2017; Sauer et al. 2018). The tight control of canonical histone levels and the severe negative impact of histone mis-expression raises the possibility that coordinate regulation between canonical and variant histones is important for genome function and stability. For instance, in the early Drosophila embryo an increase in the ratio of variant H2Av to canonical H2A causes mitotic defects and reduces viability (Li et al. 2014). Here, we address the question of coordinate regulation between variant and canonical histone genes by focusing on those encoding histone H3.

In D. melanogaster, the non-centromeric H3 variant is encoded by two genes that produce identical proteins, H3.3A and H3.3B (Fig. 1a). Variant H3.3 differs from canonical H3.2 by only four amino acid residues (Fig. 1c), and these differences are highly conserved among other animals including humans (Malik and Henikoff 2003; Szenker et al. 2011). Three of the four residues are found in the globular domain and are known to modulate interactions with the histone chaperone complexes that deposit histones into chromatin (Grover et al. 2018). Canonical H3.2 is deposited during DNA replication by CAF-1 (chromatin assembly factor 1 complex) (Smith and Stillman 1989; Verreault et al. 1996; Shibahara and Stillman 1999; Tagami et al. 2004; Sauer et al. 2018). Variant H3.3 is deposited into chromatin by the ATRX (alpha-thalassemia X-linked mental retardation protein) complex and the HIRA (histone cell cycle regulator) complex (Ahmad and Henikoff 2002; Tagami et al. 2004; Schneiderman et al. 2009; Goldberg et al. 2010; Lewis et al. 2010; Rai et al. 2011; Orsi et al. 2013; Ray-Gallet et al. 2018; Torné et al. 2020). Whereas H3.2 is deposited evenly genome-wide during replication, H3.3 is enriched at sites with high nucleosome turnover, including active regulatory elements, transcribed gene bodies, and pericentromeric regions (Ahmad and Henikoff 2002; McKittrick et al. 2004; Mito et al. 2005; Wirbelauer et al. 2005; Loyola and Almouzni 2007; Goldberg et al. 2010; Szenker et al. 2011; Martire and Banaszynski 2020). The fourth amino acid difference between H3.2 and H3.3 occurs at position 31 in the post-translationally modified N-terminal tail (Szenker et al. 2011). Position 31 is an alanine in H3.2 (H3.2A31) and a serine in H3.3 (H3.3S31), which can be phosphorylated (Hake et al. 2005; Martire et al. 2019; Armache et al. 2020; Sitbon et al. 2020). Other residues on the N-terminal tails of H3.2 and H3.3 are also differentially enriched in post-translational modifications (PTMs), likely due to their differential localization in the genome. Relative to H3.2, H3.3 is enriched with PTMs associated with active chromatin (e.g. H3K4me3) and depleted in marks associated with inactive chromatin (e.g. H3K9me2) (McKittrick et al. 2004; Hödl and Basler 2012). Although the mechanisms regulating canonical and variant histone mRNA and protein levels are distinct, we do not know if and how these mechanisms are coordinated to supply the necessary amount of each histone isotype across the genome. Here, we use D. melanogaster to explore this question by examining the consequences of manipulating the relative number of canonical and variant H3 genes.

Genetically manipulating histone gene copy number is challenging in many metazoans, including mice and humans, because canonical histones are encoded by multiple gene clusters located at distinct chromosomal locations (Marzluff and Duronio 2002). D. melanogaster is a powerful organism to investigate the effects of altering histone gene copy number because all ∼100 haploid copies of the canonical histone genes are tandemly repeated (Fig. 1a) and can be removed with a single genetic deletion, ΔHisC (Günesdogan et al. 2010). The ability to manipulate histone genes in D. melanogaster led to the discovery that canonical histone gene copy number is a modifier of position effect variegation (PEV), a genetic phenomenon associated with heterochromatin. Heterozygosity of the histone locus results in suppression of PEV, suggesting that histone abundance contributes to maintenance of epigenetic silencing of H3K9me3-marked constitutive heterochromatin (Moore et al. 1979; Moore et al. 1983; Sinclair et al. 1983). The ability to manipulate histone gene copy number in D. melanogaster has been extended in recent years. Replacement of all ∼200 copies (200xWT) of the canonical histone genes with a transgene containing 12 wild-type canonical histone gene repeat units (12xHWT,  Fig. 1a) is sufficient to support development and provides a means of altering canonical histone gene copy number with precision (McKay et al. 2015).

Here, we report that 12xHWT viability depends on expression of variant H3.3 genes, whereas 200xWT viability does not. This finding suggests that coordination of H3.2 and H3.3 protein levels is necessary for proper development when either H3.2 or H3.3 gene copy number is reduced. We conducted a screen to identify genes involved in the coordinated control of H3.2 and H3.3. We identified a deficiency that uncovers Yem, a component of the HIRA histone chaperone complex, the function of which may be particularly important when H3.2 gene copy number is reduced. Surprisingly, we also found that reduction of Polycomb (Pc) gene function decreases viability of flies that have reduced numbers of H3.3 genes. Furthermore, we found that reductions in either H3.2 or H3.3 gene copy number disrupt Polycomb-mediated gene repression. Rather than Pc being involved in the coordinate expression of canonical and variant H3, we conclude from these findings that the appropriate balance of H3.2 and H3.3 genes is critical for the proper epigenetic silencing of developmental genes and maintenance of facultative heterochromatin function.

Materials and methods

Fly stocks and husbandry

Fly stocks were maintained on standard corn medium provided by Archon Scientific (Durham, NC). See Supplementary Table 9 for a list of all stocks.

CRISPR-mediated generation of H3.3A^Δ

pCFD4 plasmids encoding dual gRNAs (gRNA_1: caaggcgccccgcaagcagc, gRNA_2: tgcaccgtgactatttcata) targeting H3.3A were injected into embryos expressing Cas9 from the nanos promoter (y1 M{nos-Cas9.P}ZH-2A w*; RRID: BDSC_54590) by GenetiVision Corporation (Houston, TX). H3.3A^Δ alleles were identified by PCR of genomic DNA and confirmed by sequencing, which revealed a 265 bp deletion removing amino acids 19 through 94 of the H3.3A open reading frame. The deletion breakpoints are indicated by “…/…” in the following sequence: CAAGGCGCCCCGCA…/…TACGGTCATGTAAT. The H3.3A^Δ allele was determined to be amorphic because H3.3B⁰; H3.3A^Δ; 12xHWT/+ animals are inviable.

CRISPR diagnostic screen primer set: H3.3A^Δ for-CCCGATGAATATAGGGTCACAC, H3.3A^Δ rev-CTGGATGTCCTTGGGCATAAT. pCFD4-U6:1_U6:3 was a gift from Simon Bullock (Addgene plasmid #49411; http://n2t.net/addgene:49411; RRID:Addgene_49411). His3.3A reference sequence: NCBI Gene ID 33736.

Viability

To examine the effect of H3.2 gene copy number on H3.3^null viability (Fig. 1d), the following four crosses were performed:

1. H3.3B⁰/H3.3B⁰; H3.3A^2×1, ΔHisC, twistGal4/CyO; x yw/Y; H3.3^A2×1, ΔHisC, UAS-YFP/CyO; 12xHWT/12xHWT;

2. H3.3B⁰/H3.3B⁰; H3.3A^2×1, ΔHisC, twistGal4/CyO; x yw/Y; H3.3^A2×1, ΔHisC, UAS-YFP/CyO; 12xHWT-VK33, 8xHWT-86F6/+;

3. H3.3B⁰/H3.3B⁰; H3.3A^2×1, ΔHisC, twistGal4/CyO; x H3.3B⁰/Y; H3.3A^2×1/CyO, twistGFP; and

4. H3.3B⁰/H3.3B⁰; H3.3A^Δ/CyO, twistGFP; x H3.3B⁰/Y; H3.3A^Δ/CyO, twistGFP.

To examine the effect of H3.3 gene copy number on 12xHWT viability (Fig. 1d), the following three crosses were performed:

1. H3.3B⁰/H3.3B⁰; H3.3A^2×1, ΔHisC, UAS-YFP/CyO; 12xHWT/12xHWT x yw; H3.3^A2×1, ΔHisC^Cadillac/CyO;

2. yw/yw; H3.3A^2×1, ΔHisC^Cadillac; 12xHWT/12xHWT x H3.3B⁰/Y; H3.3A^2×1, ΔHisC, twistGal4/CyO; and

3. yw/yw; ΔHisC, twistGal4/CyO; x yw; ΔHisC, UAS-YFP/CyO; 12xHWT/12xHWT.

Vials were maintained at 25°C and flipped every other day. Data were plotted as percent observed of expected based on Mendelian ratios for each individual cross. Chi-squared analysis was performed to determine statistical significance. A significance threshold of P < 0.05 was used in this study. See Supplementary Table 1 for all genotypes and progeny numbers. All genotypes were confirmed by PCR. Primer information can be found in Supplementary Table 11.

RNA sequencing

For each replicate sample, 25 brains were dissected from wandering 3rd instar larvae and homogenized in Trizol solution. RNA was isolated from the Trizol aqueous phase using the Zymo RNA Clean and Concentrator-5 kit (Genesee Scientific #11-352) including treatment with DNAse I, as per the manufacturer's instructions. Libraries were prepared from polyA-selected RNA using the KAPA stranded mRNA kit (Roche # 07962207001) and sequenced using the NOVASeq-S1 paired-end 100 platform. Sequence reads were trimmed for adaptor sequence/low-quality sequence using BBDuk (bbmap v38.67) with parameters: ktrim = r, k = 23, rcomp = t, tbo = t, tpe = t, hdist = 1, mink = 11. Dm6 genome files for use with the STAR aligner were generated using parameters: sjdbOverhang 99. Paired-end sequencing reads were aligned using STAR v2.7.7a with default parameters (Dobin et al. 2013). featureCounts (subread v2.0.1) was used with default parameters to count reads mapping to features (Liao et al. 2014). DESeq2 (v1.34.0) was used to identify differentially expressed genes (Love et al. 2014). Differentially expressed genes were defined as genes with an adjusted P-value less than 0.05 and an absolute log2 fold change greater than 1.

Western blots

Protein extracts from H3.3^null third instar larvae and 12xHWT and yw (200xWT) third instar larval wing discs were prepared by boiling samples for 10 minutes in Laemmli SDS-PAGE loading buffer followed by sonication using the Bioruptor Pico sonication system (Diagenode) for 10 cycles (30 sec on, 30 sec off). Samples were clarified by centrifugation. Proteins were fractionated on BioRad Any kD Mini-PROTEAN TGX Precast Protein Gels GX (BioRad #4569033) and were transferred to 0.2 µm nitrocellulose membranes (BioRad #1620112) at 100 V for 10 and 60 V for 20 minutes. Total protein was detected using G-Bioscience Swift Membrane Stain (G-Bioscience, 786677). Membranes were probed using the following antibodies: rabbit anti-H3 (1:60,000; Abcam Cat# ab1791, RRID:AB_302613), rabbit anti-H3.3 (1:1,000, Abcam Cat# ab176840, RRID:AB_2715502), and mouse anti-tubulin (1:15,000, Sigma-Aldrich Cat# T6074, RRID:AB_477582). Western blot analysis was performed using the following HRP conjugated secondary antibodies: goat anti-Mouse-IgG-HRP (1:10,000, Thermo Fisher Scientific Cat# 31430, RRID:AB_228307), donkey anti-Rabbit-IgG-HRP (1:10,000, (Cytiva Cat# NA934, RRID:AB_772206). Blots were detected using Amersham ECL Prime Western blotting Detection Reagent (Cytvia, RPN2232). ImageLab densitometry analysis was used to determine total protein, tubulin, H3.3, and H3 band intensity. Histone signal was normalized to corresponding tubulin signal. Normalized signals from different titrations of the same genotype were averaged and resulting values were set relative to the wild-type value. This process was completed for three biological replicates. See Supplementary Fig. 2 and Supplementary Table 2 for biological replicate blots and signal quantification, respectively.

EdU + RNA-FISH

Third instar larval eye discs were dissected in Grace's medium and incubated in 0.1 mg/mL EdU for 30 minutes. Samples were then washed in phosphate-buffered saline (PBS) for 5 minutes, fixed for 30 minutes in 4% paraformaldehyde (16% paraformaldehyde, diluted in PBS), washed 3× 15 minutes in PBS with Triton (PBST) (0.5% Triton X-100) and washed for 30 minutes in 3% Bovine serum albumin (BSA) in PBS. EdU was detected using the Click-iT EdU Cy5 imaging kit (Invitrogen) according to the manufacturer's instructions. After EdU detection, samples were washed in 3% BSA for 15 minutes, washed in PBS for 5 minutes, and subsequently fixed in 4% paraformaldehyde for 15 minutes. Next, RNA-FISH was performed using Stellaris H3.2 mRNA probes with TAMRA fluorophores following the manufacturer's instructions for D. melanogaster wing imaginal discs. Samples were stained with DAPI (1 µg/mL) at 37°C for 30 minutes in Wash Buffer A. The maximum projection from 4 adjacent z-slices from third instar wandering larval eye discs was used as representative images for each genotype. H3.2 RNA-FISH probe set can be found in Supplementary Table 12.

Mitotic recombination

yw; ΔHisC, FRT40A/CyO, twistGFP; 12xHWT males were crossed to yw, hsFLP; UbiGFP, FRT40A; females. Adults were removed from the vial after 24 hours. Forty-eight hours later, the vials were heat shocked at 37°C for 8 minutes. yw, hsFLP/yw; ΔHisC, FRT40A/UbiGFP, FRT40A; 12xHWT/+ third instar wandering larvae were selected based on lack of twistGFP fluorescent signal.

Genetic screen viability

All fly stocks were maintained on standard corn medium at 25˚C. Crosses were flipped every other day for 8 days. Progeny were scored once per day. All genotypes and progeny numbers can be found in Supplementary Table 3.

To examine the effect of chromosome 3L mutations on H3.3^Δ; 12xHWT viability (Fig. 3b, Fig. 4a):

Fig. 3.

A genetic screen to identify genes required for survival when H3 gene copy is reduced. a) Schematic of the screen used to identify regions of interest (ROI) on chromosome 3 that when heterozygous or hemizygous reduce adult viability of the H3.3A^Δ; 12xHWT experimental genotype compared to control siblings. H3 gene copy number indicated as in Fig. 1. Pink fill indicates gene copy number of the ROI. b) Bar plot of viability for 23 mutations on chr3L and chr3R (see Table 1). Asterisks indicate statistical significance by Chi-square test (** P < 0.01, * P < 0.05, ♦ potential haploinsufficiency irrespective of H3 gene copy number, see Supplementary Table 3 for P-values). Deficiencies covering region of interest indicated by orange bars. Data are plotted as the percent of H3.3A^Δ; 12xHWT animals inheriting the deficiency chromosome relative to sibling animals inheriting the homologous balancer chromosome, which establish the expected percentage (dashed line). c) Diagram of deficiencies that delineate a region of interest on chr3R from 98E5-99B9. The overlapping region specific to the positive hits from 98F10-98F12 contains seven genes (green bar).

Fig. 4.

Screen of chromosome 3L deficiencies identifies a genetic interaction between Polycomb and histone H3 gene copy number. a) Bar plot of viability for 65 chr3L deficiency mutations (see Table 2). Orange bars indicate two overlapping deficiencies that scored positive and uncover the Pc locus. Data are plotted as in Fig. 3. b) Diagram of deficiencies that delineate a region of interest on chr3L from 78C2–78F1. The overlapping region specific to the positive hits from 78C6–78C8 contains nine genes (green bar). Df(3L)BSC435 (Df 2) and Df(3L)BSC419 (Df 3) correspond to the orange bars in panel A. c) Bar plot of viability of H3.3A^Δ; 12xHWT animals heterozygous for the seven indicated mutations. Data are plotted as in Fig. 3. Asterisks indicate statistical significance by Chi-square test (** P value < 0.01, * P value < 0.05, ♦ potential haploinsufficiency irrespective of H3 gene copy number, see Supplementary Table 3 for P-values).

Bloomington third chromosome deficiency stock males were crossed to yw; H3.3A^2×1, ΔHisC, twistGal4/CyO; MKRS/TM6B virgin females. Subsequently, yw; H3.3A^2×1, ΔHisC, twistGal4/+; Df(3)/MKRS male progeny were crossed with yw; H3.3A^2×1, ΔHisC^Cadillac/H3.3A^2×1, ΔHisC^Cadillac; 12xHWT/12xHWT virgin females. Animals eclosing from deficiency crosses were counted beginning ten days post egg-laying based on the presence or absence of dsRed from the ΔHisC^Cadillac locus and stubble phenotype from MKRS until all adult flies eclosed. Percent of expected was determined by the ratio of yw; H3.3A^2×1, ΔHisC, twistGal4/H3.3A^2×1, ΔHisC^Cadillac; Df(3)/12xHWT was to yw; H3.3A^2×1, ΔHisC, twistGal4/H3.3A^2×1, ΔHisC^Cadillac; MKRS/12xHWT siblings. Significance was determined by a Chi-square test. Observed value is the number of yw; H3.3A^2×1, ΔHisC, twistGal4/H3.3A^2×1, ΔHisC^Cadillac; Df(3)/12xHWT progeny and expected value is the number of yw; H3.3A^2×1, ΔHisC, twistGal4/H3.3A^2×1, ΔHisC^Cadillac; MKRS/12xHWT progeny. Thresholds of P < 0.05 and P < 0.01 were used in this study.

To examine the effect of Pc³ on H3.3^Δ viability (Fig. 6a):

Fig. 6.

Polycomb-mediated gene repression is sensitive to both H3.2 and H3.3 gene copy number. a) Bar plot of viability for the indicated genotypes, depicted as in Fig. 1 with Pc gene dose in pink. Data are plotted as in Fig. 3. Asterisks indicate statistical significance by Chi-square test (** P < 0.01, see Supplementary Table 6 for P-values). b) Scanning electron micrographs of first (T1), second (T2), and third (T3) thoracic legs from adult males of the indicated genotypes. Red brackets indicate the location where sex combs developed. Percent of T2 and T3 legs with ectopic sex combs indicated for each genotype. c) Confocal images of wing imaginal discs of the indicated genotypes stained for DAPI (blue) and Ubx (magenta). Red brackets indicate the wing pouch where ectopic Ubx expression occurred. Bars, 50 μM. d) Bright field images of adult wings of the indicated genotypes. Pc³ was used in panels B and D. Pc¹⁵ was used in panel C.

H3.3B⁰/H3.3B⁰; H3.3A^2×1/CyO, twistGFP; x yw; H3.3A^2×1/CyO; Pc³/MKRS. Animals eclosing were counted beginning ten days post egg-laying based on the presence or absence of curly wings from CyO and stubble phenotype from MKRS until all adult flies eclosed. Percent of expected was determined by the ratio of H3.3B⁰/Y; H3.3A^2×1/H3.3A^2×1; Pc³/ + to H3.3B⁰/Y; H3.3A^2×1/H3.3A^2×1; MKRS/ + siblings. Significance was determined by a Chi-square test. Observed value is the number of H3.3B⁰/Y; H3.3A^2×1/H3.3A^2×1; Pc³/ + progeny and expected value is the number of H3.3B⁰/Y; H3.3A^2×1/H3.3A^2×1; MKRS/ + progeny. Thresholds of P < 0.05 and P < 0.01 were used in this study.

To examine the effect of Pc³ on 12xHWT viability (Fig. 6a):

Yw/Y; ΔHisC, twistGal4/CyO; Pc³/TM6B x yw; ΔHisC, UAS-YFP/CyO; 12xHWT/12xHWT. Animals eclosing were counted beginning ten days post egg-laying based on the presence or absence of curly wings from CyO and humoral phenotype from TM6B until all adult flies eclosed. Percent of expected was determined by the ratio of yw; ΔHisC, twistGal4/ΔHisC, UAS-YFP; Pc³/12xHWT to yw; ΔHisC, twistGal4/ΔHisC, UAS-YFP; TM6B/12xHWT siblings. Significance was determined by a Chi-square test. Observed value is the number of yw; ΔHisC, twistGal4/ΔHisC, UAS-YFP; Pc³/12xHWT progeny and expected value is the number of yw; ΔHisC, twistGal4/ΔHisC, UAS-YFP; TM6B/12xHWT progeny. Thresholds of P < 0.05 and P < 0.01 were used in this study.

Immunofluorescence

For Ubx and H3.3 staining of wing discs, third instar larval cuticles were inverted and fixed in 4% paraformaldehyde in PBS for 20 minutes at room temperature. Cuticles were washed for 1 h in PBST (0.15% Triton X-100). Mouse anti-UBX (1:30, DSHB Cat# FP3.38, RRID:AB_10805300) and mouse anti-H3.3 (1:500, Abnova Cat# H00003021-M01, RRID:AB_425473) were used overnight at 4°C. Goat anti-mouse IgG secondary antibody (1:1000, Thermo Fisher Scientific Cat# A-11029, lot #161153, RRID:AB_2534088) was used for 2 hours at room temperature. DNA was counterstained with DAPI (0.2 µg/mL) and the discs were mounted in Vectashield (VWR, 101098-042) mounting media and imaged on a Leica Confocal SP8.

Scanning electron microscopy

One- to four-day-old flies were dehydrated in ethanol and images of legs were taken using a Hitachi TM4000Plus tabletop SEM microscope at 15 kV and 500× magnification.

Results

H3.3 is required for viability when H3.2 gene copy number is reduced

To examine whether coordination between canonical H3.2 and variant H3.3 gene expression contributes to Drosophila development, we measured the effects of altering the relative number of canonical vs variant histone gene copies on viability and fertility. Zygotes lacking all canonical histone genes (ΔHisC) arrest early in embryonic development, and this lethality can be rescued with a transgene encoding 12 tandemly arrayed histone gene repeats (12xHWT), providing an opportunity to manipulate canonical histone gene dose over an ∼18-fold range (Fig. 1, a and d) (McKay et al. 2015). Null mutations of either H3.3A or H3.3B have no effect on viability or fertility of flies containing the normal complement of canonical histone genes (Supplementary Table 1), but only 50% of the expected number of H3.3A, H3.3B double mutants (H3.3^Δ; 200xWT) eclose as adult flies, which are infertile (Fig. 1d) (Sakai et al. 2009). H3.3^Δ animals heterozygous for a HisC deletion (H3.3^Δ; 100xWT) survive to adulthood at a similar frequency as H3.3^Δ; 200xWT animals (54.2 and 50% of expected, respectively) (Fig. 1d). However, reducing canonical histone gene copy number to 20 (H3.3^Δ; 20xHWT) results in only 17.1% of the expected number of adults (Fig. 1d). A further reduction to 12 histone gene repeats (H3.3^Δ; 12xHWT) results in a complete loss of viability of flies lacking variant H3.3 genes (Fig. 1d). H3.3^Δ; 12xHWT lethality is rescued by one copy of either H3.3A or H3.3B, and the adults of these genotypes are fertile (Fig. 1d and Supplementary Table 1). Thus, we conclude that H3.3 expression is necessary for completion of development when canonical histone gene copy number is reduced to 12, and that the probability of animals lacking H3.3 to complete development increases with increasing numbers of canonical histone genes. Furthermore, our data support previous observations that H3.3 is required for male and female fertility (Fig. 1d) (Sakai et al. 2009). Collectively, these data suggest that H3.3 compensates for reduced H3.2 gene copy number to maintain a critical threshold of total H3 protein.

H3.3 mRNA and protein levels do not change when H3.2 gene copy number is reduced

Mechanisms that compensate for altered variant vs canonical histone genes could operate at many levels, including transcription, translation, histone deposition into chromatin, or histone protein turnover. For instance, increased H3.3 expression could compensate for reduced H3.2 gene copy number, potentially explaining why 12xHWT animals do not survive in the absence of H3.3 genes. We reasoned that measuring steady-state mRNA and protein levels could reveal evidence of such compensatory mechanisms. To determine whether H3.3 steady-state mRNA levels are elevated in 12xHWT animals, we compared H3.3 mRNA levels in 12xHWT vs 200xWT control animals in an RNA-sequencing data set obtained from third instar larval brains. We found no significant difference in H3.3A or H3.3B mRNA levels in 12xHWT compared to 200xWT cells (Fig. 2a). Consistent with these RNA-seq data, immunoblots of third instar larval wing imaginal discs show comparable levels of H3.3 protein in 12xHWT and 200xWT controls (Fig. 2b). Moreover, immunofluorescence of genetically mosaic third instar larval wing discs generated by mitotic recombination show similar levels of H3.3 protein in clones of 12xHWT cells and 212xHWT control cells (Supplementary Fig. 1). We conclude that compensation for reduced H3.2 gene copy number in 12xHWT animals does not occur via detectable changes in the steady-state levels of H3.3 mRNA or protein, as measured by RNA-sequencing, immunoblotting, or immunofluorescence.

Fig. 2.

H3.3 mRNA and protein levels do not change when H3.2 gene copy number is reduced. a) MA plot showing fold change of normalized third instar larval brain RNA-seq signal in wild-type (200xWT) vs 12xHWT for all transcripts (y-axis). Average coverage on the x-axis represents the mean expression level of a transcript. Differentially expressed genes are indicated in dark gray. Values for H3.3A and H3.3B transcripts are indicated (green). b) Western blot of total H3, H3.3 and tubulin from 12xHWT and 200xWT wandering third instar larval wing imaginal discs. Dilution series with indicated number of wing discs for each genotype. H3.3^Δ third instar larvae were used as a negative control. Bar plot depicting the average fold change in H3.3 signal relative to 200xWT signal normalized to tubulin. Error bars represent standard deviation of three biological replicates. c) Confocal images of third instar imaginal eye discs stained for DAPI (grey), EdU (magenta), and H3.2 mRNA (cyan) for the four indicated genotypes, denoted as in Fig. 1. The maximum projection of four adjacent slices is shown. Bars, 50 μM.

We considered the possibility that expression of H3.2 becomes uncoupled from S phase upon loss of H3.3, thereby maintaining a pool of H3 outside of S phase even when H3.3 genes are absent. To determine whether H3.2 transcripts are present in cells outside of S phase in the absence of H3.3 genes, we combined EdU staining with RNA-fluorescent in situ hybridization (FISH) in the developing eye of 200xWT, H3.3^Δ, H3.3^Δ; 12xHWT, and 12xHWT animals. We observed that H3.2 mRNA is only detected in EdU-positive cells in all four genotypes, indicating that H3.2 transcription is not uncoupled from S phase when H3.2 and/or H3.3 gene copy number are reduced (Fig. 2c). In addition, we observed via FISH that the levels of H3.2 mRNA are similar in 12xHWT cells compared to 200xWT controls. Similar findings were reported previously in early D. melanogaster embryos via RT-PCR, suggesting a mechanism of histone dosage compensation at the transcriptional level (McKay et al. 2015).

A genetic screen for genes sensitive to reduced histone H3 gene copy number

The inability to detect evidence of histone gene coordination at the molecular level motivated us to instead take an unbiased genetic approach. Performing a screen in a genotype with reduced variant and canonical histone gene copy number could potentially identify genes that: (1) regulate histone gene expression, (2) coordinate expression between variant and canonical histone genes, or (3) are otherwise sensitive to reduced histone levels. As described above, 12xHWT animals are viable and fertile at wild-type frequencies (Fig. 1d); however, H3.3^Δ; 12xHWT flies are inviable (Fig. 1d). We therefore reasoned that we could identify other genes that when mutated would reduce the viability of 12xHWT animals. Because having one copy of H3.3A or H3.3B is sufficient to retain viability in a 12xHWT background (Fig. 1d), we decided to screen using a 12xHWT background that is further sensitized by the removal of both copies of H3.3A. We refer to this genotype as H3.3A^Δ; 12xHWT (Fig. 3a).

First, we conducted a proof-of-principle screen using single gene loss of function alleles or deficiencies covering genes with potential roles in histone function. We tested histone H3 chaperones (e.g. Asf1, Yem, Xnp), cell cycle regulators (e.g. E2F, Stg), and genes involved in the control of histone mRNA synthesis or chromatin regulation (e.g. Slbp, wge, Arts, Dre4). We performed crosses that produced H3.3A^Δ; 12xHWT progeny heterozygous for individual mutations or deficiencies and determined whether the progeny had reduced or increased viability compared to H3.3A^Δ; 12xHWT control siblings. Of the 23 mutations tested, heterozygosity of three deficiency mutations—Df(3R)BSC874 (Df 66), Df(3R)BSC500 (Df 67), and Df(3R)BSC501 (Df 68)—resulted in a significant reduction in viability of H3.3A^Δ; 12xHWT flies compared to control siblings, with only 8.4%, 57.9%, and 77.6% of expected surviving to adulthood, respectively (Fig. 3b, orange bar and Table 1). Interestingly, three other deficiencies Df(3R)ro80b (Df 81), Df(3R)BSC527 (Df 82), and Df(3R)Exel6210 (Df 83) resulted in a significant increase in viability of H3.3A^Δ; 12xHWT flies compared to control siblings, but we did not pursue these further (Fig. 3b and Table 1). Heterozygosity of the region covered by Df 81 results in an increase in viability independent of histone gene copy number (Fig. 3b, diamond). Notably, the three deficiencies (Df 66, Df 67, and Df 68) that reduce H3.3A^Δ; 12xHWT viability overlap the same 79.1 kb region on chromosome 3R (Fig. 3c). To further define the genomic interval responsible for the genetic interaction, we tested three additional deficiencies that partially overlap this region. Df(3R)BSC789 (Df 70), Df(3R)ED6310 (Df 72), and Df 83 did not result in a significant reduction in viability. Therefore, the region of interest is limited to a 26.4 kb region defined by the left breakpoints of Df 67 and Df 72 at genomic positions 98F10 and 98F12 (Fig. 3c). Seven annotated genes reside within this region of interest: Alg13, Atg14, Ctl2, dgt6, Pdhb, Vha100-1, and yem (Fig. 3c). Because Yem is an H3.3 specific chaperone, we next tested heterozygosity of a yem amorphic allele, which resulted in a significant reduction in viability, with only 66.7% of expected surviving to adulthood (Fig. 3b and Supplementary Table 3). One possible explanation for these genetic results is that heterozygosity of yem attenuates incorporation of H3.3 protein (derived from the H3.3B locus) into chromatin enough to reduce the viability of H3.3A^Δ; 12xHWT flies. This targeted screen confirms that our genetic paradigm can identify mutant loci that when hemizygous cause sensitivity to a reduction in H3.2 and H3.3 gene copy number.

Table 1.

Chromosome 3R and 3L mutation alleles tested in proof-of-principle screen for changes in viability.

Numbera	Deficiency	Start coordinateb	End coordinateb	P valuec
27	Df(3L)BSC289	1332329	1628100	-
45	Df(3L)ED4287	1795442	2551761	-
79	Df(3L)ED4284	1795442	1963552	-
80	Df(3L)BSC385	2259731	2417382	-
38	Df(3L)BSC730	12156077	12836424	-
51	Df(3L)ED4606	16080584	16773223	-
43	Df(3L)BSC220	18965662	19164368	-
77	Asf1[1]	19619559	19619559	-
78	E2F[rM729]	21626545	21626545	-
82	Df(3R)BSC527	22626930	23111808	+
71	Df(3R)ED6220	24543798	25183773	-
73	Df(3R)10-65	81F	81F	-
76	Df(3R)XNP[1]	25477868	25482834	-
81♦	Df(3R)ro80b	97D1	97D13	+
75	Df(3R)BSC460	27937830	28461658	-
83	Df(3R)Exel6210	28674961	28991018	+
66♦	Df(3R)BSC874	28675029	29191671	++
70	Df(3R)BSC789	28820134	29040507	-
67	Df(3R)BSC500	29112527	29675700	++
68	Df(3R)BSC501	29112527	29724685	+
84	Yem[2]	29119694	29122874	++
72	Df(3R)ED6310	29138895	29512153	-
74	Stg[4]	29252826	29255800	-
69	Df(3R)BSC547	29621303	29821399	-

♦ suggested haploinsufficiency.

Breakpoints are as determined by the Bloomington stock center.

+, P < 0.05; ++, P < 0.01. P-values obtained from a Chi-square test. See Supplementary Table 4 for P-values.

To expand our search for such loci, we screened the left arm of chromosome 3 (chr3L) using the Bloomington Stock Center Chr3L Deficiency Kit (Supplementary Table 8), which consists of 77 stocks that cover 97.1% of the chr3L euchromatic genome (Cook et al. 2012; Roote and Russell 2012). Fourteen of the deficiency mutations were excluded from the screen because they carry a mini-white genetic marker, resulting in an eye color that precludes identifying all progeny classes (Supplementary Table 8). Of the 60 deficiency mutations screened, two resulted in an increase in viability when heterozygous in H3.3A^Δ; 12xHWT flies (Fig. 4a, asterisks and Table 2), which we did not pursue further. By contrast, heterozygosity of eleven deficiencies caused significant reductions in viability of H3.3A^Δ; 12xHWT flies (Fig. 4a, asterisks and Table 2). Four of these deficiencies (Df 1, Df 4, Df 8, and Df 15) also caused a significant reduction in viability of siblings with one copy of H3.3A, one copy of HisC (100x), and the 12xHWT transgene (H3.3A^+/-; 112xHWT), suggesting haploinsufficiency (Fig. 4a, diamonds). We did not pursue these hits further. Interestingly, of the remaining hits Df(3L)BSC435 (Df 2) and Df(3L)BSC419 (Df 3) overlap the same 293 kb region of chr3L (Fig. 4b, orange). To map the genetic interaction in greater detail, we obtained two additional deficiencies—Df(3L)BSC418 (Df 52) and Df(3L)BSC836 (Df 36)—that overlap this same region of chr3L and observed no changes in viability of H3.3A^Δ; 12xHWT flies compared to control siblings (Fig. 4, a and b). Therefore, the genomic region spanning 77.8 kb on chr3L between cytological positions 78C6 and 78C8 impairs viability of flies with reduced H3.3 and H3 gene copy number.

Table 2.

Chromosome 3L deficiency alleles tested for changes in viability.

Numbera	Deficiency	Start coordinateb	End coordinateb	P valuec
54	Df(3L)ED50002	22947	128631	-
23	Df(3L)BSC362	306168	628171	-
59	Df(3L)Exel6085	548528	749210	-
1♦	Df(3L)ED4196	639583	1478937	++
27	Df(3L)BSC289	1332329	1628100	-
41	Df(3L)BSC800	1628101	1647451	-
46	Df(3L)BSC181	1688724	1841694	-
64	Df(3L)Aprt-32	1795318	2555775	+
45	Df(3L)ED4287	1795442	2551761	-
22	Df(3L)BSC119	2600282	2823614	-
35	Df(3L)Exel6092	2821245	3047162	-
32	Df(3L)BSC671	2982129	3193143	-
14	Df(3L)BSC672	3081311	3206906	-
37	Df(3L)ED4293	3226338	3250564	-
60	Df(3L)BSC368	3759821	4040635	-
55	Df(3L)BSC884	5601375	5770185	-
34	Df(3L)BSC410	5763773	6483285	-
39	Df(3L)BSC411	5969060	6618726	-
19	Df(3L)Exel6109	6736213	6936639	-
53	Df(3L)BSC27	6936605	7136086	-
58	Df(3L)BSC224	6957557	7150109	-
50	Df(3L)BSC33	7242439	7350373	-
21	Df(3L)BSC117	7242575	7328086	-
30	Df(3L)Exel8104	7353086	7522363	-
13	Df(3L)BSC375	7510880	7904179	+
12	Df(3L)BSC388	7643513	8184286	-
63	Df(3L)Exel6112	8089573	8351924	-
56	Df(3L)BSC815	8256164	8499740	-
11	Df(3L)BSC389	8415284	8582696	+
16	Df(3L)BSC816	8632181	8738462	-
24	Df(3L)ED4421	8738426	9377175	-
47	Df(3L)BSC113	9342609	9416591	-
65	Df(3L)AC1	9351951	10140553	+
17	Df(3L)BSC391	9446770	9697191	-
26	Df(3L)BSC118	9508772	9690291	-
44	Df(3L)BSC392	9671802	9892354	-
15♦	Df(3L)BSC673	9756714	10174058	+
48	Df(3L)BSC439	10507047	10964106	-
20	Df(3L)ED4470	11090089	11826330	-
49	Df(3L)ED4475	11580140	12401701	-
38	Df(3L)BSC730	12156077	12836424	-
25	Df(3L)BSC12	13037536	13221789	-
18	Df(3L)ED4543	13928325	14751140	-
33	Df(3L)BSC845	15504128	15819023	-
51	Df(3L)ED4606	16080584	16773223	-
6	Df(3L)ED4674	16654384	17042518	+
7	Df(3L)BSC414	16962973	17469226	+
62	Df(3L)Exel6132	17414682	17526191	-
57	Df(3L)ED4710	17487463	18139299	-
5	Df(3L)BSC775	17788244	18891426	+
43	Df(3L)BSC220	18965662	19164368	-
8♦	Df(3L)BSC839	20313247	20486308	+
10	Df(3L)BSC797	20445923	20942833	-
31	Df(3L)BSC449	20850015	21196030	-
40	Df(3L)BSC553	20984731	21219092	-
3	Df(3L)BSC419	21218032	21597878	++
2	Df(3L)BSC435	21304739	21770618	++
36	Df(3L)BSC836	21382499	21497772	-
52	Df(3L)BSC418	21382499	21637924	-
9	Df(3L)BSC223	21909520	22078536	-
4♦	Df(3L)BSC451	22069194	22684788	++
42	Df(3L)ED230	22127751	22827471	-
29	Df(3L)ED5017	22828597	22991401	-
61	Df(3L)1-16	24292305	24536634	-
28	Df(3L)6B-29	24977118	25115180	-

♦ suggested haploinsufficiency.

Breakpoints are as determined by the Bloomington stock center.

+, P < 0.05; ++, P < 0.01. P-values obtained from a Chi-square test. See Supplementary Table 5 for P-values.

Reduced histone H3 gene copy number disrupts Polycomb-mediated gene repression

Nine annotated genes reside within the defined genomic interval: CG12971, CG32436, CG32437, CG32440, ebd2, Pc, Rab26, Tbc1d8-9, and Tsr1 (Fig. 4b). To determine which of these genes contributes to viability of flies with reduced histone gene copy number, we generated H3.3A^Δ; 12xHWT animals heterozygous for single gene mutations. CG12971, CG32437, and CG32440 were not tested because no loss-of-function alleles exist. Heterozygous MiMIC transposon insertion alleles of ebd2, Rab26, Tbc1d8-9, Tsr1, and CG32436 did not impact viability of H3.3A^Δ; 12xHWT flies (Fig. 4c). However, three independent alleles of the Polycomb gene—Pc¹, Pc³, and Pc¹⁵—resulted in significant reductions in viability of H3.3A^Δ; 12xHWT flies (55.0, 61.8, and 52.4% of expected survive to adulthood, respectively) (Fig. 4c). These data suggest that H3.3A^Δ; 12xHWT flies are less viable when Polycomb function is reduced.

Polycomb group genes encode evolutionarily conserved regulators of cell identity. Polycomb complexes function to heritably silence expression of master regulator genes, including the Hox genes, which specify segmental identity (Kassis et al. 2017). In adult males, reduction of Polycomb function can result in homeotic transformations whereby the second (T2) and third (T3) thoracic legs acquire morphological features normally found only on the first thoracic legs (T1). This is most notably manifest by the appearance of sex combs on T2 and T3 legs, which normally only occur on T1 legs (Kaufman et al. 1980; Pattatucci et al. 1991). Based on our identification of the Pc gene in our genetic screen, we hypothesized that reduced histone gene copy number compromises Polycomb complex function. A prediction of this hypothesis is that reduced histone gene copy number would enhance Pc mutant phenotypes. Therefore, we evaluated the frequency and severity of homeotic transformations in H3.3A^Δ; 12xHWT animals heterozygous for a Pc null mutation (H3.3A^Δ; 12xHWT; Pc^3/+). We observed that H3.3A^Δ; 12xHWT; Pc^3/+ males exhibit an increased frequency of ectopic sex combs (100%, Fig. 5a, row 3) on T2 and T3 legs relative to Pc^3/+ (48%, Fig. 5a, row 1) or H3.3A^Δ; 12xHWT males (0%, Fig. 5a, row 2) (Table 3). Moreover, the expressivity of the ectopic sex comb phenotype is more severe in H3.3A^Δ; 12xHWT; Pc^3/+ animals relative to Pc^3/+ controls, often having a full set of sex combs on T2 and T3 legs (Fig. 5a, rows 1 and 3). Males and females of this genotype also exhibit defects in posterior wing morphology (Fig. 5b, row 3), suggesting partial wing to haltere transformation due to a failure to maintain proper repression of Ubx in the wing. Consistent with this hypothesis, immunostaining of H3.3A^Δ; 12xHWT; Pc^15/+ third instar imaginal wing discs revealed ectopic Ubx expression in the pouch region (Fig. 5c, row 3). We conclude that histone H3 gene copy number contributes to Polycomb function during development.

Fig. 5.

Reduced histone H3 gene copy number disrupts Polycomb-mediated gene repression. a) Scanning electron micrographs of first (T1), second (T2), and third (T3) thoracic legs from adult males of the indicated genotypes, depicted as in Fig. 1 with Pc gene dose in pink. Red brackets indicate the location where sex combs developed. Percent of T2 and T3 legs with ectopic sex combs indicated for each genotype. b) Bright field images of adult wings from the indicated genotypes. c) Confocal images of wing imaginal discs of the indicated genotypes stained for DAPI (blue) and Ubx (magenta). Red brackets indicate the wing pouch where ectopic Ubx expression occurred. Bars, 50 μM. d) Bar plots of viability of H3.3A^Δ; 12xHWT animals heterozygous for mutations of PRC1 and PRC2 Polycomb complex members. Data are plotted as in Fig. 3. Asterisks indicate statistical significance by Chi-square test (** P < 0.01, ♦ potential haploinsufficiency irrespective of H3 gene copy number, see Supplementary Table 3 for P-values). Pc³ was used in panels A and B. Pc¹⁵ was used in panel C.

Table 3.

Decrease in H3.2 and H3.3 gene copy number increases frequency of ectopic sex combs.

Genotype	# H3.3 genes	# H3.2 genes	% legs with ectopic sex combsa
H3.3B+/Y; H3.3A+/+, HisC+/+; Df(3L)BSC419/+	3	200	0%
H3.3B+/Y; H3.3A+/-, HisC+/-; Df(3L)BSC419/12xHWT	2	112	15%●
H3.3B+/Y; H3.3A-/-, HisC-/-; Df(3L)BSC419/12xHWT	1	12	40%●
H3.3B+/Y; H3.3A-/-, HisC-/-; +/12xHWT	1	12	0%
H3.3B+/Y; H3.3A+/-, HisC+/-; Pc3/12xHWT	2	112	50%▴
H3.3B+/Y; H3.3A-/-, HisC-/-; Pc3/12xHWT	1	12	100%▴
H3.3B+/Y; H3.3A+/+, HisC-/-; +/12xHWT	3	12	0%
H3.3B+/Y; H3.3A+/+; HisC+/-; Pc3/12xHWT	3	112	80%▪
H3.3B+/Y; H3.3A+/+, HisC-/-; Pc3/12xHWT	3	12	100%▪†
H3.3B-/Y; H3.3A-/-; HisC+/+;	0	200	0%
H3.3B+/Y; H3.3A+/+, HisC+/+; Pc3/+	3	200	48%
H3.3B+/Y; H3.3A+/-, HisC+/+; Pc3/+	2	200	95%*
H3.3B+/Y; H3.3A-/-, HisC+/+; Pc3/+	1	200	96%*
H3.3B-/Y; H3.3A-/-, HisC+/+; Pc3/+	0	200	100%*

Statistically significant difference in sex comb frequency between indicated genotype and H3.3B^+/Y; H3.3A^+/+, HisC^+/+; Df(3L)BSC419/+ (P < 0.0001).

Statistically significant difference in sex comb frequency between indicated genotype and H3.3B^+/Y; H3.3A^-/-, HisC^-/-; +/12xHWT (P < 0.0001).

Statistically significant difference in sex comb frequency between indicated genotype and H3.3B^+/Y; H3.3A^+/+, HisC^-/-; +/12xHWT (P < 0.0001).

Statistically significant difference in sex comb frequency between indicated genotype and H3.3B^+/Y; H3.3A^+/+; HisC^+/-; Pc³/12xHWT (P < 0.0001).

Statistically significant difference in sex comb frequency between indicated genotype and H3.3B^+/Y; H3.3A^+/+, HisC^+/+; Pc³/+ (P < 0.0001).

P-values obtained from a Fisher's exact test. See Supplementary Table 7 for P-values.

Next, we determined whether mutations in other Polycomb group genes cause effects similar to mutations of Pc when histone H3 gene copy number is reduced. H3.3A^Δ; 12xHWT flies heterozygous for null mutations in Sce and Ph, which encode members of Polycomb Repressive Complex 1 (PRC1), are viable at expected frequencies (Fig. 5d). Similarly, heterozygous mutations in Polycomb Repressive Complex 2 (PRC2) genes, E(z) and Su(z)12, do not cause reductions in viability (Fig. 5d). Moreover, RNA-seq revealed that no Polycomb group genes are differentially expressed in 12xHWT larval brains compared to control (Supplementary Table 10). Unlike other members of PRC1, Pc directly interacts with histones trimethylated at lysine 27 which may make it uniquely sensitive to histone H3.2 and H3.3 gene copy number.

Histone H3.2 and H3.3 gene copy number are each critical for Polycomb-mediated gene repression

Since both H3.2 and H3.3 gene copy number are reduced in H3.3A^Δ; 12xHWT animals, we next determined the individual requirement of either H3.2 or H3.3 gene copy number in Polycomb-mediated gene repression. We quantified viability and assessed whether Polycomb target genes were de-repressed in either 12xHWT or H3.3^Δ animals that were also heterozygous for a Pc null mutation (12xHWT; Pc^3/+ and H3.3^Δ; Pc^3/+, respectively). 12xHWT; Pc^3/+ animals are viable at expected frequencies (Fig. 6a) but have an increased frequency of ectopic sex combs on T2 and T3 legs like H3.3A^Δ; 12xHWT; Pc^3/+ animals (Fig. 6b, row 2 and Table 3). 12xHWT; Pc^3/+ animals also exhibit ectopic Ubx expression in the wing pouch of third instar imaginal discs and defects in adult posterior wing morphology (Fig. 6, c and d, row 2). 112xHWT; Pc^3/+ animals are viable at the expected frequency (Supplementary Table 6) but also exhibit increased frequencies of ectopic sex combs on T2 and T3 legs relative to Pc^3/+ controls (Table 3). Of note, the frequency of ectopic sex combs in 112xHWT; Pc^3/+ animals (80%) is significantly lower than 12xHWT; Pc^3/+ animals (100%) (Table 3). Thus, reducing H3.2 gene dose makes animals sensitive to reduced Polycomb function, and the severity of homeotic transformation is proportional to canonical histone gene copy number. We also found that normal H3.3 gene dose is necessary for Polycomb function. H3.3^Δ; Pc^3/+ animals are not fully viable, with only 59.6% of expected surviving to adulthood (Fig. 6a). H3.3^Δ; Pc^3/+ males also exhibit increased frequencies of ectopic sex combs on T2 and T3 legs (100%, Fig. 6b, row 4 and Table 3), Ubx de-repression in the wing pouch of third instar imaginal wing discs (Fig. 6c, row 4), and defects in adult posterior wing morphology (Fig. 6d, row 4). Animals with one or two copies of H3.3 survive to adulthood at a similar frequency—79.7 and 77.6% of expected, respectively—and animals with three copies of H3.3 are viable at the expected frequency (Fig. 6a). Consistent with these observations, males with only one copy of H3.3B also exhibit ectopic sex combs on T2 and T3 legs (Table 3). Taken together, these findings indicate that both H3.2 and H3.3 gene copy number are independently important for Polycomb-mediated gene repression, but viability is only affected by reduction in H3.3 gene copy number.

Discussion

In this study, we found that reducing either canonical or variant histone H3 gene copy number disrupts Polycomb-mediated gene repression. Two major protein complexes establish and maintain Polycomb-mediated repression: PRC1 and PRC2. PRC2 catalyzes H3K27me3, both PRC2 and PRC1 bind to H3K27me3, and PRC1 facilitates repression of local chromatin (Blackledge and Klose 2021). Mutations in core components of PRC1 and PRC2 disrupt the formation of these domains and cause transcriptional de-repression of Polycomb targets, such as Hox genes (Kennison and Tamkun 1988; Paro 1990; Orlando 2003). We found that reduction in canonical or variant H3 gene copy number results in homeotic transformations associated with de-repression of the Polycomb target genes Ubx (posterior wing transformation) and Scr (ectopic sex comb development in males). Consistent with these findings, a previous study in D. melanogaster found that heterozygosity of HisC suppresses homeotic transformation phenotypes in animals with a mutation that causes ectopic silencing of Polycomb target genes (Bajusz et al. 2001). Reduction of canonical histone gene copy number in D. melanogaster also modifies position-effect variegation, a phenomenon mediated by H3K9me3-marked constitutive heterochromatin (Moore et al. 1979; Moore et al. 1983; Sinclair et al. 1983). Moreover, deletion of all variant H3.3 gene copies in mouse embryonic fibroblasts results in disruption of heterochromatin domains at pericentromeric repeat regions, centromeres, and telomeres (Jang et al. 2015). Collectively, these findings indicate that H3.2 and H3.3 gene copy number play an important role in establishing efficient silencing via H3K9me3- and H3K27me3-mediated heterochromatin. We discuss below potential mechanisms for how changes in histone gene copy number might impact Polycomb-mediated repressive chromatin.

Histone protein abundance and stoichiometry may influence Polycomb-mediated repressive chromatin

Here, we show that although 12xHWT animals develop normally, reducing Pc gene dose by half in a 12xHWT background results in mutant phenotypes associated with impaired Polycomb-mediated gene repression. This genetic interaction suggests that the combination of reduced amounts of canonical histones and Polycomb prevents the proper formation of a repressive chromatin domain at Polycomb-silenced genes. However, our previous work found that canonical histone transcript levels are similar in 12xHWT and 200xWT animals, at least in early embryos (11), and here we found similar levels of total H3.3 and H3 protein in 12xHWT and 200xWT animals by western blotting. H3.2 and H3.3 are highly abundant proteins, and western blots may not provide the sensitivity needed to identify small changes in H3 protein levels that could be biologically meaningful. Moreover, a subtle decrease in histone protein abundance may result in changes in nucleosome occupancy that preferentially affect heterochromatin function. Polycomb chromatin domains have elevated nucleosome occupancy and decreased nucleosomal spacing and therefore may be particularly sensitive to changes in histone abundance (King et al. 2018). In fact, disruption of PRC1-mediated chromatin compaction in D. melanogaster results in de-repression of Hox genes (Bonnet et al. 2022). In addition to direct effects of decreased histone abundance at Polycomb target genes, it is also possible that indirect effects contribute to Polycomb target gene misregulation in 12xHWT animals. Previous work showed that reductions in the concentration of free histone H3 results in increased local histone recycling during replication in Xenopus egg extracts and D. melanogaster embryogenesis (Gruszka et al. 2020; Mühlen et al. 2023). Thus, another possibility is that reduced histone gene copy number results in an increased proportion of recycled histones within chromatin. If recycled histones carry PTMs that antagonize Polycomb function, such as H3K36me3, they could alter the PTM landscape and impact target gene repression at Polycomb domains (Finogenova et al. 2020; Bonnet et al. 2022; Mühlen et al. 2023; Salzler et al. 2023). Future studies examining chromatin accessibility and the PTM landscape at Polycomb target domains upon reduction in H3 gene copy number would help address these issues.

Nucleosomes and histone-chaperone complexes are multiprotein complexes that assemble with defined stoichiometries (Luger et al. 1997; Andrews and Luger 2011; Grover et al. 2018), and many genomic processes are sensitive to perturbations in subunit stoichiometry within these complexes. For instance, disrupting the stoichiometric balance between H2A:H2B dimers and H3:H4 dimers in yeast causes genome instability and mitotic chromosome loss (Meeks-Wagner and Hartwell 1986). In all the genotypes, we assessed that display mutant phenotypes indicative of impaired Polycomb repression, the balance between H3.2 and H3.3 gene copy number is altered, and this change in the relative abundance of H3.2 and H3.3 could impact genome regulation. Consistent with this interpretation, work done in mice shows that displacement of H3.3—and enrichment of replication dependent H3.1—at regulatory regions causes transcriptional deregulation and chromosomal aberrations (Chen et al. 2020). Thus, the stoichiometric balance between H3.2 and H3.3 in chromatin may be critical for maintaining Polycomb target gene repression in flies.

A corollary to this model is that proper stoichiometric balance between H3 proteins and their chaperones is needed for repression of Polycomb targets. Previous work in mouse cells shows that H3.3-specific chaperones interact with PRC1 and PRC2, and that these interactions are needed for the recruitment of H3.3 to H3K9me3-dependent heterochromatin and for the establishment of H3K27me3 at developmental gene promoters (Banaszynski et al. 2013; Liu et al. 2020). Therefore, one could posit that altering the stoichiometric balance between H3.3 and its chaperones may perturb the establishment or maintenance of Polycomb domains. Notably, in our genetic screen for viability we identified Yem, an H3.3-specific chaperone, suggesting that mechanisms regulating the levels of canonical and variant histone within the genome involve control of histone deposition into chromatin.

Distinct roles of canonical H3.2 and variant H3.3 in Polycomb-mediated gene repression

Canonical H3.2 and variant H3.3 differ in their expression patterns and protein sequence. Our genetic analyses demonstrate that reducing H3.3 gene copy number, but not H3.2 gene copy number, causes a decrease in viability of Pc heterozygotes, suggesting H3.2 and H3.3 may have non-identical roles in Polycomb target gene regulation. H3.3^Δ mutants do not uncouple H3.2 expression from S phase, and 12xHWT mutants still express H3.3 throughout all of interphase. Thus, depleting the pool of H3 outside of S phase in the H3.3^Δ mutants may sensitize cells to small perturbations in Polycomb-mediated gene repression during development. For example, Polycomb Response Elements (PREs), such as those that regulate Ubx, are sites of high histone turnover even though they reside within silent chromatin domains (Mito et al. 2007). As such, PREs may be particularly sensitive to the loss of available H3 protein outside of S phase. Reduced histone occupancy at PREs could impact Polycomb repression and organismal viability. Our finding that viability of animals with reduced H3.2 gene copies increases as H3.3 gene copy number increases supports this model. Alternatively, canonical H3.2 and variant H3.3 proteins may have distinct functions at PREs or Polycomb target domains. Consistent with this model, recent work in D. melanogaster shows that residue 36 on the N-terminal tail of H3.2 and H3.3 are both required for Polycomb-mediated repression but likely function through distinct mechanisms (Salzler et al. 2023). H3.2 and H3.3 differ at residue 31 on the N-terminal tail and H3.3S31 can be phosphorylated (H3.3S31ph). It is known that histone H3 PTMs can influence one another (Yuan et al. 2011; Finogenova et al. 2020). Notably, H3.3S31ph affects the local PTM landscape and binding of factors that interact with the H3 tail, like the H3K36me3 reader ZMYND11 (Armache et al. 2020; Sitbon et al. 2020). Conceivably, these H3.3S31ph-specific effects on the PTM landscape could impact Polycomb function. Future studies probing the impacts of an H3.3S31A mutation, which renders the residue non-modifiable, on Polycomb function would address this possibility.

In summary, our data investigating the control of canonical and variant histone abundance provide evidence that Polycomb-mediated gene repression is sensitive to both canonical and variant histone gene copy number. This work advances our understanding of the distinct and overlapping functions of canonical and variant histones.

Data availability

Strains are available upon request. Raw RNA-seq data were deposited to Gene Expression Omnibus (GEO) under accession GSE228058. Additional information is available from the corresponding authors upon request.

Supplemental material available at GENETICS online.

Funding

JEM and RLA were supported in part by the National Institutes of Health predoctoral traineeships: NIGMS T32GM135128 and NIGMS T32GM007092, respectively. LG was supported in part by a National Institute of Health administrative supplement to support undergraduate summer research to grant R35GM128851 (to DJM). This work was supported by the Nationals Institute of Health grants R35GM136435 (to AGM), R35GM145258 (to RJD), and R35GM128851 (to DJM).