Drosophila histone locus body assembly and function involves multiple interactions
The histone locus body (HLB) assembles at replication-dependent (RD) histone loci and concentrates factors required for RD histone mRNA biosynthesis. The Drosophila melanogaster genome has a single locus comprised of ∼100 copies of a tandemly arrayed 5-kB repeat unit containing one copy of each of the 5 RD histone genes. To determine sequence elements required for D. melanogaster HLB formation and histone gene expression, we used transgenic gene arrays containing 12 copies of the histone repeat unit that functionally complement loss of the ∼200 endogenous RD histone genes. A 12x histone gene array in which all H3-H4 promoters were replaced with H2a-H2b promoters (12xPR) does not form an HLB or express high levels of RD histone mRNA in the presence of the endogenous histone genes. In contrast, this same transgenic array is active in HLB assembly and RD histone gene expression in the absence of the endogenous RD histone genes and rescues the lethality caused by homozygous deletion of the RD histone locus. The HLB formed in the absence of endogenous RD histone genes on the mutant 12x array contains all known factors present in the wild-type HLB including CLAMP, which normally binds to GAGA repeats in the H3-H4 promoter. These data suggest that multiple protein–protein and/or protein–DNA interactions contribute to HLB formation, and that the large number of endogenous RD histone gene copies sequester available factor(s) from attenuated transgenic arrays, thereby preventing HLB formation and gene expression on these arrays.
An important organizing principle in cells is the use of membraneless compartments to spatially and temporally regulate diverse biological processes (Mitrea and Kriwacki, 2016). Numerous membraneless compartments have been identified in both the nucleus (e.g., nucleoli, Cajal bodies, histone locus bodies) and the cytoplasm (e.g., P-bodies, stress granules, germ granules) and are collectively referred to as biomolecular condensates (Banani et al., 2017). There is increasing evidence suggesting that biomolecular condensates are formed through liquid–liquid phase separation or condensation (Alberti et al., 2019). This occurs when proteins and/or nucleic acids in the nucleoplasm or cytoplasm coalesce or demix into a condensed phase that often resembles liquid droplets. Large nuclear condensates that are visible under light microscopy are most often referred to as nuclear bodies (NBs) and represent an important organizing feature of the nucleus.
The histone locus body (HLB) is a conserved NB that assembles at replication-dependent (RD) histone genes and concentrates factors required for RD histone mRNA biogenesis (Duronio and Marzluff, 2017). RD histone mRNAs are the only eukaryotic mRNAs that are not polyadenylated (Marzluff and Koreski, 2017). The unique stem loop at the 3′-end of RD histone mRNAs results from a processing reaction requiring a specialized suite of factors, some of which are constitutively localized in the HLB (Duronio and Marzluff, 2017). The HLB provides a powerful system to study how NBs form and function because it contains a well-characterized set of factors involved in producing a unique class of cell–cycle-regulated mRNAs. We previously demonstrated that concentrating factors (e.g., FLASH [FLICE-associated huge protein] and U7 snRNP) in the Drosophila melanogaster HLB is critical for efficient histone pre-mRNA processing (Wagner et al., 2007; Tatomer et al., 2016). However, a full understanding of how the HLB participates in histone mRNA biosynthesis requires knowledge of HLB assembly at the molecular level.
Prior studies of NBs have provided several important assembly concepts that are applicable to the HLB. Many NB components have an intrinsic ability to self-associate, an observation leading to two different models of NB assembly: 1) interactions among NB components occur stochastically, wherein individual factors can be recruited to the body in any order; or 2) components assemble in an ordered or hierarchical pathway, wherein the recruitment of components is predicated on prior recruitment of others (Dundr and Misteli, 2010; Rajendra et al., 2010). The HLB appears to employ a hybrid version of these two possibilities. For example, genetic loss of function experiments suggest a partially ordered assembly pathway of the Drosophila HLB with some components being required for subsequent recruitment of others (White et al., 2011; Terzo et al., 2015; Tatomer et al., 2016). The scaffolding protein Mxc (multi sex combs), the Drosophila ortholog of human NPAT (nuclear protein, ataxia-telangiectasia locus), and FLASH likely form the core of the HLB and are required for subsequent recruitment of other factors (White et al., 2011). Tethering experiments in mammalian cells indicate that ectopic HLB formation also may be induced by several different HLB components, supporting a stochastic model of assembly (Shevtsov and Dundr, 2011).
The initiation event in self-organizing NB assembly is the key step in the process and is not well understood. A prevalent model postulates a “seeding” event that initiates the nucleation of critical components that form a platform for further recruitment of other components (Dundr, 2011; Shevtsov and Dundr, 2011; Altmeyer et al., 2015; Falahati et al., 2016; Stanek and Fox, 2017; Gomes and Shorter, 2019). In some instances, RNA is thought to help seed NB assembly, and NBs such as the nucleolus and ectopic paraspeckles can form at sites of specific transcription (Matera et al., 2009; Mao et al., 2011; Falahati et al., 2016). Blocking transcription prevents complete HLB assembly in both zebrafish and flies (Salzler et al., 2013; Heyn et al., 2017; Hur et al., 2020). However, the HLB is present at RD histone genes even in G1 when the genes are not active, raising the possibility that histone genes themselves participate in seeding HLB assembly (Zhao et al., 1998; Liu et al., 2006).
In Drosophila, the RD histone genes are present at a single locus with ∼100 copies of a tandemly arrayed 5-kb repeat unit, each of which contains one copy of the divergently transcribed H2a-H2b and H3-H4 gene pairs as well as the gene for linker histone H1 (Lifton et al., 1978; McKay et al., 2015; Bongartz and Schloissnig, 2019). Using transgenes containing a wild-type or mutant derivative of a single histone repeat, we previously demonstrated that the bidirectional H3-H4 promoter stimulated HLB assembly and transcription of the single histone repeat in salivary glands (Salzler et al., 2013). We subsequently showed that the conserved GAGA repeat elements present in the H3-H4 promoter region are targeted by the zinc-finger transcription factor CLAMP (chromatin-linked adaptor for MSL proteins), and that this interaction promotes HLB assembly (Rieder et al., 2017). Thus, the H3-H4 promoter region might act to seed HLB assembly.
In this work, we leveraged transgenic histone gene arrays to test whether the H3-H4 promoter region is necessary for in vivo function of the RD histone locus. We found that replacement of H3-H4 promoters with H2a-H2b promoters results in an attenuated transgenic histone gene array that does not function in the presence of the intact endogenous RD histone locus, but surprisingly provides full in vivo function, including normal HLB assembly and histone gene expression, when the endogenous RD histone locus is absent. These results suggest that multiple elements in the histone genes and core HLB proteins are involved in HLB assembly.
To study histone gene regulation and the role of histone posttranslational modifications in chromatin, we previously developed an experimental system in which BAC-based, transgenic histone gene arrays containing 12 copies of the 5-kb histone repeat unit assemble HLBs and functionally complement homozygous loss of the ∼100 gene copies at the endogenous RD histone locus, HisC (McKay et al., 2015). To study histone gene expression, we created a 12xHWT (Histone Wild Type) transgenic construct containing a synonymous polymorphism in the H2a gene (i.e., mutation of a XhoI site) that allows us to distinguish transgenic H2a gene expression from endogenous H2a gene expression (McKay et al., 2015). Here, we extended this design and created a “Designer Wild Type” (DWT) 5-kb histone repeat unit that has all five histone genes marked in a similar manner (Figure 1A; Supplemental Figure S1A). We also introduced restriction enzyme sites around the mRNA 3′-end processing signals, the stem loop and histone downstream element (Marzluff and Koreski, 2017), to allow us to readily manipulate those sequences (Supplemental Figure S1). We used this DWT repeat to create a 12xDWT construct and found that a transgenic version of this array behaved identically to the original 12xHWT array. It rescued the lethality caused by homozygous deletion of HisC, resulting in viable, fertile adult flies. In addition, we can maintain a stock of flies lacking the endogenous RD histone locus and containing a homozygous 12xDWT transgene.
The H3-H4 promoter stimulates HLB formation
To test whether the H3-H4 promoter region is necessary for histone locus function in vivo, we engineered a derivative of the 12xDWT array in which all H3-H4 promoters were replaced with H2a-H2b promoters. In this “Promoter Replacement” (12xPR) construct, we replaced the entire 298-nt sequence between the initiation codons of the divergently transcribed H3 and H4 genes with the 226-nt sequence between the initiation codons of the divergently transcribed H2a and H2b genes, leaving the H1 and H2a and H2b genes intact (Figure 1A; Supplemental Figure S1B). One difference between the two promoters is that the H2a-H2b region lacks the CLAMP-binding GAGA repeats present in the H3-H4 promoter (Rieder et al., 2017). Otherwise it is an intact, native bidirectional histone gene promoter. Consequently, the 12xPR construct should retain the ability to initiate transcription from all RD histone genes. The 72-nt size difference between the promoters provides a way to unambiguously distinguish between the 12xPR and the 12xDWT array genotypes using PCR primers in the H3 and H4 coding regions in flies heterozygous (Figure 1B) or homozygous (Figure 1C) for the HisC deletion.
We first assessed whether the 12xPR and 12xDWT transgenes could support HLB formation in the presence of the endogenous histone genes in polytene chromosome spreads from 3rd instar larval salivary glands. In these polyploid cells, the genome is amplified greater than 1000-fold and sister chromatids line up in register, resulting in large chromosomes providing high-resolution cytology. We visualized the HLB by immunofluorescence using antibodies that recognize one of several HLB components. These components include Mxc, the Drosophila ortholog of human NPAT, which is necessary for HLB assembly and histone gene expression (White et al., 2011; Terzo et al., 2015); histone pre-mRNA processing factors FLASH (Yang et al., 2009) and Lsm11, a component of the U7 snRNP (Liu et al., 2006; Godfrey et al., 2009; Burch et al., 2011); and muscle wasted (Mute), an ortholog of human YARP (YY1AP-related protein 1) that is a putative repressor of histone gene expression (Bulchand et al., 2010). In these preparations, we also visualized both the endogenous histone locus and the ectopic, transgenic histone genes by FISH using a probe derived from the entire histone repeat unit. We observed HLB formation at the control 12xDWT transgenic locus but not at the 12xPR transgenic locus (Figure 1D). These data are consistent with our previous results that sequences within the H3-H4 promoter are necessary for HLB assembly in the presence of the endogenous histone genes (Salzler et al., 2013).
Mutant, single copy transgenic histone gene repeats that fail to form an HLB (Salzler et al., 2013), and Mxc mutants that don’t form an HLB (Terzo et al., 2015), result in reduced histone mRNA levels. These data suggest that the HLB is necessary for efficient histone mRNA biosynthesis. We therefore tested whether the 12xPR transgene could support histone gene expression in the absence of HLB assembly. To determine expression from the ectopic 12xDWT or 12xPR genes in the presence of the endogenous RD histone genes, we randomly prime cDNA from total RNA and then amplify a fragment of each coding region containing the restriction enzyme site that is changed in the DWT construct. By digesting the PCR fragment with the appropriate restriction enzyme and resolving the fragments by gel electrophoresis, we can determine the relative level of expression from the transgenic and endogenous RD histone genes. For example, RT-PCR of the H2a gene results in the same size amplification product from both the transgenic and endogenous genes, but only the product from the endogenous histone genes is sensitive to digestion with XhoI (Figure 1E). When assayed in the presence of the endogenous genes, we found that H2a was expressed from the 12xDWT at a much higher level than from the 12xPR transgene, even though identical promoters drove H2a in each transgenic array (Figure 1E; Figure 2, A, lanes 2 and 4, and B, lanes 2 and 3). These data are consistent with our previous observations using transgenes with a single histone gene repeat unit (Salzler et al., 2013). Together these results demonstrate that the H3-H4 promoter is required both for HLB formation and for histone gene expression in the presence of the endogenous RD histone genes.
A histone gene array lacking the H3-H4 promoter forms HLBs and is expressed in the absence of the endogenous genes
To assess the biological activity of the 12xPR transgene, we determined the developmental outcome of having a 12xPR transgene as the only zygotic source of histone mRNA. Due to large stores of maternal histone protein and mRNAs, embryos homozygous for a deletion that removes the endogenous RD histone gene array develop normally through S phase of cell cycle 14, but require zygotic RD histone gene expression for progression through S phase of cycle 15 and beyond (Gunesdogan et al., 2010, 2014). Consequently, embryos lacking RD histone genes arrest at nuclear cycle 15 and do not hatch. As noted above, this embryonic lethality is rescued by a single 12xDWT transgene, which supports development of histone deletion progeny into viable, fertile adult flies. Surprisingly, we found that embryos lacking endogenous histone genes and containing the 12xPR transgene hatched and developed into nearly the expected number of fertile adults without any overt developmental delays, similar to embryos lacking endogenous histone genes and rescued by the 12xDWT transgene. These data suggest that the 12xPR transgene provides normal amounts of histone gene function when the endogenous RD histone genes are absent.
We interrogated this unexpected result more thoroughly by analyzing 1) histone gene expression, 2) HLB formation, and 3) histone pre-mRNA processing in both 5- to 7-h-old embryos and 3rd instar larvae lacking endogenous histone genes and containing either the 12xDWT or the 12xPR transgene. In both genotypes, all histone mRNA in either 5- to 7-h-old embryos (Figure 2A, lanes 3 and 5) or 3rd instar larvae (Figure 2B, lanes 4 and 5) was derived only from the ectopic histone gene array, indicating that the maternal histone mRNA stores had been degraded normally. We observed robust histone gene expression from the 12xPR transgene when the endogenous histone genes are absent, in stark contrast with the low level of expression from the 12xPR transgene when endogenous histone genes are present (Figure 2, A, lane 4, and B, lane 3).
High levels of histone gene expression are strongly correlated with the ability to form an HLB (Salzler et al., 2013; Terzo et al., 2015; Rieder et al., 2017). Given that the 12xPR can support histone gene expression in the absence of the endogenous genes, we assayed for HLB formation in polytene chromosome spreads from 3rd instar larval salivary glands and in embryos. We detected robust HLB formation at the 12xPR transgenic locus with antibodies against FLASH and Lsm11 (Figure 2C), or Mxc and Mute (Figure 2D), similar to that observed at the 12xDWT locus. Thus, the 12xPR transgene, which lacks H3-H4 promoter sequences, can support HLB formation and normal levels of histone gene expression in the absence of endogenous RD histone genes.
Histone mRNA from the 12x PR transgenic locus is properly processed
An important function of the HLB is concentrating factors to promote efficient histone pre-mRNA processing (Tatomer et al., 2016). Therefore, we reasoned it was possible that the attenuated 12xPR gene array might affect other aspects of histone mRNA biosynthesis, including pre-mRNA processing. All Drosophila RD histone genes contain cryptic polyadenylation signals downstream of the normal processing sites that are only used when the histone pre-mRNA processing reaction is compromised, resulting in the production and accumulation of poly(A)+ histone mRNA (Lanzotti et al., 2002; Godfrey et al., 2009; Tatomer et al., 2016). We examined pre-mRNA processing efficiency in flies rescued by the 12xPR transgene. Our PCR-based assays to detect histone mRNA expression cannot differentiate between properly processed and misprocessed histone mRNAs. We therefore used Northern blotting with a probe against the H3 coding region to determine whether histone pre-mRNA was efficiently processed. In contrast to a FLASH mutant, which expresses large amounts of poly(A)+ histone mRNA (Tatomer et al., 2016), we did not detect poly(A)+ histone mRNA from early embryos (Figure 2E) or from whole 3rd instar larvae (Figure 2F) in HisC deletion animals rescued by the 12xPR transgene. Note also that the levels of histone mRNA from the 12xPR transgenes are similar to wild-type levels. These results indicate that the HLB formed on the 12xPR array in the absence of endogenous histone genes supports efficient histone pre-mRNA processing.
HLB assembly at the 12x PR transgenic locus occurs at the normal time during embryogenesis
The above data suggest that the 12xPR represents an attenuated histone gene array that provides normal biological function when not in competition with the wild-type endogenous histone genes. To further explore this model, we determined if an HLB assembles on the 12xPR array in the early embryo at the same time that it assembles on the 12xDWT array. The HLB begins assembling in syncytial blastoderm embryos just prior to the onset of zygotic histone transcription (White et al., 2007, 2011). We previously reported that a “proto-HLB” consisting of Mxc and FLASH forms in nuclear cycle 10, followed by the onset of zygotic histone gene expression and further recruitment of additional HLB components (Mute and Lsm11) in cycle 11 (White et al., 2011; Salzler et al., 2013). To determine if the HLB forms at the 12xPR transgenic locus with normal timing, we stained syncytial blastoderm embryos lacking endogenous histone genes and rescued by a 12xPR transgene with antibodies against Mxc (Figure 2G). In these experiments, HLB assembly during the syncytial blastoderm cycles was indistinguishable from that of histone deletion embryos rescued by the control 12xDWT transgene. Thus, in the absence of the endogenous genes, an HLB assembles on the 12xPR array at the same time in early development as it does on a wild-type array.
Note that zygotic histone transcription occurs coincident with HLB formation in syncytial blosterm cycle 11 (Terzo et al., 2015; Hur et al., 2020). HLBs increase in size in each cycle after they are formed (Hur et al., 2020). Given how short S phase is at this stage of embryogenesis, the levels of histone mRNA formed during these rapid cell cycles are likely very small and are not necessary for development since embryos can progress through cycle 14 without any histone genes using only the maternal supply of histone proteins and mRNAs. Only starting in cycle 15 are large amounts of zygotic histone mRNAs required (Smith et al., 1993; Gunesdogan et al., 2010, 2014).
CLAMP is present in the 12x PR HLB in the absence of the endogenous RD histone genes
The H3-H4 promoter is highly conserved among Drosophilids and contains conserved GAGA repeats (Salzler et al., 2013; Rieder et al., 2017), which we previously showed are essential for HLB formation and expression of RD histone genes in the presence of the endogenous histone locus (Rieder et al., 2017). Although in vitro these repeats can bind both the zinc-finger GA-repeat binding protein CLAMP and the major Drosophila GAGA repeat binding protein GAF (GAGA factor; trithorax-like) (Gilmour et al., 1989), only CLAMP is bound to the histone locus in wild-type animals (Rieder et al., 2017). The H2a-H2b promoter and the rest of the histone repeat unit do not contain any GAGA repeats longer than 4 nts, and CLAMP preferentially binds to longer GAGA repeats (Kuzu et al., 2016). We asked whether CLAMP is recruited to the 12xPR transgene in salivary gland polytene chromosomes. As with all other HLB components we tested, CLAMP is not recruited to the 12xPR transgene or a 12x histone gene array in which the GAGA sequences are replaced with lacO binding sites (GAM, Figure 3A) when the endogenous RD histone genes are present (Rieder et al., 2017). Surprisingly, we found that in the absence of endogenous RD histone genes, CLAMP (Figure 3C), but not GAF (Figure 3E), is recruited to the 12xPR transgenic locus with similar intensity to CLAMP recruitment to the 12xDWT transgenic locus (Figure 3, B and D). Furthermore, the 12x GAM array also supports H2a gene expression in the absence of the endogenous genes (Figure 3F). These data indicate that CLAMP can be recruited to the HLB in a histone gene array lacking GAGA repeats when the GAGA binding sites within the H3-H4 promoter at the endogenous RD histone locus have been removed. The H3 and H4 genes are transcribed, indicating that CLAMP is not an essential DNA-binding transcription factor for the H3 and H4 genes, but likely serves as a factor that alters chromatin structure (Rieder et al., 2017).
We carried out ChIP-qPCR and ChIP-seq experiments on HisC deletion embryos containing only the 12xDWT array or only the 12xPR array to determine how CLAMP is interacting with the histone gene array. In agreement with ChIP-seq results on the endogenous histone genes, which showed that CLAMP is localized precisely to the H3-H4 promoter (Rieder et al., 2017), CLAMP was bound to the H3-H4 promoter but not to the H2a-H2b genes or promoter on the 12xDWT array (Figure 4, A–C). In contrast, CLAMP was not enriched at any location on the 12xPR array (Figure 4, A, B, and D). As in wild-type flies, in the 12xPR embryos CLAMP bound the endogenous rox1 locus on the X chromosome, a locus to which CLAMP is recruited in males for its role in X chromosome dosage compensation (Figure 4, B, E, and F) (Soruco et al., 2013). Thus, CLAMP is not specifically enriched at any location in the entire histone gene repeat unit when the H3-H4 promoter is replaced with H2a-H2b promoter in a 12x transgenic array, despite our observation using microscopy that CLAMP is present in the HLB at the 12xPR array (Figure 3C). These data suggest that CLAMP can be recruited to RD histone genes by protein–protein interactions independently of its binding to specific DNA sequences in the histone gene repeat.
A wild-type array can activate the 12x PR array when present in trans at the homologous locus
The results above demonstrate that the 12xPR transgenic locus does not form an HLB in the presence of the endogenous histone genes unlike the wild-type 12xDWT transgene, which does. We tested whether juxtaposing a wild-type histone gene array near the 12xPR locus would nucleate a functional HLB and activate expression from the 12xPR transgene. Our BAC-based transgenes are inserted into the genome via site-specific recombination at the same chromosomal location. We created flies in which the 12xPR was placed at position VK33 on chromosome 3 in trans to the 12xHWT transgene used in our initial studies (McKay et al., 2015) and analyzed these transgenes in the presence of the endogenous RD histone genes on chromosome 2. We examined HLB formation at the ectopic VK33 location by staining whole mount salivary glands with antibodies against Mxc and FLASH. Using this approach, we detect only a single, large HLB on the endogenous RD histone genes in wild-type nuclei lacking any transgenes (Figure 5, A-1 and B). When 12xPR was the only transgene present in addition to the endogenous RD histone genes, none of the nuclei contained an ectopic HLB and we only detected the endogenous HLB (Figure 5, A-2 and B), consistent with the results of staining spread salivary gland polytene chromosomes (Figure 1D). In contrast, we observed formation of a second small HLB in addition to the single large endogenous HLB in 100% of the 12xHWT/12xPR nuclei (Figure 5, A-5 and B), as expected since this HLB also forms when only the 12xHWT array is present at this site.
We measured histone gene expression from the 12xPR transgene in these genotypes. In the 12xHWT array the histone H2a gene, and not the other histone genes, is marked with a restriction site change (McKay et al., 2015), whereas the 12xPR array has all five histone genes marked with a restriction site change. Therefore, to specifically detect histone gene expression from the 12xPR transgene, we digested embryonic H3 RT-PCR products with Sac I, whose recognition site is missing from the 12xPRH3 gene but is present in both the endogenous and the 12xHWTH3 genes (Figure 5C). Strikingly, the 12xPR transgene supports stronger H3 gene expression in the presence compared with the absence of the endogenous histone genes when present in trans with the 12xHWT transgene (Figure 5C, lanes 1 and 2). These data demonstrate that the wild-type histone sequences in the 12xHWT were able to activate the 12xPR transgenic locus in the presence of the endogenous histone genes, likely by nucleating ectopic HLB formation that encompasses the paired homologous chromosomes. This result is similar to transvection, a phenomenon in which transcription of a gene can be activated by an enhancer located in trans through pairing of homologous chromosomes (Duncan, 2002; Fukaya and Levine, 2017).
Transgenes with a single copy of the wild-type repeat partially activate a 12x array in trans
Since 12 copies of a wild-type histone repeat at the same integration site-stimulated expression from the 12xPR transgene, we tested whether a single copy could do the same using two different approaches; one where the single copy is present in trans and the other where it is present in cis. For the first approach, we inserted a 1x HWT transgene at the VK33 integration site. In these strains we detected an ectopic HLB in 90% of the salivary gland nuclei in the presence of the endogenous genes (Figure 5B), similar to what we observed previously (Salzler et al., 2013). When this 1x HWT transgene was placed in trans with the 12xPR array, we detected an ectopic HLB in 94% of the nuclei examined (Figure 5, A-6 and B). We tested whether there was any activation of expression of the 12xPR array when present in trans with 1x HWT using our RT-PCR assay. There was a modest activation of the expression 12xPR array, but much weaker than when the 12xHWT was in trans with the 12xPR as measured by the ratio of the bands (Figure 5D, lanes 1 and 3).
We next tested if including a single wild-type repeat in the 12xPR array could stimulate HLB formation and histone mRNA expression from that array in the presence of the endogenous genes. We created a 12x array (12xPR-1) containing one wild-type histone repeat unit in the center of 11 PR repeat units (Figure 5E). Like 12xPR, the 12xPR-1 transgene rescued a deletion of the endogenous histone genes, resulting in viable and fertile adults and indicating that it is likely fully active when present as the only source of RD histone genes. We then examined HLB formation in intact salivary glands from animals containing both the 12xPR-1 transgene and the endogenous RD histone genes. In this genotype, we detected HLB formation at the ectopic 12xPR-1 transgenic locus in 59% of the nuclei (Figure 5, A-4 and B), compared with 0% of nuclei from the 12xPR. We also analyzed expression from this array in the presence of the endogenous RD histone genes and observed only modest activation of histone mRNA expression (Figure 5D, lane 2) and less than the level of expression when the 1x WT was in trans to the 12x array (Figure 5D, lane 3). Thus, a single histone repeat unit can stimulate HLB formation, but activates expression of neighboring genes better when placed in trans (i.e., on the other homolog) rather than in cis (i.e., in the same 12x array).
We next tested a transgene containing only the H3-H4 promoter with no coding or 3′-end sequences, which we previously demonstrated could form an HLB in salivary glands when inserted into a ectopic site (Salzler et al., 2013). Ectopic HLBs were detected in 70% of the salivary glands when only the H3-H4p promoter transgene was present and 69% when it was placed in trans to the 12xPR array (Figure 5, A-3, A-7, and B). We assayed the expression of the H3 gene from the 12xPR when the H3-H4p transgene was in trans to the 12xPR array. We detected low level expression, similar to the expression detected when the 1x WT transgene was present opposite the 12xPR array (Figure 5D, lane 4). Collectively, these data demonstrate that the presence of an HLB nucleating sequence in trans to 12xPR can induce formation of an ectopic HLB and histone gene expression in the presence of the endogenous histone genes. Further, these data emphasize that the H3-H4 promoter is a critical element in promoting ectopic HLB formation in the presence of the endogenous histone genes, consistent with our previous observations (Salzler et al., 2013; Rieder et al., 2017).
In this study, we used our histone gene replacement platform to analyze the cis acting elements within the Drosophila histone repeat unit that are necessary for HLB formation and histone gene expression. Previously we showed using a single, transgenic histone gene repeat unit that the promoter region of the divergently transcribed H3-H4 gene pair is capable of stimulating HLB formation (Salzler et al., 2013). We subsequently further mapped this functionality using a 12x gene array to conserved GAGA repeats in this region that are targeted by the CLAMP protein (Rieder et al., 2017). Here, we present evidence that a 12xPR histone gene array devoid of the H3-H4 promoter and lacking any CLAMP binding elements cannot assemble an HLB in the presence of the ∼100 RD histone gene copies at the endogenous locus (HisC). However, the 12xPR array surprisingly can rescue homozygous deletion of HisC and fully support the entire Drosophila life cycle. In the HisC deletion background, the 12xPR array assembles an HLB and expresses the same amount of properly processed histone mRNAs as the endogenous genes or as a 12xDWT wild-type array. Below we discuss the implications of these observations on HLB assembly and organization.
The RD histone locus stimulates HLB formation in Drosophila
Biomolecular condensates form via a seeding event that promotes a high concentration of factors at a discrete location, leading to recruitment of additional factors that ultimately result in a structure that can be observed by light microscopy (Gomes and Shorter, 2019). A number of putative seeding events for biomolecular condensates have been described (Dellaire et al., 2006; Dundr, 2011; Mao et al., 2011; Shevtsov and Dundr, 2011; White et al., 2011), but in many cases the precise mechanism of seeding is not known. Nucleic acids, particularly RNA, have been proposed to seed different NBs. Both the nucleolus and the HLB are associated with specific genomic loci (Mao et al., 2011; Shevtsov and Dundr, 2011; Salzler et al., 2013), and it is likely that the DNA (or chromatin) and/or nascent RNA at the locus participates in the seeding event. The activation of zygotic transcription of rRNA leads to the precise spatiotemporal formation of the nucleolus in Drosophila embryos (Falahati et al., 2016). In the absence of rDNA, Drosophila nucleolar components still form high concentration assemblies, but these are smaller, more numerous, and do not form at the same time in the early embryo as the wild-type nucleolus.
Drosophila HLB components also stochastically assemble smaller and more unstable foci in embryos lacking the RD histone locus (White et al., 2007; Salzler et al., 2013; Hur et al., 2020), suggesting that HLBs and the nucleolus form similar seeding events. Indeed, the dynamics of HLB assembly in single early Drosophila embryos display properties consistent with liquid–liquid phase transition seeded by HisC (Hur et al., 2020). Blocking transcription in the early embryo prevents normal HLB growth (Hur et al., 2020), and a defective H3-H4 promoter (with mutated TATA boxes) does not support HLB formation in the context of a single copy histone gene repeat in salivary glands (Salzler et al., 2013). These data suggest that active transcription is essential for forming a complete HLB.
It is important to note that HLBs assemble and persist in nonproliferating Drosophila tissues that do not express histone mRNAs (Liu et al., 2006; White et al., 2007) and are also present in G0/G1 mammalian cells (Ma et al., 2000; Zhao et al., 2000). Histone gene expression is activated as a result of phosphorylation of Mxc/NPAT by Cyclin E/Cdk2, resulting in changes in the HLB that promote histone gene transcription and pre-mRNA processing (Wei et al., 2003; Ye et al., 2003; White et al., 2011). We propose that in early embryonic development the histone locus DNA and/or chromatin seeds HLB assembly in Drosophila, with the H3-H4 promoter region being particularly important. We further propose that subsequent transcriptional activation of histone genes then drives HLB growth and maturation.
Formation of an HLB on a transgenic RD histone gene array requires that this array compete effectively with the endogenous HisC locus for recruitment of HLB components. This is the situation with 12xHWT and 12XDWT arrays, which form HLBs in the presence of HisC. These results also indicate that there are no other elements within HisC that are necessary for HLB formation. Because the 12xPR array does not form an HLB in the presence of HisC, but it does so in the absence of HisC, we hypothesize that the endogenous RD histone gene array sequesters critical HLB components, likely including Mxc and CLAMP, thereby preventing HLB assembly at the transgenic locus. By removing the H3-H4 promoter from the transgene, we removed an element that provided additional interactions with HLB components, notably CLAMP, weakening the overall ability of the locus to stably nucleate an HLB.
HLB assembly involves multiple interactions among components
Interactions among multivalent proteins, or multivalent protein–nucleic acid interactions, are driving forces in the assembly of biomolecular condensates (Shin and Brangwynne, 2017). Mxc is likely the critical factor that together with histone genes seeds Drosophila HLB formation and activates histone gene expression. Mxc is a large (∼1800 aa) protein that oligomerizes in vivo and likely provides a scaffold for multivalent interactions (Terzo et al., 2015). A C-terminal truncation mutant of Mxc that fails to recruit histone pre-mRNA processing factors still forms an HLB and activates histone gene expression at sufficient levels to complete development, underscoring the multivalent nature of Mxc (White et al., 2011; Landais et al., 2014; Tatomer et al., 2016).
Surprisingly, the HLB that assembles on the 12xPR array in the absence of HisC contains CLAMP, even though we have removed all of the known CLAMP binding sites from the histone repeat. Although CLAMP may bind another sequence in the 12xPR array, no other favorable GAGA repeats are present, and we were unable to detect CLAMP bound to any other location in the histone array by ChIP-qPCR and ChIP-seq experiments. More likely, CLAMP interacts with other HLB components, possibly Mxc or the Mxc-FLASH complex, providing multivalent contacts between CLAMP and other HLB components. Deleting the GAGA sequences from the H3-H4 promoter did not affect transcription of the H3 or H4 genes in the absence of HisC, suggesting that CLAMP’s major function is to promote HLB assembly and not to act as a canonical DNA binding transcription factor. Supporting this interpretation is the observation that another, more abundant transcription factor that binds to GAGA repeats, GAF, is not found at the HLB unless CLAMP is absent (Rieder et al., 2017), consistent with CLAMP’s critical interactions with both the GAGA repeats and the HLB factors in seeding the HLB.
Because the 12xPR array is capable of assembling a completely functional HLB in the absence of HisC, the H3-H4 promoter is not absolutely essential for HLB formation. One possibility is that there are multiple pathways for assembling functional HLBs. Previous work suggests that not all seeding events are equivalent in their ability to assemble biomolecular condensates. In artificial systems, changes in scaffold stoichiometry, which can stem from changes in valence, alter the recruitment of components (Banani et al., 2016). Further, mathematical modeling has revealed that scaffolds can nucleate distinct complexes when at different concentrations, and that this can qualitatively alter the transcriptional output (Yang and Hlavacek, 2011). Additionally, P-bodies can form in multiple ways through different protein–protein or protein–nucleic acid interactions, with different interactions predominating under different conditions (Rao and Parker, 2017). Therefore, different nucleators of the HLB (i.e., the H3-H4 promoter or other sequences in the locus) may result in similar but not identical outcomes. Collectively our results suggest that HLB formation results from the contribution of many molecular interactions, and the loss of any single one may be overcome by other multivalent interactions within the body.
MATERIALS AND METHODS
Culture condition and fly strains
Transgenic histone gene array construction
Construction of the 5-kb histone repeat designed for this study was performed using the HiFi DNA Assembly system from New England Biolabs. PCR amplification of fragments from existing histone repeats, in addition to in vitro synthesized gblocks (IDT) containing the desired nucleotide changes, was employed for the building blocks of the reaction. Manufacturer’s protocol was followed with slight modification to the incubation time of the reaction. Engineered 5-kb histone repeats were then arrayed to 12 copies in pMultiBac (McKay et al., 2015; Meers et al., 2018). All histone gene arrays were inserted into the VK33 attP site on chromosome 3L (65B2) (Venken et al., 2006).
Northern analysis was performed using a 7 M 6% urea acrylamide gel to resolve histone mRNAs; ∼1 µg of RNA from embryos or larvae was used with a radio-labeled probe to the coding region of H3 as in Lanzotti et al. (2002).
Histone expression analysis
Total RNA was prepared in Trizol and cDNA synthesized with random hexamers using Superscript II (Invitrogen), according to the manufacturer’s instructions. RT-RCR was performed using gene-specific primers to H2a (McKay et al. 2015) and H3. Each reaction was performed at least three times with similar results. PCR products were digested using XhoI (H2a) or SacI (H3). Digested amplicons were run on an 8% acrylamide gel or a 1.5% agarose gel.
We used primary antibodies at the following concentrations: rabbit anti-CLAMP (1:1000), guinea pig anti-Mxc (1:2000), guinea pig anti-Mute (1:5000), rabbit anti-FLASH (1:2000), rabbit anti-Lsm10 (1:1000), mouse anti-MPM-2 (1:100; Millipore), rabbit anti-GAF (1:1000), and mouse anti-LacI (1:1000; Millipore). We used Alexa Fluor secondary antibodies (Thermo Fisher Scientific) at a concentration of 1:1000. In situ probes were detected using 15 μg/ml streptavidin-DyLight-488 (Vector Laboratories). Salivary gland dissections and squashes were performed as in Tatomer et al. (2016). Images were acquired using a Zeiss Lsm710 with ZEN DUO software. Images were analyzed using ImageJ. Ectopic HLBs in embryos were quantified as Salzler et al. (2013).
Fluorescent in situ hybridization
FISH probes were made from a PCR product that spanned the entire wild-type histone repeat (primers AAAGGAGGTTGGTAGGCAGC and ACGCTAGCGCTTTATCTGCA). Biotinylated FISH probes were made with nick translation by incubating 1 μg of purified PCR product for 2 h at 15°C in a total of 50 μl containing 1× DNAPolI buffer (Fisher Optizyme); 0.05 mM each of dCTP, dATP, and dGTP; 0.05 mM biotin-11-dUTP (Thermo Scientific); 10 mM 2-mercaptoethanol; 0.004 U of DNaseI (Fisher Optizyme); and 10 U of DNAPol I (Fisher Optizyme). The reaction was purified on a PCR purification column (Thermo Scientific) and diluted in hybridization buffer (2× SSC, 10% dextran sulfate, 50%formamide, and 0.8 mg/ml salmon sperm DNA) to a final volume of 220 μl. FISH probes were diluted in hybridization mixture and added to slides containing salivary gland polytene chromosome spreads before heating. We added a coverslip, sealed it with rubber cement, and heated the slide for 2 min on a 91°C heat block. Slides were placed in a humid box and incubated at 37°C overnight. Immunostaining was then performed by incubating the slides in primary antibody overnight at 4°C in a humid box.
ChIP was performed according to Blythe and Wieschaus (2015) with the following particulars: 2- to 4-h-old embryos (∼200 embryos for each of three biological replicates) from HisC/HisC; 12xDWT/12xDWT or 12xPR/12xPR strains were collected, fixed, and stored at −80°C until further processing. Immunoprecipitation of chromatin preparations was performed using rabbit anti-CLAMP antibody (Rieder et al., 2017) and a polyclonal rabbit IgG (Millipore-Sigma, 12-370). qPCR was performed as described in Urban et al. (2017). Three biological replicates and two technical replicates were used for each genotype. We synthesized libraries using 4 µl of inputs and 36 µl of each ChIP sample (CLAMP or IgG) according to Blythe and Wieschaus (2015). We made libraries using the NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs), pooled 12 samples per lane, and sequenced libraries on an Illumina HiSeq 2500 in 2 × 150-bp mode (Genewiz). We normalized CLAMP IP target abundance against IgG IP target abundance. We analyzed data using a Student’s t test, comparing target abundance between the DWT and PR lines.
Analysis and quantification of ChIPseq data
The 150-bp paired-end sequencing reads were quality checked using Fastqc (version 0.11.8). ILLUMINA adapter sequences were trimmed from the reads using the BBDuk tool in the BBtools package (BBmap version 38.82). Due to the presence of contaminating reads from yeast, mouse, and human genomes, the Bbsplit tool was used to obtain reads that mapped to the Drosophila genome (BBmap version 38.82). Custom reference genomes were created by removing all RD histone gene repeat sequences from genome version dm3 and by adding back a single 5-kb histone gene repeat unit with either DWT or PR sequences (McKay et al., 2015). The reads were mapped to the custom genomes using Bowtie2 (version 188.8.131.52) using default parameters (Langmead and Salzberg, 2012; Langmead et al., 2019). We used MACS2 (version 2.1.2) to identify peaks with the following parameters: -g dm –keep-dup all -B (Zhang et al., 2008). The resulting bedGraph files were visualized using Integrative Genomics Viewer. For quantitation analysis, reads overlapping with specified coordinates, chrX:3755103-3755605 (roX1), DWT:3575-4077, PR:3575-4077 (H3-H4), DWT:1797-2299, PR:1797-2299 (H2A-H2B), DWT:1811-2313, PR:1811-2313 (H1), were counted using SAMtools (version 1.10) using -c -F 260 options (Li et al., 2009). The reads for specified regions were normalized to total reads and then normalized to input. ChIP-seq data are deposited with NCBI Gene Expression Omnibus (GEO) under accession number GSE150297.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E20-03-0176) on May 13, 2020.
designer wild type
FLICE-associated huge protein
histone locus body
histone wild type
multi sex combs
nuclear protein, ataxia-telangiectasia locus
We thank Erica Larschan’s lab at Brown University for assistance with ChIP experiments and John Atwater for technical assistance. This work was supported by National Institutes of Health grant R01GM058921 to R.J.D and W.F.M and F32GM109663, K99HD092625, and R00HD092625 to L.E.R.