The Integrity of the Speciation Core Complex is necessary for centromeric binding and reproductive isolation in Drosophila

Postzygotic isolation by genomic conflict is a major cause for the formation of species. Despite its importance, the molecular mechanisms that result in the lethality of interspecies hybrids are still largely unclear. The genus Drosophila, which contains over 1600 different species, is one of the best characterized model systems to study these questions. We showed in the past that the expression levels of the two hybrid incompatibility factors Hmr and Lhr diverged in the two closely related Drosophila species, D. melanogaster and D. simulans, resulting in an increased level of both proteins in interspecies hybrids. This overexpression leads to mitotic defects, a misregulation in the expression of transposable elements and a decreased fertility. In this work, we describe a distinct six subunit Speciation Core Complex (SCC) containing HMR and LHR and analyse the effect of Hmr mutations on complex function and integrity. Our experiments suggest that HMR acts as a bridging factor between centromeric chromatin and pericentromeric heterochromatin, which is required for both its physiological function and its ability to cause hybrid male lethality.


INTRODUCTION
Eukaryotic genomes are constantly challenged by the integration of viral DNA or the amplification of transposable elements. As these challenges are often detrimental to the fitness of the organism, they frequently elicit adaptive compensatory changes in the genome.
As a result of this process, the genomes as well as the coevolving compensatory factors can rapidly diverge between individuals within a population. Such divergences can result in severe incompatibilities eventually leading to the formation of two separate species (Presgraves, 2010;Sawamura, 2012).
Arguably the best characterized system for studying the genetics of reproductive isolation and hybrid incompatibilities is constituted by the two closely related Drosophila species D. melanogaster and D. simulans (D. mel and D. sim) (Sturtevant, 1920). One century of genetic studies has led to the identification of the three fast evolving genes that are critical for hybrid incompatibility: Hmr (Hybrid male rescue), Lhr (Lethal hybrid rescue) and gfzf (GST-containing FLYWCH zinc finger protein) (Hutter & Ashburner, 1987;Barbash & Ashburner, 2003;Watanabe, 1979;Brideau et al, 2006;Phadnis et al, 2015). The genetic interaction of these three genes results in the lethality of D.mel/D.sim hybrid males.
Strikingly, all three genes are fast evolving and code for chromatin proteins suggesting that their fast evolution reflects adaptations to genomic alterations. While the molecular interaction between HMR and LHR is well established in pure species as well as in hybrids (Thomae et al, 2013;Satyaki et al, 2014), the molecular basis for their genetic interaction with GFZF is unclear. Interestingly, in interspecies hybrids or when HMR/LHR are overexpressed, HMR spreads to multiple novel bindings sites many of which have been previously characterized to also bind GFZF (Cooper et al, 2019).
In the nucleus of tissue culture cells and in imaginal discs, HMR and LHR form defined foci that are clustered around centromeres (Thomae et al, 2013;Kochanova et al, 2020;Blum et al, 2017). Super-resolution microscopy and chromatin immunoprecipitation revealed that HMR is often found at the border between centromeres and constitutive pericentromeric heterochromatin bound by HP1a (Kochanova et al, 2020;Gerland et al, 2017;Anselm et al, 2018). In addition to pericentromeric regions, HMR also binds along chromosome arms colocalizing with known gypsy-like insulator elements (Gerland et al, 2017). Depending on the tissue investigated, HMR shows slightly different binding patterns. It binds to telomeric regions of polytene chromosomes (Cooper et al, 2019;Thomae et al, 2013), colocalizes with HP1a in early Drosophila embryos (Satyaki et al, 2014) and near DAPI-bright heterochromatin in larval brain cells (Blum et al, 2017).
Flies carrying Hmr or Lhr loss of function alleles show an upregulation of transposable elements (TEs), defects in mitosis, and a reduction of female fertility in D. mel. Expression of transposable elements is increased particularly in ovarian tissue but also in cultured cells (Satyaki et al, 2014;Thomae et al, 2013). The mechanism that causes such a massive and widespread upregulation is not entirely clear as most of the TEs that respond to a reduced Hmr dosage are not bound by HMR under native conditions (Gerland et al, 2017). Due to the overexpression of the HeT-A, TART and TAHRE retrotransposons, Hmr mutants show a substantial increase in telomere length (Satyaki et al, 2014) and an increased number of anaphase bridges during mitosis presumably due to a failure of chromatid detachment during anaphase (Blum et al, 2017). The massive upregulation of transposable elements in ovaries is possibly also the cause of the substantially reduced fertility of Hmr and Lhr mutant female flies (Aruna et al, 2009).
Many of the phenotypes observed in cell lines lacking Hmr and Lhr, are mirrored by Hmr and Lhr overexpression, highlighting the importance of properly balanced Hmr/Lhr levels (Thomae et al, 2013). Hybrids show enhanced levels of both proteins relative to the pure species and consistently, are also characterized by loss of transposable elements silencing and cell cycle progression (Thomae et al, 2013;Kelleher et al, 2012;Satyaki et al, 2014).
The latter is thought to be the cause for the failure of male hybrids to develop into adults, given the almost complete absence of imaginal discs (Blum et al, 2017;Gatti & Baker, 1989;Orr et al, 1997;Bolkan et al, 2007). In addition, hybrids and Hmr/Lhr overexpressing cells, display a widespread mis-localization of HMR at several euchromatic loci at chromosome arms including the previously unbound GFZF binding sites (Kochanova et al, 2020;Thomae et al, 2013;Cooper et al, 2019).
To better understand the deleterious effects observed in the presence of an excess of HMR, we decided to investigate the binding partners of HMR under native conditions and upon overexpression. Our results suggest that a defined Speciation Core Complex (SCC) of 6 subunits exists under native conditions and that interference with the formation of this complex results in its loss of function.

Cell culture and induction
Drosophila melanogaster Schneider cell lines (SL2) were grown in Schneider's medium (Gibco Life Technologies) supplemented with 10% fetal calf serum and antibiotics (penicillin 100 units/mL and streptomycin 100 µg/mL) at 26°C. Stable SL2 cells transfected with metallothionein promoter (pMT) driven FLAG-HA-Hmr + /Myc-Lhr, FLAG-HA-Hmr 2 /Myc-Lhr, rotating wheel. Soluble proteins were extracted by increasing the NaCl to 450 mM and incubated for 1 h at 4°C on a rotating wheel. Finally, the soluble material was separated from the insoluble chromatin pellet material by centrifugation for 30 min at 20000 x g and used for immunoprecipitations.

Immunoprecipitation
Anti-FLAG immunoprecipitation was performed using 20 μL of packed agarose-conjugated mouse anti-FLAG antibody (M2 Affinity gel, A2220 Sigma-Aldrich) and were targeted either against the exogenously expressed transgenes (HMR + , HMR dC , HMR 2 ) or an endogenously FLAG-tagged HMR (HMR). The other IPs were performed by coupling the specific antibodies to 30 µL of Protein A/G Sepharose beads. Each bait was targeted with at least one antibody (rat anti-LHR 12F4, mouse anti-HP1a C1A9, rabbit anti-NLP, anti-FLAG-M2 for FLAG-BOH1 and FLAG-BOH2), while HMR was targeted with three different antibodies (rat anti-HMR 2C10 and 12F1, anti-FLAG-M2 for FLAG-HMR). Rabbit anti-NLP and mouse anti-HP1a were directly incubated with the beads, while rat anti-HMR and anti-LHR were incubated with beads that were pre-coupled with 12 μL of a rabbit anti-rat bridging antibody (Dianova,. FLAG-IPs in non-FLAG containing nuclear extracts were used as mock controls for FLAG-IPs. For all other IPs, unspecific IgG coupled to Protein A/G Sepharose or Protein A/G Sepharose alone were used as mock controls. The steps that follow were the same for all the immunoprecipitations and were all performed at 4°C. Antibody coupled beads were washed three times with IP buffer (25mM Hepes pH 7.6, 150 mM NaCl, 12.5 mM MgCl 2 , 10% Glycerol, 0.5 mM EGTA) prior to immunoprecipitation. Thawed nuclear extracts were centrifuged for 10 minutes at 20000 x g to remove precipitates and subsequently incubated with antibody-coupled beads in a total volume of 500-600 µL IP buffer complemented with a cocktail of protease inhibitors plus 0.25 μg/mL MG132, 0.2 mM PMSF, 1 mM DTT and end-over-end rotated for 2 h (anti-FLAG) or 4 h (other IPs) at 4°C. After incubation, the beads were centrifuged at 400 x g and washed 3 times in IP buffer complemented with inhibitors and 3 times with 50 mM NH 4 HCO 3 before on beads digestion.

Sample preparation for mass spectrometry
The pulled-down material was released from the beads by digesting for 30 minutes on a shaker (1400 rpm) at 25°C with trypsin at a concentration of 10 ng/μL in 100 µL of digestion buffer (1M Urea, 50 mM NH 4 HCO 3 ). After centrifugation the peptide-containing supernatant was transferred to a new tube and two additional washes of the beads were performed with 50 μL of 50 mM NH 4 HCO 3 to improve recovery. 100 mM DTT was added to the solution to reduce disulphide bonds and the samples were further digested overnight at 25°C while shaking at 500 rpm. The free sulfhydryl groups were then alkylated by adding iodoacetamide (12 mg/mL) and incubating 30 minutes in the dark at 25°C. Finally, the light-shield was removed and the samples were treated with 100 mM DTT and incubated for 10 minutes at 25°C. The digested peptide solution was then brought to a pH~2 by adding 4 μL of trifluoroacetic acid (TFA) and stored at -20°C until desalting. Desalting was done by binding to C18 stage tips and eluting with elution solution (30% methanol, 40% acetonitrile, 0.1% formic acid). The peptide mixtures were dried and resuspended in 20 μL of formic acid 0.1% before injection.

Sample analysis by mass spectrometry
Peptide mixtures (5 µL) were subjected to nanoRP-LC-MS/MS analysis on an Ultimate 3000 nano chromatography system coupled to a QExactive HF mass spectrometer (both Thermo Fisher Scientific). The samples were directly injected in 0.1% formic acid into the separating column (150 x 0.075 mm, in house packed with ReprosilAQ-C18, Dr. Maisch GmbH, 2.4 µm) at a flow rate of 300 nL/min. The peptides were separated by a linear gradient from 3% ACN to 40% ACN in 50 min. The outlet of the column served as electrospray ionization emitter to transfer the peptide ions directly into the mass spectrometer. The QExactive HF was operated in a Top10 duty cycle, detecting intact peptide ion in positive ion mode in the initial survey scan at 60,000 resolution and selecting up to 10 precursors per cycle for individual fragmentation analysis. Therefore, precursor ions with charge state between 2 and 5 were isolated in a 2 Da window and subjected to higher-energy collisional fragmentation in the HCD-Trap. After MS/MS acquisition precursors were excluded from MS/MS analysis for 20 seconds to reduce data redundancy. Siloxane signals were used for internal calibration of mass spectra.

Proteomics Data analysis
For protein identification, the raw data were analyzed with the Andromeda algorithm of the MaxQuant package (v1.6.7.0) against the Flybase reference database (dmel-all-translation-r6.12.fasta) including reverse sequences and contaminants. Default settings were used except for: Variable modifications = Oxidation (M); Unique and razor, Min. peptides = 1; Match between windows = 0.8 min. Downstream analysis on the output proteinGroups.txt file were performed in R (v4.0.1). If not otherwise stated, plots were generated with ggplot2 package (v3.3.2). Data were filtered for Reverse, Potential.contaminant and Only.identified.by.site and iBAQ values were log 2 transformed and imputed using the R package DEP (v1.10.0, impute function with following settings: fun= "man", shift = 1.8, scale = 0.3). Except for Fig. 2 and 3, where data were bait normalized, median normalization was performed. Statistical tests were performed by fitting a linear model and applying empirical Bayes moderation using the limma package (v3.44.3). AP-MS for SCC identification ( Fig. 1 and S1) were compared with a pool of all control samples (IgG and FLAG mock IPs). For Fig.   1C and Fig. S1E enriched proteins from AP-MS experiments from SCC components were first selected (cut off: log2FC > 2.5, p-adjusted < 0.05) and then intersection was quantified and plotted with UpsetR (v1.4.0). The Network graph in Fig Secondary antibodies included sheep anti-mouse (1:5000) (RRID: AB772210), goat anti-rat (1:5000) (RRID: AB772207), donkey anti-rabbit (1:5000) (RRID: AB772206) coupled to horseradish peroxidase. For proteins detection after IP, beads were boiled in Laemmli sample buffer after washing. For protein detection in ovaries, a short nuclear extraction was performed from 10 pairs of ovaries prior to boiling in sample buffer.

ChIP-Seq
Chromatin immunoprecipitation was essentially performed as in (Gerland et al., 2017). For each anti-FLAG ChIP reaction, chromatin isolated from 1-2 x 10 6 cells were incubated with 5 µg of mouse anti-FLAG (F1804, SIGMA-ALDRICH -RRID: AB262044) antibody pre-coupled to Protein A/G Sepharose. For ChIPs targeting total HMR, the same amount of chromatin was incubated with rat anti-HMR 2C10 antibody pre-coupled to Protein A/G Sepharose through a rabbit IgG anti-rat (Dianova, 312-005-046). Samples were sequenced (single-end, 50 bp) with the Illumina HiSeq2000. Sequencing reads were mapped to the Drosophila genome (version dm6) using bowtie2 (version 2.2.9) and filtered by mapping quality (-q 2) using samtools (version 1.3.1). Sequencing depth and input normalized coverages were generated by Homer (version 4.9). Enriched peaks were identified by Homer with the parameters -style factor -F 2 -size 200 for each replicate.
High confidence FLAG-HMR peaks (a pool of HMR + and HMR dC ) were called when a peak was present in at least half of the samples (5 out of 10). Coverages were centred at high confidence FLAG-HMR peaks in 4 kb windows and binned in 10 bp windows. The as-such generated matrices were z-score normalized by the global mean and standard deviation.
HP1a-proximal peaks were defined as 10 percent of the peaks with highest average HP1a ChIP signal in 4 kb windows surrounding peaks. Composite plots and heatmaps indicate the average ChIP signal (z-score) across replicates. Heatmaps were grouped by HP1a class and sorted by the average ChIP signal in HMR native in a 400 bp central window. For statistical analysis, the average ChIP signal (z-score) was calculated in a 200 bp central window across peaks for each replicate. P-values were obtained by a linear mixed effect model (R packages: lme4 version 1.1-23 and lmerTest version3.1-2), in which average ChIP signal was included as outcome, genotype (Hmr + or Hmr dC ) and peak class (HP1a-proximal or non-HP1a-proximal) as fixed effects and sample ids as random intercept.
Chromosome-wide coverage plots were generated by averaging replicates, binning coverages in 50 kb windows and z-score normalizing by the global mean and standard deviation.
The percentage of peaks on chromosome 4 relative to the total number of peaks was calculated for each replicate. P-value was obtained by a linear model (R package: stats version 3.6.1), in which percentage was included as outcome and genotype (Hmr + or Hmr dC ) as independent variable.  into account for analysis. Two different quantifications were performed. In one case cells were separated and counted based on the degree of co-localization between HMR and CENP-C: overlapping, partially overlapping or non-overlapping. In parallel, the number of CENP-C marked centromeric foci associated with HMR signal was measured. Both cells and centromeric foci were blind-counted, the experiment was repeated in 2 biological replicates and for each replicate at least two slides were measured (for each slide between 24 and 63 cells were quantified). Further details about stainings for Fig. S2 and microscopy are available as supplementary methods.

Immunofluorescent staining in ovaries
Flies were grown at 25°C for 7-9 days and fed in yeast paste for at least 3 days prior to dissection. Ovaries were dissected in ice-cold PBS, then ovarioles were teased apart with forceps and moved to 1.5 mL tubes. PBS was removed and fixation solution (400 µL PBS Paraformaldehyde 2%, Triton 0.5 %, 600 µL Heptane) was added. Samples were incubated for 15 min at room temperature on a rotating wheel.

Drosophila husbandry and stocks
Drosophila stocks were reared on standard yeast glucose medium and raised at 25°C on a

Fertility assays
Three 1-3 days old D. melanogaster females were crossed for 2-3 days with six wild type males D. melanogaster. Flies were then transferred to fresh vials and again every 5 days for 3 times in total. Vials were scored 15-18 days after first eggs were laid, to make sure all adults were eclosed but no F2 was included. Vials in which one female or more than one male was missing were not scored. The whole assay was performed at 25°C. Tested females Df(1)Hmr-; Hmr*/+ (Hmr* = an Hmr transgenic allele) were always grown with and compared to their respective control siblings Df(1)Hmr -;+/+, obtained from crosses between Df(1)Hmr-and Df(1)Hmr-; Hmr*/+. Rescue was measured as total offspring counted per female. In the rescue experiment, Hmr + served as a positive control (fertility rescue) and Hmr 2 as a negative control (no fertility rescue). For statistical testing, Wilcoxon rank sum test (non-parametric) was used for pairwise comparisons with FDR correction for multiple testing using ggpubr package (v0.4.0, using compare_means function with following settings: formula = offspring_per_mother ~ Hmr_allele, group.by = 'day', method = 'wilcox.test', p.adjust.method = 'fdr'). Annealing temperature for all the tested primers was 60°C and the list of primers used are given on request.

RNA extraction, cDNA synthesis and quantitative RT-PCR
Plots for qPCR results were generated with R using ggplot2 package. For statistical testing Welch t-test was used with FDR correction for pairwise comparisons using rstatix package (v0.5.0).

Data Sources
Published ChIP-seq data were obtained from GEO (HMR native and HP1a: GSE86106; CP190: GSE41354). New ChIP-seq datasets are available with the accession number: GSE163058. Proteomics datasets with the protein-protein interaction network of the SCC and the analysis of HMR overexpression and mutants are available with the accession numbers PXD023188 and PXD023193, respectively. Interactive network and volcano plots are available at Interactive Network and volcano plots from SCC purifications are entirely available at the following (URL) and in table S2.

Characterization of the Speciation Core Complex in Drosophila melanogaster
To identify the proteins interacting with the hybrid incompatibility (HI) proteins HMR and LHR under native conditions, we used specific monoclonal antibodies targeting HMR and LHR to perform affinity purification coupled with mass spectrometry (AP-MS) from nuclear extracts prepared from D. mel SL2 cells. For HMR, we additionally validated our results by performing AP-MS with a FLAG antibody in cells carrying an endogenously tagged HMR (HMR endo (Gerland et al, 2017). These experiments revealed the existence of a set of four stable protein interactors shared between HMR (Fig. 1A) and LHR (Fig. 1B). Besides HMR and LHR, this six-subunit complex contains nucleoplasmin (NLP) and nucleophosmin (NPH) as well as two poorly characterized proteins, CG33213 and CG4788, which we named Buddy Of Hmr 1 (BOH1) and Buddy Of Hmr 2 (BOH2), respectively. AP-MS experiments for each of the individual subunits confirmed the existence of this defined complex (Figs. 1C, S1 A-E, Table S2) and the individual components also largely colocalize in SL2 cells ( Fig. S2, (Anselm et al, 2018 1C and S1, Table S2), suggesting that the SCC components are also engaged in other complexes. In summary, our AP-MS results reveal the existence of a stable HMR/LHR containing protein complex (SCC) under physiological conditions. As several SCC subunits also contribute to other complexes, we wondered whether a surplus of HMR and LHR would affect complex composition.

Overexpression of HMR and LHR results in a gain of novel protein-protein interactions
The importance of a physiological HMR and LHR dosage has been shown in nonphysiological conditions like interspecies hybrids or cells where they are artificially co- Establishing to which extent these newly acquired interactors or the formation of a functional SCC contributes to HMR/LHR's physiological function and it's lethal function in male hybrids, would provide further mechanistic details. To this end we investigated the HMR interaction proteome upon ectopically expressing mutant HMR proteins.

Two different Hmr mutations interfere with SCC formation and HMR localization
Most of the Hmr alleles that rescue hybrid male lethality are either null mutations or mutations that dramatically reduce the level of HMR (Df (1) We next wondered whether the HMR C-terminal truncation and its concomitant loss of interaction with LHR and HP1a would influence its nuclear localization. A co-staining with antibodies against the exogenously expressed HMR and a centromeric marker (anti-CENP-C) revealed a rather diffuse nuclear localization of HMR dC , which is in sharp contrast to the full length HMR, which forms distinct bright (peri)centromeric foci (Fig. 4A, S6A) 4B,C; Fig. S5) leading to a substantial reduction of HMR dC binding in proximity to centromeres in both metacentric chromosomes 2 and 3 (Fig. 4B,C) as well as throughout the mostly heterochromatic chromosome 4 (Fig. S5C,D).
All together our results show that while the HMR C-terminus is required for HMR's interaction with LHR and HP1a and for localization in close proximity to centromeres, it is dispensable for binding to the rest of the SCC and to genomic loci unrelated to HP1a. The Hmr2 mutation in contrast does not affect HMR's ability to interact with LHR/HP1a but weakens the interaction with the other SCC components. The fact that both mutations impair the centromere proximal binding suggests that complex integrity is necessary for HMR's genomic localization. Next, we took advantage of these two mutants to investigate the importance of HMR's interactions within the SCC for its function in flies.

The HMR C-terminus is required for HMR physiological function in D. melanogaster
To investigate whether the C-terminus of HMR is required for HMR to fulfill its physiological function, we generated fly lines expressing full length HMR (HMR + ) or mutant forms of it (HMR dC , HMR 2 ). We crossed these alleles in a mutant background (Df(1)Hmr-, hereafter referred to as Hmr ko ) and performed complementation assays to assess if the transgenic alleles are able to rescue Hmr wild type functions (Fig. 5A-B). All assays were done in ovaries, since HMR has been shown to be well expressed and important for fertility and retrotransposons silencing in this tissue. After verifying the expression of the Hmr transgenic alleles (Fig. S6D) in the presence of an endogenous wild type copy, the centromeric localization of an HMR Cterminal mutant is only moderately affected in early stage follicle cells (Fig. S7B). However, in later stages HMR dC does not associate with HP1a domains but rather shows a more diffuse nuclear staining, while the wild type transgenic HMR (HMR + ) mirrors the localization of endogenous HMR (Fig. 5C).
As the silencing of transposable elements (TE) has been previously shown to be impaired by tested whether the Hmr dC or the Hmr 2 alleles were able to restore TE silencing (Fig. 5D).
Whereas full length HMR was able to strongly repress all the TE studied, neither Hmr dC nor Hmr 2 were able to do so, showing expression levels comparable to Hmr deletion mutants.
Since Hmr loss of function mutations have been shown to also cause a major reduction in female fertility (Aruna et al, 2009), we also tested Hmr dC for the complementation of this phenotype. Similar to what we have observed for the TE silencing, both mutant alleles (Hmr dC and Hmr 2 ) were unable to rescue the fertility defect (Fig. 5E, Table S4). All together, these results show that the Hmr C-terminus is required for HMR localization and physiological function in D. melanogaster ovaries.
The Hmr C-terminus is necessary for male hybrid lethality and reproductive isolation To understand whether the toxic Hmr function in interspecies hybrids also requires its Cterminus, we performed a hybrid viability rescue assay (Fig. 6A). We therefore crossed D.  (Anselm et al, 2018).

Excess of HMR and LHR interact with novel chromatin factors
As hybrid animals suffer from increased levels of HMR

HMR contains two functionally important protein-protein interaction modules
Our proteomic analysis of Hmr mutants suggests that HMR's N-terminal MADF3 domain, which is mutated in the Hmr 2 allele, mediates the interaction with NLP, NPH, BOH1 and BOH2 while its C-terminus binds LHR, and through this interaction presumably recruits HP1a (Giot et al, 2003;Aruna et al, 2009;Thomae et al, 2013;Greil et al, 2007). Further genome-wide and cytological experiments with these Hmr alleles reveal that the integrity of all these interactions as well as a balanced expression of Hmr/Lhr is vital for proper targeting and the physiological functions of the SCC. Since the experiments in SL2 cells were done in the presence of endogenous (wild type) HMR, they reflect a competitive situation, which might be more sensitive in uncovering subtle differences between wild type and mutant HMR.
Especially the centromere binding phenotype was more apparent in SL2 cells than in mitotically cycling follicle cells, in which the mutant Hmr allele was the sole source of HMR.
The loss of heterochromatin association of HMR dC , however, was very apparent in follicle cells that have undergone a mitotic to endo-cycling transition. This finding argues that HMR interacts with HP1a through LHR to recruit the SCC to HP1a containing nuclear domains.
The functional assays in flies confirmed that both mutants are not able to rescue the HMR ko phenotype, supporting the hypothesis that the integrity of the SCC complex is essential for its function. We therefore consider it unlikely that the Hmr mutant phenotype observed in D.mel (reduced female fertility, upregulation of TEs) is solely dependent on HMR's ability to bind heterochromatin, since the mutant HMR 2 protein is still able to interact with LHR and HP1a but no longer rescues these phenotypes. We therefore propose that HMR organizes the chromocenter by directly interacting with centromeric as well as heterochromatic factors and that both interactions are required for HMR to fulfil its function. In fact, defects in chromocenter bundling have been shown to result in micronuclei formation and loss of cellular viability in the imaginal discs and lymph glands (Jagannathan et al, 2019, 2018), a phenotype that is also observed in interspecies hybrids (Bolkan et al, 2007;Orr et al, 1997).

HMR bridges heterochromatin and the centromere
Mutations impairing HMR's bridging capacity between heterochromatin and the centromere not only fail to complement Hmr null phenotypes in pure species, but also do not cause lethality of male hybrids of D.mel and D.sim. This suggests that the simultaneous binding of HMR to heterochromatic and centromeric factors not only plays a role in its physiological function but also contributes to hybrid incompatibility. In addition, the novel finding that The isolation of a defined complex involving the hybrid incompatibility proteins HMR and LHR will enable us to perform a more detailed molecular analysis of its function and sets the ground for future comparative studies on the divergent evolution of its components within species and on their lethal interactions in hybrids.

DATA AVAILABILITY
ChIP-seq datasets are available with the accession number: GSE163058. Proteomics datasets with the protein-protein interaction network of the SCC and the analysis of HMR overexpression and mutants are available with the accession numbers PXD023188 and PXD023193, respectively. Interactive Network and volcano plots from SCC purifications are entirely available at the following https://wasim-aftab.shinyapps.io/SccNet-AL/. .

SUPPLEMENTARY DATA
Supplementary Data are available online.

Figure 2: Overexpression leads excess of HMR to interact beyond the SCC (A)
Differential interaction proteome between endogenously FLAG tagged HMR (HMR endo , n = 4) and ectopically expressed HMR (HMR + , n = 9). Only proteins enriched in HMR + or HMR endo vs CTRL (p < 0.05) were considered. Components of the SCC and HP1a are shown in red, all other factors in blue. To display the differences within the HMR endo and HMR + interactomes, the enrichment of each putative interactor (Log2(iBAQ HMR* /IBAQ control )) was normalized to the enrichment of the HMR protein used as bait. The resulting values were then plotted against each other. Dots below the diagonal indicate a stronger enrichment in the HMR endo pull down, the dots above the diagonal a stronger enrichment in the ectopically expressed HMR. (B) Differences of the ratio between HMR and members of the SCC and HP1a with and without ectopic expression of HMR. Plotted is the relative enrichment of each SCC member to the enrichment of the HMR used as bait (= the offset from the diagonal).  Table S3). In (A) and (B) proteins were labelled only if enriched in HMR + or HMR endo vs CTRL (p < 0.05).

Figure 3: Two different Hmr mutations interfere differently with HMR interactome and SCC formation. Effect of Hmr 2 (A) and
Hmr dC (C) mutations on the HMR interaction with the SCC components and HP1a. Y-axis represent the Log2 fold-change of HMR 2 /HMR + and HMR dC /HMR + , respectively, calculated after normalization of each sample to the enrichment of the HMR protein used as bait. Error bars reflect the SEM (HMR + : n=9; HMR dC : n=10, HMR 2 : n=3). Differential interaction proteome between ectopically expressed wild type or mutated HMR (HMR + versus HMR 2 (B) or HMR + versus HMR dC (D)). Only proteins enriched in HMR + or HMR endo vs CTRL(p < 0.05) are shown. Components of the SCC are shown in red, all other factors in blue. To display the differences within each interactome, the enrichment of each putative interactor was normalized to the enrichment of the HMR protein used as bait. The resulting values were then plotted against each other. Dots above the diagonal indicate a stronger enrichment in the HMR + pull down, dots below the diagonal a stronger enrichment in the HMR mutated alleles (HMR 2 or HMR dC ). Further details are available in Table S3.

Figure 4: The HMR C-terminus is required for HMR localization in proximity to centromeres and HP1a-bound chromatin (A) Ectopic HMR dC fails to form bright (peri)centromeric foci in SL2 cells. Immunofluorescence images of cells expressing different
Hmr transgenic alleles (HA-Hmr + , HA-Hmr dC or HA-Hmr 2 ) together with wild type LHR showing the co-staining of HA-HMR and CENP-C. Based on the overlap between HMR and CENP-C signals, cells were categorized in three groups (overlapping, partially-overlapping and non-overlapping) and the number of cells belonging to each group quantified. Error bars represent standard error of the means (n = 2). Size bars indicate 5 µm. (B) HP1a-proximal binding is specifically disrupted in HMR dC . Average plot of FLAG-HMR ChIP-seq profiles (zscore normalized) centered at high confidence FLAG-HMR peaks in 4 kb windows. HP1aproximal (left) and non-HP1a-proximal (right) peaks are shown for HMR + (light blue) and HMR dC (dark blue). (C) HMR dC genome-wide binding is impaired in proximity to centromeres and mostly unaffected at chromosome arms. Chromosome-wide FLAG-HMR ChIP-seq profiles (z-score normalized) for HMR + (light blue), HMR dC (dark blue) and HP1a (green).
Number of adult offspring per female mother assessed in a time course (females aged 5-10, 10-15 and 15-20 days). Wilcoxon rank sum test was used for pairwise comparisons with Hmr + as a reference group and FDR for multiple testing adjustment (* p  0.05, ** p  0.01, *** p  0.001, **** p  0.0001). For details about fertility assays refer to Table S4. Crosses from non-complemented Hmr ko mothers were used as control. Dots represent individual biological replicates. Wilcoxon rank sum test was used for pairwise comparisons with Hmr + as a reference group and fdr for multiple testing adjustment (* p  0.05, ** p  0.01, *** p  0.001, **** p  0.0001). For details refer to Table S5. Figure S1: Interactomes of HMR/LHR interactors provide evidence for the existence of a Speciation Core Complex (SCC) (related to Fig. 1)   and (B) proteins were labelled or considered for GO search only if enriched in HMR + or HMR endo vs CTRL (p < 0.05) and differentially enriched between HMR + and HMR endo (log 2 fold-change (HMR endo /HMR+) < 1.5). Unlabeled additional bait-specific interactors are listed in Table S3. (E) GO terms depleted upon HMR dC mutation. (F) Volcano plot highlighting interactions depleted in HMR 2 . X-axis: log 2 fold-change of FLAG-HMR 2 IPs (right side of the plot) vs FLAG-HMR + IPs (left side of the plot). Y-axis: significance of enrichment given as -log 10 p-value calculated with a linear model. SCC subunits are labelled in red. In blue are factors depleted upon HMR 2 mutation (among the endogenous or overexpression-induced interactions of HMR). Unlabeled additional bait-specific interactors are listed in Table S3. (G) GO terms depleted upon HMR 2 mutation. In (D) and (G) proteins labelled or used for GO search include endogenous or overexpression-induced interactions of HMR (i.e. enriched in HMR + or HMR vs CTRL with p < 0.05) and differentially enriched between HMR + and the HMR mutant analyzed (log 2 foldchange (HMR*/HMR+) < 1.5).
Secondary antibodies: goat anti-rabbit Alexa Fluor Plus 647, donkey anti-rat Cy3, goat antimouse Alexa Fluor Plus 488. Following washes with PBS/Triton X-100 0.05% and PBS, coverslips were mounted in Prolong Diamond antifade.  into account for analysis. Two different quantifications were performed. In one case cells were separated and counted based on the degree of co-localization between HMR and CENP-C: overlapping, partially overlapping or non-overlapping. In parallel, the number of CENP-C marked centromeric foci associated with HMR signal was measured. Both cells and centromeric foci were blind-counted, the experiment was repeated in 2 biological replicates and for each replicate at least two slides were measured (for each slide between 24 and 63 cells were quantified).

Microscopy and downstream image analysis
Confocal microscopy of data presented in Figures 5C and S2 was performed at the core facility bioimaging of the Biomedical Center using the following instruments and settings:    Figure S4 Lukacs et al.