cells Article The Histone Variant MacroH2A1 Regulates Key Genes for Myogenic Cell Fusion in a Splice-Isoform Dependent Manner Sarah Hurtado-Bagès 1 , Melanija Posavec Marjanovic 2 , Vanesa Valero 1 , Roberto Malinverni 1 , David Corujo 1 , Philippe Bouvet 3 , Anne-Claire Lavigne 4 , Kerstin Bystricky 4 and Marcus Buschbeck 1,2, * 1 Cancer and Leukemia Epigenetics and Biology Program, Josep Carreras Leukaemia Research Institute (IJC), Campus ICO-GTP-UAB, 08916 Badalona, Spain; shb.scientific.illu@gmail.com (S.H.-B.); vvalero@carrerasresearch.org (V.V.); rmalinverni@carrerasresearch.org (R.M.); dcorujo@carrerasresearch.org (D.C.) 2 Program for Predictive and Personalized Medicine of Cancer, Germans Trias i Pujol Research Institute (PMPPC-IGTP), 08916 Badalona, Spain; melanija.posavec@gmail.com 3 Université de Lyon, Ecole Normale Supérieure de Lyon, Centre Léon Bérard, Centre de Recherche en Cancérologie de Lyon, INSERM 1052, CNRS 5286, F-69008 Lyon, France; philippe.bouvet@ens-lyon.fr 4 Center for Integrative Biology (CBI), LBME, University of Toulouse, UPS, CNRS, F-31062 Toulouse, France; anne-claire.lavigne@ibcg.biotoul.fr (A.-C.L.); kerstin.bystricky@ibcg.biotoul.fr (K.B.) * Correspondence: mbuschbeck@carrerasresearch.org; Tel.: +34-93-557-2800  Received: 31 March 2020; Accepted: 23 April 2020; Published: 30 April 2020  Abstract: MacroH2A histone variants have functions in differentiation, somatic cell reprogramming and cancer. However, at present, it is not clear how macroH2As affect gene regulation to exert these functions. We have parted from the initial observation that loss of total macroH2A1 led to a change in the morphology of murine myotubes differentiated ex vivo. The fusion of myoblasts to myotubes is a key process in embryonic myogenesis and highly relevant for muscle regeneration after acute or chronic injury. We have focused on this physiological process, to investigate the functions of the two splice isoforms of macroH2A1. Individual perturbation of the two isoforms in myotubes forming in vitro from myogenic C2C12 cells showed an opposing phenotype, with macroH2A1.1 enhancing, and macroH2A1.2 reducing, fusion. Differential regulation of a subset of fusion-related genes encoding components of the extracellular matrix and cell surface receptors for adhesion correlated with these phenotypes. We describe, for the first time, splice isoform-specific phenotypes for the histone variant macroH2A1 in a physiologic process and provide evidence for a novel underlying molecular mechanism of gene regulation. Keywords: histone variants; myogenic differentiation; myotubes; cell fusion; macroH2A; PARP1; gene regulation; ADP ribose 1. Introduction Histone variants can replace replication-coupled histones in the nucleosome and thus convey unique properties to specific regions of the genome [1]. MacroH2A is a member of the group of H2A histone variants and is unique in having a tripartite domain structure consisting of a histone fold, an unstructured linker and a globular macrodomain [2]. Two genes, H2AFY and H2AFY2, and one event of alternative splicing give rise to three proteins sharing the same overall domain structure: the splice variants macroH2A1.1 and macroH2A1.2 and the second gene product macroH2A2 [3]. A number of loss-of-function studies concluded that macroH2As stabilize the epigenome of differentiated cells. Cells 2020, 9, 1109; doi:10.3390/cells9051109 www.mdpi.com/journal/cells Cells 2020, 9, 1109 2 of 17 In this context, macroH2As were found to provide robustness to differentiation and developmental processes [4,5]. Conversely, macroH2As act as barriers to somatic cell reprogramming [6–8]. In cancer, macroH2As primarily have tumor-suppressive functions, which were first described in melanoma [9]. Some isoform-related and cancer-type-related dependences have also been reported [10]. At the present, it is not clear how exactly macroH2As exert their impact on differentiation, cancer and reprogramming. The hypothesis put forward in most of these studies is that the physiological impact of macroH2As is a consequence of downstream changes in gene regulation. This may be a direct consequence of modulating transcription factor binding to DNA [11–13] or an indirect consequence of changing the three-dimensional organization of chromatin in the nucleus [14]. The difference between the two splice-isoforms, macroH2A1.1 and macroH2A1.2, is the usage of a mutually exclusive exon that encodes a number of amino acids defining the form and hydrophobicity of a cavity in the macrodomain [15]. As a consequence, macroH2A1.1, but not macroH2A1.2, is able to bind NAD+-derived ADP-ribose, poly-ADP-ribose and ADP-ribosylated PARP1 [16]. In particular, the interaction with the key stress sensor PARP1 can have different consequences. The recruitment of active PARP1 to macroH2A1.1-bound genes can positively or negatively affect their expression [17]. Furthermore, in cases where macroH2A1.1 is more abundant than PARP1, such as in muscle cells, macroH2A1.1 acts as endogenous inhibitor of PARP1 [18,19]. As PARP1 is an avid consumer of nuclear NAD+, its inhibition can have a global impact on NAD+ metabolism affecting reactions in distal organelles, such as oxidative phosphorylation in the mitochondria (discussed in [20]). The expression of both macroH2A1 splice isoforms is dynamically regulated during myogenesis. The early steps of myogenesis can be recapitulated ex vivo. Primary myoblasts or immortal myogenic cell lines proliferate rapidly in cell culture, but after a change to serum-restricted medium, they quickly exit the cell cycle, upregulate differentiation, promoting transcription factors such as MyoD and Myogenin, aggregate and fuse to form multinucleated myotubes [21]. MacroH2A1.2 is the predominantly expressed isoform in proliferating myoblasts and contributes to the activation of enhancers by recruiting the transcription factor Pbx1 at the beginning of the differentiation process [13]. During myogenesis, a switch in splicing leads to the downregulation of the macroH2A1.2 transcript and an increase in the macroH2A1.1 transcript, yielding comparable protein levels in myogenic C2C12 cells after four days of differentiation [19]. How macroH2As and, in particular, the two splice isoforms, macroH2A1.1 and macroHA1.2, influence physiological processes is not well understood. Here, we have used differentiating myotubes as a model system to examine the role of macroH2A1 splice variants in a differentiation process. We find that the two isoforms have an opposing effect on myotube fusion. The effect is mediated by the regulation of a number of genes encoding adhesion and extracellular matrix proteins. 2. Materials and Methods 2.1. Cell Culture and Isolation of Primary Myoblasts Murine C2C12 cells were obtained from ATCC (CRL-1772) and cultured and differentiated as described [19]. Muscles were collected from 2-month-old macroH2A1 knockout mice and control wild-type littermates with a mixed C57Bl/6 and 129Ola background [22]. These experiments were approved by the Ethics Committee for Animal Experimentation of the Ecole Normale Supérieure de Lyon (CECCAPP) under authorization from the National Committee for Ethical Reflection on Animal Experimentation (authorization N◦ 15) on February 22, 2012 with the registered number ENS_2012_008. Primary myoblasts were isolated, cultured and differentiated, as described elsewhere [23]. In brief, mouse primary myoblasts were maintained on collagen-coated dishes in Ham’s F10 medium complemented with 20% FBS (Invitrogen), 10 ng/mL of bFGF (Invitrogen), 0.1% Fungizone (Invitrogen) and penicillin– streptomycin. For maintenance, mouse primary myoblasts were seeded on plates coated with rat-tail collagen I (BD Biosciences). For differentiation, cells were cultured on plates coated with Matrigel™ (BD Biosciences). Cells 2020, 9, 1109 3 of 17 2.2. Plasmids, Transfection and Retroviral Infection For the knockdown of macroH2A1 isoforms, we used siRNAs (Invitrogen) that were previously described [19,24]. For all siRNAs, 10 nM was found to be the optimal concentration to achieve a maximal depletion. Briefly, in a 6-well plate, 20,000 C2C12 cells were seeded. C2C12 cells were then transfected with the transfection reaction composed of 500 µL of Opti-MEM (Gibco) 3.12 µL of lipofectamine RNAiMAX Transfection Reagent (Invitrogen) and 10 nM of siRNA. For knockdown in myotubes, siRNA was repeatedly delivered at day +1 after seeding, the day of medium change for differentiation and 2 days after. In the case of p15 plate, 300,000 C2C12 cells were seeded and transfected with 7.5 mL of Opti-MEM, 45 uL of lipofectamine and siRNA at a final concentration of 10 nM. For the reintroduction of Flag-macroH2A1.1 plasmid into primary myoblasts, we generated a retroviral vector, using standard cloning techniques. The mouse macroH2A1.1 was amplified from C2C12 cell cDNA and inserted into a retroviral pBabe.puro backbone. Infection was essentially performed as previously described [25]. Sequences of oligos used for cloning are given in Table A1 in Appendix A. 2.3. Antibodies and Immunofluorescence Please find a description of all used antibodies in Table A2 in Appendix A. The macroH2A1.1 antibody has been generated in the laboratory of Prof. Kerstin Bystricky and is characterized in [26] For immunofluorescence, 20,000 C2C12 cells were seeded in 6 well-plates and cover slips. On day 4 of differentiation, C2C12 cells were fixed in 4% paraformaldehyde for 10 min at room temperature. Cells were permeabilized for 10 min, using a solution of 0.1 M HCl and 0.5% Triton- × 100 in PBS, and washed three times with PBS containing 0.1% Tween 20 (PBST). Then, slides were pre-blocked with PBST containing 5% BSA, for 30 min, at room temperature, and incubated with a 1/50 to 1/100 dilution of specific primary antibody for 2 h, in PBST containing 5% BSA. After three washes with PBST, the slides were incubated for 1 h, at room temperature, with a 1/100 dilution of secondary goat anti-rabbit antibody conjugated to Alexa 488 (Thermo Fischer). After successive washes with PBST, slides were mounted with Vectashield® Mounting Medium with DAPI. 2.4. Image Data Analysis Images of the C2C12 immunofluorescence were obtained by using a Leica DMI6000B Advance Fluorescence microscope (Leica) microscope equipped with a 63 × /1.4 Plan-Apochromat oil immersion objective. Groups of 10 to 20 images were loaded and analyzed in Fiji, a distribution of ImageJ [27]. Myotubes were directly drawn as Regions of Interest (ROIs) on the selected image, using Fiji’s manual selection tools on the basis of eMHC staining. The images were then automatically analyzed by using a custom ImageJ script, with the aim of calculating the number of nuclei inside each myotube, the total number of nuclei in each image and measurements of each myotube. In short, after background removal with a rolling ball algorithm, the DAPI signal was automatically thresholded, and artifacts were removed with binary image functions and subject to the “analyze particles” function, to identify and count all nuclei in the image. Then, the centroid of each nucleus was extracted, and each region defined as eMHC positive myotubes was evaluated to count the number of nuclei it contains. The total number of nuclei was normalized to the total imaged surface. The differentiation index was calculated as the ratio of nuclei in eMHC-positive cells over the total nuclei. Nuclei distribution per myotube was analyzed by the distribution of nuclei in eMHC-positive cells that contain at least 2 nuclei. The myotube distribution shows different categories of myotubes containing the selected number of nuclei. Total nuclei number and differentiation index were statistically analyzed with Student’s t-test, and we used Wilcoxon test for fusion parameters that were not normally distributed. 2.5. Gene Expression Analysis RNA was extracted by fusing the PureLinkTM RNA Mini Kit (Ambion, Life Technologies), including on-column PureLink DNase treatment. RNA concentration was checked by using Nanodrop Cells 2020, 9, 1109 4 of 17 (ThermoScientific). For RNA-seq, RNA was sent for library preparation to the Genomics Unit of the Centre for Genomic Regulation (CRG, Barcelona, Spain). RNA quality control showed the RNA integrity number (RIN) was between 9.9 and 10. RNA was amplified by using mRNAseq, which selects polyA. The sequencing was performed on single read (SR), with an average length of reads of 50 bp, on an Illumina HiSeq2500 sequencer. RNAseq analysis was performed by using a pipeline assembled in our laboratory. The quantification of the expression was performed by using Salmon software (version 0.13.1) [28] with the quasi-mapping mode on Mus_musculus GRCm38 assembly genome. The statistical analysis was performed by using DESeq2 package from Bioconductor. The raw quantification data obtained with Salmon was imported in DSEq2, using tximport package. Then, we performed three different contrasts, to evaluate the difference in the expression between wild type, si macroH2A1.1 and si macroH2A1.2. The results of this test were corrected, using Benjamini Hochberg procedure (False Discovery Rate). Finally, data of the different contrasts were selected, using an adjusted p-value equal or less of 0.05. Gene ontology analysis was performed based on the annotated mouse genome from org.Mm.eg.db (Bioconductor). Pathways analysis was obtained, using Paintomics3 [29]. For RT-qPCR, Complementary DNA (cDNA) was synthesized from 1 µg of RNA by using a first-strand cDNA synthesis kit (Fermentas). Semiquantitative cDNA levels were determined by real-time PCR. Relative expression of genes of interest was determined by normalizing the levels of corresponding cDNAs to the average of the two housekeeping genes Gapdh and Rpl7. All oligos were purchased from Invitrogen (sequences are given in Table A1). 2.6. Chromatin immunoprecipitation, Sequencing and Data Analysis Chromatin immunoprecipitation was performed, essentially, as described elsewhere [30]. For each immunoprecipitation, 25 µg of chromatin and 1–3 µg of antibody were used. Final concentration was determined by QubitTM dsDNA HS Assay (Invitrogen). Precipitated DNA was analyzed by ChIP-qPCR, and the data were normalized on the diluted input (1/50). For ChIP-sequencing (ChIP-seq), 10 ng of enriched DNA was used for library generation and parallel sequencing on Illumina Genome Analyzer at the Genomics Unit of the CRG (Barcelona, Spain). After data cleaning and trimming, reads were aligned to the mouse genome (mm9), using Bowtie 2 version 2.0.6, with sensitive presetting option (-D 15 -R 2-L 22 -i S,1,1.15) [31]. Enriched genomic regions for multiple overlapping ((peaks) SICER software version 1.1) were used to identify enriched regions using the following settings: redundancy threshold = 2, window size = 600, fragment size = 250, effective genome fraction = 0.75, gap = 1200 and FDR = 0.05 [32]. Datasets: The here described RNA-seq and MacroH2A1.1 ChIP-seq data can be accessed at GEO through the accession number GSE148045. The ChIP-seq data of total macroH2A1 have been previously described [19] and are accessible through GSE78257. The ChIP-seq data of histone H3 K4me2 in myotubes were generated by others and are accessible through GSM918413. 3. Results 3.1. Loss of Total MacroH2A1 Reduces the Capacity of Primary Myoblasts to Fuse Mice lacking both isoforms of macroH2A1 are viable and fertile [22,33]. We took advantage of the availability of these mice to take a first look at a possible function of macroH2A1 in myogenic differentiation. For this, we isolated quadriceps from macroH2A1 knockout (KO) mice and sex-matched littermate controls and verified the loss of expression by immunoblotting (Figure 1A). Similar to the original studies describing the KO mice, we did not observe any compensation by macroH2A2. Previous quantification of the relative mRNA levels in myotubes indicated that macroH2A2 is expressed at a much lower level than macroH2A1 [19]. In order to investigate the physiological relevance of macroH2A1 in muscle maturation, we isolated primary myoblasts from macroH2A1 wild-type control and knockout (KO) mice and assessed their capacity to differentiate into myotubes ex vivo. As shown in a representative experiment in Figure 1B, although both WT and KO myoblast were able to differentiate Cells 2020, Cells 9, 1109 2020, 9, x FOR PEER REVIEW 55 of of 17 18 into myotubes ex vivo. As shown in a representative experiment in Figure 1B, although both WT and and expressedwere KO myoblast the differentiation marker able to differentiate andembryonic expressed myosin heavy chain, the differentiation the resulting marker embryonic myotubes myosin displayed morphological differences. The lack of large myotubes in macroH2A1 heavy chain, the resulting myotubes displayed morphological differences. The lack of large KO conditions was particularly myotubes obvious. KO in macroH2A1 As conditions cell fusion is wasa direct function particularly of cell density, obvious. we first As cell fusion is averified that the direct function number of nuclei was in a similar range in both conditions (Figure 1C). Next, we assessed of cell density, we first verified that the number of nuclei was in a similar range in both conditions commitment to differentiation (Figure 1C). Next,bywecounting assessed the number of commitment tonuclei in in mono differentiation or poly nucleated by counting the number cells expressing of nuclei in in embryonic myosin heavy chain (eMHC) as marker of differentiation. We mono or poly nucleated cells expressing embryonic myosin heavy chain (eMHC) as marker of found that the commitment to differentiation differentiation. was We conserved found that theand not affected commitment toindifferentiation KO cells. However, when counting was conserved and notthe number affected in of nuclei in poly-nucleated cells, we found that fusion events were strongly KO cells. However, when counting the number of nuclei in poly-nucleated cells, we found that reduced in the absence of macroH2A1. fusion events wereA more detailed strongly analysis reduced of absence in the the size distribution of macroH2A1. of the myotubes A more in all detailed experiments, analysis of the together, indicated a reduction of the median number of nuclei per myotube. In size distribution of the myotubes in all experiments, together, indicated a reduction of the median particular, fewer macroH2A1 KO myotubes number of nuclei withIn per myotube. a range of 4–16 particular, fewernuclei were found, macroH2A1 KO while the number myotubes of myotubes with a range of 4–16 with three or two nuclei and cells with a single nucleus were increased (Figure nuclei were found, while the number of myotubes with three or two nuclei and cells with a single 1D). Re-expression of exogenous nucleus were macroH2A1.1 increased (Figurein macroH2A1 KO myoblasts 1D). Re-expression led to amacroH2A1.1 of exogenous visible increasein in the formation macroH2A1 KO of larger myotubes (Figure 1E). Taken together, these results indicate that macroH2A1 myoblasts led to a visible increase in the formation of larger myotubes (Figure 1E). Taken together, promotes the fusion of myoblasts to myotubes. these results indicate that macroH2A1 promotes the fusion of myoblasts to myotubes. Figure 1. Myoblasts lacking macroH2A1 fail to form proper myotubes. (A) Immunoblot analysis of Figure 1. Myoblasts macroH2A1 (mH2A1), lacking mH2A2 macroH2A1 fail to and Histone H3form proper myotubes. of quadriceps (A) Immunoblot from wild-type (WT) andanalysis knockout of macroH2A1 (KO) mice. (mH2A1), mH2A2 and (B) Immunostaining ofHistone H3 of quadriceps differentiated primary mH2A1 from wild-type WT and KO (WT)myotubes and knockout with (KO) mice. (B)myosin anti-embryonic Immunostaining heavy chainof(eMHC). differentiated primary (C) Total mH2A1ofWT nuclei, percent and KO myotubes differentiated eMHC positivewith anti-embryonic cells and percentmyosin heavy chain of differentiated eMHC(eMHC). (C)cells positive Total nuclei, with percent at least of differentiated two nuclei eMHC after four days of positive cells and percent of differentiated eMHC positive cells with at least two nuclei differentiation. Data points are from sets of four to six photos obtained from three independent biological after four days of differentiation. replicates, evaluating 100 andData200 points are from myotubes and thesets sameof surface four toareasixperphotos obtained replicate. from three Lines indicate the independent median biological of replicates, *p