Deubiquitinase catalytic activity of MYSM1 is essential in vivo for hematopoiesis and immune cell development

Myb-like SWIRM and MPN domains 1 (MYSM1) is a chromatin binding protein with deubiquitinase (DUB) catalytic activity. Rare MYSM1 mutations in human patients result in an inherited bone marrow failure syndrome, highlighting the biomedical significance of MYSM1 in the hematopoietic system. We and others characterized Mysm1-knockout mice as a model of this disorder and established that MYSM1 regulates hematopoietic function and leukocyte development in such models through different mechanisms. It is, however, unknown whether the DUB catalytic activity of MYSM1 is universally required for its many functions and for the maintenance of hematopoiesis in vivo. To test this, here we generated a new mouse strain carrying a Mysm1D660N point mutation (Mysm1DN) and demonstrated that the mutation renders MYSM1 protein catalytically inactive. We characterized Mysm1DN/DN and Mysm1fl/DN CreERT2 mice, against appropriate controls, for constitutive and inducible loss of MYSM1 catalytic function. We report a profound similarity in the developmental, hematopoietic, and immune phenotypes resulting from the loss of MYSM1 catalytic function and the full loss of MYSM1 protein. Overall, our work for the first time establishes the critical role of MYSM1 DUB catalytic activity in vivo in hematopoiesis, leukocyte development, and other aspects of mammalian physiology.


Loss of catalytic activity in the MYSM1 D660N mutant protein.
To generate an allele encoding a catalytically inactive MYSM1 in mouse we chose to introduce the Mysm1 D660N point mutation at the highly conserved aspartic acid 660 residue within the JAMM motif of the MPN catalytic domain (Fig. 1A). This residue is considered essential to the catalytic mechanism and is predicted to interact with both Zn 2+ and the substrate 3,4 . To confirm that the catalytic activity of MYSM1 D660N is indeed impaired, we expressed and purified both wildtype MYSM1 and MYSM D660N proteins from Sf9 insect cells, and performed an in vitro catalytic activity assay using Ubiquitin-Rhodamine 110 as substrate. The proteins were purified from Sf9 insect cells at a similar yield (0.78 mg/L of cells for wild type MYSM1, 0.83 mg/L of cells for MYSM1 D660N ) and eluted in a retention volume slightly lower than 13 mL in size exclusion chromatography, demonstrating that they were not aggregated and likely not misfolded (Fig. S1A). We found that wild-type MYSM1 cleaved the substrate in a dose-dependent manner with a K m of 8.5 μM for the substrate, whereas the activity of MYSM D660N was completely abrogated (Fig. 1B). This confirms that the D660N mutation inactivates the catalytic activity of mouse MYSM1 protein.
Generation of the Mysm1 D660N mouse strain. We used CRISPR/Cas9-mediated genome editing in zygotes to introduce the point mutation into exon 16 of Mysm1. C57BL/6 zygotes were co-injected with Cas9 protein and a gRNA, along with two homology-dependent recombination (HDR) templates. Since conventional knockout of MYSM1 causes partial embryonic lethality, we used two HDR templates to increase the efficiency of the procedure: one HDR template for the introduction of the D660N mutation and a second template introducing silent mutations to disrupt the gRNA recognition site 24 . The founder harboring both the D660N and silent mutations was backcrossed onto C57BL/6 mice to generate heterozygous Mysm1 D660N mice lacking the silent mutations on the other Mysm1 allele. Sanger sequencing of the genomic DNA demonstrated successful introduction of the point mutation that translates into the MYSM1 D660N amino acid substitution in the protein (Fig. 1C). The sequencing window covered 371 nucleotides (NCBI GRCm39 Mysm1, Gene ID: 320713, range from 94,840,309 to 94,840,679), and included the entire Mysm1 exon 16 and ≥ 80 nucleotides of the flanking introns at both ends. This demonstrated no other mutations apart from those shown in Fig. 1C and resulting in the D660N substitution in the protein. As the mutations are located > 20 nucleotides away from the 3′ splice site of Mysm1 exon-16 they are not expected to disrupt splicing 25 ; and further analysis of the Mysm1 D660N allele with the Spliceator online tool (www. lbgi. fr/ splic eator/) predicted no changes in splicing. Furthermore, RT-qPCR analysis of mouse bone marrow cells with the primer pairs spanning Mysm1 exon junctions 15-16 and 16-17 demonstrated no changes in the levels of Mysm1 transcript successfully spliced across these exon junctions in Mysm1 DN/DN compared to Mysm1 +/+ control cells (Fig. S1B).
Mysm1 DN/DN mouse model: partial embryonic lethality and developmental phenotypes. Mysm1 DN/DN mice were born in sub-Mendelian numbers, with only ~ 4% of offspring from an intercross of two heterozygous parents having the Mysm1 DN/DN genotype, indicating that the loss of MYSM1 catalytic activity causes increased embryonic lethality (Fig. 1D). At adulthood, Mysm1 DN/DN mice were significantly smaller in length and weight than their littermates (Fig. 1E,F), and had abnormally short tails (Fig. 1F), as previously seen in the Mysm1 −/− mice 8,9 . Importantly, we demonstrated similar levels of MYSM1 protein in Mysm1 DN/ DN and control Mysm1 +/+ bone marrow cells, and the expected loss of MYSM1 protein expression in Mysm1 −/− cells (Fig. 1G). Overall, we highlight the similarity in the gross developmental phenotypes of the Mysm1 DN/DN and Mysm1 −/− mouse strains, and establish the essential role of the MYSM1 DUB catalytic activity in vivo. www.nature.com/scientificreports/ a strong depletion of MYSM1 protein in Mysm1 Δ/Δ mouse splenocytes, but comparable retention of MYSM1 protein levels in Mysm1 Δ/+ and Mysm1 Δ/DN samples (Fig. 1H). The Cre ERT2 Mysm1 fl/DN model will test the effects of the loss of MYSM1 DUB catalytic activity on the maintenance of hematopoiesis, leukocyte development, and other aspects of mammalian physiology, independently of its roles in mouse development and the significant developmental phenotypes seen in the Mysm1 DN/DN mouse strain.
Severe hematologic dysfunction in Mysm1 DN/DN and Mysm1 fl/DN Cre ERT2 mice. Hematology analyses of the blood of Mysm1 DN/DN and Mysm1 −/− mice relative to the Mysm1 +/+ controls, demonstrated severe hematopoietic dysfunction, characterized by macrocytic anemia, with reduction in blood erythrocyte counts, hematocrit, and hemoglobin concentration, as well as an increased in mean corpuscular volume (MCV, Fig. 2A). Severe depletion of leukocytes and lymphocytes in Mysm1 DN/DN relative to control Mysm1 +/+ mice was also observed ( Fig. 2A). Overall, the reported anemia and leukopenia phenotypes of Mysm1 DN/DN mice are highly consistent with those observed in the Mysm1 −/− mouse model ( Fig. 2A), and also clinically in the patients with MYSM1 loss-of-function mutations 1,5-7 .
We conducted further hematology analyses on tamoxifen-treated Cre ERT2 transgenic mice of Mysm1 +/fl , Mysm1 fl/fl , and Mysm1 DN/fl genotypes, and observed highly similar hematopoietic phenotypes in the Mysm1 Δ/DN mice, including macrocytic anemia, leukopenia, and lymphocyte depletion (Fig. 2B). We further observed an increase in platelets in Mysm1 DN/Δ mice (Fig. 2B), while platelets were not quantified in the Mysm1 DN/DN model due to increased clotting of the blood samples. Elevated platelet counts were previously reported for the Mysm1 −/− mouse strain 1,8 , and although the mechanisms remain poorly understood they may be linked to elevated inflammatory response in Mysm1 −/− mice [14][15][16] , as thrombocytosis is a common feature of systemic inflammation 27 . Overall, we demonstrate that the loss of MYSM1 DUB catalytic activity in either constitutive or inducible mouse models results in a severe hematologic dysfunction with highly similar phenotypes to the previously characterized Mysm1 −/− and Mysm1 Δ/Δ mouse strains.   www.nature.com/scientificreports/ nificance (Fig. 5A). The numbers of HSC and MPP1-3 cells were highly variable between the Mysm1 DN/DN mice, and showed trends for expansion, which however did not reach statistical significance (Fig. 5B,C), and this likely reflects the competing effects of the loss of HSC quiescence and increased cell apoptosis, as previously reported in the Mysm1 −/− mouse models 10,19 . Importantly, there was a severe depletion of the lymphoid primed MPP4 cells in both Mysm1 DN/DN and Mysm1 −/− relative to control mice (Fig. 5B,C), further supporting the essential role of MYSM1 DUB catalytic activity for lymphopoiesis. Furthermore, an increase in the proportion of dead cells was observed particularly for lymphoid biased MPP4 and CLP cells in Mysm1 DN/DN relative to control Mysm1 +/+ mice ( Fig. S2C-D).   www.nature.com/scientificreports/ meras were set up. CD45.1 + wild type bone marrow was mixed in a 1:1 ratio with the bone marrow of CD45.2 + Cre ERT2 mice of Mysm1 +/fl , Mysm1 fl/fl , or Mysm1 DN/fl genotypes, and transplanted into three independent groups of lethally irradiated recipients (Fig. 6A). The recipient mice were bled at 12-weeks to confirm the normal reconstitution with donor bone marrow across the genotypes (data not shown), and subsequently all the mice were administered with tamoxifen to induce the Mysm1 fl to Mysm1 Δ allele conversion. The mice were analyzed for the relative contributions of the CD45.2 + bone marrow to the different hematopoietic lineage, across the three Mysm1 genotypes. We observed a significant reduction in the contribution of the Mysm1 DN/Δ donor hematopoiesis to the B cell, CD4 T cell, CD8 T cell, and NK cell populations in the mouse spleen (Fig. 6B), to monocyte and neutrophil populations in both spleen and bone marrow (Fig. 6B and not shown), and to all the leukocyte populations in the www.nature.com/scientificreports/ mouse blood (Fig. S4A). Similar defects in the reconstitution were observed for the Mysm1 DN/Δ hematopoietic progenitor cells, including the lineage committed progenitors (CMPs, GMPs, CLPs, MEPs, and MkPs, Fig. 6C), all the developing B cell subsets (Fractions A-C, pre-B, and immature B cells, Fig. S4B), and the majority of T cell precursor subsets within the thymus (Fig. S4C). Among the multipotent HSC and MPP hematopoietic cells there was no defect in Mysm1 DN/Δ reconstitution of the early HSC and MPP1-2 cells, likely reflecting the balancing effects of the loss of quiescence and increase in apoptosis among these cells, as in the Mysm1 −/− mouse models 10,19 , however impaired reconstitution was seen for the latter myeloid-biased MPP3 and lymphoid-biased MPP4 subsets (Fig. S4D). Importantly, throughout the datasets presented above the Mysm1 DN/Δ phenotypes aligned very well with the Mysm1 Δ/Δ group, both showing strong impairment of hematopoietic function relative to the Mysm1 +/Δ control (Fig. 6, S4). Overall, this demonstrates the essential and cell-intrinsic role of the MYSM1 DUB catalytic activity in the regulation of hematopoiesis, and suggests lack of significant MYSM1 mechanisms of action that are independent of its catalytic function.

Discussion
In this study, we for the first time establish and characterize a mouse strain expressing a catalytically inactive MYSM1 D660N protein. We demonstrate a profound similarity in the developmental, hematopoietic, and immune phenotypes of Mysm1 Δ and Mysm1 DN mice, indicating the critical role of MYSM1 DUB catalytic activity in hematopoiesis and other aspects of mammalian physiology. While the depletion of hematopoietic cells and HSPC failure in functional assays were highly consistent between the Mysm1 −/− and Mysm1 DN/DN strains throughout this study, the Mysm1 DN/DN strain showed high embryonic lethality, with only ~ 4% of Mysm1 DN/DN offspring from an intercross of heterozygous parents, as compared to 10% for the Mysm1 −/− strain reported in previous studies 8 . Although these figures are not directly comparable between studies, it is interesting to note that increased embryonic lethality has been reported in other strains expressing catalytically inactive proteins relative to the corresponding knockout strains 28,29 . While the current study provides an in-depth analysis of the role of MYSM1 DUB catalytic activity in hematopoiesis and leukocyte development, the International Mouse Phenotyping consortium 30,31 and previously published studies report complex phenotypes in many other physiological systems in Mysm1 −/− mice, including alterations in skeletal, skin, and adipose physiology 21,22,32,33 . In future work, a broader comparison of Mysm1 −/− and Mysm1 DN/DN mouse strains will allow us to further explore the role of MYSM1 catalytic activity in these other physiological systems. Recent studies also established MYSM1 as a negative regulator of inflammatory responses to microbial stimuli in macrophages [14][15][16] . In these studies MYSM1 was shown to deubiquitinate TRAFs, RIP2, and STING proteins in the signal transduction cascades of innate immunity [14][15][16] . The Mysm1 DN/DN mouse strain developed in our current work may be used to further validate the role of MYSM1 catalytic activity in the regulation of innate immune and inflammatory responses in vivo.
Although no molecular analyses for the mechanisms of hematopoietic failure were conducted in the Mysm1 DN/ DN and Mysm1 DN/Δ mouse models in the current study, the very high concordance of their phenotypes to those of the Mysm1 −/− and Mysm1 Δ/Δ mouse strains suggests that similar molecular mechanisms are at play. Previously in the Mysm1 −/− mice the hematopoietic failure was shown to be driven by the activation of p53 and the induction of its pro-apoptotic transcriptional programs [18][19][20] . Consistently, our current data showed some increase in cell death in Mysm1 DN/DN relative to control Mysm1 +/+ hematopoietic and immune cells, supporting that similar mechanisms may also mediate the hematopoietic failure in Mysm1 DN/DN mice.  www.nature.com/scientificreports/ With the validation of the role of MYSM1 DUB catalytic activity in hematopoiesis, it will be important to identify the full range of protein targets and substrates of MYSM1, for example through proteomics approaches [34][35][36] . While histone H2A-K119ub is a highly abundant and well characterized MYSM1 substrate 2,37 , MYSM1 can also cleave K63, M1, K6, and K27 polyubiquitin in vitro 14 and regulates K63-polyubiquitination of TRAFs, RIP2, and STING proteins in macrophages [14][15][16] . Given the diverse and complex roles of ubiquitination in regulating chromatin accessibility, gene expression, genomic stability, signal transduction, protein localization and many other cellular processes [37][38][39] , such studies may lead to the discovery of further novel MYSM1 substrates beyond histone H2A-K119ub and advance the understanding of its functions and mechanisms of action.
We recently established that the loss of MYSM1 in mouse models of cMYC-driven B cell lymphoma can protect against disease onset and progression 40 . At the cellular and molecular levels, the protective effects were attributed to the role of MYSM1 in the cMYC-dependent induction of the genes encoding ribosomal proteins in the tumor cells (Rps/Rpl genes), with MYSM1-loss resulting in reduced Rps/Rpl transcript levels, reduced cellular protein synthesis rates, and the activation of p53 tumour suppressor 40 . Overall, these studies may suggest MYSM1 as a drug-target for cMYC driven hematologic malignancies; and the Mysm1 DN mouse strain described in our current work will allow to test whether the loss of MYSM1 DUB catalytic function can offer similar therapeutic benefits. This can serve as a proof-of-concept for the development of pharmacological MYSM1 inhibitors and for the assessment of their activities in experimental models of cMYC-driven hematologic malignancies.
In summary, our study establishes the primary and indispensable function of MYSM1 as a DUB in vivo in the normal progression of mammalian development, hematopoiesis, and immune cell production. This work also provides a mouse model for further analyses of the roles of MYSM1 DUB catalytic functions in vivo in many other aspects of mammalian physiology. Mouse genotyping for the Mysm1 D660N allele was performed with a custom designed TaqMan SNP Genotyping assay and TaqMan Genotyping Master Mix on a StepOnePlus instrument (all reagents from ThermoFisher Scientific). Other genotyping was performed by conventional genomic PCR with DreamTaq DNA Polymerase (ThermoFisher Scientific) and primers from Integrated DNA Technologies.

Tamoxifen mouse treatment.
For tamoxifen-induced Mysm1-gene deletion, mice of Mysm1 fl/fl Cre ERT2 and control genotypes were injected intraperitoneally with tamoxifen (Sigma-Aldrich, T5648) in sterilized corn oil at 0.12 mg per gram body weight per injection, with 8 doses administered in total over 16 days, as in our previous work 17,26,42 . Successful deletion of Mysm1 exon 3 was validated by PCR analyses of the genomic DNA from lymphoid organs of the mice, as described previously 26,42 . Mouse bone marrow transplantation. For competitive bone marrow transplantations, recipient wild type B6.SJL-PtprcaPepcb/Boy (JAX002014, congenic for CD45.1) mice were irradiated with 2 doses of 4.5 Gy, delivered 3 h apart, in an RS2000 irradiator (Rad Source). Wild-type CD45.1-marked bone marrow cells were mixed in a 1:1 ratio with bone marrow cells from mice of Mysm1 fl/+ Cre ERT2 , Mysm1 fl/fl Cre ERT2 , or Mysm1 fl/DN Cre-ERT2 genotypes, and the mixes transplanted into three independent cohorts of recipient mice via intravenous injection. The mice were kept on neomycin in drinking water (2 g/l, BioShop) for 3 week. Successful reconstitution of the hematopoietic system by donor cells was confirmed with a bleed and flow cytometry analysis at 12 weeks, and was followed with tamoxifen treatment to induce Mysm1 fl to Mysm1 Δ allele conversion and further studies to compare hematopoietic function across the Mysm1 genotypes. www.nature.com/scientificreports/ surface-markers in PBS with 2% FCS for 20 min on ice, using antibodies listed in Supplemental Table S1. Viability Dye eFluor ® 506 (ThermoFisher Scientific) was used to discriminate dead cells. Compensation was performed with BD™ CompBeads (BD Biosciences). The data were acquired on BD Fortessa and analyzed with FlowJo (Tree Star) software.
RNA Isolation and RT-qPCR. RNA was extracted using the MagMAX™ Total RNA Isolation Kit (Invitrogen, Thermo Fisher Scientific) according to manufacturer's protocol, and quantified with NanoDrop (Ther-moFisher Scientific). cDNA was prepared using the Moloney murine leukemia virus (MMLV) reverse-transcription kit and quantitative PCRs run on a StepOnePlus instrument with PowerSYBR master mix (all from ThermoFisher Scientific). Primers were purchased from Integrated DNA Technologies, and the sequences are provided in Supplemental Table S2.
In vitro fluorescence catalytic activity assay. Genes for wild-type and mutated mouse MYSM1 were sub-cloned into pFastBac vectors, which contain an N-terminal hexahistidine tag. The D660N mutation was introduced using standard site-directed mutagenesis protocols. MYSM1 proteins were expressed in Sf9 cells, infected with recombinant baculovirus and grown at 27 • C for ~ 66 h. Proteins were purified by sonicating the cells, followed by Ni-NTA affinity, Mono-Q and size exclusion chromatography (Superdex 200i). The final yield was ~ 0.8 mg/ L of cells. Enzymatic assays were conducted in assay buffer comprised of 20 mM Tris pH 8.1, 100 mM NaCl, 0.01% BSA, 1 mM DTT at room temperature in 96-well low binding black plates. 12.5 μL of either MYSM1 or MYSM1 D660N at 300 nM was added to the 96-well plate followed by the addition of 12.5μL of serially diluted substrate (0.02-40 μM) and immediately placed into a fluorescence spectrophotometer with excitation/emission wavelengths set to 485/535 nm, respectively.
Statistics. Statistical analyses used Prism 7.01 (GraphPad Inc.), with Student's two-tailed t-test for two datasets and ANOVA for multiple comparisons; p < 0.05 was considered significant. ARRIVE guidelines statement. The study is reported in accordance with ARRIVE guidelines. Experimental unit corresponds to a single animal. Mice used in the study included both males and females and were always sex-matched between the groups. Mice were also age-matched between the groups and were at least 8 weeks of age, however in the studies involving lengthy tamoxifen treatments and bone marrow transplantations they were significantly older at the endpoint of the study. Whenever possible mice of test and control groups were bred as litter-mates and maintained in shared cages. Mice were allocated to experimental groups based on genotype and with no randomization; staff carrying out the experiments was not blinded to group allocations; no a priori sample size calculations were performed.

Data availability
The relevant datasets used and analyzed during the current study are available from the corresponding author on reasonable request.