Protein arginine methyltransferases PRMT1, PRMT4/CARM1 and PRMT5 have distinct functions in control of osteoblast differentiation

Osteogenic differentiation of mesenchymal cells is controlled by epigenetic enzymes that regulate post-translational modifications of histones. Compared to acetyl or methyltransferases, the physiological functions of protein arginine methyltransferases (PRMTs) in osteoblast differentiation remain minimally understood. Therefore, we surveyed the expression and function of all nine mammalian PRMT members during osteoblast differentiation. RNA-seq gene expression profiling shows that Prmt1, Prmt4/Carm1 and Prmt5 represent the most prominently expressed PRMT subtypes in mouse calvarial bone and MC3T3 osteoblasts as well as human musculoskeletal tissues and mesenchymal stromal cells (MSCs). Based on effects of siRNA depletion, it appears that PRMT members have different functional effects: (i) loss of Prmt1 stimulates and (ii) loss of Prmt5 decreases calcium deposition of mouse MC3T3 osteoblasts, while (iii) loss of Carm1 is inconsequential for calcium deposition. Decreased Prmt5 suppresses expression of multiple genes involved in mineralization (e.g., Alpl, Ibsp, Phospho1) consistent with a positive role in osteogenesis. Depletion of Prmt1, Carm1 and Prmt5 has intricate but modest time-dependent effects on the expression of a panel of osteoblast differentiation and proliferation markers but does not change mRNA levels for select epigenetic regulators (e.g., Ezh1, Ezh2, Brd2 and Brd4). Treatment with the Class I PRMT inhibitor GSK715 enhances extracellular matrix mineralization of MC3T3 cells, while blocking formation of H3R17me2a but not H4R3me2a marks. In sum, Prmt1, Carm1 and Prmt5 have distinct biological roles during osteoblast differentiation, and different types histone H3 and H4 arginine methylation may contribute to the chromatin landscape during osteoblast differentiation.

The protein arginine methyltransferase (PRMT) family is comprised of nine evolutionarily conserved proteins that catalyze the methylation of arginine residues (Kleinschmidt et al., 2008). These enzymes have broad protein substrates ranging from histones to non-histone nuclear and cytoplasmic proteins that differ in their activity by catalyzing asymmetrical arginine methylation (Class I; e.g., PRMT1 and 4), symmetrical arginine methylation (Class II; e.g., PRMT5) or arginine monomethylation (Class III; e.g., PRMT7). Each enzyme also has selectivity for different arginine moieties in histones H3 and H4, most of which represent gene activating marks. For example, PRMT1 asymmetrically catalyzes the formation of H4R3me2a (active mark), PRMT4/CARM1 catalyzes the formation of H3R17me2a (active mark), PRMT5 catalyzes the formation of symmetrical H3R2me2s (active mark), while PRMTs also mediate methylation of other less characterized histone marks (e.g., H3R26me2a, and H3R42me2a by PRMT4/CARM1) (Beacon et al., 2020;Jahan and Davie, 2015;Wu and Xu, 2012).
Methylation of arginine residues is fundamental for chromatin remodeling to support different cellular processes (e.g., RNA splicing, transcription, signal transduction, DNA damage response, and formation of phase separated domains) (Guccione and Richard, 2019;Min et al., 2019). Multiple PRMT family members have been shown to contribute to cell fate determination and metabolic pathways associated with human diseases, as well as mediate chromatin organization and function in different biological contexts (Jahan and Davie, 2015). Aberrant expression of PRMTs is correlated with some cancers (Kim and Ronai, 2020), including osteosarcoma (Li et al., 2017). PRMT1, PRMT4/ CARM1, PRMT5 and PRMT7 are prevalent isoforms that are functionally expressed in myogenic cells and osteosarcoma cells (Moena et al., 2020a;Blanc and Richard, 2017;Rakow et al., 2020). Three PRMTs (i.e., PRMT1, PRMT4/CARM1 and PRMT5) have each been shown to control vitamin D3-dependent transcription in osteosarcoma cells via interactions with SRC-1/NCOA1 (Moena et al., 2020b), whereas PRMT5 binds to the SWI/SNF component BRG1/SMARCA4 in different cell types (Seth-Vollenweider et al., 2014). These findings collectively suggest that PRMT proteins may also have a role in osteoblast differentiation. However, there are no studies that have comprehensively analyzed PRMT expression and function during MC3T3 osteoblast differentiation that would permit a specific focus on a key PRMT protein.
Because PRMTs have broad biological functions, a key question in bone biology is whether and how PRMT family members control osteogenic differentiation of mesenchymal stem/stromal cells (MSCs) and osteoblasts, as well as bone formation and bone homeostasis (i.e., osteoblast and osteoclast interplay). Initial studies have suggested individual roles of PRMTs in musculoskeletal tissues, including skeletal myoblast differentiation, chondrocyte proliferation and osteoblastic cells (Moena et al., 2020b;Seth-Vollenweider et al., 2014;Ito et al., 2009). In this study we surveyed the expression and functional contributions of multiple PRMTs to osteoblast differentiation. The results show that three distinct PRMTs (PRMT1, PRMT4 and PRMT5) regulate early stages of osteoblast differentiation by either enhancing or inhibiting osteoblast related gene expression and calcium deposition. Furthermore, selective pharmacological inhibition of PRMTs can enhance osteoblastogenesis.

High throughput RNA sequencing and bioinformatic analysis
RNA-seq data was retrieved from human musculoskeletal tissues including bone, cartilage, growth plate, muscle and adipose tissue, as previously reported (Paradise et al., 2018). In addition, we included RNA data from bone marrow derived MSCs from two different healthy donors that were subjected to osteogenic differentiation for fifteen days. In addition, we included data from mouse callus (Khani et al., 2017;Matthews et al., 2014), as well as undifferentiated mouse MC3T3 osteoblasts, ATDC5 chondrocytes and C2C12 cells that were cultured in their respective maintenance media (Dashti et al., 2023). High throughput mRNA sequencing and bioinformatic analyses were performed as previously reported (Dudakovic et al., 2014) with mRNA levels expressed as fragments per kilobase pair per million mapped reads (FPKM).

Cell culture
Our studies used MC3T3 E1 sc4 murine calvarial osteoblasts [PMID: 10352097] that were acquired from the American Type Culture Collection. These cells were cultured at 37 • C in 5 % CO 2 with complete maintenance medium, which is α-minimal essential medium (Gibco) without ascorbic acid and supplemented with 10 % fetal bovine serum (Atlanta Biologicals), 100 units/ml penicillin (Gibco), and 100 μg/ml streptomycin (Gibco). MC3T3 cells were seeded in 6-or 12-well plates in maintenance medium (10,000 cells/cm 2 ). The following day (day 0), we divided the cultures into two groups. Cells were subjected to siRNA transfections with control (D-001810-10-20; GE Healthcare, Chicago, IL), Prmt1 siRNA (Mouse Prmt1 ON-TARGETplus siRNA SMARTpools; L-049497-00-0005; GE Healthcare), Prmt4/Carm1 siRNA (Mouse Carm1 ON-TARGETplus siRNA SMARTpools; L-048766-00-0005; GE Healthcare), Prmt5 siRNA (Mouse Prmt5 ON-TARGETplus siRNA SMARTpools; L-042281-00-0005; GE Healthcare). Transfections were performed using RNAiMAX as instructed by the manufacturer (Invitrogen). After 6 h, osteogenic medium consisting of complete maintenance medium supplemented with 50 μg/ml ascorbic acid (Sigma) and 4 mM β-glycerol phosphate (Sigma) was added, and the cells were cultured until harvest. Studies with the PRMT inhibitor GSK715 were performed similarly, except that instead of transfection, cells were treated with GSK715 one day after seeding (day 0; 1× treatment) or also after three days (day 0 and day 3; 2× treatment). For both siRNA and PRMT inhibitor studies, we selected the time of harvest based on the first visual appearance of calcified extracellular matrix nodules which typically occurs between days 21 and 32 depending on the passage number, actual seeding density and type of experimental treatment.

Quantitative real-time reverse transcriptase PCR (RT-qPCR)
Cells were lysed using TRIzol Reagent (Invitrogen) and RNA was isolated using the Direct-zol RNA MiniPrep Kit (Zymo Research) with on column DNA digestion kit (Zymo Research). Isolated RNA was reverse transcribed into cDNA using the SuperScript III First Strand Synthesis System (Invitrogen). Gene expression was quantified using real-time PCR in which each reaction was performed with 10 ng cDNA per 10 μl, QuantiTect SYBR Green PCR Kit (Qiagen), and the CFX384 Real-Time System machine (BioRad). Transcript levels were quantified using the 2 − ΔΔCt method and normalized to the housekeeping gene GAPDH/Gapdh using gene specific primer sequences (Table 1).

Alizarin Red staining
To assess calcium deposition in the extracellular matrix of MC3T3 cultures, we fixed cells in 10 % neutral buffered formalin for 20 min and stained with 2 % Alizarin Red (Sigma) for 12 min. Upon removal of the Alizarin Red, plates were rinsed for 3 times with distilled water and imaged using a standard plate scanner. Absorption of Alizarin Red stain was quantified with ImageJ software (Dudakovic et al., 2020). Because Alizarin Red staining is saturable, we harvested samples well before they reach maximal staining. We note that cultures were visually inspected for the first appearance of bone-like nodules in either the control or test samples (treated with siRNA or GSK715), and harvested simultaneously at a time that aims to maximize differences between control and the specific treatment. For this reason, there are differences in the extent of calcium deposition in the control samples in distinct experiments, because they were harvested on different days.

Statistics
Statistical analysis and generation of figures was performed with GraphPad Prism software (version 9) using Student's unpaired t-test or two-way ANOVA for multiple comparison analyses. Histochemical assays and RT-qPCR time course experiments were performed in triplicate for each time point; each RT-qPCR measurement was performed as a technical duplicate. Experiments were replicated three times and results from a representative course are shown. Significance is denoted by asterisks in the figures (p < 0.05, *; p < 0.01, **).

Selective expression of PRMT genes in mouse osteoblasts and human bone tissues
The PRMT family of methyl arginine transferases has been conserved during mammalian genome evolution, reflecting the important contributions of these epigenetic regulators to embryonic development and epigenetic regulation. There are nine different PRMT isoforms which include Class I, II and III enzymes, with distinct protein domains; the variations in protein structure may have evolved to accommodate specific epigenetic functions in different tissues (Fig. 1A). We performed expression profiling using RNA-seq datasets from our previous studies to understand where and when human PRMT or mouse Prmt genes are expressed in musculoskeletal cell types and tissues. RNA-seq data from mesenchymal versus non-mesenchymal cell types in mouse callus during fracture repair show that Prmt1, Prmt4/Carm1 and Prmt5 are preferentially expressed in mesenchymal cell types compared to nonmesenchymal cell types in callus tissue (sorted using a dsTomato fluorescent protein lineage tracer) (Matthews et al., 2014) at day 2 after fracture in the callus site (Fig. S1). Additionally, the data show that multiple PRMT isoforms are expressed in adult human musculoskeletal tissues (Fig. S2). PRMT1, PRMT4/CARM1 and PRMT5 are the three most prominently expressed PRMT members across all tissues examined (i.e., bone, cartilage, muscle, growth plate and adipose tissue). Whereas PRMT1 is the highest expressed member in each of the tissues, PRMT4/ CARM1 and PRMT5 are particularly prominent in muscle and cartilage, respectively (Fig. S2). PRMT mRNAs were also examined in human bone marrow derived mesenchymal stem cells (BMSCs) (Fig. S3). Similar to expression in musculoskeletal tissues, we find that PRMT1, PMRT4/ CARM1 and PRMT5 are the most highly expressed PRMT members in human BMSCs, and their expression remains relatively constant during osteogenic lineage differentiation (Fig. S3). Furthermore, Prmt1, Prmt4 and Prmt5 are also highly expressed in frontal and parietal bone of calvaria from two different mouse strains (Fig. 1B&C). Similarly, these Table 1 Summary of RT-qPCR primers used in this study. The mRNAs for the indicated mouse genes (left column) were amplified using gene specific forward primers (middle column) and reverse primers (right column).

RT-qPCR primers (5′ -3′)
Gene symbol Forward Reverse three PRMT members are also robustly expressed in MC3T3 cells which represent immortalized osteoblast precursors derived from mouse calvaria (Fig. 1D). Expression of Prmt1, Prmt4/Carm1 and Prmt5 mRNAs and proteins in MC3T3 cells was confirmed by RT-qPCR and western blot analysis (Fig. 2). Taken together, these data suggest that PRMT1/ Prmt1, PMRT4/Prmt4 and PRMT5/Prmt5 are selectively expressed in both human and mouse musculoskeletal tissues. Because these three mRNAs are generally more prevalently expressed than other PRMTs, we prioritized these three genes for loss of function studies during osteoblastic differentiation.

Selective modulation of osteoblast differentiation upon siRNA depletion of PRMT1, PMRT4/CARM1 and PRMT5
To investigate the functional contributions of Prmt1, Prmt4/Carm1 and Prmt5 in mouse osteoblastogenesis, we conducted loss of function studies using siRNAs in MC3T3 cells. We examined short-term effects of siRNA depletion at both mRNA and protein levels (Fig. 2), as well as longer-term effects during a 2 week time-course (Fig. 3). To establish specificity of the siRNA knockdown, we also analyzed effects of each siRNA on the expression of the other two PRMT members, while also assessing the potential for cross-regulation by different PRMT subtypes. Transfection of cells with commercial siRNA pools results in specific knockdown of all three genes at both the mRNA-and protein-levels three days after siRNA treatment, as assessed by RT-qPCR ( Fig. 2A) and western blot analysis (Fig. 2B). Time-course analysis during MC3T3 osteoblast differentiation reveals that siRNA treatment during the initial 3 days of the culturing period dramatically alters the temporal expression of each Prmt isoform and expression returns to levels observed in control cells treated with non-silencing RNA after 10 to 14 days (Fig. 3). Hence, the siRNA depletion for each Prmt gene remains effective during the first week of the culturing period.
To understand whether depletion of one Prmt member affects the expression of the other two Prmt members, we examined the expression of all three members concurrently in response to siRNA depletion. We find that siRNAs for any Prmt member do not affect the expression of other Prmt members during the first three days of the culturing period as is expected (Figs. 2 & 3). Remarkably and unexpectedly, knockdown of Prmt1 results in a transient upregulation of Prmt4/Carm1 but not Prmt 5 at later cell culture stages (i.e., at day 7 and day 10) (Fig. 3). Prmt4/ Carm1 depletion appears to have a negative effect on Prmt1 levels on day 6, but the magnitude of this effect is marginal (Fig. 3). Collectively, these results suggest that siRNA knockdown of Prmt members may have effects on Prmt mRNA expression for 7 to 10 days. The results also suggest that interpretations of Prmt1 function after mRNA depletion may be confounded by a compensatory increase in mRNA levels of Prmt4/ Carm1. The latter could potentially substitute for Prmt1 during MC3T3 osteoblast differentiation. We assessed whether loss of Prmt1, Prmt4/Carm1 or Prmt5 affects calcium deposition using Alizarin Red staining (Figs. 4 & 5). Calcium accumulation is evident for control cells (treated with nsRNA) at day 30 of the cell culturing period (Fig. 4). Loss of Prmt1 increases calcium deposition (~2 fold increase; p < 0.001), while loss of Prmt4/Carm1 has no appreciable effects on calcium accumulation. In contrast, inhibition of Prmt5 dramatically decreases calcium deposition (~10-fold reduction; p < 0.01). There is no clear correlation between mineralization and Crystal Violet staining (Fig. 5), but the presence of Crystal Violet staining across each culture dish indicates that the absence of Alizarin Red staining is not simply due to local loss of cells. Taken together, all three Prmt members have different roles in osteoblast differentiation: Prmt1 inhibits and Prmt5 is required for osteoblast differentiation, while Prmt4/Carm1 loss by itself is dispensable for calcium deposition by osteoblasts. The finding that Prmt1 inhibits osteoblast differentiation is in agreement with other studies showing that this protein promotes the undifferentiated state of cells (Goruppi et al., 2023).

Loss of Prmt1, Prmt4/Carm1 and Prmt5 alters expression of stagespecific mRNA biomarkers for osteoblast differentiation
To understand whether depletion of PRMT proteins between day 0 and day 4 affects proliferation-related or osteoblast-specific gene expression, we examined cell cycle regulated genes (i.e., Cyclin B/Ccnb1 and Ki-67/Mki67) (Fig. S4). RT-qPCR analysis shows that siRNA depletion of each PRMT protein does not have strong numerical effects on cell cycle regulated mRNAs, including expression of the mitosisrelated mRNA marker Ccnb1 or the proliferation-related Mki67 mRNA (Fig. S4). We note that expression values at some time points exhibit statistical significance. However, the observed numerical changes cannot easily be interpreted and may not necessarily be biologically relevant. At face value, loss of CARM1 at day 7 reduces the mitosis related marker CCNB2 and increases the general proliferation marker MKI67. The latter would suggest that absence of CARM1 may decrease the G2/M population and increase the number of cells in G1 phase around day 7. Because loss of CARM1 does not alter Alizarin Red staining, these molecular changes have uncertain biological relevance. Furthermore, loss of PRMT5 results in a significant albeit quantitatively modest reduction (<2 fold) of both CCNB2 and MKI67, suggesting that PRMT5 loss may induce a quiescent cell state. The latter could potentially contribute to a reduced level of calcium deposition by preventing confluence of MC3T3 cell cultures (see Fig. 4). Crystal Violet data show qualitatively that cells are distributed across the entire cell culture plate, indicating that differences in Alizarin Red staining are not due to the complete absence of cells (Fig. 5). We conclude that effects of PRMT loss on cell proliferation or quiescence are not sufficiently dominant to account for changes in calcium deposition but may contribute to changes in Alizarin Red staining.
Although siRNAs typically are effective for up to four days, our data show that siPRMT1, siCARM1/PRMT4 an siPRMT5 effectively decrease  the mRNA expression of the corresponding PRMT members for about one week. We extended this time period of analysis to 14 days to examine the recovery of MC3T3 cells after siRNA depletion and to assess effects on mature osteoblast markers (e.g., Col1a1, Alpl, Ibsp, Phospho1 and Bglap) which are typically expressed at maximal levels around day 14 (Fig. 6). Calcium deposition was analyzed by Alizarin Red (see Figs. 4 & 5) in separate cultures because this process requires a much longer time (up to 32 days); we did not directly align our mRNA analyses with histochemical examination of calcium deposition. The expression of osteoblast markers during osteoblast differentiation follows a wellestablished temporal sequence of, first, expression of collagenous (Col1a1) and non-collagenous (Ibsp) genes and, second, of expression of mineralization specific genes (Bglap and Phospho1). Reduced Prmt1 expression significantly increases expression of Ibsp and Phospho1 at day 10 (p < 0.01), but does not appear to have statistically significant effects expression of Alpl or Bglap at any timepoint. Interestingly, Ibsp expression is reduced by Prmt1 loss at day 3 (Fig. 5). Hence, the effects of Prmt1 loss could be biphasic and/or time-dependent. From a temporal perspective, time course curves can be condensed or stretched to reflect acceleration or deceleration of osteoblast differentiation. From this temporal perspective, evaluation of the entire mRNA expression time course upon Prmt1 depletion shows that Ibsp expression is modestly delayed (at day 3; p < 0.001) while Phosho1 expression may be accelerated at day 10 (Fig. 6). Diminished expression of Prmt4/Carm1 correlates with significantly decreased expression of Ibsp and Phospho1 at day 6, as well as decreased expression of Alpl and Bglap at day 14 (p < 0.01). While these changes indicate that loss of Carm1/Prmt4 has effects on osteoblast-related gene expression (Fig. 6), the biological relevance of these expression changes is uncertain because loss of Prmt4/Carm1 does not affect osteoblast mineralization (Figs. 4 & 5). More interestingly, depletion of Prmt5 expression, which inhibits osteoblast mineralization ( The results show that MC3T3 cell mineralization is enhanced upon siPrmt1 treatment, remains the same as control upon siCarm1 treatment, and is decreased after siPrmt5 knock-down. (C) Quantification of mineralization in differentiating cells on day 30 using Image J software. Statistical significance was determined by two-way ANOVA (n = 3, mean + STD; * = p < 0.05 and **** = p < 0.0001). We note that for each of the siRNAs, we stained the cultures as pairs of nsRNA vs siRNA at different times when mineralization would be apparent in either the control or treatment group; hence, mineralization in controls differs for each of nsRNA/siRNA pairs (respectively, for PRMT1, CARM1 and PRMT5). The time course endpoint of Alizarin Red staining is determined by the appearance of visual evidence for nodular calcium deposition and varies between experiments (day 21 to day 32). at days 10 and 14 (p < 0.01), and Phospho1 at day 10, while there appears to be a time-course trend for decreased expression of Bglap at days 6, 10 and 15 (Fig. 6). Collectively, these gene expression data indicate that Prmt1, Prmt4/Carm1 and Prmt5 alter the expression of distinct osteoblast markers to different degrees and with distinct temporal dynamics.
Because PRMT1, PRMT4/CARM1 and PRMT5 proteins modulate distinct osteoblast markers at different stages of osteoblast differentiation, we assessed whether these epigenetic regulators may have indirect effects on osteoblastogenesis by controlling the mRNA expression of osteoblast related transcription factors (Fig. S5) or other epigenetic regulators (Fig. S6). The only remarkable finding of these RT-qPCR results is that Prmt4/Carm1 loss reduces expression of Runx2 mRNA (at day 6, p < 0.001), with a numerical trend of decreased expression for Runx2, Osterix/Sp7 and Dlx5 during the differentiation time-course (Fig. S5) that was not statistically validated. Aside from significant reduction of Sp7 in the siPrmt5 treatment group, expression analysis of these transcription factors upon depletion of Prmt1 or Prmt5 did not yield unambiguous results that permit a straightforward interpretation (Fig. S5C). Similarly, loss of Prmt1, Carm1 and Prmt5 mRNAs does not appear to modulate expression of epigenetic regulators known to contribute to skeletal development (e.g., Ezh2, Brd4) Dudakovic et al., 2018), except Brd2 which is decreased by Prmt5 at an early stage of differentiation (day 3; Fig. S6). We conclude that PRMT1, PRMT4/CARM1and PRMT5 proteins have functional roles that are not directly rate-limiting for the expression of key osteoblast related transcription factors or other epigenetic regulators.

Pharmacological inhibition with the Class I PRMT inhibitor enhances calcium deposition in the extracellular matrix of MC3T3 cells
To investigate whether the enzymatic activity of Class I PRMT enzymes (e.g., PRMT1, CARM1) affects osteoblastogenesis, we performed studies with the pharmacological inhibitor GSK715 (Fig. 7A). Treatment of MC3T3 osteoblasts with GSK715 reduces the nuclear levels of H3R17 asymmetric methylation (H3R17me2a), but not the levels of H4R3 asymmetric methylation (H4R3me2a) (Fig. 7B&C); the H3R17me2a versus H4R3me2a marks are generated by, respectively, PRMT4/ CARM1 versus PRMT1. As expected, GSK715 also does not affect the protein levels of either H3 or H4, or the two Class I proteins (PRMT1 and CARM1) (Fig. 7B&C). These results indicate that GSK715 primarily affects nuclear levels of the CARM1 dependent methylation of H3R17me2a in MC3T3 cells.
We tested the effects of GSK715 on MC3T3 differentiation using Alizarin Red staining (Fig. 8). MC3T3 cultures were treated with doses from 1 to 6 μM GSK715 that were either applied once (from day 0 to day 3, 1× dosing) or twice (from day 0 to day3 and day 3 to day 6; 2× dosing). In general, we find that a double dosing (i.e., drug administration repeated between day 3 and day 6) is more effective than single dosing (Fig. 8). Histochemical examination using Alizarin Red staining reveals that at least one of these regimens for GSK715 (6 μM administered twice) significantly enhances calcium deposition in MC3T3 osteoblast cultures as a proxy for osteoblast differentiation (Fig. 8). As the data overlays show, we observed a numerical trend of increased average values of Alizarin Red staining. However, this numerical trend was not matched by a statistical trend based on standard t-tests compared to controls. None of these observed changes, except for one, was statistically significant (p-values > 0.05) due to the large spread in values.
The densitometric values of Alizarin Red quantification for each of the dose response curves were also fitted to a polynomial curve (Fig. S7). The relative fit for the 2× dose response curve was better than the 1× dose response curve that have R 2 values 0.5589 and 0.2185, respectively (Fig. S7A&B). Interestingly, the fitted lines for the 2× curve (which has an acceptable R 2 value that clarifies >50 % of the observed variation in values) shows a consistent 3 to 6 fold increase in densitometric values for cells treated with concentrations of between 1.4 and 5.9 μM GSK175; for example, comparison of staining of cells treated with vehicle (0 μM; black dotted line) compared to a representative low dose of GSK175 (2.9 μM; red dashed line) reveals a 4.3 fold stimulation in Alizarin Red staining by GSK715 (Fig. S7B).
We note that GSK715 has biphasic effects on MC3T3 cells with apparent positive effects at low doses and negative effects of GSK715 at higher doses (Figs. 8 & S7). Negative effects of GSK715 at high micromolar concentrations are likely due to general cytotoxicity. In our prior experience with a large panel of EZH2 related epigenetic drugs (Galvan et al., 2021), we observed similar biphasic effects. Therefore, cytotoxic effects of high doses of a PRMT-related epigenetic drug on osteoblast differentiation are not unexpected.
Collectively, the findings of these pilot studies indicate that pharmacological inhibition of Class I PRMT enzymes that control nuclear H3R17me2 levels may increase calcium deposition during osteoblast differentiation. Comparison of findings obtained with GSK715 versus siRNA treatments for Prmt1, Prmt4 and Prmt5 indicates that none of these three epigenetic enzymes by themselves can account for the Western blot analysis shows that GSK715 reduces levels of H3R17me2a (mediated by Pmrt1) but not H4R3me2a (which is mediated by PRMT4/CARM1). Administration of GSK715 at two different doses (2 μM and 5 μ) has no biological effects on total H4 or H3 protein levels (representative gel of 3 biological replicates).
positive effect of the inhibitor on osteoblast mineralization. In conclusion, histone arginine methylation contributes to osteoblast differentiation and selective pharmacological inhibition of PRMTs may promote osteoblastogenesis.

Discussion
In this study, we examined the contribution of histone arginine methylation to osteoblast differentiation by surveying the expression and function of the full complement of human PRMT proteins. Our findings establish that PRMT1, PRMT4/CARM1 and PRMT5 mRNAs are selectively expressed in human and mouse musculoskeletal tissues. These three genes are generally most prominently expressed (i) across multiple human musculoskeletal tissues examined (i.e., bone, cartilage, muscle, growth plate and adipose tissue), (ii) in frontal and parietal bone of mouse calvaria, (iii) in human bone marrow derived MSCs and mouse MC3T3 osteoblasts, and (iv) in mesenchymal cell types present in early callus tissue during fracture repair. Loss of function studies using siRNA knockdown of Prmt1, Prmt4/Carm1 and Prmt5 in MC3T3 cells indicate that each of these Prmt members has distinct roles during osteoblast differentiation: Prmt1 loss enhances while Prmt5 loss inhibits calcium deposition, while Prmt4/Carm1 loss by itself is dispensable for calcium deposition in the extracellular matrix of differentiating osteoblasts. Gene expression analysis reveals that loss of Prmt1, Pmrt4/Carm1 and Prmt5 each modulates the expression of several but not all stage-specific mRNA biomarkers for osteoblast differentiation, although not proliferation. Finally, pharmacological inhibition with the Class I PRMT inhibitor GSK715 enhances calcium deposition in the extracellular matrix of MC3T3 cells. Collectively, these data establish that PRMT proteins have distinct effects on osteoblast differentiation.
Several technical limitations qualify our interpretations and conclusions. One major assay for the effects of PRMT loss of osteoblast differentiation is Alizarin Red staining which measures calcium deposition and not necessarily deposition of bone mineral (hydroxyapatite) in MC3T3 cells. While methods that are more definitive for hydroxyapatite (e.g., Von Kossa staining) could be applied, we consider that examination of definitive effects on bone mineralization will require in vivo studies that are beyond the scope of the present work. This study does not focus on recapitulating fidelity of bone-tissue like organization, but rather we use calcium deposition as a proxy for osteoblast differentiation to understand which PRMTs modulate osteoblastogenesis. Furthermore, our study includes only a single PRMT drug (GSK715) which exhibits non-specific negative effects at higher micromolar doses. Therefore, it will be important to survey additional PRMT drugs, similar to studies we performed to screen for EZH2 inhibitors (Galvan et al., 2021) before we can commit to an in-depth analysis of the transcriptome, epigenetic landscape and histone arginine methylome that are beyond the scope of the present work.
Our study shows that PRMT1, PRMT4/CARM1 and PRMT5 each have different functions in osteoblast maturation and/or expression of osteoblast related markers. Results obtained for PRMT5 are most easily interpretable. Loss of PRMT5 suppresses expression of multiple genes (i. e., Alpl, Ibsp, Phospho1) involved in mineralization and this finding corroborates reduced Alizarin Red staining upon siPRMT5 treatment. Our finding that PRMT1 inhibits osteoblast differentiation is in agreement with other studies indicating that PRMT1 promotes the undifferentiated state of cells (Goruppi et al., 2023). The results for PRMT4/ CARM1 do not permit a straightforward interpretation, because siRNA depletion of this protein has no appreciable effects on osteoblast mineralization by Alizarin red staining. Yet, treatment of MC3T3 cells using GSK715 obliterates H3R17me2 levels (a mark that is catalyzed by PRMT4/CARM1) and appears to stimulate calcium deposition as a proxy Alizarin red staining on day 28 of osteoblast differentiation shows that single (1×, day 0 to day 3) or dual administration (2×; from day 0 to day 3, and from day 3 to day 6) of 1 to 6 μM of GSK715 increases MC3T3 mineralization compared to control. (C) Quantification of mineralization in differentiating cells on day 28 using Image J software. Statistical significance was determined by twoway ANOVA (n = 3, mean + STD).
for osteoblast differentiation. The latter may indicate that either siCARM1 effects are compensated by other PRMT proteins or that another Class I H3R17me2 methyl transferase beyond PRMT4/CARM1 is targeted by GSK715. Regardless of these two possibilities, this study provides the first evidence of the importance of H3 and H4 arginine methylation during osteogenesis. Recent studies suggest that PRMT proteins also have molecular effects on the activity of the H3K27me3 transferase EZH2 that normally suppresses osteoblast differentiation (Dudakovic et al., 2016;Dudakovic et al., 2020). One solid study by the Nimer laboratory showed that PRMT5-mediated histone arginine methylation antagonizes and prevents the formation of H3K27me3 marks that are generated by EZH2 as part of the transcriptionally repressive polycomb complex PRC2 (Liu et al., 2020). Loss of PRMT5 results in a global rise of EZH2 dependent H3K27me3 marks because PRMT5 mediated methylation of H3 blocks subsequent methylation by PRC2 (Liu et al., 2020). Our finding that PRMT5 loss blocks osteoblast differentiation is consistent with this molecular mechanism. Loss of PRMT5 is predicted to prevent arginine methylation of H3 which is predicted to enhance H3K27me3 levels (Liu et al., 2020), and the latter has been associated with suppression of osteoblast differentiation. A second mechanism by which PRMT proteins can affect EZH2 activity is via direct methylation of EZH2 (at R342) by PRMT1, as has been suggested by another recent study (Li et al., 2020). This methylation step has been reported to stabilize EZH2 and increased EZH2 levels are predicted to block osteoblast differentiation (Li et al., 2020). Importantly, PRMT5 mediated increases in H3 arginine methylation and enhanced H3K27me3 levels are consistent with our observation that siRNA depletion of PRMT5 blocks osteogenesis. Similarly, the finding that PRMT1 depletion accelerates osteogenesis is in agreement with the model that loss of PRMT1 reduces EZH2 stability.
Our findings on the distinct functions of PRMT1, PRMT4/CARM1 and PRMT5 complement and extend previous studies that examined the contribution of PRMTs and their cognate epigenetic modifications in vitamin D3-dependent transcriptional regulation in rat osteosarcoma cells. One previous study showed that PRMT1 and PRMT4/CARM1 support the vitamin D3 (1,25(OH) 2 D 3 ) dependent upregulation of CYP24A1 gene expression (Moena et al., 2020b). Examination of histone marks revealed that H4R3me2a and H3R17me2a and their corresponding writers PRMT1 and PRMT4, respectively, are recruited by vitamin D receptor (VDR) and its coactivator SRC-1/NCOA1. Both PRMTs are released when the VDR/SRC-1 complex is stimulated by (1,25(OH) 2 D 3 ). Similarly, when the CYP24A1 gene promoter remains transcriptionally silent, the CYP24A1 locus is enriched with both H3R3me2s marks and the corresponding methyl transferase PRMT5 (Moena et al., 2020b). Furthermore, PRMT5 and the SWI/SNF protein complex are involved in the regulation of 1,25(OH) 2 D 3 metabolism, which suggests that this enzyme may play a role in the maintenance of calcium homeostasis by vitamin D (Seth-Vollenweider et al., 2014). Taken together, the results from the current study and other reports (Li et al., 2017;Blanc and Richard, 2017) indicate that PRMTs control osteoblast-related gene expression.
In conclusion, our study shows that PRMT1, PRMT4/CARM1 and PRMT5 proteins each have different effects on osteoblast mineralization and/or expression of osteoblast-related differentiation markers. The mechanisms by which PRMT1 and PRMT5 control osteoblast differentiation may involve effects on the histone arginine methylome, as well as direct and indirect effects on EZH2, which is a known suppressor of osteogenesis. One translationally relevant finding from our study is that the Class I PRMT inhibitor GSK715 enhances osteoblast differentiation. This finding predicts that GSK715 may potentially have bone anabolic effects in mouse models for low bone mineral density, a prediction that will be tested in future studies.

Funding
This publication was made possible through internal funding of Erasmus University Medical Center (to BCJvdE), the National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 AR069049 to AvW & R01AR039588 to GSS), as well as the CancerCare Manitoba Foundation (761020451 to JD).

Authorship contributions
PD performed the majority of the experiments, while EAL contributed additional supporting data, reagents and experimental ideas. PD, EAL, JARG, MAM, JPTMvL, GSS, BCJvdE, JRD & AJvW provided principal conceptual contributions pertinent to the conclusions of this work. PD prepared the first draft of this manuscript, while EAL, JARG, MAM, JPTMvL, GSS, BCJvdE, JRD & AJvW each provided critical input and suggested edits. AJvW obtained the main source of funding for this study, while EAL, GSS, BCJvdE and JRD provided auxiliary funding.

Declaration of competing interest
The authors have no conflicts of interest to declare. All co-authors have seen and agree with the contents of the manuscript and there is no financial interest to report.

Data availability
RNA-seq data are available on request and can be retrieved via links available in previous papers from our group.