**2. Results**

#### *2.1. The Recombinant p38*α *Protein Is Methylated by PRMT1 on R49 and R149*

PRMT1 promotes erythroid di fferentiation through enhancing the activation of p38 α [3]. To investigate whether the enhanced activation of p38 α is mediated by arginine methylation and the underlying mechanisms, we first performed in vitro methylation then identified methylation sites by LC-MS/MS analysis. Briefly, His- p38 α was incubated with or without recombinant GST-PRMT1 (glutathione S-transferase-PRMT1) in the presence of S-adenosyl-methionine (AdoMet) as a methyl donor. The reaction mixtures were fractionated by SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis), and gels containing p38 α proteins were excised and subjected to mass spectrometric analysis. A schematic representation of the procedure is shown in Figure 1A. The sequence coverage of p38 α from all identified peptides was above 70%. Peptides harboring dimethylated arginine residues on R49 and R149 in the presence of GST-PRMT1 were identified (Figure 1B) with Xcorr ≥ 2.0. These peptides were not identified when PRMT1 was absent in the methylation reactions. These results indicate that PRMT1 methylates p38 α on R49 and R149. The R49 and R149 residues are located in the N lobe in the ATP-binding cleft and in the C lobe near the catalytic loop, respectively, based on the structure information of the kinase domain [21] (Figure 1C).

Methylation of p38 α was further shown by using HA-PRMT1 (hemagglutinin-PRMT1) expressed in and immunoprecipitated from K562 cells. The His- p38 α protein was also readily methylated by HA-PRMT1 in vitro (Figure 1D, lanes 1 and 4). The methyltransferase-deficient PRMT1G80R greatly lost the ability to methylate p38 α, indicating that PRMT1 was indeed the enzyme in the immunoprecipitates responsible for the methylation of p38 α (Figure 1D, lanes 4 and 5). The immunoprecipitated HA-PRMT6, which also catalyzes the formation of asymmetric di-methylarginine and shares some substrate recognition sequences with PRMT1 [14], did not methylate p38 α. As reported [22], PRMT6 was readily automethylated (Supplementary Figure S1). The methyltransferase-deficient PRMT6KA mutant did not self-incorporate methyl groups (Supplementary Figure S1). These results further provide evidence for the specificity of p38 α methylation by HA-PRMT1. The methylation level of p38 α was significantly reduced when the arginine residues of 49 and 149 were mutated to lysine, which is not a substrate residue for PRMT1 (Figure 1E). The methyl incorporation into either R49K or R149K was reduced to around 60% compared to the wild-type p38 α, indicating that both sites were methyl acceptors. The methylation on R49 and R149 may be a co-dependent event, since simultaneous mutation on R49 and R149 did not completely diminish methylation levels (Figure 1E, R49/149K). We cannot rule out the possibility that there are more methylation sites on p38 α since mass spectrometric analysis does not guarantee a complete coverage of all peptides, including modified and non-modified. The structural integrity of these recombinant p38 α proteins was examined by their sensitivity to protease digestion. After incubation with trypsin, all four purified proteins, WT, R49K, R149K, and R49/149K, exhibited a similar digestion pattern upon fractionation by SDS-PAGE (Figure 1F). These results indicate that R49K and R149K mutations do not cause a notable structural collapse, particularly in the surface regions.

**Figure 1.** *Cont*.

**Figure 1.** Recombinant p38α protein is methylated by PRMT1 (protein arginine methyltransferase 1) on R49 and R149. The methylation of His- p38α proteins was performed in the presence or absence of recombinant GST-PRMT1 (glutathione S-transferase-PRMT1) as described in the Methods section. Peptides containing di-methyl arginine were identified by mass spectrometric analysis as illustrated (**A**). The sequencing results of peptides harboring di-methylated R49 and R149 are shown (**B**). The localization of Arg49 and Arg149 is marked (**C**). The methylation of His-p38α was further carried out using HA-PRMT1 (hemagglutinin-PRMT1) immunoprecipitated from K562 cells. The methylation of p38α was greatly diminished with the methyltransferase-deficient mutant HA-PRMT1G80R (**D**). When R49 and R149 of p38α were mutated to K49 and K149, the methyl incorporation was significantly reduced as compared to the wild type (WT) (**E**). The quantification in (**D**) and (**E**) was carried out with results from three separate assays. Asterisks indicate non-specific bands. The recombinant p38α WT and mutant proteins (5 μg) were incubated with trypsin at 37 ◦C for various times and fractionated by SDS-PAGE to reveal their sensitivity to trypsin digestion (**F**). \*\*\* *p* < 0.005 and \* *p* < 0.05 as compared with WT proteins.

#### *2.2. The Non-Methylation R49K and R149K Mutants of p38*α *Lose the Ability to Stimulate Erythroid Di*ff*erentiation*

To investigate whether R49 and R149 mediate the promotive effect of PRMT1, we first examined how the non-methylation mutations (Arg to either Lys or Ala) might influence differentiation. The ectopic expression of wild-type p38α stimulated erythroid differentiation from 40% to 60% in p38α-knockdown K562 cells, as measured by hemoglobin accumulation (Figure 2A, Vector vs. WT). Both the R49K and R149K single mutants lost the ability to promote differentiation and so did the R49/149K double mutant (Figure 2A). Similarly, the stimulatory effect on differentiation was completely diminished in R49A and R149A mutants (Figure 2B). These indicate a crucial role of R49 and R149 in stimulating erythroid differentiation. Erythroid differentiation was further examined by the expression of key genes. GATA1 (GATA Binding Protein 1) and EKLF (erythroid Kruppel-like factor) are transcription factors critical for differentiation toward erythroid lineage and PBGD (porphobilinogen deaminase) and ALAS2 (5'-aminolevulinate synthase 2) are enzymes involved in heme synthesis [3]. These transcripts were significantly up-regulated by AraC treatment (Figure 2C, 0 h vs. 96 h); however, the extents were significantly reduced when R49 and R149 were mutated (Figure 2C, R49/149K). The R136 of p38α was also identified as a methylated arginine in our study. However, the methylation event was PRMT1 independent (unpublished results). The R136K mutant of p38α still retained the stimulatory effect on erythroid differentiation (Figure 2D), which provided supports for a selective role of R49 and R149 in modulating differentiation.

**Figure 2.** Non-methylation mutants of p38α lose the ability to stimulate erythroid differentiation. The wild-type and R49/R149 mutant p38α were expressed as Flag-tagged proteins in p38α-knockdown K562 cells. Erythroid differentiation was induced with AraC (1-beta-arabinofuranosyl) (1 μM) for 96 hours and analyzed by benzidine staining for the production of hemoglobin (**A** and **B**). Mutations of R49 and R149 to either lysine (R49K, R149K and R49/149K) or alanine (R49A, R149A and R49/149A)

abolished the ability to promote differentiation (**A** and **B**). Erythroid differentiation was also analyzed by RT-qPCR for the expression of key genes. These transcripts were significantly up-regulated by AraC treatment; however, the extents were significantly reduced when R49 and R149 were mutated (**C**). The mutation of R136 to K136 (R136K) did not affect the ability to promote differentiation (**D**). All results shown are representatives of three independent experiments. Erythroid differentiation is presented as the mean ± S.E. of three repeats. \*\*\* *p* < 0.005. N.S. means no significance.

#### *2.3. The Promotive E*ff*ect of PRMT1 on Erythroid Di*ff*erentiation Is Mediated by the Methylation of p38*α *on R49 and R149*

We have demonstrated that the PRMT1 promotes erythroid differentiation in a p38α-dependent fashion in K562 cells and human primary CD34+ hematopoietic progenitors [3]. To show whether the kinase activity was required, we ectopically expressed p38α in a p38α-knockdown context. Erythroid differentiation was increased from 40% to 55% (Figure 3A, Vector vs. WT). The AGF p38α mutant is deficient of kinase activity due to mutations on the Thr-Gly-Tyr motif to Ala-Gly-Phe and cannot be activated by phosphorylation via up-stream MKKs [2]. This mutant completely lost the ability to promote differentiation (Figure 3A, Vector vs. AGF), indicating a requirement for the kinase activity to promote differentiation. The catalytic activity of PRMT1 is also required to promote differentiation since the methyltransferase-deficient G80R mutant could not promote [3], indicating that PRMT1 promotes differentiation through the arginine methylation of downstream effector substrates. Notably, PRMT1 promoted differentiation only in the presence of wild-type p38α but not the p38α AGF mutant (Figure 3A, PRMT1 + WT vs. PRMT1 + AGF), indicating that the kinase activity of p38α is essential to mediate the effect of PRMT1. Together, these results provide further evidence suggesting that PRMT1 modulates the kinase activity of p38α likely via arginine methylation.

PRMT1 could not promote differentiation in a p38α-knockdown context (Figure 3B, Vector vs. PRMT1 + Vector). The co-expression of wild-type p38α greatly stimulated differentiation from around 35% to around 60% (Figure 3B, PRMT1 + Vector vs. PRMT1 + WT); however, PRMT1 was unable to stimulate when the R49/R149 of p38α were mutated to K49/K149 (Figure 3B, R49/149K vs. PRMT1 + R49/149K), indicating that the R49 and R149 of p38α mediates the stimulatory effect of PRMT1. Since the methyltransferase activity of PRMT1 is required for the promotive effect of PRMT1 [3] and PRMT1 methylated p38α on R49 and R149 (Figure 1), these results together sugges<sup>t</sup> that R49 and R149 mediate the promotive effect of PRMT1 on erythroid differentiation through arginine methylation.

We further examined the phosphorylation and methylation states of p38α upon AraC stimulation. The Flag-p38α was expressed in PRMT1-knockdown cells with or without the overexpression of HA-PRMT1. Flag-p38α was immunoprecipitated from cells after AraC stimulation and examined by Western Blotting using anti-phospho-p38 or anti-methylarginine antibodies. Our results clearly show that p38α was methylated in cells and PRMT1 enhanced activation phosphorylation as well as the arginine methylation of p38α by around 1.6 folds (Figure 3C), supporting our notion that PRMT1 enhances activation of p38α by arginine methylation of the kinase. To examine whether phosphorylation was a precedent event for arginine methylation, we compared the arginine methylation status of WT and phosphorylation-deficient AGF mutant p38<sup>α</sup>. The Flag-p38α proteins were expressed in p38α-knockdown cells, immunoprecipitated after AraC stimulation and examined by Western Blotting. The WT p38α was readily phosphorylated while the AGF mutant, as expected, was not phosphorylated (Figure 3D). In spite of the dramatic difference in phosphorylation status, the arginine methylation levels were very similar in WT and AGF mutant proteins (Figure 3D), indicating that phosphorylation is not a prerequisite for arginine methylation by PRMT1.

**Figure 3.** Promotive effect of PRMT1 on erythroid differentiation is mediated by methylation on R49 and R149 of p38<sup>α</sup>. The wild-type and AGF (Ala-Gly-Phe) mutant Flag-p38α were expressed in p38α-knockdown cells. The wild-type p38α (WT) promoted differentiation but the phosphorylation activation-deficient AGF mutant was unable to (**A**). HA-PRMT1 could further promote only in the presence of wild-type p38α but not the p38α AGF mutant (**A**). In the same p38α KD (knockdown) context, PRMT1 was unable to promote differentiation when R49 and R149 were mutated to K49 and K149 (**B**). Flag-p38α was expressed in the presence or absence of HA-PRMT1. After AraC stimulation, p38α was immunoprecipitated and examined by Western Blotting using anti-phospho-p38 or anti-methyl arginine antibodies. PRMT1 significantly enhanced p38α phosphorylation and arginine methylation (**C**). Upon AraC stimulation, both the WT and AGF mutant were methylated to a similar extent, although AGF was deficient in phosphorylation (**D**). All results shown are representatives of three independent experiments. Erythroid differentiation is presented as the mean ± S.E. of three repeats. The quantification in (**C**) and (**D**) was carried out with results from three separate experiments. \*\*\* *p* < 0.005. N.S. means no significance.

#### *2.4. The R49 and R149 Residues Play a Crucial Role in the Activation of p38*α

Upon stimulation, p38 MAPK is activated by phosphorylation on the characteristic Thr-X-Tyr motif [2]. AraC treatment stimulated phosphorylation of the wild-type p38α (Figure 4A, WT); however, the activation of the R49/149K mutant was greatly reduced (Figure 4A, R49/149K), as examined by Western Blotting analysis using anti-phospho-p38 antibodies. We further immunoprecipitated Flag-p38α proteins and examined the activation phosphorylation. Similarly, R49/149K mutation significantly reduced AraC-induced activation (Figure 4B). These results indicate that R49/R149 have a role in modulating the activation of p38<sup>α</sup>. To further assess whether R49 and R149 modulate the activation of p38α in conditions other than AraC-induced differentiation, we stimulated K562 cells with sorbitol, which is a well-known strong stimulator of p38α in hyperosmotic stress [23]. The activation of p38α wild type was stimulated by around 4.0 folds at 0.5 h; however, it was reduced to around 2 folds in R49/149K mutant (Figure 4C). As an internal control, the activation of endogenous p38α was similar, around 3.5 folds, in cells expressing either Flag-p38α WT or Flag-R49/149K mutant (Figure 4C). To examine whether PRMT1 has a role in the activation of p38α induced by sorbitol treatment, we used K562 and PRMT1-overexpressing R2-1 cells as a comparison and found that the sorbitol-stimulated activation of p38 was significantly higher when PRMT1 was overexpressed (Figure 4D, K562 vs. R2-1). The basal level of p38α phosphorylation in R2-1 cells was higher than K562 parental cells before stimulation (Figure 4D, 0 h), which is in agreemen<sup>t</sup> with our previous observations [3]. This is conceivable because the ectopically expressed PRMT1 is active [3,4,20]. Together, these results indicate that PRMT1 likely also plays a role in enhancing the activation of p38α via methylation on R49 and R149 upon sorbitol stimulation.

**Figure 4.** R49 and R149 residues play a crucial role in the activation of p38<sup>α</sup>. The Flag-p38α wild type or R49/149K mutant were expressed in p38α-knockdown cells. Cells were stimulated with AraC (1 μM)

and the activation of p38α was examined directly by Western Blotting using specific anti-phospho-p38 antibodies (**A**) or after immunoprecipitation (**B**). Phosphorylation was greatly reduced in R49/149K. Alternatively, the cells were treated with sorbitol (150 mM) and the activation of p38α was examined by Western Blotting (**C**). The overexpression of PRMT1 (R2-1) enhanced the activation of p38α upon sorbitol stimulation (**D**). The levels of pp38 and p38 were quantified by Multi-Gauge V3.0 analysis. All results shown are representatives of three independent experiments. Statistical analysis was performed with results from three separate experiments. \*\*\* *p* < 0.005 and \*\* *p* < 0.01 as compared with WT.

#### *2.5. PRMT1 Acts Downstream of MKK3 to Promote Erythroid Di*ff*erentiation and the R49*/*149K Non-Methylation Mutant p38*α *Exhibits a Reduced Interaction with MKK3*

MKK3 and MKK6 both are upstream kinases, which phosphorylate and activate the p38 MAPK pathway [1]. To identify which MKK is responsible for activating p38α and promoting erythroid differentiation, we established MKK3 and MKK6 knockdown cell clones (Figure 5A,B, upper) and found that only MKK3 knockdown (MKK3 KD) greatly compromised erythroid differentiation (Figure 5A, lower), indicating that MKK3 is the major MAPK kinase in AraC-induced erythroid differentiation. On the contrary, MKK6 knockdown (MKK6 KD) exhibited a remarkable stimulation on erythroid differentiation (Figure 5B, lower), indicating a previously unknown role in negatively regulating erythroid differentiation. This result suggests that MKK6 is not a p38α activating MAPK kinase in erythroid differentiation. Furthermore, the AraC-induced activation of p38 was greatly reduced in MKK3 KD cell clones (Figure 5C), confirming that MKK3 plays a role in activating p38 during differentiation. This notion was further supported by the observation that overexpression of p38α in MKK3 KD cells partially rescued the differentiation but p38β, which is not involved in erythroid differentiation [3], could not (Figure 5D). Mutations of R49 and/or R149 lost the ability to compensate the inefficiency of MKK3 activity in MKK3 KD1 cells (Figure 5E), suggesting R49 and R149 are required to mediate the activation of p38α by MKK3.

**Figure 5.** *Cont*.

**Figure 5.** PRMT1 acts downstream of M KK3 (MAPK kinase 3) to promote erythroid differentiation, and the R49/149K non-methylation mutant exhibits a reduced interaction with MKK3. MKK3 and MKK6 were knocked down (**A**) and (**B**) as described in the Methods section. AraC-induced erythroid differentiation was significantly suppressed in MKK3-knockdown (KD1 and KD2) cells (**A**); however it was stimulated in MKK6-knockdown (KD1 and KD2) cells (**B**). The activation of p38 was significantly reduced in MKK3 KD1 and KD2 cells (**C**). Ectopic expression of p38<sup>α</sup>, but not p38β, in MKK3 KD1 cells could partially rescue differentiation (**D**). However, the R49K, R149K and R49/149K mutants lost the ability to promote differentiation (**E**). Ectopic expression of HA-PRMT1 promoted differentiation in parental K562, Luc and MKK3 KD1 and KD2 cells but had no effect in MKK6 KD1 and KD2 cells (**F**). Luc is the vector control cell. To examine the interaction of p38α with MKK3 and MKK6, Flag-p38α wild type and R49/149K mutant proteins were expressed in p38α KD cells. After cells were stimulated with AraC (1 μM) for 5 h, Flag-p38α proteins were immunoprecipitated and the protein levels of MKK3 (**G**) or MKK6 (**H**) in the immunoprecipitates were examined by Western Blot. The levels of MKK3, pp38 and p38 were quantified by Multi-Gauge V3.0 analysis. All results shown are representatives of three independent experiments. Statistical analysis was performed with results from three separate experiments. \*\*\* *p* < 0.005. \*\* *p* < 0.01.

Since PRMT1 enhanced the activation of p38α (Figure 3C), we next examined whether PRMT1 acts upstream or downstream of MKK3 to enhance the activation of p38α and to promote differentiation. We overexpressed PRMT1 in MKK3 KD cells and found that erythroid differentiation was greatly stimulated to around 65% as compared to 20–40% without PRMT1 overexpression (Figure 5F). These results indicate that higher PRMT1 levels can compensate the inefficiency of MKK3 functions and sugges<sup>t</sup> that PRMT1 acts downstream of MKK3. In agreemen<sup>t</sup> with the notion that MKK6 was not the upstream kinase for p38α (Figure 5B), PRMT1 did not further stimulate differentiation in MKK6 KD cells (Figure 5F). Taken together, these results indicate PRMT1 acts downstream of MKK3 and likely by directly methylating p38α on 49 and R149.

The MAPKs are known to form a protein complex with upstream kinases, scaffold proteins and phosphatases to bring the players to a close proximity, which confers temporal and spatial regulation of signal transduction [1]. When immunoprecipitated from cells, the p38α protein interacted with MKK3 only but not MKK6 (Figure 5G,H, WT), supporting our notions that MKK3, but not MKK6, is the major MAPK kinase for p38<sup>α</sup>. The non-methylation p38α R49/149K mutant exhibited a significantly reduced interaction with MKK3 by around 50% (Figure 5G, R49/149K). Similar to the wild type, the R49/149K mutant did not interact with MKK6 (Figure 5H, R49/149K). The expression levels of MKK3 and MKK6 were similar in the parental and R49/149K mutant cells (Figure 5G,H, input). Together with the observation that the activation of the R49/149K mutant was significantly lower than the wild type (Figure 4A–C), our results sugges<sup>t</sup> that R49 and R149 up-regulates the activation of p38α through enhancing the interaction of p38α with MKK3 via arginine methylation by PRMT1.

#### *2.6. Identification of MAPKAPK2 as a Downstream E*ff*ector of P38*α *During Erythroid Di*ff*erentiation and the R49*/*149K Non-Methylation Mutant Exhibits a Reduced Interaction with MAPKAPK2*

In order to have a comprehensive understanding of the biochemical and functional role of R49 and R149 methylation, we further analyzed the protein interactors of the wild-type and mutant p38α proteins. We co-expressed either the wild-type or the R49/149K mutant with HA-PRMT1 in p38α KD cells, performed immunoprecipitation, fractionated the immunoprecipitates by SDS-PAGE and carried out mass spectrometric analysis to reveal the interacting proteins. A total of 163 proteins were found to interact with both WT and mutant p38<sup>α</sup>. We then analyzed the connection of these proteins with the p38α pathway by using Ingenuity Pathway Analysis (IPA). MAPKAPK2 and MAPKAPK3 (MAPK-activated protein kinases 2 and 3) were identified by the software within the limit of "direct interaction" and "experimentally observed". MAPKAPK2 is a known substrate of p38α [24], whereas the role of MAPKAPK3 is less described. Since, to the best of our knowledge, there was no report for the role of MAPKAPK2 in erythroid differentiation, we knocked down MAPKAPK2 and examined AraC-induced erythroid differentiation. The results showed that the knockdown of MAPKAPK2 reduced erythroid differentiation, from 50% to 35–40% (Figure 6A,B), suggesting a role of MAPKAPK2 in erythroid differentiation. The ectopic expression of p38α still promoted differentiation in the MAPKAPK-2 KD cells, likely due to the remaining low level of MAPKAPK2. However, the differentiation was significantly lower in KD-1 cells (49%) than in the K562 parental cells (61%), where the MAPKAPK2 level was normal (Figure 6B). These results sugges<sup>t</sup> that p38α signals through MAPKAPK2 to promote differentiation. This notion was further supported by the immunoprecipitation experiments, showing the interaction of MAPKAPK2 with p38α (Figure 6C, WT). Notably, the interaction was significantly reduced to about 60% when R49 and R149 were mutated (Figure 6C, R49/149K). Although R49/149K mutation reduced the interaction of p38α with its upstream kinase MKK3 as well as with its downstream substrate MAPKAPK2, the interaction of p38α with protein phosphatase 2A (PP2A) was not significantly affected (Supplementary Figure S2). To examine the influence of arginine methylation on partner interaction, we immunoprecipitated Flag-p38α in the presence or absence of HA-PRMT1 upon AraC stimulation. The results show that PRMT1 enhanced the interaction of p38α with MKK3 and MAPKAPK2 by 1.7 folds and 1.2 folds, respectively (Figure 6D). Together, our results indicate that the R49 and R149 of p38α are critical for a previously unidentified role in selective partner interactions through arginine methylation by PRMT1.

**Figure 6.** MAPKAPK2 is a downstream effector of p38α and the R49/149K non-methylation mutation reduces the interaction of p38α with MAPKAPK2. The wild-type and R49/149K mutant p38α proteins were expressed in p38α KD cells. Cells were stimulated with AraC and Flag-p38α was immunoprecipitated using anti-Flag antibodies. The interacting proteins were analyzed by mass spectrometric analysis. MAPKAPK2 was identified as an interactor of the p38α by Ingenuity Pathway Analysis (IPA). AraC-induced erythroid differentiation was reduced in MAPKAPK2 knockdown (KD) cells (**A**). The ectopic expression of p38α rescued differentiation of MAPKAPK2 KD1 cells (**B**). The Flag-p38α WT and R49/149K proteins were immunoprecipitated after AraC stimulation. MAPKAPK2 interacted with wild-type p38α; however, the R49/149K mutant exhibited a significantly lower interaction with MAPKAPK2, as examined by Western Blotting (**C**). PRMT1 promoted the interaction of wild-type p38α with MKK3 and MAPKAPK2 (**D**). The intensity of protein bands in Western Blots were quantified by Multi-Gauge V3.0 analysis. All results shown are representatives of three independent experiments. Statistical analysis was performed with results from three separate experiments. \*\*\* *p* < 0.005.

This study reveals a novel regulatory mechanism for p38<sup>α</sup>. Arginine methylation of R49/R149 by PRMT1 occurs upon stimulation, which does not require a prior phosphorylation on Thr180 and Tyr182. The methylation of R49/R149 facilitates the selective association of p38α with MKK3 and thus augments phosphorylation by MKK3 and enhances the activation of p38<sup>α</sup>. The methylation of R49/R149 also facilitates the association of p38α with a downstream effector MAPKAPK2, which enhances the propagation of signals to up-regulate erythroid differentiation. An illustrated model is presented in Figure 7.

**Figure 7.** Illustration of the novel regulatory mechanism for p38α through arginine methylation on R49/R149 by PRMT1. AraC treatment stimulates the methyltransferase activity of PRMT1, which methylates p38α on R49 and R149. This facilitates the interaction of p38α and MKK3 and enhances the activation phosphorylation of p38α by MKK3. The interaction of p38α with downstream effector MAPKAPK2 is also increased upon R49/R149 methylation by PRMT1 methylation. Together, erythroid differentiation is promoted due to the facilitated p38α signaling. Green arrow: methylation. Yellow arrow: phosphorylation.
