The coenzyme S adenosylmethionine SAM binding
The coenzyme S-adenosylmethionine (SAM)-binding pocket of Set7 is connected to the histone-tail binding groove by a conserved lysine-channel, similarly observed in vSET (Figures 2D and S3). The Set7 SAM-binding pocket is negatively charged as observed in other known HMTase structures. However, the bound coenzyme has a distinct U-shaped conformation stabilized by hydrogen bonds, with Arg20 in the conserved GxG motif of the N-terminal region and Asn79 in the pseudo-knot NHS/CxxPN motif (Cheng et al., 2005, Qian and Zhou, 2006). π-π stacking between the conserved His80 and SAM ensures a firm binding into Set7. In the X-ray structure of vSET complexed with S-adenosyl-L-homocysteine and the substrate H3K27me1, the lysine-channel is formed by conserved hydrophobic residues Tyr50, Leu51, Phe52, and Ser53 of β6 and Tyr105 and Tyr109 at the C-terminal region (Qian et al., 2006). A similar hydrophobic lysine-channel is also observed in MLL3 (Li et al., 2016). In the Set7 ternary complex bound to SAM and H3 histone-tail peptide (amino acids 33–42), the hydrophobic wall of the lysine-channel is formed by Tyr59, Thr60, and Tyr61 of β5, Ala 71 of β6, and Ser76 of α3 (Figures 2D and S3). Set7 methylates H3K37, not the adjacent K36, which is mechanistically fascinating. The H3K37 substrate extends inside this channel toward the sulfur SB-3CT mg of SAM, while having the K36 extending ∼180° away, with its ɛ-ammonium side-chain group anchored in a negatively charged pocket (Figure 2D). The K37 substrate is stabilized in the hydrophobic lysine channel by a hydrogen bond with Val62 and a cation-π interaction with Tyr61. Overall, residues surrounding K37 are firmly stabilized, allowing K37 to extend inside the lysine channel for catalysis while leaving K36 firmly anchored away from the catalytic channel, providing a distinct structural basis for K37 specificity over K36. HMTases have been mostly identified as monomeric. Out of the 41 structurally characterized HMTases across species, only four are dimeric (EHMT1-GLP, EHMT2-G9a, Ga9/GLP, and vSET). Set7 was found to be homo-dimeric in its crystal structure, with an extensive network of interactions at the dimer interface, sharing a highly similar dimeric arrangement with PBCV-1 vSET (Figures 2A, 2B, and S3). In the PBCV-1 vSET (PDB: 3KMT) dimeric structure, only one of the two ligand binding sites can exist in an open state at one time, thus displaying negative cooperativity between the two active sites (Wei and Zhou, 2010). To confirm the hypothesis that the biological unit of Set7 is a dimer, a yeast two-hybrid experiment was performed. The expression of the reporter genes, ADE2 and HIS3, was induced only in AH109 cells co-expressing GAL4-activating domain-Set7 and GAL4-binding domain-Set7, resulting in cell growth on the agar medium lacking adenine (−Ade) or histidine (−His +3-AT) (Figure 2E), suggesting an in vivo Set7-Set7 interaction. The Set7-Set7 interaction was further confirmed in vivo by Ni-NTA pull-down assays using S. pombe whole-cell extracts. Set7-HA or Set7-cMyc was pulled down with Ni-NTA-bead-bound Set7-6xHis, but not with non-tagged Set7 (Figure 2F). Taken together, these results give compelling evidences that Set7 functions in a dimeric conformation within the cell, possibly with a similar mechanism as that of PBCV-1 vSET (Wei and Zhou, 2010). S. pombe gametogenesis shares similarities with mammalian spermatogenesis and is initiated under nitrogen starvation in the presence of two mating-type cells, h+ and h−, and generates four gametes (spores) through mating, meiosis, and sporulation (Figure 3A). These processes are spatiotemporally tightly regulated (Mata et al., 2002). Approximately 2,000 genes are upregulated and hundreds of genes are downregulated during gametogenesis in S. pombe, accounting for more than 50% of the genome (Mata et al., 2002). This drastic change in gene expression implies the involvement of epigenetic regulation during gametogenesis. We found that set7+ gene deletion affected the processes of gametogenesis (Figure 3). Gametogenesis was induced in two different auxotrophic WT (YK19 and YK65) and set7Δ (AS166 and AS41) strains in nitrogen-free liquid Edinburgh minimal medium and observed under a microscope after 48 h incubation. Remarkably, more than 50% of set7Δ cells failed to initiate gametogenesis. set7Δ asci produced the abnormal number of spores (1, 2, or 3) and showed lower efficiency of four-spore formation with 35.9% and 31.6% versus 58.8% and 69.3% in WT (Figures 3B and 3C). A majority of set7Δ spores had unclear spore walls that appeared uneven in thickness in contrast to WT round-shape spores with mature spore walls (Figure 3C). Furthermore, the methylation levels of all H3K37me1-3 remarkably increase in the course of gametogenesis (Figures 3D and 3E). These results suggest that dysregulated gene expression due to set7+ gene deletion results in defective gametogenesis. Interestingly, the level of control histone H3 consistently decreased overtime during the course of gametogenesis while the level of control β-actin remained constant (Figure 3D). The decrease of histone H3 levels is concomitant with increased levels of H3K37me on the remaining histone H3 during gametogenesis (Figure 3D). The roles of Set7, H3K37, and histone H3 levels during gametogenesis will be further explored in our next study.