Exploring potential roles for the interaction of MOM1 with SUMO and the SUMO E3 ligase-like protein PIAL2 in transcriptional silencing

Abstract

The CHD3-like chromatin remodeling protein MOM1 and the PIAS-type SUMO E3 ligase-like protein PIAL2 are known to interact with each other and mediate transcriptional silencing in Arabidopsis. However, it is poorly understood whether and how the interaction is involved in transcriptional silencing. Here, we demonstrate that, while the PIAL2 interaction domain (PIAL2-IND) is required for PIAL2 dimerization, MOM-PIAL2 interaction, and transcriptional silencing, a transgene fusing the wild-type MOM1 protein with the PIAL2 protein defective in PIAL2-IND can completely restore transcriptional silencing in the mom1/pial2 double mutant, demonstrating that the artificial fusion of MOM1 and PIAL2 mimics the in vivo interaction of these two proteins so that PIAL2-IND is no longer required for transcriptional silencing in the fusion protein. Further, our yeast two-hybrid assay identifies a previously unrecognized SUMO interaction motif (SIM) in the conserved MOM1 motif CMM3 and demonstrates that the SIM is responsible for the interaction of MOM1 with SUMO. Given that eukaryotic PIAS-type SUMO E3 ligases have a conserved role in chromatin regulation, the findings reported in this study may represent a conserved chromatin regulatory mechanism in higher eukaryotes.

Fig 1. The IND domain of PIAL2 is responsible for MOM1-PIAL2 interaction and PIAL2 dimerization in vivo.

(A) Schematic diagram of mutations in the IND domain of PIAL2. Black squares represent the conserved amino acids. PIAL2-IND-M represents the mutations in the IND domain (D182A/F183A/I185A, shown in red). (B) The interaction between MOM1 and PIAL2 or PIAL2-IND-M as determined by yeast two-hybrid assays. MOM1 was fused to GAL4-BD in the pGBKT7 vector; PIAL2 and PIAL2-IND-M were fused to GAL4-AD in the pGADT7 vector. “Vec” represents the empty GAL-BD or GAL4-AD vectors. (C) The IND mutations impair the interaction between PIAL2 and MOM1 as determined by affinity purification followed by mass spectrometric analysis. Protein extraction from transgenic plants separately harboring the wild-type PIAL2 and PIAL2-IND-M transgene were subjected to affinity purification. (D) The interaction of MOM1-Flag with the wild-type PIAL2-Myc and the PIAL2-IND-M-Myc as determined by co-IP. (E) The interaction of Flag-PIAL2 with the wild-type PIAL2-Myc and the PIAL2-IND-M-Myc as determined by co-IP.

Fig 2. The IND domain of PIAL2 is involved in transcriptional silencing through interaction with MOM1.

(A) Complementation of the silencing defect in the pial2 mutant by wild-type and mutated PIAL2 transgenes. The mutated PIAL2 transgene contains D182A/F183A/I185A mutations in the IND domain of PIAL2. The expression of the PIAL2 target loci was examined by RT-PCR. ACT7 was used as a control. (B) Schematic diagram of wild-type and mutated MOM1-PIAL2 fusion genes. The wild-type and mutated PIAL2 was fused to the 3’-terminal of the wild-type MOM1 driven by the MOM1 promoter through the BamHI restriction site. Both the fusion genes harbor a Flag tag in their 3’-terminals. (C) Complementation of the silencing defect in the mom1/pial2 double mutant by MOM1, MOM1-PIAL2, and MOM1-PIAL2-IND-M transgenes. The expression of solo LTR, SDC, and ROMANIAT5 was detected by qPCR. ACT7 served as an internal control. Error bars are standard deviation (SD). *P < 0.05 or **P < 0.01 was determined by Student’s t test.

Fig 3. Mutations in the CMM2 domain affect MOM1 dimerization and MOM1-PIAL2 interaction.

(A) The L1761D/L1765D mutations in the CMM2 domain disrupted the CMM2 dimerization as determined by yeast two-hybrid assays. The wild-type CMM2 was fused to GAL4-BD; the wild-type and the mutated CMM2 were fused to GAL4-AD. “Vec” represents the empty GAL-BD or GAL4-AD vectors. (B) The full-length MOM1 harboring the CMM2 mutations still interacts with PIAL1 and PIAL2 in yeast. The wild-type and mutated MOM1 were fused to GAL4-BD; PIAL1 and PIAL2 were fused to GAL4-AD. “Vec” represents the empty GAL-BD or GAL4-AD vectors. (C) The mutations in the CMM2 domain disrupt MOM1 dimerization in vivo as determined by co-IP. The wild-type and mutated MOM1-Flag were introduced into transgenic plants harboring the MOM1-Myc transgene by crossing. F1 generation plants were subjected to co-IP. (D) The mutations in the CMM2 domain partially impair the interaction of MOM1 with PIAL2 in vivo as determined by co-IP. The wild-type and mutated MOM1-Flag were separately introduced into transgenic plants harboring PIAL2-Myc transgene by crossing. (E) The CMM2 mutations in the CMM2 domain of MOM1 are required for MOM1 complex formation in vivo as determined by gel filtration assays. Protein exaction from transgenic plants harboring the wild-type MOM1-Flag and the mutated MOM1-CMM2-M-Flag in the mom1 background was eluted on Superose 6 (10/300 GL) column. The fractions were subjected to western blotting. The arrows mean the fractions corresponding to the standard proteins of 67, 220, 669 kDa.

Fig 4. Mutations in the CMM2 domain of MOM1 affect transcriptional silencing.

(A) Silencing of indicated loci was largely restored by the wild-type MOM1-Flag transgene but was only slightly restored by the MOM1-Flag transgene harboring the mutations in the CMM2 domain. Class I loci are only regulated by MOM1, including ROMANIAT5 and TSI; Class II loci are co-regulated by MOM1 and RdDM pathway, including solo LTR and SDC. ACT7 was used as an internal control. Error bars are SD. *P < 0.05 or **P < 0.01 was determined by Student’s t test. (B) Expression levels of the wild-type and mutated MOM1 transgenes in mom1 mutant as determined by western blotting. The transgenic lines were used to determine the effect of the CMM2 mutations on transcriptional silencing. Rubisco stained by Ponceau S was shown as a loading control.

Fig 5. The CMM3 domain of MOM1 is responsible for SUMO1 and SUMO2 interaction as determined by yeast two-hybrid assays.

(A) Schematic diagram of full-length and various truncated versions of MOM1 used in yeast two-hybrid assays. P1, 1–832 aa; P2, 798–2001 aa; P3, 1660–2001 aa; P4, 1660–1860 aa. (B) Interaction of full-length and truncated versions of MOM1 with SUMO1, SUMO2 as determined by yeast two-hybrid assays. Full-length and truncated versions of MOM1 were fused to GAL4-AD, and SUMO1 and SUMO2 was fused to GAL4-BD. “Vec” represents the empty GAL-BD or GAL4-AD vectors. (C) Alignment of the CMM3 domains of Arabidopsis MOM1 and its orthologues from other plants, including Brassica rapa, Populus euphratica and Vitis vinifera. The amino acids highlighted by black and gray backgrounds indicated that the amino acids are completely and partially conserved, respectively. CMM3-M1 (V1994A/V1995A, shown in red) and CMM3-M2 (V1994A/V1995A/ C1996A/L1997A/S1998A, shown in red) represent two versions of mutations in the CMM3 domain of MOM1. (D) Mutations in the CMM3 domain disrupt the interaction between MOM1-P3 and SUMO1 or SUMO2 as determined by yeast two-hybrid assays. The wild-type and mutated MOM1-P3 were fused with GAL4-AD. SUMO1, SUMO2, and CMM2 were fused with GAL4-BD. “Vec” represents the empty GAL-BD or GAL4-AD vectors.

Fig 6. The interaction of MOM1 with SUMO proteins is dispensable for transcriptional silencing.

(A) The SIM in the CMM3 domain of MOM1 is not required for transcriptional silencing. As determined by qPCR, the silencing of solo LTR, TSI, and ROMANIAT5 was restored by the MOM1 transgene harboring the CMM3 mutations as well as by the wild-type transgene. ACT7 served as an internal control. Error bars are SD. *P < 0.05 or **P < 0.01 was determined by Student’s t test. (B) SUMO2 was not co-purified with MOM1 as determined by co-IP. MOM1-Myc transgene was introduced into transgenic plants harboring Flag-SUMO2 transgene by crossing. F1 generation plants were subjected to co-IP.