Transcriptional gene silencing by Arabidopsis microrchidia homologues involves the formation of heteromers

Significance

Members of the Arabidopsis microrchidia (AtMORC) ATPase family are involved in gene silencing and heterochromatin condensation without altering genome-wide DNA methylation patterns. Here, we examine the functional relationship between several family members and show that AtMORC6 interacts in two mutually exclusive protein complexes with AtMORC1 and its closest homologue, AtMORC2. Consistently, RNA sequencing of high-order mutants indicates that AtMORC1 and AtMORC2 act redundantly in gene silencing. We also examine the genetic interactions between AtMORC6 and the transcriptional repressor Morpheus Molecule 1 (MOM1). We observe a synergistic transcriptional regulation in the mom1/atmorc6 double mutant, indicating that these epigenetic regulators act mainly in different silencing pathways, both independently of DNA methylation.

Abstract

Epigenetic gene silencing is of central importance to maintain genome integrity and is mediated by an elaborate interplay between DNA methylation, histone posttranslational modifications, and chromatin remodeling complexes. DNA methylation and repressive histone marks usually correlate with transcriptionally silent heterochromatin, however there are exceptions to this relationship. In Arabidopsis, mutation of Morpheus Molecule 1 (MOM1) causes transcriptional derepression of heterochromatin independently of changes in DNA methylation. More recently, two Arabidopsis homologues of mouse microrchidia (MORC) genes have also been implicated in gene silencing and heterochromatin condensation without altering genome-wide DNA methylation patterns. In this study, we show that Arabidopsis microrchidia (AtMORC6) physically interacts with AtMORC1 and with its close homologue, AtMORC2, in two mutually exclusive protein complexes. RNA-sequencing analyses of high-order mutants indicate that AtMORC1 and AtMORC2 act redundantly to repress a common set of loci. We also examined genetic interactions between AtMORC6 and MOM1 pathways. Although AtMORC6 and MOM1 control the silencing of a very similar set of genomic loci, we observed synergistic transcriptional regulation in the mom1/atmorc6 double mutant, suggesting that these epigenetic regulators act mainly by different silencing mechanisms.

DNA methylation and histone posttranslational modifications are essential for silencing of transposable elements (TEs) and other repeat sequences. In the plant model organism Arabidopsis thaliana, DNA methylation sites are found in three different cytosine contexts: CG, CHG, and CHH (in which H is A, T, or C) (1). Symmetric CG and CHG methylations are mediated by DNA Methyltransferase 1 (MET1) and Chromomethylase 3 (CMT3), respectively (2, 3). Asymmetric CHH methylation is maintained at nonoverlapping sites by CMT2 and Domains Rearranged Methyltransferase 2 (DRM2) (4, 5). In the RNA-directed DNA methylation (RdDM) pathway, de novo methylation of CHH sites is established by DRM2 and involves 24-nucleotide small interfering RNAs and long noncoding RNAs (6–11). Genome-wide studies revealed that DNA methylation and repressive histone modifications such as dimethylation of histone 3 lysine 9 (H3K9me2) correlate with transcriptionally silent chromatin (12–16). Furthermore, transcriptional derepression of silenced methylated loci is accompanied by loss of DNA methylation. A prominent exception to this interdependence is the Morpheus Molecule 1 (MOM1).

MOM1 is unique to the plant kingdom and was identified in a random transfer-DNA (T-DNA) insertion screen reporting the derepression of a silenced transgene (17). The mom1 mutant shows a loss of transcriptional gene silencing at loci located predominantly in the pericentromeric regions of the chromosomes (18). Interestingly, these transcriptional gene-silencing defects occur without major changes in DNA methylation or histone marks (17–21). RNA Polymerase IV and V (PolIV and PolV), which are key components of the RdDM pathway, were identified as enhancers of the mom1 phenotype (18). To date, the extent to which MOM1 is implicated in RdDM as well as its molecular mechanism of action remain poorly understood. Because MOM1 shows partial sequence similarities to chromodomain–helicase–DNA binding proteins, it has been proposed that MOM1 is involved in heterochromatin compaction (17, 22). However, the mom1 mutant does not show any heterochromatin decondensation (20, 23).

Recently, members of the Arabidopsis microrchidia (AtMORC) ATPase family have also been shown to be involved in transposon repression and gene silencing (24–26). The MORC1 gene was originally described in mice, where it was found to be essential for male primordial germ cell development (27, 28). The Arabidopsis genome contains seven MORC homologs, which were termed AtMORC1 [NP_568000; AT4G36290; Compromized Recognition of Turnip Crinkle Virus 1 (CRT1)], AtMORC2 [NP_195351; AT4G36280; CRT1–Homolog 1 (CRH1)], AtMORC3 (NP_195350; AT4G36270; CRH2), AtMORC4 (NP_199891; AT5G50780; CRH4), AtMORC5 (NP_196817; AT5G13130; CRH5), AtMORC6 [NP_173344; AT1G19100; CRH6; Defective in Meristem Silencing 11 (DMS11)], and AtMORC7 (NP_194227; AT4G24970; CRH3) (25, 29–32). AtMORC1 and AtMORC2 are the most closely related homologs and share 80.9% amino acid sequence identity (29–32) (Fig. S1A). AtMORC6 has been identified in four independent forward genetic screens (24–26, 31) as required for gene silencing and maintenance of heterochromatin integrity. AtMORC1 is also required for gene silencing (26), although it was originally described as a master regulator in plant disease resistance signaling (30–33).

Currently, the molecular mechanisms by which the different AtMORC homologs achieve gene silencing remain to be elucidated. AtMORC proteins carry a gyrase, Hsp90, histidine kinase, and MutL (GHKL) domain together with an S5 domain that constitute an active adenosine triphosphatase (ATPase) module (27, 31, 34). They also carry a putative C-terminal coiled-coil domain (27). In vitro assays showed that both AtMORC1 and AtMORC6 are bona fide ATPases (26, 31). A modest reduction of DNA methylation and repressive histone marks at specific RdDM target sites in atmorc6 mutant suggested that AtMORC6 could also play a role in RdDM (24, 25). However, whole genome sequencing analyses of DNA methylation and H3K9me2 in atmorc1 and atmorc6 did not reveal significant differences compared with the wild-type level either in the genome at large or at sites of the highest level of gene derepression in atmorc mutants (26). Therefore, it is unlikely that the predominant function of AtMORC proteins is maintenance of DNA methylation and H3K9me2, although some interaction with the RdDM pathway seems likely.

In this study, we describe the physical interactions between three different AtMORC homologs and their functional implication in gene silencing. Biochemical analyses indicate that AtMORC6 forms mutually exclusive heteromers with AtMORC1 and its close homolog, AtMORC2. RNA-sequencing (RNA-seq) analyses of high-order mutants show that AtMORC1 and AtMORC2 act redundantly to repress a set of TEs similar to AtMORC6. Furthermore, we also examined the relationship between AtMORC6- and MOM1-mediated silencing as both pathways have only minor impacts on genome-wide DNA methylation. Interestingly, we observed a synergistic effect on transposon derepression, suggesting that these epigenetic regulators act by independent silencing mechanisms.

Results and Discussion

AtMORC6 Interacts in Vivo with AtMORC1 and AtMORC2 to Form Distinct Heteromers.

Previous analyses showed similar transcriptional derepression between the single atmorc6 single mutant and the atmorc1/atmorc6 double mutant, suggesting that AtMORC1 and AtMORC6 could interact to enforce gene silencing (26). To test this hypothesis, FLAG epitope-tagged AtMORC1 and AtMORC6 under their respective endogenous promoters were introduced into cmt3/atmorc1-3 and atmorc6-1 lines, respectively. Western blotting analyses confirmed that both AtMORC1-FLAG and AtMORC6-FLAG were expressed in their respective mutant background and could complement the suppressor of drm2 cmt3 (SDC)::GFP silencing defects (Fig. S1B). These lines were subsequently used to immunoprecipitate FLAG-tagged AtMORC proteins from leaf tissue, and mass spectrometry (MS) analyses were performed to determine potential interacting proteins. MS analyses indicated that AtMORC1 was strongly immunoprecipitated with AtMORC6-FLAG and vice versa (Table 1). This interaction was validated by coimmunoprecipitation (co-IP) using F₁ transgenic plant lines expressing complementing AtMORC1-myelocytomatosis (MYC) (26) and AtMORC6-FLAG (Fig. 1A).

Table 1.

FLAG-tagged AtMORC proteins were immunoprecipitated and interacting proteins were analyzed by MS

AtMORC6-FLAG IP
Name	Accession	Spectra		NSAF		% AtMORC6
AtMORC6	AT1G19100	77	75	2,060	539	100	100
AtMORC1	AT4G36290	62	31	1,732	233	84	43
AtMORC2	AT4G36280	35	20	992	152	48	28
AtMORC1-FLAG IP
Name	Accession	Spectra		NSAF		% AtMORC1
AtMORC1	AT4G36290	76	71	6,273	765	100	100
AtMORC6	AT1G19100	11	42	870	434	14	57
FLAG-AtMORC2 IP
Name	Accession	Spectra		NSAF		% AtMORC2
AtMORC2	AT4G36280	65	—	370	—	100	—
AtMORC6	AT1G19100	32	—	172	—	47	—

The total numbers of identified spectra, the normalized spectral abundance factor (NSAF), and the percentage relative to the bait protein are given for two biological replicates.

Fig. 1.

Redundancy of AtMORC1 and AtMORC2 in transposon silencing. (A) AtMORC1 physically interacts with AtMORC6. AtMORC6-FLAG was coimmunoprecipitated with AtMORC1-MYC in F₁ plants expressing both epitope-tagged proteins. Epitope-tagged proteins were detected by Western blotting. (B) RT-PCR assessing endogenous expression of SDC, AtCopia28, and RomaniaT5. Three biological replicates were performed for each tested genotype. Two individual alleles were used for atmorc1 and atmorc2. (C and D) Venn diagrams of overlap between TEs up-regulated (fourfold increase; FDR, 0.05; Fisher’s exact test) in each genotype. Gray regions represent categories with no TEs counted. Blue shading represents the union set of TEs up-regulated in atmorc mutants. (E) Boxplot and (F) heatmap of average reads per kilo base per million (RPKM) values between two biological replicates for TEs in a union set for different genotypes. An asterisk indicates a significant increase relative to wild-type samples (P < 1e-3, Mann–Whitney U test).

To further characterize the interaction between AtMORC1 and AtMORC6, we performed gel filtration experiments. Leaf protein extracts from epitope-tagged lines were separated on a Superdex 200 10/300GL column, and the eluted fractions were probed by immunoblotting. We observed that both AtMORC1-FLAG and AtMORC6-FLAG were predominantly eluting around 200–300 KDa, suggesting that AtMORC proteins are primarily existing in vivo as dimers (Fig. S2). Together with the co-IP experiments, these results indicate that AtMORC1 and AtMORC6 are primarily found in vivo as heteromers, most likely as heterodimers. Nevertheless, it cannot be completely ruled out that AtMORC proteins might also form heterotetramers or higher molecular weight complexes, as we observed some signal in fractions with predicted sizes up to several hundred kilodaltons.

MS analysis of FLAG-tagged AtMORC6 IPs revealed an additional interaction with the closest homolog of AtMORC1, AtMORC2 (Table 1). This result is consistent with a recent independent study that also found peptides of AtMORC1 and AtMORC2 in an IP–MS of AtMORC6 in flowers (35). Interestingly, AtMORC2 was not immunoprecipitated with AtMORC1, suggesting that AtMORC6 was interacting with AtMORC1 and AtMORC2 in two distinct complexes (Table 1). To validate the heteromerization between AtMORC6 and AtMORC2, we engineered a complementing transgenic line expressing FLAG-tagged AtMORC2 in an atmorc1/atmorc2 background (Fig. S1 C and D) and performed IP followed by MS. MS analysis showed that AtMORC6 was immunoprecipitated with FLAG-AtMORC2 (Table 1). Consistent with this interaction, gel filtration analysis of FLAG-AtMORC2 leaf extracts showed that FLAG-AtMORC2 was principally present in the elution fractions around 200–300 KDa, corresponding to similar elution fractions as AtMORC6-FLAG (Fig. S2). In summary, our biochemical analyses indicate that AtMORC6 physically interacts with AtMORC1 and AtMORC2 in the form of two mutually exclusive heteromers.

AtMORC6 was shown to interact in vitro with DMS3 when both proteins were coexpressed in Escherichia coli, providing a physical link to the RdDM pathway (25). DMS3 is a structural maintenance of chromosomes hinge domain-containing protein that lacks an ATPase domain (36). Based on the stimulation of AtMORC6 ATPase activity by in vitro interaction with DMS3, it was proposed that AtMORC6 and DMS3 cooperate to promote transcriptional repression. DMS3 has also been shown to interact with additional components of the DRD1-DMS3-RDM1 (DDR) complex including Defective in RNA-Directed DNA Methylation 1 (DRD1) or RDM1 as well as with the largest subunit of PolV (37). Furthermore, genome-wide association of PolV to chromatin and thus the production of PolV-dependent transcripts and subsequent DNA methylation are dependent on all members of the DDR complex (37, 38). However, we did not detect DMS3 or other components of the DDR complex in our IP–MS experiments. Also, previous IP–MS experiments using FLAG-tagged DRD1 and DMS3 proteins as bait did not immunoprecipitate AtMORC6 (37). Nevertheless, we cannot rule out that the interactions between components of the DDR complex and AtMORC6 are weak or ephemeral and could not be detected under our IP conditions.

A recent study found that AtMORC6 was immunoprecipitated in flowers in very small amounts with SUVH9, an SRA- (SET [suppressor of variegation 3–9 [Su(var)3–9], enhancer of zeste [E(z)], and trithorax (Trx)] and RING [really interesting new gene] associated)- and SET-domain-containing protein (35). SUVH9 and its closest homolog, SUVH2, were shown to bind methylated DNA and recruit PolV to chromatin through an interaction with the DDR complex (11, 35, 39). Yeast two-hybrid assays further indicated that the interactions between AtMORC proteins and SUVH proteins were direct (35). These data, together with the slight changes observed in DNA methylation of certain RdDM target loci (24, 25, 40), suggest that AtMORC proteins modulate RdDM through interactions with the DDR complex and SUVH proteins. Nevertheless, the mild changes of small RNAs and DNA methylation genome-wide in atmorc mutants (26) suggest that AtMORCs are unlikely to be canonical RdDM factors. It is also plausible that AtMORCs contribute to processing of target loci transcripts, thus leading to posttranslational silencing. Future experiments are needed to clarify the precise function in gene silencing and degree of involvement of AtMORCs in the RdDM pathway.

AtMORC2 Acts Redundantly with AtMORC1 to Achieve Gene Silencing.

To further study the role of AtMORC2 in gene silencing and its functional relationship with AtMORC1 and AtMORC6, we generated high-order mutants and performed transcriptional profiling analyses. Real-time PCR (RT-PCR) from RNA extracted from leaf tissue indicated that SDC was derepressed in atmorc1 but not atmorc2 (Fig. 1B), consistent with the fact that AtMORC2 was not identified in the genetic screens that identified AtMORC1 and AtMORC6 (24–26, 31). RT-PCR also showed an increased derepression of two transposons, AtCopia28 and RomaniaT5, in the atmorc1/atmorc2 double mutant compared with atmorc1 and atmorc2 single mutants (Fig. 1B), indicating that AtMORC1 and AtMORC2 act redundantly in transposon silencing. Further genome-wide characterization of the transcriptome by RNA-seq indicated that only two transposons was significantly up-regulated in atmorc2 compared with wild type [using a very stringent cutoff of fold change ≥4; false discovery rate (FDR) < 0.05], whereas nine TEs were up-regulated in atmorc1 (Fig. 1C). Transcriptional derepression of protein-coding genes was also more pronounced in atmorc1 compared with atmorc2 (Fig. 2A). Publicly available microarray data indicate that expression of AtMORC1 is higher than AtMORC2 in most tissues and developmental stages (Fig. S3A), providing a plausible explanation for the stronger silencing defects observed in atmorc1 compared with atmorc2. Interestingly, combined deletion of AtMORC1 and AtMORC2 led to significantly higher transcription of TEs and protein-coding genes compared with both single mutants (Fig. 1 C, E, and F and Fig. 2 A, C, and D), confirming that AtMORC1 and AtMORC2 are functionally redundant. In addition, the overexpression of FLAG-AtMORC2 succeeded in complementing transcriptional derepression in the atmorc1/atmorc2 double mutant (Fig. S1D).

Fig. 2.

Redundancy of AtMORC1 and AtMORC2 in gene silencing. (A and B) Venn diagrams showing relationships between sets of protein-coding genes called up-regulated (fourfold increase in expression; FDR < 0.05) for different genotypes. Gray regions represent categories with no gene counted. Blue shading represents the union set of genes up-regulated in atmorc mutants. (C) Boxplot and (D) heatmap of average RPKM values for different genotypes (two biological replicates) for protein-coding genes in a union set for different genotypes. An asterisk indicates a significant increase relative to wild-type samples (P < 1e-8, Mann–Whitney U test). Two asterisks represent a significant increase relative to wild-type samples and the atmorc1 single mutant (P < 1e-2, Mann–Whintey U test). (E) Overrepresentation in H3K9me2-enriched heterochromatin of protein-coding genes significantly up-regulated in atmorc1-2/atmorc2-1, atmorc6-3, or atmorc1-2/atmorc2-1/atmorc6-3 mutants. An asterisk indicates a significant increase relative to all protein-coding genes (P < 1e-3, Fisher’s exact test). (F) Metagene analysis of RNA-seq reads over protein-coding genes called up-regulated in atmorc1-2/atmorc2-1, atmorc6-3, or atmorc1-2/atmorc2-1/atmorc6-3 mutants. Reads are derived from previously published RNA-seq libraries for two replicates of the drm1/drm2 double mutant and the corresponding wild type (WT).

The observed redundancy between AtMORC1 and AtMORC2 and their physical interaction with AtMORC6 in two mutually exclusive heteromers predict that a loss of AtMORC6 should be phenotypically comparable to the combined loss of AtMORC1 and AtMORC2. To test this hypothesis, we compared the transcriptomes of atmorc1/atmorc2 with the atmorc6 single mutant. RNA-seq revealed a high overlap of transcriptional derepression between atmorc1/atmorc2 and atmorc6 (Fig. 1 D–F and Fig. 2 B–D), supporting the notion that AtMORC6 function is epistatic to both AtMORC1 and AtMORC2 combined. Derepressed transposons were not restricted to a specific family in any of the mutant backgrounds analyzed (Fig. S3B). Finally, the observed transcriptional derepression did not significantly increase in a triple mutant lacking AtMORC1, AtMORC2, and AtMORC6 (Fig. 1 D–F and Fig. 2 B–D). These results are consistent with the model that AtMORC6 interacts exclusively with either AtMORC1 or AtMORC2 to achieve gene silencing and that AtMORC1 is functionally redundant with AtMORC2.

It appeared that up-regulated genes were preferentially localized in H3K9me2-enriched heterochromatin (12) even though they are protein-coding (Fig. 2E). This is in agreement with the previous observations that AtMORC1 and AtMORC6 are mainly involved in silencing and compaction of heterochromatin (26). Gene ontology term analysis using AmiGO (41) of all up-regulated protein-coding genes indicated enrichments (P < 6e-4) in response to chitin and in response to organonitrogen compounds in atmorc1/atmorc2 and in atmorc1/atmorc2/atmorc6. It is interesting to note that chitin has been recognized as a general elicitor of plant defense responses (42), which is in agreement with the reported implication of AtMORC1 in plant immunity (31). To assess if protein-coding genes up-regulated in atmorc6 were also targets of the RdDM machinery, we looked at their expression in a mutant lacking the methyltransferases DRM1 and DRM2 that is thus defective in RdDM (4). These were not significantly up-regulated in drm1/drm2 (Fig. 2F), indicating that AtMORCs are unlikely to be canonical RdDM factors.

Our combined genetics and RNA-seq data show that the simultaneous absence of AtMORC1 and AtMORC2 in atmorc1/atmorc2 cannot be functionally compensated by the presence of AtMORC6 alone (Figs. 1 and 2). Also, the loss of AtMORC6 in atmorc6 cannot be compensated by the presence of AtMORC1 and AtMORC2 (Figs. 1 and 2). Furthermore, the atmorc1/atmorc2/atmorc6 triple mutant does not have a stronger phenotype than the atmorc1/atmorc2 double mutant (Fig. 1 B and D–F and Fig. 2 B–D). Together with the observation that AtMORC1 and AtMORC2 did not interact, these results lead to the conclusion that AtMORCs function as heteromers and not as homomers.

AtMORC6 and MOM1 Act Synergistically to Silence a Common Set of Transposons.

AtMORC1 and AtMORC6 were identified in a forward genetic screen reporting the derepression of an SDC::GFP transgene in wild type or in the cmt3 mutant background (26). Further screening of ethyl methanesulfonate (EMS) mutagenized seeds followed by deep genome resequencing identified two new alleles of AtMORC6 in the cmt3 background. In the first line, cmt3 262, glycine 212 was mutated to glutamic acid, and in cmt3 379, a guanine (chr1:6599258) was mutated to adenine in the splice site before exon 14. Interestingly, we also identified three loss-of-function alleles of the MOM1 gene in the same genetic screen. The EMS mutations in these new mom1 alleles were a stop codon introduced at amino acid 603 (line 337 in a wild-type background), a stop codon introduced at amino acid 586 (cmt3 265), and a substitution of Leucine 656 to Phenylalanine (cmt3 113).

MOM1 is unique to the plant kingdom and has no homologs in the Arabidopsis genome. Previous studies showed that DNA methylation in mom1 mutants was similar to the wild-type level (17–19, 21). This observation was recently confirmed by genome-wide bisulfite-sequencing (BS-seq) analyses (43). RNA-seq analyses showed that 52 TEs were significantly up-regulated in mom1 using similarly stringent cutoffs as for atmorc mutants (Fig. 3A), and we found that the DNA methylation levels of these TEs also remained unchanged in mom1 compared with wild type (Fig. 3D). Nineteen transposons were significantly derepressed in atmorc6 in this experiment, and most of these were also derepressed in mom1 (Fig. 3A). The numbers of TEs significantly up-regulated in atmorc6 slightly vary between the two RNA-seq experiments performed (Figs. 1D and 3A) because both experiments were done independently. As shown previously, DNA methylation was not significantly changed in TEs up-regulated in atmorc6 (26) (Fig. 3D). These data indicate that overall transcriptional derepression is higher in mom1 compared with atmorc6 and that MOM1 and AtMORC6 mediate the silencing of a subset of common targets as well as of a number of independent loci.

Fig. 3.

Synergy of AtMORC6 and MOM1 in transposon silencing. (A) Venn diagram showing relationships between sets of TEs called up-regulated (fourfold increase in expression; FDR < 0.05) for different genotypes. Grayed regions highlight sets with no elements, and red shading highlights TEs uniquely called up-regulated in the higher order mutant. (B) Boxplot and (C) heatmap of average RPKM values between two biological replicates for TEs uniquely called up-regulated in the mom1/atmorc6 mutant background for different genotypes. An asterisk indicates a significant increase relative to all other genotypes (P < 1e-8, Mann–Whitney U test). (D) Metagene analysis of DNA methylation levels across all Arabidopsis TEs for the atmorc6-3, mom1-2, mom1-2/atmorc6-3, and wild-type genotypes. Also shown are the methylation levels at TEs up-regulated in mutant genotypes.

To further understand the relationship between MOM1- and AtMORC6-mediated transcriptional silencing, we generated a double mutant lacking MOM1 and AtMORC6. RNA-seq analyses in mom1/atmorc6 showed a significant increase in derepression of TEs and to a smaller extent of protein-coding genes compared with both single mutants (Fig. 3 A–C and Fig. 4 A–C). RT-PCR analyses corroborated the synergistic derepression of SDC and RomaniatT5 (Fig. S4A). Overexpressed TEs in all three genotypes profiled by RNA-seq are predominantly located in the pericentromeric heterochromatin and belong to diverse families, consistent with previous reports (18, 26) (Fig. S4 B and C). Genome-wide BS-seq analysis showed that DNA methylation was unchanged in TEs up-regulated in mom1/atmorc6 (Fig. 3D). Similar to AtMORC6 target loci, protein-coding genes significantly up-regulated in mom1 were preferentially located in heterochromatin (Fig. 4D). Furthermore, transcription of these was not affected in the drm1/drm2 mutant, suggesting a limited role of MOM1 in RdDM (Fig. 4E). Altogether, these results indicate that AtMORC6 and MOM1 act synergistically to silence a largely common set of heterochromatic DNA elements through two independent pathways.

Fig. 4.

Synergy of AtMORC6 and MOM1 in gene silencing. (A) Venn diagram showing relationships between sets of protein-coding genes called up-regulated (fourfold increase in expression; FDR < 0.05) for different genotypes. Grayed regions highlight sets with no elements. (B) Boxplot and (C) heatmap of average RPKM values for different genotypes (two biological replicates) for protein-coding genes uniquely called up-regulated in the mom1/atmorc6 mutant background. An asterisk represents a significant increase relative to wild-type samples (P < 1e-2, Mann–Whintey U test). (D) Overrepresentation in H3K9me2-enriched heterochromatin of protein-coding genes significantly up-regulated in atmorc6-3, mom1-2, or mom1-2/atmorc6-3 mutants. An asterisk indicates a significant increase relative to all protein-coding genes (P < 1e-3, Fisher’s exact test). (E) Metagene analysis of RNA-seq reads over protein-coding genes called up-regulated in atmorc6-3, mom1-2, or mom1-2/atmorc6-3 mutants. Reads are derived from previously published RNA-seq libraries for two replicates of the drm1/drm2 double mutant and the corresponding wild type (WT).

Conclusion

In this study, we combined biochemistry, genetics, and genomics to understand further the mode of action of the recently discovered Arabidopsis MORC homologs. We found that AtMORC6-mediated transcriptional silencing requires the formation of mutually exclusive heteromers with AtMORC1 and its closest homolog, AtMORC2. Further biochemical studies involving domain deletions or point mutations should uncover the molecular mechanisms of the AtMORC proteins and the implication of heteromerization for ATPase activity. It is interesting to note the similarities between AtMORCs and the structural maintenance of chromosome proteins cohesin and condensin (44). These three protein families are ATPases that function in vivo as heteromers and modulate chromatin superstructure to regulate proper expression and maintenance of genomic integrity.

Genetic and RNA-seq analyses showed that AtMORC6 acts synergistically with the putative chromatin remodeler MOM1 to silence a common set of heterochromatin-localized loci. The synergistic effect observed in the mom1/atmorc6 double mutant suggests that AtMORC6 and MOM1 act in two convergent pathways that are both required for the proper silencing of pericentromeric heterochromatin. It has been previously shown that AtMORC6 and AtMORC1 accumulate in the nucleus as discrete nuclear bodies that localize in the vicinity of the heterochromatic chromocenters (26). It will be interesting to determine in the future whether MOM1 accumulates in a similar fashion in the nucleus to form distinct nuclear bodies. The identification of MOM1 interactors will also be crucial to understanding its mode of action.

Materials and Methods

Plant Material and Growing Conditions.

Wild-type and all mutant lines are from the ecotype Columbia and were grown under continuous light. Plant lines used include atmorc1-2 (SAIL_893_B06; crt1-2), atmorc1-4 (SAIL_1239_C08), atmorc1-5 (SAIL_131_H11; crt1-5), atmorc2-1 (SALK_072774C; crh1-1), atmorc2-4 (SALK_021267C; crh1-4), atmorc6-3 (GABI_599B06), cmt3-11 (SALK_148381), and mom1-2 (SAIL_610_G01). EMS mutagenized atmorc6-1 and cmt3/morc1-3 lines and complementing AtMORC1-MYC and AtMORC6-MYC lines are described in ref. 26. T-DNA insertions were confirmed by PCR-based genotyping. Primer sequences are described in Table S1.

Cloning of pAtMORC1::AtMORC1-FLAG, pAtMORC2::FLAG-AtMORC2, and pAtMORC16::AtMORC6-FLAG.

Cloning was done according to ref. 26. Briefly, AtMORC1 and AtMORC6 genomic regions were PCR amplified and the FLAG epitope was added to the C terminus of AtMORC1 and AtMORC6 and at the N terminus of AtMORC2. The amplified region includes a ∼1 Kb promoter sequence upstream of the respective transcriptional start site.

IP and MS Analysis.

Ten grams of 2-wk-old seedling tissue of each epitope-tagged line were ground in liquid nitrogen and resuspended in 45 mL ice-cold IP buffer [50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 5 mM MgCl₂, 0.1% Nonidet P-40, 10% (vol/vol) glycerol, 1× Protease Inhibitor Mixture (Roche)] and centrifuged for 10 min at 4 °C at 16,000 × g. We added 200 μL M2 magnetic FLAG-beads (SIGMA, M8823) to the supernatants and incubated it for 60 min rotating at 4 °C. M2 magnetic FLAG-beads were washed five times in ice-cold IP buffer for 5 min rotating at 4 °C, and immunoprecipitated proteins were eluted three times with 100 μL 3×-FLAG peptides (SIGMA, F4799) for 15 min at 25 °C. The eluted protein complexes were precipitated by trichloroacetic acid and subjected to MS analyses as previously described (14).

Co-IP and Immunoblotting.

We ground 1.5 g of 2-wk-old seedling tissue of each epitope-tagged line in liquid nitrogen, resuspended it in 12 mL ice-cold IP buffer [50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 5 mM MgCl₂, 0.1% Nonidet P-40, 10% (vol/vol) glycerol, 1× Protease Inhibitor Mixture (Roche)], and centrifuged it for 10 min at 4 °C at 16,000 × g. We added 100 μL M2 magnetic FLAG-beads (SIGMA, M8823) or 150 μL MYC-conjugated agarose beads (COVANCE, AFC-150P-1000) to the supernatants and incubated it for 60 min rotating at 4 °C. Beads were washed five times in ice-cold IP buffer for 5 min rotating at 4 °C, and immunoprecipitated proteins were eluted in 1× Lämmli buffer for 15 min at 80 °C.

Western blots were performed as previously described (26) with GFP-specific antibody (Invitrogen, AA1122), HRP-coupled FLAG-specific antibody (SIGMA, A8592), and MYC-specific antibody (Pierce, MA1-980).

Gel Filtration.

Gel filtration experiments were performed according to ref. 37. Briefly, 0.5 g of 2-wk-old seedling tissue of each epitope-tagged line were ground in liquid nitrogen and resuspended in 1 mL of ice-cold IP buffer [50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 0.1% Nonidet P-40, 10% (vol/vol) glycerol, 1× Protease Inhibitor Mixture (Roche)] and centrifuged for 10 min at 4 °C at 16,000 × g. The supernatants were centrifuged again for 10 min at 4 °C at 16,000 × g. The supernatants were then centrifuged through a 0.2 μm filter (Millipore), 500 μL were loaded onto a Superdex 200 10/300GL column (GE Healthcare, 17–5175-01) column, and 250 μL fractions were collected. We ran 20 μL of every collected fraction on a 4–12% SDS/PAGE. Before use, the column was equilibrated and calibrated with gel filtration standards (Biorad, 151–1901).

RNA Extraction.

We froze 100 mg of 20-d-old leaf tissue in liquid nitrogen. The frozen leaves were then added to a mortar containing liquid nitrogen. Immediately after the liquid nitrogen boiled off, the leaf tissue was crushed to powder using a pestle. We immediately added 1.2 mL of TRIzol Reagent (Life Technologies 15596) to the cold powder, and then it was pulverized further until a clear, dark brown solution was visible. The solution was transferred to a chilled Eppendorf tube, and 400 μL of chloroform was added. The tube was vortexed for 5 s at maximum power, then spun in a centrifuge at 16,000 × g (4 °C) for 10 min to separate the aqueous and organic phases. We collected 700 μL of the aqueous (top) phase. To precipitate the RNA, 700 μL of isopropanol was added to the aqueous material, the solution was vortexed for 5 s at maximum power, and then it was centrifuged for 10 min at 16,000 × g (4 °C). The supernatant was removed, and 500 μL of room temperature 80% (vol/vol) ethanol was added to the pellet, which was then spun for 5 min at 16,000 × g (4 °C). The supernatant was removed and the pellet was air-dried for 5 to 10 min. The pelleted RNA was resuspended in 100 μL water and then purified using the Qiagen RNeasy Mini (Qiagen 74104) “RNA Cleanup Protocol” according to manufacturer’s instructions. RNA was quantified using Nanodrop.

RT-PCR.

We treated 1 μg of input RNA with DNase I (Life Technologies, 18068) according to the manufacturer’s protocol. Of the 11 μL final reaction volume, 3 μL was set aside as a negative control for RT-PCR, whereas 8 μL was converted to cDNA using SuperScript III (Life Technologies 18080). We used 5% of cDNA for each RT-PCR. RT-PCR was performed using IQ SYBR Green Supermix (Bio-Rad 170–8880), with 375 nM final primer concentration using a Stratagene Mx3005p instrument. Amplification conditions were as follows: 95 °C 10:00; 40 cycles, 95 °C, 30 s, 55 °C 1:00, 72 °C 1:00; melting curve. At least two technical replicates were performed per biological replicate, and three biological replicates were used in all experiments. Relative abundance of transcripts was calculated using the difference of squares method. Primer sequences are described in Table S1.

BS-Seq, RNA-Seq, and Accession Codes.

BS-seq was done according to ref. 26. RNA-seq libraries were generated using 2 μg of input RNA using TruSeq RNA Sample Preparation Kit v2 (Illumina RS-122-2001) according to the manufacturer’s protocols. Sequencing data were deposited into Gene Expression Omnibus under accession no. GSE54677.