|Home | About | Journals | Submit | Contact Us | Français|
In addition to RNA polymerases I, II, and III, the essential RNA polymerases present in all eukaryotes, plants have two additional nuclear RNA polymerases, abbreviated as Pol IV and Pol V, that play nonredundant roles in siRNA-directed DNA methylation and gene silencing. We show that Arabidopsis Pol IV and Pol V are composed of subunits that are paralogous or identical to the 12 subunits of Pol II. Four subunits of Pol IV are distinct from their Pol II paralogs, six subunits of Pol V are distinct from their Pol II paralogs, and four subunits differ between Pol IV and Pol V. Importantly, the subunit differences occur in key positions relative to the template entry and RNA exit paths. Our findings support the hypothesis that Pol IV and Pol V are Pol II-like enzymes that evolved specialized roles in the production of noncoding transcripts for RNA silencing and genome defense.
In bacteria and Archaea, a single multisubunit RNA polymerase transcribes genomic DNA into RNA. By contrast, eukaryotes have three essential nuclear DNA-dependent RNA polymerases that perform distinct functions. For instance, 45S ribosomal RNA (rRNA) genes are transcribed by RNA polymerase I (Pol I), mRNAs are transcribed by RNA polymerase II (Pol II), and tRNAs and 5S rRNA are transcribed by RNA polymerase III (Pol III) (Grummt, 2003; Schramm and Hernandez, 2002; Woychik and Hampsey, 2002).
Bacterial DNA-dependent RNA polymerase (RNAP) is composed of only four different proteins (β′, β, ω, α; with two molecules of α in the core enzyme), but archaeal RNAP and eukaryotic Pol I, II, and III are more complex (Cramer et al., 2001; Darst et al., 1998; Hirata et al., 2008). Archaea have a fundamental subunit number of 10, with the caveat that the two largest subunits are generally split into two genes (Werner, 2007). Pol I, II, and III have 12–17 subunits that include homologs of archaeal polymerase subunits, suggesting their functional diversification from an archaeal progenitor. The crystal structures of bacterial, archaeal, and eukaryotic Pol II are fundamentally similar (Cramer et al., 2001; Darst et al., 1998; Hirata et al., 2008). In each case, the largest and second-largest subunits, corresponding to the β′ and β subunits of E. coli RNAP, respectively, are the catalytic subunits that interact to form the DNA entry and exit channels, the active site, and the RNA exit channel.
Sequencing of the Arabidopsis thaliana genome revealed genes for the expected catalytic subunits of Pol I, II, and III but unexpectedly revealed two atypical largest subunit genes and two atypical second-largest subunit genes (reviewed in Pikaard et al., 2008). Moreover, five subunits of Pol I, II, and III that are typically encoded by single genes in yeast and mammals, namely RPB5, RPB6, RPB8, RPB10, and RPB12 (named according to their discovery as Pol II subunits; aka RNA Polymerase B) (Cramer, 2002; Werner, 2007), are encoded by multigene families in Arabidopsis, as are the Pol II-specific subunits RPB3, RPB4, RPB7, and RPB9. The functional significance of the extensive subunit diversity in plants is unclear.
The genes encoding the atypical largest and second-largest polymerase subunits in Arabidopsis are not essential for viability (Herr et al., 2005; Kanno et al., 2005; Onodera et al., 2005; Pontier et al., 2005), unlike their Pol I, II, or III counter-parts (Onodera et al., 2008). However, the atypical catalytic subunits are nuclear proteins (Onodera et al., 2005; Pontes et al., 2006) required for siRNA-directed DNA methylation and silencing of retrotransposons, endogenous repeats, and transgenes (Herr et al., 2005; Kanno et al., 2005; Onodera et al., 2005; Pontier et al., 2005). The atypical catalytic subunit genes also play roles in the short-range or long-distance spread of RNA-silencing signals, responses to biotic and abiotic stresses, and the control of flowering time (Borsani et al., 2005; Brosnan et al., 2007; Dunoyer et al., 2007; Katiyar-Agarwal et al., 2007; Pontier et al., 2005; Smith et al., 2007). The atypical largest subunit genes are NRPD1 and NRPE1. NRPD1 (formerly NRPD1a) is the largest subunit of Nuclear RNA polymerase IV (Pol IV; formerly Pol IVa) (Herr et al., 2005; Onodera et al., 2005), whereas NRPE1 (formerly NRPD1b) is the largest subunit of Pol V (formerly Pol IVb) (Kanno et al., 2005; Pontier et al., 2005). The second-largest subunits of Pol IV and Pol V are encoded by the same gene, designated by the synonymous names NRPD2a (NRPD2 for simplicity) or NRPE2 (Herr et al., 2005; Kanno et al., 2005; Onodera et al., 2005; Pontier et al., 2005). Pol IV and Pol V are functionally distinct, with Pol IV required for siRNA production and Pol V generating noncoding transcripts at target loci (Wierzbicki et al., 2008). Our current model is that siRNAs bind to Pol V nascent transcripts to bring the silencing machinery to the vicinity of the chromatin at target loci (Wierzbicki et al., 2008).
Aside from their largest and second-largest subunits, the subunit compositions of Pol IV and Pol V are unknown. Here, we show that Pol IV and Pol V have subunit compositions characteristic of Pol II but make differential use of RPB3, RPB4, RPB5, and RPB7 family variants in addition to having distinct catalytic subunits. Collectively, our results support the hypothesis that Pol IV and Pol V are RNA Pol II derivatives whose molecular niche is the production of noncoding transcripts for RNA-mediated silencing.
To affinity purify Pol IV and Pol V from Arabidopsis thaliana, we engineered full-length NRPD1 (NRPD1a) and NRPE1 (NRPD1b) genomic clones, including their promoter regions and complete sets of introns and exons, adding a FLAG epitope tag to the protein’s C terminus. The transgenes rescue the loss of RNA-directed DNA methylation in their respective null mutants (nrpd1a-3 or nrpd1b-11), indicating that the recombinant proteins are functional (Pontes et al., 2006). NRPD1-FLAG and NRPE1-FLAG, and their respective associated subunits, were affinity purified on anti-FLAG resin, and tryptic peptides were identified by using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). For both Pol IV and Pol V, their two known catalytic subunits were detected, as expected. However, in each case, ten additional previously unknown subunits were identified, corresponding to the ten noncatalytic subunits of yeast RNA Pol II: RPB3, RPB4, RPB5, RPB6, RPB7, RPB8, RPB9, RPB10, RPB11, and RPB12 (Figure 1; see Table S1 and Figures S1 and S2, available online). The pairs of catalytic subunits specific to RNA Pol I, II, or III were not detected in Pol IV or Pol V samples, ruling out copurification of these polymerases as an explanation for the noncatalytic subunits detected in affinity-purified Pol IV or Pol V. Likewise, coimmunoprecipitation (colP) data show that Pol IV and Pol V do not associate with each other or with Pol I, II, or III (Figure 2A).
For Pol V, peptide sequence data typically allowed unambiguous identification of subunits that are members of protein families (see Figure S1 for peptide coverage maps and Figures S4–S12 for family alignments). An exception was the RPB8 family, for which the sole peptide identified matched both variants, which are 96% identical. Two RPB3-related variants that are 88% identical are present in Arabidopsis, and both proteins are detected in Pol V, resulting in their designation as NRPE3a and NRPE3b (Figure 1, Figure 3A). The single RPB11 subunit encoded by the Arabidopsis genome was also detected; hence we refer to this protein as NRPE11 (Figure 1). Of six homologs of RPB5 in the genome, only one (NRPE5) is detected in Pol V (Figure 1, Figure S5). Two RPB9-like subunits were identified in Pol V (Figures 1 and and2D).2D). These proteins, designated NRPE9a and NHPE9b, are 92% identical. There are four RPB7 homologs in Arabidopsis, only one of which is detected in Pol V, NRPE7. One of two RPB4-like subunits (NRPE4), one of two RPB10-like subunits (NRPE10), one of two RPB12-like subunits (NRPE12), and one of two RPB6-like subunits (NRPE6a) were also detected in Pol V (Figure 1).
Analysis of Pol IV’s subunit composition revealed similarities and differences compared to Pol V (Figure 1, Figure S2). As with Pol V, peptides for the single RPB11-like subunit were identified. In the context of Pol IV, we refer to this protein as NRPD11; in the context of Pol V, we refer to this same protein as NRPE11. Similar nomenclature rules were adopted for other subunits shared by more than one polymerase (see Figure 1 for synonyms). NRPD4, NRPD6a, NRPD8b, and NRPD10 subunits were unambiguously identified (Figure 1). Similar to Pol V, both RPB3-like variants were detected in Pol IV, but one is predominant (NRPD3; see Figure 1). Interestingly, the RPB5-like subunit of Pol IV, NRPD5, is identical to the previously identified NRPB5 subunit of Pol II but differs from the NRPE5 subunit of Pol V (Figure 1) (Larkin et al., 1999). The major NRPD7 subunit detected in Pol IV is 62% identical to the Pol V NRPE7 subunit, but low-level peptide sequence coverage for the NRPE7 subunit was detected as well. The Pol IV NRPD9b subunit corresponds to NRPE9b detected in Pol V (Figures 1 and and2D2D).
The significant number of Pol II-like subunits in Pol IV and Pol V raised questions concerning the relative similarities of Pol II, Pol IV, and Pol V. Therefore, we affinity purified Arabidopsis Pol II by exploiting epitope-tagged NRPB2 (NRPB2-FLAG) expressed from a transgene that rescues the nrpb2-1 null mutant (Onodera et al., 2008). LC/MS-MS revealed 12 subunits orthologous to their 12 yeast Pol II counterparts, with no contaminating subunits specific to Pol I, III, IV, or V (Figure 1, Figure S3). The same RPB10, RPB11, and RPB12 family subunits found in Pol IV and/or Pol V are present in Pol II (Figure 1). Sequenced peptide coverage for the RPB6, RPB8, and RPB9-like subunits in the Pol II dataset revealed that each of the two genes for these subunits encodes a subunit incorporated into Pol II (Figure S3), suggesting that the genes are redundant. A single RPB3-like subunit, NRPB3, is predominant in Pol II, consistent with a previous report (Ulmasov et al., 1996). However, peptides corresponding to the NRPE3b subunit were also detected at low frequency. The single RPB5 subunit identified in Pol II corresponds to the expected subunit based on a previous study (Larkin et al., 1999) and is identical to the NRPD5 subunit of Pol IV but distinct from the NRPE5 subunit of Pol V. Pol II also makes use of RPB4 and RPB7 variants that are distinct from the corresponding Pol IV and Pol V subunits. These NRPB4 and NRPB7 subunits correspond to subunits previously shown to associate with Pol II (Larkin and Guilfoyle, 1998).
To test subunit associations with all five nuclear RNA polymerases, we exploited Arabidopsis lines expressing FLAG-tagged Pol I, II, and III second-largest subunits (NRPA2-FLAG, NRPB2-FLAG, or NRPC2-FLAG) or FLAG-tagged Pol IV and Pol V largest subunits (NRPD1-FLAG, NRPE1-FLAG), each expressed from trangenes that rescue corresponding null mutants (Onodera et al., 2008; Pontes et al., 2006). Plants expressing FLAG-tagged genomic clones of NRPE6a, NRPE8b, NRPE10, or NRPE11 or an NRPE5 cDNA were also engineered. Each recombinant protein could be immunoprecipitated from transgenic plants and detected by immunoblotting using anti-FLAG antibody (Figure 2A). Probing immunoblots with antibodies for NRPE1 and NRPE2 (Onodera et al., 2005) revealed that these Pol V catalytic subunits are present in NRPE1, NRPE6a, NRPE8b, NRPE10, NRPE11, and NRPE5 immunoprecipitates (Figure 2A; see also the anti-NRPE1 specificity control in Figure 2B), consistent with the detection of all of these subunits in Pol V (Figure 1). Controls show that NRPE2 and NRPE1 do not coimmunoprecipitate with Pol I, II, or III; that NRPE1 does not coimmunoprecipitate with Pol IV; and that NRPE2/NRPD2 is present in Pol IV and Pol V, as expected. The anti-NRPE1 antibody consistently reveals multiple NRPE1 isoforms (Figures 2A and 2B); whether these are degradation, posttranslational modification, or alternative splicing products is unclear.
To test whether NRPE5, NRPE6a, NRPE8b, NRPE10a, and NRPE11 subunits are shared by Pol I, II, and/or III, we used an anti-peptide antibody recognizing an invariant sequence in the Pol I, II, and III second-largest subunits (Onodera et al., 2005); this antibody fails to crossreact with NRPE2/NRPD2 due to a single amino acid substitution. In NRPE6a, NRPE8b, NRPE10, and NRPE11 immunoprecipitated fractions, Pol I, II, or III second-largest subunits are detected, consistent with the LC-MS/MS analysis of Pol II (Figures 1 and and2A).2A). In yeast, RPB6, RPB8, and RPB10 are common to Pol I, II, and III, but RPB11 is Pol II specific. Second-largest subunits of Pol I, II, or III do not coimmunoprecipitate with FLAG-NRPE5, showing that NRPE5 is not a subunit of the essential polymerases (Figure 2A).
The LC-MS/MS data indicate that either of the two RPB8 homologs associate with Pol V. ColP analysis confirms that NRPE8a or NRPE8b will coimmunoprecipitate with the Pol V catalytic subunits (Figures 2A and 2E). Although LC-MS/MS identified only one RPB6 variant (NRPE6a), its paralog (NRPE6b) can also associate with Pol V in vivo (Figure 2C). Both Pol II clade RPB9-like subunits (Figure 2D) were detected in Pol V by LC-MS/MS. ColP analysis confirms that FLAG-NRPE9a associates with the Pol V NRPE1 and NRPE2 catalytic subunits in vivo (Figures 2C and 2D). NRPE6b and NRPE9a also coimmunoprecipitate the second-largest subunits of Pol I, II, or III (Figure 2C).
LC-MS/MS analysis of Pol V identified both potential RPB3 variants (Figure 3A). In confirmation of this result, HA-tagged NRPE3a and NRPE3b both coimmunoprecipitate the Pol V catalytic subunits (Figure 3B). NRPE3a, but not NRPE3b, also coimmunoprecipitates a subunit recognized by the antibody specific for Pol I, II, or III second subunits (Figure 3B); we deduce this to be the Pol II NRPB2 subunit because Pol I and Pol III use third-largest subunits distinct from RPB3. Moreover, the gene encoding NRPE3a was previously shown to encode a NRPB3 (see Figure 1) subunit present in purified Pol II (Ulmasov et al., 1996).
NRPE11, NRPE6a, NRPE8b, NRPE10, and NRPE9a all coimmunoprecipitate with the Pol IV and Pol II largest subunits (Figures 1 and and4A).4A). Upon immunoprecipitation of NRPE3b, no Pol II is detected in the immunoprecipitated fraction using an antibody recognizing the C-terminal domain (CTD) of the largest subunit. Likewise, Pol IV is detected in only trace amounts using the anti-NRPD1 antibody. We conclude that NRPE3b is used almost exclusively by Pol V (Figures 1 and and4A).4A). In contrast, NRPB3, NRPD3, and NRPE3a are encoded by the same gene. Controls show that the NRPD1 subunit of Pol IV does not coimmunoprecipitate with Pol I, II, III, or V (Figure 4A). Likewise, the NRPB1 subunit of Pol II does not coimmunoprecipitate with Pol I, III, IV, or V (Figure 4A).
Using antibodies specific for NRPB5/NRPD5 or NRPE5 (Larkin et al., 1999), we tested their associations with FLAG-tagged Pol I, II, III, IV, or V (Figures 4B and 4C). Controls show that the NRPD2/NRPE2 subunit common to both Pol IV and Pol V is detected in NRPD1 and NRPE1 IPs, as expected, but not in Pol I, II, or III IPs (Figures 4B and 4C). NRPE5 was detected only in the NRPE1-FLAG immunoprecipitated fraction (Figure 4B), confirming that this subunit is unique to Pol V. By contrast, the NRP85/NRPD5 subunit is detected in Pol I, II, III, and IV fractions, but not in Pol V (Figure 4C), in agreement with the LC-MS/MS data and previous studies showing that NRPB5/NRPD5 copurifies with Pol I, II, and III (Larkin et al., 1999) (Saez-Vasquez and Pikaard, 1997).
We affinity purified FLAG-tagged NRPE5 expressed in the nrpe5 mutant background and identified the associated RNA polymerase subunits using LC-MS/MS. The results confirmed association of NRPE5 with all Pol V subunits except NRPE7 (Table S2, Figure S18), which most likely escaped detection in this experiment due to insufficient sample mass.
Collectively, the immunological tests of Figures 2–4 confirm the Pol V association of the NRPE1, NRPE2, NRPE3a, NRPE3b, NRPE5, NRPE6a, NRPE8, NRPE9a, NRPE10, and NRPE11 subunits detected by LC-MS/MS. Likewise, the immunological tests confirm the Pol IV associations of NRPD1, NRPD2, NRPD3, NRPD5, NRPD6a, NRPD8b, NRPD9a, NRPD10, and NRPD11. Pol IV and Pol V subunits that are shared with Pol II were also confirmed immunologically.
Of the five full-length homologs of yeast RPB5 in Arabidopsis, RT-PCR analysis shows that only NRPB5/NRPD5 and NRPE5 are constitutively expressed; other family members show organ-specific expression patterns (Figure 5A, Figures S5 and S13). Homozygous nrpe5-1 mutants resulting from a T-DNA insertion (Figure 5B) are viable, as are Pol V nrpe1 and nrpe2 mutants. In contrast, homozygous nrpd5-1/nrpb5-1 T-DNA insertion mutants were not recoverable due to female gametophyte lethality, as shown by reciprocal genetic crosses (Figures S14A and S14B). Female gametophyte lethality is a characteristic of Pol I, II, and III mutants, as demonstrated previously for nrpa2, nrpb2, nrpc2, and nrpb12 (Onodera et al., 2008). A homozygous nrpe11 T-DNA insertion mutant was also unrecoverable, consistent with this gene also encoding the Pol II subunit, NRPB11 (Figures S14A and S14B).
Like Pol IV and Pol V catalytic subunit mutants, nrpe5-1 mutants lack obvious morphological phenotypes but flower later than wild-type plants under short-day conditions (Figure 5C), similar to mutants disrupting the 24 nt siRNA-directed DNA methylation pathway, including RNA-DEPENDENT RNA POLYMERASE 2 (RDR2) and DICER-LIKE 3 (DCL3) mutants (Chan et al., 2004; Liu et al., 2007; Pontier et al., 2005). Comparison of nrpe5 and wild-type individuals suggests that the delay in flowering is stochastic, with some individuals showing substantial delays and others flowering at the same time as wild-type plants (Figure S15).
We tested nrpe5-1 mutants for Pol V-dependent molecular phenotypes, including DNA hypermethylation at 5S rRNA gene clusters and at AtSN1 and AtSN2 retroelements. In nrpd1 (nrpd1a-3),nrpe1 (nrpd1b-11), and nrpd2/nrpe2 mutants, loss of methylation at 5S rDNA repeats results in increased digestion by the methylation-sensitive restriction endonucleases HpaII and HaeIII compared to wild-type plants (Figure 5D). In the nrpe5 mutant, methylation at 5S rRNA genes is reduced compared to wild-type, but to a lesser extent than in nrpe1 or nrpd2/nrpe2 mutants (Figure 5D). Transformation of the nrpe5-1 mutant with a 35S:FLAG-NRPE5 transgene restores methylation to wild-type levels, as shown in three independent transgenic lines (Figure 5D).
To test whether nrpe5 affects DNA methylation at other Pol V-dependent loci, we examined the SINE retrotransposon families, AtSN1 and AtSN2 (Myouga et al., 2001). In wild-type plants, AtSN1 and AtSN2 elements are heavily methylated such that their DNA is not cut by HaeIII and a PCR product can be obtained (Figures 5E and 5F). In nrpe1 and nrpe2/nrpd2 mutants, however, methylation is lost such that HaeIII cuts and PCR amplification fails (Figures 5E and 5F). In nrpe5-1, decreased AtSN1 and AtSN2 methylation occurs, but not as severely as in nrpe1 or nrpe2/nrpd2 mutants. Nonetheless, the decreased methylation in nrpe5-l plants is rescued by a 35S:FLAG-NRPE5 transgene (Figures 5E and 5F).
RNA-directed DNA methylation silences AtSN1 retroelements in wild-type plants such that loss of methylation correlates with increased AtSN1 transcription (Hamilton et al., 2002; Herr et al., 2005; Kanno et al., 2005). AtSN1 transcripts are barely detectable in wild-type plants but are abundant in nrpe5 mutants, as in nrpe1 or nrpe2/nrpd2 mutants (Figure 5G). In the nrpe5-1 genetic background, the 35S:FLAG-NRPE5 transgene restores AtSN1 silencing (Figure 5G). Collectively, these results demonstrate that NRPE5 is important for DNA methylation and silencing of AtSN1 elements.
In the RNA-directed DNA methylation pathway, Pol IV is required for 24 nt siRNA production (Herr et al., 2005; Onodera et al., 2005) such that siRNAs are eliminated in nrpd1 and nrpd2 mutants (Figure 5H). In contrast, siRNAs in nrpe1 mutants are reduced but not eliminated at 5S rRNA genes and COP1A elements (Figure 5H). Consistent with a Pol V mutant phenotype, siRNAs are reduced in nrpe5 mutants relative to wild-type and are restored by the 35S:FLAG-NRPE5 transgene (Figure 5H). MicroRNA and trans-acting siRNA levels are unaffected in nrpe5, nrpd1, or nrpe1 mutants, consistent with the lack of Pol IV or Pol V involvement in these pathways.
Crystallographic studies indicate that yeast RPB5 is composed of an N-terminal jaw domain and a C-terminal assembly domain separated by a short linker (Figures S5, S16, and S17A). These domains appear to be conserved in nearly all plant RPB5 homologs (Figure S16). A feature of Arabidopsis NRPE5, and its presumptive orthologs in other plants, is a short N-terminal extension compared to NRPB5 (Figure S16 and S17A). To test the functional significance of this N-terminal extension, we created a 35S:FLAG-ΔN-NRPE5 construct in which the extension was deleted (Figure S17A). This transgene fails to rescue nrpe5-1 mutant phenotypes (Figures S17B–S17D). Surprisingly, immunoprecipitation of equal volumes of soluble extracts revealed that the FLAG-ΔN-NRPE5 protein is present at very low levels relative to full-length FLAG-NRPE5, despite similar transcript levels (Figure S17E). These data suggest that the N-terminal extension is important for the stability of the NRPE5 protein in vivo, possibly because the extended sequence facilitates Pol V-specific subunit interactions.
Pol IV and Pol V are plant-specific enzymes that appear to have originated in an algal progenitor of land plants several hundred million years ago (Luo and Hall, 2007). Their specific involvement in siRNA-mediated transcriptional gene silencing, which also occurs in other metazoans and fission yeast, has begged the question as to which polymerases accomplish the functions of Pol IV and Pol V in other eukaryotes. In fission yeast, Pol II transcripts traverse silenced loci, serving as binding sites for siRNAs and as templates for the sole RNA-dependent RNA polymerase, thereby generating precursors for further siRNA biogenesis (Buhler and Moazed, 2007; Buhler et al., 2006; Grewal and Elgin, 2007; Irvine et al., 2006). Several nonlethal mutations that disrupt siRNA-mediated silencing and/or siRNA accumulation in S. pombe have been mapped to the RPB1, RPB2, and RPB7 subunits of Pol II (Djupedal et al., 2005; Kato et al., 2005; Schramke et al., 2005). Our finding that Pol IV and V have Pol II-like subunit compositions fits the hypothesis that Pol IV and Pol V are derivatives of Pol II that evolved specialized roles in RNA silencing but no longer perform Pol II functions essential for viability, in contrast to fission yeast Pol II, which appears to accomplish all of these tasks. Presumably, the subunits of Pol IV/V that are not shared by Pol II, including NRPD1, NRPE1, NRPD2/NRPE2, NRPE3b, NRPD4/NRPE4, NRPE5, NRPD7, and NRPE7, account for Pol IV- or Pol V-specific activities. It is intriguing that most of these subunits occupy key positions with regard to the template channel and RNA exit paths (Figures 6A and 6B).
Previous analyses of Pol IV and Pol V catalytic subunits had pointed to a Pol II connection. In our initial study of Pol IV, we noted that the NRPD2/NRPE2 subunit is more closely related to the second-largest subunit of Pol II than to the corresponding subunits of Pol I or Pol III (Onodera et al., 2005). Moreover, five out of eight intron positions in the beginning of NRPD1 and NRPE1 match the intron positions in NRPB1, encoding the largest subunit of Pol II (Luo and Hall, 2007). Based on phylogenetic analyses, Luo and Hall proposed that Pol IV came into existence following a duplication of the NRPB1 gene that generated the NRPD1 gene. A subsequent duplication of NRPD1 to generate NRPE1 is proposed to have led to the evolution of Pol V after the emergence of land plants but prior to the divergence of angiosperms (flowering plants). Our finding that Pol IV utilizes the same RPB5-family subunit as Pol I, II, and III whereas Pol V uses a distinct variant (NRPE5) is consistent with the hypothesis that Pol V is more distantly related to Pol II than is Pol IV.
The fact that Pol IV and Pol V share numerous small subunits with Pol II, including NRPB3, NRPB6, NRPB8, NRPB9, NRPB10, NRPB11, and NRPB12 family subunits, can explain why alleles for these genes have not been identified in genetic screens; loss-of-function mutations in the subunits of essential polymerases cause female gametophyte lethality (Figure S14) (Onodera et al., 2008). Likewise, the use of more than one NRPE3, NRPE6, NRPE8, or NRPE9 variant by Pol IV or Pol V (Figures 6C and and1)1) can be expected to make identification of mutations in these genes problematic due to functional redundancies (Figure 6C).
A number of observations in our study fill in gaps concerning the functions of RNA polymerase subunit families in Arabidopsis. For instance, Ulmasov et al. reported the existence of two RPB3-like genes in Arabidopsis, which they named AtRPB36a and AtRPB36b based on their predicted sizes of ~36 kD (Ulmasov et al., 1996). AtRPB36a was found in highly purified Pol II fractions (Ulmasov et al., 1996), but AtRPB36b was not, making the function of the latter variant unclear. Our study reveals that AtRPB36b is the NRPE3b subunit of Pol V. AtRPB36a (now NRPB3) and NRPB11 (formerly AtRPB13.6) in Pol II are the homologs and functional equivalents of the two α subunits (α and α′) of E. coli RNA polymerase. Previous studies demonstrated that NRPB3 and NRPB11 copurify with Pol II in vivo and physically interact in yeast two-hybrid assays (Ulmasov et al., 1996). Interestingly, AtRPB36b/NRPE3b also interacted with NRPB11 in yeast two-hybrid assays (Ulmasov et al., 1996), which is likely to be meaningful, occurring in the context of Pol V in a manner equivalent to the interaction of NRPB3 and NRPB11 in Pol II. Interestingly, the AtRPB36a variant also associates with Pol V in vivo; therefore, this protein serves as the NRPB3 subunit of Pol II, the NRPD3 subunit of Pol IV, and one of two alternative Pol V NRPE3 subunits (NRPE3a). How these highly similar RPB3-like subunits are differentially assembled into Pol II, IV, or V is a question deserving further study.
Although peptide coverage for the NRPD4/NRPE4 subunit was low in our study, the Jian-Kang Zhu laboratory identified the nrpd4/nrpe4 gene in a screen for defective RNA-directed DNA methylation and confirmed the Pol IV and Pol V association of the encoded protein (He, X.-J., Hsu, Y.-F., Pontes, O., Zhu, J., Lu, J., Bressan, R.A., Pikaard, C., Wang, C.-S., and Zhu, J.-K., unpublished data). In budding yeast, RPB4 forms a subcomplex with RPB7 that can be dissociated from the ten subunit Pol II core enzyme without abolishing Pol II catalytic activity in vitro (Cramer, 2004), although the subcomplex appears to be more stable in Pol II from plants (Larkin and Guilfoyle, 1998). In vivo, RPB7 is an essential protein in yeast, whereas RPB4 deletion mutants are temperature sensitive (McKune et al., 1993; Woychik and Young, 1989) and are impaired in transcription elongation and mRNA 3′ end processing (Runner et al., 2008; Verma-Gaur et al., 2008). It is intriguing that Pol II, IV, and V have unique RPB7-like subunits and that the NRPB4 subunit of Pol II is different from the NRPD4/NRPE4 subunits of Pol IV and Pol V. Given that the RPB4/RPB7 complex is thought to interact with the nascent RNA transcript (see Figure 6), these differences are likely to contribute to the unique functions of Pol II, IV, and V.
Previous studies had shown that one of the two consitutively expressed RPB5 family proteins is a subunit by Pol I, II, and III (Larkin et al., 1999; Saez-Vasquez and Pikaard, 1997). The function of the other variant, formerly designated AtRPB5b or AtRPB23.7, was unknown. Our study reveals that the latter protein is the NRPE5 subunit of Pol V. By contrast, the NRPD5 subunit of Pol IV is encoded by the same gene that encodes the Pol II NRPB5 subunit and the equivalent subunits of Pol I and III. As we have shown, nrpe5-1 mutants display defects in DNA methylation, retroelement silencing, siRNA accumulation, and flowering time, similar to nrpe1 mutants (Herr et al., 2005; Kanno et al., 2005; Onodera et al., 2005; Pontier et al., 2005). However, nrpe5-1 mutant phenotypes are typically less severe than nrpe1 or nrpe2/nrpd2 mutants. Because the T-DNA insertion is near the 3′ end of the gene, nrpe5-1 may be a partially functional allele. It is also possible that other members of the multigene family are partially redundant with NRPE5, particularly At2g41340, which shares 70% identity with NRPE5, including the N-terminal extension that is missing in the NRPB5/NRPD5 subunit (Figure 5A and Figure S5). Consistent with this hypothesis, preliminary evidence suggests that a nrpe5-1 At2g41340 double mutant has a more severe loss of DNA methylation phenotype than does nrpe5-1 (data not shown). A third possibility is that NRPE5 may not be absolutely required for Pol V transcription. The failure to identify nrpe5 alleles in genetic screens to date may stem from one or more of these reasons.
The fact that Pol V is unique in using the NRPE5 variant of the RPB5 family is likely to have functional significance. Crystal structures of yeast Pol II reveal that RPB5 Interacts with RPB1 and RPB6 to form a mobile “shelf” module that stabilizes the template DNA as it enters the polymerase (Cramer et al., 2001; Gnatt et al., 2001). RPB5 also interacts with hepatitis B transcriptional activator protein X (HBx); the general transcription factor TFIIB; TIP120, a protein which facilitates recruitment of Pol II to the preinitiation complex (Cheong et al., 1995; Lin et al., 1997; Makino et al., 1999); and the yeast chromatin remodeling complex, RSC (Soutourina et al., 2006). Therefore, the differential use of the NRPD5 or NRPE5 subunits in the context of Pol IV or Pol V could mediate different template specificity, locus targeting, or transcriptional activation processes.
A. thaliana nrpd1 (allele nrpd1a-3), nrpe1 (allele nrpd1b-11), and nrpd2/nrpe2 (nrpd2a-2 nrpd2b-1) have been described (Pontes et al., 2006). nrpe11-1 (nrpb11-1/nrpd11-1) is from T-DNA line SALK_100563 (Alonso et al., 2003), nrpd5-1/nrpb5-1 from T-DNA line SAIL_786_E02 (Sessions et al., 2002), and nrpe5-1 from GABI-KAT T-DNA line 237A08 (Rosso et al., 2003). Primers for nrpe11-1, nrpd5-1, and nrpe5-1 genotyping are listed in Table S3. Callus cultures were induced by germinating sterilized seeds on MS media containing Gamborg's vitamins (Sigma), 5% agargel (Sigma), 0.02 mg/L kinetin (Sigma), and 2 mg/L 2,4-dichlorophenoxyacetic acid (Sigma). Plates were incubated at 23°C. Callus frozen in liquid N2 was stored at −80°C.
Frozen callus (115–150 g) expressing FLAG-tagged NRPE1 or NRPD1 was ground in extraction buffer (300 mM NaCl, 20 mM Tris [pH 7.5], 5 mM MgCl2, 5 mM DTT, 1 mM PMSF, and 1:100 plant protease inhibitor cocktail [Sigma]) at 4°C, filtered through two layers of Miracloth (Calbiochem), and centrifuged twice at 10,000 g, 15 min, 4°C. Pol II and NRPE5 were purified with the same protocol from 150 g of leaf tissue expressing FLAG-tagged NRPB2 or NRPE5, respectively. Supernatants were incubated with anti-FLAG-M2 resin for 2–3 hr in a 15 ml tube using 30 µl of resin per 14 ml of extract. Resin was pelleted at 1000 rpm for 2 min and the supernatant incubated with fresh resin for 2–3 hr. Pooled resin was washed five times in 14 ml of extraction buffer containing 0.4% NP-40 (Sigma). Aliquots (125 µl) of resin were then mixed 2 min with 125 µl Ag/Ab Elution Buffer (Pierce) at 4°C. Resin was pelleted, and the eluted complex was pooled. Two ~500 µl batches of pooled complex were concentrated in YM-10 centricon columns (Millipore) at 4°C and desalted using Pierce 500 µl desalting columns. The final elute of ~70 µl containing ~10–50 µg of protein was subjected to LC-MS/MS.
Samples adjusted to 50% (v/v) 2,2,2-Trifluoroethanol (TFE) (Sigma) were sonicated 1 min at 0°C and then incubated 2 hr at 60°C with shaking at 300 rpm. Proteins were reduced with 2 mM DTT at 37°C for 1 hr, then diluted 5-fold with 50 mM ammonium bicarbonate. CaCl2 (1 mM) and sequencing-grade modified porcine trypsin (Promega) was added at a 1:50 trypsin-to-protein mass ratio. After 3 hr at 37°C, samples were concentrated to ~30 µl and subjected to reversed-phase liquid chromatography (RPLC) coupled to an electrospray ionization source and LTQ-Orbitrap mass spectrometer (ThermoFisher Scientific). Tandem mass spectra were searched against A. thaliana proteins using SEQUEST and filtering criteria, which provided a false discovery rate (FDR) <5%. See the Supplemental Data for details.
NRPD1 and NRPE1 genomic clones (Pontes et al., 2006) were cloned into a Gateway-compatible vector (A.W. and C.S.P., unpublished data) that adds a C-terminal FLAG tag, 3C protease cleavage site, and biotin ligase recognition peptide. NRPE5, NRPE6a, NRPE6b, NRPE8a, NRPE9a, NRPB7, NRPE3a, and NRPE3b cDNAs were amplified by RT-PCR from poly-T primed cDNA cloned into pENTR-D-TOPO or pENTR-TEV-TOPO. cDNAs were recombined into pEarleyGate 201 (HA tag) or 202 (FLAG tag) (Earley et al., 2006). Genomic NRPE8b, NRPE10, NRPE11 and NRPE6a clones were similarly amplified by PCR and cloned into pEarleyGate 302 (FLAG tag). NRPD1-FLAG, NRPE1-FLAG, NRPA2-FLAG, NRPB2-FLAG, and NRPC2-FLAG transgenes were previously described (Onodera et al., 2008; Pontes et al., 2006).
5S rDNA Southern blot methylation assays and AtSN1 PCR assays were performed using 250 ng−1 µg of DNA as in Onodera et al. (2005).
For AtSN1 transcripts, high-molecular-weight RNA was isolated from 300 mg of leaves using a miRVANA (Ambion) kit, and strand-specific RT-PCR was performed as described (Wierzbicki et al., 2008).
Inflorescence small RNA (7.5 µg) was analyzed by northern blot hybridization using COPIA, siR1003 (5S rRNA), 45S rRNA, miR173, and tasiR255 probes as described previously (Allen et al., 2005; Onodera et al., 2005; Pontes et al., 2006; Xie et al., 2004). Blots stripped twice with 50% formamide, 0.1 × SSC, and 1% SDS at 65°C for 2 hr were reprobed to generate multiple figure panels.
Anti-NRPE2/NRPD2, anti-NRPB5/NRPD5, and anti-NRPE5 have been described (Larkin et al., 1999; Onodera et al., 2005). Anti-FLAG antibodies were from Sigma. Anti-NRPB1-CTD (8WG16) was purchased from Abcam. NRPE1 antibodies (Covance) recognize peptide N-CDKKNSETESDAAAWG-C. NRPD1 antibodies (Covance) recognize peptide N-CLKNGTLESGGFSENP-C. Anti-NRPA2/NRPB2/NRPC2 antibodies (US Biologicals) recognize N-CGDKFSSRHGQKG-C. Antibodies were affinity purified using immobilized peptides.
Leaves (2–4 g) were ground in extraction buffer (Baumberger and Baulcombe, 2005), filtered through Miracloth, and centrifuged at 10,000 g for 15 min. Supernatants were incubated 3–12 hr at 4°C with 30 µl of anti-FLAG-M2 resin (Sigma). Beads were washed three times in extraction buffer + 0.5% NP-40 (Sigma) and eluted with two bed volumes of 2× SDS sample buffer, and 5–20 µl was subjected to SDS-PAGE and transferred to Immobilon PVDF membranes (Millipore). Blots were incubated with antibodies in TBST + 5% (w/v) nonfat dried milk. Antibody dilutions were as follows: 1:250 (NRPE1), 1:500 (NRPD1), 1:2000 (NRPB1-CTD), 1:750 (NRPB5/NRPD5), 1:750 (NRPE5), 1:250 (NRPD2/NRPE2), 1:500 (anti-Pol I, II, and/or III) and 1:2000–1:10,000 (FLAG-HRP). The secondary antibody was anti-rabbit-HRP, diluted 1:5000–1:20,000; or anti-mouse-HRP, diluted 1:5000 (GE Healthcare, Sigma). Blots were washed four times for 4 min in TBST and visualized by chemiluminescence (GE Healthcare). Blots were stripped for 35 min in 25 mM glycine (pH 2.0), 1% SDS: re-equilibrated in TBST; and probed with additional antibodies.
Sequences were aligned using ClustalW and highlighted using BOXSHADE. Construction of phylogenetic trees was performed using MegAlign. Trees are based on ClustalW alignments of full-length proteins, and bootstrap values are based on 10,000 replicates. Dotted lines represent negative branch lengths.
T.S.R. and C.S.P. designed the study and wrote the paper. T.J.G. and G.H. contributed antibodies. J.R.H. generated Pol I, II, and III transgenic lines and NRPD1 and NRPE1 antibodies. A.W. made NRPD1- and NPRE1-FLAG-biotin lines. T.S.R. performed all experiments except LC-MS/MS analyses by C.D.N., A.N., and L.P.-T. at Pacific Northwest National Laboratory (PNNL). J.-K.Z. provided NRPD4/NRPE4 insights. We thank the Washington University green-house staff far plant care and Pikaard lab colleagues for discussions. T.S.R. and C.S.P. also thank Biology 4024 students who helped clone cDNAs: Silvano Ciani and Colin Clune (At2g04630), Andrew Pazandak and Kariline Bringe (At1g54250 and At3gl6980); Caitlin Ramsey and Colin Orr (At5g59180), Wan Shi and Soon Goo Lee (At1g11475), and Lily Momper and Charu Agrawal (At5g51940). Pikaard lab research is supported by National Institutes of Health (NIH) grant GM077590. Any opinions expressed in this paper are those of the authors and do not necessarily reflect the views of the NIH. Portions of this research were supported by the NIH National Center for Research Resources (RR18522), and the W.R. Wiley Environmental Molecular Science Laboratory, a national scientific user facility sponsored by the U.S. Department of Energy’s Office of Biological and Environmental Research and located at PNNL. PNNL is operated by Battelle Memorial Institute for the U.S. Department of Energy under contract DE-AC05-76RL01830.
The Supplemental Data include Supplemental Experimental Procedures, three tables, and 18 figures and can be found with this article online at http://www.molecule.org/supplemental/S1097-2765(08)00858-7.