A partial loss-of-function mutation in an Arabidopsis RNA polymerase III subunit leads to pleiotropic defects

Highlight Analysis of the first known viable RNA polymerase III subunit mutant in Arabidopsis reveals novel roles for this complex in maintaining RNA homeostasis, adjusting the expression of a diverse suite of genes, and indirectly modulating alternative splicing.


Introduction
Transcription under both static and dynamic conditions requires the action of evolutionarily conserved multi-subunit enzymes known as DNA-dependent RNA polymerases (Pols). All eukaryotes possess three distinct RNA polymerases (Pols I, II, and III), each of which transcribes specific suites of genes (Cramer et al., 2008).
Pol I transcribes 45S rRNA, which is the precursor to 5.8S, 18S and 25S rRNAs. Pol II transcribes mRNAs as well as most small nuclear (sn)RNAs and micro (mi)RNAs. Pol III was previously thought to be primarily required for the transcription of 'housekeeping' genes such as those encoding 5S rRNA and tRNAs. However, recent reports indicate that the Pol III transcriptome is more diverse than formerly assumed (Dieci et al., 2007). There are two additional plant-specific RNA polymerases, Pol IV and Pol V, which are required for the biogenesis and functional activity of small interfering (si) RNAs (Haag and Pikaard, 2011). As knockout mutations in the genes encoding the subunits of Pols I, II, and III are lethal, there is a dearth of functional analysis of plant Pols.
While there are a number of published studies examining global transcriptomic changes in plants under various conditions (e.g. Nagano et al., 2012, Woo et al., 2012, the literature to date has largely focused on the roles played by Pol II-transcribed RNAs in regulating plants' responses to stimuli. Stimulus-induced alteration of expression of protein-coding genes has been extensively documented. Numerous recent reports have highlighted the importance of miRNAs in regulating a broad spectrum of biological processes including development (Wu, 2013), flowering time (Spanudakis and Jackson, 2014), drought stress (Ding et al., 2013), metal toxicity (Gupta et al., 2014), immunity (Staiger et al., 2013), and phytohormone crosstalk (Curaba et al., 2014), among others. Furthermore, the biosynthesis, functional mechanisms, and degradation pathways of miRNAs have been well studied (Rogers and Chen, 2013).
Comparatively little is known about Pol III-transcribed RNAs and how they aid plants in responding to intrinsic and extrinsic signals. An RNA molecule with significant sequence and structural similarity to 5S rRNA was found to regulate alternative splicing of certain pre-mRNAs in Arabidopsis (Hammond et al., 2009). Intriguingly, studies in a variety of eukaryotes indicate that Pol III-transcribed non-coding RNAs may play regulatory roles in addition to their housekeeping functions (Hu et al., 2012).
Among the various stimuli to which plants are subjected, biotic stress in the form of pathogenic infection requires that plants be able to respond rapidly and initiate signalling cascades specific to the type of pathogen being encountered. While plants possess physical barriers and broad spectrum resistance that is activated by conserved features of pathogenic microbes, many pathogens are able to deliver infectionpromoting effector molecules into the plant cell, thereby bypassing this layer of plant immunity (Bigeard et al., 2015). However, the plant genome contains a large number of genes encoding nucleotide-binding leucine-rich repeat proteins (NLRs; also referred to as Nod-like receptors due to their structural similarity to mammalian proteins of the same name), which either directly bind to pathogenic effectors or detect their activities within the plant cell in a highly specific manner (Li et al., 2015). Upon recognition of its cognate effector, NLR activation results in rapidly induced and robust defence responses. Plant NLRs can be sorted into two classes based on their N-termini: some possess a Toll-interleukin 1 receptor (TIR) domain and are thus termed TNLs, while others contain a coiled-coil (CC) domain and are referred to as CNLs.
NLR-mediated signalling must be tightly controlled under both resting and induced conditions, as improper signalling through this pathway may lead to either enhanced disease susceptibility or autoimmunity. However, the regulatory mechanisms underlying NLR-mediated signalling are only partially understood. A successful forward genetic suppressor screen previously conducted in our lab used the gain-of-function autoimmune TNL mutant suppressor of npr1, constitutive 1 (snc1; Li et al., 2001;Zhang et al., 2003) to search for positive regulators of immunity (Johnson et al., 2012). More recently, we have undertaken a forward genetic screen to identify negative regulators of NLR-mediated immunity.
Here, we report the characterization of nrpc7-1, a partial loss-of-function allele of the gene encoding the Arabidopsis orthologue of yeast Rpc25, a Pol III subunit. This mutant was isolated from our MUSE (MUTANT, snc1-ENHANCING) forward genetic screen conducted in the modifier of snc1 4 (mos4) snc1 double mutant background. A null mutation in NRPC7 is lethal, while a mutation in an intron-exon splice site junction gives rise to intronic retention in some NRPC7 transcripts, resulting in viable mutant plants. While the nrpc7-1 mos4 snc1 triple mutant displays enhanced resistance against the virulent oomycete pathogen Hyaloperonospora arabidopsidis (H.a.) Noco2, the nrpc7-1 single mutant exhibits wild type-level resistance. This correlates with the altered splicing of SNC1 observed in the triple mutant but not in the single mutant. Morphologically, the nrpc7-1 mutant is dwarf and has serrated leaves, short roots, and stunted siliques, although flowering time does not appear to be affected. The expression and potentially activity of a number of RNAs are distorted in nrpc7-1, contributing to its developmental defects. In keeping with its known function, we observed that the NRPC7 protein localizes to the nucleus. This is the first reported viable Pol III subunit mutant in Arabidopsis.

Plant growth conditions and mutant isolation
Plants were grown either on soil or on half-strength Murashige and Skoog (MS) medium supplemented with 1% sucrose and 0.3% phytagel. All plants were grown under long day conditions (16 h light-8 h dark) at 22 ºC in climate-controlled chambers. The muse4 mutant was isolated from the MUSE screen, described previously (Huang et al., 2013).

Infection assays
H.a. Noco2 infection was performed by spraying 2-week-old soilgrown seedlings with a spore suspension with a concentration of 10 5 spores per millilitre of water. Inoculated seedlings were grown for 7 d at 18 ºC in a growth chamber with ~80% humidity and a 12 h light-12 h dark cycle. Sporulation was then quantified using a haemocytometer to count the number of spores from five plants shaken in 1 ml of water. Five replicates were performed for each of three independent trials. P.s.m. ES4326 infection was performed by infiltrating the abaxial leaf surface of 4-week-old soil-grown seedlings with bacteria suspended in 10 mM MgCl 2 (OD 600 =0.0005). Leaf punches were collected at day 0 and day 3, and serial dilutions were performed and plated on LB medium. Plates were incubated at 28 ºC for 24 h before colony forming units were measured.

Positional cloning and Illumina whole-genome sequencing
Positional cloning of muse4 was performed by crossing the muse4 mos4 snc1 triple mutant (generated in the Col-0 ecotype) with wild type Landsberg erecta. Twenty-four F2 plants homozygous for all three mutations were used for crude mapping, and approximately 500 F3 plants homozygous for mos4 and snc1 and heterozygous for muse4 were used for fine mapping. The markers used in mapping were derived from insertion/deletion polymorphisms between the Col-0 and Ler Arabidopsis ecotypes (Jander et al., 2002; http:// www.arabidopsis.org). After determining that the mutation must be located on the top of chromosome 1 between 1.4 MB and 2.75 MB, extracted genomic DNA from muse4 mos4 snc1 was sequenced using the Illumina sequencing platform.

Preparation of transgenic plants and confocal microscopy
Full length At1g06790 genomic DNA, including 766 bp upstream of the start codon, was amplified via PCR, cloned into the pCam-bia1305 vector, and transformed into muse4 mos4 snc1 using the floral dip method (Clough and Bent, 1998). The full length genomic fragment was also cloned into a pCambia1305 vector containing a green fluorescent protein (GFP) tag. Transgenic plants were selected for on ½ MS plates containing 50 mg ml -1 hygromycin. Confocal images of wild type (negative control), 35S::X-GFP (positive control), and NRPC7-GFP transgenic seedlings were obtained using a PerkinElmer Ultraview VoX spinning disc confocal system (PerkinElmer) mounted on a Leica DM16 000 B inverted microscope and equipped with a Hamamatsu 9100-02 electron multiplier CCD camera (Hamamatsu). An argon 488 nm laser line with a complementary (522/36) emission band-pass filter to detect GFP or a 561 nm laser with a complementary (595/50) emission band-pass filter to detect propidium iodide was used. Images were acquired with a ×63 (water) objective lens. To stain the nuclei and the cell wall, seedlings were incubated in a 10 µg ml -1 solution of propidium iodide (Calbiochem) for 1 min, rinsed with water, and mounted on a slide and coverslip prior to imaging.

Small RNA library construction and sequencing
Small RNAs within the size range of 15-40 nt were fractionated from total RNAs by 15% polyacrylamide gel electrophoresis. These small RNAs were then ligated with the 3′ and 5′ adapters sequentially using the Small RNA Sample Preparation Kit (Illumina) according to the manufacturer's instructions. A reverse transcription reaction followed by a low cycle PCR was performed to obtain final products for deep sequencing. The wild type and muse4 libraries were barcoded and sequenced in one channel on an Illumina Hiseq2000.
Analysis of small RNA high throughput sequencing Data PERL scripts were used to process small RNA raw reads as per Lertpanyasampatha et al. (2012). To summarize, reads were passed through Illumina's quality control filter before being sorted into bins based on their barcodes and having their adaptor sequences removed. SOAP2 was used to map reads within the size range of 20-24 nt to the Tair10 Arabidopsis genome ). Differential small RNA regions were identified as previously described (Dinh et al., 2014). For analysis of differentially expressed miRNAs, all known Arabidopsis miRNAs were downloaded from miRBase (Release 20 from www.mirbase.org; Griffiths-Jones et al., 2008). PERL scripts were used to determine the expression level of known miRNAs in the small RNA libraries and then normalize these counts to reads per million (RPMs). miRNAs with <10 RPMs in both nrpc7-1 and wild type libraries were removed. The differentially expressed miRNAs were identified by comparing expression in the nrpc7-1 library with wild type. The Audic-Claverie method was used to calculate P-values (Audic and Claverie, 1997), which were subsequently adjusted as described by Benjamini and Hochberg (1995) to determine the false discovery rate (FDR). To qualify as a differentially expressed miRNA, both a fold change ≥2 between wild type and nrpc7-1 and an FDR <0.05 were necessary.

Results
The isolation, characterization, and identification of the muse4/nrpc7-1 mutant The MUSE screen was designed to identify enhancers of the dwarf autoimmune mutant snc1 and has been described previously (Huang et al., 2013). To avoid potential lethality resulting from dramatically enhanced autoimmunity, the snc1 suppressor mos4 was included in the genetic background of the screen. Seeds from the wild type-like mos4 snc1 plants were mutagenized with ethyl methanesulfonate, and the M2 population was screened for plants displaying a reversion to snc1-like morphology and resistance. A number of mutant lines were isolated, one of which (muse4) was selected for further characterization. When the triple mutant was backcrossed to mos4 snc1, all progeny appeared wild type-like, indicating that muse4 is a recessive mutation.
As shown in Fig. 1A, the muse4 mos4 snc1 plants exhibit dwarf, curled leaf morphology similar to that observed for snc1 plants. In addition, the muse4 mos4 snc1 plants have serrated and slightly chlorotic leaves. The muse4 mutation also re-establishes the constitutive expression of the defence marker PATHOGENESIS-RELATED (PR) genes observed in snc1 but absent in mos4 snc1. A pPR2-GUS reporter gene construct was used to visualize PR2 gene expression in seedlings, and GUS staining was much stronger in the triple mutant than in mos4 snc1 (Fig. 1B). Consistent with this observation, qPCR demonstrated that expression of PR1 and PR2 is elevated in the triple mutant (Fig. 1C).
To examine whether the muse4 mutation alters resistance to the virulent oomycete strain Hyaloperonospora arabidopsidis Noco2, 2-week-old triple mutant seedlings were spray-inoculated with this pathogen. The enhanced resistance observed in snc1 but lost in mos4 snc1 was found to be reconstituted in the triple mutant (Fig. 1D). Together, these data indicate that muse4 restores all examined snc1-like phenotypes in the mos4 snc1 background.
A positional cloning strategy was employed to determine the molecular lesion responsible for the observed phenotypes.
The muse4 mos4 snc1 mutant, which was generated in the Col-0 ecotype, was crossed to Landsberg erecta (Ler). From the F2 population, 24 plants displaying the triple mutant morphology were selected for crude mapping, which identified a linkage to the top of chromosome 1. Several F2 plants heterozygous at the top of chromosome 1 (but homozygous for snc1 and mos4 to prevent interference by these loci) were used to generate a fine mapping population of approximately 500 plants. The mutation was narrowed down to between the markers T7A14 (1.4 MB) and F22O13 (2.75 MB). Genomic DNA was extracted from muse4 mos4 snc1 triple mutant plants and sequenced using the Illumina whole-genome sequencing platform. The sequencing results were compared with the Arabidopsis reference genome, and five genes in this region were found to contain mutations ( Fig. 2A). The mutations in three of these genes are located in introns and the mutation in one gene was found to be silent, therefore the mutation in the remaining gene (At1g06790) was selected as the most likely candidate for muse4. This gene encodes the Arabidopsis orthologue of the yeast Pol III subunit Rpc25, NUCLEAR RNA POLYMERASE C, SUBUNIT 7 (NRPC7; Ream et al., 2015), and the muse4 mutation is at the intron-exon junction just before the sixth exon (Fig. 2B).

The mutation at an intron-exon junction of NRPC7 results in intron retention and is responsible for the muse4 phenotypes
To verify that the mutation in NRPC7 is responsible for the muse4 phenotypes, a full-length wild type copy of the gene driven by its native promoter and fused to the GFP-encoding gene at its 3′ end was transformed into the single mutant, which was generated by backcrossing the triple mutant to Col-0 and selecting plants homozygous for wild type SNC1 and MOS4 that retained the serrated leaf phenotype and dwarf size. Eight independent T2 lines displayed wild type morphology, and one representative line can be seen in Fig. 3A. These data suggest that NRPC7 can fully complement the muse4 phenotypes, and therefore that MUSE4 is indeed NRPC7. We hypothesized that the muse4 mutation in the intronexon junction of NRPC7 results in retention of the preceding intron. To test this, we designed primers flanking the intron of interest and amplified cDNA from wild type and muse4. A strong band of the expected size (465 bp) was observed in wild type while in muse4 two bands were observed, one of the expected size and one slightly larger (Fig. 3B). The larger band was excised from the gel and the PCR product was purified and sequenced. As predicted, sequencing revealed that the larger band corresponded to a transcript in which the intron preceding the muse4 mutation had been retained (Fig. 3C).
Despite strong sequence similarity between NRPC7 and known Rpc25 proteins in other species ( Supplementary  Fig. S1 at JXB online), the NRPC7 gene failed to complement a temperature-sensitive rpc25 yeast knockout line ( Supplementary Fig. S2), suggesting divergence between the plant and yeast NRPC7.
It is expected that a knockout mutation in NRPC7 would be embryo lethal, as a previous study showed that loss-of-function mutations in RNA polymerase subunits are not transmitted maternally (Onodera et al. 2008). Indeed, when we let the heterozygous nrpc7 T-DNA insertion line CS1001213 self-fertilize and then planted the progeny, we identified 23 wild type plants lacking the insertion, 46 heterozygotes, and 0 plants that were homozygous for the insertion, matching the expected 1:2:0 (wild type:heterozygote:homozygote) ratio for a lethal mutation. We also performed reciprocal crosses between this heterozygous T-DNA insertion line and muse4 and found that none of the F1 progeny contained the T-DNA insertion, indicating that the T-DNA/muse4 heterozygotes are not viable. These results, combined with the data in Fig. 3B showing that muse4 still produces some properly spliced transcripts without intron retention, as well as the fact that muse4 is a recessive mutation, suggest that muse4 is a partial loss-offunction allele of NRPC7. Therefore, we renamed muse4 as nrpc7-1 and the T-DNA allele as nrpc7-2.
Splicing of SNC1 is altered in the nrpc7-1 mos4 snc1 background In yeast, Rpc25 is required for Pol III transcription initiation (Zaros and Thuriaux, 2005). To assess whether Pol III function is affected by the nrpc7-1 mutation, we used real-time qPCR to determine whether expression of U6, a snRNA component of the spliceosome that is known to be transcribed by Pol III (Waibel and Filipowicz, 1990), is different in nrpc7-1 than in wild type. Relative to the expression levels of the Pol II-transcribed 'housekeeping' gene UBQ5, U6 expression is significantly lower in nrpc7-1 (Fig. 4A). To examine whether the nrpc7-1 mutation has a general effect on spliceosomal snRNA biosynthesis, the accumulations of Pol II-transcribed U1 and U2 snRNAs were also examined. While U1 accumulation is wild type-like, U2 expression is significantly reduced in nrpc7-1. This is likely due to an indirect effect of altered Pol III function on Pol II-transcribed genes.
The reduced expression of the spliceosome components U6 and U2 leads us to hypothesize that pre-mRNA splicing might be affected by the nrpc7-1 mutation. Specifically, as nrpc7-1 was isolated in our screen for snc1 enhancers, we hypothesized that the mutation may affect the excision of introns from the SNC1 pre-mRNA transcript. The alternative splicing of a number of plant NLR-encoding genes, including SNC1, is known to affect their function in plant immunity (Yi and Richards, 2007;Xu et al., 2011). For SNC1, the second and third introns may be either retained or removed; therefore we used primers spanning these two introns to amplify the SNC1 transcript variants. A dramatic accumulation of the largest transcript variant (with both introns retained) was observed in nrpc7-1 mos4 snc1, although the SNC1 splicing pattern in the nrpc7-1 single mutant was indistinguishable from that observed in wild type ( Fig. 4B and Supplementary Fig S3C). Alternative splicing defects were also observed in nrpc7-1 mos4 snc1 for RESISTANT TO PSEUDOMONAS SYRINGAE 4 (RPS4), another NLR-encoding gene, and to a lesser degree for SR30, which encodes a serine/arginine-rich RNA-binding protein and is known to be alternatively spliced (Fig. 4B). The relative proportions of the transcript variants in the genotypes examined are shown in Fig. 4C. These data reveal significant alternative splicing defects caused by the Pol III subunit mutation.
To determine whether this splicing defect occurs at the level of basal splicing, transcripts of a gene that is not alternatively spliced (PHYTOALEXIN-DEFICIENT 4; PAD4) were also examined. No difference from the wild type splicing pattern was detected (Fig. 4B). Both snc1 and nprc7-1 mos4 snc1 accumulated higher levels of PAD4 compared with wild type, which is consistent with previous reports that PAD4 is a defence-induced gene (Glazebrook 2001), whose expression is expected to be upregulated in autoimmune mutants.
NLR-mediated signalling is often regulated by modulating transcription and/or translation of NLRs. As such, we examined whether SNC1 expression and protein accumulation are affected by the nrpc7-1 mutation. SNC1 expression was found to be slightly reduced in nrpc7-1 as compared with wild type, while SNC1 protein levels were wild type-like ( Supplementary  Fig. S3). Similarly, the accumulation of SNC1 in nrpc7-1 mos4 snc1 was not dramatically higher than that observed in mos4 snc1. Taken together, these data indicate that SNC1 alternative splicing, but not overall gene expression or translation, is affected by the nrpc7-1 mutation.

The nrpc7-1 single mutant does not have altered immune responses
Since alterations in the splicing of SNC1 were observed in the nrpc7-1 mos4 snc1 triple mutant but not the nrpc7-1 Fig. 3. MUSE4 is NRPC7. (A) MUSE4 tagged with GFP and expressed under the control of its native promoter is able to complement the muse4 single mutant defects. Plants were grown on soil for 3 weeks. (B) The size of the MUSE4 transcript in wild type and muse4 was examined using cDNA reversetranscribed from total RNA. (C) The larger muse4 band in (B) was excised, purified, and sequenced, and found to retain the intron preceding the intronexon splice site mutation in muse4.
single mutant, we predicted that nrpc7-1 may not exhibit the enhanced disease resistance observed in the triple mutant (Fig. 1D). We challenged nrpc7-1 with the oomycete pathogen H.a. Noco2 and the bacterial pathogen Pseudomonas syringae pv. maculicola ES4326, and found no statistically significant difference in response from that observed in wild type plants ( Supplementary Fig. S4). These data support our hypothesis that the retention of introns in the SNC1 transcript in the nrpc7-1 mos4 snc1 confers enhanced disease resistance, and is likely the reason why this mutation was isolated from our snc1 enhancer screen.

nrpc7-1 has global defects in RNA levels
Pol III transcribes tRNA, 5S rRNA, and assorted other non-coding RNAs. To further explore how Pol III function is affected by the nrpc7-1 mutation, we examined the expression of a variety of RNAs by qPCR (Fig. 5A). Relative to UBQ5, 5S rRNA and three representative tRNAs (coding for Gln, Gly, and Leu, respectively) showed significantly reduced accumulation in nrpc7-1 compared with wild type. In addition, when total RNA was run on a 2% agarose gel, altered relative proportions of the various rRNAs were consistently observable in association with the nrpc7-1 allele (Fig. 5B). Relative to Pol I-transcribed 25S rRNA, there appears to be a lower abundance of chloroplast 16S and 23S rRNA associated with the nrpc7-1 allele. These data suggest that in addition to Pol III transcribed genes, the nrpc7-1 mutation also affects abundance of other RNAs, likely through indirect mechanisms.
Small RNA libraries were then prepared from BG2 plants (Col-0 with the pPR2-GUS reporter gene construct that is present in the nrpc7-1 background) and two independently isolated nrpc7-1 single mutant lines. Analysis of these small RNA libraries indicated that a number of miRNAs are differentially expressed in the mutant. Those miRNAs that exhibited a two-fold or greater change in expression are shown in Supplementary Fig. S5A. To validate these results, three representative miRNAs were selected for northern blot analysis. Although the data from the small RNA libraries indicated that expression of both miR159 and miR166 is reduced in nrpc7-1 while miR398 expression is increased, no significant alterations in the levels of these miRNAs were consistently observed via northern blotting ( Supplementary Fig. S5B), suggesting that any differences that exist between the mutant and wild type are too subtle to be detected by this method.
The general disruption in RNA equilibrium combined with the striking serrated leaf phenotype and dwarf morphology of the nrpc7-1 mutant led us to hypothesize that the expression of (i) the CUP-SHAPED COTYLEDONS (CUC) genes, which are targeted by miR164 (Mallory et al., 2004), and (ii) the HOMEODOMAIN-LEUCINE ZIPPER (HD-ZIP) genes, which are targeted by miR165/166 (Rhoades et al., 2002), might be altered in the mutant. Expression of a CUC2 transcript resistant to miR164 cleavage was previously shown to result in enhanced leaf serration; the same morphological phenotype was observed in plants containing loss-of-function mutations in the Pol II-transcribed MIR164 (Nikovics et al., 2006).
Overexpression of either miR165 or miR166 results in reduced expression of the HD-ZIP genes, which corresponds with dwarf morphology and altered rosette leaf morphology that is similar to that observed in nrpc7-1 (Jung and Park, 2007). Real-time qPCR was used to determine that CUC1 and CUC2 accumulation is elevated in nrpc7-1 (Fig. 5C), although no alteration in miR164 levels were observed in nrpc7-1 based on our small RNA library data. Expression of the HD-ZIP genes PHABULOSA (PHB) and REVULOTA (REV) was found to be decreased in nrpc7-1 (Fig. 5C). No changes in miR165 levels were observed in nrpc7-1, and although miR166 expression was elevated in the mutant according to the small RNA library sequencing data ( Supplementary Fig.  S5A), no detectable change in miR166 levels was consistently measurable by northern blotting (Supplementary Fig. S5B). The altered expressions of CUC1, CUC2, PHB, and REV in the absence of detectable changes in miR164, miR165, and miR166 abundances suggests that in nrpc7-1 the activity of a number of small RNAs may be affected; alternatively, the nrpc7-1 mutation may indirectly affect Pol II-mediated transcription of certain genes, including CUC1, CUC2, PHB, and REV, although the mechanism behind this specificity is unclear. Many of the RNAs that seem to be differentially expressed in nrpc7-1 are not transcribed by Pol III (Fig. 5), suggesting that this mutation results in a disruption of the global RNA equilibrium and homeostasis.

NRPC7 localizes to the nucleus
As part of the Pol III complex, NRPC7 is predicted to localize to the nucleus. To examine its localization, we used the complementing nrpc7-1 lines containing the transgene with NRPC7 fused to GFP under the control of the native promoter, described above. We analyzed cotyledon and root tissue using confocal microscopy, and GFP fluorescence was visible throughout the nucleus as it co-localized with nuclei stained with propidium iodide (Fig. 6). Additionally, there appeared to be intense fluorescent foci within the nucleus and along the plasma membrane.  The proportions of rRNAs in the noted genotypes were compared by running total RNA extracted from seedlings grown on MS medium for 12 d on a 2% agarose gel. Band intensities were quantified using ImageJ, and the intensities of chloroplast rRNA (16S and 23S) relative to Pol I-transcribed rRNA (25S) were compared between genotypes for three biological replicates ±SD. (C) The accumulation of CUC1, CUC2, PHB, and REV transcripts was examined using quantitative real-time qPCR. Bars represent the averages of three technical replicates of two biological replicates ±SD. *P≤0.05; **P≤0.01; ***P≤0.001; ****P≤0.0001.

RFP (PI)
NRPC7-GFP Merged GFP n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n n nrpc7-1 has pleiotropic developmental defects The roles various small RNAs play in the regulation of plant development have been well studied. As nrpc7-1 has large impacts on small RNA levels and, potentially, RNA activities, we examined the developmental phenotypes of the mutant. As described earlier, nrpc7-1 has serrated leaves (Figs 3A and 7A), and its growth is stunted (Fig. 7B). When grown on halfstrength MS medium, nrpc7-1 plants also have significantly shorter roots than wild type plants (Fig. 7C). The siliques of nrpc7-1 were consistently found to be smaller (Fig. 7D).
Flowering time was measured using several different assays, but a significant difference between nrpc7-1 and wild type was only observed when measuring the number of days until the primary stalk reached 6 cm (Fig. 7E), which is likely a reflection of the restricted growth of the mutant rather than an actual delay in flowering time. While the number and arrangement of the floral organs are wild type-like, the texture of the sepals is bumpy and irregular (Fig. 7F). These results show that the nrpc7-1 mutation is associated with a number of pleiotropic developmental defects.

Discussion
We demonstrated that Pol III function is altered by the partial loss-of-function mutation nrpc7-1 by showing that the expressions of Pol III-transcribed U6 snRNA, 5S rRNA, and a number of tRNAs are reduced in the mutant (Figs 4A and 5A). Pol II-transcribed U2 snRNA, but not U1 snRNA, also had reduced expression in nrpc7-1 (Fig. 4A), indicating that the transcriptional defects in the mutant extend to genes not directly transcribed by Pol III. The decreased accumulation of U6 and U2 snRNA led us to hypothesize that spliceosome functionality is impaired in nrpc7-1 and that alternative splicing of SNC1 is consequently affected, thereby explaining why this mutation was isolated from a screen for enhancers of the autoimmune mutant snc1. Indeed, SNC1 splicing is defective in the nrpc7-1 mos4 snc1 triple mutant background, in that there is a dramatic accretion of a transcript variant that retains both the second and third introns (Fig. 4B, C). A similar pattern was observed for the NLR-encoding gene RPS4.
The second intron of SNC1 contains an in-frame premature stop codon, and thus retention of this intron should yield a truncated version of the protein. Previous reports have shown that an accumulation of the N-termini of TNL proteins is sufficient to activate cell death and immunity (Weaver et al., 2006;Swiderski et al., 2009). This finding, combined with our data showing that transcription and translation of SNC1 are not enhanced by the nrpc7-1 mutation ( Supplementary  Fig. S3A, B), suggests that the modification in SNC1 splicing could be the primary cause of the snc1 enhancing effects of the nrpc7-1 mutation in the mos4 snc1 background (Fig. 1).
It is notable that the nrpc7-1 single mutant does not differ from wild type in either alternative splicing (Fig. 4B, C) or disease resistance (Supplementary Fig. S4). This suggests that the mos4 mutation is required in the genetic background for the nrpc7-1-associated splicing defects to become obvious. MOS4 is an integral component of the evolutionarily conserved MOS4associated complex (MAC) that functions together with the spliceosome to regulate pre-mRNA splicing (Johnson et al., 2011). Mutations in mos4 and other MAC components have previously been shown to affect the alternative splicing of both SNC1 and RPS4 (Xu et al., 2012), and are associated with a suppression of SNC1-dependent immune signalling (Palma et al., 2007). In this study we demonstrated that there is an increased accumulation of the intron-retaining SNC1 transcripts in mos4 snc1 compared with snc1 (Fig. 4C). However, this accumulation is radically enhanced in the nrpc7-1 triple mutant. One possible explanation for these results is that the mos4 mutation and, to a lesser extent, the nrpc7-1 mutation individually disrupt splicing efficiency, reducing the pool of the functional full-length SNC1 transcript variant with both introns excised but not increasing the production of alternative variants beyond the threshold required for immune activation. However, when these two mutations are combined in the snc1 background, spliceosome activity is markedly disturbed and the accumulation of intron-retaining SNC1 transcript variants is sufficiently high to yield enough truncated SNC1 to activate defence responses.
In addition to the splicing defects observed in nrpc7-1, the accumulations of rRNAs and tRNAs appear to be considerably distorted (Fig. 5A, B). Ribosomal protein gene dosage was recently found to have an effect on embryonic stem cell differentiation in mice (Fortier et al., 2015), indicating that alterations in the abundance of ribosome components can dramatically alter developmental progression. Homozygous nrpc7-1 plants exhibit certain phenotypes that may be associated with impaired stem cell differentiation including short roots (Fig. 7C) and delayed emergence of the first true leaves. This suggests that the sensitivity of ribosome function to changes in its subunit levels, as well as its role in regulating stem cell differentiation, may be conserved in plants, although the data in support of this are preliminary and additional experiments are required to fully explore this hypothesis.
We also detected a reduction in the accumulation of the chloroplast 16S and 23S rRNAs relative to Pol I-transcribed 25S rRNA (Fig. 5B). A similar rRNA abundance pattern was recently reported for atybeY-1, a mutant allele of an endoribonuclease required for chloroplast rRNA processing and development that also exhibits pale green leaves and delayed development (Liu et al., 2015). Although the mechanism by which alterations in a Pol III subunit result in changes to the transcriptional regulation of Pol I-transcribed genes and the chloroplast genome is unclear, the light green colour of the nrpc7-1 single mutant (Fig. 3A) further suggests that this mutation may be associated with impaired chloroplast function.
The serrated leaf phenotype observed in nrpc7-1 is likely linked to its elevated expression of CUC1 and CUC2 (Fig. 5C), which could be a result of reduced miR164 activity in the mutant background or an indirect effect of the nrpc7-1 mutation on Pol II function. There may also be a link between the nrpc7-1 mutant morphology and the decreased expression of the HD-ZIP genes (Fig. 5C). Other studies have demonstrated tentative links between the transcriptional activities of Pol II and Pol III. One study identified areas of the genome where protein-coding genes on one DNA strand overlapped with tRNA-coding genes on the opposite strand, and that their rates of transcription by Pol II and Pol III, respectively, were negatively correlated (Lukoszek et al., 2013). Another study found that human RPPH1 is transcribed by both Pol II and Pol III, and identified a number of transcriptional activators that associate with both Pols (Faresse et al., 2012). Our data show that a disturbance in Pol III function affects the expression of non-Pol III-transcribed RNAs, indicating that the role of Arabidopsis Pol III in transcriptional regulation is more complex than previously assumed.
There are 12 core subunits of Arabidopsis Pol III, each of which has a homologue or is itself also a component in Pols I, II, IV, and V (Haag and Pikaard, 2011;Ream et al., 2015). There are also subunits specific to individual Pols. NRPC7 encodes a core Pol III subunit with homologues in each of the other Pols, and shares significant sequence similarity with Rpc25 proteins from other model organisms ( Supplementary Fig. S1). However, Arabidopsis NRPC7 failed to complement a temperature-sensitive rpc25 yeast knockout line ( Supplementary Fig.  S2), suggesting that the functional conservation of this protein by itself between yeast and plants is limited. This is not entirely unprecedented. Rpc25 is known to form a dimer with Rpc17 within the Pol III complex (Siaut et al., 2003). The proteinprotein interaction surface of NRPC7 may be sufficiently evolutionarily divergent so as to prohibit it from dimerizing with yeast Rpc17. Although the function of the protein complex is conserved, an individual component of the complex may still be divergent enough that it fails to complement a knockout of its orthologue in a distant organism.
In summary, we have demonstrated that a perturbation in Pol III function results in modified gene splicing as well as alterations in the abundances and potentially activities of a number of RNA molecules. These effects extend to several RNAs reported to be transcribed by other polymerases, revealing a novel role for Pol III in modulating the expression of a larger complement of genes than previously described. Moving forward, the partial loss-of-function nrpc7-1 mutant provides a unique tool for performing other functional analyses of Pol III.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Sequence alignment of RPC25 from a broad range of species, based on BLAST analysis. Figure S2. Yeast complementation with NRPC7. Figure S3. SNC1 gene and protein expression in nrpc7. Figure S4. Immune characterization of nrpc7-1 single mutant plants. Figure S5. RNA defects in nrpc7-1.