Functional characterization of Polr3a hypomyelinating leukodystrophy mutations in the S. cerevisiae homolog, RPC160

Mutations in RNA polymerase III (Pol III) cause hypomeylinating leukodystrophy (HLD) and neurodegeneration in humans. POLR3A and POLR3B, the two largest Pol III subunits, together form the catalytic center and carry the majority of disease alleles. Disease-causing mutations include invariant and highly conserved residues that are predicted to negatively affect Pol III activity and decrease transcriptional output. A subset of HLD missense mutations in POLR3A cluster in the pore region that provides nucleotide access to the Pol III active site. These mutations were engineered at the corresponding positions in the Saccharomyces cerevisiae homolog, Rpc160, to evaluate their functional deficits. None of the mutations caused a growth or transcription phenotype in yeast. Each mutation was combined with a frequently occurring pore mutation, POLR3A G672E, which was also wild-type for growth and transcription. The double mutants showed a spectrum of phenotypes from wild-type to lethal, with only the least fit combinations showing an effect on Pol III transcription. In one slow-growing temperature-sensitive mutant the steady-state level of tRNAs was unaffected, however global tRNA synthesis was compromised, as was the synthesis of RPR1 and SNR52 RNAs. Affinity-purified mutant Pol III was broadly defective in both factor-independent and factor-dependent transcription in vitro across genes that represent the yeast Pol III transcriptome. Thus, the robustness of yeast to Pol III leukodystrophy mutations in RPC160 can be overcome by a combinatorial strategy.


Introduction
Transcription by RNA polymerase III (Pol III) is an essential process that generates stable non-coding RNAs involved in protein synthesis, RNA processing, protein secretion and other processes (Dieci et al., 2007). The majority of Pol III-transcribed genes encode tRNAs, the amino acid adaptors that decode mRNA sequences during protein synthesis. The Pol III transcription machinery comprises essential multi-subunit complexes that include the 17subunit polymerase and a three-subunit initiation factor, TFIIIB. Biallelic missense mutations in either Polr3a or Polr3b (which encode the two largest subunits of Pol III), Polr3k, or Polr1c (a shared subunit of Pols I and III) cause childhood-onset hypomyelinating leukodystrophy (HLD) (Wolf et al., 2014;Gutierrez et al., 2015;Thiffault et al., 2015;Dorboz et al., 2018). Pol III HLD mutations are anticipated to alter Pol III transcription throughout the body but the primary phenotypic effect is on the central nervous system. The major feature of this early-onset autosomal recessive disease is hypomyelination of cerebral white matter. Missense mutations in the TFIIIB subunits, BRF1 and BDP1, have also been linked to neurological disease, causing cerebellar hypoplasia and cognitive dysfunction (Borck et al., 2015) and early-onset severe hearing loss, respectively (Girotto et al., 2013;Girotto et al., 2014). Disease-causing mutations in Pol III and initiation factor subunits include invariant and highly conserved residues that are predicted to negatively affect Pol III activity and decrease transcriptional output (Trinh et al., 2006;Bernard et al., 2011).
The mechanisms of pathogenesis of Pol III HLD disease mutations are unknown, but decreased Pol III transcription is anticipated to alter the protein synthetic capacity of cells.
Studies of the Pol III transcriptome in HLD patient-derived samples report varied dysregulation of Pol III gene expression and gene-specific sensitivity to POLR3 mutations (Dorboz et al., 2018;Choquet et al., 2019a). Eight patient-derived fibroblast lines, representing homozygous and compound heterozygous Polr3a HLD genotypes showed a common decrease in BC200 RNA, a brain-specific RNA linked to repression of translation (Choquet et al., 2019a). Two fibroblast lines derived from homozygous mutant Polr3k HLD patient samples showed lower levels of 5S rRNA, 7SL RNA and tRNAiMet levels (Dorboz et al., 2018).

Splicing mutations in
Polr3a are linked to a striatal variant of POLR3A-related disease that is clinically distinct from HLD: An analysis of blood samples from three such patients showed a reduction across all tRNAs, negative effects on the steady-state abundance of a subset of other Pol III transcripts, and elevated 5S rRNA, and RppH1 and 7SK RNAs (Azmanov et al., 2016). The Pol II transcriptome was also altered in these samples, with a reduction of chaperone protein expression and enrichment in snoRNA maturation proteins.
Experimental models of Pol III HLD report varied defects in Pol III gene expression.
Expression of a POLR3A M852V disease mutation in a human cell line resulted in a global decrease in precursor tRNA levels and lower steady state levels of BC200 RNA and 7SL RNA.
Other Pol III-transcribed genes were unaffected, as were the levels of mature tRNAs (Choquet et al., 2019a). Expression of the M852V mutation in an oligodendroglial cell line, that can be induced to differentiate from preoligodendocyte-like cells to cells that express myelin basic protein (MBP), resulted in decreased BC200 RNA and MBP mRNA levels. While the proteome of these cells showed both positive and negative effects on a limited number of proteins, there was no global effect on protein abundance (Choquet et al., 2019a). Expression of two POLR3A disease mutations in the S.pombe homolog of POLR3A resulted in a decrease in several precursor tRNAs, elevated m 2 2 G26 modification of a subset of tRNAs and enhanced tRNA decoding of a reporter gene, thought to be due to the increased stability of the mature tRNA . This change in tRNA modification was induced in human cells by a nutrient stress-mediated shutdown of transcription, indicating that the tRNA modification enzyme Trm1 is broadly limiting for function. These reports indicate that Pol III disease mutations in patient-derived samples and in experimental models compromise the synthesis of a subset of the Pol III transcriptome and can have specific effects on protein output.
Pol III HLD remains poorly characterized at the cellular and molecular level. Mouse models of Pol III-related HLD have proven difficult to generate. Mice homozygous for the most frequent Pol III-associated HLD missense mutation, G672E in POLR3A, did not recapitulate human disease phenotypes either as pups or at advanced age. Moreover, no defect in Pol III transcription was detected in the mice or a knock-in human cell culture model (Choquet et al., 2017). Mice in which the G672E mutation was combined with a Polr3a null allele also did not exhibit neurological abnormalities or transcription phenotypes (Choquet et al., 2017). In contrast, a POLR3B R103H mutation that impairs the assembly of Pol III when ectopically expressed in human cell lines, was viable as a heterozygote knock-in in the mouse but embryonically lethal as a homozygote (Choquet et al., 2019b). A double-mutant mouse, with a Polr3a G672E/null Polr3b +/R103 genotype, also did not exhibit neurological or transcription phenotypes (Choquet et al., 2019b). Together, these phenotypes suggest that genetic robustness limits the effect of Pol III mutations and/or deficits in Pol III function in the mouse. Alternatively, genetic modifiers and/or environmental stressors in human development may contribute to HLD disease presentation and depth of phenotype (Choquet et al., 2017).
The phenotypic disparity between the human disease and mouse models of Pol IIIassociated HLD led us to screen a panel of POLR3A HLD mutations, which map to the pore region where multiple mutations are clustered, and assess their functional deficits in yeast Pol III transcription. Each POLR3A HLD-associated mutation was engineered individually or in combination with the frequently occurring pore mutation (G672E in humans) into the S.cerevisiae homolog, RPC160 (G686E, yeast amino acid numbering), and assayed for growth and transcription phenotypes. All HLD-associated single mutations were indistinguishable from wild-type in growth and transcription and all had wild-type steady state levels of 5S rRNA and mature tRNAs. Double mutations showed a spectrum of growth phenotypes as well as lethality.
Characterization of one double mutant, with Y685K and G686E mutations, showed defective tRNA synthesis and low steady-state levels of precursor and mature forms of RPR1 and SNR52 RNAs in vivo. In vitro analysis of affinity-purified mutant Pol III showed it to be defective in factor-independent transcription and in factor-dependent transcription across genes that represent the yeast Pol III transcriptome.

Mutagenesis and subcloning
Rpc160 containing a C-terminal triple hemagglutinin (HA) tag was cloned from a genomically-tagged strain (Lee et al., 2015) into pRS315 and pRS316. Mutations in the pRS315-Rpc160 plasmid were generated using the Gibson Assembly Cloning Kit (NEB) and clones were confirmed by sequencing the entire coding region. Chromosomal deletion of RPC160 in strain W303 pRS316-RPC160 used pFA6a-HphNT1 and standard PCR-based methodology (Janke et al., 2004). RPC160 mutants were subcloned into pRS405, resequenced and the plasmid linearized with PpuM1 for integration at the LEU2 locus. Colonies resistant to 5fluoroorotic acid (FOA) were identified and integration of the mutant allele was confirmed by PCR and sequencing. For multicopy expression of RPR1, 160 base pairs upstream and 199 base pairs downstream of the RPR1 coding region was cloned into pRS423. The wild-type SSD1 gene in pRS316 included 400 base pairs upstream and downstream of the coding sequence and was obtained from Ted Powers (pPL092).

RNA analysis
Yeast strains, listed in Table S1, were derived from W303 and grown in YPD or synthetic complete drop-out media as required. Yeast spot images were generated using 10-fold serial dilutions of freshly grown cultures diluted to OD 600 0.5. Doubling times of yeast strains were determined by real time monitoring in a Bioscreen C instrument and represent the average of technical triplicates and multiple biological replicates for each strain (Moir et al., 2012). Cultures 6 for RNA analysis were grown to OD 600 0.4-0.5 at 30° before RNA extraction. Hot-phenol extracted total RNA (10 μg/lane) was processed, hybridized, imaged and quantified as described in (Li et al., 2000a). Probe signal intensities were normalized to that of SNR17a/b RNA (U3). Oligo probe sequences are listed in Table S2. For uracil pulse-labeling, cells (10 OD 600 ) grown in synthetic complete media without uracil, were washed and resuspended in 900 µl warm media that contained 100 μCi 5,6-3 H-uracil, incubated at 30° for 10 mins and processed immediately for RNA. After electrophoresis, gels were soaked in Amplify Fluorographic Reagent (GE Healthcare) for 1 hour at room temperature, washed in water, dried, visualized by autoradiography to film at −80° and quantified as for Northern analysis.

Proteins
Wild-type and KE mutant strains were transformed with pNZ16 (pRS314 that carries RPC128 with six N-terminal histidine residues followed by a FLAG tag, described in . Chromosomal deletion of RPC128 was achieved using pFA6a-natNT2 and standard PCR-based methodology (Janke et al., 2004). Cells were batch grown in SC medium lacking leucine, uracil and tryptophan to OD 4.0 and Pol III was affinity-purified . Pol III peaks were identified by Western blot for HA, pooled, dialyzed, concentrated and stored at -80°C. The Pol III preparations had comparable total protein yield and were normalized by the level of Flag-tagged C128 and C34 proteins detected by Western blotting.
The Tfc8TAP strain from the yeast TAP-tagged ORF library was engineered to overexpress Brf1 (approximately 50-fold) by substitution of the natNT2 TEF1 promoter cassette (Janke et al., 2004) for the endogenous BRF1 promoter. Whole cell extracts were prepared and chromatographed by size exclusion chromatography (Sephadex 200), cation exchange (Biorex 70, BRα step elution) and anion exchange (DEAE-Sephadex, step elution) to generate TFIIIB and TFIIIC fractions as described in (Nichols et al., 1990). The TFIIIC fraction was incubated in batch with 1 ml IgG Sepharose resin for 6 hours in 10 mM Tris-HCl pH 8, 150 mM NaCl, 10% glycerol, 1 ug/ml protease inhibitors (leupeptin, pepstatin A, aprotonin and 1mM PMSF). TFIIIC was recovered after overnight incubation with TEV protease (100 units, NEB). TFIIIC was empirically determined to be saturating and TFIIIB and Pol III activities were determined to be limiting by multiple-round transcription assays.

In vitro transcription
Factor-independent transcription on a tailed template was as described in . Reactions contained EC buffer (40 mM Hepes-NaOH pH 8, 3 mM βmercaptoethanol, 5% glycerol, 100 μg/ml BSA) with 7 mM MgCl 2 , 100 mM NaCl, 100 µM each ATP, CTP and UTP, 25 µM GTP and 10 µCi [α 32 P]GTP and 20 ng to 460 ng of template and were incubated for 30 mins at 22° in a total volume of 25µl. The 3'-tailed duplex template was prepared from gel-purified annealed oligos AGTO71 and AGTO72 as described in (Arimbasseri and Maraia, 2013). Reactions were started by addition of Pol III, stopped with 70 µl proteinase K solution (1 mg/ml in 10 mM Tris-HCl pH 7.5, 0.1% SDS), phenol/chloroform/isoamyl alcohol extracted and precipitated with 2 M ammonium acetate and ethanol (Willis et al., 1992). RNA was separated by 8% denaturing polyacrylamide gel electrophoresis, before exposure to phosphorimager as either a wet gel or after fixation and drying. Transcripts were quantified with ImageQuant software.
Non-specific transcription reactions were carried out for 30 mins at various temperatures in 50 µl with 1ug polydAdT copolymer, 1 mM ATP, 500 µM UTP and 0.5µCi [α 32 P]UTP in EC buffer plus 7 mM MgCl 2 and 100mM NaCl (Huet et al., 1996). Reactions were started by addition of Pol III. Reactions were stopped and processed as described in (Roeder and Rutter, 1970) with 40µl 100 mM sodium pyrophosphate (pH 7) solution, containing 2 mg/ml salmon sperm DNA, 2 mg/ml BSA and 5 mM UTP, cooling followed by addition of 40 µl 5% SDS and 800 µl of cold 10% TCA, 40 mM sodium pyrophosphate. Acid-insoluble radioactivity was collected on Whatman GF/C filters, washed in 30 ml 5% TCA containing 40 mM sodium pyrophosphate, and counted in scintillation fluid.
Single-round factor-dependent transcription reactions were assembled on TZ1 for 45 min at 22° with TFIIIB, TFIIIC, DNA and Pol III to form preinitiation complexes (Kassavetis et al., 1989).
Stalled nascent 17 nucleotide transcripts were formed by the addition of 1.2 mM ATP, 600 µM CTP and 25 µM UTP and 10µCi [α 32 P]UTP for 5 mins. Time course experiments established that IIIB-IIIC-DNA-Pol III complex assembly was at equilibrium and that 17-mer synthesis was completed at or before 30 seconds for both wild-type and KE Pol III ternary complexes.
Reactions were either stopped or extended to full-length transcripts by the addition of GTP and UTP to a final concentration of 600 µM with 300 µg/ml heparin (to prevent Pol III reinitiation (Kassavetis et al., 1989)) for an additional 5 to 10mins. Reactions were processed with 20 µg carrier tRNA as described and analyzed on 15% denaturing gels as above.

Growth phenotypes of leukodystrophy mutations in the pore domain of Rpc160
POLR3A and its homologs are highly conserved across species and POLR3A HLD mutations identified to date are located throughout the entire protein (reviewed in (Wolf et al., 2014;Arimbasseri and Maraia, 2016b)). The most common disease-associated missense mutation in POLR3A, G672E, lies in a hotspot of missense residues that map to the pore domain of the enzyme; a region located below the floor of the DNA binding cleft that provides nucleotide access to the active site and where RNA is extruded during backtracking ( Figure 1A) (Gnatt et al., 2001). We mapped pore domain POLR3A HLD mutations onto its S. cerevisiae homolog, Rpc160, in cryo-EM structures of the preinitiation complex and elongating Pol III ( Figure 1A) (Hoffmann et al., 2015;Vorlander et al., 2018). The mutations are predicted to alter intramolecular interactions within Rpc160 and do not directly affect contacts with other Pol III subunits, subunits of TFIIIB or DNA ( Figures 1A and 1B).
Five disease-associated missense mutations were engineered at the homologous positions in Rpc160 as single mutants, including the mutation G686E (corresponding to G672E in POLR3A). Each single mutant was also engineered as a double mutant in combination with G686E. The RPC160 alleles were integrated at the LEU2 locus in a rpc160∆::hphNT1 haploid strain where viability was maintained by a plasmid copy of the wild-type RPC160 gene. The rescuing plasmid was evicted on FOA, and the mutant strains were evaluated for growth phenotypes and for Pol III transcription in vivo.
None of the single mutant strains exhibited a significant growth defect in liquid or solid media between 22° and 37° ( Figure 1C). An analysis of total RNA showed that all of the single mutants had wild-type levels of mature tRNA and 5S rRNA ( Figure S1A). The growth phenotypes of double mutant strains ranged from wild-type to lethal at 30°, with the three viable strains showing variable sensitivity to growth at 22° and 37°. The extremely slow-growing strain containing the R683G/G686E mutations exhibited a significant decrease in mature tRNA at the permissive temperature ( Figure S1A). An intermediate growth defect was measured for the strain carrying the Y685K/G686E mutations. This double mutation, which we abbreviate to KE for simplicity, had a doubling time in liquid media almost twice that of the wild-type and its parental single mutant strains ( Figure 1C) and was both cold-sensitive and heat-sensitive. The wild-type residues Y685 and G686 form a tight turn between two α-helices that pack onto other structural elements that position catalytically essential residues in the active site. We selected the KE mutant for further examination of Pol III transcription defects.

Compounding in vivo transcription defects in a RPC160 double mutant
Northern analyses showed that the KE mutant had reduced steady state levels of a subset of Pol III transcripts in vivo compared to wild-type and its parental counterparts. The KE mutant showed a significant decrease in the steady state level of both precursor (pre-) and mature forms of RPR1 RNA (the RNA component of nuclear RNase P, the enzyme that cleaves the 5' leader sequence of pre-tRNAs) (Figure 2A). Pre-and mature levels of SNR52 RNA (a nonessential snoRNA responsible for methylation of rRNA) were also decreased in the KE mutant  2F and Figure S1). The ratio of pre-to mature RPR1 RNA was approximately one to one for all strains, suggesting that the low pre-RPR1 RNA level in the KE mutant led to the reduced mature RPR1 RNA level. Processing of the pre-RPR1 RNA 5′-leader occurs after RNA assembly into the ribonucleoprotein complex (Srisawat et al., 2002) indicating that RNase P complexes might also be lower in the KE mutant. A probe specific to the 5'-end of pre-tRNA Leu RNA showed full-  Figures 2D, 2G and S1), suggesting a stochastic reduction in RNase P activity in independent KE mutant cultures. The defect in pre-tRNA processing in the KE mutant was not sufficient to reduce the total mature tRNA level ( Figure S1A).
As RPR1 RNA is essential for viability in yeast (Lee et al., 1991b), we asked whether multiple gene copies could rescue the slow growth rate of the KE mutant. The doubling time of the KE mutant remained slow and its temperature sensitivity was unchanged by transformation with a multicopy RPR1 plasmid ( Figure 3A). In wild-type cells the increased RPR1 gene copy number generated a five-fold increase in pre-RPR1 RNA and a 50% increase in mature RPR1 RNA. The effect on the KE mutant parental strains was less pronounced, generating a two to four-fold increase pre-RPR1 RNA and a 20-25% increase in mature RPR1 RNA. Increased RPR1 gene copy number did not significantly alter pre-and mature RPR1 RNA levels in the KE mutant and the activity of RNase P remained compromised ( Figures S2 and 3B). The underlying basis for the absence of RPR1 gene dosage effects on RPR1 RNA in the KE mutant is unknown (see Discussion).
Ectopic expression of the wild-type SSD1 gene (SSD1-V) has long been known to suppress the conditional growth defects of mutations in Pols I, II and III in strain backgrounds (e.g. W303 used in this study) that contain the ssd1-d2 allele, an early termination mutant of SSD1 (Stettler et al., 1993). Ssd1 is an RNA binding protein which affects the localization of specific mRNAs, limiting their translation (Hu et al., 2018) and Hsp104-mediated protein disaggregation (Mir et al., 2009). Introduction of the SSD1-V gene partially overcame the temperature-sensitive growth phenotype and suppressed the slow growth rate of the KE mutant at 30° ( Figure 3A). However, SSD1-V did not reverse the defect in generating pre-and mature RPR1 or pre-SNR52 RNAs ( Figures 3B and S2), nor did it alleviate the defect in RNase P processing of pre-tRNA Leu ( Figure   S2). Incorporation of [ 3 H]-uracil into total RNA in wild-type and KE mutant strains carrying SSD1-V was comparable, yet incorporation into mature tRNA by the KE mutant, calculated as either (i) the ratio of incorporation into 5.8S rRNA or (ii) the ratio of incorporation into total RNA, was low compared to wild-type values ( Figure 3C). There was also a defect of [ 3 H]-uracil incorporation into 5.8S rRNA relative to total RNA. Although the wild-type Ssd1 protein partially rescued the growth defects of the KE strain, the decrease in the abundance of pre-and mature RPR1 and SNR52 transcripts, and the reduction in neosynthesis of mature tRNAs, indicates that Ssd1 did not rescue the Pol III transcription defect.

Factor-dependent and factor-independent transcription is defective in vitro
Rpc160 protein abundance was equivalent in total cell extracts of both wild-type and KE mutant strains, whether extracts were prepared by native or denaturing methods ( Figure S3A and B). This indicates that the half-life of Rpc160 is not significantly affected by the KE mutations, unlike several other mutations in Rpc160 which cause Pol III to be targeted by the sumoylation machinery and degraded (Wang et al., 2018).
Wild-type and KE strains were engineered to contain N-terminal His-tag and Flag epitopes on the second largest subunit of Pol III, Rpc128 (C128) . Pol III was purified from both strains in comparable yield using immobilized metal affinity and anion exchange chromatography as per established protocols .
Affinity-purified Pol III was normalized for levels of C128 and Rpc34 (C34) proteins ( Figure S3B) to quantify activity in factor-independent and factor-dependent in vitro transcription assays.
Structures of Pol III show that C34, which is part of a trimeric C82/C34/C31 subcomplex, interacts with C160. Thus, the comparable recovery of C128 and C34 in the Pol III preparations reports the abundance of assembled Pol III complexes (Hoffmann et al., 2015;Abascal-Palacios et al., 2018;Han et al., 2018;Vorlander et al., 2018).
Factor-independent transcription of poly(dAdT), an alternating copolymer, was achieved with Pol III, ATP, and [α-32 P]UTP. In this assay, KE Pol III was 55% and 40% as active as wildtype Pol III at 22° and 30°, respectively ( Figure 4A). The activity of both polymerases was reduced at 37°, where transcription by KE Pol III was only 36% the level of wild-type Pol III.
Factor-independent transcription was also assessed on a tailed oligonucleotide duplex template containing a 3'-single-stranded overhang and a 7bp T-run terminator sequence positioned 50 nucleotides from the transcription start (Arimbasseri and Maraia, 2013). KE Pol III was 56% to 59% as active as wild-type Pol III, at template concentrations ranging of 0.8 and 18 ng/µl ( Figure   4B and data not shown). Both wild-type and KE Pol III exhibited a small but equivalent readthrough of the terminator (22 ± 3% and 20 ± 3%, respectively, n = 8), which is attributed to dissociation of the C53/37/11 Pol III subcomplex that occurs during enzyme purification (Arimbasseri and Maraia, 2013).
Factor-dependent transcription by Pol III was assayed using a plasmid-borne SUP4 tRNA Tyr gene as a template. Affinity-purified yeast TFIIIC and a partially-purified yeast TFIIIB fraction were used to achieve high factor-dependent transcription activity. Both the DNA template and TFIIIC activity were in excess and Pol III was limiting in these assays. The activity of KE Pol III on SUP4 DNA was 38 ± 10% (n = 9) the level of wild-type Pol III activity in multiple-round transcription assays ( Figure 5A). A comparable defect was observed for transcription of a tRNA Leu template ( Figure 5B). Transcription was also compromised on Pol III-transcribed genes (SCR1, RPR1 and SNR6) that deviate from the consensus sequence and/or spacing of type 2 promoter elements found in tRNA genes (Geiduschek and Tocchini-Valentini, 1988).
Transcription of a dimeric sup9-e tRNA gene from S.pombe, which contains a tRNA Ser -tRNA i Met gene where transcription is directed by the upstream tRNA (Willis et al., 1986), was also defective ( Figure 5B). These results indicate that the activity of KE Pol III in vitro is broadly and equally defective on templates that vary in promoter strength and promoter element organization. Moreover, there were no shorter or longer transcripts generated on any template indicating normal start site selection and termination site recognition. The apparently normal steady-state level of mature tRNAs ( Figure S1), SCR1 and U6 RNAs (Figures 2E and F) in the KE mutant strain, coupled with defective Pol III activity in vivo and in vitro, indicates that there are homeostatic mechanisms at play to maintain a constant level of these transcripts in vivo.
Factor-dependent single-round transcription assays were performed in which preinitiation complexes were assembled on SUP4 DNA with TFIIIC, TFIIIB and Pol III. Addition of ATP, CTP and 32 P-labelled UTP (25µM UTP, in the absence of GTP) allowed the formation of a ternary complex with a stalled nascent 17 nucleotide transcript (17-mer). 17-mer synthesis was lower for KE Pol III (averaging 35.2 ± 12% of the wildtype Pol III value, n = 9). Addition of unlabelled GTP and heparin to the stalled 17-mers allowed RNA synthesis to resume and limited Pol III elongation to a single-round of transcription (Kassavetis et al., 1989). These assays demonstrated that all labelled 17-mers formed by KE Pol III could be fully extended into fulllength transcripts and that these generated a comparable differential; KE Pol III activity was 31 ± 6% the wild-type value ( Figure 5C). No gross elongation or stalling phenotype was observed and KE Pol III transitioned from a stalled transcript into elongation mode to generate full-length RNA like wild-type ( Figure 5C). The differential in single-round synthesis of full-length transcripts was maintained at high UTP concentrations (400µM) indicating that the defect in 17mer synthesis was not caused by hypersensitivity of KE Pol III to low UTP concentrations ( Figure   5D). We noted that factor-dependent transcription by KE Pol III was modestly more defective than factor-independent transcription, suggesting that the TFIIIB, TFIIIC factors and/or supercoiled templates may impose further constraints on step(s) at which KE Pol III is defective.
Together, the in vitro transcription experiments indicate that KE Pol III is globally defective at an early step in transcription on a variety of templates.

Discussion
The essential function of Pol III and the high conservation between POLR3A and Rpc160 (51% identical, 67% similar) suggested that pathological missense substitutions at orthologous positions in yeast would compromise Pol III transcription. This view was further supported by the fact that gene deletions involving the majority of Pol III subunits in yeast, all of which are essential, can be complemented by their human orthologs (Kachroo et al., 2015). HLDassociated missense mutations that map to the pore domain of Pol III were all able to complement the lethality of a yeast RPC160 deletion. However, these single mutants exhibited no growth phenotypes and no measurable differences in the level of RPR1 RNA or of a precursor tRNA that is normally sensitive to changes in Pol III transcription (Li et al., 2000b;Upadhya et al., 2002;Lee et al., 2015). The absence of transcription phenotypes for any single mutation in this region may result from yeast-specific amino acids within Rpc160 that buffer functional deficits in the protein and which negate, to a variable extent, the effect of a second mutation. Alternatively, as POLR3A is a substrate for both Hsp90 and co-chaperone binding in 13 human cells (Taipale et al., 2014), species-specific differences in mutant protein folding, stability and complex assembly may also contribute to the suppression of a phenotype for the single mutations in yeast.
Purified KE Pol III has a global defect in function in both factor-independent and factordependent transcription in vitro. These findings and the location of the KE mutations on the opposite side of Pol III from its interface with TFIIIB (Abascal-Palacios et al., 2018;Vorlander et al., 2018), suggests that Pol III recruitment to the TFIIIB-DNA complex is unlikely to be affected by the mutations. Transcription by Pol III can be described as a series of fundamental steps including DNA binding to form a closed preinitation complex, DNA opening, transcription initiation, escape from abortive initiation, elongation, termination and facilitated recycling (reviewed in ). Nucleotide incorporation directed by poly (dAdT) reports on all of these steps except terminator recognition and facilitated recycling. The tailed DNA template used in this work additionally reports on termination. Both assays indicated that KE Pol III was ~60% as active as wild-type Pol III. KE Pol III was uniformly defective in factordirected multiple-round transcription on a variety of Pol III-transcribed genes, with no indication of gene-specific defects. KE Pol III was also defective in the production of stalled nascent 17mer transcripts, which encompass the DNA binding and opening, transcription initiation and escape from abortive initiation steps. KE Pol III was as competent as wild-type Pol III in supporting further polymerization of the 17-mer to produce a single full-length transcript. These in vitro transcription properties of KE Pol III are distinct from those of rpc160-112 (a double substitution flanking the catalytic site) which showed a reduced elongation rate, increased pausing at intrinsic pause sites and slippage of nascent RNA in the stalled 17-mer (Dieci et al., 1995). As the activity of KE Pol III in single-round transcription and multiple-round transcription was comparably defective, at ~40% of the wild-type activity, we suggest that KE Pol III may partition into two forms: one as competent for transcription as WT Pol III and the other form, functionally inactive. Whether the inactive KE Pol III population is defective in DNA binding or at an early step in transcription initiation has not been examined.
Although KE Pol III transcription in vitro was clearly defective, its defects in vivo were less apparent. The abundance of full-length pre-tRNA Leu transcripts, an otherwise useful proxy for Pol III transcription, was not significantly altered in the KE mutant as tRNA processing was compromised. However, 3 H-uracil incorporation into mature tRNAs was decreased compared to wild-type, clearly indicating a defect in global tRNA neosynthesis. Elevated nuclear exosome activity, estimated to specifically degrade pre-tRNA in wild-type cells such that less than 50% of pre-tRNA ends up as mature tRNA (Gudipati et al., 2012;Axhemi et al., 2020), could also contribute to the net decrease in labeled mature tRNA. However, as the level of total mature tRNA was not changed in the KE mutant relative to wild-type, the homeostatic mechanisms presumed to exist to maintain a constant level of mature tRNA in the cell are not affected in the KE mutant (Bonhoure et al., 2015). We have not assessed whether the composition of the tRNA pool, known to change in response to cell stress (Pang et al., 2014;Torrent et al., 2018), has been altered in the KE mutant.
The lower levels of pre-and mature RPR1 and SNR52 RNAs detected in the KE mutant in vivo likely stem from a combination of their non-canonical type 2 promoter structure, their unique processing and/or a requirement for specific RNA binding proteins to prevent their degradation (Lygerou et al., 1994;Srisawat et al., 2002;Preti et al., 2006;Palsule et al., 2019).
Both RPR1 and SNR52 are synthesized as precursor RNAs from tRNA-like promoter elements that are positioned upstream of the mature RNAs and thus produce precursor RNAs with long 5' leader sequences (Lee et al., 1991b;Harismendy et al., 2003;Lee et al., 2003;Preti et al., 2006). Although both genes contain A-and B-box internal promoter elements, the B-box in RPR1 deviates from consensus (Lee et al., 1991a) and the distance between the boxes in SNR52 is significantly longer than is typical. Additionally, the SNR52 sequence contains a run of six T residues, normally a strong termination signal in the context of other Pol III-transcribed genes (Braglia et al., 2005;Arimbasseri and Maraia, 2016a). Transcription of RPR1, both in vivo and in vitro, is particularly sensitive to an internal deletion mutant of the Bdp1 component of TFIIIB (bdp1∆269-312) (Ishiguro and Kassavetis, 2003) that is predicted to alter DNA binding by Bdp1 in the TFIIIB-Pol III preinitiation complex (Abascal-Palacios et al., 2018). Mature RPR1 RNA (and RNase P function) remained low in the KE mutant despite additional copies of the RPR1 gene, indicating elevated turnover of this pre-RNA. Cleavages to generate the mature 5'ends of the RPR1 and SNR52 RNAs occur after assembly into their respective ribonucleoprotein complexes (Lygerou et al., 1994;Preti et al., 2006;Palsule et al., 2019), suggesting that low abundance of their specific binding proteins might underlie the low pre-RNA levels.
Pulse-labelling with 3 H-uracil indicated that 5.8S rRNA biosynthesis was also decreased in the KE mutant. Crosstalk between defects in Pol III transcription and Pol I transcription is well documented. Mutations in RPR1 RNA have been reported to affect 5.8S rRNA maturation in vivo (Chamberlain et al., 1996) and mutations in RPC160 and 5S rRNA genes affect pre-rRNA processing (Hermann-Le Denmat et al., 1994;Dechampesme et al., 1999;Briand et al., 2001).
It is unlikely that the reduction in mature SNR52 RNA in the KE mutant significantly affects ribosome function, as a complete deletion of SNR52 has limited phenotypic consequences (Esguerra et al., 2008). However, the reduction in 5.8S rRNA maturation and pre-rRNA processing may limit translation and underlie the slowed growth rate of the KE strain.
Although the in vitro transcription activity of KE Pol III was broadly defective, the growth phenotype of the KE strain was affected by the genetic background and the transcription defect of the KE mutation was detected at only a subset of Pol III-transcribed genes in vivo. In metazoans, the Pol III transcriptome varies in response to proliferation state, nutritional signaling, cell origin and stress (Dittmar et al., 2006;Coughlin et al., 2009;Canella et al., 2010;Canella et al., 2012;Ishimura et al., 2014;Carnevali et al., 2017;Mange et al., 2017;Yeganeh et al., 2019), and includes additional genes whose products affect pol II elongation (Peterlin et al., 2012), regulate autophagy (Horos et al., 2019) and, in humans, are implicated in translational repression and histone 3'-end maturation (Samson et al., 2018). The list of RNase P substrates is also expanded to include two long non-coding RNAs with pleiotropic activities (Wilusz, 2016). The additional complexity of the Pol III transcriptome, coupled with developmental, differentiation and tissue-specific gene expression, suggests that background modifiers of transcriptional defects will complicate readouts of defects in Pol III transcription in more complex organisms.

Conclusions
This work shows that the most frequent POLR3A HLD missense mutation, G672E, did not alter growth or Pol III transcription when engineered into its yeast homolog, Rpc160, as seen for the corresponding mutation in mice (Choquet et al., 2017). The addition of a second diseaseassociated mutation in the same region of Rpc160 caused intermediate defects in growth and Pol III transcription. This raises the possibility that introducing the corresponding double-mutant Polr3a gene in the mouse might generate a viable yet dysfunctional Pol III transcription phenotype and provide a mammalian model for the study of Pol III-associated leukodystrophy.
Author Contributions: Each author has made substantial contributions to the work; has approved the submitted version; agrees to be personally accountable for the work.     transcription. Equal amounts of wild-type (black) and KE (red) affinity-purified Pol III was titrated into non-specific transcription reactions that used poly (dAdT) copolymer as a template.
Transcription activity is expressed relative to that of wild-type Pol III at each temperature. KE Pol III was 55%, 40% and 36% as active as wild-type Pol III at 22, 30 and 37°, respectively. (b) Nonspecific tailed template transcription. Equal amounts of wild-type (black) and KE (red) affinitypurified Pol III were titrated into non-specific transcription assays of a tailed oligo duplex.
Transcription activity is expressed relative to that of wild-type Pol III. A representative titration at 250ng template is shown. KE Pol III had 59 to 56% the activity of wild-type Pol III on 20ng and 250ng template respectively. T, correctly terminated products; RT, read-through transcripts (22 ±3% and 20 ±3% read-through, for wildtype and KE).