The Non-Canonical CTD of RNAP-II Is Essential for Productive RNA Synthesis in Trypanosoma brucei

The carboxy-terminal domain (CTD) of the largest subunit (RPB1) of RNA polymerase II (RNAP-II) is essential for gene expression in metazoa and yeast. The canonical CTD is characterized by heptapeptide repeats. Differential phosphorylation of canonical CTD orchestrates transcriptional and co-transcriptional maturation of mRNA and snRNA. Many organisms, including trypanosomes, lack a canonical CTD. In these organisms, the CTD is called a non-canonical CTD or pseudo-CTD (ΨCTD. In the African trypanosome, Trypanosoma brucei, the ΨCTD is ∼285 amino acids long, rich in serines and prolines, and phosphorylated. We report that T. brucei RNAP-II lacking the entire ΨCTD or containing only a 95-amino-acid-long ΨCTD failed to support cell viability. In contrast, RNAP-II with a 186-amino-acid-long ΨCTD maintained cellular growth. RNAP-II with ΨCTD truncations resulted in abortive initiation of transcription. These data establish that non-canonical CTDs play an important role in gene expression.


Introduction
RNA polymerase II (RNAP-II) is the eukaryotic enzyme responsible for synthesis of mRNA. This enzyme is an ,550 kDa complex consisting of twelve subunits, RPB1-12 [1]. The carboxy-terminal domain (CTD) of the largest subunit, RPB1, is essential for cell survival. In well-studied organisms, such as metazoa and yeast, the CTD consists of consensus heptapeptide repeats having the sequence Y 1 S 2 P 3 T 4 S 5 P 6 S 7 ( [2,3] and reviewed in [4,5]). The CTD plays a pivotal role in RNA production by recruiting multiple proteins that modulate transcription initiation, transcription elongation, transcription termination, co-transcriptional mRNA 59 capping, RNA splicing, and RNA polyadenylation [6][7][8]. Interaction of proteins with the CTD is orchestrated by dynamic and differential phosphorylation. The patterns of phosphorylation make up the ''CTD code'' [5,9]. Non-canonical CTDs, which lack the heptapeptide repeats and thus are pseudo CTDs (YCTDs), exist in a wide variety of eukaryotic organisms [10,11]. For example, the early branching protozoan, Trichomonas vaginalis, and the multicellular red alga, Bonnemaisonia hamifera, as well as members of the Apicomplexian, Stylonichia, and Kinetoplastid groups contain YCTDs as part of their RNAP-II machinery [12,13]. We sought to determine whether a YCTD is essential for gene expression.
Among the Kinetoplastids, the largest subunit of RNAP-II has an ,300 amino acid-long YCTD that lacks heptapeptide repeats and exhibits little amino acid similarity to CTDs of other organisms [14][15][16][17]. In the African trypanosome, Trypanosoma brucei, the YCTD of RNAP-II is an ,285 amino acid-long carboxyl terminal extension of RPB1 ( Figure 1 and [15,16]). In T. brucei, RNAP-II synthesizes two very different types of RNA, a long polycistronic precursor mRNA (pre-mRNA) and a short, capped Spliced Leader (SL) RNA. mRNA maturation require addition of SL RNA to pre-mRNAs through a trans-splicing reaction. Different forms of RNAP-II, associated with specific transcription factors, likely transcribe SL RNAs and pre-mRNAs. Indeed the T. brucei homolog of the yeast transcription factor Spt16 appears to associate with protein-coding genes but not with the SL RNA gene [18]. In contrast, the T. brucei homolog of the mammalian snRNA transcription factor SNAPc appears to associate only with the SL RNA gene promoter, although one subunit of SNAPc in Leishmania major, a relate parasite, occupies regions upstream of protein-coding genes [19]. Whether RNAP-II YCTD is important for transcription of either the SL RNA gene and/or protein-coding genes is unknown.
To address if non-canonical CTDs play a fundamental role in gene expression, we undertook a study of T. brucei RNAP-II YCTD. We altered RNAP-II and tested its function using an in vivo assay system and discovered that the YCTD is essential for cell survivial and production of both SL RNA and mRNA. Nascent transcription analysis demonstrated that the YCTD is required specifically for productive transcription initiation, as an RNAP-II severely truncated within the YCTD caused abortive initiation. These results demonstrate that a non-canonical CTD is a vital component of the RNAP-II machinery in eukaryotic cells.

T. brucei YCTD undergoes phosphorylation
It has been shown previously that T. brucei RNAP-II is modified by phosphorylation, despite the lack of consensus heptapeptide repeat sequences [20]. To determine whether the phosphorylation occurs within the YCTD, we directly visualized RNAP-II labeled with 32 P-orthophosphate. For this work, we tagged RNAP-II in a T. brucei procyclic cell line by stably-expressing the RPB3 subunit of the enzyme with a tandem affinity tag. These transgenic cells were metabolically labeled with 32 P-orthophosphate and RNAP-II was The carboxy terminus of T. brucei RPB1 lacks the highly conserved YSPTSPS repeats that orchestrate multiple co-transcriptional processes in the well-studied eukarya. Four different RPB1 proteins analyzed in this study are diagrammed: RPB1 containing no YCTD (DYCTD, truncated at amino acid 1481), onethird of YCTD (1/3 YCTD, truncated at amino acid 1576), two-thirds of YCTD (2/3 YCTD, truncated at amino acid 1667), and full length YCTD (full YCTD). The heavy line at the C-terminus of each truncation represents the five conserved amino acids (E/DEEEQ). The location of the ,27 kDa Lys-C fragment is indicated with a horizontal line over the YCTD. (B) Alignment of the human and yeast CTD with the YCTD sequence of T. brucei RPB1. A portion of the H domain is included (gray bar) as reference. The CTD's linker region (thin line) and the heptapeptide repeat domain (thick line) are shown. Thirty-one of the 52 heptapeptide repeats in the mammalian CTD are shown. The YCTD truncations are defined by arrows following the final amino acid contained in each mutant protein. Grey boxes show similar amino acids in two sequences, black boxes show similar amino acids in all three sequences. doi:10.1371/journal.pone.0006959.g001 purified from nuclear extracts using protein A and streptavidinbinding peptide affinity chromatography. An autoradiograph of the purified protein, separated into polypeptides by denaturing gel electrophoresis, indicated that the largest subunit (RPB1) is modified by phosphorylation (Figure 2A). A stained gel of the purified proteins revealed both phosphorylated and non-phosphorylated forms of RPB1 ( Figure 2B). Seven of the eleven other RNAP-II subunits, as verified by mass spectrometry, were also seen [21]. To determine whether phosphorylation occurs in the YCTD of RPB1, 32 P-labeled RPB1 was subjected to in-gel proteolysis by endoproteinase Lys-C, which cleaves the peptide bond at the carboxy side of lysine. Lys-C digestion of RPB1 generated a 253 amino acid long, ,27 kDa, peptide consisting only of the YCTD (from amino acid 1491-1744) minus its terminal 21 amino acids ( Figure 1A). The remainder of RPB1 was fragmented into ,76 amino acid-long peptides. Immunoblotting with anti-YCTD antibodies detected the ,27 kDa peptide ( Figure 2C). Lys-C digestion products from 32 Porthophosphate-labeled RPB1 are shown in Figure 2D. The heterogeneous collection of radiolabeled peptides migrating more slowly than ,27 kDa are YCTD peptides that differ in phosphorylation patterns. The immunoblot in Figure 2C shows most clearly the non-phosphorylated YCTD because the phosphorylated peptides migrate as a broad band that eludes antibody detection. Thus, our analysis demonstrates that although T. brucei RNAP-II YCTD lacks a heptapeptide repeat sequence, it undergoes phosphorylation.
RNAi-targeting of mRNA 39UTR allows regulated depletion of RPB1 To test directly the importance of RNAP-II YCTD in transcription, we used RNAi to deplete endogenous RNAP-II (referred to as 'endo') and replace it with mutant RNAP-II, conditionally expressed from an exogenously provided DNA cassette. Endo RNAP-II relies upon RPB1 that is encoded by two sets of allelic gene copies on chromosomes 4 and 8. These genes differ slightly in their protein-coding regions. Fortunately, RPB1 mRNAs generated from all four alleles contain identical 39UTRs, as determined by analysis of RPB1 mRNA by 39 RACE-PCR (data not shown). We generated cell lines containing inducible RNAi that targeted the RPB1 39UTR. As expected, targeting 39UTR of RPB1 mRNA by tetracycline-inducible RNAi expression stops cellular growth of procyclic T. brucei by day 2, in contrast to uninduced cells. The induced cells progressed to cell death by day 4 ( Figure 3A). Western analysis of total cellular protein showed decrease in RPB1 beginning by day 1 ( Figure 3B).

RNAP-II containing 1/3 YCTD is toxic and causes cell death
To study whether RPB1 mutant proteins can support transcription, we expressed tagged versions of full, 2/3, 1/3 and null (D) YCTD domain-containing proteins ( Figure 1A). The full YCTD or 2/3 YCTD RPB1 transgenic cell lines were readily established. However, cell lines with a 1/3 YCTD or D YCTD RPB1 were difficult to establish: only one cell line for each construct was obtained after multiple transfections. The 1/3 YCTD or D YCTD RPB1-containing parasites grew poorly and had low levels of non-phosphorylated tagged-RPB1 ( Figure S1). These results indicated that expression of either the 1/3 YCTD or D YCTD poison cells and prohibit selection of robustly expressing cell lines.
To investigate whether 1/3 YCTD poisons the RNAP-II enzyme, we tailored RPB1 transgenes to express in an inducible fashion ( Table 1). The discovery that RNAP-II with 2/3 YCTD was well tolerated and modified by phosphorylation ( Figure S1), whereas 1/3 YCTD was not, prompted us to compare 2/3 YCTD with 1/3 YCTD RNAP-II in our studies. Induced expression of 2/3 YCTD RNAP-II resulted in normal cellular growth ( Figure 4A). However, induced expression of 1/3 YCTD RNAP-II was toxic to cells. The crippling effect of 1/3 YCTD RNAP-II was evident in the rapid growth arrest and loss of viable cells. Microscopic observation following expression of 1/3 YCTD RNAP-II revealed striking changes: cells became rounded and formed large aggregates. In contrast, cell morphology was normal following expression of RNAP-II with 2/3 YCTD ( Figure 4A).
To verify that the 1/3 YCTD RPB1 was incorporated into RNAP-II, we analyzed enzyme complexes ( Figure 4B). Fractionation of nuclear extract by sucrose density-gradient centrifugation confirmed that tagged-2/3 YCTD and tagged-1/3 YCTD RPB1 assembled into complexes that migrated near the bottom of the gradient. These same fractions were enriched for endogenous RNAP-II. Transcriptionally active endogenous RNAP-II also migrated at this position in similar sedimentation velocity measurements (data not shown). Western blot analysis of immunoprecipitated proteins captured by polyclonal anti-RPB4 antibody demonstrated the presence of tagged-1/3 YCTD, tagged-2/3 YCTD, or endogenous RPB1 in RNAP-II ( Figure 4C). These results confirmed that 2/3 YCTD or 1/3 YCTD RPB1 were incorporated into RNAP-II. Thus, RNAP-II with 1/3 YCTD is toxic to cells and produces a dominant negative effect over endogenous RNAP-II.

RNAP-II containing 2/3 YCTD rescues cellular growth in absence of endogenous YCTD
Although a 1/3 YCTD RNAP-II had a dominant-negative effect, cells expressing both full-length YCTD and 2/3 YCTD RNAP-II grew normally. We speculated that RNAP-II harboring a 2/3 YCTD might be as proficient as the endogenous enzyme in supporting cell growth. Thus, a 2/3 YCTD RNAP-II was tested for its ability to rescue cells depleted of endogenous RNAP-II ( Figure 5). We established stable cell lines containing a regulated RNAi construct and a regulated tagged-RPB1 gene. Addition of tetracycline induced RNAi production and tagged RPB1 expression. Thus, endogenous RPB1 was selectively destroyed and the tagged-RPB1 proteins that contain either the 2/3 YCTD or 1/3 YCTD were produced. Growth curves showed that 2/3 YCTD RNAP-II sustained cellular growth after RNAi-depletion of endogenous RPB1 ( Figure 5A). In contrast, 1/3 YCTD RNAP-II failed to rescue the cells.
Endogenous RPB1 depletion was monitored by immunoblotting with anti-YCTD antibody. The top, left blot in Figure 5B shows that both phosphorylated and non-phosphorylated forms of the endogenous RPB1 decreased after RNAi-targeting. In addition, both phosphorylated and non-phosphorylated forms of the tagged-2/3 YCTD RPB1 were present after induction. Similar results are shown in the top, right blot in Figure 5B, although the diminished number of epitopes on the 1/3 YCTD resulted in weak antibody detection. Expression of tagged proteins was independently monitored in immunoblots by using anti-tag antibodies ( Figure 5B, lower blots). As noted above, the 1/3 YCTD RPB1 failed to undergo phosphorylation, and thus a single band corresponding to the tagged protein was observed. These data confirmed that induced expression of tagged 2/3 and 1/3 YCTD RPB1 occurred in our experiments. Taken together, these results argue that YCTD contributes to cell viability, likely as an essential component of functional RNAP-II.   RNAP-II requires a 2/3 YCTD to maintain steady-state transcript levels To determine the role of YCTD on RNAP-II transcription, first we analyzed transgenic cells for steady-state transcript levels following concomitant endogenous RPB1 depletion and tagged-RPB1 induction. Northern blot analyses of total RNA confirmed that expression of the tagged-RPB1 mRNA from transgenes containing either the 2/3 YCTD or 1/3 YCTD was tightly regulated and induced with tetracycline ( Figure 6A). These data are shown schematically in the right portion of the panel. Levels of 7SL RNA, which requires RNAP-III, were unaffected during the two-day course of the experiment. In contrast, levels of b-tubulin mRNA, which requires RNAP-II for synthesis, were markedly reduced in cells expressing the 1/3 YCTD RNAP-II; only ,40% of the b-tubulin mRNA remained by day 2 after induction. In cells containing 2/3 YCTD RNAP-II, no decrease in b-tubulin mRNA was observed. The differential effect on b-tubulin levels supported the observation that while RNAP-II with 1/3 YCTD was crippled, RNAP-II with 2/3 YCTD was functional. Finally, levels of procyclin/EP1 mRNA, which requires RNAP-I, indicated that ,60% of the procyclin/EP1 mRNA was present in 1/3 YCTDexpressing cells in comparisons to cells expressing endogenous enzyme. The decrease in EP1 mRNA steady-state levels was likely due to inefficient mRNA maturation, which requires the RNAP-II-dependent SL RNA transcript.
To test this prediction, we assessed SL RNA levels by primer extension analysis ( Figure 6B). Cells that relied on 1/3 YCTD RNAP-II exhibited a striking decrease in SL RNA steady-state levels; there was a 70% decrease on day 1 and a 85% decrease by day 2. In contrast, cells containing 2/3 YCTD RNAP-II maintained SL RNA levels. They experienced only a ,30% decrease on day 1 that did not change by day 2. Thus, 1/3 YCTD RNAP-II was probably incapable of making new SL RNA transcripts and thus failed to maintain a reservoir of SL RNA necessary for efficient mRNA maturation.

RNAP-II with 1/3 YCTD is defective in transcription and causes abortive initiation
To examine how the 1/3 YCTD impairs RNAP-II function, we assessed nascent RNA synthesis in transgenic cells before and after expression of 1/3 YCTD RPB1-containing enzyme. In comparison to uninduced cells, cells utilizing 1/3 YCTD RNAP-II poorly synthesized b2tubulin and SL RNA ( Figure 7A). Reduced synthesis of b2tubulin and SL RNA was also observed in cells utilizing 2/3 YCTD-containing RNAP-II, although this reduction was minimal. For the cells relying on the 1/3 YCTD RNAP-II, we observed an overall decrease in RNAP-I (procyclin/EP1 and 18S rRNA) and RNAP-III (7SL RNA and 5S rRNA) transcription, which probably indicates a general reduction in their transcriptional capacity.
We verified that the 1/3 YCTD RNAP-II is less efficient in transcription than its wild-type or 2/3 YCTD-containing homolog by direct visualization of newly synthesized, radiolabeled SL RNA ( Figure 7B). Nascent SL RNA was significantly more reduced in cells containing 1/3 YCTD RNAP-II compared to cells containing 2/3 YCTD RNAP-II (,60% compared to 20%), after two days of expression. Interestingly, we also observed an increase in small radiolabeled molecules migrating near the bottom of the gel, as shown in the brief exposure in Figure 7B. Compared to wild-type, cells expressing either 2/3 YCTD RNAP-II or 1/3 YCTD RNAP-II produced increased level of very short (,4 nt long) transcripts. The amount of short transcripts was greater in cells containing 1/3 YCTD RNAP-II (eight-fold over wild-type) than in cells containing 2/3 YCTD (four-fold over wild-type). We postulate that these RNAs were generated from abortive initiation.
We first determined that the radiolabelled molecules were RNA, and thus sensitive to alkali treatment ( Figure 7C). Mononucleotides were produced from gel-purified, short transcripts (see Figure 7B, brief exposure). To confirm that the short RNAs indeed originated from transcription initiation events, we characterized their 59 end. Short transcripts, treated with Tobacco Acid Pyrophosphatase (TAP), were ligated to form a ladder of longer RNAs ( Figure 7D). No ligation products were observed when TAP-treatment was omitted. Taken together, the nascent transcription experiments supported the conclusion that RNAP-II requires a YCTD to enter into a productive transcription process.

Discussion
The present work describes the first characterization of an RNAP-II non-canonical CTD, which is found in many single-cell organisms. The T. brucei RNAP-II uses a non-canonical CTD for cell growth and viability. The growth defect associated with YCTD deletions was likely a direct result of decreased mRNA production. Specifically, when the YCTD is truncated, defective RNAP-II produced less mRNA and SL RNA and increased abortive initiation products.
A consequence of a non-canonical CTD is the opportunity for RNAP-II to partner with unknown proteins. This contention is best illuminated by the elegant biochemical and structural studies in Trichomonas vaginalis. Here, a non-canonical CTD partners with a Trichomonas-specific protein, IBP39, to stably assemble RNAP-II at the initiator element of promoters [12]. Affinity-purification of RNAP-II from T. brucei revealed several copurified proteins that appear to be trypanosome-specific [21,22]. We speculate that some of these proteins participate in gene expression through specific interaction with YCTD. Indeed, a ten amino acid motif (SSYHLQ-SVAP) is shared among trypanosomes within the serine/proline-rich subdomain of YCTD (Figure 1). This motif is within the 2/3 YCTD truncation that rescued cells when wildtype RNAP-II was depleted (Figures 5 and 6). In summary, these observations are consistent with non-canonical CTD having significant roles in organism-specific RNAP-II machineries.
Modification of canonical CTD by phosphorylation at specific serine residues is required for orchestrating transcriptional activities in better studied eukaryotes. It is known that a S 5 phospho CTD-code dictates that an enzyme leaves a preinitiation complex, and a S 2 phospho CTD-code dictates that an enzyme becomes elongation competent [6]. In T.brucei, we found phosphorylated forms of YCTD within RNAP-II (Figure 2). Proteolytic digestion of phospho-YCTD produced what was most likely a heterogeneous set of differentially phosphorylated YCTDs. We note that there are ,30 potential phosphorylation sites within the 2/3 YCTD that are absent from the 1/3 YCTD. Thus, we speculate that a phosphorylation-code occurs within the YCTD in T. brucei that may account for the two different RNAP-IIdependent reactions, namely, transcription of short, capped SL RNAs and long pre-mRNAs. Such a YCTD code may allow RNAP-II to specifically function at the SL RNA gene promoter, much like the S 7 phosphorylation mark on the mammalian CTD directs polymerase specifically during small nuclear RNA transcription [4].
The structure of the canonical CTD is divided into three subdomains: the linker region, the central reiterated region that harbors intact and degenerate heptapeptide repeats, and the terminal short acidic peptide [23]. The YCTD parallels the CTD in structure; it contains a linker region of size similar to CTD, a central serine/proline-rich region, and a terminal short acidic peptide (Figure 1). In the case of 1/3 YCTD, the RNAP-II contained a YCTD having only the linker region plus a small part of the central serine/proline-rich region. This enzyme failed to rescue cell growth when wild-type RNAP-II was depleted. In addition, we did not observe phosphorylation of the 1/3 YCTD. In contrast, 2/3 YCTD of RNAP-II contained the linker region plus half of the central serine/proline-rich region. This enzyme was phosphorylated within the YCTD and allowed cell growth in the absence of wild-type RNAP-II. Taken together, it appears that the central serine/proline-rich region of the YCTD is important for polymerase function.
Although 2/3 YCTD could sustain cell growth, it generated increased levels of abortive transcripts. Nascent transcription in cells containing the 1/3 YCTD generated even higher levels of abortive transcripts. These transcripts were confirmed to be abortively initiated RNA, as they possessed 59 tri-phosphate ends, a signature of transcription initiation ( Figure 7B, C and D). Studies of the mammalian CTD indicate that an RNAP-II having only 5 heptapeptide repeats can bind a gene promoter but is defective in initiating transcription [24]. This mutant CTD is structurally similar to our 1/3 YCTD, because it contains the linker region, a very small part of the central region, and the terminal acidic region. Thus, the central region of both the YCTD and CTD plays a role in the transition from an initiating polymerase, which generates abortive RNAs, to an elongating enzyme, which produces full-length mRNAs.
An unexpected and interesting observation was made when an RNAP-II with 1/3 YCTD was expressed in cells along with wildtype levels of endogenous RNAP-II. Within 16 hours of 1/3 YCTD expression, cells ceased to move, changed their shape, formed large aggregates, and soon died ( Figure 4A). These dramatic phenotypic changes were not observed when 1/3 YCTD was expressed in cells along with very low levels of endogenous RNAP-II. In this case cells merely stopped dividing by day 2 and slowly died ( Figure 5A). This differential response may indicate that a wild-type RNAP-II, is required to synthesize protein-coding RNAs used in a stress response [25]. A stress response may account for the observed, abrupt morphological alterations in the cells expressing both the mutant and the endogenous polymerase. An actively transcribing polymerase in cells with abundant wildtype RNAP-II enables cells to respond to the stress imposed by the mutant YCTD. In contrast, cells with limited number of active polymerase could not. Indeed, Drosophila stress responses require transcription of heat shock proteins within minutes of the initial stress [26]. This rapid response is accompanied by the release of RNAP-II from its transcriptionally paused state on heat shock genes and depends on transcription elongation factors [27]. If a mechanism similar to this type of elongation control exists in trypanosomes, it raises the intriguing possibility of RNAP-II transcriptional control, heretofore unrecognized, in these organisms.

Plasmid constructs
The RNAi construct was made in p2T7-177 by inserting a 260 bp region of the RPB1 39UTR at BamHI and XhoI sites [28]. This plasmid integrates in the highly repressed 177 bp satellite region of a mini-chromosome. The construct was linearized with NotI, transfected into procyclic T. brucei cells and transgenic clonal cell lines selected by limiting dilution in the presence of phleomycin [29]. pAD71, designed to express a transgene constitutively, was derived from pUC19 and contains sequences for homologous recombination into the procyclin/GPEET locus, following linearization by NotI digestion. The transgene uses the endogenous procyclin/GPEET promoter and a 39UTR derived from the aldolase mRNA. A blasticidin-resistance gene, which follows the transgene, is co-expressed using actin 59 SAS and 39UTR regions. pAD74 was derived from pLEW100, minus the luciferase open reading frame and containing a blasticidin-resistance gene in place of the phleomycin-resistance gene and an rRNA promoter in place of the procyclin/GPEET promoter.
Plasmid RPB3-SBP was generated by joining the RPB3 open reading frame (Tb927.3.550) with the protein A open reading frame and the Streptavidin binding protein (SBP) open reading frame (a gift from Larry Simpson) and cloned into pAD71 downstream from the procyclin/GPEET 59UTR. To establish mutant RNAP-II, we generated epitope tagged versions of RPB1, with varying truncations of YCTD. A Ty1-tag (EVHTNQDPLD) [30] at the N-terminus allowed us to monitor the expression of mutant RPB1 proteins. Transgenes expressing mutant RNAP-II were designed with a well-characterized 39UTR of aldolase (ALD) mRNA and thus resistant to the RNAi-targeting of endogenous RPB1 mRNA. The C-terminal five conserved amino acids, which contribute to RPB1 stability [24], was included in each of the four transgene constructs ( Figure 1A). For constitutive expression of the transgenes, the tagged and mutant RPB1 ORFs were cloned into pAD71. For inducible expression of the transgenes, they were cloned into pAD74. All plasmid constructs were confirmed by DNA sequencing.

Culturing and transfection of parasites
Procyclic (tsetse midgut form) wild-type T. brucei Lister 427 was grown in SDM-79 supplemented with 10% FCS as described [29]. For all transgene experiments, cells were selected and grown in medium containing tetracycline-free serum. The 29-13 cell line contains the T7 RNAP and the tetracycline repressor that are maintained by their co-expression with G418 and hygromycinresistance genes [29]. p2T7-177 derivatives were transfected into 29-13 and stable cell lines were selected using 2.5 mg/ml phleomycin. pAD71 and pAD74 derivatives were transfected into 29-13 and selected using 10 mg/ml blasticidin. Inducible RNAi and transgene expression were regulated by addition of 500 ng/ml tetracycline. For metabolic labeling, 5610 8 cells were incubated with 500 mCi of 32 P-orthophosphate (9000 Ci/mmole) in 5 mls of phosphate-free DMEM (40 min, 26uC).

Preparation of T. brucei nuclear extracts and protein purification
Nuclear extracts were made from the RPB3-SBP cells as previously described [31]. Protein was purified using tandem affinity chromatography; IgG-bound protein was washed and released from the matrix using TEV-protease. Eluted protein was bound to Streptavidin-agarose, washed and eluted with 2 mM biotin-containing buffer. For velocity sedimentation, nuclear extracts were loaded onto a 10-30% sucrose gradient as described [31]. Fractions were collected from the bottom of the tube and analyzed by immunoblotting. 32 P-containing RPB1 was in-gel digested with endoproteinase Lys-C (Roche) as recommended by the manufacturer.

Antibodies and western analyses
Polyclonal anti-YCTD antibody, produced in rabbits, recognizes endogenous RPB1 as well as the full-length, 2/3, and 1/3 YCTD RPB1 derived from transgenes. All four transgene-derived tagged-RPB1 proteins are also detected with monoclonal anti-Ty1 antibody, generated in the Gull laboratory and kindly provided by the Cross laboratory [30]. Anti-RPB4 antibody, produced in rabbits from recombinant protein as antigen, was used for immunoprecipitation experiments. For immunoprecipitations, nuclear extracts were incubated with anti-RPB4 antibody bound to protein A-sepharose, washed in buffer (20 mM Hepes [pH 7.9], 150 mM KCL, 150 sucrose, 2.5 mM MgCl 2 , 1 mM EDTA, 2.5 mM DTT, 0.1%NP-40, 1 mM each of pepstatin A, leupeptin, and PMSF) and analyzed by immunoblotting using ECL TM kit from Amersham.

RNA analysis
Total cellular RNA from wild-type and transgenic cells was prepared using Trizol TM reagents. For Northern analysis, 5 mg of total RNA was separated on a formaldehyde-1.2% agarose gel in MOPS buffer [32]. RNA was fragmented by alkali-treatment, transferred to nylon membranes and hybridized with 32 P-labelled oligonucleotides. Steady-state SL and U2 RNA levels were determined by primer extension analysis, using 32 P-labeled oligonucleotides complementary to these RNAs. Nascent RNA was synthesized in lysolecithin-permeabilized cells using 32 P-UTP [33] and profiled on 7M urea 10% polyacrylamide gel. Short transcripts were extracted from the gel, purified according to [34] and resolved on a 7M urea 25% polyacrylamide gel. Tobacco acid pyrophosphatase (Epicentre) and T4 RNA ligase (New England Biolabs) were used according to manufactures' protocols. Data were captured on a Typhoon PhosphorImager TM and quantification was performed using ImageQuant TM software.

Microscopy
An Olympus BX61 microscope with Nomarski DIC optics was used to visualize live cells.