Transcription Termination and 3′-End Processing of the Spliced Leader RNA in Kinetoplastids

Abstract

Addition of a 39-nucleotide (nt) spliced leader (SL) by trans splicing is a basic requirement for all trypanosome nuclear mRNAs. The SL RNA in Leishmania tarentolae is a 96-nt precursor transcript synthesized by a polymerase that resembles polymerase II most closely. To analyze SL RNA genesis, we mutated SL RNA intron structures and sequence elements: stem-loops II and III, the Sm-binding site, and the downstream T tract. Using an exon-tagged SL RNA gene, we examined the phenotypes produced by a second-site 10-bp linker scan mutagenic series and directed mutagenesis. Here we report that transcription is terminated by the T tract, which is common to the 3′ end of all kinetoplastid SL RNA genes, and that more than six T’s are required for efficient termination in vivo. We describe mutants whose SL RNAs end in the T tract or appear to lack efficient termination but can generate wild-type 3′ ends. Transcriptionally active nuclear extracts show staggered products in the T tract, directed by eight or more T’s. The in vivo and in vitro data suggest that SL RNA transcription termination is staggered in the T tract and is followed by nucleolytic processing to generate the mature 3′ end. We show that the Sm-binding site and stem-loop III structures are necessary for correct 3′-end formation. Thus, we have defined the transcription termination element for the SL RNA gene. The termination mechanism differs from that of vertebrate small nuclear RNA genes and the SL RNA homologue in Ascaris.

The spliced leader (SL) RNA is central to kinetoplastid nuclear gene expression. The SL RNA, or mini-exon-derived RNA, is a primary transcript that is synthesized independently of the pre-mRNA and trans spliced onto all nuclear mRNAs (1), most of which are synthesized as polycistronic precursors. SL RNA transcription represents approximately 10% of total RNA synthesis (10, 42). The approximately 100 copies of the SL RNA gene are tandemly repeated in a head-to-tail manner in the chromosomal locus; the transcription of each gene is directed by an upstream promoter (26, 30, 54, 62). It has been demonstrated by nuclear run-on analysis that transcription does not proceed into the SL RNA gene-flanking region (40, 62); therefore, the intergenic region (256 bp of a 363-bp repeat in Leishmania tarentolae, 1.2 kb of a 1.35-kb repeat in Trypanosoma brucei) can be considered a nontranscribed spacer (62). The presence of some termination element near the 3′ end of the SL RNA sequence is thus indicated experimentally.

Although the 39-nucleotide (nt) SL exon is well conserved in two domains, the primary sequence of the SL RNA intron is not conserved among the trypanosomes (73). However the secondary structure of the SL RNA, composed of three stem-loops and a single-stranded region containing a putative Sm-binding site (Fig. 1A), is consistent (11). This structure has been confirmed by physical-chemical and enzymatic studies (29, 44) and examined by mutagenesis (47). An equivalent structure is also conserved in the nematode SL RNAs (11, 53). The Sm-binding site (11) is an element found in the U-RNAs of higher eukaryotes but, with the exception of U5 (77), apparently lacking in the small nuclear U-RNAs of kinetoplastids (50).

FIG. 1.

Secondary structure of the L. tarentolae SL RNA and linker-scanning mutations in the intron. (A) Secondary structure of L. tarentolae SL RNA, based on the L. collosoma structure (form II) that predominates in vivo (29, 44); numbering is relative to the start of transcription. The stem-loop structures are labeled I, II, and III, and the exon-intron junction is indicated by the arrow after nt 39. The Sm-binding-site sequence AUUUUGG is indicated. The 7meG cap is shown at the 5′ end, along with the methylated nucleotides (∗) that comprise the cap 4 structure (21, 55). (B) A linker-scanning series was generated through the intron of the L. tarentolae SL RNA gene and beyond. Numbering is relative to the start site of SL RNA transcription. All constructs contained an exon tag (tSL) sequence at nt 28 to 39, indicated by lowercase letters. The scan sequence, CTCGAGCTCA, included XhoI and SacI sites; positions within the 10-bp blocks that were not altered from WT are indicated by dots. The mature 3′ end of the SL RNA is indicated.

A poly(T) tract is a common feature of the 3′ end of SL RNA genes in kinetoplastids. The poly(T) tract ranges in length from 5 T’s in Leishmania donovani (18) to 31 in Trypanosoma cruzi (15). This homopolymeric motif has been postulated to be a termination element for transcription of the SL RNA genes. The ability of four T’s to terminate RNA polymerase III (pol III) transcription has been used as an argument in favor of pol III transcription of SL RNA genes (24).

No pol II transcription terminators have been identified in trypanosomatids. In vertebrates, small nuclear RNA snRNA genes transcribed by pol II possess conserved downstream sequence elements that determine 3′-end formation, most probably through transcription termination (33). Although a 12-of-13 match with the vertebrate U snRNA transcription termination box was found 9 nt downstream of the SL RNA gene coding region in Ascaris (53), mutagenesis of this box and beyond did not affect SL RNA 3′-end formation (28), leading the authors to conclude that the signals for 3′-end formation lay entirely within the transcribed region and adjacent 12 nt.

The identification of the polymerase that is responsible for the transcription of the SL RNA gene has been approached by several groups with at least three kinetoplastids. Drug inhibition studies with disrupted or permeabilized cells and in vitro transcription systems support pol II transcription by the criteria of Tagetitoxin (26, 62) and Sarkosyl (60, 62) responses. Similar studies found different results with α-amanitin (pol II [5, 41], pol III [24], and unclear [38]) and 1,10-phenanthroline with or without Mn²⁺ (pol II [61] and pol III [24]). The presence of putative box A and box B pol III promoter elements in the SL RNA gene has also been described; however, it has since been shown that disruption of these elements does not affect transcription (2, 47, 54, 62). Studies of the K⁺ ion requirement support pol III transcription (70). Because these studies were carried out with several kinetoplastids under variable reaction conditions, it is difficult to resolve the polymerase issue; however, in L. tarentolae, three of four indicators tested indicated pol II activity while the fourth was ambiguous (62).

We have recently identified the core promoter elements of the L. tarentolae SL RNA gene (80). To define SL RNA gene termination and 3′-end formation elements that might be distinct from elements related to trans splicing, we took an in vivo approach to challenge various structures and sequence elements by using stable transfection of L. tarentolae cells with exon-tagged markers. The results obtained in vivo were complemented by using an in vitro transcription assay. In this paper, we report that the SL RNA intron and poly(T) tract have distinct effects on transcript termination and 3′-end formation but do not alter transcriptional initiation. We demonstrate that transcription is terminated in the poly(T) tract and is followed by removal of 6 to 8 nt to the base of stem-loop III, which is dependent on both Sm-binding site sequences and a stem-loop III structure. Transcription termination, but not the 3′-processing activity, was achieved in transcriptionally competent L. tarentolae nuclear extracts.

MATERIALS AND METHODS

Generation of mutations and transfectants.

Mutagenesis of the L. tarentolae SL RNA gene was performed with the Sculptor mutagenesis kit (Amersham) or by PCR (62). Mutated fragments were cloned into the pX transfection plasmid (43). L. tarentolae transfections were performed by electroporation as described previously (66).

Nucleic acid isolation and gel analysis.

RNA was purified with TriZOL reagent (Gibco/BRL) and was electrophoresed through 6% (0.4 mm thick, 40 cm long for high resolution) or 8% (1.5 mm thick, 15 cm long for medium resolution) acrylamide–8 M urea gels, blotted, and hybridized as previously described (62, 67). Quantitation was performed with a PhosphorImager (Molecular Dynamics).

5′- and 3′-end analyses.

Transcript initiation was assayed by primer extension of total RNA as described previously (62, 66).

The 3′-end mapping of wild-type (WT) and mutant SL RNAs was performed by poly(A) tailing total RNA followed by reverse transcription with an oligo(dT) primer and PCR with the oligo(dT) primer and SL-specific oligonucleotide LtSL5′RI or 30/39-5′HI (66) as described previously (58).

In vitro transcription assay.

Nuclear extracts were prepared and in vitro transcription reactions were carried out for L. tarentolae essentially as described for T. brucei (25, 26).

RESULTS

Intron mutations do not affect transcription initiation.

To localize specific elements within the primary sequence or secondary structure of the SL RNA (Fig. 1A) that play a role in the maturation of the molecule, a systematic mutagenesis approach was adopted (Fig. 1B). To differentiate mutants from the endogenous SL RNA population, an exon mutated at positions 28 and 30 to 39 (28/39), which was previously shown to trans splice accurately and efficiently (66), was used as a molecular tag (tSL RNA) for detection by hybridization. A series of 10-bp linker-scanning (CTCGAGCTCA) mutations in the tSL RNA gene was created between positions 42 and 129 (Fig. 1B) for transfection as stable episomes in L. tarentolae. Two mutations in the region from positions 42 to 49 were created: a mutant with alterations in bases 43 and 44 (43/44) was used to test a postulated SL RNA-U6 snRNA interaction (76), and a second mutant altered at positions 42 to 48 (42/48) changed all but the splice donor site with the equivalent linker-scanning sequences. Subsequent intron mutations continued from position 50 (52/59) and proceeded through the end of the intron. Three mutations lay downstream of the mature 3′ end of the SL RNA transcript (position 96) and were included to identify possible adjacent expression elements.

To examine the tSL RNA expression, total-cell RNA was separated on a medium-resolution gel system (Fig. 2). All tSL RNA genes were transcribed. RNA from mutants 42/48, 43/44, 52/59, 62/69, 100/109, 110/119, and 120/129 yielded the typical 96-nt tSL RNA. The 3′-most intron mutants, 70/79, 80/89, and 90/99, showed discrete increases in the tSL RNA length of less than 10 additional nucleotides, plus some faint higher-molecular-weight products that migrate above 215 nt. 100/109 showed, in addition to the 96-nt tSL RNA, a pattern of at least eight larger transcripts (∼110 to 220 nt).

FIG. 2.

Mutated tSL RNA genes are transcribed, but mutations in the 3′ end of the gene yield tagged-SL RNAs with altered sizes. Gels consisting of 8% acrylamide and 8 M urea provided medium-range resolution over 50 to 500 nt. Size markers correspond to the small rRNA fragments and tRNAs visualized by ethidium bromide staining of the gel prior to transfer (data not shown).

Intron mutations affect 3′-end and cap 4 formation.

Since gene-internal elements can affect the specificity of transcription initiation (27, 69), we tested the possibility that the changes in transcript size were due to incorrect transcription initiation. Primer extension analyses were performed with an oligonucleotide complementary to the exon tag sequence (Fig. 3). Extension products from accurately initiated and methylated tSL RNAs are expected to terminate predominantly 5 nt downstream of the true 5′ end due to base methylation at the fourth position in the cap 4 structure (21, 55). Similar to WT SL RNA methylation in L. tarentolae (62), the control tSL RNA showed approximately 75% extension to position +5, as did the downstream mutations (110/119 and 120/129). Although all tSL RNAs showed primer extension products consistent with initiation at +1, only 42/48 and 43/44 conformed with the extension pattern expected for cap 4 methylation. 100/109 showed 50% methylation at position +4. The tSL RNAs mutated from 52/59 through 90/99 showed greatly reduced methylation (<5%) at +4 but appeared to retain base methylation of nucleotide +1, based on the predominant extension products at position +2.

FIG. 3.

Tagged-SL RNAs initiate accurately but have reduced cap 4 methylation. Primer extension analysis was performed with γ-³²P-labeled 28/39-tag oligonucleotide. The interpretation of the extension products is indicated in schematic form to the right of the data, where cap 4 structures including the 7meG cap, 2′-O-ribose methylations (m), and base methylations on nt 1 (m6,6) and 4 (m,3) are indicated.

Since transcription initiation appears to be unaffected, these data indicate that the 100/109, 70/79, 80/89, and 90/99 mutations affect SL RNA 3′-end maturation.

Sm-binding site and stem-loop III mutations result in extended 3′ ends.

To determine the precise sizes of the transcripts, the tSL RNAs were examined in single-nucleotide resolution gels (Fig. 4A). The tSL, 43/44, 100/109, 110/119, and 120/129 RNAs resolved predominantly as single bands. Doublet bands of equal intensity were observed for 42/48, 52/59, and 62/69 (see the shorter exposure of 62/69 in the right panel), in which the upper band corresponded to the band seen in tSL. The 70/79, 80/89, and 90/99 tSL RNAs were larger and comprised four major bands each. Based on the uniform 5′ ends mapped by primer extension and on the ladder pattern of the single-nucleotide-resolution gel, the 3′ ends of the 70/79, 80/89, and 90/99 transcripts terminated in a staggered fashion within the poly(T) tract, corresponding U residues 5 to 8.

FIG. 4.

Staggered 3′ ends in the poly(T) tract. (A) Single-nucleotide size heterogeneity visible at high resolution. RNA samples were electrophoresed in a 6% acrylamide–8 M urea sequencing gel, blotted, and hybridized with the 28/39-tag oligonucleotide. (B) In the steady-state population, a minor percentage of WT SL RNA shows the poly(T)-tract termination profile. Total RNA from WT L. tarentolae and poly(T)-tract termination mutant 70/79 were resolved in a 6% acrylamide–urea gel, blotted, and probed with LtSLintron, which hybridizes to positions 40 to 59 or 28/39-tag oligonucleotides, respectively. The WT lane is overexposed to visualize additional SL RNA species.

The 3′ ends of the SL and tSL RNAs were mapped by poly(A)-RT-PCR (data not shown). The 3′ end of tSL and WT transcripts coincided and mapped to nt 96 to 97; the ambiguity was due to the presence of an A in the sequence at position 97. Products with extra 3′ U residues corresponding to poly(T)-tract termination products were found in 70/79, 80/89, 90/99, and SL RNA populations.

WT SL RNA poly(T)-tract termination.

Because transcription initiation was accurate, the lengths of the tSL RNAs seen in 70/79, 80/89, and 90/99 suggested that transcription termination occurs in a staggered fashion in the poly(T) tract downstream of the SL RNA gene. Consistent with this proposal, transcription of the SL RNA does not proceed significantly beyond the poly(T) tract in nuclear run-on assays (40, 62). To determine if WT SL RNAs also terminate in the poly(T) tract, WT and 70/79 total RNA was resolved in a 6% acrylamide–urea gel (Fig. 4B). An overexposure of the WT lane showed the 96-nt SL RNA and a collection of larger bands that mirrored the tSL RNA accumulation pattern in 70/79, indicating that poly(T) tract termination products are produced during normal SL RNA synthesis.

These results suggest that formation of the SL RNA 3′ end is a two-step process: transcription termination in the poly(T) tract followed by nucleolytic processing to/at the base of stem-loop III. Alternatively, two distinct termination signals may be operating in the WT SL RNA, one leading to the 96-nt transcript directly and the second resulting in the staggered termination in the poly(T) tract, or 3′-end formation may be a three-step process with transcription termination occurring downstream of the poly(T) tract followed by primary cleavage within the poly(U) tract and secondary cleavage at the mature 3′ end.

Transcription termination by the 3′ poly(T) tract in vivo.

The poly(T) tract location of 3′ ends in some of the tSL RNAs, coupled with the increased length in the 100/109 mutant, suggested that the poly(T) tract itself was a termination element. Transcription termination mediated by poly(T) tracts is a hallmark of pol II (37) and pol III (8). Since the identity of the RNA polymerase that transcribes the SL RNA gene has been a subject of some contradictory data and since little is known about transcription termination in trypanosomatids, we performed a mutational analysis of the poly(T) tract.

Because 100/109 (2 T’s) showed extended 3′-end products, the length of the poly(T) tract was altered in two nucleotide increments (Fig. 5A). These mutations were designed to define the number of T’s necessary for termination. An 8-bp poly(T) tract terminated transcription efficiently, while 2-, 4-, and 6-bp poly(T) tracts accumulated higher-molecular-weight bands (Fig. 5B). The largest transcript (∼220 nt) corresponds approximately to a 3′ end within a pyrimidine-rich region in the pX sequence (TTTTGTTCCCTTT, at positions 201 to 213 relative to the start of transcription; positions 1415 to 1427 on the pX vector). The other products may correspond to termination by a low-processivity polymerase, to pause sites in an exonucleolytic pathway, or to secondary endonucleolytic cleavage sites exposed due to the transcription of the downstream sequences. The sizes of the high-molecular-weight products are not identical in all samples; this may be the result of secondary- or tertiary-structure effects on polymerase processivity or nucleolytic processing.

FIG. 5.

The poly(T) tract is a transcription terminator. (A) Diagram of the poly(T)-tract scan mutation 100/109 and additional mutations reducing or relocating the poly(T) tract. The poly(T) tract was reduced in situ to 2, 4, 6, or 8 T’s. A second poly(T) tract was introduced at position 150/159 with or without the 100/109 scan mutation. (B) At least 6 T’s are required for efficient transcript termination. An 8% acrylamide–urea gel was used to visualize intermediate-molecular-weight termination products. The size markers are an MspI digest of pBR plasmid that was denatured and run alongside the RNA samples and a 96-nt marker, which is based on the SL RNA. (C) Staggered termination and discrete processing sites visible at high resolution. A 6% acrylamide–8 M urea sequencing gel was used to separate tSL RNA products that were blotted and probed with 28/39 tag oligonucleotide. A DNA sequencing reaction was run alongside to provide an approximate size ladder.

To test further the termination ability of the poly(T) tract motif, a 10-bp poly(T) tract was relocated 52 bp downstream of the 2 T mutation (100/109 + 150/159 [Fig. 5A]). As predicted for a termination element, the poly(T) tract displacement mutant displayed a transcript of approximately 160 nt; it also showed low levels of 96-nt SL RNA (Fig. 5B). High-resolution analysis of the products revealed a heterogeneous group of transcripts in the 160-nt size range (Fig. 5C). The absence of high-molecular-weight products in the 150/159 double-poly(T) tract control mutant indicated that normal termination is highly efficient in vivo.

Examination of the termination products at high resolution (Fig. 5C) allowed further differentiation among the transcripts. The products terminating in the poly(T) tracts showed a staggered group of bands, as seen in tSL, 106/107 (nt 100 to 104), and 100/109 + 150/159 (nt 153 to 157) and faintly in the intron-tag-containing mutant tSL+IT (nt 151 to 155) (62). The additional products seen around nt 116 in lanes 100/109 and 100/109 + 150/159 may be termination products, although their absence from the other lanes containing mutants RNA suggests that they are the result of secondary processing events. The remaining higher-molecular-weight bands also displayed discrete migration patterns, suggesting that they are also products of secondary 3′-processing events.

These results are consistent with a model in which transcription of the SL RNA gene terminates in homopolymeric poly(T) tracts of greater than 6 nts. The phenotype of the poly(T) tract relocation mutant demonstrates that transcription termination is caused directly by the presence of a homopolymeric T stretch.

Poly(T) tract termination with in vitro transcription extracts.

We have demonstrated accurate transcription initiation from the SL RNA gene by using in vitro transcription assays (79) (Fig. 6A) similar to those described for Trypanosoma and Leptomonas (26, 35); as in the related systems, cap 4 methylation is lacking in L. tarentolae extracts. Because in vitro products could be detected by primer extension and blotting methods, we used this assay to monitor transcription termination. The in vitro products from the constructs with altered poly(T) tracts (Fig. 5A) were compared with their in vivo counterparts and analyzed by gel electrophoresis and RNA hybridization (Fig. 6B) for transcription termination efficiency as measured by an all-or-nothing appearance of bands in the in vivo samples relative to the in vitro samples. We first noted that the in vitro tSL-positive control did not comigrate with the in vivo tSL product but migrated with transcripts that had staggered 3′ ends in the poly(T) tract. The differential migration cannot be ascribed to variable 5′-end methylation, since the latter does not affect electrophoretic mobility (71). We also noted three bands near the top of the gel at ≥300 nt in vitro that were not present in the in vivo RNA samples. This transcript pattern was also seen for in vitro products from 52/59 and 62/69 (data not shown), which resulted in WT-sized transcripts in vivo.

FIG. 6.

In vitro transcription termination directed by poly(T) tracts. (A) Accurate initiation of in vitro transcription system. Primer extension products from exon-tagged in vitro transcription products were run on a 6% acrylamide–8 M urea sequencing gel next to a sequencing ladder from a cloned tSL RNA gene. (B) An 8% acrylamide–urea gel was used to separate in vitro transcription products on the basis of total size. In vivo RNA products (boxed) were electrophoresed in the same gel for comparison.

In vitro, the products from the templates with altered poly(T) tracts reflected the in vivo results in that the staggered poly(T)-tract termination pattern was abolished by 6 or fewer T’s (100/109, 102/107, and 104/107); however, the accumulating products were significantly larger (≥250 nt) than those seen in vivo (<215 nt). The constructs with 8 and 10 T’s (106/107 and tSL) led to poly(T)-tract-mediated termination, as seen by the accumulation of poly(T)-tract products, but was not as efficient as in vivo, since some higher-migrating bands were also evident. The poly(T)-tract displacement mutation (100/109 + 150/159) showed primary termination products that corresponded to those seen in vivo at around 155 nt, but none of the smaller accumulation products, including no WT tSL RNA; small amounts of the ≥250-nt in vitro-specific products were also visible. The 150/159 double poly(T)-tract mutant accumulated products at both WT and the displaced poly(T) tracts, confirming that in vitro termination was less efficient than in vivo termination.

These results demonstrate the accurate and efficient termination in nuclear extracts directed by poly(T) tracts of ≥8 nt. The lack of WT-sized products in any of the in vitro products supports the conclusion that primary transcript termination is mediated by the poly(T) tract alone. In addition, the absence of minor products in the range of 110 to 200 nt seen in the in vivo termination knockout mutants (Fig. 5B) indicates that the presence of in vivo products is due to processing and not to primary termination events. The combined results of the in vivo and in vitro studies indicate that transcriptional termination and 3′ maturation are two distinct events. While it is formally possible that the in vitro transcription system does not recognize some alternative termination signal in the Sm-binding site or stem-loop III region, it is unlikely based on additional mutagenesis studies (see below).

The Sm-binding site affects 3′-end formation in vivo.

Because 70/79 contained sequences affecting both 5′- and 3′-end formation and altered most of the Sm-binding site, we examined this sequence block in fine detail. A 2-bp transversion series was created from positions 70 through 81; in addition, 70/79 and 75/81 transversion mutations (TV) were made (Fig. 7A). Mutations 70/71 TV and 72/73 TV comprise part of stem-loop II (Fig. 1A).

FIG. 7.

Position 76 marks the start of an element for 3′-end formation. (A) Mutations in the 70/79 region and Sm-binding site. Nucleotide transversions (TV) were used to maximize structural disruption. (B) Mutations beyond position 76 are unable to form mature SL RNA 3′ ends. An 8% acrylamide–urea gel was used to examine the mature sizes of the 70- to 81-region mutants; the hybridizing portion of the blot is shown. The sizes correspond to the SL RNA (96 nt) and the average size of poly(T)-tract termination products (105 nt).

Mutants 70/71 TV and 74/75 TV displayed comparable migration to the tSL RNA control (Fig. 7B). In contrast, mutants 70/79 TV, 75/81 TV, 76/77 TV, 78/79 TV, and 80/81 TV produced exclusively (or predominantly in the case of 80/81 TV) transcripts extending into the poly(T) tract. Mutant 72/73 TV contained the 96-nt transcript and increased levels of 3′-extended transcripts. Thus, of the AUUUUGG consensus, the A was not essential for 3′-end formation, the 4 U’s were essential, and mutation of the 2 G’s resulted in an intermediate phenotype.

The structure of stem-loop III is necessary for proper 3′-end maturation.

Mutations 80/89 and 90/99 disrupted stem-loop III and resulted in larger, undermethylated tSL RNAs (Fig. 3 and 4A). To examine further the effect of stem-loop III structure, a series of mutations that either disrupted or compensated 1 or 3 bp of the stem and that altered the loop sequence independent of the stem were created (Fig. 8). The single-base disruptions were expected to disrupt only one rung of the stem and thus lead to a minor size difference in the tSL RNA, while the triple-base mutations were designed to disrupt the stem thermodynamically.

FIG. 8.

The structure of stem-loop III directs 3′-end processing. The stem-loop III structure and its disruption by various mutations are shown. The scan mutations 80/89 and 90/99 are shown in Fig. 1B; single (83 and 96) and triple (83/85 and 94/96) stem disruptions and their compensatory stem structure mutations (83 + 96 and 83/85 + 94/96) are shown here. A loop transversion mutation was also created (88/91). All mutated positions are boxed. (A) Single-nucleotide mutants show minor size differences. The 83, 96, and 83 + 96 mutant SL RNAs were resolved in a high-resolution 6% acrylamide–urea gel. The sizes of the major products are indicated. (B) Triple-nucleotide stem disruptions abolish 3′-end formation. A medium-resolution 8% acrylamide–urea gel was used to analyze triple-base-pairing mutants and the effect of structural restoration (83/85 + 94/96) on 3′-end formation. The relevant portion of the blot is shown. The sizes of the major products are indicated. (C) In vitro products from IT-containing templates show reduced processing. The position of the IT in the loop of stem-loop III is shown schematically. An 8% acrylamide–8 M urea gel shows the comigration of the in vitro product with the in vivo (boxed) −18 artifact band described previously (62, 80).

Analysis of the 83 and 96 tSL RNAs in high-resolution gels showed minor size shifts that were completely compensated by the 83 + 96 double mutation (Fig. 8A): 83 shifted 0.5 nt up, and 96 shifted 1 nt down. Mutant 96 also showed a higher accumulation of poly(T)-tract termination products that may reflect impaired 3′ processing. Mutants 83/85 and 94/96 displayed size heterogeneity around 104 nt, corresponding to 3′ ends in the poly(T) tract. This heterogeneity was not seen in the 96-nt products of the compensating double mutation 83/85 + 94/96 (Fig. 8B). The loop III mutation, 88/91, did not show a size shift. Thus, the stem structure but not the sequence of stem-loop III is important for 3′-end formation.

The availability of the in vitro transcription system allowed us to address a problematic artifact seen in past studies relative to transcription termination and 3′ processing. Our previous studies used SL RNA gene constructs containing an intron tag (IT) to monitor SL RNA gene transcription and expression (62, 66, 80). The location of this tag is within the loop of stem-loop III (Fig. 8C); the expression of the constructs was accompanied consistently by an additional artifactual band that was visible in blotting analyses only in combination with an active promoter (62, 80). The presence of primer extension products at nt −12 to −18 in the presence (62, 66) and absence (66) of SL RNA gene promoter activity prompted the suggestion that the artifacts were the same and were a by-product of the runaround transcription in the expression vector pX (62). However, when the IT is placed in the stem-loop II structure (61) or removed altogether (66), the −18 artifact does not appear in blotting analyses. Thus, the placement of the IT in stem-loop III may be responsible directly for the additional band seen in blotting analyses, perhaps by interfering with 3′-end formation. This hypothesis was tested by comparing IT-containing products generated in vivo to products synthesized by the in vitro transcription system (79) (Fig. 6). The in vitro products comigrated with the −18 product, and no mature-sized product was seen (Fig. 8C). Thus, we conclude that the −18 artifact is due to the accumulation of poly(T) tract termination products, consistent with its association with active promoter elements and with impaired 3′ processing due to the introduction of an additional 44 nt to the loop of stem-loop III, and does not correspond to the upstream primer extension products.

DISCUSSION

We demonstrate that the SL RNA intron structure, content, and downstream poly(T) tract affect SL RNA termination and processing in an independent fashion. In L. tarentolae, SL RNA is most probably transcribed by pol II (62). Most aspects of pol II transcription in trypanosomatids are poorly understood, and little is known about how pol II transcription terminates in trypanosomatids (6, 39) and other organisms (64). However, termination of small RNA gene transcription is controlled by a well-defined 3′ box (64). Transcription termination and 3′-end formation of the SL RNA can be separated into a two-step process: staggered transcription termination in the 3′ poly(T) tract enhanced by the presence of stem-loop III, followed by nucleolytic processing to a defined 3′ end at position 96 that is dependent on stem-loop III and the Sm-binding site.

Regarding the important question of the identity of the polymerase that transcribes the SL RNA gene, it should be noted that relatively high α-amanitin resistance in pol II genes has been documented in kinetoplastids (5, 24, 38, 41) as well as in other lower eukaryotes (45, 57). The resulting differences in pol II and pol III α-amanitin sensitivities create difficulties in the discrimination between these two polymerases on the basis of this criterion, probably accounting for many of the conflicting conclusions seen in the literature. Since the data from most inhibitors support pol II transcription of the SL RNA gene in L. tarentolae (62), we will proceed with our discussion accordingly.

Termination and the poly(T) tract.

Transcription termination by pol II has been studied primarily on protein-coding genes (74). The transcription termination signals are bipartite and include the polyadenylation signal (7) in conjunction with a variety of downstream elements, such as poly(T) tracts with intrinsic bent DNA but no adjacent upstream secondary structure (37), trans-acting factors binding to the DNA (4, 13, 14) or the RNA polymerase (75), or the structure of the transcript (56). Since long (>6-nt) poly(T) tracts may cause pausing but should not effect pol II termination on protein-coding genes, a different mechanism must exist for mRNA termination in trypanosomatids. Because transcription of kinetoplastid pre-mRNA is polycistronic and includes extensive polypyrimidine tracts as part of the trans-splicing signals, the factors involved in termination would have to be differentially recruited by the same polymerase. In this regard, it has been suggested that transcription driven by the T. brucei SL RNA gene promoter is terminated by the actin gene upstream polypyrimidine tract (49). In T. cruzi, transcription of an SL RNA-chloramphenicol acetyltransferase gene is driven by the SL RNA gene promoter, perhaps because the construct lacks the polypyrimidine splicing signal and has no poly(T) tracts longer than 5 nt in its coding region and 7 nt in the flanking sequences (46); it should also be noted that T. cruzi SL RNA genes have particularly lengthy poly(T) tracts, up to 31 nt (15), which may reflect a functional constraint for termination in this organism.

In the case of pol III, transcription terminates in a 4 T tract and is processed to the mature 3′ end (8); termination is stimulated by La autoantigen (22, 23). In higher eukaryotes, termination-polyadenylation elements function only in the presence of the cognate snRNA-mRNA promoter (33, 34, 52). With the identification of the SL RNA gene termination element, a similar association between the SL RNA gene promoter and terminator elements can now be tested functionally in vivo and in vitro.

The role of the poly(T) tract located at nt 98 to 107 as a transcription terminator is clear, since its disruption results in a variety of 3′-extended transcripts and since poly(T)-tract relocation 52 bp downstream generates the predicted ∼155-nt products (Fig. 5). As a result of the disruption and relocation experiments, transcription proceeds into the flanking region, which has been shown in nuclear run-on assays to constitute a nontranscribed spacer between the tandemly arranged SL RNA genes (40, 62). The requirement of at least 6 T’s for efficient transcriptional termination of the SL RNA gene in vivo and at least 8 T’s in vitro argues against transcription by pol III, which terminates transcription at runs of 4 to 8 T’s that have GC-rich flanking regions and lack flanking AA nucleotide pairs (8). Since tRNA and 5S genes in trypanosomatids generally follow these rules (see below), pol III might be expected to terminate prematurely at the 4 T’s located in the Sm-binding site. In addition, by the same analogy, we would expect pol III to terminate transcription efficiently in the 4-T poly(T)-tract mutant (102/107), which lacks AA nucleotide pairs in the immediate flanking regions. Transcription termination at a poly(T) tract downstream of the SL RNA gene and the presence of an upstream modulating stem-loop structure are reminiscent of rho-independent termination in Escherichia coli and provide an alternative mechanism for pol II transcription termination for small-RNA genes in eukaryotes. Our model of pol II termination on the trypanosomatid SL RNA gene differs from that used for eukaryotic small-RNA genes that are transcribed by pol II and have downstream GTTN_1–4AANARNAGA elements (32). Although present downstream of the pol II-transcribed SL RNA gene in Ascaris, the 3′-box sequence does not function as a terminator; termination signals lie within the transcribed region (28).

A variety of poly(T) tracts are found downstream of the SL RNA genes in kinetoplastids. Within the leishmanias, L. donovani displays a range of 5 to 9 T’s, L. major ranges from 8 to 9 T’s, L. mexicana ranges from 8 to 10 T’s, L. braziliensis ranges from 10 to 13 T’s, and L. guyanensis, L. panamensis, and L. naiffi have interrupted T tracts T₉CT₁₀, T₁₁CT₆, and T₈CT₁₃, respectively (18). The trypanosomes display large homopolymeric poly(T) tracts ranging from 20 to 31 T’s in T. cruzi (15, 19) and tracts interrupted by purine residues such as T₅AT₈AT₇ in T. brucei (12). Purine interruptions are also seen in the nonpathogens Blastocrithidia (T₁₀AT₁₃) (17), Crithidia desouzai (T₄GT₃) (20), and Herpetomonas (T₁₀AT₈) (3).

Because the possibility remains that the SL RNA gene is transcribed by pol III, we include a summary of putative pol III termination elements in L. tarentolae. Eighteen tRNA genes and two 5S rRNA genes have been characterized (48, 63, 68). The tRNA genes are presumed to be transcribed by pol III due to the presence of box A/box B or box A/box C motifs and downstream poly(T) tracts. Kinetoplastid tRNAs are typical in their transcription profiles with regard to the inhibitors α-amanitin and Tagetitoxin (16, 62). Most of the L. tarentolae tRNAs are encoded in clusters, and transcription termination is likely to be effected by 4 to 6 T residues present downstream of the transcribed region, although one exception is found in relative isolation and has a putative termination signal of T₈AC₇ (48). All the poly(T) tracts lack adjacent AA dinucleotides, and the majority are flanked by GC-rich sequences. Among other genes in L. tarentolae presumed to be transcribed by pol III promoters in associated tRNA genes (51), U4 ends in a run of 4 T’s (61), while U-RNA B (the U3 homolog) has a stretch of 11 T’s downstream of its transcribed region (63) and U2 has a stretch of 9 T’s (78). The increased poly(T) tract length may be relevant in the U-RNA B gene because the transcript itself has an internal stretch of 4 T’s. Functional assays will be necessary to determine if any of these genes require more than 4 T’s for pol III termination.

3′-end processing: Sm-binding site and stem-loop III.

Disruption of the UUUUGG Sm-binding-site sequences or stem-loop III structure results in a loss of 3′-end processing in L. tarentolae (Fig. 7 and 8). Single-nucleotide disruptions of stem-loop III produce slightly modified 3′ ends, but major disruptions of the stem portion of stem-loop III result in the accumulation of poly(T)-tract termination products ranging in size from 5 to 9 nt longer than the mature, WT transcript.

The presence of 3′-extended products in stably infected mutants (70/79, 80/89, and 90/99) suggests that transcription termination and 3′-end processing are separate events. This model is supported by the inability of an in vitro transcription extract to 3′ process transcripts terminated in the poly(T) tract. The lack of 3′ and 5′ processing of tSL RNAs in the in vitro transcription nuclear extract suggests that both of these processing events occur in the cytoplasm of the cell rather than in the nuclear compartment. Consistent with this hypothesis and the nature of snRNP-processing pathways in other systems (36), fractionation studies indicate that the kinetoplastid SL RNA is present in its unspliced form in the cytoplasm as well as in the nucleus (31, 59). The lack of 5′ methylation seen in our in vitro transcription system mirrors the results obtained in T. brucei (26). However, in other studies, nuclear extracts were found to perform as well as whole-cell extracts did for the 5′ methylation of in vitro-transcribed SL RNA (72). Future studies will explore the ability of Leishmania cytoplasmic, nuclear, or whole-cell extracts to process poly(T)-tract-terminated tSL RNA substrates and to identify the nuclease responsible. 3′ processing is a critical step for SL RNA maturation, since unprocessed tSL RNAs are unable to participate in the trans-splicing reaction (65).

The mode of 3′ processing to the base of stem-loop III remains to be determined. It is likely to be an enzymatic reaction that could include endonucleolytic attack at the base of secondary structures (for example, RNase P processing of the 5′ ends of tRNAs of Escherichia coli) or exonucleolytic trimming (for example, exonucleolytic trimming of U6 [9]). In this regard, we note that the related kinetoplastid T. cruzi possesses an exonuclease activity that removes uridine but stops at adenosine (81), which is a candidate for SL RNA 3′ processing.

Given the general lack of knowledge about transcription in trypanosomes, our work on the promoter (80) and transcription termination elements is the first step toward a global understanding of pol II function in these primitive eukaryotes. The definition of these elements can now be exploited for continued study of the SL RNA itself, which is so central to kinetoplastid pre-mRNA processing. In addition, these elements can be used to drive the expression of other RNAs, natural or engineered, that will expedite the study of other genes from a functional rather than transcriptional point of view.