|Home | About | Journals | Submit | Contact Us | Français|
Thermococcus kodakarensis (formerly Thermococcus kodakaraensis) strains have been constructed with synthetic and natural DNA sequences, predicted to function as archaeal transcription terminators, identically positioned between a constitutive promoter and a β-glycosidase-encoding reporter gene (TK1761). Expression of the reporter gene was almost fully inhibited by the upstream presence of 5′-TTTTTTTT (T8) and was reduced >70% by archaeal intergenic sequences that contained oligo(T) sequences. An archaeal intergenic sequence (tmcrA) that conforms to the bacterial intrinsic terminator motif reduced TK1761 expression ~90%, but this required only the oligo(T) trail sequence and not the inverted-repeat and loop region. Template DNAs were amplified from each T. kodakarensis strain, and transcription in vitro by T. kodakarensis RNA polymerase was terminated by sequences that reduced TK1761 expression in vivo. Termination occurred at additional sites on these linear templates, including at a 5′-AAAAAAAA (A8) sequence that did not reduce TK1761 expression in vivo. When these sequences were transcribed on supercoiled plasmid templates, termination occurred almost exclusively at oligo(T) sequences. The results provide the first in vivo experimental evidence for intrinsic termination of archaeal transcription and confirm that archaeal transcription termination is stimulated by oligo(T) sequences and is different from the RNA hairpin-dependent mechanism established for intrinsic bacterial termination.
Archaea are prokaryotes, with only one RNA polymerase (RNAP), genes cotranscribed in operons, and translation and transcription coupled (5, 10, 12). Archaeal RNAPs, however, are more similar to the three eukaryotic RNAPs (polymerases [Pol] I, II, and III) than to bacterial RNAPs (9, 19, 24, 25, 29, 48). In the absence of genetics, virtually all studies of archaeal transcription have used purified proteins and in vitro assays, and almost all have focused on initiation (13, 15, 29, 33, 36). Consistent with the Pol II-like structure of archaeal RNAPs, these studies have established that archaeal promoters have TATA boxes and transcription factor B recognition elements (BRE) and that initiation requires homologues of the eukaryotic TATA box binding protein (TBP) and transcription factor IIB (designated TFB in Archaea). Initiation is regulated by sequence-specific transcription factors that prevent or stimulate TBP, TFB, and/or RNAP binding to the promoter region (3, 13).
In contrast to initiation, and despite the substantial differences between bacterial and eukaryotic transcription termination (4, 6, 8, 14, 16, 21, 27, 30, 49-51), very little research has addressed transcription termination in Archaea. The few in vitro studies reported (36, 41, 43) are consistent in concluding that termination is stimulated by oligo(T) sequences, but the extents of termination observed differed substantially, and identical oligo(T)-containing sequences positioned at different sites on the same template elicited different termination responses. Unlike intrinsic bacterial termination, termination was not limited to one or two discrete sites but often occurred at multiple adjacent sites, and not always within the oligo(T) sequence. Adding to the difficulty in defining the requirements for termination, the sites and efficiencies of termination in vitro were very sensitive to the assay conditions, changing with changes in the reaction temperature, nucleotide concentrations, salt solutions, or DNA template topology (36, 41, 43).
Given these observations, validating in vivo studies of archaeal transcription termination are needed, but such studies have been precluded by the lack of a facile genetics. Fortunately, this barrier has been removed by the development of genetics for Thermococcus kodakarensis (formerly Thermococcus kodakaraensis) (35, 38, 39), and we have taken advantage of this technology to place sequences predicted to be transcription terminators between a constitutive promoter and a reporter gene (34, 35), at a fixed site, in the T. kodakarensis genome (11). Based on quantitative effects on reporter gene expression, the results obtained confirm that an oligo(T) tract is sufficient to terminate archaeal transcription in vivo, and they establish that intrinsic archaeal termination is very different from intrinsic bacterial termination. The results are consistent with intrinsic archaeal termination resembling Pol III termination (36, 41), and they predict the existence of a mechanism that suppresses termination in vivo at oligo(T) sequences within genes.
T. kodakarensis TS372 was constructed previously from T. kodakarensis KW128 by transformation with plasmid pTS372 DNA (34). In the genome of T. kodakarensis TS372, homologous recombination upstream and downstream of TK1761 replaced the natural promoter and intergenic region upstream of this β-glycosidase-encoding reporter gene with the strong constitutive archaeal promoter PhmtB and the intergenic sequence present upstream of TK1761 in pTS372 (34). Control experiments confirmed that TK1761 expression in T. kodakarensis TS372 is fully dependent on PhmtB and that the encoded β-glycosidase is not essential and can be expressed at different levels without any detectable effects on growth (34, 35).
All of the strains used in this study were constructed by the same plasmid transformation, homologous recombination, and selection procedures used previously to produce T. kodakarensis TS372 (33, 34, 39). The donor plasmid DNAs were identical to pTS372 except for the additional presence of the experimental DNA sequence of interest (listed in Fig. Fig.1)1) at position +10 relative to the site of PhmtB-directed transcription initiation (+1) and 6 bp upstream of the ribosome binding site of TK1761 (see Fig. Fig.2A).2A). Each of these plasmids was constructed by cloning the sequence of interest into pTS372 (34) at the identical target site between PhmtB and TK1761. PCR amplification and sequencing of genomic DNA confirmed that the construct was integrated intact, at the desired location, in the genome of a representative transformant that was then used for further study. The T. kodakarensis strains constructed had only a single genomic copy of TK1761 and retained no additional plasmid-derived sequences. Aliquots of each plasmid DNA, and of the linear DNA molecules generated by PCR amplification from the genomic DNAs of transformants, were also used as templates for in vitro transcription assays. The primers used to generate templates for transcription added a 5′-biotinylated nucleotide to the template.
T. kodakarensis cultures were grown in ASW-YT medium (39), supplemented with 2 g of sulfur per liter, at 85°C to an optical density at 600 nm of ~0.6. The cells were harvested by centrifugation and lysed, and the β-glycosidase activity present in the cleared lysate was measured by ortho-nitrophenyl-β-d-glucopyranoside (ONPG) hydrolysis, as previously described (34). Cleared lysates were obtained from at least three independent cultures of each strain, which had been grown on different days, and each lysate was assayed in triplicate.
The dinucleotide ApC (Sigma, St. Louis, MO) was incubated with T4 polynucleotide kinase and [γ-32P]ATP for 1 h at 37°C to synthesize [32P]ApC. The reaction mixture was then placed at 85°C for 10 min to inactivate the polynucleotide kinase. T. kodakarensis RNAP, TBP, and TFB2 (TK2287) were generated and purified as previously described (40). To allow open complex formation, an aliquot (10 nM) of the template DNA was incubated for 10 min at 85°C with 75 μM [32P]ApC in a reaction mixture that contained 40 nM RNAP, 80 nM TFB2, 80 nM TBP, 20 mM Tris-HCl (pH 8), 250 mM KCl, 5 mM MgCl2, and 5 mM dithiothreitol. Transcription was initiated by adding a mixture of the four nucleoside triphosphates, each at a 1 mM concentration, and was allowed to proceed for 3 min at 85°C. The linear template DNA plus any attached paused or stalled transcripts were removed from an aliquot of each reaction mixture following transcription by binding and sedimentation on streptavidin-coated magnetic beads. Fully terminated transcripts were identified as transcripts that were released from the template and so remained in the supernatant. All of the [32P]-labeled transcripts synthesized in vitro were separated by electrophoresis through denaturing 20% (wt/vol) polyacrylamide gels, detected, and quantified by using a Storm 840 phosphorimager (GE Healthcare; Piscataway, NJ), as previously described (33, 36).
In in vitro assays, an oligo(T) sequence was sufficient to terminate archaeal transcription (36, 41, 43). T. kodakarensis TS552 was therefore constructed with 5′-TTTTTTTT (T8) positioned between the site of PhmtB-directed transcription initiation and TK1761 (Fig. (Fig.1A1A and and2A).2A). The presence of T8 reduced the expression of the downstream TK1761 reporter gene by ~98% from that in T. kodakarensis TS372 (Fig. (Fig.1A).1A). To investigate if an oligo(T) tract was necessary, or if an A/T-rich sequence was sufficient for this reduction in TK1761 expression, T. kodakarensis TS553 and TS554 were constructed with the 5′-AAAAAAAA (A8) sequence and the TA-repetitive sequence 5′-GCTAGTATATATA [designated (TA)4], respectively, positioned upstream of TK1761 (Fig. (Fig.1A).1A). The presence of A8 had no inhibitory effect on TK1761 expression, but the presence of (TA)4 reduced β-glycosidase activity by 64% relative to that in T. kodakarensis TS372 (Fig. (Fig.1A1A).
Because the transition from archaeal transcription initiation to elongation is complete before position +10 (36, 40), the sequences assayed for termination activity were positioned beginning at +10. Nevertheless, to confirm that the inhibitory effect of T8 on TK1761 expression was not dependent on the +10-to-+17 location, T. kodakarensis TS557 was constructed with an additional 20 bp preceding T8. The expression of TK1761 was reduced by ~97% in T. kodakarensis TS557, showing that T8 from +30 to +37 was just as effective in terminating transcription as T8 from +10 to +17 (Fig. (Fig.1A).1A). As a control, T. kodakarensis TS437 was constructed with only the 20-bp sequence positioned upstream of TK1761, not followed by T8. The expression of TK1761 in T. kodakarensis TS437 was essentially the same as that in T. kodakarensis TS372 (data not shown).
The GeSTer algorithm (45) predicts and ranks the likely efficiency of intrinsic bacterial terminators by their location and the presence of an inverted repeat followed by an oligo(T) (T-trail) sequence. This algorithm, with the inverted-repeat sequence requirement eliminated, was used to identify candidate transcription terminators within the T. kodakarensis genome. Twelve intergenic sequences so identified, which had one to three oligo(T) tracts ranging from T4 to T7, were PCR amplified and positioned upstream of TK1761 (Fig. (Fig.1B).1B). In every strain, the presence of the T. kodakarensis intergenic sequence reduced TK1761 expression by >70%, and in 8 of the 12 strains, the expression of TK1761 was reduced by >90% (Fig. (Fig.1B).1B). Mutations were introduced into several of these intergenic sequences to reduce any potential that the sequence upstream of the T tracts had to form a RNA hairpin. The mutated sequences were just as effective as the wild-type intergenic sequences in reducing TK1761 expression in vivo (data not shown).
Intrinsic bacterial terminators classically have an inverted-repeat sequence followed by an oligo(T) (T-trail) sequence (16, 27, 30, 37, 50), and mutations that reduce the ability of the inverted repeat to form a stable hairpin-loop dramatically reduce termination efficiency (32, 42, 46). Given this well-documented precedent, a sequence that conformed exactly to the bacterial terminator motif located immediately downstream of the very highly expressed mcrBDCGA operon in Methanothermobacter thermautotrophicus was predicted to be a transcription terminator. Northern blotting confirmed that termination in vivo occurred at or near this sequence, designated tmcrA (31). T. kodakarensis TS442 was constructed with tmcrA positioned upstream of TK1761, and this reduced TK1761 expression by ~88%, consistent with tmcrA functioning as an efficient transcription terminator in T. kodakarensis (Fig. (Fig.1C).1C). To determine if the bacterial terminator motif was required for this activity, T. kodakarensis TS443, TS444, TS445, TS555, and TS556 were constructed with derivatives of tmcrA positioned upstream of TK1761 (Fig. (Fig.1C).1C). The activity of tmcrA was not reduced by the presence of a point mutation in the inverted repeat (T. kodakarensis TS443) and was increased, rather than decreased, by the deletion either of parts of the inverted-repeat and loop region (T. kodakarensis TS444 and TS555) or of the entire region (T. kodakarensis TS556). In contrast, when the inverted-repeat and loop region was retained and the T-trail deleted (T. kodakarensis TS445), TK1761 expression increased relative to that in the strains containing an tmcrA-derived sequence that retained the T-trail (Fig. (Fig.1C).1C). The β-glycosidase activity in T. kodakarensis TS445 was nevertheless reduced to ~ 40% of that of T. kodakarensis TS372, consistent with the T6 sequence in the loop region of tmcrA also having the capacity to decrease downstream gene expression. As controls, T. kodakarensis TS452 and TS454 were constructed, with M. thermautotrophicus coding sequences amplified from within mcrA, and similar in length to tmcrA, positioned between PhmtB and TK1761. The presence of these coding sequences reduced TK1761 expression by only ~25% and ~50%, respectively (Fig. (Fig.1C1C).
To establish that the sequences inserted upstream of TK1761 did direct transcription termination in vitro, templates with a biotinylated 5′ nucleotide were PCR amplified from genomic DNA isolated from the T. kodakarensis strains. These templates were incubated in in vitro transcription reaction mixtures under conditions in which every transcript synthesized was labeled equally, regardless of length, by the incorporation of [32P]ApC as the 5′ dinucleotide. Because all of the templates had the same promoter and sequence to +10, as expected, initiation occurred with equal efficiency, and the patterns of abortive transcripts produced were very similar on all templates (Fig. (Fig.2B2B and and3A).3A). Template removal by binding to streptavidin-coated beads removed ~25% of each runoff transcript, but all the other transcripts synthesized were fully terminated and released from the templates.
The amount of [32P]ApC incorporated into all transcripts 10 nucleotides long or longer was measured, and this provided a value for the total number of transcripts initiated and extended beyond +10. The percentage of this amount that was incorporated into transcripts shorter than full-length transcripts provided a measure of termination on a particular template. Based on this calculation, ~35% of the transcripts initiated and extended beyond abortive transcripts on the control 372 template, amplified from T. kodakarensis TS372, were terminated before the RNAP reached the end of the template. Termination occurred most noticeably at positions from +12 to +20 but also at many other positions along the length of the 372 template (Fig. (Fig.2B).2B). In contrast, on template 552, amplified from T. kodakarensis TS552, >75% of the transcripts were terminated, and termination occurred predominantly at positions 4, 5, and 6 within the T8 sequence (Fig. (Fig.2B).2B). Termination occurred to essentially the same extent (~82%) on template 557, amplified from T. kodakarensis TS557, again predominantly within the T8 sequence (Fig. (Fig.2B).2B). As anticipated, based on the ~64% reduction of TK1761 expression in vivo by (TA)4 in T. kodakarensis TS554 (Fig. (Fig.1A),1A), termination was ~20% higher on template 554 than on template 372 and occurred within the TA-repetitive sequence. In contrast, and in conflict with the lack of an inhibitory effect of A8 on TK1761 expression in T. kodakarensis TS553, ~75% of the transcripts initiated on template 553 were terminated, and substantial termination occurred at almost every position within the A8 sequence (Fig. (Fig.2B2B).
To further investigate the A8 anomaly, in vitro transcription was repeated, but an aliquot of the supercoiled, circular plasmid DNA (pTS553) that had been used to transform T. kodakarensis KW128 to produce T. kodakarensis TS553 was used as the template. On this circular template, in addition to abortive and terminated transcripts, very long read-through transcripts were generated by continued transcription through terminators and around the circular DNA. Termination was calculated as the amount of [32P]ApC incorporated into terminated transcripts 10 nucleotides long or longer and was expressed as a percentage of the incorporation of [32P]ApC into all read-through and terminated transcripts. In contrast to the ~75% termination that occurred on the linear 553 template, only ~27% of the transcripts initiated on pTS553 DNA were terminated (Fig. 2B and C). Given this difference, transcription termination was also measured on pTS372, pTS552, pTS554, and pTS557, the plasmids used to generate T. kodakarensis TS372, TS552, TS554, and TS557 (Fig. (Fig.2B).2B). There was a small (~10%) reduction in termination on pTS372 from that on the linear 372 template (Fig. (Fig.2C).2C). The percentages of transcripts that terminated on pTS372 and pTS553 were essentially identical, consistent with the almost identical levels of β-glycosidase activity present in T. kodakarensis TS372 and TS553 (Fig. (Fig.1A).1A). In contrast to the topology-dependent termination at A8, topology had little effect on the extent of termination directed by T8 and (TA)4, but the predominant termination occurred at fewer adjacent sites in T8 and (TA)4 on the plasmid pTS552, pTS557, and pTS554 templates.
We also investigated and compared transcription in vitro from both linear and plasmid templates that had the tmcrA and tmcrA-derived sequences (Fig. (Fig.3B).3B). The extents of termination were essentially the same on the linear and circular templates, except for a small (~13%) increase in termination on pTS445 over that on the linear 445 template (Fig. (Fig.3C).3C). The overall patterns of termination were, however, different. Consistent with the in vivo reporter assays, termination occurred on all templates at the T-trail sequence, with or without the presence of the inverted-repeat and loop region, but whereas this accounted for almost all of the termination on the plasmid templates (Fig. (Fig.3B),3B), detectable termination also occurred at many other sites along the linear templates (Fig. (Fig.3A),3A), including at the T6 sequence in the loop region of tmcrA.
All of the synthetic and natural oligo(T)-containing sequences positioned between PhmtB and TK1761 reduced the expression of the reporter gene in vivo, and the extents of reduction correlated well with the effectiveness of the sequence at terminating transcription in vitro by T. kodakarensis RNAP. It seems reasonable, therefore, to conclude that these sequences reduced TK1761 expression in vivo by directing intrinsic transcription termination upstream of TK1761. In contrast to intrinsic bacterial termination (9, 16, 27, 30, 37, 44, 50), this did not require the presence of an inverted repeat in the template sequence but was stimulated by the presence of an oligo(T) sequence, as is also the case for Pol III intrinsic termination (4, 26, 41). If the common ancestor of the archaeal RNAPs, Pol I, II, and III, was sensitive to oligo(T)-directed termination, the retention of this by Pol III may have been facilitated by the dedication of Pol III to the transcription of short RNA-encoding genes with few oligo(T) sequences. This argument cannot be extrapolated to all extant Archaea, since many have very A+T rich genomes, and although the T. kodakarensis genome is 52% G+C, it contains 7, 127, and 1,081 copies of T8, T7, and T6, respectively, of which 3, 80, and 713 are within rather than between genes. Given the demonstration of operon polarity (34), tight coupling of translation and transcription presumably suppresses intragenic transcription termination, but it seems likely that additional sequence information must inform a transcribing RNAP, either directly or via a transcription termination factor(s), to terminate or read through some oligo(T) sequences. T. kodakarensis grows optimally at 85°C, and such a hyperthermophilic lifestyle may preclude the use of nascent transcript folding to regulate gene expression during transcription. Given that this is a hyperthermophily limitation, it remains possible that spontaneous RNA duplex formation during transcription does direct termination in other Archaea, but inverted-repeat sequences occur only infrequently downstream of archaeal genes, arguing that this is not a common event (45, 47).
To date, we have mapped the sites at which archaeal transcription is terminated in vitro on >50 archaeal and bacterial intergenic sequences and synthetic templates, with a precision of 1 to 2 bp. With this database, we have sought to identify common sequence features and the driving force that destabilizes an otherwise very stable archaeal elongation complex (33, 36). Archaeal termination in vitro is stimulated by T-rich sequences of differing lengths and contexts, and it occurs most often within T tracts. However, as illustrated by the (TA)4 results (Fig. (Fig.2B),2B), archaeal termination is not strictly dependent on an extended T tract, and in this regard, it may differ from Pol III termination. The predominance of termination at oligo(T) tracts is increased in vitro on circular versus linear templates (Fig. (Fig.22 and and3).3). In contrast to intrinsic bacterial termination, transcript release is not limited to one or two positions with defined spacing relative to a recognizable terminator motif but occurs at several adjacent sites, and not preferentially at sites where an RNA-DNA hybrid would have inherently weak base-pairing (7, 20, 23). With no requirement for an inverted repeat, there is no support for a mechanism dependent on a change in RNAP structure driven by RNA hairpin formation, and the data also provide no evidence for transcript slippage facilitating transcript release (1, 2, 9, 17, 28, 37, 44, 50, 51). Given this lack of support for any established mechanism, archaeal termination may be directed by a novel mechanism, but establishing this mechanism will require a combination of in vivo and in vitro assays. As the results with A8 exemplify, a sequence that terminates the transcription of a purified archaeal RNAP on a linear template in vitro may not terminate transcription in vivo. Similarly, in vivo validation is needed for the in vitro observation that archaeal transcription through an upstream sequence can influence an elongation-termination decision at a remote downstream sequence (36). The occurrence of termination at multiple adjacent sites in vitro does, however, already have in vivo support. Based on high-resolution hybridization data, archaeal transcripts synthesized in vivo have a defined 5′ terminus but multiple 3′ termini (22).
In Bacteria, many attenuators and riboswitches that regulate gene expression by controlling intrinsic termination have been identified (16, 18, 27, 30, 50). Intriguingly, to date, there is no experimental evidence or bioinformatics prediction for such regulation in Archaea, although long untranslated leader regions have been identified. Most analyses of archaeal genomes have likely searched primarily for sequences that conform to the bacterial intrinsic terminator motif, whereas the results reported here demonstrate that relatively simple oligo(T)-rich sequences could be sites at which regulation is imposed. The extent of intrinsic termination directed by the T-trail sequence of tmcrA, for example, could be reduced, as was observed in vivo under very high expression conditions (31), by the presence or absence of a factor interacting with the upstream inverted repeat.
This research was supported by grants from the DOE (DE-FG02-87ER13731) and NIH (GM53185), awarded to J.N.R., and by NIH fellowship (1F32-GM073336-01) support to T.J.S.
Published ahead of print on 11 September 2009.