Supplemental Data


Abstract
Full Text
RNA-Catalyzed RNA Polymerization: Accurate and General RNA-Templated Primer Extension
Wendy K. Johnston, Peter J. Unrau, Michael S. Lawrence, Margaret E. Glasner, and David P. Bartel

Supplementary Material

Secondary structural elements and other sequence features conserved among the polymerase isolates

Sequences of the most active isolates from rounds 14 and 18 were compared with the round-10 parental sequence (Web fig. 1). Analysis of conserved and covarying residues led to the secondary structure model of the parental ribozyme presented in Fig. 1C. Many residues were highly conserved within the accessory domain (spanning C110 to A204 of the parental sequence), which had been mutagenized at a level of 20% per residue. Within this domain, 29 residues were conserved in all 22 isolates; another 14 residues were conserved in 21 of the 22 isolates. All but a few of these residues must be important for ribozyme function, because the chance conservation of a residue not important for activity is low (P = 0.0074 for conservation in 22 of 22 isolates, P = 0.048 for conservation in at least 21 of 22 isolates). Four residues consistently differed from the parent sequence and are likely to confer increased polymerization activity.

Interspersed among and flanking the highly conserved residues were five pairs of covarying residues (G151:C200, A153:U198, C154:G197, U175:A183, C176:G182; Web fig. 1), which support the existence of two critical paired regions within the accessory domain (segments shaded in yellow and pink in Web fig. 1). The region spanning positions 115 to 149 had little sequence conservation yet still contained a large number of complementary segments (many of which are underlined in gray in Web fig. 1). Within this region, the propensity for pairing appears to have been conserved, but the precise register of the potential pairing was not highly conserved. The activity of constructs with deletions within this region (e.g., constructs 5 and 21, described below) matches that of the parent, suggesting that pairing in this region directs nonfunctional segments away from the ribozyme, while bringing the important residues of the accessory domain near to the ligase core domain.

Engineering shorter and more active polymerase constructs

Site-directed mutagenesis was based on 18.12 (Web fig. 2), a round-18 isolate particularly proficient with long template RNAs. Sequences and activities of 20 derivatives are presented here (Web fig. 3, Web table 1). In the first set of constructs (18.12.1 to 18.12.8), three deletion derivatives were as active as the parental isolate, showing that the bulk of the 5´ PCR primer-binding site was dispensable (18.12.1) and that terminal loops U45 to U52 and U122 to G142 could each be replaced with the UUCG tetraloop (18.12.2 and 18.12.5, respectively). Deleting A199 improved activity (8.12.7). This residue was predicted to bulge from the G149 to U154:A197 to C203 pairing (yellow shading in Web fig. 1) and was not present in any of the other improved isolates; its removal may stabilize important flanking pairs. Other deletions were detrimental. Deleting the RT-PCR primer-binding site substantially lowered activity (8.12.8), whereas replacing C81 to G86 with the analogous terminal loop of the ligase ribozyme slightly lowered activity (8.12.3).

The 18.12 isolate has changes at positions 106 and 156, two residues that were otherwise conserved in the other 21 improved polymerase isolates. Reverting A156 to the consensus sequence slightly decreased activity (8.12.6). The change at 106 was more intriguing. It lies within the segment G104 to C109, which was designed to pair with the 7-nt RNA that completes the ligase core domain (Web fig. 2). Furthermore, segment G104 to C109 was not mutated in the degenerate pool, because it also corresponded to a Ban I restriction site needed for construction of the DNA template. Nevertheless, all five of the improved isolates from round 18 have a change within this segment (Web fig. 1), and reverting C106 to the consensus dramatically decreased activity (8.12.4). Rare mutations within G104 to C109 that occurred after pool construction apparently conferred a substantial selective advantage in subsequent rounds of selection.

Combining changes from constructs that had negligible (8.12.1, 8.12.2, and 8.12.5) or beneficial effects (8.12.7) produced a composite derivative (8.12.9) that was improved over the parent and served as the starting point for subsequent derivatives. The primary goal was to eliminate or replace the helix formed between the RT-PCR primer-binding site and the reverse-transcription primer (Web fig. 2), so that a DNA oligonucleotide would not be needed for maximal ribozyme activity. With respect to this goal, the favored derivative was 8.12.23, which has this stem replaced with a 3´-terminal stem-loop, yet retains activity superior to that of isolate 18.12. Construct 8.12.23 was renamed the "round-18 ribozyme" (Fig. 1D) and served as the prototype polymerase ribozyme for the remainder of our study.

Other notable constructs include 8.12.16 and 8.12.21, two derivatives that are shorter than 8.12.23 yet retain polymerization activity. Constructs 8.12.17 and 8.12.18 demonstrate that the 7-nt RNA that completes the ligase domain can be appended to the end of the accessory domain to generate a unimolecular polymerase construct, although the possibility remains that these ribozymes function as homodimers.

Nuclease analysis of the primer-extension product

Unlabeled primer (CUGCCAA) was extended using α-32P-CTP in a polymerase reaction with 5 μM ribozyme, 2 μM primer, and 2.5 μM template encoding 11 residues (Fig. 4). Primer extended by one nucleotide was gel purified, then digested with ribonuclease T2 (which does not cleave at 2´,5´-linkages) and analyzed by two-dimensional TLC, as described previously (1). Synthetic 2´-linked dinucleotide (adenylyl-2´,5´-cytidine; Sigma) and tRNA were included in the digest to serve as carrier and to provide nonradiolabeled chromatography standards, observable with ultraviolet shadowing. Radiolabeled digestion product comigrating with the dinucleotide standard was not detected (detection limit, 0.5%). Instead, the digestion product comigrated predominantly with the adenosine 3´-phosphate standard, indicating that the ribozyme had extended the primer by a normal RNA linkage.

Generating, purifying and cloning full-length primer-extension product (methods for Fig. 4C)

Primer RNA CAGCCAA was thiophosphorylated using T4 polynucleotide kinase with a trace amount of γ-35S-labeled ATP, followed by a chase with excess cold adenosine-5´-(γ-thio)triphosphate (ATP-γ-S). Primer with a 5´-thiophosphate was purified on an APM gel (excising RNA immobilized at the APM interface because of the 5´-thiophosphate), then used in a polymerization reaction like that of Fig. 4, with a template, dAdAGACGGUUGGCACGCUUCG (bold type indicates pairing to primer; dA is 2´-deoxyA), analogous to the one coding for 11 nt in Fig. 4B. The reaction was stopped after 24.5 hours, heated in the presence of competitor RNA, CUGCCAACCGUGCGAAGCGUCUCC (bold type indicates segment hybridizing to the template), and electrophoresed on a sequencing gel. The portion of the gel extending from just above the 17-nt band, through the 18-nt (full-length) and 19-nt bands was excised. To avoid contamination by competitor and template RNAs, the extended primer eluted from this gel slice was purified further on an APM gel (excising RNA at the APM interface). An RT-PCR primer-binding site was appended to the 3´ terminus of the purified RNA using an adenylylated DNA oligonucleotide, A-5´-pp-5´-GAAGAGCCTACGACGA, and T4 RNA ligase in the absence of ATP. The ligation yield was greater than 75%, suggesting that sequence preferences of T4 RNA ligase did not substantially influence the population of cloned sequences. The ligated product was purified on a sequencing gel, reverse-transcribed (SuperScriptII, Gibco BRL), PCR-amplified using the primers TTGGAATTCAGCTGCCAA, and TCGTCGTAGGCTCTTC, and cloned (Topo cloning kit, Invitrogen).

The precautions taken to avoid cloning the competitor or template RNAs were successful; none of the sequences matched the products expected from cloning the competitor or template RNAs. Note that rare degradation products of the competitor RNA that match the sequence of the extended product would have a phosphate at their 3´(2´) terminus and thus would not be suitable substrates for T4 RNA ligase. Data from 100 of the 101 sequenced clones are tabulated in Fig. 4C. The other clone was that of a product missing the last residue (although each of the 10 nucleotides that had been added by the ribozyme was a Watson-Crick match to the template).

Sequences of the primer and template RNAs used in Fig. 5

Primer RNAs were GGAGCAAAAC or GGCGUAGACAACCCUGUGUUUAGCCUGCGUUUUGUGCCAUCCUAAUGCUUGGAGCAAAAC.
Template RNAs were CCUACGCCUCGUUUUGCUCC, GGAGCGAGGACGCUCUACAAACUGGCUAAACAACCAUCCCGUUGUGCGUUCAACUUCAACUUCGGAACAGGGAGCAACCGCCUUUUAAACCCUACGCCUCGUUUUGCUCC, or GGAGCGAGGACGCUCUACAAACUGGCUAAACAACCAUCCCGUUGUGCGUUCAACUUCAACUUCGGAACAGGGAGCAACCGCCUUUUAAACCCUACGCCUCGUUUUGCUCCAAGCAUUAGGAUGGCACAAAACGCAGGCUAAACACAGGGUUGUCUACGCC.

References

1. P. J. Unrau, D. P. Bartel, Nature395, 260 (1998).
2. C. Kao, M. Zheng, S. Rudisser, RNA5, 1268 (1999).


Supplemental Figure 1. Comparative sequence analysis of improved polymerase variants isolated from selection rounds 14 and 18. DNA sequences of the 22 most active variants are aligned with respect to the sequence of the round-10 isolate (parent). Residues of the parent sequence are colored as in Fig. 1C; in blue are positions that were mutagenized at 20% when synthesizing the degenerate pool; in black are positions mutagenized at 3% or less; in green are positions of the primer-binding site. Segments with conserved potential for base pairing are shaded (large colored regions), with the pairing partners shaded the same color. The pairing partner for the purple shaded segment is the 7-nt RNA, GGCACCA, added to complete the ligase core. A region of more diffuse complementarity spanned positions 115 to 149 (complementary segments underlined in grey). Residues within the shaded segments that covaried such that the predicted pairing was maintained are marked with dark blue vertical bars. Other residues that differed from the parent sequence are boxed (dark blue boxes). Red asterisks mark the four positions that consistently differed from the parent sequence.


Medium version | Full size version


Supplemental Figure 2. Secondary structure model of 18.12. Dashes indicate conserved potential for pairing; thick dashes indicate pairing supported by covariation as well as conservation (Web fig. 1). Residues of the ligase core are in black and purple. Regions mutagenized at 20% are predominantly blue, with scattered pink residues. Residues mutagenized at 20% yet conserved in all 22 improved isolates (Web fig. 1) are shown in blue uppercase. Residues that differed from the round-10 ribozyme are in pink. The four residues that frequently differed from the parent sequence are shown in pink uppercase. Primer-binding sites are in green. The reverse-transcription primer is also shown (ttcagattgtagccttc, in green) paired to the RT-PCR primer-binding site (positions 204-220), because this DNA oligonucleotide was included during ribozyme incubations to prevent residues of the primer-binding site from interfering with ribozyme activity. Isolate 18.12 and the ribozyme pools from rounds 14 and 18, were all more active in the presence of this primer.


Medium version | Full size version


Supplemental Figure 3. Engineered derivatives of the 18.12 isolate (parent). Residues of the parent are colored as in Web fig. 2; segments of the constructs that differ from the parent are colored red. Paired segments are shaded as in Web fig. 1. The 7-nt RNA that completes the ligase core domain (7-nt oligo) was added as a separate RNA oligonucleotide, except with constructs 17 and 18, wherein the 7-nt oligo was appended to the 3´ terminus of the ribozyme using a short linker segment. For constructs 1 to 7, 9, 10, and 15, the RT primer was also included. The PCR DNA used as a transcription template for constructs 17, 18, 21, and 23 had penultimate 2´-methoxynucleotides incorporated with the intent of minimizing T7 RNA polymerase transcripts with extra, nontemplated nucleotides at their 3´ termini (2).


Medium version | Full size version


Supplemental Table 1.Activity of the 18.12 derivatives shown in Web fig. 3.
ConstructActivity* ConstructActivity*
Round-10 isolate+8.12.9+++++
8.12 isolate++++8.12.10+++
8.12.1++++8.12.11++++
8.12.2++++8.12.12++
8.12.3+++8.12.13+++
8.12.4++8.12.14++
8.12.5++++8.12.15+
8.12.6+++8.12.16++
8.12.7+++++8.12.17++
8.12.8+8.12.18++
8.12.21++++
8.12.23+++++
*Activity measured using the primer-template complex CUGCCAA:GCUUCGCACGGUUGGCAG in assays like those shown in Fig. 4. Scale: +, activity similar to that of the round-10 isolate; ++, activity greater than that of the round-10 isolate yet substantially less than that of 18.12; +++, activity nearly that of 18.12; ++++, activity indistinguishable from that of 18.12; +++++, activity discernibly greater than that of 18.12.