Pool design and construction. (A) Secondary structure of the Round 18 polymerase, which was used to construct the starting pool. The ligase core is shown in red, while the accessory domain is depicted in blue, and the P2 oligonucleotide, found to be important for the WT polymerase ribozyme, is drawn in green. Black residues indicate fixed sequence that was required for transcriptional initiation, PCR amplification, or ligation of the two domains after EarI digestion. A random insert of variable length was also introduced, (NN)_0–5, 3′ to the pair of initiating guanosine residues (magenta). (B) The starting pool was made from three ligation reactions using the molar amounts shown. (C) Pool processing for the selection (T7P indicates T7 promoter). The DNA pool was digested with BtgI, leaving a 4-nt overhang that was used to attach the single-stranded selection primer using T4 DNA ligase (P-pool). This DNA–RNA hybrid (RNA in red) was then annealed to an RNA template in preparation for the selection process (PT-pool).

Large-scale emulsion manufacture. (A) Photographs of the steps used to make 50 mL of emulsion. First (left to right), DEPC-treated glass beads are added into a 50 mL Falcon tube; second, the aqueous phase (bromophenol blue is included for visualization) is added; third, the oil-phase is poured on top of the aqueous phase before the whole mixture is vortexed to make the water-in-oil droplets. (B) Microscope images of emulsion containing RNA (180-nt-long transcripts at 5 μM) that was stained with SYBR green. Fluorescence resulting from 450 to 490 nm excitation colocalized with the DIC image. The droplets are 1–3 μm in diameter. (C) A denaturing polyacrylamide gel showing that emulsified transcription takes place with a yield one-third that of the normal one.

The WT polymerase is active in the selection context. (A) The PT-WT or the PT-R0 genome constructs were incubated with gel-purified WT (2 μM final concentration) ribozyme polymerase or T7 RNAP, and incubated in the selection buffer. Time points were taken at 0, 1, 2, 4, and 20 h, and the DNA genomes were subsequently digested with HinfI to improve gel resolution. Both constructs were extended when the WT ribozyme was added, but only the PT-WT was extended in the presence of T7 RNAP, indicating that transcripts produced by T7 RNAP were catalytically active. (B) The WT ribozyme was transcribed in the presence of varying ^4SUTP concentrations (relative to 2 mM UTP which was always present) and gel purified. The four resulting ribozymes were then assayed for their ability to extend PT(D4P12:T21) in the selection buffer supplemented with 0.5 mM ^4SUTP. Time points were taken at 0.5 h, 4.5 h, and 24 h and resolved on denaturing PAGE plus or minus APM. Ribozymes transcribed in the presence of 1 mM ^4SUTP and 2 mM UTP were almost as active as ribozymes that were transcribed in the absence of ^4SUTP. All ribozymes were capable of incorporating ^4SUMP as indicated by the mercury-dependent gel shift shown in the right panel (marked as primer ^4SU in the figure).

The hybridization approach is efficient in isolating extended DNA constructs. A ³³P-end labeled D11P12 primer was extended by dNTPs in the presence of T21 using superscript reverse transcriptase (Invitrogen), gel purified, and ligated to the DNA pool. This material was then mixed with an equal amount of DNA that had been ligated to a ³²P-end-labeled unextended D11P12 primer. A mock round of selection recovered 21.4 ± 0.7% of the ³³P-labeled material (mock extended), while only 0.21 ± 0.03% of the ³²P-labeled construct (unextended) was retained, as determined by spectrum gated scintillation counting.

Selection scheme used in the branch B selection. (A) DNA constructs, which were ligated to an RNA primer, are annealed to the selection template. (B) This PT-pool is then compartmentalized within water-in-oil droplets together with NTPs, ^4SUTP, and T7 RNAP. Incubation results in T7 RNAP-mediated transcription of the RNA phenotype. (C) Active polymerase ribozymes recognize the PT complex that was attached to their parent DNA genotype and extend it in a template-directed manner while also incorporating ^4SUMP. (D) The emulsion is broken open and the recovered DNA genomes separated on an APM gel to enrich in those modified with ^4SUMP. (E) The recovered material is then hybridized to a biotinylated probe having a sequence complementary to the extended primer, and applied to streptavidin magnetic beads. (F) The purified pool is eluted from the beads, amplified, digested with BtgI, and ligated to the selection DNA/RNA primer ready for the next round of selection.

Progress and outcome of the selection. (A) Polymerization activity of gel-purified selection rounds under the conditions indicated. Aliquots were taken at the indicated time intervals and extension products separated on 20% PAGE. Length of extension products are shown on the right. After six rounds of the branch B selection, extension activity was enriched by a factor of 60. (B) Extension of PT(D4P12:T21) complex by variants from branches A and B of the selection. B6.61 extended >6 nt of the PT complex threefold faster than the WT, while A7.15 and A7.29 were slightly slower. (C) Sequence alignment of the selected isolates relative to the WT sequence. Elements highlighted in the same color are proposed to form a secondary structure with each other based on the previous secondary structure determination of the class I ligase and WT polymerase. Residues that are different from the WT sequence are boxed. The initiating guanosine residues are underlined, while the underlined CAA stretch indicates the fixed sequence that was required to ligate the two modules together to make the starting pool.

B6.61 polymerizes a PT complex by at least 20 nt. The PT complexes were incubated in the presence of WT or B6.61 ribozymes using optimum conditions. Reactions were stopped at the specified time intervals and resolved using 20% PAGE. Although both polymerases extended PT(P7:T21) by 14 nt (the end of the template), B6.61 was threefold faster in synthesizing > 9 nt extension products (left panel). B6.61 was able to extend PT(P7:T31) by at least 20 nt in contrast to the 11 nt of extension detectable by the WT ribozyme (right panel).

The fidelity of B6.61 is superior. (A) The extension pattern of PT(D4P12:T31) by both polymerases was compared with the pattern observed for PT(D4P12:T21). Reactions were conducted using optimum conditions, and aliquots were taken at the indicated time points. Expected extension products are numbered, while incorrect products are marked by asterisks. Note that B6.61 produces approximately threefold more of the expected products relative to the WT. Products beyond the 8-nt addition were not labeled as there was no simple standard to determine whether they were correct additions. (B) Effect of P2 on ribozyme fidelity. Ribozyme-catalyzed extension of PT(P7:T26) was conducted in the absence or presence of the P2 oligonucleotide using optimum conditions. In the absence of the P2 element, the WT polymerase adds an incorrect product after the first addition (marked by an asterisk), while B6.61 produces three- to fourfold less of the same anomalous product. In the presence of P2, the accumulation of this product by the WT ribozyme was inhibited, but the addition of an incorrect nucleotide after the second addition was enhanced by both polymerases.

The WT polymerase utilizes wobble base pairing during extension more often than B6.61. Polymerization of PT(P7:T26) by both ribozymes was conducted in the absence of NTPs (B); CTP (C); CTP and GTP (C+G); CTP, GTP, and UTP (C+G+U); and all four NTPs (N). Only in the presence of UTP is the misincorporation observed, where the WT ribozyme incorporates UMP across a guanosine residue approximately threefold more than the B6.61 ribozyme.