Ribozyme motif structure mapped using random recombination and selection

Abstract

Isolating the core functional elements of an RNA is normally performed during the characterization of a new RNA in order to simplify further biochemical analysis. The removal of extraneous sequence is challenging and can lead to biases that result from the incomplete sampling of deletion variants. An impartial solution to this problem is to construct a library containing a large number of deletion constructs and to select functional RNA isolates that are at least as efficient as their full-length progenitors. Here, we use nonhomologous recombination and selection to isolate the catalytic core of a pyrimidine nucleotide synthase ribozyme. A variable-length pool of approximately 10(8) recombinant molecules that included deletions, inversions, and translocations of a 271-nucleotide-long ribozyme isolate was constructed by digesting and randomly religating its DNA genome. In vitro selection for functional ribozymes was then performed in a size-dependent and a size-independent manner. The final pools had nearly equivalent catalytic rates even though their length distributions were completely different, indicating that a diverse range of deletion constructs were functionally active. Four short sequence islands, requiring as little as 81 nt of sequence, were found within all of the truncated ribozymes and could be folded into a secondary structure consisting of three helix-loops. Our findings suggest that nonhomologous recombination is a highly efficient way to isolate a ribozyme's core motif and could prove to be a useful method for evolving new ribozyme functions from pre-existing sequences in a manner that may have played an important role early in evolution.

FIGURE 1.

The random recombination scheme used to synthesize the starting pool. (A) Colored lines indicate arbitrary positions within the defined input B.01 DNA sequence (four copies shown). Black lines terminated with a dot and bar denote the 5′- and 3′-PCR primer-binding regions, respectively. The primers used to generate the pool are not phosphorylated. (B) dsDNAs are cut into randomly distributed fragments using DNase I. The resulting overlapping ends are removed or filled into blunt ends using T4 DNA polymerase. (C) T4 DNA ligase randomly reassembles the blunt-end fragments into new molecules with a broad length distribution. These fragments may now have a different location or orientation within the ligated construct. Nicked dsDNA with inner fragments containing 5′ hydroxyl primer sequence and dsDNA with inner fragments containing the majority of the primer region result in shorter DNA after PCR amplification. (D) After agarose gel purification of a suitable size range and PCR amplification, the final recombined DNA pool consist of molecules with 5′- and 3′-primer sequence at each end. This pool is then transcribed into the initial RNA pool and is ready for in vitro selection. (E) The recombined DNA pool construction steps summarized on a 3% agarose gel. (Lane 1) Input 291-bp B.01 DNA; (lane 2) DNA fragments after DNase I digestion; (lane 3) DNA pool after T4 DNA ligase step; (lane 4) the final recombined DNA pool after length selection and PCR. (F) Distribution of DNase I cut sites within the B.01 DNA sequence. A total of 126 cut sites (of 237 possible) were inferred from 40 recombined sequence isolates (see text). Long and short black bars indicate the 37-bp 5′-PCR primer and 17-bp 3′-RT-PCR primer locations, respectively.

FIGURE 2.

In vitro selection strategy. (A) The input RNA pool is transcribed and a suitable size range is selected using a denaturing polyacrylamide gel. The purified pool is ligated to pRpp using adenylylated pRpp (AppRpp) and T4 RNA ligase. (B) The resulting pool now tethered to pRpp at its 3′-terminus is incubated with 4-thiouracil. The reacted pool consists of ribozymes able to catalyze the synthesis of tethered ^4SU. (C) Active ribozymes in a size range indicated by synthetic markers terminated with ^4SU are purified using one or two APM gels, which retard the mobility of sulfur-containing RNA. (D) The eluted RNA pool is reverse-transcribed into DNA, ready for the next round of selection. Two selection pressures can be applied. The first is by varying the RNA size recovered from polyacrylamide and APM gels, and the second is by changing the 4-thiouracil incubation time.

FIGURE 3.

RNA pool distribution resulting from the size-independent selection. Equivalent amounts of input RNA from each round of selection were loaded on a 6% polyacrylamide gel, with RNA size markers (70, 137, and 274 nt) loaded on the right. Round 0 represents the initial recombined RNA pool. In each round, a segment of RNA from 70 nt to ~250 nt was excised from gels as indicated by the bars on the left. The 4-thiouracil incubation times from Round 1 to 4 were kept at 24 h and then decreased to 0.5 h for Round 5, and 0.05 h by Round 6. By Round 4 the RNA pool distribution showed two dominant bands, labeled upper (U) and lower (L), which were spanned by a middle region (M).

FIGURE 4.

RNA pool distribution resulting from the size-dependent selection. Starting in Round 4, a gel slice extending from 70 nt to the shortest RNA dominating the previous round was used to select the RNA for the next round of selection. The 4-thiouracil incubation times used for each round of selection are indicated.

FIGURE 5.

Sequence alignment of active ribozyme DNA sequence isolated from both selections and sorted by length. Ribozymes from the size-independent selection are indicated by I-X.Y, where X is the round of selection and Y is the isolate number. Sequences marked with a D correspond to isolates from the size-dependent selection. U, M, or L correspond to the range of lengths expected from the upper, middle, and lower bands observed in Figure 3, respectively. Pairs of vertically colored bars, within the four island regions, represent helical structural elements as predicted by the Pknots algorithm. The ribozyme length (Leng) and number of recombined fragments (Frag) for each isolate are shown in the lower right.

FIGURE 6.

Proposed core motif secondary structure of the family B nucleotide synthase ribozyme family as predicted using the Pknots algorithm. The three regions of deletion are found within the three terminal loops of the structure, and their size ranges are indicated. The sequence corresponds to the shortest isolate (D-12.1) in the alignment shown in Figure 5 and is numbered from its 5′-end.