In vitro selection of self-cleaving ribozymes. (A) Selection of self-cleaving ribozymes from random-sequence RNA. (I) RNAs are incubated under permissive reaction conditions (see text). (II) The 5′ fragments of cleaved RNAs are separated from uncleaved precursors by PAGE and (III) are amplified by RT-PCR, which introduces a T7 promoter sequence and restores the nucleotides that were lost upon ribozyme cleavage. The resulting double-stranded DNAs are transcribed in vitro using T7 RNAP to generate the subsequent population of RNAs. (B) RNA construct used to initiate the in vitro selection process. N₄₀ depicts 40 random-sequence nts. Underlined nucleotides (1–17) identify the region that represents all 16 possible nearest neighbor combinations. (C) An autoradiogram of a denaturing 10% PAGE revealing the self-cleavage activities of the starting RNA population (G0) and the populations at G6, G9, G12, and G15 after incubation in the absence (−) or presence (+) of 20 mM Mg²⁺ under otherwise permissive reaction conditions. The locations of the precursor (Pre), 5′ cleavage (5′ Clv), and 3′ cleavage (3′ Clv) fragments are indicated.

Characterization of class I and class II self-cleaving ribozymes. (A) Sequence variations for artificial phylogenetic analysis are depicted for the nts corresponding to the original random-sequence domain (numbered). Dots indicate no change from the prototype (listed first). Underlined bases retain base complementary. For example, the asterisk identifies a class I ribozyme variant that carries mutations at positions 18 and 26 that retain the ability to base pair. (B) Determination of cleavage sites for bimolecular (Fig. 2) class I and class II ribozymes (Left and Right, respectively). In each case, a trace amount of 21-nt substrate (5′ ³²P-labeled S21) was incubated with (+) or without (−) 500 nM ribozyme under the permissive reaction conditions. The 5′ cleavage fragments (5′ Clv) were separated from uncleaved S21 by denaturing 20% PAGE, and their sizes were compared with S21 5′ cleavage fragments prepared by partial ribonuclease (T1) and partial alkaline (⁻OH) digests. Class I and class II ribozymes cleave S21 after the G9 and G6 nts, respectively. (C) Cleavage reaction profiles for bimolecular class I (filled circles) and class II (open circles) ribozymes (Fig. 2) under the permissive reaction conditions. Observed rate constants for class I and class II ribozymes are 0.01 and 0.05 min⁻¹, respectively, as determined by the negative slope of the lines representing the progression of each reaction. (D) Compilation of the cleavage sites of the 12 classes of ribozymes. Numbers identify the nts within the nearest neighbor domain depicted in Fig. 1B.

Cleavage site versatility of the X motif. (A) Three 43-nt RNAs carrying the conserved features of the X motif ribozyme but differing in base pairing potential of arms I and IV were generated by in vitro transcription and targeted to cleave a 51-nt RNA. The lines represent nts within binding arms I and IV that are complementary to the X-E1 target site (arrow). Likewise, ribozymes X-E2 and X-E3 carry the arms that allow unique base pairing (dashes) with their corresponding target sites. (B) Cleavage of the RNA substrate (sub) upon incubation with three engineered ribozymes. Substrate (trace, 5′ ³²P-labeled) is not cleaved when incubated under in vitro selection conditions for 1 h in the absence of ribozyme (lane 1), but undergoes site-specific cleavage (clv) at the expected locations when incubated with 500 nM ribozymes X-E1, X-E2, and X-E3 (lanes 2–4, respectively).

Secondary structure models for 12 classes of self-cleaving ribozymes. Noted for each class is the generation at its first appearance, the rate constant for RNA cleavage (min⁻¹), and the cleavage site (arrowhead). Secondary structure models are based on artificial phylogenetic data (e.g., see Fig. 3A) or otherwise reflect the most stable structure predicted by the Zuker RNA mfold program that is accessible on the internet (44). Putative Watson–Crick base pairing and G⋅U wobble interactions are represented by dashes and dots, respectively. For most classes, the predicted secondary structure was used as a guide to generate bimolecular ribozyme–substrate complexes that are truncated relative to the original clones. Classes VI and VII were examined as unimolecular constructs because preliminary efforts to generate bimolecular constructs were unsuccessful. For class VII, the boxed regions identify variable (shaded) or conserved nts that reside in the original random-sequence domain.

Comparison of the secondary structures of two hammerhead ribozymes and the dominant X motif. (A) Variant hammerheads i and ii isolated from the G12 and G15 populations, respectively. Stem elements and nts in i are numbered according to the nomenclature defined by Hertel et al. (45) for hammerhead ribozymes. (B) The dominant ribozyme isolated after a total of 25 rounds of in vitro selection conforms to the X motif. This ribozyme can be reorganized in a bimolecular construct with a 43-nt enzyme and S21 substrate. Arrowheads identify the sites of ribozyme-mediated cleavage.