Nucleic acid enzymes: Playing with a fuller deck

Poor RNA. Being comprised of only four subunits, and chemically similar subunits at that, it is a wonder that RNA can have efficient enzymatic activity. Nonetheless, beginning with the discovery of RNA enzymes in nature (1, 2), and continuing with the development of novel RNA enzymes through in vitro evolution (3, 4), it is now abundantly clear that RNA can be an efficient and versatile catalyst. The repertoire of chemical reactions that can be catalyzed by RNA is broad (Table 1) and expanding steadily. Catalytic rates exceeding 1 sec⁻¹ and catalytic rate enhancements of up to 10¹³-fold have been reported (for recent reviews see refs. 5 and 6).

Table 1.

Reactions catalyzed by RNA and DNA enzymes

Reaction	RNA enzyme		DNA-enzyme
	Natural	Non-natural	Non-natural
Phosphoester transfer	X	X
Phosphoester hydrolysis	X	X	X
Polynucleotide ligation		X	X
Polynucleotide phosphorylation		X
Mononucleotide polymerization		X
Aminoacyl transfer		X
Amide bond cleavage		X
Amide bond formation		X
Peptide bond formation	X	X
N-alkylation, S-alkylation		X
Porphyrin metallation		X	X
Diels-Alder cycloaddition		X
Oxidative DNA cleavage			X

Natural RNA enzymes are derived from catalytic RNAs that exist in biological systems. Non-natural RNA and DNA enzymes are derived from catalytic motifs obtained by in vitro evolution. 

In seeking to explain the catalytic activity of RNA enzymes, several authors have pointed to the critical role that is played by metal cofactors, most commonly Mg²⁺ (7–9). Until recently it was thought that all RNA enzymes are metalloenzymes. Divalent metal cations help fold the RNA into a well-defined structure and assist in catalysis by functioning as either a Lewis acid or, when coordinated with water, a general acid base. During the past year, however, it has become clear that divalent metals are not essential for RNA catalysis. The hairpin ribozyme, for example, is as active in the presence of Mg²⁺ as it is in the presence of the exchange-inert metal complex cobalt hexamine (10–12). This strongly suggests that the functional groups of the RNA itself are responsible for catalysis.

The nucleotide components of RNA are lacking in chemical diversity, certainly as compared with the amino acid components of proteins. Just think what RNA could do if it possessed an imidazole side chain like histidine, a carboxyl group like glutamic acid, or a sulfhydryl group like cysteine. But RNA also must function as a genetic macromolecule, and in that context its chemical uniformity may be advantageous because it simplifies the tasks of replication and transcription. With regard to catalytic function, RNA lacks a general acid base with a pK_a in the neutral range, as occurs in histidine. On the other hand, the pK_a of N-1 of adenine (normally 3.5) or N-3 of cytosine (normally 4.2) can be shifted upward within a special local environment (13, 14). The phosphate groups along the RNA backbone may take on some of the functions of the carboxylate group of glutamic acid, and the internal 2′-hydroxyls and terminal 2′,3′-diol of RNA may be viewed as a partial substitute for the cysteine sulfhydryl.

If RNA is to be pitied for its lack of chemical functionality, then what are we to say about DNA? Those familiar with RNA catalysis and RNA tertiary structure have come to appreciate the importance of the 2′-hydroxyl group. DNA, lacking a 2′ hydroxyl, is thus expected to be further handicapped as a catalyst. Yet, a number of DNA enzymes have been described in recent years. Nature may not have had the opportunity or incentive to invent DNA enzymes, but this has been accomplished in the laboratory through in vitro evolution (for review see ref. 15). The range of chemical reactions that have been catalyzed by DNA is still limited compared with RNA (Table 1). However, more DNA enzymes are on the way, and at present it is a DNA rather than RNA enzyme that holds claim to being the nucleic acid enzyme with the highest catalytic efficiency (16). Furthermore, there are examples of DNA enzymes that do not require a metal cofactor for their catalytic activity (17, 18). The latest example of a metal-independent DNA enzyme, one that instead uses a histidine cofactor, is reported in this issue of the Proceedings (19).

We have grown accustomed to the fact that enzymes need not be comprised of amino acids; nucleotides can do very nicely. We have come to realize that metal cofactors, although useful, are not essential for nucleic acid catalysis, nor is the 2′-hydroxyl group that distinguishes RNA from DNA. What else can be jettisoned? Are all four nucleotides really necessary? What about the phosphates? Clearly there will come a point of diminishing return, beyond which further reduction of chemical diversity will be incompatible with efficient catalytic function. However, the effectiveness of Darwinian evolution should not be underestimated. Whether operating in nature or in the laboratory, evolution is a powerful optimization strategy that can drive even the most feeble heteropolymer to perform a chemical reaction.

It is perhaps more intriguing to think about the problem in the opposite way: how can the chemical diversity of nucleic acids be increased so as to increase their catalytic potential? One possibility would be to replace one of the standard nucleotides with a nucleotide analogue that has enhanced functionality. In developing an RNA enzyme with amide-bond-forming activity, Eaton and colleagues (20) replaced uridine with a 5-imidazole derivative of uridine, thereby capturing the flavor of histidine in a nucleic acid. The resulting amide synthase exhibited a catalytic rate enhancement of 10⁴- to 10⁵-fold, depending on the choice of metal ion cofactor. There are three 5-imidazole uridine residues within the catalytic domain of the enzyme. Their role in catalysis has not been established, but replacement of these residues with unmodified uridines abolished catalytic activity.

In developing an RNA enzyme that catalyzes a Diels-Alder cycloaddition reaction, Eaton and colleagues (21) replaced uridine by a 5-pyridyl derivative of uridine, anticipating a fruitful interaction between the pyridine “side chains” and a supplied Cu²⁺ cofactor. The resulting catalyst, which requires both the pyridine modifications and Cu²⁺, is the first example of a nucleic acid enzyme with carbon–carbon bond-forming activity. This activity could not be obtained when the four standard nucleotides were used (21).

The use of transcriptionally incorporated nucleotide analogues is a powerful approach for increasing the functional diversity of nucleic acids. A variety of ribonucleotide (22) and deoxyribonucleotide (K. Sakti and C. F. Barbas, personal communication) analogues can be incorporated in this way. However, this approach faces three important restrictions. First, in the context of in vitro evolution experiments, the genetic properties of the nucleotide analogue must be maintained. Thus the analogue, bearing a 5′-triphosphate, must be incorporated by a polymerase, and the resulting polymer must serve as a template for copying by a polymerase. Second, substitution of the analogue must occur globally. A mixed distribution of, say, uridines and 5-substituted uridines could not be maintained at specific locations along the polymer because both are templated by adenosine. Third, no more than four nucleotide analogues can be used at one time. Even if a third base pair was available that could be distinguished from the other two (23), the polymer would accommodate no more than six different subunits, as compared with the 20 subunits of proteins.

Another approach to increasing the functional diversity of nucleic acids is posttranscriptional modification. This is the way that modified bases are produced within transfer, ribosomal, and other biological RNAs. Posttranscriptional modifications can be made either chemically or enzymatically. If the modifications are made chemically, then it is difficult to restrict them to a particular nucleotide position. One important exception is the site-specific modification of a nucleotide analogue that has been incorporated at the 5′ end of an in vitro-transcribed RNA or within the primer portion of an in vitro-prepared DNA. If the modifications are made enzymatically, then the site of modification can be specific, as determined by the recognition properties of the modifying enzyme. The modifying enzyme may be a separate protein or nucleic acid enzyme, but also may be the nucleic acid that is itself undergoing modification. The latter case is exemplified by a group I ribozyme that catalyzes the self-incorporation of nicotinamide adenine dinucleotide (NAD⁺) or dephosphorylated coenzyme A (CoA) at a specific location close to its active site (24). But if one nucleic acid enzyme (or catalytic domain) is required to enhance the functionality of another, then a third might be needed to enhance the activity of the modifying enzyme, and so on. Fortunately, not every task will require modified nucleotides, and it is possible that a single modifying enzyme can operate on a family of nucleic acid catalysts.

Yet another approach to increasing the chemical diversity of nucleic acid enzymes is to provide a separate cofactor molecule that is bound by the enzyme and used in catalysis. This approach is potentially far-reaching, because it places few restrictions on the chemical nature of the cofactor. Near the top of the list of desirable cofactors is the amino acid histidine, coveted for its ability to act as a general acid base at near-neutral pH. In this issue of the Proceedings, Roth and Breaker (19) report the development of a DNA enzyme that is dependent on l-histidine for its ability to cleave a target RNA phosphoester. The imidazole group of the histidine cofactor appears to function as a general base in facilitating deprotonation of the 2′ hydroxyl that lies adjacent to the cleaved phosphoester. The enzyme does not require a divalent metal ion cofactor, and thus mechanistically is more akin to the protein enzyme ribonuclease A than to other RNA-cleaving nucleic acid enzymes (Fig. 1).

Figure 1.

Three approaches to the catalyzed cleavage of a RNA phosphoester. (A) Ribonuclease A uses two histidine residues, one as a general base (His-12) and the other as a general acid (His-119) (25). (B) Nucleic acid enzymes such as the hammerhead ribozyme (26) and 10–23 DNA enzyme (16) use Mg(OH)⁺ as a general base. (C) The Roth and Breaker DNA enzyme (19) uses a histidine cofactor as a general base.

The Roth and Breaker study opens the door to what will be a bounty of in vitro-evolved nucleic acid enzymes that are assisted by nonmetal cofactors. Evolving populations of RNA or DNA molecules could be challenged to use a variety of cofactors, including biological and nonbiological amino acids, peptides, small organic molecules, and even compounds that are selected from a library of potentially useful ligands.

An important lesson that emerges from the Roth and Breaker study concerns the hurdle that must be overcome in developing a cofactor binding site within a nucleic acid enzyme. In principle, this can be accomplished through in vitro evolution, provided that the bound cofactor contributes to the fitness of the enzyme. It may be difficult, however, to achieve both tight binding of the cofactor and high catalytic efficiency, because these two tasks could place conflicting demands on the preferred sequence of the enzyme. Studies involving the in vitro selection of aptamers (nucleic acids that bind a target ligand) have made it clear that selection for binding alone typically leads to the development of a high-affinity binding site (for review see ref. 27). But for catalysis, the bound cofactor must be properly oriented with respect to the substrate, making high-affinity binding less assured. The histidine-dependent DNA enzyme of Roth and Breaker has an observed rate of 0.2 min⁻¹ in the presence of 100 mM histidine, corresponding to a catalytic rate enhancement of ≈10⁶-fold. Even at this high concentration of cofactor, the histidine binding site remains unsaturated. Variant forms of the enzyme that have a higher affinity for histidine also have a lower maximum catalytic rate (19).

The co-optimization of cofactor binding and cofactor-dependent catalysis is a surmountable problem. One solution would be to increase the number of nucleotides in the enzyme so that it has sufficient complexity to accomplish both tasks. Another solution would be to choose cofactors that are known to bind readily to nucleic acids and are expected to contribute significantly to catalysis. What cofactors will prove to be most favorable? Drawing on chemical intuition and in vitro selection experiments, it should be possible to delineate a basic toolbox of cofactors that can be applied to a broad range of catalytic tasks. These cofactors could be sought one at a time, but it would be preferable to allow the evolving population of nucleic acids to choose the most useful cofactors from a library of compounds.

This process of cofactor discovery is reminiscent of a nucleic-acid-based selection experiment that took place on a much grander scale during the early history of life on earth. If one accepts the notion of an “RNA world,” an ancestral genetic system based on RNA genomes and RNA enzymes, then one must face up to RNA’s functional limitations. In the RNA world, individuals that could use compounds from the environment as catalytic cofactors would have had a selective advantage. They would have enjoyed greater advantage if they could synthesize those cofactors from readily available starting materials. Apparently, particular advantage was offered by peptide cofactors that could be synthesized by RNA. Thus, what may have begun as dabbling with amino acid cofactors eventually led to the origin of a translation apparatus and the transfer of catalytic responsibilities from RNA to proteins.