Analogous Enzymes: Independent Inventions in Enzyme Evolution

Origin of Analogous Enzymes

The most likely mechanism for the evolution of analogous enzymes appears to be recruitment of existing enzymes that take over new functions by virtue of changed substrate specificity or a modified catalytic mechanism. Such a scenario could be inferred for about one half of the analogous enzyme sets by showing that at least one of them is homologous to a family of enzymes catalyzing a different, albeit related, reaction (Table 4). The argument for recruitment is particularly convincing when one of the analogous enzyme forms is found in a limited number of species, whereas the family to which it belongs is common. For example, gluconate kinase fromBacillus subtilis (GNTK_BACSU, EC 2.7.1.12) seems to have orthologs with the same activity only in otherBacillus species and is unrelated to gluconate kinases from other organisms. However, it belongs to a large kinase family that includes xylulose kinases and glycerol kinases from a variety of organisms (Galperin and Koonin 1998). Thus, the Bacillusgluconate kinase most likely evolved as the result of a duplication of a gene for xylulose or glycerol kinase in the Gram-positive lineage, which was followed by a shift in the substrate specificity. A similar mechanism is likely to account for the origin of the 2,3-bisphosphoglycerate-independent phosphoglycerate mutase from an alkaline phosphatase-like enzyme and of the second phosphofructokinase of Escherichia coli (PfkB) from an enzyme of the ribokinase family (Table 4).

Some of the analogous enzymes can be described as “second edition” (Doolittle et al. 1986), that is, enzymes that perform functions related to adaptation to new environments and life styles and usually have a limited phylogenetic distribution. In cases like this, the recruitment hypothesis seems plausible for all analogous forms. For example, the three analogous apyrases (ATP-diphosphohydrolases, EC3.6.1.5), typically extracellular or membrane-bound enzymes that are involved in such specific functions [e.g., prevention of blood clotting (Champagne et al. 1995)], seem to have been derived from three large, unrelated enzyme families, each with its own 3D fold (Tables 2 and 4).

When neither of the analogous isoforms has detectable sequence similarity to any enzymes of different specificity, the recruitment scenario may still apply, but the connection to the progenitor enzyme might have become undetectable owing to the recruitment having occurred very early in evolution and/or because of rapid change associated with the acquisition of the new function.

The Most Diverse Groups of Enzymes

Table 3 shows that analogous enzymes are more or less uniformly distributed among all enzyme classes. A closer examination of their functions, however, shows that the majority of analogous enzymes belongs to several functional groups. The most conspicuous one contains enzymes involved in synthesis and hydrolysis of polysaccharides (Davies and Henrissat 1995; Henrissat and Davies 1997). This group includes 51 different enzyme forms, belonging to 24 EC nodes, and contains glycosyl transferases, glycosyl hydrolases, and pectate and alginate lyases. The most striking examples are 6-phospho-β-glucosidase (EC 3.2.1.86), which is found in bothE. coli and B. subtilis genomes in two unrelated forms, each represented by three paralogous genes, and cellulase (EC3.2.1.4), found in six forms, belonging to at least three different folds (Table 1).

The second large group of analogous enzymes (nine EC nodes) deals with the effects of oxygen on cellular components. It includes, among others, cytochrome c peroxidase, catalase, peroxidase, chloroperoxidase, superoxide dismutase, glutathioneS-transferase, and glutathione synthetases I and II.

The third group (seven EC nodes) consists of enzymes involved in the synthesis and turnover of bacterial and eukaryotic cell walls. It includes lysozyme (N-acetylmuramidase), hyaluronidase, penicillin acylase, N-acetylmuramoyl-l-alanine amidase, β-lactamase, phosphomannose isomerase, and phosphomannomutase.

Notably, these groups of enzymes are mostly involved in specific rather than universal cellular functions, suggesting that they might have evolved relatively recently.

Analogous Enzymes in Central Metabolism

Although less abundant than in functions like cell wall biosynthesis or cell defense, analogous enzymes can be found in a variety of metabolic pathways; however, no central pathways were detected that would consist exclusively of such enzymes. Rather, analogous enzymes typically are sandwiched between those that are universally conserved. Among glycolytic enzymes, phosphofructokinase, phosphoglycerate mutase, and lactate dehydrogenase have analogous forms, whereas glucokinase and aldolase are represented by highly divergent, albeit structurally similar, forms, in bacteria and eukaryotes. The presence of analogous enzymes in the early steps of glycolysis and the abundance of analogous enzymes catalyzing other reactions of hexose metabolism support the conclusion that only the lower part of glycolysis from glyceraldehyde 3-phosphate to pyruvate is ubiquitous and indispensable for life and that the original function of the Embden–Meyerhof pathway could have been biosynthetic, rather than glycolytic (Romano and Conway 1996).

The same patchy distribution of analogous enzymes can be seen in reactions of amino acid metabolism. Unrelated forms of asparaginase, 3-dehydroquinate dehydratase, and arginine decarboxylase are found in pathways that can serve both biosynthetic and biodegradative functions. Predictably, biosynthetic forms are mostly present in autotrophs, for example, cyanobacteria and plants, whereas biodegradative forms are typical of heterotrophs. In two of the three cases above, E. coli encodes both versions of the enzyme.

Analogous Enzymes in DNA Replication, Transcription, and Translation

DNA replication systems present prominent examples of analogous enzymes, although these are not easily captured by the automatic procedure used here owing to the multisubunit and/or multidomain structure of the key components (e.g., DNA polymerases) and the difficulties of their placement within the framework of current enzyme classification (e.g., lack of an appropriate rule to distinguish DNA-dependent RNA polymerases involved in transcription and DNA primases). None of the essential DNA replication enzymes are orthologous in bacteria as compared with archaea and eukaryotes, leading to the speculation that the last common ancestor of all extant life forms might not have had a DNA genome at all (Mushegian and Koonin 1996). In particular, no sequence similarity can be detected between the dNTP polymerization domains of the replicative polymerases (DNA polymerase III α-subunit in bacteria and B family DNA polymerases in archaea and eukaryotes), suggesting that these central components of the DNA replication machinery are indeed analogous; the final conclusion, however, should await the determination of polymerase III 3D structure. A striking example of analogy in replicative enzymes that was automatically detected by our procedure is type I DNA topoisomerases, with two unrelated families including bacterial topoisomerases I together with bacterial and eukaryotic topoisomerase III, and eukaryotic topoisomerase I, respectively (Table 1). Another crucial step in DNA replication in bacteria and archaea/eukaryotes is catalyzed by apparently unrelated DNA ligases, though the EC numbers are different in this case as NAD and ATP, respectively, are used as substrates.

Although the transcription and translation machineries are generally uniform in all life forms and analogous enzymes are not typical, there are notable exceptions. Thus, the archaeal lysyl-tRNA synthetase, orthologs of which have been unexpectedly discovered in spirochaetes, belongs to aminoacyl-tRNA synthetase class I as opposed to the class II lysyl-tRNA synthases found in the rest of bacteria and in eukaryotes (Ibba et al. 1997b). In this case, unrelated 3D folds are obvious, and evolution by recruitment of a class I enzyme appears most likely (Koonin and Aravind 1998).

Analogous Enzymes in Signal Transduction

As enzymes with functions in signal transduction may have originally evolved from metabolic enzymes, independent inventions in this area seem particularly likely. Because cAMP plays substantially different roles in bacterial and eukaryotic cells (Saier 1996), it is perhaps unsurprising that its synthesis is catalyzed by apparently unrelated enzymes in bacteria and eukaryotes. Pertussis and anthrax toxins also have adenylate cyclase activity, constituting yet another, third class of adenylate cyclases.

The two analogous enzymes that hydrolyze the intracellular signal molecule diadenosine tetraphosphate have been apparently recruited from two ancient classes of hydrolases, namely the HIT superfamily (APH1_SCHPO) and the NTP pyrophosphohydrolase (MutT) superfamily (AP4A_HUMAN) (Table 4).

The epitome of regulatory enzymes, serine–threonine protein kinases, exist in three unrelated forms. Although the great majority of them have the classical protein kinase fold (typified by Src), bacterial protein kinases containing P loops have been described recently (Galinier et al. 1998; Reizer et al. 1998), and furthermore, histidine kinase homologs have been described that phosphorylate specific serines in their target proteins (Popov et al. 1993; Yang et al. 1996). Although not detected automatically by our procedure owing to problems with assigning EC numbers, this is a striking example of apparent evolution of analogous enzymes by recruitment.

Bacteria and eukaryotes use analogous enzymes to synthesize glutathione, the universal regulator of the thiol–disulfide ratio in the cell. Both reactions of glutathione biosynthesis in E. coli are catalyzed by enzymes, namely glutamate–cysteine ligase (EC 6.3.2.2) and glutathione synthetase (EC 6.3.2.3), that are unrelated to the respective enzymes from yeast and humans. Remarkably, plants have a typical eukaryotic glutathione synthetase, but their glutamate–cysteine ligase is unrelated to either bacterial or yeast enzyme.

Phylogenetic Distribution of Analogous Enzymes

Comparison of microbial genomes has shown that species with larger genomes contain multiple paralogous genes, coding for enzymes with similar catalytic properties. In organisms with small genomes, not only the absolute number but also the fraction of proteins that belong to paralogous families is considerably lower (Koonin et al. 1997). In the same vein, the data in Table 5 show that organisms with small genomes, both parasitic and free-living ones, encode disproportionally small numbers of analogous enzymes. This observation is confirmed by the analysis of specific metabolic pathways such as glycolysis and purine biosynthesis, in which organisms with larger genomes have analogous enzymes for certain steps, whereas organisms with small genomes typically have only one form (Koonin et al. 1998). Thus, biochemical diversity, manifest in the variety of both analogs and paralogs, is a luxury enjoyed mostly by organisms with large genomes.

Analogous Enzymes and Enzyme Classification

We showed that numerous biochemical reactions may be catalyzed by enzymes without detectable sequence similarity and, in some cases, with demonstrably distinct 3D structures. In other words, many enzymatic activities have been independently invented in evolution on more than one occasion. This is not to say that these enzymes have nothing in common—although not sharing common ancestry, they still may have similar reaction mechanisms and even similar local active center geometries. Identification of such common features in analogous enzymes seems to be an interesting direction for future research that may shed new light on mechanisms of enzymatic catalysis.

The current classification of enzymes on the basis of the catalyzed reactions is an indispensable tool for enzymologists, but in the case of analogous enzymes, it inevitably fails. A hierarchical system of protein classification constructed by sequence and structure comparison, like the ones already proposed for peptidases (Barrett 1994) and glycosidases (Henrissat and Davies 1997), can handle these cases and will provide a wealth of information complementary to the information currently embodied in the EC system.

Implications for Genome Annotation and Drug Design

Functional annotations of proteins identified in the course of genome sequencing projects rely mostly on sequence similarities between newly identified proteins and those already in the databases. Although this process can produce many important insights into structure, evolution, and catalytic properties of various proteins (e.g., Galperin and Koonin 1997; Aravind et al. 1998), it can also lead to misannotations, which tend to spread as newly sequenced proteins are assigned functions based on their similarity to previously misannotated proteins (Bhatia et al. 1997). Enzyme recruitment, leading to a significant change of function accomplished by relatively minor sequence changes, is a phenomenon that can significantly affect the quality of similarity-based functional annotation. The examples of enzyme recruitment, identified here (Table 1) and available in the WWW supplement, may be useful to prevent such erroneous annotations.

In many cases, the phylogenetic distribution of analogous enzymes is such that one enzyme form is found in bacteria and the other one in eukaryotes. The enzyme forms that are absent in eukaryotes (particularly humans) can be used as targets for the development of new antibacterial drugs that would have a low chance of side effects. This strategy may be especially promising for targeting pathogenic bacteria as their genomes generally have a low level of enzyme redundancy and code for only one form of the analogous enzymes (Table 5).