Dichotomic Phylogenetic Tree of the Pyruvate Kinase Family

K+ dependence was assumed to be a feature of all pyruvate kinases until it was discovered that some enzymes express K+ -independent activity. Almost all the K+-independent pyruvate kinases have Lys at position 117, instead of the Glu present in the K+-dependent muscle enzyme. Mutagenesis studies show that the internal positive charge substitutes for the K+ requirement (Laughlin, L. T. & Reed, G. H. (1997) Arch. Biochem. Biophys. 348, 262–267). In this work a phylogenetic analysis of pyruvate kinase was performed to ascertain the abundance of K+ -independent activities and to explore whether the K+ activating effect is related to the evolutionary history of the enzyme. Of the 230 studied sequences, 46% have Lys at position 117, and the rest have Glu. Pyruvate kinases with Lys117 and Glu117 are separated in two clusters. All of the enzymes of the Glu117 cluster that have been characterized are K+-dependent, whereas those of the Lys117 cluster are K+-independent. Thus, there is a strict correlation between the dichotomy of the tree and the dependence of activity on K+. 77% of the pyruvate kinases that possess Lys117 have Lys113/Gln114; they also have Ile, Val, or Leu at position 120. These residues are replaced by Glu117 and Thr113/Lys114/Thr120 in 80% of K+-dependent pyruvate kinases. Structural analysis indicates that these residues are in a hinge region involved in the acquisition of the catalytic conformation of the enzyme. The route of conversion from K+-independent to K+-dependent pyruvate kinases is described. A plausible explanation of how enzymes developed K+ dependence is put forth.

The dependence of enzyme activity on monovalent cations is widespread. Indeed more than 100 enzymes exhibit structural and catalytic requirement for either K ϩ or Na ϩ (1)(2)(3)(4). In some enzymes, the monovalent cations participate in the formation of a ternary complex of enzymes and substrates, whereas in other, cations act allosterically or stabilize the protein structure (3)(4)(5)(6)(7). Structural information on the molecular origin of these effects has begun to emerge; however, there are still many questions that remain to be solved. Pyruvate kinase, the first enzyme in which an absolute requirement for K ϩ has been documented (8), is an excellent model enzyme for studying the effect of monovalent cations particularly because in the rabbit muscle enzyme the activating effect of K ϩ is about 10,000-fold (9,10), which, to our knowledge, is the highest reported thus far. In regard to the mechanism of action of K ϩ , it is known that this cation changes the ultraviolet and fluorescence spectra of pyruvate kinase (11)(12)(13) as well as its immunoelectrophoretic pattern (14). K ϩ also changes the structure of the active site (15)(16)(17)(18). In this respect, NMR studies of Tl ϩ bound to pyruvate kinase show that the alkali metal ion binds within 0.8 nm of the Mn 2ϩ -binding site (16) and that this distance shifts to 0.49 nm upon binding of the substrate phosphoenolpyruvate (19). In addition, it has been shown that K ϩ changes the kinetic mechanism of rabbit muscle pyruvate kinase and is involved in the acquisition of the active conformation of the enzyme (20). The K ϩ -binding site is highly conserved in all crystallographic structures reported to date. K ϩ is coordinated by O-␦1 of Asn 74 , O-␥ of Ser 76 , O-␦2 of Asp 112 , the carbonyl oxygen of Thr 113 (21), a water molecule, and a phosphate oxygen of phosphoenolpyruvate analogs (22), or an oxygen from the ␥-phosphate of ATP (5).
For a long time, it was thought that the absolute dependence of K ϩ was a common feature to all pyruvate kinases (8,(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38). However, as the number of characterized enzymes increased, it became apparent that the activity of several pyruvate kinases was K ϩ -independent, for example those from Escherichia coli (Type II) (39), Phycomyces blakesleeanus (40), Corynebacterium glutamicum (41), Zymomonas mobilis (42), Thermoproteus tenax (43), Synechococcus pcc 6301 (44), Archaeoglobus fulgidus, Pyrobaculum aerophilum, Aeropyrum pernix, and Thermotoga maritima (45). To explore the molecular basis of this different behavior, Laughlin and Reed (46) compared the amino acid sequence of rabbit muscle pyruvate kinase with those of E. coli type II and of C. glutamicum and found that Glu 117 of the rabbit enzyme, which is close to the K ϩ -binding site, is a Lys in the bacterial enzymes. The authors constructed the E117K mutant of the rabbit enzyme and found that the mutant is Ͼ200-fold more active than the wild type in the absence of K ϩ and exhibits no stimulation by monovalent cations. They also constructed two other mutants, E117A and E117D; these enzymes were activated by K ϩ . The data clearly indicate that the positive charge of Lys mimics the action of K ϩ (46). Mimicry of monovalent cation activation by a Lys has also been reported in chaperon HSC70 (47) and H ϩ -translocating inorganic pyrophosphatase (48). In consonance with the data of Laughlin and Reed (46), the sequences of five of the character-ized K ϩ -independent enzymes also contain Lys in position 117 (39,41,42,44,45). However, pyruvate kinase from T. tenax (43) and P. aerophilum (44) exhibit K ϩ -independent activity, albeit with a Ser at position 117. These observations thus indicate that the expression of K ϩ -independent activity is not completely explained by the charge of residue 117 and that other residues in the vicinity of the active site may be involved in conversion of the K ϩ -independent to K ϩ -dependent pyruvate kinase.
In this work we performed an extensive phylogenetic study of the pyruvate kinase family. The purpose of the study was to ascertain the abundance of K ϩ -independent activities and to explore whether the K ϩ activating effect is related to the evolutionary history of the enzyme. Phylogenetic analyses of pyruvate kinase family have been reported previously (43,45,49). Schramm et al. (43) and Johnsen et al. (45) found a dichotomic structure of the phylogenetic tree that does not coincide with the universal tree topology (Bacteria, Archaea, and Eukarya) (50,51); this unusual topology may be the result of ancient gene duplication or lateral transfer events (43). The latter studies were an attempt to correlate the major subfamilies or clusters to the type of allosteric effector of the enzymes. According to Schramm et al. (43), cluster I mostly includes enzymes activated by sugar phosphates, whereas cluster II comprises enzymes regulated by nucleotides. Given that some enzymes of cluster I are regulated by effectors of cluster II and vice versa and some even lack heterotropic allosteric regulation, a classification based on allosteric effectors of the enzymes may be more complex.
Our phylogenetic data show that there are numerous enzymes that have Lys at position 117 (the numbering used is that of rabbit muscle pyruvate kinase). Because pyruvate kinases that contain Lys 117 and the E117K mutant of the rabbit enzyme are K ϩ -independent, it may be safely assumed that K ϩ -independent pyruvate kinases are broadly present in nature (106 sequences from 230 have Lys). In accordance with Schramm et al. (43), we found a dichotomic topology of the pyruvate kinase phylogenetic tree. However, in our case, cluster I groups sequences that include Glu 117 correspond to K ϩ -dependent enzymes, whereas cluster II comprise sequences that have Lys 117 and are therefore K ϩ -independent enzymes. In addition, we found that changes in the residue at position 117 are consistently accompanied by conservative changes in residues 113, 114, and 120. In all likelihood these residues are mechanistically important, because they localize in a hinge bending region that participates in the acquisition of the active catalytic conformation of the enzyme.

EXPERIMENTAL PROCEDURES
The search for pyruvate kinase sequences was performed at the World Wide Web site of the National Center for Biotechnology Information (52). The integrated data base retrieval system ENTREZ (52) was used to access the National Center for Biotechnology Information data base. The search for homologous sequences was performed with gapped BLASTP and PSI-BLAST using default gap penalties and BLOSUM62 substitution matrix (53). The sequence of rabbit muscle pyruvate kinase was the starting point in the search for homologous sequences.
Translated nucleotide sequences were retrieved from the Gen-Bank TM of the National Center for Biotechnology Information.
Progressive multiple sequence alignment was calculated with the ClustalX package (54), using secondary structure-based penalties. The alignment was manually corrected according to the results of gapped BLASTP and three-dimensional alignments obtained from the three-dimensional structure data base of Entrez (55). Phylogenetic analyses were performed with MEGA 3.1 software (56) using maximum parsimony, minimum evolution, and neighbor-joining methods with the aid of the empirical Jones-Taylor-Thornton amino acid substitution model. Differences between amino acid sequences were corrected for multiple substitutions assuming gamma distribution for rate variations among sites. The gamma-shaped parameter (a ϭ 1.0) was estimated with the Whelan-Goldman matrix of substitutions and the eight-category discrete-gamma model using TREE-PUZZLE 5.2 (57). Confidence limits of branch points were estimated by 500 bootstrap replications.
Structural analysis of domain movements and hinge bending regions of pyruvate kinase were performed with DynDom software. DynDom is a program that determines dynamic protein domains, hinge axes, and amino acid residues involved in hinge bending (58).

RESULTS AND DISCUSSION
Phylogenetic Analysis-The search for sequences of pyruvate kinase comprised the data reported up to August 2004. A total of 230 sequences were retrieved: 69 eukarya (21 animals, 17 fungi, 16 plants and 15 protists), 143 bacteria, and 18 archaea (Fig. 1). Position 117 is occupied by Glu in 121 sequences and by Lys, Ser, and Arg in 106, 2, and 1 sequence, respectively. The requirement for K ϩ by 21 pyruvate kinases that have Glu at position 117 has been examined, and the activity of all of them is K ϩ -dependent (8,(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35)(36)(37)(38). The activities of eight pyruvate kinases that do not have Glu at that position have been characterized; in all cases the activity is K ϩ -independent. Therefore, there is an excellent correlation between the dependence of activity on K ϩ and the existence of Glu at position 117. About half of all the sequences analyzed (46%) have Lys 117 ; thus, the K ϩ -independent activity is as frequent as the K ϩ -dependent activity.
Our phylogenetic analysis shows a dichotomic tree structure ( Fig. 1) that, similar to a previous phylogenetic study (43), does not coincide with the universal tree topology (Bacteria, Archaea, and Eukarya) (50,51). As discussed by Schramm et al. (43), this unusual topology may be caused by an early gene duplication event. However, a remarkable feature of our phylogenetic data is the high correlation between the dichotomic nature of the tree and the dependence on K ϩ . As shown in Fig.  2, the red branch corresponds to the enzymes that contain Glu 117 , which include the 21 experimentally characterized K ϩdependent enzymes. The only exception found in this branch is pyruvate kinase from the anaerobic bacteria Clostridium perfringens, (GenBank TM accession number Q46289) that has Lys at position 117. To gain insight into the peculiar position of pyruvate kinase of C. perfringens, we examined the sequences that had been deposited (in GenBank TM ) in the last year. We found four new sequences belonging to the genus Clostridium.
One of these sequences corresponds to a pyruvate kinase of C. perfringens that surprisingly has Glu in position 117. We also found that Clostridium acetobutylicum has K117 pyruvate kinase, in addition to the E117 enzyme included in the phylo-genetic tree. This property is also shared by Clostridium beijerinckii where both Glu 117 and Lys 117 pyruvate kinases were found. The data on the Clostridium genus are highly suggestive that the two enzymes resulted from recent gene duplication. In Available protein sequences belonging to the pyruvate kinase family were used to construct the phylogenetic topology with the minimum evolution method. Similar topologies were obtained using the neighbor-joining and maximum parsimony methods. Trees were calculated using MEGA 3.1 (26). Dots indicate nodes supported in Ն70% (open), Ն80% (gray), or Ն90% (closed ) of 500 random bootstrap replicates of all neighbor-joining, minimum evolution, and maximum parsimony trees. Taxonomic groups are indicated by the following colors: animal in red, fungi in brown, kinetoplastida in orange, apicomplexa in olive green, plants in green, ␥-proteobacteria in blue, firmicutes in pink, chlamidiae in gray, euryarchaeota in navy blue, crenarchaeota in purple, diplomonanida in dark green, alfa-proteobacteria in magenta, actinobacteria in dark yellow, cyanobacteria in wine red, ␤-proteobacteria in cyan, and single represented bacterial taxonomic groups in black. Sequence names are indicated according to a Swiss-Prot-like identifier (accession number ϩ organism). Characterized K ϩ -dependent and -independent enzymes are indicated by red and blue asterisks. A complete list of the names of the organisms included in the tree is provided in supplemental Appendix 1.
fact, the identity between the Glu and Lys enzymes of C. perfringens is 52.6%, whereas identity between early duplicated pyruvate kinases of E. coli in cluster I and cluster II is 36.5%.
The blue branch (Fig. 1) contains enzymes with Lys 117 that comprise the eight experimentally studied K ϩ -independent enzymes; it also includes the K ϩ -independent enzymes with Ser and Arg at position 117 (indicated in black) and three archaeal enzymes that contain Glu 117 . Thus the branching of the phylogenetic tree concurs with the residue at position 117. Accordingly, the existing pyruvate kinases may be divided into two The branches of the sequences with Glu, Lys, and Ser or Arg in position 117 are illustrated in red, blue, and black, respectively. Characterized K ϩ -dependent (Glu 117 ) and K ϩ -independent enzymes (Lys, Ser, or Arg at position 117) are indicated by red and blue asterisks, respectively. As shown, residues 113, 114, and 120 vary depending on residue 117. The numbering used is according to the sequence of rabbit muscle pyruvate kinase. The phylogenetic topology and node dots are described in the legend for Fig. 1. groups: group I is formed by the K ϩ -dependent enzymes and Group II by the K ϩ -independent enzymes. As this classification is based on an all-or-none property, i.e. K ϩ -independent or -dependent activities, it has advantages over classification based on allosteric effectors (43,45). It is relevant to acknowledge that K ϩ dependence has been used as a criterion to explain the dichotomic topology of the phylogenetic tree of the H ϩ -translocating inorganic pyrophosphatase family (48). However the lack of structural information available on these enzymes does not allow the establishment of structural correlations between amino acid substitutions and the K ϩ -binding site.
Co-evolving Amino Acids with Residue 117-The phylogenetic analysis also shows that there is a correlation between the residue that exists at position 117 and some of its neighboring residues (Table 1). When Glu is in position 117, in 99% of the cases the residue in position 113 is Thr. Likewise, when Lys is in position 117, position 113 is occupied by Leu in 98% of the enzymes. A similar, but slightly less strict pattern was observed between residue 117 and those at positions 114 and 120. The data show that: 1) pyruvate kinases that have Glu 117 have Lys 114 in 94% of the cases; 2) when Lys is in position 117, position 114 is occupied by Gln in 83% of the enzymes; 3) enzymes with Glu in position 117 have Thr in position 120 in 83% of the cases; 4) enzymes with Lys 117 have a hydrophobic residue in position 120 (Leu, Val, or Ile) in 93% of the cases.
The main chain carbonyl oxygen of Thr 113 participates in K ϩ binding. As shown in Table 1, this residue is almost invariably present when Glu is in position 117 (cluster I). It is likely that the conservation of Thr 113 is the result of the evolutionary pressure to keep the integrity of the K ϩ -binding site intact. In the K ϩ -independent enzymes (cluster II), Leu 113 is practically invariable (98%). This is rather atypical because there is no obvious evolutionary benefit for the conservation of this residue. Leu 113 is associated to a positive charged residue in position 117 (Lys, Arg) or to serine. The constancy of Leu 113 in the enzymes of cluster II suggests that it is involved in the expression of the K ϩ -independent activity. It is also noteworthy that in the K ϩ -independent activity cluster, aside from the enzymes that contain Ser 117 , three archaeal enzymes that have Glu in position 117 have Leu in position 113 (see Fig. 2). These enzymes have not been characterized, but their localization in the phylogenetic tree and the presence of Leu 113 suggest that they are K ϩ -independent enzymes.
Concerning residue 114, Laughlin and Reed (46) found that in the rabbit muscle E117K mutant, the presence of Lys (the most abundant residue found in position 114 in K ϩ -dependent enzymes) or Gln (found mostly in K ϩ -independent enzymes) did not affect the kinetics of the enzyme. Therefore, it is possible that only the pair Leu 113 /Gln 114 participates in K ϩ -independent activity. In fact, there is not a single pyruvate kinase that has the pair Thr 113 /Gln 114 .
No functional role has been assigned to residue 120; however the co-evolution between this amino acid and the residue 117 (83% in cluster I and 93% in cluster II) makes it an attractive target to study. Finally, it is noted that in the transition between clusters I and II, there are about 30 sequences that do not show the conserved pattern in residues 114 and/or 120. This suggests that these positions are not as critical as positions 113 and 117 for monovalent cation dependence.
The overall analysis of K ϩ -dependent pyruvate kinases (cluster I) shows that the sequence Thr 113 /Lys 114 /Glu 117 /Thr 120 is conserved in 80% of the cases, whereas in the K ϩ -independent enzymes (cluster II) the sequence Leu 113 /Gln 114 /Lys 117 /(Leu-Val-Ile) 120 is conserved in 77% of the enzymes. The co-evolution of residues 113, 114, and 120 with that in position 117 suggest that these residues are involved in the activating effect of K ϩ in the K ϩ -dependent enzymes or in the expression of the K ϩ -independent activity. In this regard, it is relevant that the data presented of Laughlin and Reed (46) show that in the rabbit muscle enzyme, Lys 117 may replace K ϩ ; however the activity of the mutant is about 10% of the activity of the wild type with K ϩ . Therefore, further substitutions are needed to attain full enzymatic activity.
Transition from K ϩ Independence to K ϩ Dependence-The broad phyletic distribution of K ϩ -independent pyruvate kinases (cluster II) in the three domains of life (Archaea, Bacteria, and Eukarya) in comparison with K ϩ -dependent pyruvate kinases (cluster I, found in Bacteria and Eukarya only) suggest that K ϩ -independent activity preceded the appearance of the K ϩ -sensitive activity. To explain the possible route through which K ϩ -independent are converted to K ϩ -dependent pyruvate kinases, we analyzed the amino acid codons for positions 113, 114, 117, and 120. As shown in Fig. 3, all of the changes in positions 113, 114, 117, and 120 took place by single nucleotide substitutions, except in the Antarctic hair grass, Deschampsia antarctica (GenBank TM accession number AAM22747), and the sulfate-reducing proteobacteria, Desulfovibrio desulfuricans (GenBank TM accession number ZP_00130421), where more than one substitution was needed. Because the frequency of mutations of amino acid codons occurs in the order of

Co-evolving residues of consensus sequences in clusters I and II
Clusters I and II are formed by 119 and 111 sequences, respectively. Cluster I has Glu 117 in 118 cases, and cluster II has Lys 117 in 105 cases; these numbers represent 100% for each group. The first column indicates the amino acids co-evolving with position 117, and the second column indicates the percent in which the respective association is present.

Cluster I K ؉ -dependent enzymes Cluster II K ؉ -independent enzymes
Residues co-evolving with Glu 117 % Residues co-evolving with Lys 117 % third Ͼ second Ͼ first nucleotide, it is possible to trace the probable series of events that led to the changes of the residues (Fig. 3). For example, in position 117 (Fig. 3A), the mutation in the third base would account for the conversion of Ser to Arg; likewise the mutation in the second nucleotide leads to the transformation of Arg to Lys and in the first nucleotide to the transformation of Lys to Glu. By the same rationale, Fig. 3, B and C, shows the routes of conversion of residues 113 and 120. In regard to residue 114 of group I, it is possible that variations of this residue (Arg, Ser, and Asn) stemmed from single nucleotide substitutions of the Lys codon, whereas the changes in group II (Pro, Ala) derived from single substitutions of the Gln codon (Fig. 3D).
Conservation of the K ϩ -binding Site-To explore whether K ϩdependent and independent enzymes differ in the region of the K ϩ -binding site, the level of conservation of residues that contribute to the coordination of the cation was determined. It is noted, however, that all of the crystallographic structures of pyruvate kinase reported so far are K ϩ -dependent. In all of them K ϩ is coordinated through O-␦1 of Asn 74 , O-␥ of Ser 76 , O-␦2 of Asp 112 , and by the backbone carbonyl oxygen of Thr 113 (21). In the 230 sequences that were studied, Asn 74 and Asp 112 are conserved except for the pyruvate kinases from Giardia intestinalis and D. antarctica. Ser 76 is less conserved; it is replaced by Ala in 12 and by Cys in 5 pyruvate kinases sequences. Collectively, the data indicate that the residues that form the K ϩ -binding site, with the exception of Thr 113 , which is almost exclusive of the K ϩ -dependent pyruvate kinases, are highly conserved even in enzymes that do not require K ϩ for activity.
Structural Analysis of the Consensus Sequence-To explore the contribution of residues 113, 114, 117, and 120 to the dynamics of pyruvate kinase, the enzyme was analyzed with the DynDom software (58). DynDom is a program that identifies domains, hinge axes, and hinge bending residues in proteins. The program may be applied when there are crystallographic data of a protein in two or more conformations. The DynDom data base of all the crystallographic structures of pyruvate kinase available in the Protein Data Bank was used for the analysis.
The enzyme is formed by four structural domains: the N-terminal and the A, B, and C domains. The active site is in a cleft formed by A and B domains, which are linked by two loops. Rotation of the B domain over the A domain closes the active site cleft. DynDom assigns domain B as the mobile domain, and the N-terminal and A and C domains as the fixed domain. Two hinge bending regions were assigned: a short hinge formed by the sequence 114 -118 (green in Fig. 4) and a long hinge formed by residues 209 -224 (yellow in Fig. 4). Therefore residues 113 and 120 are, respectively, at the beginning and end of the short hinge, whereas residue 114 and 117 are part of the hinge (Fig. 4).
Crystallographic data have clearly demonstrated that K ϩ provides a positive charge to coordinate the phosphate group of phosphoenolpyruvate (22) or ATP (5). In addition, NMR stud-  ies (15)(16)(17)(18)(19) have shown that monovalent cations induce structural rearrangements in the active site of rabbit muscle pyruvate kinase. Similar findings have also been documented in yeast pyruvate kinase where the interaction of the monovalent cation with the wild type enzyme (59,60) and Thr 298 mutants (60) was investigated by 205 Tl ϩ NMR. The main results indicate that monovalent cation induces conformational alterations at the active site of yeast pyruvate kinase (59,60). Moreover, kinetic and intrinsic fluorescence anisotropy measurements suggest that K ϩ induces the closure of the active site and the acquisition of the active conformation of rabbit muscle pyruvate kinase (20). In K ϩ -dependent enzymes, the high conservation of Glu 117 and its flanking residues Thr 113 /Lys 114 /Thr 120 , which are in a strategic structural location, indicates that these residues may be involved in the transduction of K ϩ binding to the movement of domain B. In the K ϩ -independent enzymes, Lys 117 , along with Leu 113 /Gln 114 /(Val-Leu-Ile) 120 may induce similar conformational changes but in the absence of K ϩ .
Why Do Enzymes Use Monovalent Cations?-There are a large number of enzymes for which activity is dependent on either Na ϩ or K ϩ . A salient feature of these enzymes is that extracellular enzymes are Na ϩ -dependent, whereas K ϩ -dependent enzymes are intracellular. Thus it may be asked, what is the evolutionary advantage of having Na ϩ or K ϩ as essential activators of these enzymes? Di Cera (4) has put forth that Na ϩ and K ϩ cannot function as regulators of enzyme activity because of their tightly controlled concentrations. Instead, he considers that monovalent cations provide a driving force for substrate binding and catalysis by lowering energy barriers (4). Thus monovalent cation-activated enzymes evolved to optimize their catalytic function by taking advantage of the large concentrations of Na ϩ and K ϩ in their surrounding media.
An alternative or additional explanation for the existence of the numerous proteins that require monovalent cations could involve the chemical rescue of an inactive enzyme. That is, we hypothesize that the biological activity of a mutated protein can be restored by the introduction of a chemical group that compensates for or overcomes the detrimental effect of a mutation. Indeed, it has been reported that imidazole compensates the loss of histidines that are essential for catalysis (61) and that guanidinium ions restore the substitution of arginine residues (62). Accordingly, it is possible that the binding of monovalent cations compensate for the loss of essential positive charged amino acids followed by additional mutations as the result of functional adaptations to the presence of the new ligand. In pyruvate kinase, the substitution of Lys for Glu 117 is compensated by the binding of K ϩ . Mutations in residues 113, 114, and 120 could be the evolutionary adaptations required to convert the enzyme into an efficient K ϩ -dependent catalyst. Concerning cation selectivity, cation-dependent pyruvate kinases emerged to select K ϩ over Na ϩ because of evolutionary pressure that led to the optimization of the interactions of the protein with the K ϩ that abounds in the intracellular milieu.
In sum, the phylogenetic analysis of pyruvate kinase showed that the K ϩ -independent activity is broadly present in nature; about half of the reported sequences possess Lys in position 117, whereas the rest have Glu in that position and are K ϩdependent. The K ϩ dependence feature clearly explains the dichotomic topology of the phylogenetic tree of pyruvate kinase family; the K ϩ -dependent and -independent enzymes belong to each of the branches of the tree. Because organisms that evolved at a later time have K ϩ -dependent pyruvate kinases, it may be suggested that they emerged from K ϩ -independent enzymes through single nucleotide substitutions. The high association found between the amino acid in position 117 and those in positions 113, 114, and 120, along with their central location in the structure of the enzyme, indicates that these residues might be involved in the activating effect of K ϩ or in the expression of the K ϩ -independent activity. Therefore, the primary structure of pyruvate kinases, related to the K ϩ requirement, provides a precise criterion for their classification.