Arginine Kinase from Nautilus pompilius, a Living Fossil

Arginine kinases were isolated from the cephalopods Nautilus pompilius, Octopus vulgaris, and Sepioteuthis lessoniana, and the cDNA-derived amino acid sequences have been determined. Although the origin and evolution of cephalopods have long been obscure, this work provides the first molecular evidence for the phylogenetic position of Cephalopoda in molluscan evolution. A crystal structure forLimulus arginine kinase showed that four amino acid residues (Ser63, Gly64, Val65, and Tyr68) are hydrogen-bonded with the substrate arginine. We introduced three independent mutations, Ser63→ Gly, Ser63 → Thr, and Tyr68 → Ser, inNautilus arginine kinase. One of the mutants had a considerably reduced substrate affinity, accompanied by a decreasedV max. In other mutants, the activity was lost almost completely. It is known that substantial conformational changes take place upon substrate binding in arginine kinase. We hypothesize that the hydrogen bond between Asp62 and Arg193stabilizes the closed, substrate-bound state. Site-directed mutagenesis studies strongly support this hypothesis. The mutant (Asp62→ Gly or Arg193 → Gly), which destabilizes the maintenance of the closed state and/or perhaps disrupts the unique topology of the catalytic pocket, showed only a very weak activity (0.6–1.5% to the wild-type).

Phosphagen (guanidino) kinases constitute a family of highly conserved enzymes, which catalyzes the reversible transfer of phosphate from phosphagen such as creatine phosphate to ADP. Creatine kinase (CK), 1 the most widely studied member of this family, plays a central role in both temporal and spatial ATP buffering in cells that display high and variable rates of energy turnover (1,2). The crystal structures have been determined for chicken mitochondrial CK (3) and muscle and brain CKs (4,5). The remaining major members of this family (arginine kinase (AK), glycocyamine kinase (GK), lombricine kinase (LK), taurocyamine kinase (TK)) have not been investigated as extensively as CK but likely play a similar physiological role (2,6,7).
Arginine kinase (AK) is widely distributed in the lower and higher invertebrate groups and is present in many lower chor-dates but absent in the vertebrates (6,8). Interestingly, AK evolved at least twice during evolution of phosphagen kinases: first, at an early stage of phosphagen kinase evolution (its descendants are molluscan and arthropod AKs) and second, from CK at a later time in metazoan evolution (9). Conventional wisdom would suggest that AK is the most primitive member of the phosphagen kinase family and that the other members, including CK, arose from tandem gene duplications and subsequent divergence (6,10,11).
From an earlier amino acid sequence alignment of CKs, GK, LK, and AKs, we proposed that a region displaying remarkable amino acid deletions (referred to as the guanidino specificity (GS) region) was a possible candidate for the guanidino recognition site (12). This GS region is overlapped partly by the so-called flexible loop in the crystal structures of chicken mitochondrial CK (3) and Limulus AK (13). There is a proportional relationship between the size of the deletion in the GS region and the mass of guanidine substrate used (12). Among the amino acid residues in the GS region, we noted previously that the seventh residue, Asp, is conserved in all AK sequences but not in other phosphagen kinases. This residue is not associated with the substrate binding (13), but we assumed that it has a special role in the recognition of the substrate arginine (14).
Mollusks constitute one of the most diverse animal phyla. Living representatives are classically divided into seven classes, of which major ones are Bivalvia, Gastropoda, and Cephalopoda. There are several hypotheses concerning molluscan phylogeny based on morphological features, but the results are still controversial. Sequence comparisons using 18S RNA, 12S rRNA, or mitochondrial DNA have been used to solve the complicated phylogeny of molluscan classes except Cephalopoda (15)(16)(17)(18)(19). However, molecular analyses have not resulted in a clear determination for the phylogeny. Thus, the origin and evolution of cephalopods have long been obscured by the poor fossil record and molecular data.
Nautilus pompilius is the only living representative of the large extinct group of nautiloids, ammonites and belemnites, and is sometimes referred to as a living fossil, like Latimeria and Lingula. In this report, first, we determined the cDNA-derived amino acid sequences of AKs from three cephalopods N. pompilius, Octopus vulgaris, and Sepioteuthis lessoniana (squid) to elucidate the evolution of molluscan AKs. Next, we examined the functional role of several amino acid residues on the GS region of Nautilus AK by site-directed mutagenesis. In particular, we focused on functional significance of Asp 62 , which is not associated with the substrate binding but is conserved in all AK sequences.
Throughout this report, the sequence numbering of Limulus AK (20) is used. The actual sequence number in Nautilus AK, used in the mutagenesis studies, can be obtained by subtracting 6 from the Limulus numbering.

MATERIALS AND METHODS
Isolation of Arginine Kinases-All procedures were carried out at 4 -8°C. The body wall muscle (10 g) of Nautilus, Sepioteuthis, or Octopus was homogenized with 50 ml of 10 mM Tris acetate buffer (pH 8.1) containing 0.1 mM dithiothreitol. The tissue extract was fractionated with 60 -80% saturated ammonium sulfate. The precipitate was dissolved in a minimum volume of the same buffer, and applied to a Sephadex G-75 column (3 ϫ 115 cm) equilibrated with the same buffer. The fractions possessing AK activity were pooled and applied to a DEAE-5PW column (7.5 ϫ 75 mm, Tosoh) equilibrated with 20 mM Tris acetate buffer (pH 8.1) containing 0.2 mM dithiothreitol. The column was washed with the same buffer and then eluted with a linear gradient of 0 -500 mM NaCl in 20 mM Tris acetate buffer. The purified AK was stored on ice or at Ϫ80°C until use.
Enzyme activity was assayed spectrophotometrically at 25°C (7). Protein concentration was estimated from the absorbance at 280 nm (the absorbance 0.77 at 280 nm in a 1-cm cuvette corresponds to 1 mg of protein/ml).
Partial Amino Acid Sequence Determination-The isolated proteins were pyridylethylated and digested with lysyl endopeptidase at an enzyme/substrate ratio of 1:100 in 0.1 M Tris-HCl (pH 8.0) at 37°C for 6 h. The digested products were isolated by reverse-phase chromatography. The column (Cosmosil 5C 18 -300, 4.6 ϫ 150 mm) was equilibrated with 0.1% trifluoroacetic acid and eluted with a linear gradient of 0 -90% acetonitrile in 0.1% trifluoroacetic acid over 90 min at a flow rate of 1 ml/min. Peptides for sequence analysis were purified further by rechromatography on the same column with a linear gradient of acetonitrile in 10 mM ammonium acetate. The amino acid sequences of the whole protein and peptides were determined using an automated protein sequencer (Applied Biosystems 476A).
cDNA Sequence Determination-mRNA was prepared from the radular muscle of Nautilus, Sepioteuthis, or Octopus with a QuickPrep Micro mRNA purification kit (Amersham Pharmacia Biotech). The single stranded cDNA was synthesized with reaction beads (Ready-To-Go You-Prime First-Strand Beads, Amersham Pharmacia Biotech) using the oligo-dT adapter as a primer. The 3Ј half of the cDNA was first amplified for 30 cycles, each consisting of 1 min at 94°C for denaturation, 1 min at 55°C for annealing, and 1.5 min at 72°C for primer extension by polymerase chain reaction (PCR). Extra Taq DNA polymerase (Takara) was used as the amplifying enzyme. The primers used are the oligo-dT adapter and a 512-fold "universal" redundant oligomer, 5Ј-GT(A/C/G/T)TGG(A/G)T(A/C/G/T)AA(T/C)GA(A/G)GA(A/G)GA(T/ C)CA, designed for amplification of phosphagen kinases (21). The amplified products were sequenced directly with dye terminator cycle sequencing FS ready reaction kit using a Model 373-18 DNA sequencer (Applied BioSystems). The universal redundant oligomer was used as a sequence primer.
The 5Ј half of the cDNA was amplified as follows. The single stranded cDNA was newly synthesized with the nonredundant primer (Nauti.R1 for Nautilus, Sepio.R1 for Sepioteuthis, and Octo.R1 for Octopus), and the poly(A) ϩ tail was added to the 3Ј end with terminal deoxynucleotidyl transferase. Then the 5Ј half of the cDNA was amplified by the method described above, using the oligo-dT adapter and the nonredundant primer (Nauti.R2 for Nautilus, Sepio.R2 for Sepioteuthis, and Octo.R2 for Octopus). The amplified products were sequenced directly.

Expression of Wild-type Nautilus AK and Site-directed Mutagenesis-
The open reading frame of 1053 base pairs (bp) of Nautilus AK was cloned into the BamHI/HindIII site of pMAL-c2 (pMAL/NautilusAKwild). The maltose binding protein-Nautilus AK fusion protein was expressed in Escherichia coli TB1 cells by induction with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at 27°C for 18 h. The soluble protein was extracted with B-PER reagent (Pierce), and the fusion protein was purified by affinity chromatography using amylose resin (New England BioLabs). The enzyme assay was done using the purified fusion protein.
The mutations (Asp 62 (GAC) 3 Gly (GGC), Asp 62 (GAC) 3 Glu (GAG), Ser 63 (AGT) 3 Gly (GGT), Tyr 68 (TAC) 3 Ser (TCC), and Arg 193 (AGA) 3 Gly (GGA)) were introduced in the template of pMAL/Nauti-lusAK-wild by PCR using mutation-primers (the sequences are shown in Table I). Pfu DNA polymerase was used as the amplifying enzyme. The PCR products were digested with DpnI, and the target 7000-bp DNA was recovered by EasyTrap (version 2, Takara). After blunting and kination, the DNA was self-ligated. The mutated protein was expressed as described above. The mutated cDNA insert was sequenced completely, and it was confirmed that there is no unexpected mutation.

RESULTS
Arginine kinases (AKs) from the body wall muscles of Nautilus, Octopus, and Sepioteuthis were purified to homogeneity using ammonium sulfate fractionation, Sephadex G-75 gel filtration and DEAE-5PW chromatography. SDS-PAGE, in the presence of a reducing agent, showed that the isolated AKs were highly purified, and the molecular masses were about 40 kDa. The AKs were confirmed to be monomers on the gel filtration column. No N-terminal amino acid of the AKs was detected by protein sequencing, indicating that N termini were blocked. The isolated enzyme was pyridylethylated and digested with lysyl endopeptidase. The peptides were purified by reverse-phase chromatography, and some peptides were sequenced (Fig. 1, A-C).
The enzymatic parameters, K m for arginine and V max , of the isolated AK enzymes from Nautilus, Octopus, and Sepioteuthis were obtained at 25°C for the forward reaction (Table II) The 3Ј halves of the cDNAs of Nautilus, Octopus, and Sepioteuthis AKs were amplified by PCR with the phosphagen consensus primer and oligo-dT adapter. Then two new oligomers were designed for amplification of the 5Ј halves. These 3Ј and 5Ј products were sequenced directly, and the nucleotide sequences of 1227 bp for Nautilus, 1226 bp for Octopus, and 1306 bp for Sepioteuthis were determined (Fig. 1, A-C). The open reading frame for Nautilus AK is 1053 nucleotides in length and encodes a protein with 350 amino acid residues; those for Octopus AK, 1044 nucleotides and 347 residues; those for Sepioteuthis AK, 1047 nucleotides and 348 residues. The validity of the cDNA-derived amino acid sequences was supported by chemical sequencing of internal lysyl endopeptidase peptides (81 amino acid residues determined for Nautilus, 36 residues for Octopus, 43 residues for Sepioteuthis). The molecular masses were calculated to be 39,276, 39,176, and 39,040 Da, respectively, consistent with the values estimated by SDS-PAGE. The nucleotide and amino acid sequences has been submitted to the DDBJ (accession numbers AB042330 for Nautilus, AB042331 for Octopus, and AB042332 for Sepioteuthis).
The recombinant maltose binding protein/Nautilus AK (wild type) gave enzymatic parameters K m (0.68 mM) and V max (7.64 mol of P i /min/mg of protein) similar to those of the native enzyme. Those for the mutated Nautilus AKs are listed in Table II.

Evolution of Cephalopod AKs and Phylogenetic Position of Nautilus
The cDNA-derived amino acid sequences of Nautilus, Octopus, and Sepioteuthis AKs were aligned with that of Limulus (horseshoe crab) AK (20) in Fig. 2, which is the only AK whose crystal structure is known (13). The sequence of Nautilus AK showed high percentage identity (69 -73%) with those of Octopus and Sepioteuthis AKs. The identity between Octopus and Sepioteuthis AKs is 83%. These cephalopod AK sequences still have a strong homology (53-55%) with that of the arthropod Limulus AK.
We have constructed a phylogenetic tree with the maximum likelihood method (22) for 36 cDNA-derived amino acid sequences of AK, CK, GK, and LK so far known, including four major molluscan classes Bivalvia, Polyplacophora, Gastropoda, and Cephalopoda. In Fig. 3, a part of the tree with all molluscan sequences is shown. The similar topology is obtained with the Neighbor-Joining method (23). The tree clearly shows that the sequences of the clams Pseudocardium, Solen, and Corbicula (Bivalvia) diverged first in molluscan AKs and was followed by the radiation of Polyplacophora (chiton Liolophura), Gastropoda (Nordotis, Battilus, Cellana, and Aplysia), and Cephalopoda (Nautilus, Octopus, and Sepioteuthis). Although the tree could not make clear the order of divergence of the three classes Polyplacophora, Gastropoda, and Cephalopoda, it suggests that the biological position of the chiton Liolophura, which is believed to be one of the primitive molluscs, should be reconsidered. Although the origin and evolution of cephalopods have long been debated, this work provides the first molecular evidence, as far as we know, for the phylogenetic position of Cephalopoda in molluscan evolution. Consistent with the morphological data, the branching pattern in Cephalopoda suggests that Nautilus (subclass Tetrabranchia, order Nautiloidea) is the most primitive, and Octopus (subclass Dibranchia, order Octapod) and Sepioteuthis (subclass Dibranchia, order Decapoda) share a common immediate ancestor.
We described before that the amino acid sequences of molluscan AKs could be used as a molecular clock to elucidate the divergence time of molluscs, because the enzymes appear to be derived from a single gene strain and have a moderate evolutionary rate (24). A maximum likelihood clock-like tree constructed from the 36 amino acid sequences allows us to estimate the divergence time of the interested molluscan species. If we assume the divergence between Mollusca and Arthropoda occurred about 550 million years before present (25), the radiation time of Polyplacophora, Gastropoda, and Cephalopoda is estimated to be 290 million years before present (data not shown).

Role of Amino Acid Residues of the GS Region in AK Function
Substrate-binding Residues-Amino acid sequences around the GS region of phosphagen kinases are shown in Fig. 4. This GS region is overlapped partly by the so-called flexible loop in the crystal structure of chicken mitochondrial CK, which has been speculated to move nearer to the active site to exclude water during catalysis (3). Clearly, there is a proportional relationship between the size of the deletion in GS region and the mass of guanidine substrate used. Namely, LK and AK, which use relatively large guanidine substrates, lombricine and arginine, have a five-amino acid deletion in this region, CK has one amino acid deletion, and GK has no deletion. It is possible that the introduction of an amino acid deletion as in the GS region in Fig. 4 shortens the flexible loop and allows the active site to accommodate a larger guanidine substrate.
Recently, a crystal structure for the transition state analog complex (TSAC) of the AK from the horseshoe crab Limulus has appeared (13). The structure showed that four amino acid residues (Ser 63 , Gly 64 , Val 65 , and Tyr 68 in Limulus AK sequence), located in the GS region (see Fig. 4), are hydrogenbonded with the amino and carboxylate groups of the substrate arginine. These residues are conserved in all three AK sequences from Nautilus, Octopus, and Sepioteuthis, suggesting that the substrate-binding mechanism of cephalopod AKs is the same as that of Limulus AK.
Among molluscan AKs, the four residues associated with substrate binding are conserved in most AKs. However, residue 64 or 65 is replaced by Ser, Lys, or Cys in the two-domain AKs from the clams Pseudocardium, Corbicula, and Solen and the sea anemone Anthopleura. Thus it appears that the two-domain AKs may have a unique substrate-binding system.
We introduced two independent mutations, Ser 63 3 Gly and Ser 63 3 Thr, in the 63rd position of Nautilus AK (Table II). At least one of the mutant proteins (Ser 63 3 Gly) had a considerably reduced substrate affinity (K m ϭ 3.45 mM) compared with the wild type (K m ϭ 0.68 mM), accompanied also by a decreased V max (less than 5% to the wild-type). In the case of Ser 63 3 Thr mutation, the AK activity was lost almost completely, probably due to a steric hindrance.
The Tyr 68 is conserved in all AKs, including two-domain clam AKs, except in the sea cucumber Stichopus AK, which evolved not from the AK gene but from the CK gene (9). Stichopus AK has a completely different substrate-binding system (see the sequence of Stichopus AK in Fig. 4), which will be described elsewhere. The aromatic side chain of Tyr 68 forms a hydrogen-bond with the amino group of the substrate arginine in Limulus AK, and thus the Tyr 68 appears to be the most important residue in substrate binding. Consistent with this, the enzyme activity of Nautilus Tyr 68 3 Ser mutation was lost  completely (Table II).
Role of Asp 62 -We noted previously that the Asp 62 in the GS region is conserved in most AK sequences, including Stichopus AK, which evolved from the CK gene, but not in other phosphagen kinases CK, GK, and LK (9, 14) (see Fig. 4). The TSAC crystal structure of Limulus AK indicates that Asp 62 is not associated with the substrate binding, and the authors have not referred to any functional role (13). On the other hand, we assumed previously that Asp 62 has a special role in the recognition of the substrate, arginine (14), and the first domain of Pseudocardium two-domain AK might not retain a complete enzyme activity, because the Asp 62 was replaced by a Gly residue (Fig. 4).
The TSAC structure of Limulus AK indicates that it consists of two domains; an N-terminal small ␣-helical domain (Ndomain) of residues 1-111 and a C-terminal domain (C-domain) possessing an 8-stranded antiparallel ␤-sheet flanked by 7 ␣-helices of residues 112-357 (13). One of the two substrates, ATP or ADP, is accommodated in the C-domain, and another substrate, arginine or arginine phosphate, contacts mainly with the N-domain. The catalytic center, where reversible transfer of the phosphate occurs, is located in the C-domain. We found that in the TSAC structure of Limulus AK, viewed by SwissPdbViewer (PDB Id: 1BG0), the side chain of Asp 62 in the N-domain is hydrogen-bonded with that of Arg 193 in the Cdomain (Fig. 5), and that Arg 193 is conserved in all AKs. Here we hypothesize that the hydrogen bond between Asp 62 and Arg 193 plays a crucial role in AK function, namely, that the two domains are maintained in a favorable topology by the hydrogen bond, where the two kinds of substrates are accessible enough for reaction.
If the 3-dimensional structure of Nautilus AK is homologous with that of Limulus AK (13), site-directed mutagenesis studies using Nautilus AK strongly support the above hypothesis. Both of the mutations Asp 62 3 Gly and Arg 193 3 Gly, which break the hydrogen bonding completely, markedly reduced enzyme activity (less than 1.5% of the wild-type). The same result was obtained from a mutagenesis study using the Asp 62 3 Gly mutant of Stichopus AK. 2 In addition, the results of mutation Asp 62 3 Glu provides strong evidence for the functional role of Asp 62 . This mutation, like the former two mutations, could be expected not to change the substrate affinity for arginine, but it could be expected to have ability to form a hydrogen bond with Arg 193 . Enzymatic parameters are consistent with this idea; the K m (0.67 mM) is the same as that (0.68 mM) of wild-type, and the V max is reduced largely but not so much as that of the Asp 62 3 Gly or Arg 193 3 Gly mutations, probably due to a partial formation of hydrogen bond between Glu 62 and Arg 193 .
Zhou et al. (13) have suggested that it is the precise orientation of the substrates in the catalytic pocket of AK that dominates the forces driving catalysis. Substantial conformational changes take place upon substrate binding in CK (26). The recent crystal structures of CK (3)(4)(5) and AK (13) provide very compelling insight into the nature of these structural changes. The extent domain and flexible loop movements during catalysis have recently been modeled by Forstner et al. (27) for CK and by Zhou et al. (28) for AK.
Our results and the above models show that, upon substrate binding, the flexible loop bearing Asp 62 moves nearer to the active center to exclude water during catalysis, and the hydrogen bond between Asp 62 and Arg 193 is formed to link the N-and C-domain of AK (Fig. 5). This bond stabilizes the closed, substrate-bound state. The mutant (Asp 62 3 Gly or Arg 193 3 Gly) destabilizes the maintenance of the closed state and/or perhaps disrupts the unique topology of the catalytic pocket; therefore, these mutants show only a very weak activity (0.6 -1.5% to the wild-type, see Table II). The mutant (Asp 62 3 Glu), while allowing hydrogen bonding between Glu 62 and Arg 193 , may also disturb the topology of the catalytic pocket.
Our results are highly consistent with prevailing views on the catalytic mechanism of phosphagen kinases. Maintenance of the precise three-dimensional structure of the active site is critical for optimum efficiency of enzyme function. Thus, we conclude that Asp 62 (or Arg 193 ) is one of the key residues in AK function. The GS region in the N-domain, which is overlapped partly by the so-called flexible loop in the crystal structure of chicken mitochondrial CK (21) and Limulus AK (12), is colored in green. The substrate arginine, nitrate, and ADP are shown by blue, green, and red balls, respectively.