Evolution of insect arylalkylamine N-acetyltransferases: Structural evidence from the yellow fever mosquito, Aedes aegypti

Abstract

Arylalkylamine N-acetyltransferase (aaNAT) catalyzes the transacetylation from acetyl-CoA to arylalkylamines. aaNATs are involved in sclerotization and neurotransmitter inactivation in insects. Phyletic distribution analysis confirms three clusters of aaNAT-like sequences in insects: typical insect aaNAT, polyamine NAT-like aaNAT, and mosquito unique putative aaNAT (paaNAT). Here we studied three proteins: aaNAT2, aaNAT5b, and paaNAT7, each from a different cluster. aaNAT2, a protein from the typical insect aaNAT cluster, uses histamine as a substrate as well as the previously identified arylalkylamines. aaNAT5b, a protein from polyamine NAT -like aaNAT cluster, uses hydrazine and histamine as substrates. The crystal structure of aaNAT2 was determined using single-wavelength anomalous dispersion methods, and that of native aaNAT2, aaNAT5b and paaNAT7 was detected using molecular replacement techniques. All three aaNAT structures have a common fold core of GCN5-related N-acetyltransferase superfamily proteins, along with a unique structural feature: helix/helices between β3 and β4 strands. Our data provide a start toward a more comprehensive understanding of the structure–function relationship and physiology of aaNATs from the mosquito Aedes aegypti and serve as a reference for studying the aaNAT family of proteins from other insect species. The structures of three different types of aaNATs may provide targets for designing insecticides for use in mosquito control.

Arylalkylamine N-acetyltransferases (aaNATs) catalyze the transacetylation of acetyl-CoA to arylalkylamines (1, 2). In mammals, aaNAT is involved primarily in the synthesis of N-acetylserotonin. The N-acetylation of serotonin (5-hydroxytryptamine) is a rate-limiting step for the synthesis of melatonin in the pineal gland. Physiologically, aaNAT regulates circadian rhythm in mammals (3, 4). In insects, aaNATs play multiple roles. In addition to melatonin synthesis, aaNAT enzymes are involved in aromatic neurotransmitter inactivation and cuticle sclerotization (5). aaNATs inactivate arylalkylamines, such as octopamine, dopamine, and serotonin (6, 7), because insects lack the monoamine oxidase (8) used by mammals to inactivate arylalkylamines (9). Studies dealing with an aaNAT from Bombyx mori clearly established its function in adult cuticle sclerotization through the production of N-acetyldopamine, a key cuticle protein crosslinking precursor (10, 11).

Unlike mammals, insects have evolved multiple aaNATs (2, 7, 12). In the yellow fever mosquito, Aedes aegypti, 13 putative aaNATs have been identified by a BLAST search using a positively verified Drosophila melanogaster aaNAT sequence; two of these aaNATs have been biochemically identified as aaNAT1 and aaNAT2 (2). The presence of multiple aaNATs in mosquitoes suggests the importance of arylalkylamine acetylation in these insect species.

The cuticular exoskeleton provides insects with protection against physical injury and infection, rigidity for muscle attachment, and mechanical support and flexibility for joint movement. During development and growth, insects periodically shed their old cuticle and produce a new one. New cuticle formation involves sclerotization, a process in which the cuticle becomes hardened and darkened. The cuticle is critical for insect survival, and thus is one of the primary targets for insect pest and disease vector control (13, 14). There are two sclerotization precursors, N-acetyldopamine and N-β-alanyldopamine. Both are synthesized from dopamine, in a pathway starting with tyrosine and including tyrosine hydroxylase and l-DOPA decarboxylase enzymes, among others downstream (15, 16). N-acetyldopamine is one of the aaNAT-catalyzed enzyme products. Therefore, aaNATs occupy a special position in insect cuticle sclerotization. Biochemical and structural characterization of insect aaNATs may provide insight into the in vivo functions of these enzymes and thus a means to manipulate enzyme activities to control the sclerotization process. Up to now, no single aaNAT crystal structure from any insect has ever been solved. Although the crystal structure of an aaNAT, also known as serotonin N-acetyltransferase (SNAT), from Ovis aries was solved previously (17), it shares only 2% and 7% sequence identity with A. aegypti aaNAT2 and aaNAT1, respectively (2).

Although 13 putative aaNATs have been identified by a BLAST search (2), most of them have not been biochemically identified as aaNAT. In this report, we provide crystal structures for aaNAT2, aaNAT5b, and putative aaNAT7 (paaNAT7) and screen more possible substrates for these three enzymes.

Results and Discussion

Evolution and Substrate Screening of Insect Putative aaNATs.

Mammal, avian, and anuran genomes possess a single copy of the aaNAT gene (18). Teleost fish have up to three paralogs (18), and amphioxus has seven aaNAT homolog genes (19). Insects do not have mammalian aaNAT homologs, based on sequence similarity searches of sequenced genomes and phylogenetic analysis (18, 19); however, insects do have multiple aaNAT enzymes that show very low sequence identity with mammalian aaNATs (2, 7, 12). The identified insect aaNAT sequences and the available genome sequences for a number of insect species make it possible to predict hypothetical aaNAT sequences in insects via a bioinformatics approach.

A BLAST search using the activity-verified D. melanogaster aaNAT sequence against three currently available mosquito species (A. aegypti, Anopheles gambiae, and Culex quinquefasciatus) and three nonmosquito insect species, identified 8 sequences from A. gambiae, 13 from A. aegypti, 15 from C. quinquefasciatus, 13 from D. melanogaster, 3 from B. mori, and 1 from Periplaneta americana showing recognizable similarity. To understand the phylogenetic relationship among the putative aaNATs in insects, we constructed a phylogenetic tree (Fig. 1). Phyletic distribution analysis confirmed three major clusters in insects, termed clusters 1, 2, and 3. The proteins in cluster 1 likely represent typical insect aaNATs, because all positively identified insect aaNATs are located in this cluster. aaNAT1 and aaNAT2 from A. aegypti are classified in this cluster. By screening more substrates, we identified two more substrates of aaNAT2, histamine, and bromoethylamine (Tables S1 and S2). No proteins have yet been identified as aaNATs in clusters 2 and 3. We found that one of the proteins in cluster 2, aaNAT5b, uses histamine, hydrazine, and hexamethylenediamine as substrates (Tables S1 and S2). Thus, cluster 2 proteins represent a different type of aaNATs found in insects, and we have named these polyamine NAT-like aaNATs based on their substrate profiles (20). Although both aaNAT2 and aaNAT5b use histamine as a substrate, their pH profiles are quite different (Fig. S1). The finding of histamine NAT activity is intriguing. In arthropods, histamine is an important neurotransmitter, particularly in photoreceptors. In Drosophila visual tissues, histamine is metabolized to carcinine (β-alanyl histamine) by Ebony (21, 22). In visual tissues of the horseshoe crab Limulus polyphemus, most histamine is metabolized to imidazole acetic acid and γ-glutamyl histamine (23). Although N-acetylation of histamine in arthropods has not been studied, N-acetyl histamine has been detected in locusts and Drosophila (24, 25), suggesting that arthropods might have a histamine NAT. The identification of histamine acetylation activity of aaNAT2 and aaNAT5b clearly demonstrate that A. aegypti has histamine NAT.

Fig. 1.

A phylogenetic tree of identified and putative aaNATs from mosquitoes and other insects. Sequences were aligned using ClustalW, and the dendrogram was generated using the neighbor-joining method. Numbers at nodes show bootstrap values based on 1,000 replicates. The National Center for Biotechnology Information gi number is provided for each protein sequence. The three proteins for which crystal structures were solved in this report are shown with a gray background. Aea, Aedes aegypti; Ag, Anopheles gambiae; Bm, Bombyx mori; Cq, Culex quinquefasciatus; Dm, Drosophila melanogaster; Pa, Periplaneta americana.

We also screened 61 possible substrates for paaNAT7, one of the proteins in cluster 3, but detected no transacetylation activity (Table S1). It must be noted that cluster 3 proteins might be unique to mosquitoes, however, given that the other insects studied have no putative aaNAT sequences in the cluster.

Overall Structures of aaNAT2, aaNAT5b, and paaNAT7.

An aaNAT2i (i.e., aaNAT2 crystal soaking with iodine ions) crystal was diffracted to 1.64 Å. Seven theoretical iodine sites were found in aaNAT2i with hkl2map, giving an overall figure of merit of 0.464 as calculated by PHENIX.Autosol. A further PHENIX run using AutoBuild resulted in an initial model with approximately 80% of the residues built. This model was subjected to iterative cycles of refinement and model rebuilding until the final model reached acceptable parameters in geometry and fitting. aaNAT2n (i.e., native crystal of aaNAT2), aaNAT5b, and paaNAT7 crystal structures were solved by molecular replacement, using the solved aaNAT2i structure as a search model. The data collection, refinement statistics, and statistics on Ramachandran plots as defined with PROCHECK (26) of four crystal structures are provided in Table S3.

The overall structures of aaNAT2, aaNAT5b, and aaNAT7 appear similar even though they have low sequence identity among one another (Table S4). Overall, the structures of aaNAT2 and aaNAT5b contain seven β-strands (β1–β7), and the structure of paaNAT7 contains eight β-strands (β0–β7). In all three proteins, the β-strands are mostly antiparallel to one another and form a semicylindrical β-sheet. The C-terminal ends of the neighboring strands β4 and β5 (the only two parallel β-strands in the sheet) are splayed apart from each other (Fig. 2), which splits the β-sheet into two groups of antiparallel β-strands composed of β0/β1–β4 and β5–β7. The tunnel thus created in the center of the β-sheet is necessary for the binding of the CoA cofactor and the substrate in other GCN5-related NAT (GNAT) proteins (27). All three proteins have four conserved sequence motifs, termed C, D, A, and B in N- to C-terminal order, and contain a common fold core that includes α-helices on both sides of six-stranded mixed β-sheets (β1–β6) of the GNAT superfamily (28) (Fig. 2). Similar to the SNAT structure, all three mosquito aaNATs have an additional β strand, β7, at the C-terminus. In addition, paaNAT7 has one more extra β strand, β0, at the N-terminus.

Fig. 2.

Overall structures of aaNAT2, aaNAT5b and paaNAT7 and 3D structural alignment of above three proteins and SNAT from O. aries. (A) Cartoon representation of aaNAT2. Motifs C, D, A, and B are labeled in purple, green, yellow, and red, respectively. Other regions of the structure are in gray. β-strands are numbered β1–β7. (B) Cartoon representation of aaNAT5b. The same scheme as used in A is used to color and label motifs, other parts of the structure, and β strands. (C) Cartoon representation of paaNAT7. The same scheme as used in A is used to color and label motifs and other parts of the structure. β strands are labeled β0–β7. (D) Superposition of the three mosquito aaNAT structures onto the structure of SNAT from O. aries. The same scheme as used in A is used to color and label motifs and β strands. Regions with major structural differences are colored gray for mosquito aaNATs and cyan for SNAT. (E) 3D structural alignment of aaNAT2, aaNAT5b, paaNAT7, and SNAT from O. aries (Oa-aaNAT). The second structures of SNAT are labeled under the sequences, and motifs C, D, A, and B are in purple, green, yellow, and red background boxes.

Structural Comparison of all Three Structures and Other Selected GNAT Superfamily Structures.

A Dali search (29) with the aaNAT2 structure revealed that aaNAT2 is most similar to histone acetyltransferase HPA2 of Corynebacterium glutamicum (PDB code 2qec; Z-score = 15.4; rmsd = 3.1) (DOI:10.2210/pdb2qec/pdb,) N-terminal acetylase ARD1 from Sulfolobus solfataricus P2 (PDB code 2x7b; Z-score = 15.4; rmsd = 2.6) (30), and ribosomal S18 N-α-protein acetyltransferase from Salmonella typhimurium LT2 (PDB code 2cnt; Z-score = 15.4; rmsd = 2.2) (31). Comparison of aaNAT2 and the SNAT from O. aries (PDB code 1kuy) (17) revealed a Z-score = 13.4 and rmsd = 2.9. A Dali search with the aaNAT5b structure found that aaNAT5b is most similar to histone acetyltransferase HPA2 of C. glutamicum (PDB code 2qec; Z-score = 15.3; rmsd = 2.8) (DOI:10.2210/pdb2qec/pdb,) PaiA from Bacillus subtilis (PDB code 1tiq; Z-score = 15.2; rmsd = 3.3) (32), and a phosphinothricin acetyltransferase from Agrobacterium tumefaciens (PDB code 1yr0; Z-score = 15.40; rmsd = 2.8) (DOI:10.2210/pdb1yr0/pdb). Comparison of aaNAT5b and the SNAT from O. aries showed a Z-score = 13.5 and rmsd = 2.9. A Dali search with paaNAT7 structure revealed that paaNAT7 is most similar to histone acetyltransferase HPA2 of C. glutamicum (PDB code 2qec; Z-score = 14.3; rmsd = 2.8) (DOI:10.2210/pdb2qec/pdb,) the acetyltransferase GNAT family (NP_688560.1) from Streptococcus agalactiae 2603 (PDB code 2pc1; Z-score = 14.2; rmsd = 3.1) (DOI:10.2210/pdb2pc1/pdb,) and a putative GNAT Ta0374 from Thermoplasma acidophilum (PDB code 3fix; Z-score = 14.1; rmsd = 2.5) (33). Comparison of paaNAT7 and the SNAT from O. aries revealed a Z-score = 12.0 and rmsd = 3.2.

Although GNAT superfamily members have four conserved sequence motifs (34) and contain a common fold core (28), they all display limited sequence similarity and retain few invariant residues (27), particularly in the regions where a more inconsistent stock of structures is built around the common core. This makes homology detection a challenge for GNAT superfamily members using sequence comparison alone (27, 35). Because the characteristic GNAT fold is well conserved (36), we used the 3D structural alignment to compare three mosquito aaNATs and other related GNAT superfamily proteins. GNATs catalyze a diverse arrays of substrates, including N-acetylation of lysine residues of proteins (histones and transcription factors) and small molecules such as polyamines, aminoglycosides, and arylalkylamines (37). Based on substrate specificities, a number of major GNAT superfamily proteins have been identified, including histone NAT (HAT), aminoglycoside NAT (AAC), aaNAT, and glucosamine 6-phosphate NAT (gpNAT). We chose representative structures from the foregoing enzymes for a structural comparison: a HAT from yeast (MYST; PDB code 1fy7) (38), a HAT from Tetrahymena (Gcn5/PCAF; PDB code 1puA) (39), human HAT (p300/CBP; PDB code 3biy) (40), a different HAT from yeast (Rtt109; PDB code 3qm0) (41), an AAC6 from Enterococcus faecium (PDB code 1n71) (42), an AAC6 from Salmonella enterica (PDB code 1s3z) (43), an AAC2 from Mycobacterium tuberculosis (PDB code 1m4g) (44), a gpNAT from Trypanosoma (PDB code 3i3g) (45), human gpNAT (PDB code 3cxs) (46), a gpNAT from Aspergillus fumigatus (PDB code 2vxk) (47), a gpNAT from yeast (PDB code 1i21) (48), and SNAT from O. aries (PDB code 1cjw) (49). Superposition of five GNAT structures and three mosquito aaNATs revealed that the common fold cores are nearly universally conserved (Fig. 3A) despite structure-based sequence alignments showing low pairwise sequence identity. Phyletic distribution analysis identified three mosquito aaNATs located in close proximity in the phylogenetic tree based on the 3D structural alignment and different from gpNAT, AAC, and mammalian aaNAT, SNAT. Interestingly, mosquito aaNATs are structurally close to HAT (Fig. 3B). Careful examination of the structural features of all three mosquito aaNATs revealed a unique structural feature, the helix/helices between β3 and β4 (Fig. 4). aaNAT2 and aaNAT5b have two helices, and paaNAT7 has one helix. Future investigations into the possible role of this structural feature might prove quite interesting.

Fig. 3.

Superposition of aaNAT2, aaNAT5b, aaNAT7, and other representative GNAT structures and the phylogenetic tree. (A) Superposition of eight GNAT structures (PDB codes 1PUA, 1I21, 1S3Z, 1CJW, and 1M4G, as well as the three mosquito putative aaNATs) reveals that GNAT common fold core structures are nearly universally conserved despite structure-based sequence alignments that reveal very low pairwise sequence identity. Motifs C, D, A, and B are well superposed, labeled as C, D, A, and B, respectively. (B) The phylogenetic tree based on the 3D structural alignment of 15 GNAT structures, including aaNAT2, aaNAT5b, and paaNAT7. After 3D alignment, the “PhylogeneticTree_TreeTop” function in STRAP was used to draw a cluster algorithm-type phylogenetic tree.

Fig. 4.

Unique structural features of aaNAT2, aaNAT5b, and paaNAT7. (A–C) Unique structural features between β3 and β4 strands of aaNAT2 (A), aaNAT5b (B), and paaNAT7 (C). (D) Superposition of 15 GNAT structures, including aaNAT2, aaNAT5b, and paaNAT7. Only the three mosquito aaNATs have this unique helix/helices structural feature; the other 12 GNATs, representing different HATs, different AACs, different GNATs, and SNAT shown in Fig. 3, do not have this structural feature.

Catalytic Funnels and Cofactor Binding.

GNAT superfamily proteins catalyze the transfer of an acetyl group to their substrates. The kinetic mechanism of most GNATs involves the ordered formation of a ternary complex. The reaction begins with acetyl-CoA binding, followed by binding of substrate, then direct transfer of the acetyl group from acetyl-CoA to the substrate, and finally product and subsequent CoA release (27). His120, His122, Leu124, and Tyr168 play important roles in the enzyme catalysis of sheep SNAT (36, 50), and by superimposing the three mosquito (putative) aaNATs onto the SNAT structure, we identified the equivalent residues in mosquito aaNATs. aaNAT5b shares three of the four residues, whereas aaNAT2 and paaNAT7 share only one of the four residues (Fig. 5). It has been suggested that Tyr168 in sheep SNAT functions as a general acid, serving to protonate the thiolate anion of CoA after decomposition of the tetrahedral intermediate formed by attack of the primary amine of serotonin on the acetyl-CoA thioester (36). However, aaNAT2, aaNAT5b, and paaNAT7 have Val180, Ala176, and Ala197, respectively, at the Tyr168 position in SNAT, suggesting that mosquito aaNATs likely use a different mechanism in catalysis. We evaluated general surface and electrostatic properties to assess the similarities and differences in the catalytic funnels of the four enzymes, aaNAT2, aaNAT5b, paaNAT7, and SNAT (Fig. S2). Interestingly, aaNAT2 had a wider catalytic funnel than aaNAT5b, which might explain why aaNAT5b catalyzes only smaller substrates, hydrazine and histamine. The surface of the paaNAT7 catalytic funnel was more positively charged, suggesting that paaNAT7 may use a more negatively charged substrate. This possibility should be considered in future substrate screening for aaNAT7. Using fixed molecular docking, we obtained binding positions of CoA molecules in aaNAT2, aaNAT5b, and paaNAT7 (Fig. S3). paaNAT7 was seen to have a similar CoA-binding funnel as aaNAT2 and aaNAT5b, demonstrating its characteristics as an NAT, although we have not yet identified its substrate.

Fig. 5.

Catalytic residues of aaNAT2, aaNAT5b, paaNAT7, and SNAT (PDB code 1cjw). Parts of motifs A and B are shown in ribbons, colored cyan for aaNAT2, yellow for aaNAT5b, blue for paaNAT7, and green for SNAT. The residues involved in the enzyme catalysis in SNAT and the equivalent residues in other three mosquito (putative) aaNATs are shown as sticks, with red oxygen and blue nitrogen atoms. Carbon atoms and residue labels are shown using the same color scheme as for the ribbon presentations.

Potential Biological Implications.

Mosquitoes transmit malaria, filarial worms, and numerous arboviruses (e.g., dengue fever, West Nile virus), which are major threats to human health throughout much of the world. Among these, malaria is considered the most prevalent life-threatening disease, with annual estimates of new cases ranging from 300 million to 660 million (51). aaNATs play critical roles in mosquito survival and development, including biogenic amine and polyamine detoxification, melatonin synthesis, and cuticle formation. The identification of mosquito aaNATs provides a starting point for more comprehensive investigations of the biochemistry and physiology of aaNATs in mosquitoes and serves as a reference for studying the aaNAT family of proteins in other insect species. The crystal structures of the three different types of mosquito aaNATs that we have identified should aid the design of new types of insecticide compounds aimed at affecting mosquito aaNAT activity or stability. Mosquito aaNATs differ greatly from mammalian aaNATs in both evolution and molecular structure, have a unique structural feature (the helix/helices between β3 and β4 strands) and may have a different catalytic mechanism. Thus, inhibitors of mosquito aaNATs might not affect mammalian aaNATs. Potent mosquito aaNAT inhibitors might provide another pathway to effective mosquito control. However, more work is needed to define the catalytic mechanisms of aaNAT2, aaNAT5b, and paaNAT7 and identify substrates of paaNAT7.

Materials and Methods

Expression and Purification of A. aegypti, aaNAT-2, aaNAT-5b, and paaNAT-7.

Amplification of full-length cDNA for three proteins—aaNAT2 (GenBank accession no. XP_001663122), aaNAT5b (GenBank accession no. XP_001649916), and paaNAT7 (GenBank accession no. XP_001663019)—was achieved using the forward and reverse primers corresponding to 5′ and 3′ regions of the coding sequences (2). Their amplified cDNA sequences were cloned into an Impact-CN plasmid (New England Biolabs) for expression of recombinant proteins. All recombinant proteins had three extra amino acid residues added, AlaGlyHis in the N-termini, because of the extra nucleotide sequence in the Impact-CN plasmid. The recombinant proteins were then produced in Escherichia coli cells and purified as described previously (2). The purity of the recombinant proteins was assessed by the presence of a single band with anticipated molecular weight on an SDS/PAGE gel. Protein concentrations were determined using a Bio-Rad protein assay kit with BSA as a standard.

Phylogenetic Analysis of aaNATs and Putative aaNATs in Insect Species.

Insect aaNAT homologs from different insect species were identified through a search of the nonredundant database of protein sequences using the BLAST and PSI-BLAST search features (52). Putative aaNAT sequences from A. aegypti, A. gambiae, C. quinquefasciatus, D. melanogaster, B. mori, and P. americana were selected, and a multiple sequence alignment of the respective protein sequences was constructed using the MEGA4-ClustalW alignment program (53). A maximum likelihood phylogenetic tree was constructed using the neighbor-joining method and bootstrap analysis (53).

Crystallization.

The crystals were grown by hanging-drop vapor diffusion methods with a 150-μL volume of reservoir solution at and a 2-μL drop volume, containing 1 μL of protein sample and 1 μL of reservoir solution. The optimized crystallization buffer for aaNAT2 consisted of 0.17 M ammonium acetate, 25.5% polyethylene glycol 4000, and 17% glycerol, and 0.1 M sodium citrate (pH 5.6). For aaNAT5b, the buffer consisted of 0.5 M sodium chloride, 25.5% polyethylene glycol 8000, 15% glycerol, and 0.1 M sodium acetate (pH 4.6). The buffer for paaNAT7 comprised 2 M ammonium sulfate, 5% polyethylene glycol 400, and 0.1 M Hepes sodium (pH 7.5). aaNAT2 and aaNAT5b crystals were mounted without the addition of extra cryoprotectant agents. The iodine derivative of paaNAT7 was cryogenized for 2 min using 22% ethylene glycol in the crystallization buffer as a cryoprotectant solution, which also contained 0.5 M sodium iodine. The iodine derivative of aaNAT2 (aaNAT2i) was prepared by soaking a single crystal in the crystallization buffer, which contained 0.5 M sodium iodine, for 2 min.

Data Collection and Processing.

Crystal diffraction data were collected at the Brookhaven National Synchrotron Light Source beam line X29A (λ =1.075 or 1.540 Å). Data were collected using an ADSC Q315 CCD detector. All data were indexed and integrated using HKL-2000 software. Scaling and merging of diffraction data were performed using the SCALEPACK program (54). The parameters of the crystals and data collection are listed in Table S3.

Structure Determination.

The structure of aaNAT2 was determined by the SAD phasing technique with iodine anomalous signals. Heavy atom sites were located using hkl2map (55), and the initial SAD phases were calculated by PHENIX.Autosol (56). Initial models were built using PHENIX.AutoBuild (56). The model was refined using SAD refinement with optimization of the iodide ion occupancy in Refmac 5.2 (57). Final models were produced after numerous iterative rounds of manual rebuilding in Coot (58) and refinement in Refmac 5.2 (57). The structures of native aaNAT2, aaNAT5b (21% sequence identical with aaNAT2), and paaNAT7 (19% sequence identical with aaNAT2) were determined by the molecular replacement method using the solved aaNAT2 structure. The Molrep (59) and Phaser (60) programs were used to calculate both the cross-rotation and translation of the model. The initial model was subjected to iterative cycles of crystallographic refinement with Refmac 5.2 (57) and graphic sessions for model building using Coot (58). Solvent molecules were automatically added and refined with ARP/warp (61) together with Refmac 5.2.

Structure Analysis.

Superposition of structures was done using Lsqkab (62) in the CCP4 suite. Figures were generated using Pymol (63). Protein surface properties were analyzed using Pymol (63). 3D structural alignment and creation of the phylogenetic tree based on this 3D alignment were done using STRAP (64).