• No results found

Dissecting the Unique Nucleotide Specificity of Mimivirus Nucleoside Diphosphate Kinase

N/A
N/A
Protected

Academic year: 2019

Share "Dissecting the Unique Nucleotide Specificity of Mimivirus Nucleoside Diphosphate Kinase"

Copied!
9
0
0

Loading.... (view fulltext now)

Full text

(1)

0022-538X/09/$08.00⫹0 doi:10.1128/JVI.00511-09

Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Dissecting the Unique Nucleotide Specificity of Mimivirus Nucleoside

Diphosphate Kinase

Sandra Jeudy,‡ Audrey Lartigue,‡ Jean-Michel Claverie, and Chantal Abergel*

Structural and Genomic Information Laboratory, CNRS-UPR2589, IFR88, 163 avenue de Luminy, Case 934, 13288 Marseille, Cedex 9, France

Received 3 March 2009/Accepted 29 April 2009

The analysis of theAcanthamoeba polyphagamimivirus genome revealed the first virus-encoded nucleoside diphosphate kinase (NDK), an enzyme that is central to the synthesis of RNA and DNA, ubiquitous in cellular organisms, and well conserved among the three domains of life. In contrast with the broad specificity of cellular NDKs for all types of ribo- and deoxyribonucleotides, the mimivirus enzyme exhibits a strongly preferential affinity for deoxypyrimidines. In order to elucidate the molecular basis of this unique substrate specificity, we determined the three-dimensional (3D) structure of theAcanthamoeba polyphagamimivirus NDK alone and in complex with various nucleotides. As predicted from a sequence comparison with cellular NDKs, the 3D structure of the mimivirus enzyme exhibits a shorter Kpn loop, previously recognized as a main feature of the NDK active site. The structure of the viral enzyme in complex with various nucleotides also pinpointed two residue changes, both located near the active site and specific to the viral NDK, which could explain its stronger affinity for deoxynucleotides and pyrimidine nucleotides. The role of these residues was explored by building a set of viral NDK variants, assaying their enzymatic activities, and determining their 3D structures in complex with various nucleotides. A total of 26 crystallographic structures were determined at resolutions ranging from 2.8 Å to 1.5 Å. Our results suggest that the mimivirus enzyme progressively evolved from an ancestral NDK under the constraints of optimizing its efficiency for the replication of an AT-rich (73%) viral genome in a thymidine-limited host environment.

Mimivirus, a DNA virus infectingAcanthamoeba, is the

larg-est and most complex virus isolated to date (8, 37). It is the first

representative and prototype member of theMimiviridae, the

latest addition to the large nucleocytoplasmic DNA viruses, including the poxviruses, the phycodnaviruses, (infecting al-gae), the iridoviruses (infecting invertebrates and fishes), and asfarvirus (the agent of a swine fever in Africa) (18). The mimivirus’s record genome size (1.2 Mb) and gene content (911 encoded proteins), as well as the presence of genes pre-viously thought to be specific to cellular organisms (such as aminoacyl-tRNA synthetases [3]), revived the debate about the evolutionary origin of DNA viruses and their putative role in the emergence of the eukaryote nucleus (reviewed in reference 7) or in the advent of DNA genomes (13).

In this peculiar context, we found the discovery of the first virus-encoded nucleoside diphosphate kinase (NDK) within the mimivirus genome of great interest and warranting a de-tailed study of the structural and biochemical properties of this unique viral enzyme. Ubiquitous in cellular organisms, NDKs

are responsible for the last step of 2⬘-deoxynucleoside

triphos-phate (dNTP) pathways and as such play an essential role in the replication of DNA by providing the basic precursors for its synthesis. Acting indiscriminately on ribonucleotides and

de-oxyribonucleotides, the cellular NDKs are also responsible for supplying energy to various essential synthetic pathways, pro-ducing NTPs for RNA synthesis, CTP for lipid synthesis, UTP for polysaccharide synthesis, and GTP for protein synthesis elongation, signal transduction, and microtubules polymeriza-tion. Besides their direct role in the above metabolic pathways, cellular NDKs have been involved in the regulation of cell growth and differentiation in vertebrates (22).

Cellular NDKs are small proteins of about 150 amino acids, the sequences of which are highly conserved among the three

domains of life (⬎40% identity). They are most often

hexam-eric enzymes, with a few occurrences of tetramhexam-eric and dimhexam-eric NDK structures in bacteria (19, 25, 26, 31, 38). They all cata-lyze the transfer of a phosphate group from an NTP onto a

nucleotide diphosphate (NDP) through an Mg2⫹-dependent

reaction. In vivo, the phosphate donor is usually the nonlimit-ing ATP nucleotide.

In agreement with their implication in various metabolic pathways, cellular NDKs exhibit little substrate specificity and are equally able to act on purine and pyrimidine nucleotides, in

their 2⬘OH and deoxyribonucleotide forms. In clear contrast,

our characterization of the mimivirus NDK revealed its en-hanced affinity for deoxypyrimidine nucleotides (20). This marked difference between the viral and cellular NDKs offered a good opportunity to explore the sequence and structure features governing substrate specificity. For instance, cellular NDKs exhibit a conserved loop, the Kpn loop, involved both in substrate binding and in oligomerization of the enzyme (19). Interestingly, a sequence comparison predicted this loop to be

shorter in the Acanthamoeba polyphaga mimivirus NDK

(NDKapm) sequence. However, many other single-residue

changes could also be involved in modifying the enzyme

prop-* Corresponding author. Mailing address: Structural and Genomic Information Laboratory, CNRS-UPR2589, IFR88, 163 avenue de Luminy, Case 934, 13288 Marseille Cedex 9, France. Phone: (33) 491 825422. Fax: (33) 491 825421. E-mail: Chantal.Abergel@igs .cnrs-mrs.fr.

† Supplemental material for this article may be found at http://jvi .asm.org/.

‡ These authors contributed equally to this work.

Published ahead of print on 13 May 2009.

7142

on November 8, 2019 by guest

http://jvi.asm.org/

(2)

erties. To explore these issues, we performed a detailed

struc-ture-function analysis of the NDKapmprotein in a variety of

mutated forms and substrate-enzyme complexes. Despite its markedly different sequence, the three-dimensional structure of the mimivirus NDK was found to be very similar to that of cellular enzymes. Its peculiar substrate specificity is not attrib-utable to a single sequence feature but rather appears to result from the conjunction of several factors, suggesting the progres-sive optimization of an ancestral enzyme for the replication of an AT-rich (73%) genome in a thymidine-limited host envi-ronment.

MATERIALS AND METHODS

NDKapmwas cloned and expressed as previously described (20, 21) (see the

supplemental material). The N62L and R107G mutations, as well as the insertion of the Kpn sequence (⫹Kpn) fromDictyostelium discoideum, were produced using a QuikChange II site-directed mutagenesis kit (Stratagene). Oligonucleo-tide sequences are presented in the supplemental material.

Protein purification and crystallization.The recombinant mimivirus NDK protein was purified and crystallized as described earlier (2, 21). NDK-nucleotide stoichiometric complexes were produced by incubating the freshly purified re-combinant NDKapmprotein in the presence of 5 mM Mg2⫹with 5 mM

nucle-otides (dTTP, dGTP, GTP, dUTP, UTP, and CTP). Each complex was purified on a HiTrap chelating column charged with Ni2⫹and recovered in the specific elution fractions. After 2 weeks, 90% of the protein was spontaneously trans-formed into a truncated form. N-terminal sequencing revealed that 15 residues of the 21-residue tag had been removed. Complexes were concentrated in 10 mM CHES buffer [2-(cyclohexylamino)ethanesulfonate, pH 9.0] and repurified after 2 weeks to recover the tag-free complexes. NDKapm-dTTP was concentrated to

26 mg䡠ml⫺1and complexes of NDK

apmwith dGTP, GTP, CTP, UTP, or dUTP

to 11 mg䡠ml⫺1

.

The NDK mutants were concentrated in 10 mM CHES buffer (pH 9.0) and incubated with 1 mM MgCl2and 1 mM NTP overnight prior to crystallization

without further purification. Due to the ATPase activity of NDK enzymes, the NTP nucleotides are spontaneously transformed into NDP (25), as evidenced by the complex structures. Consequently, all NDK complexes contain NDP even if NTP was used for crystallization.

Crystallization of the NDKapmand the NDK mutant nucleotide complexes was

performed by vapor diffusion experiments. Conditions were initially identified through a previously described screening protocol (2) (see the supplemental material). For the nucleotide complexes of NDKapm(26 mg䡠ml⫺1), NDKN62L

(11 mg䡠ml⫺1), NDK

R107G(11.7 mg䡠ml⫺1), NDKR107G-N62L(11 mg䡠ml⫺1),

NDK⫹Kpn-N62L(11.8 mg䡠ml⫺1), and NDK⫹Kpn-N62L-R107G(11.5 mg䡠ml⫺1),

crystals were obtained using MPD (2-methyl-2,4-pentane-d12-diol)

concentra-tions from 40 to 45% (vol/vol) in 0.1 M HEPES (pH 7.5) or 0.1 M MOPS (pH 7.4). These crystals belong to the C2221space group, with six monomers per

asymmetric unit. Crystals of the NDK⫹Kpn(13 mg䡠ml⫺1), NDK⫹Kpn-N62L(13.3

mg䡠ml⫺1), NDK

⫹Kpn-R107G(10.7 mg䡠ml⫺1), and NDK⫹Kpn-N62L-R107G(11.5

mg䡠ml⫺1

) mutants were obtained using 0.8 to 1.2 M sodium citrate, 0.1 M Tris buffer (pH 7.6). They belong to the P63space group, with two monomers per

asymmetric unit.

Data collection and processing.Crystals produced in sodium citrate were soaked for 1 min in a reservoir solution containing 10% glycerol as a cryopro-tectant, collected in Hampton Research loops, flash frozen to 105 K in a cold nitrogen gas stream, and subjected to X-ray diffraction. MOSFLM and SCALA from the CCP4 package (9) were used for the processing, scaling, and data reduction. Data sets were collected at the European Synchrotron Radiation facility (see Tables S1 and S2 in the supplemental material) or in-house using a Xcalibur PX-Ultra diffraction system (Oxford Diffraction). A previously de-scribed evaporation protocol (1) was used to improve the diffraction of some crystals.

Structure determination. (i) Native form.The structure of the NDKapmnative

form was solved as previously described (21) (see the supplemental material). Refinement of the molecular replacement solution structures was performed using crystallography and nuclear magnetic resonance (5), including simulated annealing and positional and B-factor refinement. Addition of solvent molecules and a final round of refinement resulted in a final model made of residues⫺4 to 129 and 0 to 129 for the two molecules and a finalRworkof 20.2% andRfreeof

25.2%. (see Table S1 in the supplemental material). The difference density map

exhibited two peaks near the substrate binding site corresponding to two sulfate ions.

(ii) NDKapmin complex with nucleotides.A crystal of the NDKapm-dTDP

complex was used to compute a self rotation function and determine the entities constituting the crystal using AMoRe software (33). A trimer of the apo form was successfully used to search for a molecular replacement solution using AMoRe software. Refinement was performed using crystallography and nuclear magnetic resonance (5), which resulted in a final model made of six molecules in the asymmetric unit, including residues⫺1 to 131, 0 to 131, and 0 to 131 for the first trimer and residues 1 to 130, 0 to 131, and 0 to 129 for the second one, with a final

Rworkof 19.9% andRfreeof 23% (see Table S1 in the supplemental material).

Other complexes were solved using this structure as a reference and the differ-ence density maps to build the nucleotide and position the Mg2⫹ion (see Table S1 in the supplemental material).

(iii) NDKapmmutants in complex with nucleotides.The NDKN62L, NDKR107G, and

NDKN62L-R107Gcomplexes were crystallized like the native form in the same

space group and with comparable cell dimensions. The native hexameric struc-ture was used as the initial template, with a first round of rigid body refinement using Refmac (32). The interactive COOT software (11) was then used to build the different ligands and to manually refine each of the monomer structures using the difference density maps. We performed 10 cycles of refinement using the integrated Refmac program (32) after each round of manual building.

For the NDK⫹Kpn-N62L and NDK⫹Kpn-N62L-R107G nucleotide complexes,

other conditions were identified (see above) for which crystals exhibited better diffraction and contained only two molecules per asymmetric unit. Molecular replacement was used to determine the corresponding structures, and COOT software (11) was used for manual building and refinement.

To verify that the nucleotide orientation into the enzyme active site was not influenced by the crystallization conditions, we determined and compared the NDK⫹Kpn-N62L-dTDP structures in both space groups.

Enzymatic activity measurements.TheSaccharomyces cerevisiaeNDK enzyme (NDKsc; Sigma-Aldrich no. N0379) was used as a control. All NDKapmmutants

were compared to the native NDKapmrecombinant protein using a coupled assay

on microtiter plates by measuring the absorbance at 340 nm (20). Activity was measured through NADH oxidation using a␮quant spectrophotometer (Bio-tech Instrument) and 96-well UV plates (PS-Microplate; Greiner Bio-one). The reaction was performed in a 100-␮l volume containing 5 mM MgCl2, 1.5 mM

phosphoenolpyruvate, 0.2 mM NADH, 1 mM ATP, and 1 active unit of pyruvate kinase/lactate dehydrogenase in 100 mM Tris-HCl buffer (pH 7.5), with the various (d)NDPs at various concentrations and with optimized quantities of NDK enzymes.

For the native enzyme as well as for all mutants, the protein concentrations were determined so as to be in the linear part of the curve of the enzyme activity obtained for 75␮M dGDP and increasing concentration of enzymes (between 0.1 and 26␮g䡠ml⫺1), as follows: NDK

apm, 1.6␮g/ml; NDK-N62L

and NDK-N62L-R107G, 3.25 ␮g/ml; NDK⫹Kpn-N62L-R107G, NDK⫹Kpn,

NDKKpn-N62L, and NDK-R107G, 6.5 ␮g/ml; and NDKsc, 0.64␮g/ml. This

assay was used for all (d)NDPs as phosphate acceptors except for ADP (for which ATP is the phosphate donor). Initial velocities were determined at least twice for each substrate concentration, with duplicate measurements varying by less than 5%.

Protein sequence accession numbers.The PDB accession numbers for the structures determined in this study are 2B8P, 2B8Q, 3B6B, 3EE3, 3EIC, 3EJM, 3ELH, 3ETM, 3EM1, 3EVM, 3EVO, 3EVW, 3EMT, 3ENA, 3DKD, 3DDI, 3FC9, 3FCV, 3FCW, 3FBB, 3FBC, 3FBE, 3FBF, 3G2X, 3GP9, and 3GPA.

RESULTS AND DISCUSSION

Structures of the NDKapm enzyme alone and in complex with various substrates.As expected from negative stain elec-tron microscopy images (Irina Gutsche [UMR 5233 UJF-EMBL-CNRS], personal communication) and gel filtration studies (20), the mimivirus NDK is hexameric (dimer of trim-ers) (see Fig. S2 in the supplemental material) in the crystal with a root mean square deviation (RMSD) of 2.5 Å based on

C␣ superimposition with the D. discoideum NDK hexamer

structure (NDKdd; PDB no. 1NDC) (6). The trimer formation

involves interactions between the␣3 and ␣k helices and

be-tween the␣0and␣1helices. The dimer formation involves the

on November 8, 2019 by guest

http://jvi.asm.org/

(3)

␣1helix and an antiparallel arrangement of the␤2strands of

two NDK monomers (Fig. 1 and 2).

The surface comparison of all available NDK structures showed conserved hydrophobic patches at the two trimer in-terfaces, while the hexamer surfaces were found to differ from one structure to another. NDK enzymes, besides their main enzymatic functions, have been reported to be involved in signaling cascades through the activation of guanine nucleotide-binding proteins (10, 12, 27) and in protein-DNA-RNA inter-actions (28). Endonuclease, exonuclease, DNase, and RNase activities have also been reported (15, 23, 36, 39). The varia-tions in the hexameric surfaces may be correlated to the dif-ferent localizations or cellular functions of these NDK pro-teins. A common feature of all NDK structures is a central channel formed through a salt bridge involving two conserved residues of the same monomer (R24 and D101) and two non-conserved residues between two monomers. In the mimivirus NDK structure, this interaction corresponds to a salt bridge between E20 of one monomer and K27 of the second mono-mer. There is also a ring of positively charged residues around

the central channel and negatively charged patches on the hexamer external edges. While the buried surface area is

roughly the same between two monomers (1,000 Å2) and

be-tween the two trimers (2,500 Å2) in all NDK structures, the

buried surface area is much smaller in the NDKapmtrimer than

in other NDK structures (2,140 Å2instead of 3,130 Å2forD.

discoideum) on account of a shorter Kpn loop. In previous studies, this loop was identified as a central element for NDK enzymatic activities and for the association of monomers in a hexameric structure (19).

We analyzed the structure of the NDKapmin complex with

various nucleotides, in order to understand the molecular basis of its enhanced affinity for deoxypyrimidine nucleotides (see Table S1 in the supplemental material). As for the other NDK structures, the nucleotide fixation at the substrate binding site does not induce major conformational changes. Despite the use of different crystallization conditions, the maximum mea-sured RMSD between the free and the bound enzyme is 0.5 Å,

based on C␣superimposition of the hexamers. The

compari-son of these complexes with other NDK complexes

demon-FIG. 1. Detail of the structural alignment of NDKapmwith other NDK structures. The complete alignment is presented in Fig. S1 in the

supplemental material. Red boxes indicate strictly conserved residues; residues in red type have conserved properties. The ESPript program (14) was used to indicate secondary structure elements in NDKddand NDKph(top) and in NDKapmand NDKpa(bottom). Black arrows indicate the

point mutations in NDKapm.

on November 8, 2019 by guest

http://jvi.asm.org/

[image:3.585.82.501.66.436.2]
(4)

strates that the residues known to be important for the enzyme

activity are conserved in the NDKapmstructure. The

phospho-ryl␥is transferred to the N␦of a conserved histidine residue,

H112 in NDKapm(H122 in NDKdd), in the presence of the

donor nucleotide and Mg2⫹ions, and the Nεof the imidazole

ring is stabilized by the conserved S114 and E123 residues

(S124 and E133 in NDKdd), both of which maintain the

histi-dine residue in the proper orientation. In all NDK structures, there is a hydrophobic residue that adopts a disallowed con-formation to keep the substrate accessible in the active site,

and this residue corresponds to L110 in NDKapm (I120

in NDKdd). The conserved phenylalanine residue (F64 in

NDKdd) stabilizing the base moiety through aromatic stacking

is replaced by a superimposable tyrosine residue in the viral enzyme (Y58). The phosphate environment is the same in all NDK structures with protein interactions between the NDK protein and the nucleotide phosphate groups involving the conserved residues K9, Y50, R86, R99, and N109 (K16, Y56,

R92, R109, and N119 in NDKdd). The ribose 3⬘ OH makes

hydrogen bonds with K9 and N109. Finally, the comparison of

the NDKapm/dTDP structure highlighted three main

differ-ences with other NDK binding sites which could explain the peculiar affinity of the mimivirus NDK for deoxypyrimidine nucleotides (Fig. 2).

The most obvious sequence difference, the shorter Kpn loop in the viral protein, results in a wider substrate binding site compared to other structures. The shortening of the Kpn loop

is precisely located in the LKpnI region (Fig. 1). The NDK

nucleotide binding site can be described as a tong-like

struc-ture made of the␣2helix on one side and part of the Kpn loop

(LKpn2) on the other side. Interestingly, there are two ordered

water molecules at the position of the missing LKpnIsegment

(Fig. 2) which appear to stabilize the NDKapm active site

through hydrogen bonds between the shorter LKpnIand LKpn2

parts of the Kpn loop. Another difference between the mim-ivirus NDK and other NDK structures is the replacement of an

otherwise conserved glycine residue in LKpn2by an exposed

arginine residue (R107) on the edge of the active site. This arginine residue forms hydrogen bonds with the dTDP by a

direct interaction between the R107 Nεand the base moiety

(Fig. 2; also see Fig. S3A in the supplemental material). Finally, deep in the nucleotide binding site, an otherwise

strictly conserved leucine in the ␣2 helix is replaced by an

asparagine (N62) in the viral NDK. The side chain (O␦1, N␦2)

of this asparagine residue forms hydrogen bonds with the

ri-bose moiety of the nucleotide, as well as with the oxygen (O2)

of the pyrimidine base (see Fig. S3B in the supplemental ma-terial). To explore the respective contributions of these changes on the substrate specificity of the mimivirus NDK, we generated a set of mutants restoring a standard sequence. The tested mutations included (i) the reinsertion of a standard Kpn

loop (⫹Kpn), (ii) two single mutations (R107G and N62L),

and (iii) the combination of all mutations, including the triple mutant. We then determined the crystal structures of the NDK mutants in complex with various nucleotides and assayed their activities on deoxyribo- and ribonucleotides.

Structure comparison of the NDK mutants with the NDKapm nucleotide complexes. The comparison of mimivirus NDK structure with the structures of the mutant NDKs did not reveal any overall differences. The structures remained

hexam-eric, and the C␣-based RMSDs between the seven mutants and

the native enzyme in complex with nucleotides and the apo enzyme structures are of the same magnitude. When visible,

the Mg2⫹ion associated with the nucleotide remained at the

same position in all NDK structures (see Fig. S3 in the sup-plemental material), forming a bridge between the two oxygens

of the␣- and␤-phosphates and stabilized by ordered water

molecules.

The most striking differences were observed at the nucleo-tide binding site, with variations in the nucleonucleo-tide position depending on its nature and on the enzyme crystallized.

(i) Complexes with deoxypyrimidine nucleotides.In the de-oxypyrimidine complexes, the nucleotide base penetrates

deeper into the active site of the native NDKapmenzyme than

in theD. discoideumNDK-dTDP complex structure (1NDC)

(6). In the N62L and R107G mutants, the nucleotide is less deeply buried than in the native enzyme, but the overall com-plex structures are very agitated (see Table S2 in the supple-mental material). In the N62L-R107G NDK mutant in com-plex with deoxypyrimidine, the nucleotide is barely visible. This is compatible with a decrease in affinity for the dTDP in the double mutant. For Kpn, Kpn-N62L, and the triple mutants, the positioning of the nucleotide is close to that observed in the

[image:4.585.55.271.68.334.2]

D. discoideum NDK complex structure (see Fig. S3 in the supplemental material). We verified that the complex struc-tures did not depend on crystallization conditions by compar-ing the Kpn-N62L mutant structure in complex with dTDP in

FIG. 2. Unique features of the NDKapmstructure. Cartoon

repre-sentation of the NDKapmstructure (yellow), with the Kpn loop

high-lighted in red and the LKpnI-␣K segment absent from the NDKapm

structure in green. Residues conserved in other NDK structures are presented in green, and the corresponding residues in NDKapmare in

blue. The dTDP/Mg2⫹nucleotide is in cyan, and two water molecules

at the LKpn1location are in gray.

on November 8, 2019 by guest

http://jvi.asm.org/

(5)

the two crystal forms (C2221and P63). Both protein and ligand

structures are superimposable (0.3-Å RMSD main chain;

⬍0.2-Å ligand atoms).

(i) Complexes with pyrimidine ribonucleotides.In the N62L,

N62L-R107G, and Kpn-N62L mutant complexes with a 2⬘OH

pyrimidine, the nucleotide base is positioned as in the native

NDKapmcomplex structure, deeper in the active site than in

the Halobacterium salinarum NDK-CDP complex structure

(2AZ3) (4), while in the triple NDKapmmutant complex, the

nucleotide position is restored to one close to that observed in the 2AZ3 structure (Fig. 3).

(iii) Complexes with deoxypurine nucleotides.In the

struc-ture of NDKapmin complex with dGDP, we observed an

agi-tated dGDP ligand exhibiting three different conformations

(Fig. 4). One of them (chain D; B factordGDP⫽ 58 Å2)

cor-responds to the conformation observed in theArabidopsis

thali-anaNDK enzyme in complex with dGDP (1S59) (16) (B

fac-tordGDP ⬎ 80 Å2). The second conformation shows a 180°

rotation of the nucleotide base (chains A and E; B factordGDP⫽

66 Å2), while in the last conformation, both the ribose and

the base parts are rotated by 180° (chains B, C, and F; B

factordGDP ⫽ 56, 54, and 66.5 Å2). Interestingly, in the

NDKKpn-R107Gdouble mutant and NDKKpn-N62L-R107G tri-ple mutant, the nucleotide is well ordered and adopts the

conformation observed in the 1S59 structure (B factordGDP

17.9 and 14.9 Å2).

(iv) Complexes with purine ribonucleotides.In complex with the native enzyme, the GDP is highly disordered despite the high-resolution data. The electronic density permits the posi-tioning of only the phosphate groups, while the base and ribose

moiety are barely visible (B factorsGDP between 50 and 100

Å2). The GDP nucleotide exhibits a 180° rotation compared to

its orientation in other available structures (2DXE [M. Kato-Murayama, K. Kato-Murayama, T. Terada, M. Shirouzu, and S.

Yokoyama, unpublished data], B factorGDP ⬍30 Å2; 1WKK

[S. Takeishi, N. Nakagawa, R. Masui, and S. Kuramitsu,

un-published data], B factorGDP⬃50 Å2; 1NUE [30], B factorGDP

⬃20 Å2; and 1BHN [24], B factor

GDP⬃90 Å

2), with the 2

OH and 3⬘OH being exposed to the solvent (Fig. 5). The R107

residue is disordered in the native enzyme. For the NDKKpn

mutant, the nucleotide, clearly visible, adopts two main con-formations, one similar to that observed in other NDK-GDP complexes and the other having the base rotated by 180° as in the native enzyme GDP complex, suggesting that the Kpn loop stabilizes this nucleotide in the enzyme binding site, even if not in an orientation optimal for phosphorylation. In both

mole-cules, the arginine residue (R111) NH2is pointing toward the

base and located 3.1 Å from the N2of the GDP base in the first

GDP orientation and 3.56 Å from the O6oxygen in the second

GDP orientation (Fig. 5). The B factorsGDP are roughly the

same,⬃30Å2, for both GDP molecules. In the NDK

⫹ Kpn-N62L-R107Gtriple mutant, where the arginine residue is replaced by

a glycine, the nucleotide conformation is restored to that

ob-served in other NDK structures (B factorGDP, 14 Å2) in

agree-ment with the stabilization of the purine nucleotide by the longer Kpn loop and the restoration of a standard orientation by the two other mutations.

Characterization of the NDKapm enzymatic activity. To

complement our structural study, we measured the enzymatic activities of the seven mutants on various nucleotide sub-strates. For the native enzyme as well as for some mutants, the high affinity for deoxypyrimidine nucleotides prevented the

direct measure of theKmvalues. We thus could only estimate

the affinity of the enzymes for these nucleotides (Table 1;

Fig. 6). It is worth noting that for each NDK enzyme, theVmax

was of the same order of magnitude for all nucleotides.

FIG. 3. Comparison of the CDP nucleotide orientations in the NDK structures. Red, NDKapm; yellow, NDKhs; pink, NDKN62L;

[image:5.585.67.259.69.252.2]

green, NDKKpn-N62L; cyan, NDK⫹Kpn-N62L-R107G.

FIG. 4. Comparison of the dGDP nucleotide orientations in NDKapm(red, chain D; blue, chain A; green, chain B), NDK⫹Kpn-R107G, and

NDK⫹Kpn-N62L-R107G(cyan). The dGDP nucleotide in NDKatis in transparent yellow.

on November 8, 2019 by guest

http://jvi.asm.org/

[image:5.585.136.448.578.704.2]
(6)

As expected from our structural analyses, the triple mutant

restored a “cellular” NDK phenotype, with equivalentKm

val-ues for all tested nucleotides in the 0.1 mM range (Table 1; Fig. 6). We thus dissected the respective contributions of each mutation on all types of nucleotides. Our results for the

NDKKpnmutant were at odds with our hypothesis that the

shorter Kpn loop of the mimivirus NDK was the main cause of the enhanced deoxypyrimidine affinity, since the activity of this mutant was still inhibited by concentrations of dTDP as low as 0.025 mM. On the other hand, restoring the longer Kpn loop induced an almost twofold increase in affinity for dGDP. A narrower active site thus appears to stabilize the binding of purine nucleotides, as suggested by the ordered GDP

nucleo-tide in the NDKKpnmutant, even if not in an optimal

con-formation. Overall, the most prominent sequence difference between the mimivirus NDK and its cellular homologues, i.e., a shorter Kpn loop, is not sufficient to account for the en-hanced affinity for pyrimidine nucleotides.

The second variation, leucine to asparagine (N62), involves a residue deep in the active binding site, close to the ribose of the nucleotide in complex structures. This residue change

could be responsible for the lower affinity of the enzyme for 2⬘

OH nucleotides due to a repulsive effect. Our activity mea-surements are consistent with this hypothesis, since the resto-ration of the leucine induces a twofold increase in CDP and

UDPKmvalues compared to the value for the native enzyme

(Table 1). There is also a significant decrease in the affinity for the deoxypyrimidine nucleotides and almost no change for deoxypurine nucleotides. This observation can be explained by a second effect of the N62 residue: the stabilization of the pyrimidine nucleotide through a hydrogen bond between the

N62-N␦2and the O2of the base moiety. Since this oxygen is

absent from the purines, the absence of interaction in the N-to-L mutant does not affect the binding of GDP and ADP

nucleotides. In conclusion, the N␦2 of N62 stabilizes the

de-oxypyrimidine nucleotide through a direct interaction with the

O2of the base, and the O␦1of the N62 residue repels 2⬘OH

nucleotides. This is supported by the Kpn-N62L double mutant and triple mutant structures, for which we obtained high-res-olution data (1.5 and 1.6 Å, respectively), allowing us to clearly

position the side chain and the N␦2versus O␦1atoms relative

to the nucleotide.

Finally, the most dramatic effect was observed for the R107G mutation, which caused a drastic decrease of the en-zyme affinity for pyrimidine nucleotides, permitting us to

di-rectly measure the kinetic constants of the NDKR107Gmutant.

TheKmvalues increased more than threefold for

deoxypyri-midine and twofold for pyrideoxypyri-midine nucleotides (Table 1). These results confirm the central role of R107 in nucleotide

stabilization through hydrogen bonds between the R107-Nε

and the O2of the pyrimidine base moiety.

We examined the numerous available sequences of NDK homologues to identify additional variations at the three conserved positions divergent in the mimivirus NDK. The shorter Kpn loop seems to be a unique feature of the

NDKapm. No occurrence of a similar mutation could be found in

the available databases, including environmental sequences. A direct consequence of this shorter loop is the widening of the binding site, which in turn causes a decreased affinity for GDP confirmed by its agitation in the structures of the

NDKapmenzyme and in the NDKN62L-R107Gdouble mutant

(see Tables S1 and S2 in the supplemental material). This

nucleotide becomes ordered in the NDKKpnmutant and in

[image:6.585.138.450.70.179.2]

the NDKKpn-N62L-R107Gtriple mutant. The wider binding

FIG. 5. Comparison of the GDP nucleotide orientations in NDKapm(red chain), NDK⫹Kpnfirst (GDP1, solid blue) and second (GDP2,

transparent green) GDP orientation, and NDKKpn-N62L-R107G(cyan). The GDP nucleotide in NDK⫹hsapis shown in transparent yellow.

TABLE 1. Apparent kinetic constants of NDKapmand its mutants on various nucleotides

Nucleotide

Km(mM)/Vmax(mM/min)

NDKapm NDK⫹Kpn NDKN62L NDKR107G NDK⫹Kpn-N62L NDKN62L-R107G NDK⫹Kpn-N62L-R107G

dGDP 0.16/0.19 0.11/0.11 0.18/0.18 0.29/0.25 0.094/0.096 0.06/0.13 0.05/0.06

dADP 0.25/0.25 NMa NM NM NM NM 0.07/0.07

dTDP ⬍⬍0.05 ⬍⬍0.05 0.05/0.07 0.15/0.18 0.07/0.08 0.086/0.13 0.12/0.07

dCDP ⬍⬍0.05 ⬍0.05 0.03/0.07 0.11/0.12 0.08/0.09 0.08/0.11 0.11/0.044

dUDP 0.03/0.05 NM 0.1/0.12 0.19/0.17 NM NM 0.11/0.06

CDP 0.1/0.12 0.07/0.05 0.05/0.05 0.21/0.14 0.034/0.07 0.19/0.15 0.12/0.05

UDP 0.1/0.15 NM 0.11/0.13 0.22/0.19 NM NM 0.14/0.08

a

NM, not measured.

on November 8, 2019 by guest

http://jvi.asm.org/

[image:6.585.43.542.615.716.2]
(7)

site of the mimivirus NDK may also allow larger substrates to be accommodated, eventually broadening its enzymatic activity. The biological relevance of the shorter Kpn loop might thus go beyond the sole variation in nucleotide spec-ificity.

The point N62L variation in the NDKapmsequence was

found in another variant of the otherwise strictly conserved leucine residue (L62). It is replaced by a methionine in the

Saccharomyces cerevisiae(G⫹C content ⬍ 40%) NDK se-quence. Finally, the position corresponding to the R107G variation in the mimivirus NDK is less conserved. There are a number of natural NDK sequence variants, with an

ex-treme case being theP. aerophilumNDK (Fig. 1), where this

residue is replaced by a 10-amino-acid insert, which also affects the substrate specificity (17). Interestingly, a

[image:7.585.43.543.65.590.2]

ran-domly generated G3R mutant of the archaealH. salinarum

FIG. 6. Comparison of activities of NDKapmand NDK⫹Kpn-N62L-R107Gas a function of dTDP concentration (A) and as a function of dGDP

concentration (B).

on November 8, 2019 by guest

http://jvi.asm.org/

(8)

NDK (67% G⫹C) also resulted in an increase in the enzyme affinity for pyrimidine nucleotides (35), thus confirming the critical role of this amino acid for substrate specificity.

In conclusion, there are multiple causes of the viral NDKapm

enzyme specificity pattern, since no single change was found to restore a “cellular” behavior. This is consistent with the viral enzyme’s progressive evolution from an ancestral NDK in

re-sponse to the AT richness of the viral genome (A⫹T content,

73%). Since the genome of the mimivirus host,Acanthamoeba,

has a 40% A⫹T content, dTTP is probably the limiting

nucle-otide for the virus replication. It thus makes some sense that

the NDKapmadapted its specificity pattern to better take

ad-vantage of the small concentration of dTDP in the amoeba host cell. It is worth noting that mimivirus also possesses a mitochondrial carrier protein that exhibits a strong preference for the dTTP and dATP deoxynucleotides (29). This again suggests that dTTP is in limited supply for viral replication. Moreover, the mimivirus genome encodes additional enzymes involved in dTTP biosynthesis, such as a thymidine kinase (responsible for dT/dU transformation into dTMP/dUMP) and a thymidylate synthase (producing dTMP from dUMP). Mim-ivirus also possesses its own deoxyribonucleotide monophos-phate kinase transforming a deoxynucleoside monophosmonophos-phate into a deoxynucleoside diphosphate and is thus able to produce dTDP from dTMP; a ribonucleoside reductase responsible for the conversion of ribonucleoside diphosphates into deoxyribo-nucleoside diphosphates (e.g., dADP, dGDP, and dCDP) is also present. In contrast, mimivirus lacks a dUTPase, an en-zyme found in most DNA viruses to prevent the

misincorpo-ration of dU in DNA. The strong affinity of the NDKapmis

consistent with its possible role in preventing the accumulation

of dUTP in vivo, as already suggested for theEscherichia coli

enzyme (34). It thus appears that throughout its evolution, mimivirus has maintained its capacity to actively shape the pool of deoxynucleotides necessary to the biosynthesis of its DNA, a process for which the NDK appears increasingly cen-tral (34).

The phylogenetic position of the mimivirus enzyme among NDK sequences from all domains (bacteria, archaea, and

eu-karya) illustrates the distance between theA. castellaniiNDK

and its viral counterpart (see Fig. S4 in the supplemental material). The mimivirus enzyme clusters with a group of eu-ryarcheal, crenarcheal, actinobacterial, and spirochete

se-quences while theA. castellaniisequence clusters with fungal,

alveolate, and metazoan sequences. This result favors an

an-cestral origin of the NDKapmover its recent horizontal transfer

from an eukaryotic organism. It is also consistent with the

pairwise sequence comparison of the NDKapm with the A.

castellaniiNDK (mRNA accession number, EC110217), show-ing that the latter lacks all the specific features of the viral enzyme.

ACKNOWLEDGMENTS

We gladly acknowledge the use of the proteomics and bioinformat-ics facilities from Marseille-Nice Ge´nopole and partial financial sup-port from the CNRS and the French Genome Research Network (RNG).

We thank Ioan Lascu for helpful discussions, Irina Gutsche for negative stain electron microscopy studies, Laeticia Frisch and Laura Mary for technical assistance, and Garry Duncan for reading the manuscript. We also thank for helpful assistance Didier Nurizzo on the

ID23 ESRF beam line, Joanne McCarthy and Christoph Mueller-Dieckmann at the ID29, ID14-EH1-EH2 and EH3 beam lines, the BM30A team with Michel Pirocchi, Laurence Serre and Jean-Luc Ferrer.

This work is dedicated to the memory of Lilian Jacquemet from the BM30A team.

REFERENCES

1.Abergel, C.2004. Spectacular improvement of X-ray diffraction through fast desiccation of protein crystals. Acta Crystallogr. D60:1413–1416. 2.Abergel, C., B. Coutard, D. Byrne, S. Chenivesse, J. B. Claude, C.

Dereg-naucourt, T. Fricaux, C. Gianesini-Boutreux, S. Jeudy, R. Lebrun, C. Maza, C. Notredame, O. Poirot, K. Suhre, M. Varagnol, and J. M. Claverie.2003. Structural genomics of highly conserved microbial genes of unknown func-tion in search of new antibacterial targets. J. Struct. Funct. Genomics4:141– 157.

3.Abergel, C., J. Rudinger-Thirion, R. Giege´, and J.-M. Claverie.2007. Virus-encoded aminoacyl-tRNA synthetases: structural and functional character-ization of mimivirus TyrRS and MetRS. J. Virol.81:12406–12417. 4.Besir, H., K. Zeth, A. Bracher, U. Heider, M. Ishibashi, M. Tokunaga, and

D. Oesterhelt.2005. Structure of a halophilic nucleoside diphosphate kinase from Halobacterium salinarum. FEBS Lett.579:6595–6600.

5.Brunger, A. T., P. D. Adams, G. M. Clore, W. L. DeLano, P. Gros, R. W. Grosse-Kunstleve, J. S. Jiang, J. Kuszewski, M. Nilges, N. S. Pannu, R. J. Read, L. M. Rice, T. Simonson, and G. L. Warren.1998. Crystallography and NMR system: a new software suite for macromolecular structure determi-nation. Acta Crystallogr. D54:905–921.

6.Cherfils, J., S. More´ra, I. Lascu, M. Ve´ron, and J. Janin.1994. X-ray structure of nucleoside diphosphate kinase complexed with thymidine diphosphate and Mg2⫹at 2-Å resolution. Biochemistry33:9062–9069. 7.Claverie, J.-M.2006. Viruses take center stage in cellular evolution. Genome

Biol.7:110.

8.Claverie, J.-M., C. Abergel, and H. Ogata.2009. Mimivirus. Curr. Top. Microbiol. Immunol.328:89–121.

9.Collaborative Computational Project Number 4. 1994. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D50:760–763. 10.Crawford, R. M., K. J. Treharne, O. G. Best, C. E. Riemen, R. Muimo, D. C.

Gruenert, S. Arnaud-Dabernat, J. Y. Daniel, and A. Mehta.2006. NDPK-A (but not NDPK-B) and AMPK␣1 (but not AMPK␣2) bind the cystic fibrosis transmembrane conductance regulator in epithelial cell membranes. Cell Signal.18:1595–1603.

11.Emsley, P., and K. Cowtan.2004. Coot: model-building tools for molecular graphics. Acta Crystallogr. D60:2126–2132.

12.Fan, Z., P. J. Beresford, D. Y. Oh, D. Zhang, and J. Lieberman.2003. Tumor suppressor NM23-H1 is a granzyme A-activated DNase during CTL-medi-ated apoptosis, and the nucleosome assembly protein SET is its inhibitor. Cell112:659–672.

13.Forterre, P.2006. Three RNA cells for ribosomal lineages and three DNA viruses to replicate their genomes: a hypothesis for the origin of cellular domain. Proc. Natl. Acad. Sci. USA103:3669–3674.

14.Gouet, P., E. Courcelle, D. I. Stuart, and F. Metoz.1999. ESPript: multiple sequence alignments in PostScript. Bioinformatics15:305–308.

15.Hammargren, J., T. Salinas, L. Mare´chal-Drouard, and C. Knorpp.2007. The pea mitochondrial nucleoside diphosphate kinase cleaves DNA and RNA. FEBS Lett.581:3507–3511.

16.Im, Y. J., J. I. Kim, Y. Shen, Y. Na, Y. J. Han, S. H. Kim, P. S. Song, and S. H. Eom.2004. Structural analysis of Arabidopsis thaliana nucleoside diphos-phate kinase-2 for phytochrome-mediated light signaling. J. Mol. Biol.343:

659–670.

17.Ishibashi, M., S. Tatsuda, K. Izutsu, K. Kumeda, T. Arakawa, and M. Tokunaga.2007. A single Gly114Arg mutation stabilizes the hexameric sub-unit assembly and changes the substrate specificity of halo-archaeal nucleo-side diphosphate kinase. FEBS Lett.581:4073–4079.

18.Iyer, L. M., L. Aravind, and E. V. Koonin.2001. Common origin of four diverse families of large eukaryotic DNA viruses. J. Virol.75:11720–11734. 19.Janin, J., C. Dumas, S. More´ra, Y. Xu, P. Meyer, M. Chiadmi, and J. Cher-fils.2000. Three-dimensional structure of nucleoside diphosphate kinase. J. Bioenerg. Biomembr.32:215–225.

20.Jeudy, S., J.-M. Claverie, and C. Abergel.2006. The nucleoside diphosphate kinase from mimivirus: a peculiar affinity for deoxypyrimidine nucleotides. J. Bioenerg. Biomembr.38:247–254.

21.Jeudy, S., B. Coutard, R. Lebrun, and C. Abergel.2005. Acanthamoeba polyphaga mimivirus NDK: preliminary crystallographic analysis of the first viral nucleoside diphosphate kinase. Acta Crystallogr. F61:569–572. 22.Kimura, N., N. Shimada, Y. Ishijima, M. Fukuda, Y. Takagi, and N.

Ish-ikawa.2003. Nucleoside diphosphate kinases in mammalian signal transduc-tion systems: recent development and perspective. J. Bioenerg. Biomembr.

35:41–47.

23.Kumar, P., A. Verma, A. K. Saini, P. Chopra, P. K. Chakraborti, Y. Singh, and S. Chowdhury.2005. Nucleoside diphosphate kinase from

on November 8, 2019 by guest

http://jvi.asm.org/

(9)

rium tuberculosis cleaves single strand DNA within the human c-myc pro-moter in an enzyme-catalyzed reaction. Nucleic Acids Res.33:2707–2714. 24.Ladner, J. E., N. G. Abdulaev, D. L. Kakuev, M. Tordova´, K. D. Ridge, and

G. L. Gilliland.1999. The three-dimensional structures of two isoforms of nucleoside diphosphate kinase from bovine retina. Acta Crystallogr. D55:

1127–1135.

25.Lascu, I., and P. Gonin.2000. The catalytic mechanism of nucleoside diphos-phate kinases. J. Bioenerg. Biomembr.32:237–246.

26.Lascu, I., A. Giartosio, S. Ransac, and M. Erent.2000. Quaternary structure of nucleoside diphosphate kinases. J. Bioenerg. Biomembr.32:227–236. 27.Leung, S. M., and L. E. Hightower.1997. A 16-kDa protein functions as a

new regulatory protein for Hsc70 molecular chaperone and is identified as a member of the Nm23/nucleoside diphosphate kinase family. J. Biol. Chem.

31:2607–2614.

28.Levit, M. N., B. M. Abramczyk, J. B. Stock, and E. H. Postel.2002. Inter-actions between Escherichia coli nucleoside-diphosphate kinase and DNA. J. Biol. Chem.277:5163–5167.

29.Monne´, M, A. J. Robinson, C. Boes, M. E. Harbour, I. M. Fearnley, and E. R. Kunji.2007. The mimivirus genome encodes a mitochondrial carrier that transports dATP and dTTP. J. Virol.81:3181–3186.

30.More´ra, S., M. L. Lacombe, Y. Xu, G. LeBras, and J. Janin.1995. X-ray structure of human nucleoside diphosphate kinase B complexed with GDP at 2 A resolution. Structure3:1307–1314.

31.Moynie´, L., M. F. Giraud, F. Georgescauld, I. Lascu, and A. Dautant.2007. The structure of the Escherichia coli nucleoside diphosphate kinase reveals a new quaternary architecture for this enzyme family. Proteins67:755–765.

32.Murshudov, G. N., A. A. Vagin, and E. J. Dodson.1997. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crys-tallogr. D53:240–255.

33.Navaza, J.2001. Implementation of molecular replacement in AMoRe. Acta Crystallogr. D57:1367–1372.

34.Nordman, J., and A. Wright.2008. The relationship between dNTP pool levels and mutagenesis in an Escherichia coli NDP kinase mutant. Proc. Natl. Acad. Sci. USA105:10197–10202.

35.Pedelacq, J. D., G. S. Waldo, S. Cabantous, E. C. Liong, and T. C. Terwil-liger.2005. Structural and functional features of an NDP kinase from the hyperthermophile crenarchaeon Pyrobaculum aerophilum. Protein Sci.14:

2562–2573.

36.Postel, E. H. 1999. Cleavage of DNA by human NM23-H2/nucleoside diphosphate kinase involves formation of a covalent protein-DNA complex. J. Biol. Chem.274:22821–22829.

37.Raoult, D., S. Audic, C. Robert, C. Abergel, P. Renesto, H. Ogata, B. La Scola, M. Suzan, and J.-M. Claverie.2004. The 1.2-megabase genome se-quence of Mimivirus. Science306:1344–1350.

38.Yonezawa, Y., K. Izutsu, H. Tokunaga, H. Maeda, T. Arakawa, and M. Tokunaga.2007. Dimeric structure of nucleoside diphosphate kinase from moderately halophilic bacterium: contrast to the tetrameric Pseudomonas counterpart. FEMS Microbiol. Lett.268:52–58.

39.Yoon, J. H., P. Singh, D. H. Lee, J. Qiu, S. Cai, T. R. O’Connor, Y. Chen, B. Shen, and G. P. Pfeifer.2005. Characterization of the 3⬘35⬘exonuclease activity found in human nucleoside diphosphate kinase 1 (NDK1) and sev-eral of its homologues. Biochemistry44:15774–15786.

on November 8, 2019 by guest

http://jvi.asm.org/

Figure

FIG. 1. Detail of the structural alignment of NDKapmpoint mutations in NDKsupplemental material
FIG. 2. Unique features of the NDKapmlighted in red and the Lat the Lstructure in green
FIG. 4. Comparison of the dGDP nucleotide orientations in NDKNDKapm (red, chain D; blue, chain A; green, chain B), NDK�Kpn-R107G, and�Kpn-N62L-R107G (cyan)
FIG. 5. Comparison of the GDP nucleotide orientations in NDKtransparent green) GDP orientation, and NDKapm (red chain), NDK�Kpn first (GDP1, solid blue) and second (GDP2,�Kpn-N62L-R107G (cyan)
+2

References

Related documents

• Dum Dums • Chewy Pops • Saf-T-Pops • Candy Canes • Chewy Canes • Circus Peanuts • Shrek Ogreheads SURF SWEETS • Gummy Worms • Gummy Swirls • Gummy Bears • Fruity Bears

We introduce new sufficient conditions for asymptotic stability of equilibrium states of nonlinear systems considering some inequalities for the higher order time derivatives

A total of 33 cases suspected for neoplastic growths were screened by cytological and histopathological procedures and 29 cases werediagnosed as tumors.Results of

Lucy came across some old school books yesterday. I came across my old love letters in a shoebox. I came across Julie’s glasses in the bathroom. Get on with means: 1) Have a

2011 Advanced Methods in Tobacco Control , Johns Hopkins Bloomberg School of Public Health Baltimore, MD.. 2011 Qualitative Methods in Tobacco Control , Johns Hopkins Bloomberg

In this study, we used physiologically immature corn silage (CS) and matured whole corn plants (WCP) as feedstocks to produce ethanol using ammonia fiber expansion (AFEX)

However if we are willing to use the available big data, combine with our own experience, leadership skills and courage we can make better decisions today than 10 years ago....

In the case of the Black River restoration, the MCDA analy- sis was used to narrow down and rank an initially large set of aquatic ecosystem monitoring metric alternatives based