proteinsSTRUCTURE O FUNCTION O BIOINFORMATICS

STRUCTURE NOTE

Crystal structure of Mycobacterium tuberculosisRv3168: A putative aminoglycosideantibiotics resistance enzymeSangwoo Kim, Chi My Thi Nguyen, Eun-Jung Kim, and Kyung-Jin Kim*

Pohang Accelerator Laboratory, Pohang University of Science and Technology, Pohang, Kyungbuk 790-784, Korea

Key words: tuberculosis; aminoglycoside; phosphotransferase; structure.

INTRODUCTION

Aminoglycosides are a broad-spectrum antibiotics fam-

ily originally isolated from soil bacteria, and this family

of antibiotics includes many clinically relevant drugs such

as streptomycin, kanamycin, neomycin, gentamicin, and

amikacin. Target of these compounds is the 16S rRNA of

the bacterial 30S ribosomal subunit, where they particu-

larly bind to the decoding aminoacyl site and stabilize

the tRNA binding to a cognate mRNA codon, resulting

in the decrease of the dissociation rate of aminoacyl-

tRNA and promote miscoding.1,2

Nowadays, use of aminoglycoside for the treatment of

pathogenic bacteria also has a problem of antibiotics

resistance, primarily through the deactivation of the anti-

biotics by enzymatic modification. One of the main chal-

lenges of aminoglycoside resistance is the considerable

quantity and diversity of modifying enzymes. Three fami-

lies of enzymes have been found to be responsible: ATP-

dependent phosphotransferases (APH), ATP-dependent

adenylyltransferases (ANT), and acetyl CoA-dependent

acetyltransferases (AAC). Many structures of the APH

enzymes are known until now, and enzymes such as

APH(30)-IIIa,3–5 APH(30)-IIa,6 APH(2@)-IIa,7 APH(2@)-IVa,8 and APH(90)-Ia9 give the molecular insights for the

enzyme mechanism.

Tuberculosis is the second leading infectious cause of

death after HIV with one-third of the human population

already infected. Every 4 s, there is one newly infected,

and every 15 s there is one died of tuberculosis. With

about 2 billion carriers of Mycobacterium tuberculosis

worldwide and the emergence of multidrug-resistant

strains, there are urgent needs to develop more effective

drugs. Many studies have been done to understand more

about the antibiotics resistance mechanisms in M. tuber-

culosis.10 The complete genome sequence of M. tubercu-

losis strain H37Rv has revealed some mycobacterial pro-

tein sequences that have homology to the aminoglycoside

phosphotransferase enzymes.11,12 The structures of these

putative aminoglycoside phosphotransferase enzymes in

M. tuberculosis are still not investigated until now.

In an effort to find a crucial enzyme responsible for

the aminoglycoside resistance through phosphorylation

of aminoglycoside, we searched M. tuberculosis whole ge-

nome and selected three phosphotransferase candidates

such as Rv3168, Rv3225c, and Rv3817. In this study, we

report a crystal structure of Rv3168 that assigned as a

putative aminoglycoside phosphotransferase.

MATERIALS AND METHODS

Cloning, expression, purification, and crystallization of

Rv3168 will be described elsewhere (Nguyen et al., manu-

script in preparation). Briefly, the recombinant Rv3168

protein was expressed using the bacterial expression sys-

tem and purified through sequential chromatographic

Sangwoo Kim and Chi My Thi Nguyen contributed equally to this work

*Correspondence to: Kyung-Jin Kim, Pohang Accelerator Laboratory, Pohang

University of Science and Technology, Pohang, Kyungbuk 790-784, Korea.

E-mail: kkj@postech.ac.kr

Received 9 February 2011; Revised 16 April 2011; Accepted 4 May 2011

Published online 15 July 2011 in Wiley Online Library (wileyonlinelibrary.com).

DOI: 10.1002/prot.23119

Grant sponsor: Korean Government (MEST) [National Research Foundation of

Korea Grant (NRF)]; Grant numbers: NRF-M1AXA002-2010-0029768, NRF-2009-

C1AAA001-2009-0093483

VVC 2011 WILEY-LISS, INC. PROTEINS 2983

steps. Suitable crystals for diffraction experiments were

obtained at 228C within 5 days from the precipitant of

0.2M Ca(OAc)2, 0.1M Tris-HCl, pH 7.0, and 20% PEG

3000. Rv3168 bound with ATP molecule was crystallized

with the same crystallization condition supplemented

with 10 mM ATP. SeMet crystals of Rv3168 were

obtained from the same crystallization condition as that

of the native protein crystals. The crystals were trans-

ferred to cryoprotectant solution containing 0.2M

Ca(OAc)2, 0.1M Tris-HCl, pH 7.0, 20% PEG 3000, and

30% glycerol, fished out with a loop larger than the crys-

tals, and flash frozen by immersion in liquid nitrogen at

100 K. The data were collected to a resolution of 1.67 A

at 6C1 beamline (MXII) of the Pohang Accelerator Labo-

ratory (PAL, Pohang, Korea) using a Quantum 210 CCD

detector (ADSC, USA). The data were then indexed, inte-

grated, and scaled using the HKL2000 suite.13 Crystals

belonged to space group P212121, with unit cell parame-

ters of a 5 56.74 A, b 5 62.37 A, and c 5 103.61 A.

Assuming one molecules of Rv3168 per asymmetric unit,

the crystal volume per unit of protein mass was 2.91 A3

Da21,14 which corresponds to a solvent content of

�57.76%. SAD data with SeMet crystal were collected at

the 6C1 beamline (MXII) of the PAL at the wavelength

of 0.97953 A. Nine of the 10 Se atoms in the asymmetric

unit were identified using program SOLVE15 at 2.2-A re-

solution. The electron density was improved by density

modification using the RESOLVE,16 resulting in 79.9%

of the residues automatically built. Further model build-

ing was performed manually using the program Win-

Coot, and the refinement was performed with CCP4

refmac5 and CNS. The data statistics are summarized in

Table I. The refined model of apo form and ATP-bound

form of Rv3168 was deposited in the Protein Data Bank

(pdb code 3ATS and 3ATT for apo form and ATP-bound

form of Rv3168, respectively).

RESULTS AND DISCUSSION

The structure of Rv3168 protein was solved by SAD

analysis with the SeMet-substituted protein crystal (Table

I). The asymmetric unit of the crystal contained one

Rv3168 molecule. Among 378 residues of the full-length

protein, an N-terminal region (residues 1–21) could not

be traced because of the poor electron density map. The

Rv3168 monomer is composed of two lobes, a smaller

N-terminal lobe and a larger C-terminal lobe. The N-ter-

minal lobe consists of five-stranded antiparallel b-sheet(b1–b5) and a helix (a1) that is located at the center of

the hollow formed by five-stranded b-sheets. Two 14-res-

idue loop structures are found at the N-terminal lobe:

one loop (loop 1, residues 45–58) connects b1 and b2,and there is no contact to other residues, which subse-

quently forms a big cavity; the other loop (loop 2, resi-

dues 81–94) connects b3 and a2, and the folding is sta-

bilized by the hydrophobic interactions with a5 and a10of the C-terminal lobe [Fig. 1(A,B)]. The C-terminal lobe

consisted mainly of a-helices (a2–a12) with two short

stretches of b-sheet (b6–b9). A long loop (loop 4, resi-

dues 241–275) incorporates the two b-sheet stretches andforms a central core region of the protein together with

two helices (a3 and a4). Four helices at the C-terminal

lobe (a5, a6, a10, and a12) form a four-helical bundle,

and a one-turn helix (a11) incorporated into a long loop

(loop 5, residues 351–366) stabilizes an exposed hydro-

phobic region of the long helix a10 [Fig. 1(B)]. Three

helices of a7, a8, and a9 are inserted between a central

core region and a four-helical bundle. At the center of

the cavity, one magnesium ion was found to be coordi-

nated by three conserved residues (Glu57, Asp267, and

Glu269) [Figs. 1(B) and 2(A)].

Structural alignment using Dali server reveals that the

structure of Rv3168 has the highest similarity score to that

of APH(30)-IIIa from Enterococcus faecalis (EfAPH(30)-IIIa,pdb code 2B0Q), although there is no significant amino

acid sequence similarity between these two proteins with

less than 10% [Fig. 1(A,C)]. Because EfAPH(30)-IIIa is

known to be a aminoglycoside phosphotransferase and a

crucial enzyme that confers aminoglycoside resistance to

E. faecalis,3–5 the structural similarity between these two

Table IData Collection and Refinement Statistics

Apo form SeMet ATP-bound form

Data collectionSpace group P212121 P212121 P212121Cell dimensionsa, b, c (�) 56.74, 62.37,

103.6156.45, 62.25,

103.4355.88, 62.10,

103.27Wavelength 1.23985 0.97953 1.23985Resolution (�) 50.00–1.67

(1.73–1.67)50.00–2.06(2.13–2.06)

30.00–2.00(2.07–2.00)

Rsyma (%) 5.8 (30.5) 7.2 (16.6) 10.3 (46.3)

I/rI 35.78 (3.09) 32.55 (7.74) 23.28 (2.86)Completeness (%) 95.5 (90.6) 98.8 (97.4) 92.4 (74.1)Redundancy 9.8 5.1 6.1RefinementResolution (�) 50.00–1.67 30.00–2.00No. reflections 40,527 21,869Rwork

b/Rfreec 19.1/23.3 19.0/24.9

No. atomsProtein 2817 2817Ligand/ion 14 41Water 320 219

B-factors 23.01 31.82RMS deviationsBond lengths (�) 0.023 0.022Bond angles (8) 1.747 1.973

Values in parentheses are for the highest resolution shell.aRsym 5 Shkl Sj|Ij 2|/ShklSjIj, where is the mean intensity of reflection hkl.bRfactor 5 Shkl||Fobs| 2 |Fcalc||/Shkl|Fobs|, where Fobs and Fcalc are, respectively, the

observed and calculated structure factor amplitude for reflections hkl included in

the refinement.cRfree is the same as Rfactor but calculated over a randomly selected fraction (5%)

of reflection data not included in the refinement.

S. Kim et al.

2984 PROTEINS

proteins leads us to postulate that Rv3168 could be

assigned as a member of a phosphotransferase enzyme

family. To confirm the function of Rv3168 as a phospho-

transferase and identify the ATP-binding mode, we then

determined an ATP-bound form of the protein at 2.0 A.

An ATP molecule was found to tightly bind in a cavity

formed by three long loops (loop 2, loop 3, and loop 4).

Unlike an apo form of Rv3168 where only one magnesium

ion was bound, two magnesium ions were found to be

coordinated in the Rv3168–ATP complex structure by

three residues of Asn254, Asp267, and Glu269 [Fig. 2(B)].

The binding of ATP to the enzyme is mediated by the

Figure 1Overall shape of Rv3168. (A) Alignment of amino acid sequences of Rv3168 and E. faecalis APH(30)-IIIa (EfAPH(30)-IIIa). Amino acid sequences of

Rv3168 and EfAPH(30)-IIIa were aligned based on the structural information. Secondary structure elements are shown and labeled based on the

structure of Rv3168. Identical and highly conserved residues are presented in red- and blue-colored characters, respectively. The xxDxxxxNx kinase

motif is indicated by rectangles of dark-orange color, and conserved residues involved in the magnesium coordination and enzyme catalysis are

marked with purple-colored stars. Mt and Ef are abbreviations for M. tuberculosis and E. faecalis, respectively. (B) Folding of Rv3168. The

secondary structure elements of a-helix and b-sheet are distinguished by cyan and magenta colors, respectively, and loops connecting the secondary

structures are by salmon color. (C) Superposition of Rv3168 and EfAPH(30)-IIIa. The structures of Rv3168 (cyan color) and EfAPH(30)-IIIa (orange

color) are superposed.

Crystal Structure of M. tuberculosis Rv3168

PROTEINS 2985

coordination of triphosphate portion of the ATP molecule

to the magnesium ions, and the Arg79 residue contributes

the binding of ATP by forming hydrogen bonds

with hydroxyl groups of a and b phosphates [Fig. 2(B)].

An adenine ring of ATP binds to a hydrophobic pocket

constituted by a cluster of nonpolar residues such as Ile60,

Val77, Val113, Pro114, Met133, Val136, Val140, Leu256,

Ala264, and Leu266. A hydrogen bond formed between

an amine base of the adenine ring and a hydroxyl group

of the main chain of Asp134 assists the binding of ATP to

the enzyme.

Interestingly, upon the binding of ATP to the enzyme,

a structural change was observed on a Glu57-containing

loop (56ser-57glu-58thr). As described above, the Glu57

residue was involved in the magnesium coordination in

an apo form of the enzyme; however, in the ATP-bound

form of the enzyme, the residue was replaced by b-phos-phate of ATP for the coordination of the magnesium ion

and moved toward the potential substrate-binding pocket

by 6.02 A [Fig. 2(A,B)]. The structural change on the

Glu57 residue upon the binding of ATP seems to mediate

an environmental change on the potential substrate-bind-

ing pocket and allows us to postulate a procedure of en-

zymatic reaction; first, ATP binds the enzyme and then

the substrate binds in its binding pocket with the aid of

structural change of Glu57.

As in EfAPH(30)-IIIa, a large pocket, speculated to

serve a substrate-binding site, is observed near the ATP-

binding pocket, and these two pockets are penetrated

each other forming a short tunnel [Fig. 2(C)]. The

pocket is constituted by highly negatively charged resi-

dues such as Asp249, Asp254, Asp267, and Glu269. Con-

sidering that aminoglycosides are almost positively

charged molecules, the presence of negatively charged

substrate-binding pockets leads us to speculate that

Rv3168 could be a candidate for aminoglycoside phos-

photransferase enzyme. In fact, it is known that APHs

and many other aminoglycoside-modifying enzymes like

AAC or ANT possess a large negatively charged sub-

strate-binding pocket.1,17–19 Putative substrate-binding

residues in the pocket are also predicted by amino acid

sequence alignment with other APH (30) proteins. Resi-

dues Gly248, Asp249, Asn254, and Asp267 are located

at the same positions as the corresponding residues

Gly189, Asp190, Asn195, and Asp268 of EfAPH(30)-IIIa.These residues form the xxDxxxxNx kinase motif and

interestingly are located in the tunnel that connects two

pockets [Fig. 1(A)].

Figure 2ATP-binding mode of Rv3168. (A) Magnesium coordination of an apo

form of Rv3168. One magnesium ion is coordinated at the apo form of

Rv3168. The bound magnesium ion is shown by a yellow-colored

sphere. Residues involved in the magnesium coordination are shown bycyan-colored stick model. E57-containing loop that undergoes structural

changes upon the binding of ATP is distinguished by green color. (B)

ATP-binding mode of Rv3168. Two magnesium ions are coordinated at

the ATP-bound form of Rv3168. The magnesium ions and the bound

ATP molecule are shown by yellow-colored sphere and magenta-colored

stick models. Residues involved in the magnesium coordination and

ATP-binding are shown by cyan-colored stick model. Residue of E57

and E57-containing loop that undergoes structural changes upon the

binding of ATP is distinguished by green color. (C) Electrostatic

potential models of the ATP-binding and the substrate-binding pockets.

The ATP-binding and the substrate-binding pockets are presented by

electrostatic potential models and indicated by green-dotted and blue-

dotted circles, respectively. Two magnesium ions and the bound ATP

molecule are shown by yellow-colored spheres and magenta-colored

stick model, respectively. The acetate molecule bound at the substrate-

binding pocket of Rv3168 is shown by green-colored stick models.

S. Kim et al.

2986 PROTEINS

As described by Hon et al., Asp190 has been proved to

play a crucial role in EfAPH(30)-IIIa catalysis. Mutagene-

sis of Asp190 resulted in an EfAPH(30)-IIIa enzyme with

only residual activity.5 This Asp residue is predicted to

have interaction with the incoming substrate hydroxyl

group like the conserved Asp residue from several ki-

nases, in which the active-site Asp residue has a role of

catalytic base needed for deprotonation of the substrate

hydroxyl group for efficient attack at the g-phosphate of

ATP. In aminoglycoside-bound structure of EfAPH(30)-IIIa, we can see that the Asp-30OH hydrogen bond is sig-

nificant, and the aminoglycoside 30hydroxyl group is the

site for phosphorylation.4 These facts gave us a clue to

predict that the Asp249 residue in our structure, which is

located at the same position of the corresponding

Asp190 residue of EfAPH(30)-IIIa, obviously plays an im-

portant role in the catalysis of Rv3168 as well [Fig.

1(A)]. Interestingly, an acetate molecule that might be

driven from the crystallization solution was observed to

be bound at the active site of the Rv3168; moreover, the

hydroxyl group of the acetate molecule was located at the

similar position of the phosphate-accepting hydroxyl

group of the neomycin bound at the active site of

EfAPH(30)-IIIa [Fig. 2(C)], which leads us to speculate

that the bound acetate molecule mimics the substrate

binding of Rv3168, and the enzymatic mechanism of

Rv3168 might be similar to that of EfAPH(30)-IIIa.In summary, we determined the structure of Rv3168

protein, the first structure among the putative aminogly-

coside phosphotransferases in M. tuberculosis. Without

significant amino acid sequence similarity, the overall

structure of Rv3168 was similar to that of E. faecalis

APH(30)-IIIa known as an aminoglycoside phosphotrans-

ferase and a crucial enzyme that confers aminoglycoside

resistance to the strain. Together with the structural simi-

larity between M. tuberculosis Rv3168 and E. faecalis

APH(30)-IIIa, an ATP-bound form of structure and a

large negatively charged substrate-binding pocket located

near the ATP-binding pocket lead us to conclude that the

Rv3168 protein might be a phosphotransferase family

enzyme and to postulate that the Rv3168 protein could

be a candidate for conferring aminoglycoside resistance

to M. tuberculosis. In fact, Rv3168 showed only the resid-

ual aminoglycoside phosphotransferase activity (data not

shown), which could be explained by the fact that the

protein is originated from the wild-type M. tuberculosis

strain H37Rv that does not show a resistance to amino-

glycoside family of antibiotics. It has been known that

mutations at the 30S ribosomal subunit confer the ami-

noglycoside resistance to the wild-type M. tuberculosis

strain H37Rv. Taken together, we speculate that not the

native but the mutant Rv3168 protein might function as

a aminoglycoside-modifying enzyme by phosphorylation

of aminoglycoside antibiotics, and examinations of the

Rv3168-coding sequences from the aminoglycoside-resist-

ant M. tuberculosis strains are strongly suggested.

REFERENCES

1. Kotra LP, Haddad J, Mobashery S. Aminoglycosides: perspectives on

mechanisms of action and resistance and strategies to counter

resistance. Antimicrob Agents Chemother 2000;44:3249–3256.

2. Karimi R, Ehrenberg M. Dissociation rate of cognate peptidyl-tRNA

from the A-site of hyper-accurate and error-prone ribosomes. Eur J

Biochem 1994;226:355–360.

3. Burk DL, Hon WC, Leung AK, Berghuis AM. Structural analyses

of nucleotide binding to an aminoglycoside phosphotransferase.

Biochemistry 2001;40:8756–8764.

4. Fong DH, Berghuis AM. Substrate promiscuity of an aminoglyco-

side antibiotic resistance enzyme via target mimicry. EMBO J 2002;

21:2323–2331.

5. Hon WC, McKay GA, Thompson PR, Sweet RM, Yang DS, Wright

GD, Berghuis AM. Structure of an enzyme required for aminogly-

coside antibiotic resistance reveals homology to eukaryotic protein

kinases. Cell 1997;89:887–895.

6. Nurizzo D, Shewry SC, Perlin MH, Brown SA, Dholakia JN, Fuchs

RL, Deva T, Baker EN, Smith CA. The crystal structure of amino-

glycoside-30-phosphotransferase-IIa, an enzyme responsible for anti-

biotic resistance. J Mol Biol 2003;327:491–506.

7. Paul DJ, Seedhouse SJ, Disney MD. Two-dimensional combinatorial

screening and the RNA Privileged Space Predictor program

efficiently identify aminoglycoside-RNA hairpin loop interactions.

Nucleic Acids Res 2009;37:5894–5907.

8. Toth M, Vakulenko S, Smith CA. Purification, crystallization and

preliminary X-ray analysis of Enterococcus casseliflavus aminoglyco-

side-2@-phosphotransferase-IVa. Acta Crystallogr Sect F Struct Biol

Cryst Commun 2010;66(Part 1):81–84.

9. Fong DH, Lemke CT, Hwang J, Xiong B, Berguis AM. Structure of

the antibiotic resistance factor spectinomycin phosphotransferase

from Legionella pneumophila. J Biol Chem 2010;285:9545–9555.

10. Johnson R, Streicher EM, Louw GE, Warren RM, van Helden PD,

Victor TC. Drug resistance in Mycobacterium tuberculosis. Curr

Issues Mol Biol 2006;8:97–111.

11. Philipp WJ, Poulet S, Eiglmeier K, Pascopella L, Balasubramanian

V, Heym B, Bergh B, Bloom BR, Jacobs WR, Jr, Cole ST. An inte-

grated map of the genome of the tubercle bacillus, Mycobacterium

tuberculosis H37Rv, and comparison with Mycobacterium leprae.

Proc Natl Acad Sci USA 1996;93:3132–7312.

12. Wright GD. Aminoglycoside-modifying enzymes. Curr Opin Micro-

biol 1999;2:499–503.

13. Otwinowski Z, Minor W. Processing of X-ray diffraction data col-

lected in oscillation mode. Methods Enzymol 1997;276:307–326.

14. Matthews BW. Solvent content of protein crystals. J Mol Biol

1968;33:491–497.

15. Terwilliger TC, Berendzen J. Automated MAD and MIR structure

solution. Acta Crystallogr D Biol Crystallogr 1999;55:489–861.

16. Terwilliger TC. Maximum-likelihood density modification. Acta

Crystallogr D Biol Crystallogr 2000;56:965–972.

17. Pedersen LC, Benning MM, Holden HM. Structural investigation of

the antibiotic and ATP-binding sites in kanamycin nucleotidyltrans-

ferase. Biochemistry 1995;34:13305–13311.

18. Wolf E, Vasilev A, Makino Y, Sali A, Nakatani Y, Burley SK. Crystal

structure of a GCN5-related N-acetyltransferase: Serratia marcescens

aminoglycoside 3-N-acetyltransferase. Cell 1998;94:439–449.

19. Wybenga-Groot LE, Draker K, Wright GD, Berghuis AM. Crystal struc-

ture of an aminoglycoside 60-N-acetyltransferase: defining the GCN5-

related N-acetyltransferase superfamily fold. Structure 1999; 7:497–507.

Crystal Structure of M. tuberculosis Rv3168

PROTEINS 2987