FEBS Letters 582 (2008) 623–626

Structure of Escherichia coli tetrahydrodipicolinateN-succinyltransferase reveals the role of a conserved C-terminal

helix in cooperative substrate binding

Long Nguyen, Guennadi Kozlov, Kalle Gehring*

Department of Biochemistry, McGill University, Montreal, 3655 Promenade Sir William Osler, Quebec, Canada H3G 1Y6

Received 16 January 2008; revised 18 January 2008; accepted 21 January 2008

Available online 31 January 2008

Edited by Hans Eklund

Abstract Tetrahydrodipicolinate N-succinyltransferase is anenzyme present in many bacteria that catalyzes the first stepof the succinylase pathway for the synthesis of meso-diamino-pimelate and the amino acid LL-lysine. Inhibition of the synthesisof meso-diaminopimelate, a component of peptidoglycan presentin the cell wall of bacteria, is a potential route for the develop-ment of novel anti-bacterial agents. Here, we report the crystalstructure of the DapD tetrahydrodipicolinate N-succinyltrans-ferase from Escherichia coli at 2.0 A resolution. Comparisonof the structure with the homologous enzyme from Mycobacte-rium bovis reveals the C-terminal helix undergoes a large rear-rangement upon substrate binding, which contributes tocooperativity in substrate binding.� 2008 Federation of European Biochemical Societies. Publishedby Elsevier B.V. All rights reserved.

Keywords: Protein structure; Succinyltransferase; THDP;DapD; Lysine biosynthesis; X-ray crystallography

1. Introduction

meso-Diaminopimelate (DAP), a component of peptidogly-

can, plays important role in the maintenance of bacterial cell

wall structural integrity. The three routes to synthesize meso-

diaminopimelate are the succinylase pathway [1], dehydroge-

nase pathway [2], and the acetylase pathway [3]. The central

route of the DAP/LL-lysine pathway is the succinylase route

used by many microorganisms including gram-negative bacte-

ria, gram-positive cocci, blue-green algae, and higher plants.

The enzyme tetrahydrodipicolinate (THDP) N-succinyltrans-

ferase, product of the Escherichia coli dapD gene, catalyzes

the committed step of the succinylase route, which involves

the conversion of 2,3,4,5-tetrahydrodipicolinate (THDP) and

succinyl-CoA to N-succinyltetrahydrodipicolinate and coen-

zyme A (CoA). The enzyme displays sequential binding kinet-

ics with roughly a twofold enhancement of substrate binding in

the presence of succinyl-CoA [1].

Abbreviations: THDP, 2,3,4,5-tetrahydrodipicolinate; DAP, diamino-pimelate; LbH, left-handed parallel b-helix; CoA, coenzyme A; Tris,tris(hydroxymethyl)aminomethane

*Corresponding author. Fax: +1 514 398 7384.E-mail address: kalle.gehring@mcgill.ca (K. Gehring).

0014-5793/$34.00 � 2008 Federation of European Biochemical Societies. Pu

doi:10.1016/j.febslet.2008.01.032

The crystal structure of THDP N-succinyltransferase from

Mycobacterium bovis at 2.2 A resolution revealed a trimeric

structure with each protomer consisting of three domains: a

helical N-terminal domain, a b-sheet C-terminal domain and

a distinctive b-helix middle domain [4]. This unique domain

contains imperfect tandem-repeated hexapeptide motifs, [LIV]–

[GAED]–X2–[STAV]–X, which are shared by a number of

enzymes, mostly acyltransferases [4]. These motifs encode b-

strands, which fold into a triangular prism-shaped, left-handed

b-helix (LbH) [5]. Crystal structures with bound substrates and

CoA have identified the active site of enzyme as a long narrow

groove located at the interface between two left-handed paral-

lel b-helices (LbH) of the trimeric enzyme [6,7].

In this study, we present the crystal structure of THDP

N-succinyltransferase from E. coli at 2.0 A resolution without

substrate and CoA. Comparison with the previous structure of

the apoenzyme from M. bovis reveals the C-terminus of the

E. coli enzyme is structured and forms an a-helix, which blocks

the CoA binding site. Upon binding CoA, this helix moves

20 A to participate in substrate binding [6,7]. Together the

E. coli and M. bovis structures reveal the conformational

changes responsible for the cooperative binding of CoA and

substrate by the enzyme.

2. Materials and methods

2.1. Protein expression, preparation, and purificationTHDP N-succinyltransferase from E. coli (residues 2–274) was

cloned into pFO4 vector, a derivative of pET15b vector (Amersham-Pharmacia), and expressed in E. coli BL21 (DE3) in rich (LB) mediumas a fusion with the N-terminal uncleavable His-tag (MGSSHH-HHHHGS). The fusion protein was purified by affinity chromatogra-phy on Ni2+-charged chelating sepharose resin and additionallypurified using size-exclusion chromatography.

2.2. CrystallizationCrystallization conditions were identified utilizing hanging drop

vapor diffusion with the JCSG+ crystallization suite (QIAGEN). Thebest crystals were obtained by equilibrating a 1.0 ll drop of THDPN-succinyltransferase (15 mg/ml) in 20 mM tris(hydroxymethyl)-aminomethane (Tris)–HCl (pH 7.5), mixed with 1.0 ll of reservoirsolution containing 0.9 M tri-sodium citrate and 0.1 M Tris–HCl(pH 8.0). Crystals grew in 7–10 days at 20 �C. For data collection,the crystals were picked up in a nylon loop and flash cooled in aN2 cold stream. The solution for cryoprotection contained the reser-voir solution with addition of 20% (v/v) glycerol. The crystals con-tain one THDP N-succinyltransferase subunit in the asymmetric unitcorresponding to Vm = 3.41 A3 Da�1 and a solvent content of 63.9%[8].

blished by Elsevier B.V. All rights reserved.

Table 1Data collection and refinement statistics

DapD

Data collectionSpace group P63

Cell dimensionsa, b, c (A) 102.92, 102.92, 69.23a, b, c (�) 90.00, 90.00, 120.00

Resolution (A) 50–2.0 (2.07–2.00)a

Rsym 0.114 (0.375)I/rI 14.5 (6.0)Completeness (%) 99.9 (100.0)Redundancy 9.8 (9.1)

RefinementResolution (A) 44.59–2.00

624 L. Nguyen et al. / FEBS Letters 582 (2008) 623–626

2.3. Structure solution and refinementDiffraction data from a single crystal of THDP N-succinyltransfer-

ase were collected on an ADSC Quantum-210 CCD detector (AreaDetector Systems Corp.) at beamline A1 at the Cornell High-EnergySynchrotron Source (CHESS) (Table 1). Data processing and scalingwere performed with HKL2000 [9]. The structure was determined bymolecular replacement with Phaser [10], using the coordinates ofTHDP N-succinyltransferase apoenzyme from M. bovis (PDB entry1TDT). The initial model obtained from Phaser was completed andadjusted with the program Xfit [11] and was improved by several cyclesof refinement, using the program REFMAC 5.2. [12] and model refit-ting. At the latest stage of refinement, we also applied the translation-libration-screw (TLS) option [13]. The refinement statistics are given inTable 1. The final model has good stereochemistry according to theprogram PROCHECK [14]. The coordinates and structure factorshave been deposited in the RCSB Protein Data Bank (accession num-ber 3BXY).

No. reflections 26848Rwork/Rfree 0.175/0.205No. atoms 2294

Protein 2119Water 185

B-factorsProtein 20.54Water 38.10

R.m.s. deviationsBond lengths (A) 0.014Bond angles (�) 1.36

Ramachandran statistics (%)Most favored regions 90.7Additional allowed regions 9.3

aHighest resolution shell is shown in parentheses.

3. Results and discussion

3.1. Structure

THDP N-succinyltransferase from E. coli was crystallized

and its structure determined at 2.0 A resolution by molecular

replacement using the structure of the homologous (95%

identity) enzyme from M. bovis (PDB entry 1TDT). Out of

285 residues in the His-tag fusion protein, three residues

Gly262-Val264 from THDP N-succinyltransferase and seven

N-terminal residues from the cloning linker were missing

from the electron density map. Noteworthy, the N-terminal

His-tag assisted in crystallization, as five of the His-tag resi-

dues are ordered with some of them participating in crystal

contacts.

Fig. 1. (A) Overlay between the crystal structures of THDP N-succinyltransmagenta). Two orthogonal views are shown with the domains and N- andshowing the electron density for the hydrophobic residues at the interfaceN-succinyltransferase. Figures are made with PyMOL (http://pymol.sourcefo

The structure contains four N-terminal a-helices ranging

from residues 1–17, 25–40, 57–69, and 95–101 with b-strand

hairpin loop structures between a2–a3 and a3–a4 (Fig. 1A).

ferase protomers from E. coli (green) and M. bovis (PDB entry 1TDT;C-termini labeled. (B) Composite omit map of the C-terminal helix

with the LbH domain. (C) Structure of the trimer of E. coli THDPrge.net/).

Fig. 2. The C-terminal helix of the THDP N-succinyltransferase. (A)Enlarged view of the interactions between the C-terminal helix and theb-sheet domains from the same subunit (green) and adjacent subunit(cyan). The side chains of interacting residues are shown and labeled.(B) Sequence conservation of C-terminus from the proteins of THDPN-succinyltransferase family. THDP N-succinyltransferase sequencefrom E. coli (gi:16128159) is aligned with homologous proteins fromMycobacterium bovis (gi:6014910), Haemophilus influenzae Rd KW20(gi:30995463), Alteromonas macleodii (gi:88793834), Pseudoalteromon-as tunicata D2 (gi:88857980), Chromobacterium violaceum ATCC12472 (gi:34495905), Neisseria meningitidis MC58 (gi:15676250), andAcinetobacter baumannii ATCC 17978 (gi:126642574). The helixobserved in THDP N-succinyltransferase from E. coli is indicated.

L. Nguyen et al. / FEBS Letters 582 (2008) 623–626 625

The LbH domain consists of residues 102–233. Following the

LbH domain is a third domain consisting of b-strands, which

cap the C-terminal end of the LbH domain. Unexpectedly, the

structure also reveals a C-terminal a-helix (residues 265–271)

that was not observed in the M. bovis structure (Fig. 1B).

Superposition of two unliganded structures shows that the

structures are otherwise nearly identical with RMSD of

0.48 A for backbone atoms of residues 2–256. Aside the C-ter-

minus, the most prominent differences are located in the loop

between two C-terminal b-strands and the N-terminal helix,

which adopts a slightly different angle (Fig. 1A). These regions

are involved in crystal contacts arising between the helix a1 of

one subunit of the trimer with the C-terminal b-sheet cap of an

adjacent trimer.

While there is a single subunit in the asymmetric unit due to

the high-symmetry P63 space group, the trimer is easily assem-

bled via crystallographic threefold symmetry (Fig. 1C). As

described earlier [4], the main contributions to trimer forma-

tion come from the LbH and C-terminal b-sheet domains.

Interestingly, the narrow junction between the C-terminal

domains of the trimeric enzyme contains two ordered water

molecules shared by hydrophilic amino acid side chains from

each of the three subunits. The first water molecule interacts

with side chains from Ser246, while the second water molecule

interacts with side chains from Asn237.

The C-terminal helix observed in the E. coli THDP N-succi-

nyltransferase structure is folded onto the same subunit with

numerous hydrophobic interactions: Ile266 with Phe183,

Leu270 with Val201 and Val234, Leu269 with Val256,

Val201, and Val232, and Ile273 with Val253 (Fig. 2A). The

helix has a strong amphipathic character and through its

hydrophilic surface makes contacts with residues Ser211 and

Arg213 from the LbH domain of the adjacent subunit and

crystal contacts with the N-terminus of another trimer. One

third of the surface of the helix is exposed to solvent.

Analysis of sequence conservation for the THDP N-succi-

nyltransferase family suggests that the C-terminal helix in the

E. coli enzyme structure is not an artifact of crystallization.

The residues comprising the helix are remarkably conserved

(Fig. 2B). The Leu269–Leu270 tandem is invariant across the

family, and Ile266 is highly conserved, only substituted by

leucine or valine. While the overall level of similarity in the

THDP N-succinyltransferase protein family is relatively high,

the C-termini are fairly divergent outside of residues 263–273.

Fig. 3. Conformational change of the C-terminal region of THDP N-succipimelate bound state (B, PDB entry 1KGT). The C-terminal helix is coloredloops green. The adjoining subunit is colored gray. Succinyl-CoA and pimelathe disordered loop (dashed red line) forms an a-helix. The C-terminal helix

THDP N-succinyltransferase undergoes a significant confor-

mational changes upon CoA and substrate binding (Fig. 3). In

the apo-structure (Fig 3A), the C-terminal helix blocks the

CoA binding site and engages many of residues such as

Ser211 that are involved in binding CoA. In the complex with

CoA and substrate, the C-terminus shifts and becomes more

ordered [6,7]. A key step is the electrostatic interaction of ly-

sine residues Lys259 and Lys263 with the pyrophosphate of

CoA. This displaces the C-terminal helix (residues 265–271),

nyltransferase in the unliganded state (A) and the succinyl-CoA andblue, other helices from the same subunit are red, b-sheets yellow andte are shown as space-filling models. Upon CoA (light green) binding,(dark blue) shifts and binds the substrate (light blue).

626 L. Nguyen et al. / FEBS Letters 582 (2008) 623–626

which then reverses orientation to bind the substrate at the

other end of the LbH domain (Fig. 3B). Additional structural

changes occur in a flexible loop (residues 166–175) of the LbH

domain which shifts to contact the substrate, helix a3 from the

N-terminal domain and the newly positioned C-terminal helix.

The dynamic nature of the C-terminal helix and preceding

loop (Val256-Arg261) is confirmed by B-factors twice those

of more rigid elements in the apo-structure.

These structures provide a mechanistic explanation for pre-

vious kinetic studies that have shown that THDP N-succinyl-

transferase functions through the sequential binding of

succinyl-CoA and substrates, tetrahydrodipicolinate or L-2-

aminopimelate (L-2-AP). With increasing concentrations of

succinyl-CoA, the apparent Km for L-2-AP decreases 2.4-fold

[1]. This occurs since CoA binding releases the C-terminal helix

and promotes formation of the L-2-AP binding pocket. Con-

versely, L-2-AP binding recruits the C-terminal helix to its

alternative position (Fig. 3B) and increases the affinity of the

enzyme for CoA binding [1].

Acknowledgements: We thank E. Zdanovich for technical assistance.This work was funded by a Canadian Institutes of Health Research(CIHR) Genomics Grant GSP-48370 for the Montreal-Kingston Bac-terial Structural Genomics Initiative (M. Cygler). K.G is a ChercheurNational of the Fonds de la recherche en sante de Quebec (FRSQ).Data acquisition at the Macromolecular Diffraction (MacCHESS)facility at the Cornell High Energy Synchrotron Source (CHESS) weresupported by the National Science Foundation award DMR 0225180and the National Institutes of Health award RR-01646.

References

[1] Berges, D.A., DeWolf Jr., W.E., Dunn, G.L., Newman, D.J.,Schmidt, S.J., Taggart, J.J. and Gilvarg, C. (1986) Studies on theactive site of succinyl-CoA:tetrahydrodipicolinate N-succinyl-transferase. Characterization using analogs of tetrahydrodipico-linate. J. Biol. Chem. 261, 6160–6167.

[2] Ishino, S., Mizukami, T., Yamaguchi, K., Katsumata, R. andAraki, K. (1987) Nucleotide sequence of the meso-diaminopime-late DD-dehydrogenase gene from Corynebacterium glutamicum.Nucl. Acids Res. 15, 3917.

[3] Sundharadas, G. and Gilvarg, C. (1967) Biosynthesis of alpha,epsilon-diaminopimelic acid in Bacillus megaterium. J. Biol.Chem. 242, 3983–3984.

[4] Beaman, T.W., Binder, D.A., Blanchard, J.S. and Roderick, S.L.(1997) Three-dimensional structure of tetrahydrodipicolinate N-succinyltransferase. Biochemistry 36, 489–494.

[5] Raetz, C.R. and Roderick, S.L. (1995) A left-handed parallel betahelix in the structure of UDP-N-acetylglucosamine acyltransfer-ase. Science 270, 997–1000.

[6] Beaman, T.W., Blanchard, J.S. and Roderick, S.L. (1998) Theconformational change and active site structure of tetrahydro-dipicolinate N-succinyltransferase. Biochemistry 37, 10363–10369.

[7] Beaman, T.W., Vogel, K.W., Drueckhammer, D.G., Blanchard,J.S. and Roderick, S.L. (2002) Acyl group specificity at the activesite of tetrahydridipicolinate N-succinyltransferase. Protein Sci.11, 974–979.

[8] Matthews, B.W. (1968) Solvent content of protein crystals. J.Mol. Biol. 33, 491–497.

[9] Otwinowski, Z. and Minor, W. (1997) Processing of X-raydiffraction data collected in oscillation mode. Meth. Enzymol.276, 307–326.

[10] Read, R.J. (2001) Pushing the boundaries of molecular replace-ment with maximum likelihood. Acta Crystallogr. D Biol.Crystallogr. 57, 1373–1382, (Epub 2001 Sep 21).

[11] McRee, D.E. (1999) XtalView/Xfit–A versatile program formanipulating atomic coordinates and electron density. J. Struct.Biol. 125, 156–165.

[12] Murshudov, G.N., Vagin, A.A., Lebedev, A., Wilson, K.S. andDodson, E.J. (1999) Efficient anisotropic refinement of macro-molecular structures using FFT. Acta Crystallogr. D Biol.Crystallogr. 55, 247–255, Epub 1999 Jan 1.

[13] Winn, M.D., Murshudov, G.N. and Papiz, M.Z. (2003) Macro-molecular TLS refinement in REFMAC at moderate resolutions.Meth. Enzymol. 374, 300–321.

[14] Laskowski, R.A., MacArthur, M.W., Moss, D.S. and Thornton,J.M. (1993) PROCHECK: a program to check the stereochemicalquality of protein structures. J. Appl. Crystallogr. 26, 283–291.