Crystal Structure of the Nucleotide-binding Domain of the ABC-transporter Haemolysin B:...

Crystal Structure of the Nucleotide-binding Domain ofthe ABC-transporter Haemolysin B: Identification of aVariable Region Within ABC Helical Domains

Lutz Schmitt1*, Houssain Benabdelhak2, Mark A. Blight2

I. Barry Holland2* and Milton T. Stubbs3

1Institut fur BiochemieBiozentrum N210, JohannWolfgang Goethe UniversitatFrankfurt, Marie-Curie Str. 960439 Frankfurt, Germany

2Institut de Genetique etMicrobiologie, Bat. 409Universite de Paris XI91405 Orsay, France

3Institut fur BiotechnologieMartin-Luther-UniversitatHalle WittenbergKurt-Mothes-Strabe 306120 Halle, Germany

The ABC-transporter haemolysin B is a central component of the secretionmachinery that translocates the toxin, haemolysin A, in a Sec-independentfashion across both membranes of E. coli. Here, we report the X-ray crystalstructure of the nucleotide-binding domain (NBD) of HlyB. The moleculeshares the common overall architecture of ABC-transporter NBDs. How-ever, the last three residues of the Walker A motif adopt a 310 helical con-formation, stabilized by a bound anion. In consequence, this results in anunusual interaction between the Walker A lysine residue and the WalkerB glutamate residue. As these residues are normally required to be avail-able for ATP binding, for catalysis and for dimer formation of ABCdomains, we suggest that this conformation may represent a latent mono-meric form of the NBD. Surprisingly, comparison of available NBD struc-tures revealed a structurally diverse region (SDR) of about 30 residueswithin the helical arm II domain, unique to each of the eight NBDs ana-lyzed. As this region interacts with the transmembrane part of ABC-trans-porters, the SDR helps to explain the selectivity and/or targeting ofdifferent NBDs to their cognate transmembrane domains.

q 2003 Published by Elsevier Science Ltd

Keywords: ABC-transporter; ATP-binding domain; X-ray structure;structural diversity; signaling domain*Corresponding authors

Introduction

Haemolysin B (HlyB) is an essential componentof the non-classical, type I secretion system1 forthe Escherichia coli RTX-toxin, HlyA.2 This 110 kDaprotein is transported in a Sec-independent fashionacross both membranes of E. coli. The haemolysinA (HlyA) transport machinery is composed of theATP-binding cassette (ABC) transporter HlyBlocated in the inner membrane, haemolysin D(HlyD), also anchored in the inner membrane, andTolC,3 which resides in the outer membrane. HlyD

apparently forms a continuous channel thatbridges the entire periplasm, interacting with TolCand HlyB. This arrangement prevents the appear-ance of periplasmic intermediates of HlyA duringsubstrate transport.4 Little is known about themolecular details of HlyA transport, but it isevident that ATP-hydrolysis by the ABC-trans-porter HlyB is a necessary source of energy. Atwhich stage of transport the energy is generatedor how many ATP molecules are consumed duringthe transport cycle is unknown.

ABC-transporters are ubiquitous, ATP-depen-dent transporters or pumps.5 Despite wide-rangingsubstrate specificity, ABC-transporters share acommon architecture. The basic blueprint consistsof four domains: two transmembrane domains(TMD) and two ATP-binding or nucleotide-bind-ing domains (NBD), in different combinations.Recent structural studies on NBDs from diverseABC-transporters have established the moleculararchitecture of this domain.6 – 11 This includes twoarchael ABC proteins,8,9 presumed to be importersbut for which no transport substrate has been

0022-2836/03/$ - see front matter q 2003 Published by Elsevier Science Ltd

E-mail addresses of the corresponding authors:[email protected]; [email protected]

Abbreviations used: ABC, ATP-binding cassette; CCD,charge-coupled device; CNS, crystallography and NMRsystem; Hly, haemolysin; rmsd, root-mean-squaredeviation; ICD, intracellular domain; MAD,multiwavelength anomalous diffraction; NBD,nucleotide-binding domain; PEG, polyethylene glycol;RTX, repeat in toxins; SDR, structurally diverse region;TMD, transmembrane domain.

doi:10.1016/S0022-2836(03)00592-8 J. Mol. Biol. (2003) 330, 333–342

identified. For a recent review and detailed discus-sion see Kerr.12 In addition, two structures of intactE. coli ABC-transporters, the lipid A exporterMsbA13 and the vitamin B12 importer BtuCD14

have been reported recently.The NBD is an L-shaped molecule divided into

two domains, termed arm I (catalytic domain) andarm II (signaling domain). The nucleotide bindingsite is located in arm I, which adopts an a?b fold.The second domain, arm II, is composed entirelyof a-helices and contains the C-loop or signaturemotif.15 Arms I and II are connected by the Q-loopand a second loop we term the Pro-loop. In thecrystal structure of the MsbA exporter,13 a thirddomain juxtaposes the NBD, the intercellulardomain (ICD), composed of the cytoplasmic loopsbetween the transmembrane helices. It has beensuggested that the ICD shuttles signals arisingfrom ATP or substrate binding between the NBDand TMD. In the vitamin B12 importer BtuCD,14

the analogous contact point with the NBD is thecytoplasmic or L-loop between TM-helix 6 andTM-helix 7 of BtuC, related to the EAA motif ofbacterial importers.16 This loop adopts an a-helicalstructure and interacts with the NBD of BtuD. Asthe structures of NBDs are very similar in theisolated state and as assembled in their mem-braneous environments, many of the mechanisticand structural principles governing ATPhydrolysis can be deduced from NBD structuresin the absence of their cognate TMD.

Despite the available three-dimensional infor-mation for NBDs, the inter- and intramolecular sig-naling mechanism involving the NBDs and TMDsremains an unsolved question. Since successfulreports of the capacity of one NBD to replaceanother have been restricted to extremely closerelatives, it seems likely that contacts between theNBD and TMD, necessary for signal transmissionand coordination of energy generation with thecognate transport substrate, must be highlyspecific. Earlier mutational studies indicated thatarm II is involved in the cross-talk with the TMDand might act as a signaling domain.17,18 This issupported by the finding that a chimeric NBDcomposed of the catalytic domain of HisP and thesignaling domain of MalK19 is able to transportmaltose but not histidine. Nevertheless, since theoverall fold and domain organization of NBDstructures appeared to be essentially identical, thebasis of this signaling specificity has hithertoremained obscure.

Here, we report the X-ray structure of thenucleotide-free NBD of HlyB (HlyB-NBD) at 2.6 Aresolution. A novel 310 helix in part of the WalkerA motif is observed, resulting in an interactionbetween Lys508 and Glu631 in the Walker A andWalker B motifs, respectively. This interaction mayexplain our observation that the HlyB NBD showsreduced binding of ATP at high ionic strength(approximately 10% at 100 mM KCl) and mightserve as a molecular switch to regulate the ATPaseactivity (Benabdelhak et al., in preparation). A

detailed comparison of eight different NBD struc-tures demonstrates that each contains a uniqueregion within the signaling domain. We designatethis a structurally diverse region (SDR). Theseresults are discussed in the light of inter-domainsignaling or fine-tuning of the individual NBDs ofABC-transporters to meet the needs of particulartransported substrates.

Results and discussion

Structure determination

The C-terminal NBD of the E. coli ABC-trans-porter HlyB (HlyB-NBD, residues 467–707) wasexpressed, purified as described as in Materialsand Methods, and crystals were obtained in spacegroup P41212 with one monomer in the asymmetricunit.20 The structure was solved using multiwave-length anomalous diffraction of a single SeMet-substituted crystal (see Materials and Methodsand Table 1). The final model with an Rfree valueof 26.0% at 2.6 A resolution consists of residues467–707, one sulphate ion, and 130 watermolecules, with no visible density for theN-terminal His tag. A sulphate or phosphate ion,with a peak height of 6.4s in a difference electrondensity map, is located at the presumed positionof the b phosphate of ATP or ADP. Since crystalli-zation conditions contained 300 mM ammoniumsulphate but only 10 mM phosphate, the ion ismore likely to be sulphate. The dimer interfacebetween symmetry-related pairs of protein mol-ecules in the crystal is rather small and exhibits nosimilarity to any of the previously describeddimer interfaces of NBDs.6,7,11,14 We thereforebelieve that our structure corresponds to a mono-meric species.

Overall fold

As first observed for HisP,6 the NBD of HlyBadopts an L-shaped conformation with a two-domain architecture (Figure 1). The first domain(arm I) consists of two b-sheets and sevena-helices, and adopts a fold most similar to that ofRecA.21 The topology is depicted in Figure 2; anno-tation of strands and helices follows that of HisP.6

Sheet 1 consists of four antiparallel strands thatpack against helix 1. This helix contains the lastthree C-terminal residues of the Walker A motif(residues 502–510). Surprisingly, these residuesform a 310 helix before continuing on into a regulara-helix. This is in marked contrast to other struc-tures of ABC NBDs, where these three residuesadopt a regular a-helical conformation at the startof helix 1. Sheet 2 is formed of four parallelstrands, interrupted by two helices (6 and 70), andtwo additional parallel strands. Helices 8 and 9are located at the C-terminal end of the protein.The conserved motifs, the Walker A and Walker B(residues 626–631), the D-loop (residues 634–637),

334 Structural Plasticity within NBDs of ABC-transporters

switch II (residues 656–663), and the Gly-loop(residues 678–680) are all located within arm I(see Figure 2). The Gly-loop contains the highlyconserved Gly679, found in all NBDs. Arms I andII are connected from b-strand 8 to helix 3 via the

Q-loop (residues 550–558) and by the loop contain-ing Pro624 (the Pro-loop). Several substitutions ofthis Pro residue have a severe impact on the ther-mal stability and amount of overexpression of theHlyB-NBD (our unpublished results). Moreover, a

Table 1. Summary of the crystallography data

A. Crystal parametersSpace group P41212Cell constants at 100 K

SeMet Native

a, b (A) 104.12 105.2c (A) 124.80 125.4a, b, g (deg.) 90 90

B. Data collection and processingNative Inflection Maximum Remote

Wavelength (A) 1.005 0.9778 0.9773 0.95Resolution (A) 20–2.6 20–3.2 20–3.2 20–3.2Mean redundancy 5.2 6.9 6.9 6.9Completenessa (%) 97.3 (99.3) 98.1 (99.8) 98.0 (99.8) 98.0 (99.7)I/sa 19.8 (1.9) 23.4 (5.4) 23.0 (5.7) 25.1 (5.3)Rsym

a 7.6 (34.9) 7.0 (29.3) 7.0 (28.2) 6.7 (31.3)

C. RefinementRF

a (%) 23.4Rfree

a (%) 26.0rmsd bond length (A) 0.007rmsd bond angles (deg.) 1.4Average B-factor (A2) 54.2

Ramachandran plotb

Most favored (%) 87.3Allowed 12.2Generously allowed (%) 0.5

D. Model contentProtein residues 467–707Ligands One sulfateWater molecules 130

Rsym ¼,P

h0 lIh0 2 Ihl . =ðP

h0 IhÞ: RF ¼P

kFol2 lFck=ðP

lFolÞ; where Fc is the calculated structure factor. Rfree is as RF but calculatedfor 10% randomly chosen reflections that were omitted from the refinement procedure.

a Values in parentheses correspond to the last resolution shell (3.25–3.2 A for SeMet and 2.65–2.6 A for native HlyB-NBD).b The analysis was performed using PROCHECK.45

Figure 1. Stereo representation of the three-dimensional structure of the HlyB-NBD. Helices are shown in red andgreen, b-strands in blue, and loops in yellow. The sulphate ion is shown in ball-and-stick representation. Helices inarm I are colored in red, those located in arm II are colored in green. The C-loop is located in helix 5: the precedingresidues 578–605 are shown in grey and correspond to the SDR (see text).

Structural Plasticity within NBDs of ABC-transporters 335

Pro624Leu mutation makes the secretion of HlyAtemperature-sensitive in vivo22 and the correspond-ing mutation in HisP renders the ATPase con-stitutively active in vitro.23 These results imply animportant role for Pro624 in the structure and func-tion of this second connecting loop. The smallerdomain, arm II, also termed the signaling domain,is composed of five a-helices and contains the

C-loop or signature motif (residues 607–613),24

which extends into helix 5.

Architecture of the ligand-binding site

A sulphate or phosphate ion is located at thenucleotide-binding site in the HlyB-NBD (Figures 1and 3). Superposition of the individual NBDsreveals the anion to be in a location equivalent tothat of the b-phosphate group in ATP. In adenylatekinase, a sulphate ion at the b-phosphate positionhas been observed in the nucleotide-free state ofthe enzyme.25,26 Furthermore, it has been shownthat either free sulphate or phosphate can mimicthe b-phosphate ion of ATP, thereby stabilizingthe loop containing the Walker A motif.27 In theHlyB-NBD, the anion interacts with the aminoacid residues of the Walker A motif via hydrogenbonds to the main-chain amide nitrogen atoms ofGly505, Ser506, Gly507 and Ser509, and to theoxygen side-chain atoms of Ser506 and Ser509. Inaddition, two water molecules form hydrogenbonds to the sulphate ion.

In contrast to other published structures ofNBDs, in which Lys508 of the Walker A motif con-tacts the b-phosphate group of ATP or ADP, thisresidue is not available for ligand binding in thepresent structure (Figure 3). Instead, it forms anion pair with the carboxyl group of Glu631, thefinal amino acid residue of the Walker B motif.This interaction is mediated by an additionalbridging water molecule. In other NBDs, although

Figure 2. Topology of the HlyB-NBD. Color-coding asin Figure 1. The SDR shown in grey in Figure 1 hasbeen omitted. Conserved motifs are shown in black andlabeled. In our topology diagram, the Walker B motif isextended by the glutamate residue C-terminal to the con-served aspartate residue.

Figure 3. Stereo representation of the Walker A motif and interaction with residues of the HlyB-NBD. The Walker Amotif is shown in grey (residues 502–507) including side-chains. The continuing a-helix (residues 511–516) is shown inred. The side-chains of Asp630 and Glu631 and the backbone of the Walker B motif are shown. Color-coding as inFigure 1. The sulphate ion is shown in stick representation. The interaction between Lys508 and Glu631 is highlightedby a broken line. For simplicity, water molecules have been omitted. For comparison, the Walker A motif, helix 1,Walker B motif, and the corresponding side-chains of the superposed are shown in green. The interaction betweenAsp170 and Ser45 of MJ0796 is highlighted by a broken line.


often crystallized under similar high-salt con-ditions, the equivalent of Glu631 does not contactany residue of the Walker A sequences; instead,the preceding Asp residue hydrogen bonds to theside-chain of Ser509. In the present structure,however, Ser509 coordinates the bound anion(see above). In common with other NBDs,however, Gly679 interacts with Ser506 of theWalker A motif.

Implications of the Lys508–Glu631 interaction

Replacement of the equivalent of the Walker BGlu631 by Gln in the Methanococcus janaschii 0796and 1276 ABC proteins results in mutant proteins(E171Q) that form stable dimers in the presence ofATP, but are unable to hydrolyse ATP.28 This isin contrast to the wild-type protein, as in the caseof the HlyB-NBD, where only transient dimer

Figure 4. Sequence and secondary structure alignment. Primary structure of exporters HlyB-NBD (haemolysin A;this study), hTAP1-NBD (peptides; 1JJ7), MsbA (lipid A; 1JSQ), and importers HisP (histidine; 1B0U), MalK (maltose;1G29, chain A), MJ0796 (1F3O), MJ1267 (1G6H (ADP) and 1GAJ (SO4)), and BtuD (vitamin B12; 1L7V) were alignedusing ClustalW (www.ebi.ac.uk/clustalw). The transport substrate, were known unambiguously are indicated inparentheses together with the PDB entry). Conserved motifs are underlined and labeled. Secondary structureelements were determined from the corresponding PDB entries using PROCHECK.45 Helices are indicated by the redsymbols and sheets by the blue arrows. Core residues are indicated by an asterisk. Amino acids located in arm II areboxed.


http://www.ebi.ac.uk/clustalw

formation is observed (in preparation). In the crys-tal structure of MJ0796 (E171Q),11 a dimer interfaceis observed similar to that in the BtuCD14 andRad5029 structures. Moody et al. therefore proposedthat this structure, where the C-loop of monomer 1participates in the dimer interface and contactsATP in the second monomer, represents the phys-iological dimer.28 In the structure of the humanTAP1-NBD,10 the residue corresponding to Glu631(Asp667) coordinates a magnesium ion that isessential for ATP hydrolysis. Thus, the equivalentsof residues Lys508 and Glu631 are of extremeimportance for nucleotide binding, inter-monomerassociation and ATP-hydrolysis. In the presentHlyB-NBD crystal structure, these residues aremoved away from any potential ligand contactand are stabilized by interaction with one another(Figure 3).

This organisation may represent a stable “latent”form of the NBD that is not immediately availablefor ligand binding, and could explain the reducedATPase activity and ATP binding of HlyB at phys-iological salt concentrations that we have observed.The ATPase activity of HlyB-NBD is sensitive toionic strength: At salt concentrations above300 mM (NaCl or KCl), no steady-state ATPhydrolysis is detectable. Binding of ATP is corre-spondingly reduced under these conditions. How-ever, these effects are completely reversible whenthe concentration of salt is reduced. Nevertheless,the observed interaction between Lys508 andGlu631 in the HlyB-NBD is in all probability aresult of the 310-helical conformation describedabove. We speculate that the ability to switch from

the 310 to the normal a-helical form may constitutean important physiological control of ATPaseactivity. A similar 310 to a-helical transition hasbeen described for lactate dehydrogenase uponligand binding and has been implicated in theregulation of protein function.30

Comparison of the secondary structuresof NBDs

In contrast to the catalytic domain, the helicaldomain of HlyB appeared to show an obviousdifference from, for example, HisP. Therefore, inorder to examine systematically possible variationsin the helical domains of the NBDs of ABCproteins, we first aligned the secondary structuresof the NBDs of all the known three-dimensionalstructures (Figure 4). This confirmed the extensiveconservation of structural features in the NBD. Incontrast, however, the helical region between helix2 and helix 5 (C-loop) showed considerable vari-ability between different NBDs (highlighted as abox in Figure 4). The secondary structural featuresshowing the closest similarity in arm II are thepeptide transporter hTAP1-NBD and the HlyB-NBD. In arm II of the ABC domain of MsbA andBtuD, unique patterns of secondary structure arepresent. In the case of MsbA, although an exporterlike TAP and HlyB, helix 4 is shortened by sixamino acid residues, helix 40 is extended by fiveresidues, whilst helix 400 is completely absent. InBtuD, which energizes the import of vitamin B12, astretch of 12 amino acid residues usuallyencompassing helix 40 is missing.

Table 2. Summary of the structurally diverse region of nucleotide-binding domains

HlyB-NBD HisP MalK MJ0796 MJ1267 TAP1-NBD BtuD

HlyB – 536–542,577–613

566–613 576–614 578–613 577–579,582–593

577–583,588–591

NBD 610–619 614–616HisP 73–86,

126–153– 71–85,

107–11071–85,

206–21171–85,

104–11071–85,

127–15373–86,

127–142120–133,145–156

145–153 154–156

MalK 100–138 68–75,94–97

– 94–120,123–135

69–75,91–123

110–137 70–76,110–124

108–122,123–136

134–136

MJ0796 109–143 70–77,199–204

94–129,136–145

– 70–77,93–101

109–143 111–133,144–146

136–145MJ1267 128–151 73–79,

92–9874–79,92–134

74–79,92–100

– 108–159 110–143,152–154

144–151 144–154TAP1 615–618,

621–631570–577,616–639

616–645 614–639 614–648 – 615–622,627–630

NBD 636–647 642–644BtuD 108–116,

121–12367–72,

109–12392–97,

109–123109–123,125–127

109–123,125–127

108–116,120–123

–

125–127 125–127 125–127 125–127

Summary of the SDR of NBDs as determined from the analysis of rmsd values, where Walker A/Walker B and layer 1 were used.From left to right, numbering of the amino acid residues refers to the NBD given in column 1. Residues that fall into the SDR are high-lighted in bold. Note, as in Figure 4 the malK sequence is from Thermococcus litoralis.


Proposed functional role of the helical domain

The studies of MalK-HisP chimeras described bySchneider & Walter indicated clearly that transportsubstrate identity is conferred by specific recog-nition of the TMD by the NBD.19 On the otherhand, biochemical and genetic studies of HisP andMalK have demonstrated that a sub-region of thehelical domain is involved in specific and func-tional contacts between the TMDs and theNBDs.17,18,31 Interestingly, this sub-region overlapswith the region of variability described in theprevious section. Thus, residues located in helices3, 4, and 40, of the helical domain of MalK havebeen identified17,18,31 that interact with the highlyconserved EAA motif16 in a cytoplasmic loop ofthe TMD. Moreover, the interface between theNBD and TMD in the structure of BtuCD (nonucleotide present)14 includes residues that arelocated in the Q-loop, helix 3, and helix 4 of theNBD. Finally, in HisP, the loops connecting arm Iwith arm II (Q-loop and Pro-loop) are probablyinvolved in relaying conformational informationbetween the helical and the catalytic domains.

Analysis of the helical domain reveals astructurally diverse region

The studies summarized above indicated that aregion in the helical domain in each ABC transporter

may be different. Therefore, we aligned all the pub-lished structures of NBDs, using layer 1 of the cataly-tic domain as the fixed point for superposition. Theresults indicated large structural variations within asub-region of arm II (Table 2). In Figure 5(A), a super-position of the structures of HlyB-NBD/SO4 andMJ1276/SO4 demonstrates that the helical domainsof both proteins show substantial structural diver-sity. On the other hand, the stereo representation ofthe superimposition of the structures of MJ1276/SO4 and MJ1276/ADP demonstrates the similarityof the helical domain in these two functional states(Figure 5(B)). Therefore, the influence of ligands orthe functional state of the NBD can be ruled out asan explanation of these differences, since the coreresidues of the MJ1267/sulphate and MJ1267/ADPstructures show no significant increase in rmsdvalues. The cross-comparison of all NBDs is sum-marized in Table 2 and supports the observationthat a region of structural diversity (SDR) existswithin a sub-region of arm II. As indicated above,the detection of the SDR is independent of the func-tional state of the NBD, e.g. nucleotide-free, ATP-bound or ADP-bound. The region in MalK detectedby the comparison (residues 566–613; Table 2) coversall the residues that have been identified in thepast as being involved in NBD-TMDintercommunication.17,18 In HlyB, the core of theSDR corresponds to residues 578–605 (see Figure 1).

It is evident from the work of Hunt and

Figure 5. (A) Stereo representation of superposition of the structure of MJ1267/SO4 complex with the structure ofHlyB-NBD/SO4 using layer 1. For the sake of clarity, only the catalytic domain of HlyB-NBD/SO4 is shown (lightyellow). The helical domain of HlyB-NBD/SO4 MJ1267/SO4 is shown in yellow and the helical domain of MJ1267/SO4 in green in a cartoon-like representation. (B) Stereo representation of the superimposition of the MJ1267/sulphatestructure with the MJ1267/ADP structure using layer 1. Only the catalytic domain of MJ1267/SO4 (light green) isshown. The helical domain of MJ1267/SO4 (green) and MJ1267/ADP (blue) are shown in a cartoon-like representation.


co-workers9 that rigid body movements of arm IIwith respect to arm I are associated with changesin the functional state of the ligand binding site,e.g. bound ATP, nucleotide-free state or ADPbound.9,11 However, the SDR is fundamentallydifferent from the rotational movement of arm IIthat is associated with ligand-binding andcatalysis. Thus, this rigid-body movement of armII towards arm I in the presence of different ligandsof a given ABC NBD was still detectable in ouranalysis, whilst the structure of the SDR remainedthe same (see Figure 5(b)).

As the SDR is currently based on a limited numberof high-resolution structures of NBDs, the study ofadditional examples, will be an important test of thehypothesis. However, of the eight structures ana-lysed so far, at least six are known to transport differ-ent molecules, so that we propose the intriguingpossibility that the SDR represents a specificityregion24 that controls targeting of NBDs to their cog-nate TMDs. In support of this, a HisP-MalK chimeracomposed of the catalytic domain of HisP and thesignaling domain of MalK energizes the import ofmaltose but not the import of histidine.19 In this con-nection, it is noteworthy that the SDR is sandwichedbetween two conserved motifs, the Q-loop and theC-loop, regions that are involved in TMD contact14

and ATP-sensing, respectively.11,32 Such anarrangement is reminiscent of GTPases33 and Ras-effector proteins,34 where a “bipartite recognitionprocess” has been postulated,34 with one part of theenzyme conserved to sense the functional state ofthe GTPase whilst the other part confers specificity.

Conclusions

We suggest that the observed 310 helix within theWalker A motif might act as a reversible, molecularswitch to regulate the ATPase activity of HlyB. Fur-thermore, we propose that arm II includes a speci-ficity module for the NBD-TMD interaction. If thisis confirmed by more structures, one might envi-sage that the SDR identified here could serve as astarting point for rational design of drugs to modu-late the activity of individual ABC transporters.Until now, only the TMDs of ABC-transporterswere thought to encode substrate specificity, anassumption stemming originally from the obvioussequence diversity of these domains or subunitscompared to the NBDs. Therefore, these domainshave been presumed to be the best targets in thesearch of specific inhibitors. However, theobserved SDR encoded in NBDs of ABC-transpor-ters described here should open new avenues fortargets for drug research.

Materials and Methods

Purification and ATPase activity of the HlyB-NBD

The HlyB-NBD (residues 467–707) was overexpressed

as an N-terminal His6-tagged fusion protein in E. coli andpurified as described elsewhere. Briefly, the NBD wascloned under the control of an arabinose-inducible pro-motor in pBAD1835 by PCR using an Nde I-EcoR I frag-ment. Plasmid pLG57036 was used as DNA template.After transformation of E. coli strain DH5a, cultureswere grown at 25C and protein production was inducedat A600 ¼ 2.0 with 0.01% (w/v) L(þ)-arabinose. Afterthree hours, cells were harvested by centrifugation andbroken by ultrasonication. The clarified lysate wasloaded onto a 5 ml HiTrap chelating column and theHlyB-NBD was eluted with a linear imidazole gradient(10–300 mM imidazole). The HlyB-NBD fractions werepooled, concentrated by ultrafiltration and purifiedfrom a Superdex S200 column by elution with 10 mMsodium phosphate (pH 8.0), 100 mM KCl. The HlyB-NBD with a purity greater than 99% (as judged from asilver-stained SDS/polyacrylamide gel), was concen-trated immediately to 10 mg/ml by ultrafiltration.Protein concentration was determined spectroscopicallyat 280 nm assuming a molar extinction coefficient of15,400 M21 cm21. The final yield of HlyB-NBD wasaround 10 mg/l of E. coli culture. The NBD behaved asa monomer in gel-filtration at all salt concentrations andin the presence of ATP. Details of the ATPase activity ofpurified HlyB-NBD, measured by a continuous ATP-regenerating assay,37 will be given elsewhere.

Crystallization

Crystals of the HlyB-NBD were obtained by the hang-ing-drop, vapor-diffusion technique at 277 K asdescribed.20 In brief, 2 ml of 10 mg/ml of HlyB-NBD in10 mM sodium phosphate (pH 8.0), 100 mM KCl pre-incubated with 50 mM ATP (pH 7.0) (30 minutes at277 K), were mixed with 2 ml of 150 mM diamino aceticacid (pH 6.2), 16% (w/v) PEG 8000, 300 mM ammoniumsulphate. Crystals, which formed within two days andgrew to their final dimensions within two weeks, weretransferred directly into cryo-buffer (150 mM diaminoacetic acid (pH 6.2), 25% PEG 8000, 80 mM ammoniumsulfate, 15% (w/v) ethylene glycol) and flash-frozen inliquid nitrogen. For seleno-methionine (SeMet) substi-tution, cells were grown in glycerol minimal mediumsupplemented with the appropriate amino acids andSeMet,38 and the protein was purified as describedabove. Crystallization of the SeMet-substituted proteinwas achieved using exactly the same conditions asabove but replacing PEG 8000 by 14% PEG 5500-mono-methylether. However, only small, thin needles of theSeMet-substituted protein could be obtained. Crystalssuitable for X-ray analysis of the SeMet-substitutedprotein and crystals of the wild-type protein, whichdiffracted to higher resolution, were obtained only afteraddition of 0.15–0.3% (w/v) low-melting agarose to thecrystallization drops.39 Crystals of the SeMet-substitutedprotein were transferred directly into cryo-buffer (seeabove) and flash-frozen in liquid nitrogen.

Data collection, structure determinationand refinement

Data from wild-type and SeMet-substituted crystalswere collected at beamline BW-6 (DESY, Hamburg) witha MAR CCD. All data were processed using DENZOand SCALEPACK.40 An initial electron density wasobtained from a three-wavelength MAD data set(Table 1) using SOLVE,41 which found three of the five


expected SeMet sites. The N-terminal SeMet residue, forwhich no electron density was visible, and SeMet646,which is in close proximity to SeMet648 were notdetected. After solvent flattening and initial model build-ing with O,42 CNS43 was used in the subsequent refine-ment steps. After application of an overall anisotropicB-value, and bulk solvent correction, the structure wasrefined to 2.6 A by alternating cycles of manual buildinginto 2Fobs 2 Fcal electron densities and crystallographicrefinement. Automated water picking was performedusing CNS with a cut-off of 4s for individual watermolecules that were checked manually for appropriatedensity. In all, 10% of the data were excluded fromrefinement to calculate the free R value for cross-validation.44 Although wild-type and SeMet-substitutedcrystals were obtained in the presence of ATP, no nucleo-tide was detectable in the electron density. The absenceof nucleotide in the ligand-binding pocket may not besurprising, since greatly reduced steady-state binding ofATP to the HlyB NBD is observed above ionic strengthsof 300 mM salt (see above). The quality of the modelwas analyzed using PROCHECK.45 Crystallographicstatistics are given in Table 1.

Comparison with other NBDs

For structural comparison of the published NBDs andABC-transporters with the HlyB-NBD, individual pro-teins were aligned using LSQMAN46 by explicitly super-positioning either the residues of the Walker A and Bmotifs or layer 1. Layer 1 consists of b-strands 1, 2 and4, and the C-terminal part of helix 1, excluding the lastthree residues of the Walker A motif (see Figure 2). Onthe basis of the Walker A/Walker B and layer 1 align-ment, “core residues” were determined. These weredefined as residues that contain a structural counterpartin all NBDs excluding MsbA and BtuD. The latter twowere excluded from the core residue determination,since no electron density is visible for the N-terminalpart of the NBD of MsbA,13 and in the case of BtuD, theNBD of the vitamin B12 importer, there is a deletion of12 amino acid residues preceding the C-loop.14

Figure preparation

Structure Figures were prepared using PYMOL:(http://pymol.sourceforge.net/).

Protein Data Bank accession code

Coordinates have been deposited in the pdb underaccession code 1MT0.

Acknowledgements

We thank Laszlo Kranitz and Carsten Horn fortheir assistance in crystallization trials, and Chrisvan der Does for critical reading of the manuscript.We are indebted to the staff of BW6 (DESY,Hamburg) especially Gleb Bourenkov for help indata collection, Gerhard Klebe for in-house facili-ties and Robert Tampe for constant encourage-ment. We gratefully acknowledge the support (toI.B.H.) of CNRS and Universite de Paris-Sud and,

in particular, the generous support of ABCF(Association de lutte contre la Mucoviscidose).H.B. acknowledges the support of the FRM(Fondation pour la Recherche Medicale) and theSociete de Secours des Amis des Science. Thiswork was supported by the Deutsche Forschungs-gemeinschaft (Emmy Noether Programm, grantSchm1279/2-2 to L.S.).

References

1. Thanassi, D. G. & Hultgren, S. J. (2000). Multiplepathways allow protein secretion across the bacterialouter membrane. Curr. Opin. Cell Biol. 12, 420–430.

2. Blight, M. A., Chervaux, C. & Holland, I. B. (1994).Protein secretion pathway in Escherichia coli. Curr.Opin. Biotechnol. 5, 468–474.

3. Koronakis, V., Sharff, A., Koronakis, E., Luisi, B. &Hughes, C. (2000). Crystal structure of the bacterialmembrane protein TolC central to multidrug effluxand protein export. Nature, 405, 914–919.

4. Blight, M. A. & Holland, I. B. (1994). Heterologousprotein secretion and the versatile Escherichia colihaemolysin translocator. Trends Biotechnol. 12,450–455.

5. Higgins, C. F. (1992). ABC transporters: from micro-organisms to man. Annu. Rev. Cell Biol. 8, 67–113.

6. Hung, L. W., Wang, I. X. Y., Nikaido, K., Liu, P. Q.,Ames, G. F. L. & Kim, S. H. (1998). Crystal structureof the ATP-binding subunit of an ABC transporter.Nature, 396, 703–707.

7. Diederichs, K., Diez, J., Greller, G., Muller, C., Breed,J., Schnell, C. et al. (2000). Crystal structure of MalK,the ATPase subunit of the trehalose/maltose ABCtransporter of the archaeon thermococcus litoralis.EMBO J. 19, 5951–5961.

8. Yuan, Y. R., Blecker, S., Martsinkevich, O., Millen, L.,Thomas, P. J. & Hunt, J. F. (2001). The crystal struc-ture of the MJ0796 ATP-binding cassette. Impli-cations for the structural consequences of ATPhydrolysis in the active site of an ABC transporter.J. Biol. Chem. 276, 32313–32321.

9. Karpowich, N., Martsinkevich, O., Millen, L., Yuan,Y. R., Dai, P. L., MacVey, K. et al. (2001). Crystal struc-tures of the MJ1267 ATP binding cassette reveal aninduced-fit effect at the ATPase active site of anABC transporter. Structure, 9, 571–586.

10. Gaudet, R. & Wiley, D. C. (2001). Structure of theABC ATPase domain of human TAP1, the trans-porter associated with antigen processing. EMBO J.20, 4964–4972.

11. Smith, P. C., Karpowich, N., Millen, L., Moody, J. E.,Rosen, J., Thomas, P. J. & Hunt, J. F. (2002). ATP bind-ing to the motor domain from an ABC transporterdrives formation of a nucleotide sandwich dimer.Mol. Cell, 10, 139–149.

12. Kerr, I. D. (2002). Structure and association of ATP-binding cassette transporter nucleotide-bindingdomains. Biochim. Biophys. Acta, 1561, 47–64.

13. Chang, G. & Roth, C. B. (2001). Structure of MsbAfrom E. coli: a homolog of the multidrug resistanceATP binding cassette (ABC) transporters. Science,293, 1793–1800.

14. Locher, K. P., Lee, A. T. & Rees, D. C. (2002). TheE. coli BtuCD structure: a framework for ABC trans-porter architecture and mechanism. Science, 296,1091–1098.


http://pymol.sourceforge.net/

15. Schmitt, L. & Tampe, R. (2002). Structure and mech-anism of ABC transporters. Curr. Opin. Struct. Biol.12, 754–760.

16. Mourez, M., Hofnung, M. & Dassa, E. (1997). Subunitinteractions in ABC transporters: a conservedsequence in hydrophobic membrane proteins ofperiplasmic permeases defines an important site ofinteraction with the ATPase subunits. EMBO J. 16,3066–3077.

17. Schmees, G., Stein, A., Hunke, S., Landmesser, H. &Schneider, E. (1999). Functional consequences ofmutations in the conserved “signature sequence” ofthe ATP-binding-cassette protein MalK. Eur.J. Biochem. 266, 420–430.

18. Hunke, S., Mourez, M., Jehanno, M., Dassa, E. &Schneider, E. (2000). ATP modulates subunit–sub-unit interactions in an ATP-binding cassette trans-porter (MalFGK2) determined by site-directedchemical cross-linking. J. Biol. Chem. 275,15526–15534.

19. Schneider, E. & Walter, C. (1991). A chimeric nucleo-tide-binding protein, encoded by a hisP-malK hybridgene, is functional in maltose transport in Salmonellatyphimurium. Mol. Microbiol. 5, 1375–1383.

20. Kranitz, L., Benabdelhak, H., Horn, C., Blight, M. A.,Holland, I. B. & Schmitt, L. (2002). Crystallizationand preliminary X-ray analysis of the ABC-domainof the ABC-transporter HlyB from E. coli. ActaCrystallog. sect. D, 58, 539–541.

21. Story, R. M. & Steitz, T. A. (1992). Structure of therecA protein–ADP complex. Nature, 355, 374–376.

22. Blight, M. A., Pimenta, A. L., Lazzaroni, J. C., Dando,C., Kotelevets, L., Seror, S. J. & Holland, I. B. (1994).Identification and preliminary characterization oftemperature-sensitive mutations affecting HlyB, thetranslocator required for the secretion of haemolysin(HlyA) from Escherichia coli. Mol. Gen. Genet. 245,431–440.

23. Petronilli, V. & Ames, G. F. (1991). Binding protein-independent histidine permease mutants.Uncoupling of ATP hydrolysis from transmembranesignaling. J. Biol. Chem. 266, 16293–16296.

24. Holland, I. B. & Blight, M. A. (1999). ABC-ATPases,adaptable energy generators fuelling transmembranemovement of a variety of molecules in organismsfrom bacteria to humans. J. Mol. Biol. 293, 381–399.

25. Dreusicke, D. & Schulz, G. E. (1986). The glycine-richloop of adenylate kinase forms a giant anion hole.FEBS Letters, 208, 301–304.

26. Dreusicke, D., Karplus, P. A. & Schulz, G. E. (1988).Refined structure of porcine cytosolic adenylatekinase at 2.1 A resolution. J. Mol. Biol. 199, 359–371.

27. Vetter, I. R. & Wittinghofer, A. (1999). Nucleoside tri-phosphate-binding proteins: different scaffolds toachieve phosphoryl transfer. Quart. Rev. Biophys. 32,1–56.

28. Moody, J. E., Millen, L., Binns, D., Hunt, J. F. &Thomas, P. J. (2002). Cooperative, ATP-dependentassociation of the nucleotide binding cassettesduring the catalytic cycle of ATP-binding cassettetransporters. J. Biol. Chem. 277, 21111–21124.

29. Hopfner, K. P., Karcher, A., Shin, D. S., Craig, L.,Arthur, L. M., Carney, J. P. & Tainer, J. A. (2000).Structural biology of Rad50 ATPase: ATP-driven con-formational control in DNA double-strand break

repair and the ABC-ATPase superfamily. Cell, 101,789–800.

30. Gerstein, M. & Chothia, C. (1991). Analysis of proteinloop closure. Two types of hinges produce one motionin lactate dehydrogenase. J. Mol. Biol. 220, 133–149.

31. Mimura, C. S., Holbrook, S. R. & Ames, G. F.-L.(1991). Structural model of the nucleotide-bindingconserved component of periplasmatic permeases.Proc. Natl Acad. Sci. USA, 88, 84–88.

32. Fetch, E. E. & Davidson, A. L. (2002). Vanadate-catalyzed photocleavage of the signature motif of anATP-binding cassette (ABC) transporter. Proc. NatlAcad. Sci. USA, 99, 9685–9690.

33. Vetter, I. R. & Wittinghofer, A. (2001). The guaninenucleotide-binding switch in three dimensions.Science, 294, 1299–1304.

34. Herrmann, C. (2003). Ras-effector interactions: afterone decade. Curr. Opin. Struct. Biol. 13, 122–129.

35. Guzman, L.-M., Belin, D., Carson, M. J. & Beckwith,J. (1995). Tight regulation, modulation, and high-level expression by vectors containing the arabinosepBAD promoter. J. Bacteriol. 177, 4121–4130.

36. Mackman, N., Nicaud, J. M., Gray, L. & Holland, I. B.(1985). Genetical and functional organisation of theEscherichia coli haemolysin determinant 2001. Mol.Gen. Genet. 201, 282–288.

37. Senior, A. E., al-Shawi, M. K. & Urbatsch, I. L. (1998).ATPase activity of Chinese hamster P-glycoprotein.Methods Enzymol. 292, 514–523.

38. Doublie, S. (1997). Preparation of selenomethionylproteins for phase determination. Methods Enzymol.276, 523–530.

39. Zhu, D. W., Lorber, B., Sauter, C., Ng, J. D., Benas, P.,Le Grimellec, C. & Giege, R. (2001). Growth kinetics,diffraction properties and effect of agarose onthe stability of a novel crystal form of Thermusthermophilus aspartyl-tRNA synthetase-1. ActaCrystallog. sect. D, 57, 552–558.

40. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.In Methods in Enzymology (Carter, C. W. & Sweet,R. M., eds), Vol. 276, Academic Press, London.

41. Terwilliger, T. C. & Berendzen, J. (1999). AutomatedMAD and MIR structure solution. Acta Crystallog.sect. D, 55, 849–861.

42. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M.(1991). Improved methods for binding protein modelsin electron density maps and the location of errors inthese models. Acta Crystallog. sect. A, 47, 110–119.

43. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).Crystallography and NMR system: a new softwaresuite for macromolecular structure determination.Acta Crystallog. sect. D, 54, 905–921.

44. Brunger, A. T. (1992). Free R value: a novel statisticalquantity for assessing the accuracy of crystal struc-tures. Nature, 355, 472–475.

45. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein struc-tures. J. Appl. Crystallog. 26, 283–291.

46. Kleywegt, G. J. (1996). Use of non-crystallographicsymmetry in protein structure refinement. ActaCrystallog. sect. D, 52, 842–857.

Edited by R. Huber

(Received 26 February 2003; received in revised form 1 May 2003; accepted 2 May 2003)


Crystal Structure of the Nucleotide-binding Domain of the ABC-transporter Haemolysin B:...

Documents

Transcript of Crystal Structure of the Nucleotide-binding Domain of the ABC-transporter Haemolysin B:...