Molecular Cell 21, 109–122, January 6, 2006 ª2006 Elsevier Inc. DOI 10.1016/j.molcel.2005.10.034

Molecular Basis for G Protein Controlof the Prokaryotic ATP Sulfurylase

Joseph D. Mougous,1,2,4 Dong H. Lee,1,3

Sarah C. Hubbard,1,3 Michael W. Schelle,1,3

David J. Vocadlo,3,5 James M. Berger,2,*

and Carolyn R. Bertozzi1,2,3,*1Howard Hughes Medical Institute2Department of Molecular and Cell Biology3Department of ChemistryUniversity of California, BerkeleyBerkeley, California 94720

Summary

Sulfate assimilation is a critical component of both pri-mary and secondary metabolism. An essential step in

this pathway is the activation of sulfate through adenyl-

ation by the enzyme ATP sulfurylase (ATPS), formingadenosine 50-phosphosulfate (APS). Proteobacterial

ATPS overcomes this energetically unfavorable reac-tion by associating with a regulatory G protein, cou-

pling the energy of GTP hydrolysis to APS formation.To discover the molecular basis of this unusual role

for a G protein, we biochemically characterized andsolved the X-ray crystal structure of a complex be-

tween Pseudomonas syringae ATPS (CysD) and its as-sociated regulatory G protein (CysN). The structure of

CysN�D shows the two proteins in tight association;however, the nucleotides bound to each subunit are

spatially segregated. We provide evidence that con-served switch motifs in the G domain of CysN alloste-

rically mediate interactions between the nucleotidebinding sites. This structure suggests a molecular

mechanism by which conserved G domain architec-ture is used to energetically link GTP turnover to the

production of an essential metabolite.

Introduction

G proteins are a mechanistically and evolutionarily con-served superfamily of proteins that function in a diversearray of cellular processes (Sprang, 1997). G proteinstypically serve as molecular ‘‘switches’’ or ‘‘timers’’ bycycling through large conformational changes associ-ated with the binding and hydrolysis of GTP (Takaiet al., 2001). Some of the processes regulated by thisprotein family in eukaryotes include cellular proliferationand differentiation, vesicle trafficking, and cytoskeletalrearrangements. Bacterial G proteins are also used fora myriad of purposes (Caldon et al., 2001). For example,Obg is a widely distributed G protein that has been im-plicated in the cellular differentiation of Streptomycescoelicolor (Okamoto and Ochi, 1998) and in the stressresponse pathway of Bacillus subtilis (Scott and Halden-

*Correspondence: jmberger@berkeley.edu (J.M.B.); crb@berkeley.

edu (C.R.B.)4 Present address: Department of Microbiology and Molecular Ge-

netics, Harvard Medical School, Boston, Massachusetts 02115.5 Present address: Department of Chemistry, Simon Fraser Univer-

sity, Burnaby, British Columbia, Canada, V5A 1S6.

wang, 1999). Era, another conserved bacterial G protein,is essential for cell cycle control in E. coli (Gollop andMarch, 1991).

A conserved nucleotide binding domain of w200 resi-dues, the G domain, is responsible for controlling G pro-tein function. Structural, biochemical, and genetic workhas been performed on this domain in the context of sev-eral G proteins, including members of the Ras (Takaiet al., 2001), elongation factor (Andersen et al., 2003),and heterotrimeric G protein families (Cabrera-Veraet al., 2003; Preininger and Hamm, 2004). These studieshave shown that the enzyme undergoes large conforma-tional changes while cycling between the GTP and GDPbound states. Two motifs, termed switches 1 and 2, un-dergo reorientation, and in many cases even refolding,by virtue of contact exchanges with GTP�Mg2+ and/orGDP�Mg2+. The typical outcome of these conformationalchanges is that the G protein oscillates between bindingone or a set of proteins when bound to GTP and an en-tirely different protein or set of proteins when bound toGDP. The cycling rate between states varies by severalorders of magnitude among G protein families and isoften controlled in trans by a combination of GTPase ac-tivating proteins (GAPs), guanine nucleotide exchangefactors (GEFs), and guanine nucleotide dissociation in-hibitors (Andersen et al., 2003; Li and Zhang, 2004).

CysN is a bacterial G protein that deviates signifi-cantly from the common paradigms through whichthese proteins typically function. This enzyme is an es-sential component of the proteobacterial ATPS, whichcatalyzes the production of APS and pyrophosphate(PPi) from ATP and sulfate (Figure 1A). As the first stepin the sulfate assimilation pathway, this process is cru-cial for the biosynthesis of sulfur-containing amino acidsand cofactors, and sulfated metabolites. E. coli ATPS isknown to consist of CysN and another, smaller subunit,CysD (Leyh et al., 1988). CysD belongs to the ATP pyro-phosphatase (ATP PPase) family of proteins, membersof which include PAPS reductase (Savage et al., 1997),GMP synthetase (Tesmer et al., 1996), asparagine syn-thetase (Larsen et al., 1999; Oda et al., 1999), andNAD+ synthetase (Rizzi et al., 1996). This subunit is re-sponsible for directly forming APS under control of theG protein.

The reaction to form APS is extremely thermodynam-ically unfavorable (Keq = 1.1 3 1028 at pH 8.0). As a con-sequence, the energy resulting from GTP hydrolysis byCysN drives the formation of a single high-energy chem-ical bond, the phosphoric-sulfuric acid anhydride bondof APS. In E. coli ATPS, this coupling harnesses the fullenergetic potential of GTP hydrolysis, shifting the equi-librium by six orders of magnitude toward APS for-mation (Liu et al., 1994b). Leyh and colleagues haveprovided extensive insights into the biochemical mech-anism of this coupling. One key aspect of this mecha-nism is that the affinity of the enzyme for GTP increasesdramatically as ATP progresses to the a,b-cleaved state(Wei et al., 2000). Prior to GTP hydrolysis, the enzymeisomerizes and becomes committed to APS synthesis(Wei and Leyh, 1999).

Molecular Cell110

E. coli

N CD

M. avium

M. tuberculosis

N CD

P. syringae

N CD N C

N C* D C

RO–SO3–

APSKinase(CysC)

APS/PAPSReductase

(CysH)

APS PAPS

SO32–

SO42–

Reduced SulfurMetabolites

ATPSulfurylase(CysN•D)

Sulfo-transferases

ATP + GTPPPi + GDP

SO32–

ATPADP 3',5'-ADP

B

C

A

ATP

SO42–

+ + PPi

ATPSulfurylase

APS

S

O

O

O–O

P

O–

O

O

O

O OH

N

N

N

N

NH2

H

Figure 1. Overview of Bacterial Sulfate As-

similation

(A) Generalized reaction catalyzed by ATPS

enzymes.

(B) Schematic of the bacterial sulfate assimi-

lation pathway. Solid lines represent enzy-

matic transformations present in all bacterial

sulfate assimilation pathways. Dashed lines

represent steps that are variable among bac-

teria. The catalytic domain of ATPS (CysD)

adenylates sulfate, forming APS and PPi.

This subunit associates with CysN, a GTPase

regulatory subunit. In bacteria with an APS

reductase (CysH), the sulfate moiety of APS

is reduced to sulfite for eventual incorpora-

tion into reduced sulfur metabolites. APS

can also be phosphorylated by APS kinase

(CysC), forming PAPS. In some bacteria,

PAPS is the substrate for CysH and the entry

point for the reductive branch of the pathway.

PAPS is also the universal sulfuryl group do-

nor for sulfotransferase enzymes.

(C) Examples of the various genetic struc-

tures of bacterial PAPS-synthesis genes. In

E. coli, cysD, cysN, and cysC are individual

open reading frames on a single operon. In

M. tuberculosis and many other bacteria,

cysN and cysC are fused, forming a bifunc-

tional cysNC gene. M. avium possesses two

apparent PAPS-synthesis loci. One locus re-

sembles that found in M. tuberculosis, and

the other, unlinked locus, appears to encode

separate CysN and CysC genes. P. syringae

also has two loci involved in PAPS synthesis.

As we demonstrate in our current work, the

APS kinase domain of the apparent CysNC

in this organism is inactive (indicated as C*).

A functional CysC is encoded at an unlinked

locus.

Sulfur metabolism is an essential component of everyliving organism (Komarnisky et al., 2003; Marzluf, 1997;Saito, 2004; Townsend et al., 2004). Sulfur prototrophs,including plants and microorganisms, are able to uptakeand survive with sulfate as a sole sulfur source. Sulfate isunreactive under physiological conditions and thereforemust be activated in order to enter cellular metabolism(Figure 1B). Activation begins with sulfate adenylationcatalyzed by ATPS, leading to the production of APS.

In many sulfur prototrophs, APS can be reduced (Bicket al., 2000; Kopriva and Koprivova, 2004) and the sulfuratom is incorporated into metabolites such as cysteineand methionine. Alternatively, APS can be phosphory-lated by APS kinase to produce 30-phosphoadenosine50-phosphosulfate (PAPS; Figure 1B) (Renosto et al.,1984). In E. coli and some plants, PAPS serves as thesubstrate for the reductive branch of the pathway (Kroneet al., 1991). PAPS also serves as the universal sulfurylgroup donor for sulfotransferases (Figure 1B), whichare responsible for generating sulfated metabolites. Sul-fated molecules are common mediators of cell-cell inter-actions in eukaryotes (Hemmerich and Rosen, 2000;Hemmerich et al., 2004; Moore, 2003). A growing list ofbacteria, including several species of the plant symbiontgenus Rhizobium (Cronan and Keating, 2004; Haninet al., 1997; Roche et al., 1991) and the human and plantpathogens Mycobacterium tuberculosis (Mougous et al.,2002a, 2002b, 2004) and Xanthomonas ooryzae (da Silvaet al., 2004), respectively, also produce sulfated metab-

olites that have been implicated in interactions betweenthe bacterium and their eukaryotic host.

The structure of ATPS enzymes from Saccharomycescerevisiae, Penicillium chrysogenum, Thermus thermo-philus, and a bacterial symbiont of the hydrothermalvent tubeworm Riftia pachyptila have been solved.These enzymes, however, do not require GTP and bearno sequence similarity to the bacterial enzyme (Beynonet al., 2001; MacRae et al., 2001; Taguchi et al., 2004; Ull-rich et al., 2001). Indeed, there is currently no suitablestructural model for the energetic linkage of a G proteinto the production of a core metabolite.

To provide molecular insight into the linkage betweenGTP hydrolysis and APS formation in ATPS, we bio-chemically characterized and solved the 2.7 A X-raycrystal structure of a complex between Pseudomonassyringae ATPS and its regulatory G protein. Our bio-chemical analyses reveal that P. syringae possessesa previously undescribed type of bacterial sulfate assim-ilation gene cluster, in which the resident APS kinasegene is enzymatically inactive yet crucial for oligomeri-zation and full activity of the complex. We demonstratethat GTP hydrolysis by CysN is not dependent on ATPhydrolysis by CysD, whereas ATP hydrolysis by CysDis highly dependent on the GTP hydrolytic activity ofCysN. Structural studies of ATPS show that CysN andCysD form a tight 1:1 complex, with CysD bound toa central cavity formed by the junction of the three do-mains of CysN. CysN, bound to GDP (CysNGDP) in our

Structure of a G Protein-ATP Sulfurylase Complex111

(Phe � Ile)

(Phe � Gln)

(Asp � Gln)

(Arg � Gln)

GTP

M. tbNC•D

P. syNC*•D

P. syN•D

P. syNC*•D/C

APS

PAPS

B

A

C

P-LoopAPS Binding

P. chrysogenumS. cerevisiae

H. sapienM. tuberculosis

E. coli

P. syringae cysNC*P. syringae cysC

Figure 2. The APS Kinase Portion of the P. syringae CysNC Fusion Is Inactivated by Mutation of Conserved APS Binding Residues

(A) Partial multiple sequence alignment of evolutionarily diverse APS kinase enzymes. Residues marked with circles and diamonds are involved

in APS binding and are absent or mutated in P. syringae CysNC*.

(B) View of APS and residues implicated in APS binding in the active site of the APS kinase-like region of P. chrysogenum ATPS (PDB code, 1I2D).

Mutations found in P. syringae CysNC* in conserved residues that are crucial for APS binding are indicated in parenthesis. Of particular note are

the mutations in two perfectly conserved phenylalanine residues that p stack with the nucleotide base (indicated with diamonds in [A]).

(C) In vitro activity of M. tuberculosis and P. syringae sulfate assimilation enzyme complexes. The indicated protein complexes were heterolo-

gously expressed and purified from E. coli JM81A lDE3. 35SO422 was used to follow APS and PAPS production from each complex in the pres-

ence and absence of GTP. Each complex required GTP for activity. The complete M. tuberculosis complex synthesizes PAPS, whereas the

P. syringae CysNC*�D complex is only capable of synthesizing APS. A P. syringae complex lacking the APS kinase-like domain (CysN�D)

also is capable of catalyzing APS formation. The addition of the separately encoded P. syringae CysC protein restores PAPS production

(CysNC*�D/C).

structure, is highly similar to the GDP bound state ofelongation factor 1A (EF1AGDP). Comparison of CysDto related structures shows that it possesses a uniqueinsertion that makes extensive contacts to both switchdomains of CysN. Limited proteolysis of CysN in thepresence of various nucleotides indicates that confor-mational changes occur in the switch I region duringthe catalytic cycle. These data suggest that the energyof GTP hydrolysis by CysN is transmitted to the activesite of CysD through changes in switch domain confor-mation.

Results and Discussion

The APS Kinase Domain of P. syringae CysNC

Is InactiveThe genes encoding CysN and CysD were originallyidentified and characterized in E. coli, where cysN andcysD were found to reside in an operon together withAPS kinase (cysC; Figure 1C) (Leyh et al., 1988). Subse-quent genomic studies of diverse bacteria have revealedseveral genetic arrangements of the cysN, cysC, andcysD genes. Notably, a fusion between cysN and cysChas been observed that leads to the production of a bi-functional protein (CysNC) with GTPase and APS kinase

activities (Figure 1C) (Schwedock et al., 1994). In thecontext of this fusion, CysN maintains its interactionswith CysD, forming a single complex (CysNC�D) that uti-lizes GTP hydrolysis to drive the formation of PAPS fromATP and sulfate.

Sequence comparisons of bacterial ATPS enzymes in-dicate that several Pseudomonas sp., including P. syrin-gae, possess CysN proteins fused N-terminally to a pro-tein with a high degree of homology to CysC (Figure 1C).Close examination of the sequence of the CysC regionin these organisms, however, reveals the presence ofmultiple mutations in conserved motifs involved in APSbinding (Figure 2A) (MacRae et al., 2001). In particular,two phenylalanine residues that form critical stacking in-teractions with APS in the crystal structure of the closelyrelated P. chrysogenum APS kinase are mutated in anonconservative fashion to Gln513 and Ile591 in eachof the Pseudomonas CysNC fusions (Figures 2A and2B). An additional deletion of four conserved residuesin the region adjacent to Ile591 (corresponding to resi-dues 160–163 of P. chrysogenum APS kinase) shouldfurther disrupt APS binding. These observations, com-bined with structural modeling, suggested to us thatthe Pseudomonas CysNC lacks APS kinase activity (Fig-ure 2B). An apparent deteriorated APS kinase has also

Molecular Cell112

been observed as a fusion to the fungal ATPS enzymes(Lalor et al., 2003; MacRae et al., 2001). In this case, how-ever, the kinase domain has P loop mutations that abol-ish activity. The domain retains PAPS and APS binding,which allosterically regulate the enzyme. Henceforth, werefer to the P. syringae open reading frame as cysNC*.We also noted the presence of a separately encodedapparent APS kinase paralog, CysC2, in the genomesof P. syringae and P. aeuruginosa (Figure 1C).

To investigate the activity of the Pseudomonas ATPS,we heterologously coexpressed P. syringae CysNC* andCysD in E. coli. A complex containing the two proteinswas isolated from lysates utilizing a His6 tag appendedto the N terminus of CysD. To avoid contamination withendogenous E. coli APS kinase, we obtained an E. coliAPS kinase-deficient strain (JM81A) and prepared itfor inducible overexpression by transduction with thelDE3 lysogen (JM81A lDE3) (Leyh et al., 1988). Thisstrain was used for protein expression in all subsequentbiochemical studies. We also purified the M. tuberculo-sis CysNC�D complex by similar methods, as this com-plex has recently been shown to perform both ATPS andAPS kinase reactions (Sun et al., 2005).

Using these purified samples, APS and PAPS pro-duction was measured through an assay that allowsproduct visualization by 35SO4

22 labeling. As expected,M. tuberculosis CysNC�D produced PAPS, whereasP. syringae CysNC*�D synthesized APS (Figure 2C).Importantly, both complexes required GTP for APS pro-duction. To ensure that the CysC* domain of CysNC* wasnot required for APS synthesis, we designed a truncation(residues 1–438, CysN) of the protein that lacked thisdomain. The truncation retained association with CysD(Figures 3A and 3B) and readily produced APS in aGTP-dependent manner (Figure 2C).

Our observation of a second, potentially functionalAPS kinase gene prompted us to investigate whethercertain Pseudomonas sp. have the capability to producePAPS. P. syringae CysC2, expressed and purified fromE. coli JM81A lDE3, worked in concert with CysNC*�Dto produce PAPS (Figure 2C). This result demonstratesthat P. syringae and likely other Pseudomonas sp. havea functional APS kinase and therefore are able to synthe-size PAPS. Because these organisms possess an APS,rather than a PAPS reductase, CysC2 is likely not re-quired for sulfur prototrophy and is essential only forthe production of sulfated molecules (Figure 1B) (Bicket al., 2000). Indeed, an open reading frame predictedto encode for a sulfotransferase has previously beenidentified on the avrD locus of P. syringae pv. tomato(Hanin et al., 1997; Kobayashi et al., 1990). Interestingly,the genome of P. putida, a nonpathogenic strain, doesnot appear to encode for a functional APS kinase.

The GTPase Activity of CysN Is Enhancedby Dimerization through CysC*

We next sought to determine the function of the inacti-vated APS kinase portion of CysNC*. Prior biochemicaland structural studies have shown APS kinase to be a di-mer; therefore, we hypothesized that one function of theCysC* domain could be to multimerize the complex(MacRae et al., 2000). To determine the contributionof CysC* to oligomerization, purified CysNC*�D andCysN�D were subjected to analytical gel filtration chro-

matography (Figures 3A and 3B). Both complexeseluted as a single species at masses of 285 and60 kDa for CysNC*�D and CysN�D, respectively. Integra-tion of SDS band intensities indicated a 1:1 stoichiom-etry of the two purified subunits. Using this restraint,the measured mass of CysNC*�D is intermediate tothat of a calculated a2b2 tetramer and an a3b3 hexamer.To resolve this ambiguity and provide direct evidence ofoligomerization through CysC*, we purified a truncationof CysNC* containing only the CysC* portion (residue448 to the C terminus of CysNC*) and subjected it togel filtration (Figure S1 available in the SupplementalData with this article online). CysC* eluted at a massclosely matching that of a dimeric species (44.2 kDameasured versus 44.8 kDa calculated for the dimer).These results confirm that CysC* mediates oligomeriza-tion of CysNC*�D.

To determine the effect of oligomerization on the en-zymatic activity of CysNC*�D, we measured the intrinsic(2ATP) and stimulated (+ATP) rate of GTP hydrolysis byCysN�D and CysNC*�D (Figure 3C). The intrinsic GTPhydrolysis rate of CysN�D was below the detection limitof our assay; however, at saturating ATP, kcat

GTP wasmeasured to be 0.24 s21. The intrinsic GTPase activityof the full complex was similar to that of the stimulatedrate of the CysN�D complex (kcat

GTP = 0.28 s21), andthis was further enhanced 26-fold (kcat

GTP = 7.22 s21)in the presence of ATP. Assays performed to measurethe efficiency of APS production by these complexesmirrored the GTPase activities (data not shown).

To ascertain the contribution of CysC* versus CysD tothe intrinsic GTPase activity of CysN, we expressed andpurified CysNC* and CysN in the absence of CysD(Figure 3B). These proteins expressed in a soluble fash-ion and were monodisperse by gel filtration chromatog-raphy. Although GTPase activity was not detectable forCysN alone, CysNC* displayed a low intrinsic GTPaseactivity (kcat

GTP = 0.010 s21, Figure 3C). Neither subcom-plex lacking CysD responded to the addition of ATP,demonstrating that ATP stimulation of the full complexis mediated through CysD. Although it is difficult to en-tirely rule out nonspecific factors that could result fromtruncating CysNC*, we believe these data, combinedwith our structural and biochemical data (see below),support a role for multimerization by CysC* in producingan inherently activated state of the enzyme complex.As evidenced by the intrinsic GTPase activities ofCysNC*�D and CysNC*, this state is further potentiated28-fold by the addition of CysD. This suggests that CysDbehaves analogously to GAP proteins by lowering thetransition state energy of GTP hydrolysis by CysN.

GTP Hydrolysis Depends on the Ligand Stateof CysD and Not on APS Formation

To investigate the relationship between GTP hydrolysisby CysNC* and the catalytic state of CysD, we measuredGTPase rates in the presence of various CysD ligands(Figure 3D). At saturating ATP, GTP, SO4

22, and Mg2+,kcat

GTP is 7.2 s21, a value in good agreement with thepublished value of 7.8 s21 for the closely relatedM. tuberculosis ATPS (Pinto et al., 2004). Consistentwith studies of the E. coli ATPS, substitution of ATP withAMP has only a slight effect on kcat

GTP (Wang et al.,1995; Yang and Leyh, 1997). It is important to note that

Structure of a G Protein-ATP Sulfurylase Complex113

Figure 3. Biochemical Characterization of P. syringae CysNC*�D(A) The tetrameric oligomeric state of CysNC*�D is unaffected by the nucleotide state of CysN. Normalized analytical gel filtration chromato-

graphs of CysNC*�D in the presence of 250 mM GDP (solid) or GMPPNP (long dashes), and CysN�D in the presence of 250 mM GDP (short

dashes). As judged by their similar elution volumes, the nucleotide state of CysNC*�D does not affect association, oligomerization, or the overall

conformation of the complex. The CysN�D complex elutes at a mass consistent with a dimeric species.

(B) SDS-PAGE analysis of purified P. syringae ATP sulfurylase components used in biochemical investigations.

(C) The intrinsic and stimulated rates of GTP hydrolysis by CysN are enhanced through association with CysD and oligomerization mediated by

CysC*. GTP hydrolysis was determined by using an in vitro coupled assay that measures GDP release. The names of the protein complexes

analyzed have been abbreviated for clarity. Reactions contained saturating Mg2+ and GTP. Error bars indicate one standard deviation from

triplicate experiments.

(D) Adenine nucleotide binding promotes GTP turnover. Results of in vitro assay measuring GTP hydrolysis by CysNC*�D in the presence or

absence of SO422, ATP, or an ATP analog (‘‘variable’’ column). Reactions contained saturating Mg2+ and GTP. Error bars indicate one standard

deviation from triplicate experiments.

(E) Hydrolysis of ATP by CysD is highly sensitive to the nucleotide state of CysN. Results of in vitro coupled assay measuring PPi release by

CysNC*�D in the presence or absence of saturating sulfate, GTP, a nonhydrolyzable GTP analog (GMPPNP), or GDP. All reactions contain sat-

urating Mg2+ and ATP. The rate of the reaction containing GMPPNP is significantly above background. Error bars indicate one standard deviation

from triplicate experiments.

under these conditions, APS is not synthesized (data notshown). Replacing ATP with AMP-CPP, an ATP analogthat cannot be hydrolyzed at the a,b position, results ina 10-fold decrease in kcat

GTP. Interestingly, half of

this decrease can be rescued by removing sulfate, sug-gesting that sulfate binding can stall the complex ina nonproductive state awaiting displacement of PPi.The opposite effect is observed in the presence of

Molecular Cell114

Table 1. Crystallographic Data Collection and Structure Refinement

Data Collection Se-Met Peak 1 Se-Met Peak 2 Se-Met Remote

Resolution (A) 50–2.7 50–2.9 50–2.9

Wavelength (A) 0.9794 0.9793 0.9184

Space group H3 H3 H3

Unit cell dimensions (a = b, c) A 110.2, 170.9 110.0, 170.8 110.0, 170.8

I/sa 16.1 (2.8) 11.5 (2.2) 10.3 (2.0)

Rsym (%)b 0.066 (0.346) 0.053 (0.280) 0.060 (0.296)

Completeness % 99.7 (98.5) 95.6 (75.4) 97.7 (92.5)

Redundancy 5.8 (5.2) 2.9 (2.4) 2.9 (2.8)

Unique reflections 42,225 32,800 33,280

Number of sites 2 21 21

Refinement Statistics Structure and Stereochemistry

Rwork (%)c 22.2 Number of atoms

Rfree (%) 27.6 Protein 6489

Mean temperature factors Ligand 61

Protein 42.5 Water 87

Ligand 46.2 Rmsd bond lengths (A) 0.005

Solvent 36.1 Rmsd bond angles (º) 1.059

Ramachandran plot (%)

Favored 90

Additional 10

a Numbers in parenthesis represent values for the highest resolution bin.b Rsym = SSjjIj 2 CIDj / SIj, where Ij is the intensity measurement for reflection j, and CID is the mean intensity for multiply recorded reflections.c Rwork, free = SkFobsj2 jFcalck / jFobsj, where the working and free R factors are calculated by using the working and free reflection sets, respec-

tively. The free reflections (5%) were held aside throughout refinement.

ATP; here, removal of sulfate causes an approximate 2-fold decrease in kcat

GTP. Under these conditions, ATP ishydrolyzed at the same rate as measured in the pres-ence of saturating sulfate (Figure 3E). Thus, the rapid hy-drolysis of GTP in this scenario represents ‘‘leakiness’’in the interdependence of GDP and APS formation.Given this property of the enzyme, it is not surprisingthat the complex is tightly transcriptionally regulatedby the presence of sulfate (Pinto et al., 2004).

Using an assay that detects PPi, we also performedexperiments in which we probed the linkage betweenthe catalytic activity of CysD and the ligand state ofCysNC* (Figure 3E). Substitution of GTP with the non-hydrolyzable analog GMPPNP results in an 85-folddecrease in ATP hydrolysis activity of CysD. This resultdemonstrates that although PPi production is not af-fected by the presence or absence of sulfate, its produc-tion is strongly dependent on the ability of CysNC* to hy-drolyze GTP. The low level of GTP hydrolysis in thepresence of GMPPNP was not detected in similar stud-ies of the E. coli CysN�D enzyme (Liu et al., 1994a). NoPPi formation could be detected in the presence ofGDP or the absence of a guanosine nucleotide. Thesefindings show that the ligand state of CysNC* is inti-mately linked to the catalytic state of CysD. WhereasCysN does not require ATP hydrolysis by CysD for max-imal activity, CysD does require GTP hydrolysis by CysNin order to efficiently hydrolyze ATP.

Overall Structure of the CysN�D Dimer Boundto GDP and ATPgS

As a first step toward understanding the molecular basisfor the energetic coupling between GTP hydrolysis andAPS production, we determined the X-ray crystal struc-ture of the P. syringae CysN�D heterodimer to 2.7 A res-olution. Importantly, the truncation we chose for crystal-lographic analysis retains GTP-dependent ATPS activity

(Figure 2C). Crystals of CysN�D were grown in the pres-ence of GDP and ATPgS. Crystal formation was de-pendent on GDP and inhibited by the addition of Mg2+

or other divalent cations. Our CysN�D crystals belongto the space group R3 and contain one CysN�D hetero-dimer per asymmetric unit. After phasing by MAD meth-ods, the model was built and refined to final Rwork andRfree values of 0.222 and 0.276, respectively (Table 1).Evidence for GDP binding to CysN and ATPgS bindingto CysD was unambiguous in experimental electrondensity maps. Because GTP hydrolysis has been shownto precede ATP hydrolysis in the closely related E. coliCysN�D system, this combination of nucleotides likelyrepresents a catalytically-relevant intermediate in theenzymatic reaction (Liu et al., 1994a).

In agreement with our biochemical studies, the crystalstructure of the CysN�D complex shows that the sub-units associate in a 1:1 stoichiometry (Figure 4). TheN-terminal domain of CysD resembles PAPS reductaseand other members of the ATP PPase family, and itbinds ATPgS through contacts with residues in the PPi

binding loop (Pilloff and Leyh, 2003). As predicted byprimary sequence comparisons (Inagaki et al., 2002),CysN contains three domains and shares striking struc-tural similarity to EF1A (Figure 5). Using the nomencla-ture adopted for the homologous EF1A, CysN containsa nucleotide binding domain (G domain) consisting ofthe N-terminal 234 residues. This region connects to do-main 2 by an extended six residue linker. Domains 2 and3 each consist of w100 residues and are six-strandedb barrels whose axes are roughly perpendicular. The ori-entation of the three CysN domains bestows a ‘‘saucer’’shape to the molecule, with CysD bound on the concaveface. This configuration allows CysD, a relatively smallglobular protein, to make significant contacts to eachof the three domains of CysN (Figures 4A and 4B). A par-ticularly large number of dimer contacts occur between

Structure of a G Protein-ATP Sulfurylase Complex115

GMP synthetase PAPS reductase

90°

A

90°

B

C

N

D

G domain

domain IIIdomain II

Figure 4. Structure of the P. syringae CysN�D Dimer

(A) Cartoon representation of the CysN�D dimer. The G domain and domains 2 and 3 of CysN are colored red, green, and blue, respectively. CysD

is colored yellow. GDP and ATPgS are shown in sticks. Disordered loops are represented with spheres.

(B) CysD binds to a large depression on the concave face of CysN. CysN is shown in surface representation with bound GDP as spheres. A Ca coil

trace of CysD is shown with bound ATPgS as sticks. The color scheme is the same as in (A), with the exception that residues in the switch I and II

regions of CysN are colored orange and white, respectively. The sequence element of CysD that mediates contacts to both switch domains of

CysN is colored purple.

(C) The sequence element of CysD that mediates switch domain contacts is a distinguishing feature relative to other ATP PPase family members.

Left, superposition of the ATP PPase domain of GMP synthetase (gray) and CysD (coloring scheme same as [B]). PPi bound in the structure

of GMP synthetase is colored cyan and shown in stick representation. For clarity, bound AMP and the amidotransferase and oligomerization

domains of GMP synthetase have been omitted. Right, superposition of PAPS reductase (gray) and CysD.

Molecular Cell116

β1 α1

G1

β6α5 α6

G5

β2 β3α2

G2 G3Switch I

β4 β5α3 α4

G4Switch II

A

P. syringae CysNC*E. coli EF1A

T. aquaticus EF1A

P. syringae CysNC*E. coli EF1A

T. aquaticus EF1A

P. syringae CysNC*E. coli EF1A

T. aquaticus EF1A

P. syringae CysNC*E. coli EF1A

T. aquaticus EF1A

B C

EF1AGDP

EF1AGTP

CysNGDP

α1α2

α3

α4

α5

α6

β1

β2 β3 β4β5β6

Structure of a G Protein-ATP Sulfurylase Complex117

the long N-terminal helix of CysD and three of theb strands of CysN domain 2. The N-terminal six residuesof CysD form a strap that binds in a groove between theG domain and domain 2 of CysN.

CysNGDP Resembles EF1AGDP: Evidencefor Conserved Conformational Changes

Studies of E. coli and M. tuberculosis ATPS have pro-vided evidence for an isomerization step that precedesand rate limits GTP hydrolysis (Sun et al., 2005; Weiand Leyh, 1998, 1999). This isomerization is dependentupon allosteric interactions between the two active sitesof the enzyme. A significant finding in our structure ofATPS is that the nucleotides bound to CysN and CysDare spatially (w40 A) and sterically isolated from eachother (Figure 4A). This finding is noteworthy because itprecludes the possibility of direct conformational cou-pling between residues of the subunits that are in im-mediate contact with the nucleotides. As illustrated inFigures 5B and 5C, the structures of CysNGDP andEF1AGDP share striking similarity. Several structures ofEF1A have revealed that the enzyme undergoes a dra-matic structural rearrangement as it cycles betweenGDP and GTP bound states (Figure 5C) (Andersenet al., 2003; Berchtold et al., 1993; Jurnak, 1985; Kjeldg-aard and Nyborg, 1992; Sprang, 1997). Given the factthat CysN possesses a full complement of the sequenceelements implicated in the conformational changes ofEF1A (Figure 5A), we questioned whether the isomeriza-tion step of ATPS would be similar to the conformationalchanges observed for EF1A.

If CysN assumes an EF1AGTP-like structure whenbound to GTP, the dimer interface observed in the struc-ture of ATPS would necessarily be significantly altered,or more dramatically, the dimer could dissociate. To as-say for dimer dissociation, we subjected purified proteinto analytical gel filtration analysis in the presence of100 mM GDP or GMPPNP (Figure 3A). As judged by a sin-gle peak elution profile at a mass consistent with thetetramer species, CysN and CysD remained associatedunder both nucleotide states. The two nucleotide statesalso showed no change in their measured mass, sug-gesting that the overall shape of the complex does notdiffer dramatically between them. To next investigatewhether potential structural changes take place in CysN,we performed limited trypsinolysis with purified CysNin the presence of Mg2+ and either GDP or GMPPNP(Figure 6). This experiment revealed that CysNGDP�Mg2+

was partially stabilized relative to CysNGMPPNP�Mg2+.Removal of Mg2+ from the reactions had no effect onCysNGDP; however, it significantly reduced the stabilityof CysNGMPPNP.

Mass spectrometric and N-terminal sequence analy-sis of the trypsin-treated protein showed that the differ-

entially protected trypsin cleavage sites of CysN arefound in a disordered loop within the switch I region (Fig-ures 5A and 5B). We interpret these nucleotide andMg2+-dependent changes in the resistance to proteasestability to reflect altered conformational states thatthe region adopts during its catalytic cycle. It is interest-ing to note that the presence of Mg2+ does not appear toaffect the conformation of CysNGDP. Close inspection ofa comparison between the active sites of CysNGDP andEF1AGDP may explain this phenomenon. The conservedG3 loop of the switch II domain, comprising residues110–112 of CysN, is displaced from the active site by sev-eral angstroms relative to its position in EF1AGDP (Fig-ure 6A) (Song et al., 1999). In EF1A, residues of this loopare critical for forming water-mediated Mg2+ contacts.

Mg2+ binding to another EF1A-like G protein, the yeasttranslation termination factor eRF3, also was found re-cently to be heavily dependent on the nucleotide state(Kong et al., 2004). In this case, Mg2+ weakened GDP bind-ing but strengthened GTP binding. The authors proposedthat this reflects a novel means of GEF-independent nu-cleotide exchange, whereby cellular Mg2+ levels woulddiminish GDP binding and promote GTP binding. Thismay also hold true for CysN, which does not appear torequire a specialized GEF for nucleotide exchange (al-though CysD could be argued to fulfill for this function).Unlike eRF3, however, our data suggest that the effectorregion of CysN does undergo conformational changesduring the course of nucleotide exchange. Becausethe alternative nucleotide states of eRF3 were obtainedby soaking nucleotides into preformed crystals, it re-mains unclear whether the absence of conformationalchanges observed in these structures might have arisendue to crystal packing constraints.

Relationship of CysD and N Type

ATP PyrophosphatasesThe structure of CysD bears no detectable similarity toeukaryotic ATPS. Rather, the N-terminal domain of theprotein, consisting of the first 211 residues, is structur-ally homologous to the adenylation domain of N typeATP PPases. Structurally characterized members ofthis family include the GMP, NAD+, and asparagine syn-thetase enzymes (Larsen et al., 1999; Nakatsu et al.,1998; Rizzi et al., 1996; Tesmer et al., 1996). In these pro-teins, the function of this domain is to activate substratethrough adenylation of a carbonyl or carboxylate group.A nucleophile (ammonia in these cases) is generated bya second domain, termed the amidotransferase domain,and subsequently displaces the adenylate group to formthe aminated product. The reaction catalyzed by CysDand other ATPS enzymes bears some similarity to thosemediated by N type ATP PPases; however, adenylatedintermediates have not been directly observed. Previous

Figure 5. CysN Resembles EF1A and Possesses Conserved G Protein Sequence Elements

(A) Multiple sequence alignment of the G domains of P. syringae CysNC*, E. coli EF1A, and T. aquaticus EF1A. Conserved motifs involved in GTP

binding (G1–G5), switch domains, and secondary structural elements of CysN are indicated. A dashed line is shown above residues of CysN that

are disordered in the structure. Arrows indicate the position of trypsin cleavage sites (Figure 6).

(B) Cartoon representation of CysN with bound GDP. The switch I and II domains are colored yellow and black, respectively. Secondary structural

assignments within the G domain are indicated. The approximate position of differential trypsin sensitivity (from [A]) is denoted with an arrow.

(C) The crystal structure of T. aquaticus EF1A in the GDP (top; PDB code, 1TUI) and GTP bound (bottom; PDB code, 1EFT) states. The color

scheme of both structures is the same as that used for CysN in (B). The orientation of the core b sheets of the G domain is fixed to aid in com-

parison of the two structures.

Molecular Cell118

Figure 6. The Switch I Region of CysN

Undergoes Mg2+-Dependent Conformational

Changes in the GTP Bound State

(A) Conformational changes are required for

CysNGDP to bind Mg2+. Stereo diagram com-

parison of the active sites of E. coli EF1AGDP

(green; PDB code, 1EFC) and CysNGDP (red).

Conserved residues directly interacting with

GTP were used to superimpose CysN and

E. coli EF1A (green). These residues and GDP

superimpose well (Ca root-mean-square de-

viation of 0.1 A); however, CysN residues

(110–112) structurally analogous to those

that make water-mediated contacts to Mg2+

in EF1A (80–82) are several angstroms re-

moved from the active site. An experimental

electron density map contoured at 1.3 s is

shown around GDP.

(B) SDS-PAGE analysis of CysN trypsinolysis

time course in the presence or absence of

Mg2+ and saturating GDP or GMPPNP. The

asterisk denotes the position of the band cor-

responding to CysN that is cleaved in the

switch I domain.

(C) Quantification of the data shown in (B).

mechanistic studies have shown that sulfate acts as anucleophile (Liu et al., 1994a), but the nature of the elec-trophile—ATP itself or an adenylated intermediate—hasnot been defined. One possibility involves formation ofan adenylated enzyme intermediate. We found no evi-dence for this in biochemical studies (data not shown)nor does our structure of CysD bound to ATPgS identifyan obvious position for adenylation in the active site.Thus, the presence of intermediates in the reaction cat-alyzed by CysD remains an open question.

Sequence alignment and structural comparison of theadenylation domain of CysD and other N type ATPPPases domains illuminates a conserved sequence ele-ment as a distinguishing feature of CysD (residues 144–178 of CysD) (Figures 4B and 4C). Interestingly, this ele-ment contains b5 and an extended loop of CysD thatforms intimate contacts with both switch regions ofCysN. Given that the insertion is highly conserved andspecific to CysD, and that it is responsible for all ob-served direct contacts with both switch domains of

CysN, it is likely to be critical for relaying switch domainconformational changes to the active site of CysD.

Although the C-terminal 88 residues of CysD werelargely disordered in our structure (see ExperimentalProcedures), we did observe weak experimental elec-tron density that suggested this domain caps the opennucleotide binding site of CysD (data not shown). Dueto the poor quality of the density and the absence ofan available model for the domain, this region was notincluded in our final structure. In GMP synthetase, theopen face of the adenylation domain and the distancebetween the two active sites of the enzyme promptedspeculation that the amidotransferase domain closesas a lid in order to facilitate the chemistry required forproduct formation (Tesmer et al., 1996). If the C-terminaldomain of CysD were to function in an analogous man-ner, i.e., as a lid that binds sulfate and brings it into prox-imity with ATP for nucleophilic attack, this may explainwhy the domain is disordered in our current sulfate-free structure and, additionally, why low levels of sulfate

Structure of a G Protein-ATP Sulfurylase Complex119

- GGGHHH -

C

D

APSsynthesis

APS consumption

N

Figure 7. Topological Model of APS Synthesis and Consumption in a CysNC�D PAPS Synthase Complex

Schematic of the relative locations of APS synthesis by CysD and APS consumption by CysC (gray). CysN�D subunits are colored as in Figure 4.

potently inhibit crystallization of ATPS under otherwiseidentical conditions.

Topological Model of a CysNC�D PAPS SynthaseCoregulation and substrate channeling have been pro-posed as evolutionary pressures for gene fusion events(Enright et al., 1999; Huang et al., 2001; Marcotte et al.,1999). When the activity of two proteins is necessaryfor the output of a particular biological pathway, a fusionevent can ensure perfect transcriptional and transla-tional coordination with minimal intervening regulatorymachinery.Substratechannelingcanactasanevolution-ary pressure for gene fusion if the coordinated activitiesof two proteins are required and sequential in a biologicalpathway. Substrate channeling is especially common inpathways that produce reactive or toxic intermediates(Huang et al., 2001). Either or both of these factors mighthave contributed to the formation of a PAPS synthasecomplex comprising a fusion of CysN and CysC.

Based on our structural and biochemical studies, wepropose a topological model for the complete CysNC*�Dcomplex. Because the inactivating mutations of CysNC*are not expected to affect the overall fold of the enzyme,we anticipate this model is equally valid for fully func-tional PAPS synthase complexes (CysNC�D; Figure 7).By virtue of the location of the C terminus of CysN andthe short linker between domains (GGGHHH in CysNC*,residues 439–445), there are strict spatial constraints onthe location of the APS domain. Interestingly, this modelsuggests that in a fully functional CysNC�D complex,CysN acts as a physical barrier between APS kinase ac-tivity, on its convex face, and the activity of CysD on itsconcave face. Therefore, physical channeling of APSfrom CysD to CysC is unlikely to occur in CysNC�D. Thisquaternary arrangement also takes into account our bio-chemical data and previous structural analysis of APSkinase, which strongly suggest that this domain is re-sponsible for tetramerization of the complex (Figure 2A).

Conclusions

The activation of sulfate by ATPS is required for sulfurprototrophy and for the production of sulfated mole-cules in a broad spectrum of organisms. Our genomicand biochemical analysis of the P. syringae sulfate as-

similation locus has shown that this organism possessesa previously unrecognized variant of the bacterial sulfateassimilation gene cluster, wherein the APS kinase por-tion of the CysNC gene has been inactivated by active-site mutations. Although not active as an APS kinase, theCysC* domain of CysNC* is responsible for oligomeriza-tion and is required for maximal activity of the associ-ated ATPS. Our biochemical dissection of P. syringaeATPS also revealed unique features of the enzyme. Wefound that binding of CysD to CysN induces its intrinsicGTPase activity and that this event is further potentiatedby oligomerization of the complex mediated through theAPS kinase-like domain of CysNC*. In the fully assem-bled complex, the rate of GTP hydrolysis by CysN islargely unlinked to ATP turnover by CysD. That is, robustGTPase activity requires adenosine nucleotide binding,but not sulfurylation. Conversely, ATP hydrolysis byCysD is highly responsive to GTP turnover by CysN.

The structure of P. syringae ATPS in complex with itsregulatory GTPase provides a glimpse into the unusualcoupling of a G protein to a core metabolic process. Asignificant finding in the structure is that the distance be-tween the nucleotides of CysN and CysD necessitatestransmittance of an intersubunit conformational changefor energetic coupling. Our structural and proteolyticdata, combined with the existing body of knowledge re-garding G protein mechanism, strongly implicate theswitch I domain of CysN in this process.

As a critical component of primary metabolism in anumber of pathogenic bacteria, ATPS represents anattractive target for antibiotics. Its divergence from eu-karyotic ATPS suggests the possibility of target singu-larity in human hosts. Ongoing efforts in the pharmaceu-tical industry to target G proteins involved in cell cycleregulation may provide a useful platform for the devel-opment of drugs that target sulfate assimilation in bac-teria (Walker and Olson, 2005).

Experimental Procedures

Protein Preparation

For crystallographic analysis, P. syringae CysN (GI:28854819, resi-

dues, 1–434 of CysNC) was coexpressed with P. syringae CysD

(GI:37999564) in E. coli BL21(DE3) by using pET21A and pET28B, re-

spectively. Protein expression was carried out with individual

Molecular Cell120

cotransformants overnight at 16ºC with 0.35 mM IPTG induction.

Cells were harvested and resuspended in 15 ml of lysis buffer

(50 mM Tris [pH 7.5], 0.5 M NaCl, 10% glycerol [v/v], 15 mM imidaz-

ole, 2 mM b-mercaptoethanol, and protease inhibitors) per liter of

culture. The CysN�D complex was purified under standard condi-

tions by using a PorosMC (Perspective Biosystems) column fol-

lowed by a concentration step and an S-300 sizing column. The

complex was subsequently concentrated to 20 mg ml21, stored,

and dialyzed into a minimal solution of 100 mM NaCl, 10 mM Tris

(pH 7.5), and 2 mM tris-(2-carboxyethyl)-phosphine (TCEP) over-

night prior to crystallization. Selenomethionine protein was over-

expressed in minimal media by the protocol of Van Duyne and cow-

orkers and purified similar to native CysN�D. P. syringae ATP

sulfurylase complexes used for biochemical analysis were coex-

pressed in E. coli JM81A (DE3). P. syringae CysC2 (GI:28872662)

was expressed and purified separately in E. coli JM81A (DE3).

E. coli JM81A was obtained from the E. coli Genetic Stock Center

(#8514) at Yale University. The DE3 lysogen was introduced by using

the lDE3 Lysogenization Kit (Novagen). Purification from this strain

was performed as above.

Crystallization and Structure Determination

Crystals of CysN�D were obtained by the vapor diffusion method by

mixing 1 ml of dialyzed protein with 1 ml of a well solution (14%–20%

PEG 4K [v/v], 100 mM Tris [pH 8.5], and 200 mM sodium acetate).

Nucleotides (GDP and ATPgS) were added to the crystallization drop

at a final concentration of 7 mM each. Crystals typically reached op-

timal size and diffraction quality after one week at room temperature.

The crystals were harvested and introduced into a final cryoprotec-

tant solution consisting of well solution with the addition of 10% eth-

ylene glycol (v/v).

Data were collected at Beamline 8.2.1 at the Advanced Light

Source (ALS) by using an ADSC Q210 detector. Diffraction data were

processed with HKL2000 (Otwinowski and Minor, 1997) (Table 1). A

two-wavelength dataset from a single selenomethionine protein

crystal was used for phasing. Selenium positions were initially found

in a SAD experiment by using SOLVE (Terwilliger and Berendzen,

1999). These sites were then refined by using both the peak and re-

mote wavelengths with MLPHARE (CCP, 1994) running under ELVES

(Holton and Alber, 2004). RESOLVE was used to extend the resolu-

tion and to calculate initial experimental electron density maps.

Model building was carried out with O (Jones et al., 1991) and refine-

ment with REFMAC5 (Murshudov et al., 1997) using TLS restraints

(Winn et al., 2001). A large fraction (15%) of the residues in the asym-

metric unit were disordered and therefore not included in the final

model. Water molecules were added with ARP/WARP (Lamzin and

Wilson, 1993). Final Rwork and Rfree values of 22.2% and 27.6%, re-

spectively, were obtained. All molecular figures were generated

with Pymol (http://www.pymol.org).

Biochemical Analyses

CysN�D complexes were purified as described above (see Protein

Preparation). GTP hydrolysis was measured by following GDP for-

mation using the coupled PK/LDH assay (Sigma). ATP hydrolysis was

measured with the Pyrophosphate Reagent (Sigma) coupled enzy-

matic assay. The standard reaction mixture contained 0.50 mM ATP,

0.25 mM GTP, 1.0 mM Na2SO4, 2.0 mM MgCl2, 50 mM HEPES (pH

8.0), and 1.2 mM enzyme. Where noted (see Figures 3D and 3E), aden-

osine 50-(a,b-methylene) triphosphate (AMP-CPP, 0.5 mM), pyro-

phosphate (PPi, 0.5 mM), adenosine 50-monophosphate (AMP,

0.5 mM), b,g-imido guanosine 50-triphosphate (GMPPNP, 0.25 mM),

or guanosine 50-diphosphate (GDP, 0.5 mM) was included. Reactions

were performed at room temperature in 96-well microtiter format.

The products of radioactive reactions were separated by using PEI

cellulose TLC plates (J.T. Baker, Phillipsburg, NJ) using 1 M LiCl. Re-

action products were quantified by using phosphorimaging followed

by densitometry. Limited proteolysis assays contained 2 mg ml21

CysN, 0.1 mg ml-1 Trypsin, 0.5 M NaCl, 50 mM Tris (pH 7.5), 10% glyc-

erol, 2 mM TCEP, 3 mM MgCl2, and 1 mM of the indicated nucleotide.

Supplemental Data

Supplemental Data include one figure and can be found with this

article online at http://www.molecule.org/cgi/content/full/21/1/109/

DC1/.

Acknowledgments

We are grateful to staff at the ALS for assistance with data collection

and processing, to David Akey, Scott Classen, Kevin Corbett, and

Emmanuel Skordalakes for crystallography advice, to Julie Roden

and Mary Beth Mudgett for P. syringae insights, to Aimee Shen for

helpful discussions, and to the laboratories of John Mekalanos

and Stephen Harrison for the use of their facilities. J.D.M. was sup-

ported by a fellowship from the Ford Foundation. This work was

supported by grants from the National Institutes of Health to

C.R.B. (AI51622) and J.M.B. (P50-GM62410).

Received: June 20, 2005

Revised: September 19, 2005

Accepted: October 27, 2005

Published: January 5, 2006

References

Andersen, G.R., Nissen, P., and Nyborg, J. (2003). Elongation factors

in protein biosynthesis. Trends Biochem. Sci. 28, 434–441.

Berchtold, H., Reshetnikova, L., Reiser, C.O., Schirmer, N.K.,

Sprinzl, M., and Hilgenfeld, R. (1993). Crystal structure of active elon-

gation factor Tu reveals major domain rearrangements. Nature 365,

126–132.

Beynon, J.D., MacRae, I.J., Huston, S.L., Nelson, D.C., Segel, I.H.,

and Fisher, A.J. (2001). Crystal structure of ATP sulfurylase from

the bacterial symbiont of the hydrothermal vent tubeworm Riftia

pachyptila. Biochemistry 40, 14509–14517.

Bick, J.A., Dennis, J.J., Zylstra, G.J., Nowack, J., and Leustek, T.

(2000). Identification of a new class of 50-adenylylsulfate (APS) re-

ductases from sulfate-assimilating bacteria. J. Bacteriol. 182, 135–

142.

Cabrera-Vera, T.M., Vanhauwe, J., Thomas, T.O., Medkova, M.,

Preininger, A., Mazzoni, M.R., and Hamm, H.E. (2003). Insights into

G protein structure, function, and regulation. Endocr. Rev. 24,

765–781.

Caldon, C.E., Yoong, P., and March, P.E. (2001). Evolution of a mo-

lecular switch: universal bacterial GTPases regulate ribosome func-

tion. Mol. Microbiol. 41, 289–297.

CCP4 (Collaborative Computational Project, Number 4) (1994). The

Ccp4 suite: programs for protein crystallography. Acta Crystallogr.

D Biol. Crystallogr. 50, 760–763.

Cronan, G.E., and Keating, D.H. (2004). A Sinorhizobium meliloti Sul-

fotransferase that Modifies Lipopolysaccharide. J. Bacteriol. 186,

4168–4176.

da Silva, F.G., Shen, Y., Dardick, C., Burdman, S., Yadav, R.C., de

Leon, A.L., and Ronald, P.C. (2004). Bacterial genes involved in

type I secretion and sulfation are required to elicit the rice Xa21-

mediated innate immune response. Mol. Plant Microbe Interact.

17, 593–601.

Enright, A.J., Iliopoulos, I., Kyrpides, N.C., and Ouzounis, C.A.

(1999). Protein interaction maps for complete genomes based on

gene fusion events. Nature 402, 86–90.

Gollop, N., and March, P.E. (1991). A GTP-binding protein (Era) has

an essential role in growth rate and cell cycle control in Escherichia

coli. J. Bacteriol. 173, 2265–2270.

Hanin, M., Jabbouri, S., Quesada-Vincens, D., Freiberg, C., Perret,

X., Prome, J.C., Broughton, W.J., and Fellay, R. (1997). Sulphation

of Rhizobium sp. NGR234 Nod factors is dependent on noeE,

a new host-specificity gene. Mol. Microbiol. 24, 1119–1129.

Hemmerich, S., and Rosen, S.D. (2000). Carbohydrate sulfotrans-

ferases in lymphocyte homing. Glycobiology 10, 849–856.

Hemmerich, S., Verdugo, D., and Rath, V.L. (2004). Strategies for

drug discovery by targeting sulfation pathways. Drug Discov. Today

9, 967–975.

Holton, J., and Alber, T. (2004). Automated protein crystal structure

determination using ELVES. Proc. Natl. Acad. Sci. USA 101, 1537–

1542.

Structure of a G Protein-ATP Sulfurylase Complex121

Huang, X., Holden, H.M., and Raushel, F.M. (2001). Channeling of

substrates and intermediates in enzyme-catalyzed reactions.

Annu. Rev. Biochem. 70, 149–180.

Inagaki, Y., Doolittle, W.F., Baldauf, S.L., and Roger, A.J. (2002). Lat-

eral transfer of an EF-1alpha gene: origin and evolution of the large

subunit of ATP sulfurylase in eubacteria. Curr. Biol. 12, 772–776.

Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Im-

proved methods for building protein models in electron-density

maps and the location of errors in these models. Acta Crystallogr.

A 47, 110–119.

Jurnak, F. (1985). Structure of the GDP domain of EF-Tu and location

of the amino acids homologous to ras oncogene proteins. Science

230, 32–36.

Kjeldgaard, M., and Nyborg, J. (1992). Refined structure of elonga-

tion factor EF-Tu from Escherichia coli. J. Mol. Biol. 223, 721–742.

Kobayashi, D.Y., Tamaki, S.J., and Keen, N.T. (1990). Molecular

characterization of avirulence gene D from Pseudomonas syringae

pv. tomato. Mol. Plant Microbe Interact. 3, 94–102.

Komarnisky, L.A., Christopherson, R.J., and Basu, T.K. (2003). Sul-

fur: its clinical and toxicologic aspects. Nutrition 19, 54–61.

Kong, C., Ito, K., Walsh, M.A., Wada, M., Liu, Y., Kumar, S., Barford,

D., Nakamura, Y., and Song, H. (2004). Crystal structure and func-

tional analysis of the eukaryotic class II release factor eRF3 from

S. pombe. Mol. Cell 14, 233–245.

Kopriva, S., and Koprivova, A. (2004). Plant adenosine 50-phospho-

sulphate reductase: the past, the present, and the future. J. Exp.

Bot. 55, 1775–1783.

Krone, F.A., Westphal, G., and Schwenn, J.D. (1991). Characterisa-

tion of the gene cysH and of its product phospho-adenylylsulphate

reductase from Escherichia coli. Mol. Gen. Genet. 225, 314–319.

Lalor, D.J., Schnyder, T., Saridakis, V., Pilloff, D.E., Dong, A., Tang,

H., Leyh, T.S., and Pai, E.F. (2003). Structural and functional analysis

of a truncated form of Saccharomyces cerevisiae ATP sulfurylase:

C-terminal domain essential for oligomer formation but not for activ-

ity. Protein Eng. 16, 1071–1079.

Lamzin, V.S., and Wilson, K.S. (1993). Automated refinement of pro-

tein models. Acta Crystallogr. D Biol. Crystallogr. 49, 129–147.

Larsen, T.M., Boehlein, S.K., Schuster, S.M., Richards, N.G., Tho-

den, J.B., Holden, H.M., and Rayment, I. (1999). Three-dimensional

structure of Escherichia coli asparagine synthetase B: a short jour-

ney from substrate to product. Biochemistry 38, 16146–16157.

Leyh, T.S., Taylor, J.C., and Markham, G.D. (1988). The sulfate acti-

vation locus of Escherichia coli K12: cloning, genetic, and enzymatic

characterization. J. Biol. Chem. 263, 2409–2416.

Li, G., and Zhang, X.C. (2004). GTP hydrolysis mechanism of Ras-like

GTPases. J. Mol. Biol. 340, 921–932.

Liu, C., Martin, E., and Leyh, T.S. (1994a). GTPase activation of ATP

sulfurylase: the mechanism. Biochemistry 33, 2042–2047.

Liu, C., Suo, Y., and Leyh, T.S. (1994b). The energetic linkage of GTP

hydrolysis and the synthesis of activated sulfate. Biochemistry 33,

7309–7314.

MacRae, I.J., Segel, I.H., and Fisher, A.J. (2000). Crystal structure of

adenosine 50-phosphosulfate kinase from Penicillium chrysogenum.

Biochemistry 39, 1613–1621.

MacRae, I.J., Segel, I.H., and Fisher, A.J. (2001). Crystal structure of

ATP sulfurylase from Penicillium chrysogenum: insights into the al-

losteric regulation of sulfate assimilation. Biochemistry 40, 6795–

6804.

Marcotte, E.M., Pellegrini, M., Ng, H.L., Rice, D.W., Yeates, T.O., and

Eisenberg, D. (1999). Detecting protein function and protein-protein

interactions from genome sequences. Science 285, 751–753.

Marzluf, G.A. (1997). Molecular genetics of sulfur assimilation in fila-

mentous fungi and yeast. Annu. Rev. Microbiol. 51, 73–96.

Moore, K.L. (2003). The biology and enzymology of protein tyrosine

O-sulfation. J. Biol. Chem. 278, 24243–24246.

Mougous, J.D., Green, R.E., Williams, S.J., Brenner, S.E., and Ber-

tozzi, C.R. (2002a). Sulfotransferases and sulfatases in mycobacte-

ria. Chem. Biol. 9, 767–776.

Mougous, J.D., Leavell, M.D., Senaratne, R.H., Leigh, C.D., Williams,

S.J., Riley, L.W., Leary, J.A., and Bertozzi, C.R. (2002b). Discovery of

sulfated metabolites in mycobacteria with a genetic and mass spec-

trometric approach. Proc. Natl. Acad. Sci. USA 99, 17037–17042.

Mougous, J.D., Petzold, C.J., Senaratne, R.H., Lee, D.H., Akey, D.L.,

Lin, F.L., Munchel, S.E., Pratt, M.R., Riley, L.W., Leary, J.A., et al.

(2004). Identification, function and structure of the mycobacterial

sulfotransferase that initiates sulfolipid-1 biosynthesis. Nat. Struct.

Mol. Biol. 11, 721–729.

Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement

of macromolecular structures by the maximum-likelihood method.

Acta Crystallogr. D Biol. Crystallogr. 53, 240–255.

Nakatsu, T., Kato, H., and Oda, J. (1998). Crystal structure of aspar-

agine synthetase reveals a close evolutionary relationship to class II

aminoacyl-tRNA synthetase. Nat. Struct. Biol. 5, 15–19.

Oda, Y., Huang, K., Cross, F.R., Cowburn, D., and Chait, B.T. (1999).

Accurate quantitation of protein expression and site-specific phos-

phorylation. Proc. Natl. Acad. Sci. USA 96, 6591–6596.

Okamoto, S., and Ochi, K. (1998). An essential GTP-binding protein

functions as a regulator for differentiation in Streptomyces coeli-

color. Mol. Microbiol. 30, 107–119.

Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction

data collected in oscillation mode. Methods Enzymol. 276, 307–326.

Pilloff, D.E., and Leyh, T.S. (2003). Allosteric and catalytic functions

of the PPi-binding motif in the ATP sulfurylase-GTPase system.

J. Biol. Chem. 278, 50435–50441.

Pinto, R., Tang, Q.X., Britton, W.J., Leyh, T.S., and Triccas, J.A.

(2004). The Mycobacterium tuberculosis cysD and cysNC genes

form a stress-induced operon that encodes a tri-functional sulfate-

activating complex. Microbiol. 150, 1681–1686.

Preininger, A.M., and Hamm, H.E. (2004). G protein signaling:

insights from new structures. Sci. STKE 2004, re3.

Renosto, F., Seubert, P.A., and Segel, I.H. (1984). Adenosine

50-phosphosulfate kinase from Penicillium chrysogenum. Purifica-

tion and kinetic characterization. J. Biol. Chem. 259, 2113–2123.

Rizzi, M., Nessi, C., Mattevi, A., Coda, A., Bolognesi, M., and Galizzi,

A. (1996). Crystal structure of NH3-dependent NAD+ synthetase

from Bacillus subtilis. EMBO J. 15, 5125–5134.

Roche, P., Debelle, F., Maillet, F., Lerouge, P., Faucher, C., Truchet,

G., Denarie, J., and Prome, J.C. (1991). Molecular basis of symbiotic

host specificity in Rhizobium meliloti: nodH and nodPQ genes en-

code the sulfation of lipo-oligosaccharide signals. Cell 67, 1131–

1143.

Saito, K. (2004). Sulfur assimilatory metabolism. The long and smell-

ing road. Plant Physiol. 136, 2443–2450.

Savage, H., Montoya, G., Svensson, C., Schwenn, J.D., and Sinning,

I. (1997). Crystal structure of phosphoadenylyl sulphate (PAPS) re-

ductase: a new family of adenine nucleotide alpha hydrolases.

Structure 5, 895–906.

Schwedock, J.S., Liu, C., Leyh, T.S., and Long, S.R. (1994).

Rhizobium meliloti NodP and NodQ form a multifunctional sulfate-

activating complex requiring GTP for activity. J. Bacteriol. 176,

7055–7064.

Scott, J.M., and Haldenwang, W.G. (1999). Obg, an essential GTP

binding protein of Bacillus subtilis, is necessary for stress activation

of transcription factor sigma(B). J. Bacteriol. 181, 4653–4660.

Song, H., Parsons, M.R., Rowsell, S., Leonard, G., and Phillips, S.E.

(1999). Crystal structure of intact elongation factor EF-Tu from

Escherichia coli in GDP conformation at 2.05 A resolution. J. Mol.

Biol. 285, 1245–1256.

Sprang, S.R. (1997). G protein mechanisms: insights from structural

analysis. Annu. Rev. Biochem. 66, 639–678.

Sun, M., Andreassi, J.L., 2nd, Liu, S., Pinto, R., Triccas, J.A., and

Leyh, T.S. (2005). The trifunctional sulfate-activating complex

(SAC) of Mycobacterium tuberculosis. J. Biol. Chem. 280, 7861–

7866.

Taguchi, Y., Sugishima, M., and Fukuyama, K. (2004). Crystal struc-

ture of a novel zinc-binding ATP sulfurylase from Thermus thermo-

philus HB8. Biochemistry 43, 4111–4118.

Molecular Cell122

Takai, Y., Sasaki, T., and Matozaki, T. (2001). Small GTP-binding pro-

teins. Physiol. Rev. 81, 153–208.

Terwilliger, T.C., and Berendzen, J. (1999). Automated MAD and MIR

structure solution. Acta Crystallogr. D Biol. Crystallogr. 55, 849–861.

Tesmer, J.J., Klem, T.J., Deras, M.L., Davisson, V.J., and Smith, J.L.

(1996). The crystal structure of GMP synthetase reveals a novel cat-

alytic triad and is a structural paradigm for two enzyme families. Nat.

Struct. Biol. 3, 74–86.

Townsend, D.M., Tew, K.D., and Tapiero, H. (2004). Sulfur containing

amino acids and human disease. Biomed. Pharmacother. 58, 47–55.

Ullrich, T.C., Blaesse, M., and Huber, R. (2001). Crystal structure of

ATP sulfurylase from Saccharomyces cerevisiae, a key enzyme in

sulfate activation. EMBO J. 20, 316–329.

Walker, K., and Olson, M.F. (2005). Targeting Ras and Rho GTPases

as opportunities for cancer therapeutics. Curr. Opin. Genet. Dev. 15,

62–68.

Wang, R., Liu, C., and Leyh, T.S. (1995). Allosteric regulation of the

ATP sulfurylase associated GTPase. Biochemistry 34, 490–495.

Wei, J., and Leyh, T.S. (1998). Conformational change rate-limits

GTP hydrolysis: the mechanism of the ATP sulfurylase-GTPase. Bio-

chemistry 37, 17163–17169.

Wei, J., and Leyh, T.S. (1999). Isomerization couples chemistry in the

ATP sulfurylase-GTPase system. Biochemistry 38, 6311–6316.

Wei, J., Liu, C., and Leyh, T.S. (2000). The role of enzyme isomeriza-

tion in the native catalytic cycle of the ATP sulfurylase-GTPase sys-

tem. Biochemistry 39, 4704–4710.

Winn, M.D., Isupov, M.N., and Murshudov, G.N. (2001). Use of TLS

parameters to model anisotropic displacements in macromolecular

refinement. Acta Crystallogr. D Biol. Crystallogr. 57, 122–133.

Yang, M., and Leyh, T.S. (1997). Altering the reaction coordinate of

the ATP sulfurylase-GTPase reaction. Biochemistry 36, 3270–3277.

Accession Numbers

Atomic coordinates have been deposited in the Protein Data Bank

under accession code 1ZUN.