mammalian heterotrimeric g-protein-like proteins in mycobacteria: implications for cell signalling...

Mammalian heterotrimeric G-protein-like proteins inmycobacteria: implications for cell signalling andsurvival in eukaryotic host cells

Sandeep Shankar, Vinayak Kapatral and A. M.Chakrabarty *

Department of Microbiology & Immunology, University ofIllinois College of Medicine, 835 South Wolcott Avenue,Chicago, IL 60612, USA.

Summary

Mammalian heterotrimeric GTP-binding proteins (Gproteins) are involved in transmembrane signallingthat couples a number of receptors to effectors medi-ating various physiological processes in mammaliancells. We demonstrate that bacterial proteins suchas a Ras-like protein from Pseudomonas aeruginosaor a 65 kDa protein from Mycobacterium smegmatiscan form complexes with human or yeast nucleosidediphosphate kinase (Ndk) to modulate their nucleo-side triphosphate synthesizing specificity to GTP orUTP. In addition, we demonstrate that bacteria suchas M. smegmatis or Mycobacterium tuberculosis har-bour proteins that cross react with antibodies againstthe a-, b- or the g-subunits of heterotrimeric G pro-teins. Such antibodies also alter the GTP synthesizingability of specific membrane fractions isolated fromglycerol gradients of such cells, suggesting that amembrane-associated Ndk–G-protein homologue com-plex is responsible for part of GTP synthesis in thesebacteria. Indeed, purified Ndk from human erythro-cytes and M. tuberculosis showed extensive complexformation with the purified mammalian a and b G-pro-tein subunits and allowed specific GTP synthesis,suggesting that such complexes may participate intransmembrane signalling in the eukaryotic host.We have purified the a-, b- and g-subunit homologuesfrom M. tuberculosis and we present their internalamino acid sequences as well as their putative homo-logies with mammalian subunits and the localizationof their genes on the M. tuberculosis genome. Usingoligonucleotide probes from the conserved regionsof the a- and g-subunit of M. tuberculosis G-proteinhomologue, we demonstrate hybridization of theseprobes with the genomic digest of M. tuberculosis

H37Rv but not with that of M. smegmatis , suggestingthat M. smegmatis might lack the genes present in M.tuberculosis H37Rv. Interestingly, the avirulent strainH37Ra showed weak hybridization with these twoprobes, suggesting that these genes might have beendeleted in the avirulent strain or are present in limitedcopy numbers as opposed to those in the virulentstrain H37Rv.

Introduction

Mycobacterium tuberculosis, the causative agent of thedisease tuberculosis, is known to evade the host immunesystem, replicate successfully in macrophages and lungtissues, and may lie in a dormant state for years withoutbeing completely eliminated (Wayne, 1994). Exactly howM. tuberculosis evades the host immune system andsurvives in the host tissues is largely unknown, althoughmany cell surface and metabolic components are believedto be involved (Chan and Kaufman, 1994). Recent attemptsin developing a genetic system in mycobacteria haveprovided significant preliminary information on the natureof some virulence genes (Jacobs and Bloom, 1994); forexample a gene allowing entry of M. tuberculosis tomouse macrophages and that demonstrates 27% homo-logy with the internalin gene of Listeria monocytogeneshas been described (Arruda et al., 1993). Mycobacterialgenes that appear to be induced in vivo or after phago-cytosis by macrophages have also been described (Kingerand Tyagi, 1993; Plum and Clark-Curtiss, 1994). In vivoexpression of genes in Salmonella typhimurium has beenhelpful in the isolation of a large number of genes thatallow enhanced intracellular growth or enhanced infectionin BALB/c mice and/or murine cultured macrophages(Heithoff et al., 1997) and such techniques might provideimportant clues regarding the role of specific gene pro-ducts in mycobacterial virulence. As survival of M. tuber-culosis in the alveolar macrophages is most probablydependent on the interference of the macrophage-inducedsignalling mechanism and the overall phagocytic processes,an alternative approach might be to look for the presenceof mammalian G proteins, which are known to be involvedin cell signalling and that might specifically be expressedunder conditions of growth in the macrophages. In thisreport, we describe the characterization of mammalian

Molecular Microbiology (1997) 26(3), 607–618

Q 1997 Blackwell Science Ltd

Received 4 June, 1997; revised 27 August 1997; accepted 27 August1997. *For correspondence. E-mail [email protected];Tel. (312) 996 4586; Fax (312) 996-6415.

m

G-protein-like GTP-binding proteins in M. smegmatis andM. tuberculosis that differ in the GTP-binding a-subunitcomponents. We demonstrate that such G-protein-likemycobacterial proteins form complexes with the bacterialnucleoside diphosphate kinase (Ndk) to modulate the levelsof GTP, and the M. tuberculosis, but not the M. smegmatis,Ndk–G-protein complexes resemble human erythrocyteNdk–G-protein complexes with respect to GTP synthesisand complex formation. Although G-protein-like proteins,cross-reactive to antibodies against Ga subunits, havebeen reported to be present in Escherichia coli cellsduring phagocytosis (Didenko et al., 1996) or in Stigma-tella aurantiaca, a myxobacterium (Derijard et al., 1989),their abilities for forming complexes with Ndk or function-ality for GTP synthesis have not been demonstrated.

Results and discussion

Bacterial proteins modulate GTP or UTP synthesisthrough complex formation with eukaryotic nucleosidediphosphate kinase

We have recently reported the presence of several pro-teins in Pseudomonas aeruginosa and M. smegmatis thatform complexes with the enzyme Ndk to modulate itsspecificity for nucleoside triphosphate (NTP) synthesis.Normally Ndk can generate all NTPs from a mixture ofnucleoside diphosphates (NDPs) and ATP. However,three different proteins from P. aeruginosa, namely pyru-vate kinase (Pk) (Sundin et al., 1996), a Ras-like proteinPra (Chopade et al., 1997) and the elongation factor Tu(EF-Tu, Mukhopadhyay et al., 1997) have been demon-strated to form complexes with Ndk and modulate itssubstrate specificity preferentially to GDP to allow pre-dominant formation of GTP. Similarly, at least four pro-teins P50, P60, P65 and P70 in M. smegmatis have beenshown to form complexes with M. smegmatis or P.aeruginosa Ndk to alter their NTP-synthesizing specificityto either GTP or CTP or UTP (Shankar et al., 1997). TheP70 protein has been shown to have pyruvate kinaseactivity, whereas the P50 protein has extensive N-terminalamino acid sequence homology with EF-Tu (Shankar etal., 1997). The nature of P60 that generates preferentiallyCTP and shows N-terminal amino acid sequence homo-logy with Hsp60, a heat shock protein, or P65 whoseN-terminal sequence has not been determined, is unknown(Shankar et al., 1997). In human cells, Ndk exists in twoforms, Nm23-H1 and Nm23-H2, which are highly homo-logous. Nm23-H1 has been shown to be a suppressor ofsome forms of cancer, particularly breast cancer, meta-stasis (De la Rosa et al., 1995). As GTP is a signallingmolecule and GTP-binding proteins, particularly theheterotrimeric G proteins are known to be involved in ionchannel opening, hormonal effect, developmental angio-genesis, etc. (Gilman, 1987; Kaziro et al., 1991; Offermanns

et al., 1997), it was of interest to determine if any of thebacterial proteins that are known to form complexes withNdk from diverse genera, such as Pseudomonas andMycobacteria (Shankar et al., 1997), may also form com-plexes with eukaryotic Ndks such as human or yeast Ndk.Among the seven bacterial proteins tested, two proteins,Pra from P. aeruginosa (Chopade et al., 1997) and P65

from M. smegmatis (Shankar et al., 1997), specificallyaltered the NTP-synthesizing activity of either human oryeast Ndk to GTP or UTP (Fig. 1). Recently, UTP hasbeen demonstrated to activate G protein coupled receptorssuch as the P2Y4 family of receptors (Charlton et al., 1996),or to trigger the activation of the stress-activated proteinkinase module in rat mesangial cells by a pathway inde-pendent of protein kinase C but requiring a pertussistoxin-sensitive G protein and tyrosine kinase activation(Huwiler et al., 1997). As Pra is a Ras-like protein, andthe GTP-binding Ras is known to be involved in cell signal-ling and oncogenesis (Barbacid, 1987), the ability of Pra tomodulate the specificity of human Ndk to preferentiallysynthesize GTP appeared to suggest that human GTP-binding proteins similar to Ras might be involved in com-plexing with human Ndk to modulate its NTP-synthesizingspecificity to GTP.

Apart from low molecular weight G proteins such asRas, another family of GTP-binding proteins is the hetero-trimeric G proteins, involved in transmembrane signalling(Gilman, 1987; Kaziro et al., 1991). The ability of bacterialproteins to form complexes with human or other eukaryo-tic Ndks, altering their NTP-synthesizing specificity to thesignalling molecule GTP, also suggested the potentialinvolvement of heterotrimeric G proteins, having a-, b-and g-subunits, in prokaryotic interference with eukaryotichost cell signalling mechanism. As M. tuberculosis and M.smegmatis have extensive similarity in the nature andcomposition of the cell wall and cell envelope that arebelieved to play a role in the survival of M. tuberculosis,but not of avirulent M. smegmatis, in the macrophages(Besra and Chatterjee, 1994; Chan and Kaufman, 1994),it is possible that M. smegmatis may have a mechanismthat precludes its ability to interfere with the signallingmechanism of the macrophages and is therefore suscep-tible to their killing action, whereas M. tuberculosis caneffectively interfere and survive. To examine if M. smeg-matis and M. tuberculosis may show differences in thenature of complexes leading to GTP synthesis, weattempted to characterize GTP-synthesizing complexesfrom M. smegmatis as well as virulent and avirulent strainsof M. tuberculosis.

Multiple complexes allow formation of GTP in bothM. tuberculosis as well as M. smegmatis

Our initial experiments demonstrated that crude extracts

Q 1997 Blackwell Science Ltd, Molecular Microbiology, 26, 607–618

608 S. Shankar, V. Kapatral and A. M. Chakrabarty

of M. smegmatis have a high phosphatase or ATPase/GTPase activity that interfered with our assay for detect-ing the synthesis of NTPs from [g-32P]-ATP and non-radioactive NDP mixtures (CDP/GDP/UDP), as describedpreviously (Shankar et al., 1996; 1997). To separate theNTP-synthesizing Ndk and its complexes from the phos-phatase activities, we isolated the membrane prepara-tions from M. smegmatis, M. tuberculosis H37Ra andM. tuberculosis H37Rv, centrifuged them in a glycerolgradient to separate the putative high molecular weightcomplexes from other free enzymes and conducted an

NTP-synthesizing assay with each fraction, as describedpreviously (Shankar et al., 1997). The results in Fig. 2Aindicate that the bulk of the phosphatase or ATPase activ-ity sedimented at 10% glycerol, whereas a GTP synthesiz-ing activity was present in the 60% glycerol fraction. Todetermine if indeed different complexes were formed inM. smegmatis and M. tuberculosis, but the glycerol gradi-ent was unable to dissociate them, we saturated themembrane complexes with aluminium ammonium sulphate(aluminium alum), which is known to dissociate various pro-tein–protein aggregates, and re-ran the complexes in aglycerol gradient. Membrane complexes isolated from M.smegmatis cells that had been grown in low pH with hydro-gen peroxide showed the presence of GTP-synthesizingcomplexes that sedimented at glycerol percentages of30, 40, 50 and 60 (Fig. 2B). A similar pattern wasobserved for M. tuberculosis H37Ra and M. tuberculosisH37Rv cells that had been grown to late stationary phase(Fig. 2B). Interestingly, M. smegmatis cells that had beengrown to late stationary phase without the added hydrogenperoxide or lowering the pH of the growth media did notdemonstrate the presence of complexes at glycerol per-centages of 40 and 60 (data not shown).

Specificity of GTP synthesis by the 60% glycerolfraction (F60) is abolished by antibodies to certainmammalian G proteins

To gain an insight into the nature of the various GTP-synthesizing glycerol fractions, we conducted an evalu-ation of the susceptibility of the GTP synthesizing potentialof such fractions with antibodies directed against P.aeruginosa Pk, Pra, EF-Tu, etc. Although most of the frac-tions showed susceptibility to such antibodies, the 60%glycerol fraction (F60) not only did not respond to such anti-bodies but gave some curious and interesting results withrespect to an alteration of GTP synthetic specificity in pre-sence of antibodies directed against specific anti-a- andanti-b-subunits of the heterotrimeric mammalian GTP-binding (G) proteins. Such G proteins consist of an a-, b-and g-subunit and are characterized by the diversity ofthe a-subunits, whereas the b-subunit is fairly conserved(Gilman, 1987; Kaziro et al., 1991). The a-subunits havebeen divided into four families (Gas, Gai /o, Gaq, and Ga12)based on sequence similarities of their genes (Simon etal., 1991). It was observed that whereas GTP synthesisby the F60 fraction was inhibited by anti-Ndk antibody(Fig. 3, lanes 6, 12 and 18), the specificity of GTP synthe-sis was altered to all NTP synthesis in the presence of anti-bodies directed against Gai in the case of M. smegmatis(Fig. 3, lane 2) or Gb (lane 4). A change in specificitywas observed in the case of M. tuberculosis strains aswell, but this effect was brought about by antibodiesagainst Gai-3 and not Gai (lanes 9 and 15, Fig. 3). This is


Fig. 1. Effect of the proteins from P. aeruginosa and M.smegmatis reported to modulate the nucleoside diphosphate kinaseactivity in these two systems on the Ndks from yeast as well ashuman erythrocytes. An aliquot (10 pmol) of either the humanerythrocyte Ndk or the yeast Ndk were incubated in a final volumeof 20 ml reconstituted in 50 mM Tris-HCl, pH 7.6, 10 mM MgC12

along with 1 mM of each of the NDPs and 10 mCi of [g-32P]-ATP.The proteins tested for their effect on the Ndks were each presentat 10 pmol. The reaction was started by the addition of the NDPsand terminated by the addition of 4× stop buffer. The lanes in thefigure have been marked appropriately. ‘cont.’ indicates controlhuman or yeast Ndk preparations without added protein; Pk, Praand EF-Tu are from Pseudomonas aeruginosa and P70, P50, P60

and P65 are from Mycobacterium smegmatis.

Mammalian G-protein-like proteins in mycobacteria 609


Fig. 2. A. Separation of the non-specific phosphatase/ATPase activity by glycerol gradient centrifugation of the M. smegmatis membranepreparation. Glycerol gradients were prepared as described previously (Shankar et al., 1997). An aliquot (10 ml) of each of the gradientfractions were analysed for NDP kinase activity as described previously (Shankar et al., 1997).B. Glycerol gradient analysis of the GTP-synthesizing membrane complex by fractionation on a glycerol gradient saturated with 1% aluminiumalum (aluminium ammonium sulphate). The gradient was prepared as described previously but aluminium alum was present in each of thegradient solutions. The figure indicates the source of the membrane preparation being fractionated along with the percentage of glycerol (5,10, 20, 30, 40, 50, and 60% corresponding to lanes 1–7, 8–14 and 15–21) at each stage. M. tb H37Rv and M. tb H37Ra represent thevirulent and avirulent forms of Mycobacterium tuberculosis. Lane 22 represents a typical Ndk reaction showing the formation of all three NTPs(GTP, CTP, UTP) from NDPs and [g-32P ]-ATP.

Fig. 3. Effect of the mammalian G-proteinantibodies on specific GTP-synthesizingactivity of the 60% glycerol fractions (F60).The reaction for NTP synthesis wasconducted as previously described (Shankaret al., 1997). Antibodies were added beforethe addition of the NDPs or the radioactive[g-32P]-ATP. Lane 1, M. smegmatis F60

complex; lane 2, þ anti-Gai antibody; lane 3,þ anti-Gai-3 antibody; lane 4, þ anti-Gb

antibody; lane 5, þ anti-Pk antibody; lane 6,þ anti-Ndk antibody; lane 7, M. tuberculosisH37Ra F60 complex; lane 8, þ anti-Gai

antibody; lane 9, þ anti-Gai-3 antibody; lane10, þ anti-Gb antibody; lane 11, þ anti-Pkantibody; lane 12, þ anti-Ndk antibody; lane13, M. tuberculosis H37Rv F60 complex; lane14, þ anti-Gai antibody; lane 15, þ anti-G ai-3

antibody; lane 16, þ anti-Gb antibody; lane17, þ anti-Pk antibody; and lane 18, þ

anti-Ndk antibody.


reminiscent of a change of GTP-synthesizing specificity ofP. aeruginosa membrane-associated Ndk by anti-Pk anti-bodies to all NTPs because of the dissociation of Pk fromthe complex (Sundin et al., 1996). The specificity for GTPsynthesis by the F60 fractions from all the three strains waschanged in the presence of antibodies directed against Gb

(Fig. 3, lanes 4, 10 and 16). Evidently, GTP syntheses bythe membrane-derived F60 fraction of M. tuberculosisH37Ra, Rv as well as M. smegmatis cells are probablythe result of complex formation between Ndk and mam-malian G-protein-like proteins present in the membranesof these cells.

Western blotting indicates the presence of proteinspecies cross-reacting with mammalian G proteinantibodies in mycobacteria

Heterotrimeric mammalian G proteins are involved intransmembrane signalling, a characteristic demonstratedby macrophages while they engulf foreign objects includ-ing intracellular pathogens. An interesting question was ifindeed homologues of such G-protein subunits are pre-sent in the virulent and avirulent strains of M. tuberculosis,such as strains H37Rv and H37Ra. Attempts to detect aband by Western blotting responsive to Gai antibodieswere unsuccessful in the extracts and membranes ofeither M. smegmatis or M. tuberculosis strains (data notshown), presumably because of a low cellular level ofsuch proteins or a low response to the antibodies. How-ever, Western blots using anti-Gb antibodies demonstratedthe presence of cross-reactive bands in the crude extracts

of M. smegmatis, as well as the virulent and avirulentstrains of M. tuberculosis (data not shown). Interestingly,the crude extract of M. smegmatis demonstrated the pre-sence of two bands cross-reactive to anti-Gb antibodies,whereas the crude extracts of the two M. tuberculosisstrains showed the presence of a single band, as didextracts from other pathogens such as P. aeruginosa(data not shown).

As the heterotrimeric G proteins are known to bind GTPthrough the a-subunit, it was of interest to us to examine ifany of the M. smegmatis or M. tuberculosis crude extractswould demonstrate the presence of proteins cross-reac-tive to antibodies against a number of a-subunits of mam-malian G proteins. As previously mentioned, no visiblebands using Western blotting were detected in any ofthe strains with antibodies directed against Gai . However,Western blotting using anti-Gai-3 antibodies (epitope corre-sponding to amino acids 345–354) showed the presenceof cross-reactive bands in crude extracts of M. tubercu-losis strains H37Ra and H37Rv (lanes 3 and 4, Fig. 4)but not in the extracts of M. smegmatis (Fig. 4, lane 2).This correlates very well with our observation that anti-bodies directed against the Gai-3 subunit changed thespecificity of the 60% glycerol fraction for GTP synthesisin the case of M. tuberculosis but not in the case of M.smegmatis. Interestingly, such a band was absent in thecrude extracts of a non-pathogenic P. putida strain (lane7, Fig. 4) but present in pathogenic P. aeruginosa orStaphylococcus aureus extracts (Fig. 4, lanes 5 and 6).Whereas the M. tuberculosis or S. aureus proteins cross-reactive to anti-Gai-3 antibodies had a size of about 53 kDa


Fig. 4. Presence of mammalian G-protein-like proteins in various bacterial pathogens. Mycobacterial cells were prepared from late stationaryphase using the modified Middlebrook media enriched with OADC (Difco laboratories). Cells were harvested by centrifugation at 7000 × g,suspended in 3 vols (w/v) buffer A and lysed by French Press at a dial setting of 16000 psi. Protein was estimated by the Bradford methodand 10 mg per lane was used. The other bacterial extracts were prepared in the same manner. Western blot analysis was performed usingthe ECL method of detection (Amersham) according to the manufacturer’s instructions. The control mammalian G-protein was obtained fromSanta Cruz Biologicals and so were the antibodies. Probing with anti-Gai-3 antibodies: lane 1, Control G ai-3 polypeptide, 10 ng; lane 2, 10 mgM. smegmatis extract; lane 3, 10 mg M. tuberculosis H37Ra extract; lane 4, 10 mg M. tuberculosis H37Rv extract; lane 5, 10 mg P. aeruginosaextract; lane 6, 10 mg S. aureus extract, and lane 7, 10 mg P. putida extract.


(Fig. 4, lanes 3, 4 and 6), the P. aeruginosa protein had asize of about 77 kDa (lane 5).

The alteration of the specificity for GTP synthesis,resulting in non-specific NTP synthesis by the M. tuber-culosis F60 fraction in the presence of anti-Gai-3 and anti-Gb antibodies (Fig. 3), also raised an interesting questionif the specificity for GTP synthesis by the F60 fraction isdue to complex formation of Ndk with M. tuberculosisGai-3 and Gb homologues of mammalian heterotrimericG proteins. Such modulation of GTP synthesis may haveimportant consequences in the macrophage signal trans-duction system during engulfment of M. tuberculosis bythe macrophages. In addition, it was important to deter-mine whether such mammalian G-protein subunits them-selves interact with M. smegmatis or M. tuberculosisNdk to modulate its NTP-synthesizing specificity.

Mammalian Gb and Gai-3 polypeptides influencespecificity of NTP synthesis by Ndks from M.tuberculosis and human erythrocytes but not that fromM. smegmatis

We determined the nature of NTP synthesis by M. smeg-matis, M. tuberculosis and human Ndk in the presence ofvarious purified mammalian G-protein subunits. As a con-trol, P. aeruginosa 16 kDa and 12 kDa Ndks were alsotested for their ability to complex with such mammalianproteins with an attendant change in NTP synthesizingspecificity. Neither P. aeruginosa nor M. smegmatis Ndkhad any altered substrate specificity when various G-pro-tein subunits were added in an NTP-synthesizing assay

(Fig. 5, lanes 1–7 and lanes 22–27). That such Ndkswere enzymatically active was demonstrated not only bytheir ability to synthesize all NTPs from [g-32P]-ATP andNDPs but also by their ability to specifically synthesizeGTP when complexed with Pk (Sundin et al., 1996) orP50 (Shankar et al., 1997) (Fig. 5, lanes 28 and 29). Inte-restingly, the change of specificity from NTP synthesisto GTP synthesis by purified M. tuberculosis or humanerythrocyte Ndk was accomplished in the presence of amixture of Gb and Gai-3 (Fig. 5, lanes 13 and 20), suggest-ing that specific G-protein subunits of mammalian origincan form complexes with M. tuberculosis or human ery-throcyte Ndk to change their NTP synthesizing specificityto GTP. As b- and g-subunits of heterotrimeric G proteinsform a tight complex, it is likely that the Gb subunits alsoharbour small amounts of the g-subunit.

Glycerol gradient centrifugation confirms complexformation between various Ndks and the G-proteinsubunits

We also demonstrated experimentally, using glycerol gra-dient separation techniques as described previously(Shankar et al., 1997), complex formation between M.tuberculosis and human Ndk and Gb or Gai-3, but notGai , proteins as well as Pk (Fig. 6A and B). Complexformation allowed a shift in the migration of Ndk from 5%glycerol fraction to a higher percentage glycerol fraction,as detected by anti-Ndk antibody. A control bovine serumalbumin (BSA) showed no complex formation. In contrast,


Fig. 5. Extent of reconstitution of the GTPsynthesizing activity by the combination ofNdks with the mammalian G proteins. Proteinswere mixed before the addition of NDPs and[g-32P ]-ATP. An aliquot (10 pmol) of each ofthe proteins was used when assayedseparately and when used in combination. Forthe combination assay, the proteins weremixed in separate vials and 10 pmol of themixture was used and the assays performedas usual. Lane 1. M. smegmatis Ndk; lane 2,þ Gb ; lane 3, þ Gai ; lane 4, þ Gai-3 ; lane 5,þ Gbþ Gai ; lane 6, þ Gb þ Gai-3; lane 7,þ Gai þ Gai-3 ; lane 8, M. tuberculosis H37RvNdk; lane 9, þ Gb ; lane 10, þ Gai ; lane 11,þ Gai-3 ; lane 12, þ Gb þ Gai ; lane 13,þ Gbþ Gai-3 ; lane 14, þ Gai þ Gai-3 ; lane 15,Human erythrocyte Ndk; lane 16, þ Gb; lane17, þ Gai ; lane 18, þ Gai-3; lane 19,þ Gb þ Gai ; lane 20, þ Gb þ Gai3 ; lane 21,þ Gai þ Gai-3 ; lane 22, P. aeruginosa Ndk16 kDa; lane 23, þ Gb þ Gai-3; lane 24,þ Gb þ Gai ; lane 25, P. aeruginosa Ndk12 kDa; lane 26, þ Gb þ Gai ; lane 27,þ Gb þ Gai-3; lane 28, þ Pk; lane 29, M.smegmatis Ndk þ P50 (homologous to EF-Tu).


M. smegmatis Ndk could form a complex with Pk and aweak complex with Gb and Gai mixture (Fig. 6C), but notwith Gb or Gai alone. Thus, the alteration of M. tubercu-losis or human Ndk’s substrate specificity towards GDP(Fig. 5, lane 13 and 20) is most probably the result ofcomplex formation between the mammalian G-proteinsubunits and human Ndk or the prokaryotic M. tuber-culosis H37Rv Ndk.

To investigate the identity of the mycobacterial G-pro-tein homologues further, we decided to attempt immuno-precipitation of these proteins using the correspondingantibodies against mammalian G-protein subunits.

Immunoprecipitation of the homologues of themammalian Gai-3 , Gb and Gg from M. tuberculosis andinternal amino acid sequence determination

Using antibodies against Gai-3 and Gb, we recovered elec-trophoretically homogeneous polypeptides of sizes 53 kDafor the Ga homologue and 47 kDa for the Gb homologue.Attempts at N-terminal sequence analysis of the polypep-tides were not successful but internal sequence analysisyielded useful information. As indicated in Fig. 7A, thedetermined sequence for the polypeptide precipitated bythe Gai-3 antibodies had 100% identity with the translated


Fig. 6. Complex formation between various Ndks and G-proteinsubunits. Individual proteins were used in 10 pmol amounts beforeloading. Protein mixtures were used at a final amount of 10 pmoleach. The glycerol gradient separation of complexes has beendescribed previously (Shankar et al., 1997). The detection methodused alkaline phosphatase conjugated second antibody. Laneshave been marked as such. Numbers on top are glycerol percen-tages.A. purified M. tuberculosis H37Rv Ndk (TbRv-Ndk), in presence ofPk, various G-protein subunits or bovine serum albumin (BSA) as acontrol. Aliquots of various glycerol fractions were tested for thesedimentational shift of Ndk, using anti-Ndk antibody.B. As above with human erythrocyte Ndk (H-Ndk).C. As above with M. smegmatis Ndk (Msm-Ndk).


sequence of gmt 43 cosmid clone (Sanger Center, unfin-ished sequence) and limited homologies with gmt 179and gmt 990. The sequence GGPGXGK was identifiablein a G-protein a-subunit reported from the molluscan ner-vous system (Knol et al., 1995). A part of the sequencemeets the minimal consensus requirements of GXXXXGK,as deduced for a variety of nucleotide-binding proteins(Dever et al., 1987). We were also able to determine aninternal sequence for the M. tuberculosis polypeptideimmunoprecipitated by the anti-mammalian Gb antibodies.As Fig. 7B indicates, the translated sequence of cosmidclone gmt 411 showed 100% identity to the internal sequ-ence of the polypeptide isolated as part of the immuno-precipitate. A striking similarity was also observablebetween this sequence and a sequence stretch thatdirects specificity of recognition between b1-, b2- withthe g1- or g2-subunits in the bovine brain (Spring andNeer, 1994). The sequence defines part of a small regionin the g-subunit from bovine brain that determines selec-tivity of binding to the b1- or b2-subunit. It should benoted that in the mammalian system, although amino

acid sequences for b1 and b2 are 90% identical, g1forms dimers with b1 but not with b2 (Schmidt et al.,1992). Such selective dimer formation may contribute tothe regulation of G-protein-mediated signal transduction.

Using antibodies against the g1 mammalian G-proteinsubunit, we were able to immunoprecipitate a 12 kDa poly-peptide whose internal sequence showed 100% identityto the translated sequence in the cosmid clone gmt 11(Fig. 7C). A limited homology was recognizable betweenthis sequence and a G-protein g-subunit previously reportedin bovine brain (Gautam et al., 1989). The primary sequenceof this G-protein g-subunit has been reported to havefunctional homologies to the transducin g (Tg)-subunit aswell as to the functional domains of mammalian Ras pro-teins (Gautam et al., 1989).

To ascertain the presence of genes for Ga and Gg

homologues of G-protein-like proteins in mycobacteria,we designed an oligonucleotide probe for Ga and a probefor Gg using conserved DNA sequences of these two M.tuberculosis G-protein subunit homologues. These probeswere used in DNA–DNA hybridization experiments


Fig. 7. A. Sequence analysis of thepolypeptide recovered in immunoprecipitationwith anti-Gai-3 antibodies.B. Sequence analysis of the polypeptiderecovered in the immunoprecipitationexperiment with anti-Gb antibodies.C. Sequence analysis of the polypeptiderecovered in the immunoprecipitationexperiment with anti-Gg antibodies. In allcases, homology search was conducted usingThe Institute for Genome Research (TIGR)tblastN database.


(Sambrook et al., 1989) using genomic digests of M.tuberculosis strain H37Rv, its non-virulent derivativeH37Ra and M. smegmatis. The results in Fig. 8 indicatethe presence of both Ga and Gg genes in M. tuberculosisH37Rv. It is interesting that the genomic digests of boththe avirulent M. tuberculosis H37Ra and the M. smeg-matis showed weak hybridization with Ga- and Gg-specificDNA probes. As Ga-specific antibodies cross-reacted withboth M. tuberculosis H37Rv and H37Ra extracts but notwith M. smegmatis cell extract (Fig. 4), it is not clear if thestrain H37Ra might have a less homologous Ga gene(s)whose product is cross-reactive to antibodies against mam-malian Gai–3. Whether the loss of the gene for Ga and Gg

may have contributed, at least partially, to the avirulenceof the strain H37Ra is, of course, unknown. It would be inte-resting to see if null mutations in the genes for a, b and g

G-protein homologues may abrogate the ability of M. tuber-culosis to survive and proliferate in the macrophages.

The physiological significance of the presence of humanG-protein subunit-like proteins in prokaryotes, and particu-larly the ability of purified mammalian G-protein subunitsto interact with human or M. tuberculosis Ndk, specificallygenerating GTP, could be substantial. In human, Ndk, par-ticularly Nm23-H1, is known to be a suppressor of cancermetastasis (De la Rosa et al., 1995). Transfection studies,in which human nm23-H1 cDNA was expressed in themetastatic human MDA-MB-435 breast carcinoma cellline, indicate that nm23-H1 suppresses in vivo metastaticpotential by 50–90% (Steeg et al., 1993). Similarly, acorrelation between the mutation in nm23-H1 gene andmetastasis in colorectal adenocarcinomas has beenobserved (Wang et al., 1993). Very little is known abouthow Nm23-H1 may suppress colorectal or breast cancermetastasis. Our demonstration that prokaryotic proteins(Fig. 1) as well as mammalian G-protein subunits (Fig. 5)form complexes with human Ndk, leading to the formationof signal transducing GTP molecules, may provide inte-resting clues to the role of specific transmembrane signal-ling pathways in cancer metastasis, in which GTP-binding

proteins such as Ras play a role (Kohn and Liotta, 1995).Similarly, the presence of mammalian G-protein subunit-like proteins in M. tuberculosis, and the ability of mam-malian G-protein subunits to form complexes with M.tuberculosis Ndk, altering its NTP-synthesizing specificityto GTP, may provide important information on the poten-tial role of such proteins in evading the immune systemthrough an altered signalling process in the macrophages.

Our demonstration that Ndk from M. tuberculosis formscomplexes with mammalian heterotrimeric G proteins isreminiscent of eukaryotic Ndks that have been shown tointeract with the corresponding G proteins; for exampleNdk-dependent phosphorylation of GDP on the hetero-trimeric GO was reported with adenosine 58(3-0-thio) tri-phosphate as the phosphate donor. Another purified Gprotein, GS, was readily ribosylated by cholera toxin inthe presence of Ndk, ATP and an ADP-ribosylation factor(Kikkawa et al., 1990). In the cellular slime mould Dictyo-stelium discoideum, extracellular cAMP induces an increaseof phospholipase C activity via a surface cAMP receptorand G proteins. Ndk was shown to be associated with theDictyostelium membranes that produced GTP, therebyactivating G proteins as monitored by altered G proteinreceptor interaction and the activation of the effectorenzyme phospholipase C (Bominaar et al., 1993). Thewasp venom, mastoparan, was shown to be an activatorof reconstituted pertussis toxin-sensitive G proteins andNdk. G-protein activation was demonstrated in the mem-branes of human neuroblastoma, erythroleukaemia andother cell lines by Ndk in response to mastoparan, andpertussis toxin catalysed ADP ribosylation of a-subunitswas postulated to inhibit the transfer of GTP from Ndk toG proteins (Klinker et al., 1996). Given such a role of Ndkin G-protein-mediated transmembrane signalling, it wouldbe of interest to examine the role of the G-protein-like pro-teins in human and animal pathogens, including M. tubercu-losis. Such studies are currently underway in our laboratory.

Experimental procedures

Growth of M. tuberculosis H37Rv, H37Ra and M.smegmatis

Cells were grown in the Middlebrook 7H9 broth supplementedwith 2% glycerol and 0.02% Tween 80 and prepared accord-ing to the manufacturer’s instructions (Difco Laboratories). Analiquot (4.7 g) of the Middlebrook 7H9 medium was added to900 ml of distilled water along with 2 ml of Tween 80 and20 ml of glycerol. Sterilization was carried out at 15 psi for20 min and the media were cooled to 458C. A 100 ml sampleof the Difco OADC enrichment supplement was added tothe 7H9 broth and the media were made ready for inoculation.The mycobacterial cultures were routinely stored as glycerolstocks at ¹708C and recovered by plating on Luria–Bertani(LB) medium supplemented with 0.02% Tween 80. A 24-h-oldplate was used for the inoculation of a single colony into


Fig. 8. DNA–DNA hybridizations showing the presence or absenceof the genes for Ga and Gg in M. tuberculosis H37Ra, H37Rv andM. smegmatis. About 200 ng of mycobacterial DNA was spottedonto a nitrocellulose paper and was probed with 32P-labelled M.tuberculosis Ga and Gg specific probes.


100 ml of the 7H9 broth and propagated with vigorous agita-tion (300 r.p.m.) for 36 h. An aliquot (1 ml) of the inoculumthus prepared was added to 100 ml of fresh medium and pro-pagated for another 36–44 h to obtain an absorbance of 2.2 at650 nm. For the low pH/H2O2 stress effects, the culture nor-mally grown to an absorbance of 1.1 (<24 h) was adjustedwith respect to the pH by the addition of orthophosphoricacid until the final pH of the culture was 6.4. 500 ml sampleof the culture was mixed with 500 ml of fresh medium pre-adjusted to a pH of 6.4 and containing H2O2 at a concentrationof 0.2% (final concentration 0.1%). The cells were harvestedafter 8 h of continuous aeration at which time the absorbanceof the culture was in the range of 1.9–2.2 at 650 nm.

Preparation of cell extracts and membrane fractions

Cells from different stages of growth or from different growthconditions were processed similarly. The cell cultures werepelleted by centrifugation at 6500 × g for 20 min at 48C. Thecells were suspended in 3 vols (w/v) of buffer A (Shankar etal., 1997) and lysed by French Press (SLM Aminco) at adial setting of 18000 psi. For isolating the membrane fraction,the lysate was centrifuged at 7000 × g for 20 min at 48C, thepellet discarded and followed by a 55 000 × g centrifugationat 48C for 1 h. The pellet thus obtained was referred to asthe membrane preparation and was resuspended in the origi-nal volume. These membrane preparations were used for theglycerol gradient experiments involving aluminium alum. Thesupernatant of the 55 000 × g spin was considered the celllysate supernatant and was used as the starting material forthe Ndk isolations. The Western blot analysis was conductedusing the crude extracts without the removal of the membranes.

Assay for the biochemical activity of nucleosidediphosphate kinase

Nucleoside triphosphate synthesis by Ndk was carried out asdescribed previously (Sundin et al., 1996; Shankar et al.,1997). The assay was reconstituted in a final volume of 20mlof buffer A containing the nucleoside diphosphates (CDP,GDP, UDP) at a final concentration of 1.0 mM in the assay.An aliquot (10 mCi) of [g-32P]-ATP (3000 Cimmol¹1) was usedas a phosphodonor per assay. An aliquot (10 pmol) of Ndkwas used where indicated to start the reaction and was car-ried out for 20 s at room temperature followed by the additionof 8 ml of 4× SDS-Stop buffer. The proteins to be analysed fortheir effect on the biochemical activity of Ndk were all added atequimolar concentrations and the system preincubated in icebefore the addition of the nucleotides. One microlitre of thereaction products was spotted on the lower end of a polyethyl-eneimine thin layer chromatography (PEI-TLC) plate andchromatographed using 0.75 M KH2PO4, pH 3.75, buffer asthe mobile phase. After the buffer had run to the end, theplate was dried and the reaction products visualized by auto-radiography (Shankar et al., 1997).

Glycerol gradient analysis in the absence and presenceof ammonium alum

Glycerol gradient analysis was performed as described pre-viously (Sundin et al., 1996; Shankar et al., 1997). Glycerol

solutions of percentages 60, 50, 40, 30, 20, 10 and 5 were pre-pared using buffer A as the diluent. A 25 ml sample of a 50%solution of aluminium alum was added to each 1 ml batch ofthe gradient. The protein sample to be analysed was layeredon top of the 5% glycerol gradient band. The tubes were cen-trifuged at 40 000 r.p.m. (55 000 × g) for 1 h and the fractionswere withdrawn in batches of 1 ml using a syringe. Membranepreparations were used in 10 mg amounts to avoid trailingeffects. Pure protein preparations were premixed in 10 pmolamounts where indicated and kept in ice for 30 min beforeloading on the gradient. Two microlitres of the sample wasspotted on a nitrocellulose filter, dried and probed with anti-Ndk antibodies diluted 1:500. The signal was detected usingthe alkaline phosphatase conjugated goat anti-rabbit second-ary antibody diluted 1:5000 according to the manufacturer’srecommendations (Sigma Chemical).

Purification of the nucleoside diphosphate kinaseactivities from M. tuberculosis H37Rv as well as H37Ra

An aliquot (100 mg) of the mycobacterial pyruvate kinase wasmixed with 5 ml of Sepharose 4B matrix activated usingcyanogen bromide (Sigma Chemical) suspended in an equalvolume of 0.5 M NaHCO3 buffer, pH 8.5. The protein was con-tinuously mixed with the gel matrix for 16 h at 48C by end-onshaking. Glycine was added to a final concentration of 1 Mand the shaking continued for another 1 h at room tempera-ture. The gel was washed sequentially with 100 vols each of0.5 M sodium acetate, pH 4.5, 4.0 M urea, 0.5 M NaHCO3,pH 8.3 and buffer A. Mycobacterial crude extracts were pre-pared as described above. Five milligrams of extracts wasloaded onto 1 ml batches of the matrix at a flow rate of0.5 ml min¹1. The column was washed extensively with bufferA until no effluent protein could be detected. The gel wassequentially washed with 1 ml of increasing concentrationsof NaCl (25 mM, 50 mM, 100 mM and 150 mM respectively)that had been prepared in buffer A. The activity of the elutedfractions was measured using the TLC method describedabove with 10 ml of the collected fraction and the purity ofthe fraction containing the NDP kinase activity was assessedby SDS–PAGE analysis.

Purification of the G-protein homologues from M.tuberculosis

M. tuberculosis H37Rv cell suspension was prepared asdescribed earlier. The extract was centrifuged at 11 000 × gat 48C for 15 min, the supernatant was transferred to a freshtube and recentrifuged at 45 000 × g for 1 h. The pellet obtainedwas resuspended in the original volume (this exercise wasconducted using 2 g of cells suspended in 7 ml of buffer A)of the same buffer but containing 0.1% Tween 20 to facilitateproper suspension. Antibodies corresponding to the Ga- , Gb-

or the Gg-subunits were added into the suspension at a finaldilution of 1:500 and mixed end-on at room temperature for1 h; 0.8 g of protein A sepharose beads (Sigma Chemical)was then added to the suspension, mixed thoroughly andleft on ice for 30 min. The suspension was centrifuged at5000 × g for 10 min. The sepharose beads were washedsequentially with 100 ml each of 2 M and 4 M MgCl2. Samples



were analysed for purity on a standard SDS-polyacrylamidegel and submitted for sequence analysis.

Acknowledgements

This work was supported by Public Health Service Grants AI31546-04 and AI 16790-16 from the National Institutes ofHealth. We are grateful to Dr John Belisle of the ColoradoState University, Fort Collins for supplying us with largeamounts of M. tuberculosis cells.

References

Arruda, S., Bomfim, G., Knights, R., Huima-Byron, T., andRiley, L.W. (1993) Cloning of an M. tuberculosis DNA frag-ment associated with entry and survival inside cells.Science 261: 1454–1457.

Barbacid, M. (1987) Ras genes. Annu Rev Biochem 56:779–827.

Besra, G.S., and Chatterjee, D. (1994) Lipids and carbo-hydrates of Mycobacterium tuberculosis. In Tuberculosis:Pathogenesis, Protection and Control. Bloom, B.R. (ed.).Washington, DC: American Society for Microbiology, pp.2285–2306.

Boominaar, A.A., Molijn, A.C., Pestel, M., Veron, M., and VanHaastert, P.J.M. (1993) Activation of G-proteins by recep-tor-stimulated nucleoside diphosphate kinase in Dictyo-stelium. EMBO J 12: 2275–2279.

Chan, J., and Kaufmann, H.E. (1994) Immune mechanismsof protection. In Tuberculosis: Pathogenesis, Protectionand control. Bloom, B.R. (ed.). Washington, DC: AmericanSociety for Microbiology, pp. 389–415.

Charlton, S.J., Brown, C.A., Weisman, G.A., Turner, J.T.,Erb, L., and Boarder, M.R. (1996) Cloned and transfectedP2Y4 receptors: characterization of a suramin and PPADS-insensitive response to UTP. Br J Pharmacol 119: 1301–1303.

Chopade, B.A., Shankar, S., Sundin, G.W., Mukhopadhyay,S., and Chakrabarty, A.M. (1997) Characterization ofmembrane-associated Pseudomonas aeruginosa Ras-likeprotein Pra, a GTP-binding protein that forms complexeswith truncated nucleoside diphosphate kinase and pyru-vate kinase to modulate GTP synthesis. J Bacteriol 179:2181–2188.

De la Rosa, A., William, R.L., and Steeg, P.S. (1995) Nm23/nucleoside diphosphate kinase: toward a structural andbiochemical understanding of its biological functions. Bio-essays 17: 53–62.

Derijard, B., Ben, A.M., Lubochinsky, B., and Cenatiempo, Y.(1989) Evidence for a membrane-associated Res GTP-binding protein in Stigmatella aurantiaca, a prokaryoticcell. Biochem Biophys Commun 158: 562–568.

Dever, T.E., Glynias, M.J., and Merrick, W.C. (1987) GTPbinding domain: three consensus sequence elements withdistinct spacing. Proc Natl Acad Sci USA 84: 1814–1818.

Didenko, L.W., Buchwalow, I.B., Schulze, W., Augsten, K.,Susa, M., and Unger, E. (1996) Localization of G proteinsin macrophages and E. coli during phagocytosis. ActaHistochem 98: 399–409.

Gautam, N., Baetscher, M., Aebersold, R., and Simon, M.I.

(1989) A G protein gamma subunit shares homology withras proteins. Science 244: 971–974.

Gilman, A.G. (1987) G proteins: transducers of receptor-generated signals. Annu Rev Biochem 56: 615–649.

Heithoff, D.M., Conner, C.P., Hanna, P.C., Julio, S.M., Hent-schel, U., and Mahan, M.J. (1997) Bacterial infection asassessed by in vivo gene expression. Proc Natl Acad SciUSA 94: 934–939.

Huwiler, A., van Rossum, G., Wartmann, M., and Pfeilschifter,J. (1997) Stimulation by extracellular ATP and UTP of thestress-activated protein kinase cascade in rat renal mes-angial cells. Br J Pharmacol 120: 807–812.

Jacobs, W.R.Jr., and Bloom, B.R. (1994) Molecular geneticstrategies for identifying virulence determinants of Myco-bacterium tuberculosis. In Tuberculosis: Pathogenesis,Protection and Control. Bloom, B.R. (ed.). Washignton,DC: American Society for Microbiology, pp. 253–268.

Kaziro, Y., Itoh, H., Kozasa, T., Nakafuku, M., and Satoh, T.(1991) Structure and function of signal-transducing GTP-binding proteins. Annu Rev Biochem 60: 349–400.

Kinger, A.K., and Tyagi, J.S. (1993) Identification and cloningof genes differentially expressed in the virulent strain ofMycobacterium tuberculosis. Gene 131: 113–117.

Kikkawa, S., Takahashi, K., Takahashi, K.-I., Shimada, N.,Ui, M., Kimura, N., and Katada, T. (1990) Conversion ofGDP into GTP by nucleoside diphosphate kinase on theGTP-binding proteins. J Biol Chem 265: 21536–21540.

Klinker, J.F., Laugwitz, K.L., Hageluken, A., and Seifert, R.(1996) Activation of GTP formation and high affinity GTPhydrolysis by mastoparan in various cell membranes:G-protein activation via nucleoside diphosphate kinase, apossible general mechanism of mastoparan action. Bio-chem Pharmacol 51: 217–223.

Knol, J.C., van der Slik, A.R., van Keseren, E.R., Planta, R.J.,van Heerikhuizen, H., and Vreugdenhil, E. (1995) A novelG protein a-subunit containing atypical guanine nucleotide– binding domains is differentially expressed in molluscannervous system. J Biol Chem 270: 18804–18808.

Kohn, E.C., and Liotta, L.A. (1995) Molecular insights intocancer invasion: strategies for prevention and interven-tion. Cancer Res 55: 1856–1862.

Mukhopadhyay, S., Shankar, S., Walden, W., and Chakra-barty, A.M. (1997) Complex formation of the elongationfactor Tu from Pseudomonas aeruginosa with nucleosidediphosphate kinase modulates ribosomal GTP synthesisand peptide chain elongation. J Biol Chem 272: 17815–17820.

Offermanns, S., Mancino, V., Revel, J.-P., and Simon, M.I.(1997) Vascular system defects and impaired cellchemokinesis as a result of Ga13 deficiency. Science275: 533–536.

Plum, G., and Clark-Curtiss, J.E. (1994) Induction of Myco-bacterium avium gene expression following phagocytosisby human macrophages. Infect Immun 62: 476–483.

Sambrook, J., Fritsch. E.F., and Maniatis, T. (1989) Molecu-lar Cloning, Vol. 1, 2nd edn. Cold Spring Harbor, NY: ColdSpring Harbor Laboratory Press.

Schmidt, C.J., Thomas, T.C., Levine, M.A., and Neer, E.J.(1992) Specificity of G protein beta and gamma subunitinteractions. J Biol Chem 267: 13807–13810.

Shankar, S., Kamath, S., and Chakrabarty, A.M. (1996) Two



forms of the nucleoside diphosphate kinase of Pseudo-monas aeruginosa 8830: altered specificity of nucleosidetriphosphate synthesis by the cell membrane-associatedform of the truncated enzyme. J Bacteriol 178: 1777–1781.

Shankar, S., Hershberger, C.D., and Chakrabarty, A.M. (1997)The nucleoside diphosphate kinase of Mycobacteriumsmegmatis: identification of proteins that modulate specifi-city of nucleoside triphosphate synthesis by the enzyme.Mol Microbiol 24: 477–487.

Simon, M.I., Strathmann, M.P., and Gautam, N. (1991) Diver-sity of G-proteins in signal transduction. Science 252:802–808.

Spring, D.J., and Neer, E.J. (1994) A 14 amino acid region ofthe G protein gamma subunit is sufficient to confer selectiv-ity of gamma binding to the beta subunit. J Biol Chem 269:22882–22886.

Steeg, P.S., De la Rosa, A., Flatow, U., MacDonald, N.J.,Benedict, M., and Leone, A. (1993) Nm23 and breastcancer metastasis. Breast cancer Res Treat 25: 175–187.

Sundin, G.W., Shankar, S., Chugani, S.A., Chopade, B.A.,Kavanaugh-Black, A, and Chakrabarty, A.M. (1996) Nucleo-side diphosphate kinase from Pseudomonas aeruginosa:characterization of the gene and its role in cellular growthand exopolysaccharide alginate synthesis. Mol Microbiol20: 965–979.

Wang, L., Patel, U., Ghosh, L., Chen, H.-C., and Banerjee,S. (1993) Mutation in the nm23 gene is associated withmetastasis in colorectal cancer. Cancer Res 53: 717–720.

Wayne, L.G. (1994) Dormancy of Mycobacterium tubercu-losis and latency of disease. Eur J Clin Microbiol InfectDis 13: 908–914.



mammalian heterotrimeric g-protein-like proteins in mycobacteria: implications for cell signalling...

Documents