Social Media

A partner for the resuscitation-promoting factors of Mycobacterium tuberculosis

A partner for the resuscitation-promoting factors of Mycobacterium tuberculosis
of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  A partner for the resuscitation-promoting factors of Mycobacterium tuberculosis  Erik C. Hett, 1 Michael C. Chao, 1 Adrie J. Steyn, 2 Sarah M. Fortune, 1 Lynn L. Deng 3 andEric J. Rubin 1 * 1 Department of Immunology and Infectious Diseases,Harvard School of Public Health, Boston, MA 02115,USA. 2 Department of Microbiology, University of Alabama,Birmingham, AL 35294, USA. 3 Department of Medicine, VA Medical Center, Boston,MA 02130, USA. SummaryMany cases of active tuberculosis are thought toresult from the reactivation of dormant  Mycobacte- rium tuberculosis   from a prior infection, yet remark-ably little is known about the mechanism by whichthese non-sporulating bacteria reactivate. A familyof extracellular bacterial proteins, known asresuscitation-promoting factors (Rpfs), has previ-ously been shown to stimulate growth of dormantmycobacteria. While Rpf proteins are clearly pepti-doglycan glycosidases, the mechanism and role ofRpf in mediating reactivation remains unclear. Herewe use a yeast two-hybrid screen to identify potentialbinding partners of RpfB and report the interactionbetween RpfB and a putative mycobacterial endopep-tidase, which we named Rpf-interacting protein A(RipA). This interaction was confirmed by  in vitro   and in vivo   co-precipitation assays. The interactingdomains map to the C-termini of both proteins, nearpredicted enzymatic domains. We show that RipAis asecreted, cell-associated protein, found in the samecellular compartment as RpfB. Both RipA and RpfBlocalize to the septa of actively growing bacteria byfluorescence microscopy. Finally, we demonstratethat RipAis capable of digesting cell wall material andis indeed a peptidoglycan hydrolase. The interactionbetween these two peptidoglycan hydrolases at theseptum suggests a role for the complex in cell divi-sion, possibly during reactivation.Introduction Mycobacterium tuberculosis   has the unusual ability tosurvive in humans for extended periods of time withoutproducing apparent symptoms. However, as long asdecades after initial infection, these apparently dormantbacteria can once again replicate and cause disease.Approximately one-third of the world’s population isinfected with latent  M. tuberculosis   and are at riskfor reactivation. Yet the mechanism by which mycobac-teria reactivate from this unusual state is poorlyunderstood.Evidence suggests that the  in vivo   conditions underwhich dormant mycobacteria persist are nutrient andoxygen limited (Timm  et al  ., 2003; Tsai  et al  ., 2006). Many in vitro   dormancy studies have attempted to simulatethese conditions (Wayne and Hayes, 1996) and foundthey yield bacteria that have both a lower plating effi-ciency and an extended lag phase (Biketov  et al  ., 2000;Shleeva  et al  ., 2004). Recently, resuscitation-promotingfactor (Rpf), a secreted protein from  Micrococcus luteus  ,was shown to stimulate the growth of  M. luteus  , as well asseveral species of mycobacteria, that have been starvedof nutrients (Mukamolova  et al  ., 1998; 2002a). Whilesome debate exists regarding the reproducibility and bio-logical relevance of this resuscitation phenomenon,similar results have been reported with Rpf in otherspecies of bacteria (Zhu  et al  ., 2003; Hartmann  et al  .,2004; Panutdaporn  et al  ., 2006). In fact, genes encodingRpf-like proteins have been found in a wide range ofbacteria (Ravagnani  et al  ., 2005).The  M. tuberculosis   genome contains five  rpf  -likegenes ( rpfA–E  ) that are defined by a conserved70-amino-acid region. In  M. tuberculosis  , the  rpf   geneshave similar expression profiles, with transcripts and pro-teins encoded by each  rpf   gene detectable throughoutvegetative and stationary-phase growth (Tufariello  et al  .,2004). Also, strains with single deletions of each genelack  in vitro   phenotypes (Downing  et al  ., 2004; Tufariello et al  ., 2004), suggesting functional redundancy. However,a strain of  M. tuberculosis   lacking the  rpfB   gene has adelayed reactivation in a mouse dormancy model(Tufariello  et al  ., 2006). Also, two strains of  M.tuberculosis   in which three  rpf   genes were deleted wereboth attenuated in mice and did not reactivate in an  in vitro  reactivation assay (Downing  et al  ., 2005). Accepted 30 August, 2007. *For correspondence. E-mail; Tel. ( + 1) 617 432 3337; Fax ( + 1) 617 738 7664. Molecular Microbiology (2007)  66 (3), 658–668 doi:10.1111/j.1365-2958.2007.05945.xFirst published online 4 October 2007 © 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd  All Rpf proteins contain a conserved domain bearingstriking homology to C-type lysozyme and the  Escherichia coli   soluble lytic transglycosylase 70 (Cohen-Gonsaud et al  ., 2005), both known to degrade peptidoglycan, theprimary stress-bearing polymer of bacterial cell wall. Puri-fied  M. luteus   Rpf has recently been shown to degradepeptidoglycan (Mukamolova  et al  ., 2006), suggesting arole in peptidoglycan metabolism. In fact, mutation of thepredicted active site glutamate of Rpf not only decreasedits peptidoglycan hydrolytic activity (Mukamolova  et al  .,2006), but also abolished its growth stimulatory activity(Cohen-Gonsaud  et al  ., 2005).Certainly, for bacteria to grow and divide, they need tocleave peptidoglycan both to create openings for theinsertion of new peptidoglycan monomers and to separatedaughter cells connected at the septum. How, though,might the lysozyme-like Rpf protein stimulate escape fromdormancy? At least two mechanisms have been sug-gested (Ravagnani  et al  ., 2005; Keep  et al  ., 2006). Rpfmight enzymatically release a molecule from the cell wallcapable of signalling downstream events, as seen with b -lactamase induction (Jacobs  et al  ., 1997). Alternatively,the direct enzymatic activity of Rpf on the cell wall couldbe sufficient to release the bacteria from a growthrestraint, such as an inert cell wall. However, becauselysozyme alone does not stimulate dormant bacteria(Mukamolova  et al  ., 2006), there is likely something inaddition to the enzymatic activity of Rpf that is required forRpf-based reactivation.Here we show that RpfB physically interacts with apeptidoglycan hydrolase in yeast two-hybrid,  in vitro  , and in vivo   binding assays. Furthermore, both proteins werefound to localize to the septa of dividing bacteria. Thisinteraction between a lytic transglycosylase and a pepti-doglycan hydrolase presents a potential mechanism forRpf-based reactivation of dormant mycobacteria. Results Yeast two-hybrid screen identifies Rpf-interacting protein A (RipA)  Based on the observation that a secreted protein, knownas Rpf, stimulated growth of dormant mycobacteria in anautocrine manner (Mukamolova  et al  ., 1998), we hypoth-esized that the mechanism of the reactivation may involvean interaction between Rpf and another secreted orsurface-associated protein. Therefore, we conducted ayeast two-hybrid screen to find proteins that interact withRpf.We fused DNA encoding a 70-amino-acid portion ofRpfB, conserved among the Rpf proteins (Fig. 1A) andhighly homologous to the  M. luteus   Rpf, to DNAencodingthe yeast GAL4 DNA-binding domain (BD-Rpf) andscreened against a random library of  M. tuberculosis  gene fragments fused to DNA encoding the GAL4-activating domain (AD) (Steyn  et al  ., 2002). Approxi-mately 1.5  ¥  10 6 independent clones were screened forinteraction with RpfB. Strains were initially selected forhistidine prototrophy and  b -galactosidase activity, tworeporter phenotypes activated when proteins interact inthe yeast two-hybrid assay. Candidates were thenscreened for adenine prototrophy, counterscreened fornon-specific interactions, and evaluated by quantitative Fig. 1.  Yeast two-hybrid screen identifies aputative endopeptidase, RipA.A. Predicted domains of  M. tuberculosis   RpfB.The lytic transglycosylase domain is theconserved domain defining the Rpf family ofproteins. This region of RpfB (amino acids289–362) was used as bait in a yeasttwo-hybrid screen against an  M. tuberculosis  genomic DNA library. DUF348, domain ofunknown function; NAG,  N  -acetylglucosamine.B. Predicted domains of  M. tuberculosis   RipA.NLPC_P60 predicted domain has homologyto the hydrolytic endopeptidase domain of L. monocytogenes   P60. RipA amino acids200–472 comprise the region initially hit in theyeast two-hybrid screen. COG3883 predicteddomain does not have a defined function. A hydrolytic partner for Rpf   659  © 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd,  Molecular Microbiology  ,  66 , 658–668  b -galactosidase assays to further confirm true positiveinteractions. We found that the two independent strainswith the highest  b -galactosidase activity each containedplasmids that encoded similar, in-frame fusions of the M. tuberculosis rv1477   gene. Therefore, we proposenaming Rv1477 Rpf-interacting protein A (RipA). RipA is a predicted endopeptidase that is secreted and cell-associated Mycobacterium tuberculosis rv1477   encodes RipA(Fig. 1B), a 472-amino-acid protein with a C-terminal 105-amino-acid predicted domain that has 40% identity withthe  Listeria monocytogenes   p60 protein shown to be a cellwall hydrolase by its ability to digest cell wall material(Pilgrim  et al  ., 2003; Boneca, 2005).Because RpfB is a secreted, cell-associated protein(Mukamolova  et al  ., 2002b), we hypothesized that it inter-acts with RipA on the surface of the bacterium. RipA hasa predicted signal sequence and similar homologues fromboth  Mycobacterium avium   and  M. tuberculosis   havebeen shown to be secreted (Braunstein  et al  ., 2000;Carroll  et al  ., 2000). To confirm that the  M. tuberculosis  RipA is also secreted, we created a gene fusion thatencoded the first 200 amino acids of RipA fused to  E. coli  alkaline phosphatase (PhoA) and expressed this in Mycobacterium smegmatis  . Enzymatic activity of PhoArequires oxidation of a disulphide bond, an event that onlyoccurs upon secretion. We found that the RipA–PhoAfusion protein was active and turned the colorimetric sub-strate blue, while PhoA lacking a signal sequence wasinactive, leaving the colonies white ( Supplementary material  ).To directly assay the distribution of RipA in M. smegmatis  , we induced RipA-myc from an episomalplasmid and performed immunoblot analysis of M. smegmatis   cell pellets and concentrated culture fil-trates using myc polyclonal antibodies ( Supplementary material  ). We found that approximately 99% of detectableRipA protein was cell-associated. Thus, the vast majorityof RipA is in the same cellular compartment as RpfB. RpfB interacts with RipA  in vitro  and   in vivoTo confirm that RipA and RpfB physically interact  in vitro  we performed a co-precipitation procedure. We producedfusion proteins, one containing the C-terminal 283 aminoacids of RipA fused to glutathione-S-transferase (GST),and another containing the 70-amino-acid conservedregion of RpfB fused to maltose-binding protein (MBP).Proteins were mixed in equimolar amounts, incubatedand purified by affinity chromatography with glutathionesepharose. Co-purifying proteins were detected usingimmunoblotting with an MBP polyclonal antibody. Wefound that MBP–RpfB co-purified with GST–RipA, whilethere was no detectable interaction between RpfB andGST, MBP and GST, or MBP and RipA controls (Fig. 2A).To test if RpfB and RipA interact  in vivo   we producedtagged proteins in  M. smegmatis  . We used inducible pro-moters to avoid the potential for toxicity from proteinoverproduction. Full-length RipA was fused to aC-terminal polyhistidine tag while full-length RpfB wasfused to a C-terminal myc epitope tag. Proteins or controlswere co-produced in  M. smegmatis   strains. Soluble pro-teins from lysates were bound to nickel columns andeluted. RpfB was visualized using immunoblotting withmyc polyclonal antibodies. We found that RpfB-myc spe-cifically co-purified with RipA-His (Fig. 2B), thus confirm-ing their interaction  in vivo  . Interaction maps to the C-terminus of RipA We mapped the region of interaction between RpfB andRipA using two techniques. First, using a yeast three- Fig. 2.  RpfB co-purifies with RipA from  in vitro   and  in vivo   lysates.A. Proteins were separately purified from  E. coli  , combined as indicated in equimolar amounts, incubated, then purified on glutathionesepharose. Samples were taken before (lysate) and after (eluate) GST purification. Co-purifying MBP fusion proteins were detected byimmunoblotting using MBP polyclonal antibody. Combinations with GST and MBP were used to test the specificity of the interaction.B. Immunoblot analysis of proteins coexpressed in  M. smegmatis   and purified by Ni-NTA affinity chromatography. Samples were taken before(lysate) and after (eluate) the Ni-NTA purification. Co-purifying myc proteins were detected by immunoblotting using myc polyclonal antibody.An empty vector was used as a negative control. 660  E. C. Hett   et al.  © 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd,  Molecular Microbiology  ,  66 , 658–668  hybrid assay encoding RipA fused to the GAL4-activatingdomain and RpfB fused to the GAL4-binding domain, wescreened regions of unfused RipA for the ability tocompete with fused RipA-AD, and thus inhibit the interac-tion (Fig. 3A). Interaction was measured by growth onselective media and LacZ activity.The initial screen, usinglarger peptides, suggested that the C-terminus was nec-essary for inhibition. Production of several peptides (pep-tides 1, 3 and 6 from Fig. 3B) was tested by immunoblotand indicated similar amounts of protein production andstability (data not shown). Therefore, we conducted asecond screen, using overlapping 25 amino acid peptidescovering the last 75 amino acids of RipA. The last 25amino acids of the C-terminus (Fig. 3B, peptide 12) weresufficient for inhibition.We then tested regions of RipAfor interaction with RpfBusing the yeast two-hybrid assay and confirmed that theC-terminus is important for the interaction (Fig. 3C). Delet- Fig. 3.  RipA requires C-terminal region for interaction with RpfB.A. Schematic of a yeast three-hybrid experiment where two proteins are fused to GAL4 domains (BD, binding domain; AD, activating domain)while a third protein lacks either GAL4 domain. Interaction between the two tagged proteins results in growth and LacZ activity. If theuntagged protein is able to compete for binding with one of the tagged proteins, the interaction between tagged proteins is inhibited, resultingin a lack of growth and a reduction in LacZ activity.B. Regions of the last 300 amino acids of the C-terminus of RipA were tested for their ability to inhibit the interaction between RipA and RpfBin the yeast three-hybrid assay. RipA (amino acids 197–472) was fused to the activating domain of GAL4 (AD), while RpfB (amino acids289–362) was fused to the DNA-binding domain of GAL4 (BD). The inhibiting peptides, derived from regions of RipA, compete with AD-RipAfor interaction with BD-RpfB. A reduction in growth on selective plates and decreased LacZ activity was interpreted as inhibition of theAD-RipA:BD-RpfB interaction. Peptide 1 (amino acids 172–472 of RipA), 2 (172–322), 3 (172–247), 4 (248–397), 5 (323–472), 6 (398–472),7 (398–423), 8 (408–432), 9 (418–442), 10 (428–452), 11 (438–462), 12 (448–472). Data shown are from a representative experimentperformed in triplicate. Data are represented as mean  SEM. Growth is represented as  +++  strong,  ++  moderate,  +  minimal, but evident,– lacking.C. Regions of  ripA  lacking DNA encoding the 40-amino-acid signal sequence were cloned into the yeast two-hybrid system to confirm theregion of interaction with RpfB. RipA-1 (amino acids 40–472), RipA-2 (40–447), RipA-3 (40–397), RipA-4 (40–305), RipA-5 (197–472). Growthon selective plates and LacZ activity indicate an interaction with RpfB. Data shown are from a representative experiment performed intriplicate. An empty vector was used as a negative control and subtracted as background. Growth is represented as  +++  strong,  ++  moderate, +  minimal, but evident, – lacking. Data are represented as mean  SEM. A hydrolytic partner for Rpf   661  © 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd,  Molecular Microbiology  ,  66 , 658–668  ing the last 25 amino acids from RipA resulted in a morethan two-fold reduction in activity, with further deletionsresulting in further loss of activity.Afusion protein of RipAlacking the N-terminus had undiminished activity in theassay. These results confirmed that the C-terminus ofRipA is important for the interaction with RpfB. RipA also interacts with RpfE  The region of RpfB screened in the yeast two-hybrid wasfrom the highly conserved region found in all Rpf proteins,suggesting that other Rpf proteins might also interact withRipA. Thus, we created fusions of each of the five full-length (minus signal sequence) Rpf proteins (A–E) to theGAL4-AD and tested individual interactions with RipAin ayeast two-hybrid assay. We found that yeast expressingRipA and full-length (minus signal sequence) RpfB orRpfE grew on selective plates and had measurable b -galactosidase activities, indicating that each indepen-dently interacts with RipA (Fig. 4). The conserved regionsof RpfB and RpfE show greater sequence similarity witheach other than with the other Rpf proteins (66% identity).Thus, at least two Rpf proteins interact with RipA. RpfB and RipA localize to the septa of dividing bacteria  The interaction between RpfB and RipA led us to hypoth-esize that RipAand RpfB may localize to the same regionin mycobacteria. We used the membrane staining dyeTMA-DPH to visualize the septa of dividing bacteria andto provide a landmark that could be used to show local-ization in relation to the septum. Each protein was pro-duced in  M. smegmatis   as a fusion with monomeric redfluorescent protein (RFP) (Campbell  et al  ., 2002) underthe control of a tetracycline-inducible promoter.We found that both the  M. tuberculosis   RipA–RFP andRpfB–RFP proteins localized to the septa of dividing M. smegmatis   (Fig. 5A). While cells expressing RpfB–RFP appeared morphologically normal, bacteria thatexpressed RipA–RFP often contained multiple septa. Incases where septa did not have corresponding RipA orRpfB localization, these proteins were observed asslightly diffuse. In occasional cells, RipA localized to thepoles, suggesting recent cell division such that RipAstain-ing may denote remnants of the division machinery at thenewly formed pole. Uninduced RipA–RFP yielded nodetectable fluorescence and RFP alone remained diffuseand cytosolic, with no observable bands of localization(Fig. 5). Usage of the same constructs in BCG resulted inlocalization of RipAto the septa of dividing cells and to thepoles of non-dividing cells (Fig. 5B). RpfB–RFP did notexpress well enough in BCG to conduct localizationexperiments.These data demonstrate that RipAand RpfBlocalize to the septa of dividing bacteria, suggesting thatthese two proteins colocalize and likely perform their func-tion at the septum during cell division. RipA degrades peptidoglycan  RipA is predicted to degrade peptidoglycan (Ananthara-man and Aravind, 2003; Pilgrim  et al  ., 2003). To test ifRipA hydrolyses peptidoglycan, we determined its enzy-matic activity using Remazol Brilliant Blue-labelled, cellwall as substrate. We expressed RipA as a fusion proteinwith GST in  E. coli   and purified the fusion protein usingaffinity chromatography. We found that GST–RipA, butnot GST alone, was able to hydrolyse cell wall derivedfrom  M. luteus   (Fig. 6). Lysozyme was used as a positivecontrol. Therefore, RipA is a hydrolase capable of break-ing down cell wall material. Discussion In this work, we identified a protein (RipA) that specificallyinteracts with the peptidoglycan glycosidases, RpfB andRpfE.This interaction occurs within growing mycobacteriaand is likely to occur on the outside of the cell membrane,as both proteins are secreted and cell-associated. Wefurther demonstrated that both RpfB and RipA localize tothe septa of dividing cells and that RipA is indeed a cellwall hydrolase. Taken together, these data suggest amodel where dormant mycobacteria could utilize Rpf tointeract with RipA and efficiently hydrolyse peptidoglycanduring reactivation, thus releasing the bacteria fromdormancy. Mycobacterium tuberculosis   encodes a bicistronicoperon (Rv1477 and Rv1478), of which Rv1477 has been Fig. 4.  RipA interacts with both RpfB and RpfE. Full-length RpfA–Ewere tested for interaction with RipA in a yeast two-hybrid assay.Empty vector and the conserved region of RpfB were used asnegative and positive controls respectively. Growth is representedas  +++  strong,  ++  moderate, but evident, – lacking. Data shownare from a representative experiment performed in triplicate andare represented as mean  SEM. 662  E. C. Hett   et al.  © 2007 The AuthorsJournal compilation © 2007 Blackwell Publishing Ltd,  Molecular Microbiology  ,  66 , 658–668
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks