PknB kinase activity is regulated by phosphorylation in two Thr residues and dephosphorylation by PstP, the cognate phospho-Ser/Thr phosphatase, in Mycobacterium tuberculosis

Molecular Microbiology (2003)

49

(6) 1493ndash1508 doi101046j1365-2958200303657x

copy 2003 Blackwell Publishing Ltd

Blackwell Science LtdOxford UKMMIMolecular Microbiology 1365-2958Blackwell Publishing Ltd 200349

614931508

Original Article

Mycobacterial serinethreonine kinase and phosphataseB Boitel et al

Accepted 5 June 2003 For correspondence E-mail alzaripasteurfr Tel (

+

33) 145688607 Fax (

+

33) 145688604 daggerTheseauthors contributed equally to this work

PknB kinase activity is regulated by phosphorylation in two Thr residues and dephosphorylation by PstP the cognate phospho-SerThr phosphatase in

Mycobacterium tuberculosis

Brigitte Boitel

1dagger

Miguel Ortiz-Lombardiacutea

1dagger

Rosario Duraacuten

2

Freacutederique Pompeo

1

Stewart T Cole

3

Carlos Cerventildeansky

2

andPedro M Alzari

1

1

Uniteacute de Biochimie Structurale URA 2185 CNRS Institut Pasteur 25 rue du Dr Roux 75724 Paris cedex 15 France

2

Laboratorio de Bioquiacutemica Analiacutetica Instituto de Investigaciones Bioloacutegicas Clemente Estable and Facultad de Ciencias Avenida Italia 3318 11600 Montevideo Uruguay

3

Uniteacute de Geacuteneacutetique Moleacuteculaire Bacteacuterienne Institut Pasteur 28 rue du Dr Roux 75724 Paris cedex 15 France

Summary

Bacterial genomics revealed the widespread pres-ence of eukaryotic-like protein kinases and phos-phatases in prokaryotes but little is known on theirbiochemical properties regulation mechanisms andphysiological roles Here we focus on the catalyticdomains of two

trans

-membrane enzymes the SerThr protein kinase PknB and the protein phosphatasePstP from

Mycobacterium tuberculosis

PstP wasfound to specifically dephosphorylate model phos-pho-SerThr substrates in a Mn

2++++

-dependent mannerAutophosphorylated PknB was shown to be a sub-strate for Pstp and its kinase activity was affected byPstP-mediated dephosphorylation Two threonine res-idues in the PknB activation loop found to be mostlydisordered in the crystal structure of this kinasenamely Thr171 and Thr173 were identified as the tar-get for PknB autophosphorylation and PstP dephos-phorylation Replacement of these threonine residuesby alanine significantly decreased the kinase activityconfirming their direct regulatory role These resultsindicate that as for eukaryotic homologues phos-

phorylation of the activation loop provides a regula-tion mechanism of mycobacterial kinases andstrongly suggest that PknB and PstP could work as afunctional pair

in vivo

to control mycobacterial cellgrowth

Introduction

Tuberculosis (TB) is a major public health problem withone-third of the worldrsquos population infected by its aetiologicagent

Mycobacterium tuberculosis

and over two millionpeople dying from the disease each year (Dye

et al

1999) (httpwwwwhoint) The Global Alliance for TBDrug Development has proposed that the current treat-ment could be improved considerably by developing morepotent therapeutic agents that reduce the duration oftherapy and by including drugs that act on latent bacilli(Global Alliance for TB Drug Development 2001) Facedwith the urgency to develop new therapeutic strategies itappears crucial to understand better the physiopathologyof the causative agent and its complex relationship withthe immune system of the host

After inhalation infectious bacilli are phagocytosed byalveolar macrophages in the lung and induce a local pro-inflammatory response which leads to the recruitment ofmonocytes from the bloodstream into the site of infection(Dannenberg 1999 Russell 2001) By blocking fusionof phagosomes with lysosomes in these non-activatedmacrophages (Brown

et al

1969 Sturgill-Koszycki

et al

1996)

M tuberculosis

escapes killing and multi-plies As the immune response progresses macroph-ages and T cells accumulate to form a granuloma inwhich the pathogen is contained in a latent state (Parrish

et al

1998 Manabe and Bishai 2000) It can lie dor-mant for years only to rise again when the immune sys-tem wanes through old age malnutrition or AIDS(acquired immuno-deficiency syndrome) The centre ofthe granuloma then liquefies and

M tuberculosis

repli-cates profusely and is discharged into the bronchial treeproducing an infectious cough (Dannenberg 1999) Tounderstand the bacterial response to these changes inhost environment the study of regulatory proteins

1494

B Boitel

et al

copy 2003 Blackwell Publishing Ltd

Molecular Microbiology

49

1493ndash1508

involved in mycobacterial signal transduction is thereforeof the utmost importance

Phosphorylation a simple and efficient means ofreversibly changing the biochemical properties of a pro-tein is a major mechanism for signal transduction andregulation of almost all biological functions There are twomain phosphorylative signal transduction systemsProkaryotes predominantly use the two-component sys-tem comprising in its simplest form a signal sensor witha histidine kinase domain and a response regulator oftena transcriptional factor (Wurgler-Murphy and Saito 1997Stock

et al

2000) This simple unidirectional mechanismallows a quick response to abrupt environmental changesThe second system depends on the reversible phospho-rylation of serine threonine and tyrosine residues and iswidely used in eukaryotes (Hanks and Hunter 1995Hunter 1995 Barford

et al

1998 Hunter 2000) Thismechanism involves the action of protein kinases andphosphoprotein phosphatases in cascades and networks(Hunter 2000) providing an efficient means for the rapidmodulation of the transduced signal to serve highly regu-lated functions

Since the identification of the first bacterial homologuea few years ago (Muntildeoz-Dorado

et al

1991) genomicshas now demonstrated that serine threonine and tyrosineprotein kinases and phosphatases are also widespread inprokaryotes (Zhang 1996 Kennelly 2002) The two phos-phorylation mechanisms (two-component systems andSerThrTyr kinases and phosphatases) in prokaryotesmay regulate distinct functions or act together in the samesignalling pathway The presence of SerThr and Tyrkinases and phosphatases in prokaryotes appears to beassociated with a complex multistage developmentalcycle and possible roles in regulating growth and devel-opment (heterocyst fruiting-body or spore formation)have been proposed (Zhang 1996 Shi

et al

1998) Thedormant state of

M tuberculosis

although poorly under-stood may be considered in some regards analogous tosporulation (Demaio

et al

1996) and thus involve theseenzymes

Mycobacterium tuberculosis

employs both systems ofprotein phosphorylation It has 15 sensor His kinases and15 response regulators forming at least 11 functionalpairs together with 11 putative SerThr protein kinases(STPKs) one phospho-SerThr phosphatase (

ppp

renamed here

pstP

) and two Tyr phosphatases (

ptpAptpB

) (Cole

et al

1998) (httpwwwgenolistpasteurfrTubercuList) There appears to be no counterpart Tyrkinase for the two Tyr phosphatases PtpA and PtpBwhich can moreover be secreted (Koul

et al

2000 Cow-ley

et al

2002) Eight of the 11 STPKs are predicted tobe transmembrane proteins with a putative extracellularsignal sensor domain and an intracellular kinase domainSix STPKs (PknA B D E F G) have already been

expressed as recombinant proteins and shown to befunctional kinases (Peirs

et al

1997 Av-Gay

et al

1999Koul

et al

2001 Chaba

et al

2002 data not shown forPknE)

At this time no physiological role has been clearly dem-onstrated for any of the STPKs or phosphatases from

Mtuberculosis

and knock-out mutants have not yet beenreported Here we have focused our interest on PknB andPstP as indirect data suggest they could play an essentialrole in the biology of

M tuberculosis

The

pknB

and

pstP

genes along with

pknA

are found in an operon (Fig 1) thatalso includes

rodA

and

pbpA

(Cole

et al

1998) twogenes encoding morphogenic proteins involved in pepti-doglycan synthesis during cell growth (Matsuhashi 1994)Furthermore this genomic region remains unchanged inthe close relative

M leprae

(Fsihi

et al

1996) in spite ofthe extensive gene decay in this bacillus which hasremoved or inactivated over 2400 genes including thosefor all other STPKs (except for PknL and PknG) and bothTyr phosphatases (Eiglmeier

et al

2001) Thus the con-servation of the

pknA

pknB

and

pstP

genes near thechromosomal origin of replication in

M leprae

stronglysuggests that the corresponding enzymes could regulateessential functions possibly related to cell growth orlatency of mycobacteria

We demonstrate here that Pstp dephosphorylates spe-cifically phospho-SerThr residues and that its activity isstrictly dependent on the presence of divalent cations Wealso report that the catalytic domain of PknB as definedby homology modelling is an active protein kinase in itsphosphorylated state Pstp is capable of dephosphorylat-ing PknB which subsequently exhibits decreased kinaseactivity Mass spectrometry analysis identified two phos-phothreonine residues in the activation loop of PknBMutagenesis of these threonines in alanine demonstratetheir role in regulating PknB kinase activity We suggestthat Pstp and PknB could interplay

in vivo

in the sametransduction pathway and discuss the putative regulatoryroles of these enzymes in mycobacteria

Results

PstP is a SerThr protein phosphatase

The

pstP

gene (Rv0018c) encodes a putative transmem-brane protein of 514 aa (Cole

et al

1998) with a C-termi-nal extracellular domain (196 aa) rich in proline and serineresidues (Fig 2A) The putative intracellular domain(301 aa) is homologous to members of the eukaryotic SerThr protein phosphatase PPM family (Bork

et al

1996)The sequence alignment of the catalytic domains of PstPand human PP2C the prototype member of the PPMfamily is shown in Fig 2B Although PstP displays only17 identity with the human enzyme all the motifs corre-

Mycobacterial Serinethreonine kinase and phosphatase

1495

copy 2003 Blackwell Publishing Ltd

Molecular Microbiology

49

1493ndash1508

sponding to key structural elements (Bork

et al

1996) arepresent in the PstP sequence The crystal structure of thehuman PP2C has revealed a metal ion-catalysed dephos-phorylation mechanism (Das

et al

1996) As indicated inFig 2B all the residues involved in the binding of metalcations and phosphate are conserved in PstP suggestinga common mechanism of phosphate recognition andcatalysis

The multiple alignment of PstP with other members ofthe PPM phosphatase family predicted Asp 240 as thelast residue of the catalytic domain Thus the His-taggedconstruction PstP

1-240

was produced as a soluble proteinin

E coli

(Fig 3A) The protein phosphatase activity andthe specificity towards phospho-amino acids were testedusing different substrates The myelin basic protein (MBP)and

a

-casein were phosphorylated either on serine andthreonine residues with the protein kinase A (PKA) or ontyrosine residues with the Abl kinase using radiolabelledATP As shown in Fig 3B PstP dephosphorylated pho-pho-SerThr substrates but showed little or no activity withphospho-Tyr substrates Furthermore PstP phosphataseactivity was strictly dependent on divalent cations with apreference for Mn

2

+

versus Mg

2

+

(data not shown) Thusin agreement with sequence homology-based predictionsthese results demonstrate that the intracellular region ofPstP is a SerThr protein phosphatase that belongs to thePPM family

The C-terminal domain of PknB is similar to that found in various other bacterial STPKs

PknB is predicted to be a 626 aa transmembrane proteinwith an intracellular N-terminal kinase domain (331 aa)and an extracellular C-terminal domain (276 aa) (Fig 4A)This structural organization for STPKs is found in plantsand as receptors for the transforming growth factor

b

(TGF

b

) family cytokines in vertebrates where the C-ter-minal domain is a signal sensor This could also be thecase for the transmembrane STPKs from prokaryotesThe C-terminal domain of PknB shows some degree ofsequence similarity with the C-terminal domain of severalprokaryotic STPKs including actinobacteria (corynebac-terium streptomyces bifidobacterium) and other Gram-positive bacteria (listeria bacillus streptococcus)(Fig 4B) These proteins display a diverse number of cop-ies four in PknB of the recently described PASTA domain(for penicillin-binding-protein and serinethreonine kinaseassociated domain Yeats

et al

2002) This suggests thatall these kinases could respond to a similar type of ligandActually it has been speculated that the PASTA domainscould bind unlinked peptidoglycan (Yeats

et al

2002)although no experimental evidence is available to sub-stantiate this claim It is noteworthy that a gene coding fora putative SerThr protein phosphatase is found in thesame genomic region for the above mentioned organisms

Fig 1

Conserved structure of the putative operon including the

pknB

and

pstP

genes in several actinobacteria The genes coding for the following signal tranduction elements PknA PknB PstP and two proteins with a FHA domain are co-localized with two genes involved in peptidoglycan synthesis namely

pbpA

and

rodA

This gene cluster is conserved in all actinobacteridae genomes known to date including those presented here

M tuberculosis

(httpwwwgenolistpasteurfrTubercuList)

M leprae

(httpwwwgenolistpasteurfrLeproma)

C glutamicum

(httpwwwtigrorg) and

S coelicolor

(httpwwwsangeracuk) (note that the

pknA

gene is missing in

S coelicolor

genome) and also such as

C diphteriae

C efficiens

Thermobifida fusca

and

Bifidobacterium longum

(httpwwwncbinlmnihgov)

1496

B Boitel

et al

copy 2003 Blackwell Publishing Ltd

Molecular Microbiology

49

1493ndash1508

suggesting a functional association with the STPKIndeed it has recently been described that the PrkCkinase and the PrpC phosphatase from

Bacillus subtilis

form such a couple

in vivo

with opposite effects onstationary-phase physiology (Gaidenko

et al

2002)

The catalytic domain of PknB is a functional protein kinase

The full-length recombinant PknB protein has been previ-ously characterized and shown to possess STPK activity(Av-Gay

et al

1999) To allow detailed biochemical andstructural studies we have chosen to focus on its catalyticdomain Multiple sequence alignment with members ofthe SerThr protein kinase family and homology modellingbased on available three-dimensional structures pointedto Gly 279 as the last residue in the C-terminal

a

-helix ofthe catalytic domain Thus the domain corresponding toamino acid residues 1ndash279 of PknB (PknB

1-279

) has beenproduced in

E coli

as a soluble monomeric His-taggedprotein (Fig 5A)

The kinase activity of PknB

1-279

was tested either in anautophosphorylation assay or using MBP as a modelsubstrate Like the full-length renatured PknB (Av-Gay

et al

1999) PknB

1-279

autophosphorylates and phospho-rylates MBP (Fig 5A) Thrombin-digested PknB

1-279

(iewithout the His-Tag) is also autophosphorylated indicat-ing that specific autophosphorylation sites exist on thePknB catalytic domain (data not shown) Kinase activitydepends on divalent cations (Fig 5A) PknB

1-279

showinga clear preference for Mn

2

+

versus Mg

2

+

ions (Fig 5B)These observations demonstrate that when separatelyexpressed the catalytic domain of PknB has intrinsickinase activity implying that other regions of the protein(such as the juxtamembrane region) are not required tostabilize an active conformation

Fig 2

A Structural organization of PstP JM juxta-membrane region TM

trans

-membrane regionB Primary sequence alignment of the catalytic domain of PstP and the human PP2CConserved residues are boxed The amino acids of PP2C involved in the binding of the metal ions and the phosphate are indicated with a star Secondary structural elements are indicated above the sequence

Fig 3

A Purification to homogeneity of PstP

1-240

His-tagged Pstp

1

-

240

was purified by affinity and size exclusion chromatography The purity was then checked by SDS-PAGE electrophoresis PstP1-240 appears as a single discrete band on the gel with an apparent MW (32 kDa) slightly higher than the expected value (276 kDa)B Analysis of the specificity of PstP1-240 towards phosphoresidues MBP (left panel) and a-casein (right panel) were phosphorylated either on serine and threonine residues or on tyrosine residues with [g-33P]ATP Release of radiolabelled inorganic phosphate was measured after incubation of increasing concentrations of the purified PstP1-240 with the different phosphosubstrates

Mycobacterial Serinethreonine kinase and phosphatase 1497

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

The recently determined structure of the catalytic coreof PknB in complex with nucleotide at 22 Aring resolution(Ortiz-Lombardiacutea et al 2003) and 3 Aring resolution (Younget al 2003) lends further support to these observations

The PknB catalytic domain was found to be very similarto its eukaryotic homologues and shares a number ofessential hallmarks first described for PKA (Knightonet al 1991) In particular all amino acid residues and

Fig 4 A Structural organization of PknBB Sequence alignment of the putative sensor domain of bacterial STPKs A BLAST search was conducted to detect the protein sequences most similar to the PknB C-terminal domain We then selected among them the nine STPKs most similar to M tuberculosis PknB ie STPKs from M leprae Corynebacterium glutamicum C efficiens Thermobifida fusca Bifidobacterium longum Streptomyces coelicolor and Bacillus subtilis The sequences of the C-terminal domains of these proteins were aligned with CLUSTALW The extracellular domain of these STPKs consists of three to four PASTA domains represented as different blocks These repeated domains may have arisen by duplication events

1498 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

other structural elements important for catalysis are foundin their active conformation (Ortiz-Lombardiacutea et al 2003)

Different preparations of PknB1-279 produced a relativelybroad complex mass peak in MALDI-TOF mass spectrom-etry experiments with maximum intensity at mz = 32 538and smaller signals close to 80 Da 98 Da or 160 Da apart(data not shown) After treatment with PstP or alkalinephosphatase the peak shifted to mz = 32 291 (thesequence-predicted average mass of uncleaved PknB1-279

is 32 281 Da) indicating the removal of at least threephosphate groups linked to the protein (Fig 6) Howeverwe have failed to detect any phosphorylated residue in the

3D structure of PknB (Ortiz-Lombardiacutea et al 2003) Asthe whole catalytic domain (except for residues A164-T179 covering most of the activation loop) is well-definedin the electron density map this suggests that the putativephosphoresidues should be found in the disordered ormobile parts of the protein ie at the N-terminal peptideextension outside the catalytic core andor within the acti-vation loop itself in agreement with the putative phospho-rylation sites recently proposed for this region by Younget al (2003)

PstP dephosphorylates PknB and inhibits its kinase activity

Full-length PknB has been shown to be autophosphory-lated on Ser and Thr residues (Av-Gay et al 1999) andthe question arises whether PknB1-279 could be a substratefor PstP To address this possibility PknB1-279 was auto-phosphorylated with radioactive ATP before incubationwith PstP in the presence or absence of MnCl2 As shownin Fig 7A PstP is capable of dephosphorylating PknBPhosphate hydrolysis is also reflected by the shift in PknBmigration on the gel concomitant with loss of label the

Fig 5 A Kinase activity of PknB1-279 autophosphorylation and MBP phosphorylation assays Purified PknB1-279 alone or with the model kinase substrate MBP was incubated with [g-33P]ATP in the presence or absence of MnCl2 The reaction products were resolved on a SDS-PAGE gel that was Coomassie blue stained (left panel) then dried and autoradiographied (right panel) As observed for other phosphop-roteins the apparent MW of the protein in SDS-PAGE (40 kDa) is significantly higher than the expected value of 32 kDaB Effect of divalent cations on the kinase activity of PknB1-279Various concentrations of MnCl2 or MgCl2 were used in the MBP phosphory-lation assay Relative quantification of the incorporated phosphate on MBP was obtained after PhosphorImager analysis

Fig 6 MALDI spectra of PknB before (A) and after (B) dephosphory-lation with alkaline phosphatase

A

B

Mycobacterial Serinethreonine kinase and phosphatase 1499

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

lower band corresponding to dephosphorylated PknBThese differences in gel mobility were exploited to furthermonitor the phosphatase reaction without previous radio-active labelling (Fig 7B) The dephosphorylation of PknB

by PstP also indicates that the recombinant kinase pro-duced in E coli is phosphorylated in vivo

We then asked whether the dephosphorylation of PknBcould have an effect on its kinase activity To address thisquestion PknB was preincubated with Pstp and ATP wasreplaced by thio-gATP in the kinase reaction The rationalfor this assay resides in the ability of PknB of thiophos-phorylating substrates whereas PstP is not active onthese thiophosphosubstrates (data not shown) Underthese conditions the kinase activity can be measuredwithout interference from the phosphatase activityFigure 7C shows that prior dephosphorylation of PknB byPstP inhibits kinase activity on MBP These resultsstrongly suggest that the phosphorylation state of PknBis important in maintaining a fully active kinase

Identification of two phosphothreonines in the activation loop of PknB

Mass spectrometry was used to identify the phosphores-idues detected in PknB1-279 Comparison of the reverse-phase chromatograms of the trypsin digestion products ofeither PknB1-279 or PstP-treated PknB1-279 (covering 90of the PknB1-279 sequence) revealed changes in the elutionpattern of some selected peptides (Fig 8A) This obser-vation was consistent with results from MS in both reflec-tor and linear modes obtained from the correspondingwhole peptide mixture (data not shown) In linear modetwo phosphopeptides could be identified from untreatedPknB1-279 A signal at mz = 18501 was assigned to theHis-tag peptide plus one phosphate group (calc averagemass = 18499 for the [MH]+ peptide) and a strong signalat mz = 29813 was assigned to the di-phosphorylatedtryptic peptide A162-R189 (calc mass = 29810) whichincludes a large fraction of the activation loop It is note-worthy that no MS signal was detected for the non-phos-phorylated A162-R189 peptide (calc mass = 28211)except when PknB1-279 was pretreated with a phosphatasesuch as alkaline phosphatase or PstP Only in such con-ditions a prominent mass signal (at mz = 28208) wasobserved in both linear and reflector modes

These results were further confirmed when the sepa-rate peptide fractions were identified by MS measure-ments in reflector mode Thus peaks numbered 1 and 2(Fig 8A) were assigned to the monophosphorylated andunphosphorylated His-tag peptide respectively whereaspeak 3 was assigned to the diphosphorylated A162-R189peptide Upon treatment with PstP peak 1 was reducedin size peak 2 increased and peak 3 almost disappearedpresumably giving rise to peak 4 which corresponds tothe unphosphorylated A162-R189 peptide

Post-source decay mass spectrometry (PSD-MS) mea-surement of a sample from peak 3 confirmed the pres-ence of two phosphate groups in this peptide (Fig 8B)

Fig 7 Dephosphorylation assay using PknB1-279 as a substrate for PstP1-240 and effect of the dephosphorylation of PknB1-279 by PstP1-240 on its kinase activityA Autophosphorylated PknB1-279 in presence of [g-33P]ATP was used as substrate for PstP1-240 As a control MnCl2 was omitted from the reaction buffer The products of the reaction were subjected to elec-trophoresis on a denaturing gel Left panel the Coomassie blue stained gel right panel the autoradiographB Without prior labelling dephosphorylation of PknB is followed with the shift in protein migration in SDS-PAGEC PknB1-279 was preincubated with PstP1-240 for the indicated time The kinase activity was then assayed using MBP and thio-gATP as substrates Relative quantification of the kinase activity obtained with the PhosphorImager was plotted

1500 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

Definitive identification and localization of the phosphory-lated residues was achieved by PSD-MS sequencing ofHPLC peak 3 purified from independent batches of PknBThis analysis showed that A162-R189 peptide was phos-phorylated on Thr 171 and Thr 173 (Fig 8C) In all casesphosphorylation of these sites was close to 100 indi-cating that these threonines are systematically and homo-geneously linked to a phosphate The HPLC patterns ofPknB tryptic digests were extremely constant and repro-ducible over the time and with different preparations of theprotein However in some experiments a shoulder or evena small peak (just before peak 3 in Fig 8A) could beobserved with a mz = 30611 (data not shown) This wasidentified as a triphosphorylated species of the A162-R189 peptide (calc mass = 30613) The third phospho-

site is a serine that could not be unambiguously identifiedby sequencing and could correspond to either Ser 166 orSer 169

The above MS results identify two threonine residuesfrom the activation loop Thr 171 and Thr 173 as targetsfor PknB autophosphorylation and PstP dephosphoryla-tion These residues are part of a disordered region in thetwo PknB crystal structures (Ortiz-Lombardiacutea et al 2003Young et al 2003) However inspection of the chargedistribution at the molecular surface of the protein revealsan exposed cluster of basic residues that are favourablypositioned to provide an anchoring site for the phospho-threonine residues (Fig 9A) These arginine residueshave partially disordered or mobile side-chains in the crys-tal structure probably reflecting the absence of bound

Fig 8 Identification of phosphorylation sites in PknB1-279A HPLC separation of tryptic digests from PknB1-279 before (upper panel) and after treatment with PstP (lower panel) Fractions were manually collected and analysed by MALDI-MS with partial sequencing by PSD-MS when necessary for conclusive peptide identification Only peptides relevant to this work are annotated in the chromatograms peak 1 monophosphorylated His-tag peptide (mz = 184861 calc monoisotopic mass = 184884) peak 2 His-tag peptide (mz = 176891 calc monoisotopic mass = 176884 sequence GSSHHHHHHSSGLVPR) peak 3 diphos-phorylated S162-R189 peptide (mz = 297917 calc monoisotopic mass = 297934) and peak 4 S162-189 peptide (mz = 281953 calc monoisotopic mass = 281941)B Detailed PSD spectra obtained with a sample from peak 3 The signals corresponding to -80 Da -98 Da -(80 + 98) Da -(98 + 98) Da are strongly indicative of presence of two phosphate groups in serine andor threonine residues in the analysed sampleC Integrated PSD spectra to confirm peptide identification by sequencing and to localise phosphorylated residues (measured values from the y -ion series in Da y3 = 3740 y5 = 6001 y6 = 6872 y7 = 7998 y8 = 9620 y9 = 10910 y10 = 11623 y11 = 12625 y12 = 13194 y13 = 14331 y14 = 15332 y15 = 16033 y16 = 16744 y17-98 = 17573 y18-98 = 18861 y19-98ndash98 = 19690 y19-98 = 20674 y19 = 21654)

Mycobacterial Serinethreonine kinase and phosphatase 1501

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

substrate When compared with a similar cluster in PKA(Knighton et al 1991) that binds phospho-Thr 197 in theactivation loop (Fig 9B) the positively charged region inPknB is found to cover a more extended surface arearaising the possibility of this region binding the phosphategroups of both Thr 171 and Thr 173

Activation loop mutants of PknB

To confirm and further analyse the role of the identifiedphospho-threonines in PknB kinase activity these resi-dues were mutated to alanine singly or in combinationThe single mutants T171A T173A and the double mutantT171173 A were produced and analysed in the MBPphosphorylation assay Comparison of the kinetics ofphosphorylation of MBP by the mutants (Fig 10) showsthat the kinase activity is affected by each single mutationto a similar extent being 15- and 20-times less active thanPknB respectively The double mutant is 300-fold lessactive suggesting a combined effect of the two phospho-

threonines on kinase activity These results confirm thatdouble phosphorylation of the activation loop is requiredfor full kinase activity and demonstrate unambiguously theinvolvement of both phosphothreonines

These mutants were also tested for the presence andlocalization of phosphorylated amino acid residues andthe degree of phosphorylation at each site following thesame experimental protocol described above for the wild-type enzyme (Table 1) The N-terminal His-tag peptideshowed a consistently lower degree of phosphorylation inthe three mutants when compared to the wild-typeenzyme reflecting the lower activity of the mutants As forthe wild-type enzyme the mutant T171A is mainly diphos-phorylated in the activation loop the residues involvedbeing now Ser 169 and Thr 173 However phosphoryla-tion of Ser 169 does not restore wild-type activity andseems to play no functional role On the other hand theT173A mutant appears to be mainly monophosphorylatedin Thr 171 (a much smaller HPLC signal could beassigned to a diphosphorylated species at residues Thr171 and either Ser 166 or Ser 169) Analysis of peptidesfrom the trypsin-digested double mutant T171173 A dem-onstrated the occurrence of unphosphorylated (36) andone monophosphorylated (at either Ser 166 or Ser 169)A162-R189 peptide species In summary both singlemutants appear still fully phosphorylated on the remainingthreonine and the activity decrease of the single and dou-ble mutants did not show co-operative behaviour suggest-ing that Thr 171 and Thr 173 are independent phospho-sites Moreover a similar decrease in kinase activity isobserved upon the lost of each phosphosite suggestingthat the two phosphothreonines are equally important forPknB activity

Discussion

Biochemical characterization of PstP and PknB

Although M tuberculosis encodes 11 STPKs (Cole et al1998) there is only one clear serinethreonine proteinphosphatase PstP which is a member of the PPM family(Bork et al 1996) We show here that its catalytic domainPstP1-240 dephosphorylates substrates previously phos-phorylated on serine or threonine but not on tyrosineresidues Furthermore its activity is strictly dependent onMn2+ or Mg2+ ions which is consistent with the deducedmetal-ion catalysed dephosphorylation mechanism forthis family (Das et al 1996)

On the basis of its amino acid sequence PknB (and allother mycobacterial STPKs) have been classified in thePkn2 family of prokaryotic STPKs (Leonard et al 1998)the cluster that most closely resembles their eukaryoticcounterparts and that could have arisen by early horizon-tal transfer from eukarya to bacteria with complex devel-

Fig 9 The putative phosphate-binding site in PknBA Surface representation of PknB (PDB code 1O6Y) colour-coded according to charge A cluster of four exposed arginine residues could provide a binding site for the two phosphorylated threonine residues Thr171 and Thr173 Sixteen residues from the activation loop (con-necting Ile163 to Ala180 and including the two phosphothreonines) are disordered in the crystal structureB Equivalent view of mouse PKA (PDB code 1ATP) in which the region corresponding to that missing in PknB is shown in stick rep-resentation The phosphate group of phospho-Thr197 makes hydro-gen-bonding interactions with the side chains of two arginine and one histidine residues

1502 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

opment cycles Recombinant full-length PknB hasalready been shown to possess kinase activity and auto-phosphorylation sites on both serine and threonine resi-dues (Av-Gay et al 1999) Here we studied a constructlimited to the catalytic core domain PknB1-279 as definedby sequence homology We found that this construct is anactive kinase showing that the juxtamembrane region is

not required for activity although it may still be involved infurther stabilization or activity regulation (see below)

PknB is regulated by phosphorylation of two Thr residues in the activation loop

Various mechanisms of eukaryotic protein kinase regula-

Fig 10 Kinase activity of the activation loop mutants of PknB MBP phosphorylation assays have been performed in parallel for the alanine mutants and the wild-type PknB1-279 Relative quantification of the kinase activity was obtained with the PhosphorImager T171A T173A and T171173 A mutants are ordf15 20 and 300 times less active than PknB1-279 respectively

Table 1 Phosphorylation status of wild-type and mutants PknB1-279

Protein

Phosphorylation statusa and amino acid(s) involvedb

His-Tag peptide Peptide S162-R189 Phosphorylated residues

PknBc 45ndash60 non-P close to100 di-P Thr171 and Thr17340ndash55 mono-P trace of tri-P d Thr171 Thr173 and (Ser169 or Ser166)

T171A 82 non-P close to100 di-P Thr173 and Ser16918 mono-P

T173A 87 non-P 96 mono-P Thr17113 mono-P 4 di-P Thr171 and (Ser169 or Ser166)

T171173 A 89 non-P 36 non-P ----11 mono-P 64 mono-P (Ser169 or Ser166)

a Refers to relative amounts of phosphorylated species present in Nt His-Tag peptide or in peptide S162-R189 populations Non-P mono-P di-P or tri-P indicates absence one two or three phosphate groups present respectively Peptide samples were isolated and quantified after proteintreatment with trypsin followed by HPLC and peak identification by MS as mainly described in Fig 8 and in Experimental proceduresb Modified amino acid(s) by phosphorylation were localized in the sequence S162-R189 by PSD-MS as exemplified in Fig 8B and C followingthe protocols described in Experimental procedures The phosphorylated serine of the Nt His-Tag peptide (MGSSHHHHHHSSGLVPR) was notidentifiedc Samples from three independently produced batches of PknB1-279 were testedd The phosphorylation of the third residue in the activation loop Ser 169 or Ser 166 appears of minor importance as the degree ofphosphorylation detected was systematically low or nul

Mycobacterial Serinethreonine kinase and phosphatase 1503

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

tion have been described (Johnson et al 1996 Hubbardand Till 2000 Huse and Kuriyan 2002) The transitionbetween active and inactive forms may occur via controlof access to the catalytic andor the substrate-binding siteor by rearrangement of structural elements involved incatalysis or substrate recognition Furthermore interac-tion with other protein domains or cofactors may takeplace It is noteworthy that a large number of these regu-lation mechanisms involve phosphorylationdephosphory-lation (inside or outside the catalytic domain) through anautocatalytic mechanism or by the action of other inter-vening kinases and phosphatases

The present study shows that the catalytic domain ofPknB autophosphorylates in vitro and is phosphorylatedwhen expressed in E coli To see whether PknB auto-phosphorylation could play a regulatory role we firstidentified phosphorylated residues in PknB Mass spec-trometry analysis indicated that two threonine residues ofthe activation loop (Thr 171 and Thr 173) are systemati-cally phosphorylated (presumably autophosphorylated)Other eukaryotic protein kinases also display two phos-phorylation sites in their activation loops such as MKK1(two Ser residues Alessi et al 1994) or ERK2 (a Thr anda Tyr residues both of which have to be phosphorylatedto form the active enzyme Robbins et al 1993) Theactivation loop is a major control element of an activeinactive conformational switch in numerous kinases(Steinberg et al 1993 Johnson et al 1996 Huse andKuriyan 2002) whose conformation often depends ontheir phosphorylation state (Johnson et al 1996) Fromits structural location this loop may control both theaccessibility to the catalytic site and the binding of thesubstrate A broad range of regulatory properties hasbeen assigned to this loop such as contributing to theproper alignment of the catalytic residues correcting therelative orientation of the two lobes permitting substratebinding andor stimulating ATP binding (Huse andKuriyan 2002)

The inhibitory effect of dephosphorylation of PknB onits kinase activity shows that phosphorylation is requiredfor full activity This is further confirmed by the mutagen-esis study of activation loop threonine residues Com-pared to the wild-type enzyme the two single mutantsstill phosphorylated on the remaining threonine displaycomparable reduced activities whereas the double-muta-tion further decreases the activity Hence Thr 171 and Thr173 play independent and equivalent but complementaryroles to reach maximal kinase activity

The structural role of the phosphothreonine residues inPknB remains unexplained because the activation loop isdisordered in the crystal structures (Ortiz-Lombardiacuteaet al 2003 Young et al 2003) This is not unusual inkinase structures It has been observed both in active andinactive kinases and does not indicate a particular phos-

phorylation state In some kinases phosphorylation of theloop fixes its conformation (Johnson et al 1996) and dis-order could thus indicate partial phosphorylation How-ever this does not seem to be the case for PknB as theactivation loop has no defined structure in the crystalstructure despite complete phosphorylation of both thre-onines Instead stabilization of the PknB loop could occurupon the binding of the peptide substrate through aninduced-fit mechanism or by additional intra- or intermo-lecular interactions with other factors outside the kinasecore In any case a positively charged region is observedin the PknB structure at the expected phosphothreonine-binding site equivalent to a similar cluster that in PKAbinds the single phosphorylated threonine Thr197(Fig 9)

Taken together these results strongly suggest thatPknB kinase activity can be regulated by the state ofphosphorylation of its activation loop in vivo through anautophosphorylation mechanism Interesting observa-tions can be drawn from the inspection of the activationloop sequences from the other M tuberculosis STPKsOne or both threonines are conserved in all but twoSTPKs (PknG and PknI have shorter loops) suggestingthat these enzymes should also be regulated by auto-phosphorylation in their activation loops Thus besidesthe same overall 3D structure and catalytic mechanismeukaryotic and prokaryotic kinases would also share thismechanism of regulation in spite of previous claims sug-gesting the absence of this process in prokaryotes (Motleyand Lory 1999) Further investigations are obviouslyrequired to determine the physiological relevance of PknBdephosphorylation by PstP and the effect of this proteinphosphatase on other kinases in particular PknA whichis present in the same operon

Other possible mechanisms of PknB regulation

Other mechanisms of kinase regulation could exist PknBis presumed to be a transmembrane protein with a puta-tive external ligand binding domain an organization sim-ilar to that found in sensor histidine kinases (Parkinson1993) and receptor tyrosine kinases (Schlessinger 2000)Binding of a ligand to the extracellular domain of the latterusually promotes receptor dimerization andor a structuralrearrangement that induces autophosphorylation andhence activation of the kinase domain Interestinglydimerization has recently been reported for PrkC (Madecet al 2002) a transmembrane STPK from B subtilis withhomology to PknB both in its Nt and Ct domains (Fig 4B)Another regulation mechanism described for both thetype I TGF-b receptor serinethreonine kinase (Huseet al 1999) and the ephrin receptor tyrosine kinase(EphB2)(Wybenga-Groot et al 2001) involves the main-tenance of an inactive state via the interaction of the

1504 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

juxtamembrane region with the kinase domain Uponligand stimulation of EphB2 the autophosphorylation ofTyr residues in the juxtamembrane sequence releases theinhibition and renders this sequence available for furtherinteraction with SH2 domains of target proteins(Wybenga-Groot et al 2001) The juxtamembrane regionis missing in PknB1-279 A recombinant construct of PknBcorresponding to the catalytic core of the kinase plus thejuxtamembrane sequence was also produced (see Exper-imental procedures) On preliminary analysis three phos-phorylation sites including Thr 294 and Thr 309 wereidentified in the juxtamembrane sequence (data notshown) Whereas the relevance of these phosphorylationevents in vivo remains to be determined it is worth notingthat these phosphoresidues could also provide recruit-ment sites for specific Forkhead-associated (FHA)domains see below

PknB and PstP may regulate mycobacterial cell growth

In prokaryotes genes involved in the same cellular pro-cess are frequently clustered often forming an operonThus co-localization of the pknB and pstP genes in thesame genomic region (Fig 1) reinforces the hypothesisthat these enzymes could intervene in the same signaltransduction pathway Furthermore the organization ofthis genomic region suggests the participation of addi-tional signal transduction elements including a secondSTPK (namely PknA) and two proteins harbouring FHAdomains (Rv0019c and Rv0020c) all of which are alsoconserved in other actinobacteria (Fig 1) The FHAdomains are small (Aring 130 aa) protein modules that medi-ate proteinndashprotein interaction via the recognition of aphosphorylated threonine on the target molecule (Duro-cher and Jackson 2002) In eukaryotes they are presentin numerous signalling and regulatory proteins such askinases phosphatases RNA-binding proteins and tran-scription factors Rv0019c (155 aa) corresponds to a sin-gle FHA domain whereas Rv0020c (527 aa) has twodomains a Ct FHA domain and a Nt domain that showsno homology with any known protein except with its ortho-logue in M leprae (ML0022) The FHA domain of Rv0020chas recently been characterized for its ability to bind phos-phorylated peptide ligands (Durocher et al 2000)

Also found in the same conserved operon (Fig 1) aretwo genes pbpA and rodA encoding proteins involved incontrolling cell shape and peptidoglycan synthesis duringcell growth (Matsuhashi 1994) Cell growth and develop-ment require the cell wall to have a dynamic structureIndeed the cell wall changes continuously during growthand developmental processes such as sporulation and inresponse to changes in the environment Moreover mor-phological adaptation like cell wall thickening could be animportant determinant for survival of the slow-growing

pathogenic mycobacteria in anaerobiosis (Cunninghamand Spreadbury 1998) Cross-linked peptidoglycan amajor component of the bacterial cell wall is synthesizedby penicillin-binding proteins (PBP) which are membraneanchored enzymes with two external catalytic modulesSome PBPs are only involved in specific phases of growthor development and for transglycosylase activity they areeach associated with a membrane protein partner Thusin E coli PBP2 and RodA are responsible for peptidogly-can synthesis during cell elongation and for determinationof the rod shape whereas PBP3 and FtsW are involvedin peptidoglycan synthesis during cell division (septation)In B subtilis a homologous couple (PBP and SpoVE) isthought to be engaged in spore formation

One reasonable working hypothesis that is currentlybeing tested involves PknA PknB and PstP along withother signalling modulators co-ordinately regulating cellelongation during growth Indeed recent data suggest aregulatory role for PknA in cell elongation (Chaba et al2002) and it has been speculated that the extracellulardomain of PknB could bind unlinked peptidoglycan (Yeatset al 2002) Kinases and phosphatase might have oppos-ing effects on the control of such a complex integratedpathway Tight regulation of the process of cell elongationcould therefore be a key element in mycobacterial devel-opment and provide a link between the intraextracellulargrowth phase and the latent lifestyle within the granulomaIf this model is correct inhibitors of STPK or even PstPwould represent attractive lead compounds for develop-ment into antitubercular agents capable of targeting Mtuberculosis in the different stages of its life cycle

Experimental procedures

Sequence analysis and modelling

For biochemical and structural (Ortiz-Lombardiacutea et al 2003)studies the catalytic kinase core of PknB was originallydefined using a homology modelling approach The 10 clos-est sequences from the Protein Data Bank were selectedand a multiple alignment was carried out using CLUSTALWAfter manual editing of the alignment the five sequencessharing highest identity with PknB (namely C elegansTwitchin kinase rabbit phosphorylase kinase mouse PKAand human CDK6 and CDK2) were used as templates forhomology modelling Using different combinations of thesetemplates various families of models were constructed andrefined with the program MODELLER (v 40) A comparison ofthe most self-consistent models allowed us to identify Gly 279as the likely end point for the a-helix I defining the C-terminusof the kinase catalytic core

Cloning and mutagenesis

Cosmid MTCY10H4 containing pknB (Rv0014c) and pstP(Rv0018c) was used in subcloning experiments A PknB con-

Mycobacterial Serinethreonine kinase and phosphatase 1505

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

struct corresponding to the putative cytoplasmic domain (cat-alytic domain + juxtamembrane sequence ndash aa 1ndash331) wasfirst obtained as some regions outside the kinase core couldstabilize the catalytic domain The following primers wereused for PCR amplification forward primer (with NdeI site)5cent-GATAGCCATATGACCACCCCTTCC-3rsquo and reverse primer(5cent-TAA codon + HindIII site) 5cent-AAACCGAAGCTTAACGGCCCACCG-3rsquo The digested and purified PCR product wasligated into the pET28 expression vector using the engi-neered NdeI and HindIII sites PknB1-331 was expressed as abroad heterogeneous protein probably reflecting heteroge-neity of its phosphorylation state as various phosphorylatedresidues were detected in the juxtamembrane region (datanot shown) A shorter construct corresponding to the corecatalytic domain (aa 1ndash279) was thus obtained introducinga stop codon by site-directed mutagenesis PknB mutants(T171A T173A T171173 A) were all obtained from this lastconstruct by the same method

The complete pstP gene was subcloned into pET28expression vector using the following primers forwardprimer (with NdeI site) 5cent-CGGGGGCATATGGCGCGCGTGA-3rsquo and reverse primer (TAA codon + HindIIIsite) 5cent-GCAGTCGTAAGCTTATGCCGCCG-3rsquo The con-struct corresponding to the catalytic domain of PstP (aa 1ndash240) was then obtained by introducing a stop codon throughsite-directed mutagenesis

All mutagenesis was done according to the Quick ChangeStratagene procedure Enzymes were purchased as followsthe T4 DNA ligase NdeI and DpnI restriction enzymes fromBiolabs HindIII and BglII restriction enzymes from Pharma-cia the Pfu and Pfu turbo polymerases from Stratagene Allconstructs were verified by DNA sequencing

Protein expression and purification

Escherichia coli BL21 (DE3) bacteria transformed with theappropriate plasmid were grown at 37infinC until late log phasein LuriandashBertani (LB) medium with antibiotic (kanamycin30 mg ml-1) Induction of expression was conducted for 12ndash16 h at low temperature (15infinC) after addition of 1 mM IPTGBacterial pellet was resuspended in 50 mM Hepes bufferpH 7 02 M NaCl in the presence of protease inhibitors andsonicated The lysate was cleared by centrifugation(20 000 g 30 min to 1 h) The supernatant containing solubleproteins was applied to Ni-column (Pharmacia) using anFPLC system and eluted by an imidazol gradient (0ndash05 M)A further step of gel filtration (Superdex 75) was required toseparate the aggregated material from the monomeric pro-teins and to remove imidazol and most of the Ni2+ cationsProteins were subsequently concentrated by means ofMacro- and Micro-sep concentrators (PallGellman) Proteinconcentration was determined using the Bio-Rad proteinassay Purity of the samples was checked by SDS-PAGEelectrophoresis

Protein kinase assays

The kinase assays were carried out in 20 ml of kinase buffer(Hepes 50 mM pH 7 DTT 1 mM Brij35 001) containing2 mM MnCl2 100 mM ATP and 1 mCi of [g-33P]-ATP For the

analysis of divalent cation preference various concentrationsof MnCl2 or MgCl2 were used as indicated in the Fig 1B Forautophosphorylation 5 mM final of the purified PknB wasused For phosphorylation of the MBP substrate by PknB orthe PknB mutants the enzymesubstrate ratio was 120 with05 mM kinase The reaction was started with the addition ofthe kinase and conducted at 30infinC for 10 min For the kineticsof MBP phosphorylation by PknB and the PknB mutants10 ml-aliquots of a scaled-up 60 ml reaction mixture were with-drawn at each indicated time The reaction was stopped bythe addition of SDS-PAGE sample buffer plus EDTA (25 mMfinal) Ten ml of the reaction were subjected to electrophore-sis In each case the reaction products were separated ona 12 SDS-polyacrylamide gel and the radiolabelled pro-teins visualized by auto-radiography To obtain relative quan-tification of the incorporation of radiolabelled ATP theradioactive samples were also analysed using a Phospho-rImager apparatus (STORM Molecular Dynamics) For test-ing kinase activity of PknB after various incubation times withPstP ATP and [g-33P]ATP were replaced by thio-gATP and[35S]ATP-gS respectively [g-33P]ATP and [35S]ATP-gS werepurchased from AmershamBiosciences MBP was fromInvitrogen

Protein phosphatase assays

Dephosphorylation of phosphoSerThr or phosphoTyr pro-teins by PstP was assayed using either MBP or a-casein(SIGMA) Phosphorylated [33P]SerThr-substrates or [33P]Tyr-substrates were prepared by phosphorylation of the proteinsusing either the catalytic subunit of PKA or the Abl proteintyrosine kinase In each case the kinase reaction was per-formed in 200 ml of buffer (50 mM Hepes pH 75 5 mMMgCl2 1 mM EGTA 2 mM DTT 001 Brij35) with 1 mMATP 75 mCi [g-33P]ATP 200 mM substrate and 25 units of PKAor 10 units of Abl kinase The reaction was incubated for 5 hat 30infinC Phosphorylated substrate was recovered by TCAprecipitation and extensively dialysed at 4infinC against a 25 mMTris buffer pH 75 with 01 mM EDTA 2 mM DTT and 001Brij35 Dephosphorylation assays were carried out in a 25 mlreaction mixture containing 50 mM Hepes buffer pH 7501 mM EDTA 1 mM DTT and 001 Brij35 5 mM MnCl2Phosphorylated [33P] substrates were used to a final concen-tration corresponding to 10 mM of incorporated phosphatesThe reaction was started with the addition of various concen-trations of the purified PstP (up to 200 ng25 ml ordf 03 mM)and incubated for 10 min at 30infinC The reaction was termi-nated by adding cold 20 TCA After centrifugation solublematerials were added to scintillation fluid and counted for therelease of inorganic phosphate The serinethreonine phos-phatase PP1 and the Tyrosine phosphatase T-Cell PTP wereused as control for the dephosphorylation of the phosphoSerThr substrates and the phosphoTyr substrates respectively(not shown) The dephosphorylation of PknB by PstP wasfirst performed using autophosphorylated [33P]-PknB that wasprepared according to the above protocol except that noextra kinase was added The reaction was performed in 15 mlof Hepes buffer 50 mM pH 7 DTT 1 mM Brij35 001 with2 mM MnCl2 [33P]-PknB and PstP were used at 5 mM and1 mM respectively and incubated 30 min at 30infinC The reac-tion products were resolved on a SDS-PAGE gel and the lost

1506 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

of labelling was visualized on the auto-radiography of thedried gel The dephosphorylation of PknB by PstP was alsosimply assayed by the appearance of a lower band on a gelcorresponding to dephosphorylated PknB The reaction wascarried out in 10 ml of the same buffer for 10 min at 30infinCexcept that PknB substrate was used at 1 mM various con-centrations of the phosphatase PstP were added from 50 to300 nM

Mass spectrometry analysis

Identification of phosphorylated sites was performed by massmeasurements in whole peptide mixtures and in purifiedHPLC fractions of proteins digested with trypsin (Promega05 mg per 20ndash50 mg of protein sample in 50 mM ammoniumbicarbonate buffer pH 84 overnight incubation at 35infinC)Twenty-six tryptic peptides covering 90 of the PknB1-279

sequence were thus identified (data not shown) whereasdigestion peptide products smaller than five amino acid res-idues could not be detected In some experiments proteinswere treated with a phosphatase before proteolytic cleavagealkaline phosphatase from Roche Diagnostics (20 enzymeunits per 20ndash40 mg of protein incubated in an assay mixtureaccording to instructions supplied by the manufacturer for1 h at 35infinC) or purified PstP enzyme as described elsewherein this section

MALDI-TOF MS was carried out in a Voyager DE-PROsystem (Applied Biosystems) equipped with a N2 laser source(l = 337 nm) Mass spectra were acquired for positive ionsin linear and reflector modes at an accelerating voltage of20 kV The matrix was prepared with a-cyano-4-cinnamicacid for peptides or with sinnapinic acid for proteins as sat-urated solutions in 02 trifluoroacetic acid in acetonitrile-H2O (50 vv) Measurement of peptide masses in reflectormode was performed under conditions of monoisotopic res-olution with the accuracy close to 50 ppm attained withexternal calibration For this purpose a mixture of the follow-ing peptide mass standards was included ([MH]+ monoiso-topic mass concentration) angiotensin I (129668 2 pmolml-1) ACTH 1ndash17 clip (209308 2 pmol ml-1) ACTH 18ndash39clip (246520 15 pmol ml-1) and ACTH 7ndash38 clip (3657933 pmol ml-1) Better accuracy was obtained when internalmass calibration was sometimes performed with alreadycharacterised peptides present in PknB tryptic digests Formass measurements of PknB proteins in linear mode eno-lase of Bakerrsquos yeast (average mass of the protonated molec-ular ion [MH]+ = 46672 and [MH2]+2 = 23336) was used asa calibration standard Samples for MS were usually pre-pared by spotting 05 ml of matrix solution and 05 ml of pep-tide solution or tiny droplets from a desalting microcolumneluted with matrix solution (see below) directly on the sampleplate

Selected peptides isolated from HPLC runs weresequenced by PSD-MS analysis of the y- ion series gener-ated from the samples (Kaufmann et al 1993) followinginstructions provided by the instrument manufacturer Whenadditional data were required to confirm a phosphorylationsite by MS sequencing the corresponding tryptic peptide wassubmitted to Ba(OH)2 treatment for dephosphorylation ofserine or threonine residues following published procedures(Resing et al 1995)

HPLC separations were performed in a reverse-phasecolumn (Vydac C18 150 yen 21 mm) equilibrated with 01trifluoroacetic acid in H2O (solvent A) and eluted with agradient of 007 trifluoroacetic acid in acetonitrile (solventB) Chromatographic conditions were as follows flow rate02 ml min-1 chart paper 2 mm min-1 gradient was from0 min to 20 min up to 10 B from 20 min to 100 min up to30 B from 100 min to 110 min up to 50 B from110 min to 115 min up to 100 B and then 100 B iso-cratic for 5 min more detection was by UV recording at220 nm

Relative amounts of the tryptic peptide A162-R189 show-ing different degrees and patterns of phosphorylation werecalculated for wild-type and mutant PknB enzymes (Table 1)Peak size of purified and identified HPLC peaks (accordingto MS and PSD-MS measurements) was measured and cor-rected according to the chromatographic response of eachpeptide tested in advance under identical chromatographicconditions as described above

For mass measurements HPLC fractions were sometimesconcentrated under a N2 gas flow freeze-dried or immobil-ised on reverse-phase Poros 10 R2 beads (Applied Biosys-tems) The latter was also a useful procedure to desalt smallpeptide or protein samples in batch or in home-made micro-columns (Gobom et al 1999) Virtual tryptic digestions andother mass calculations were performed with the GPMAW32(v402) program (Lighthouse Data)

Acknowledgements

We thank Nadine Honoreacute and Tong Nguyen for technicalassistance This work was funded by the Institut Pasteur(PTR no 46) the Genopole programme the PRFMMIP(MENRT France) and the European Community (QLRT-2000ndash02018) MOL was recipient of a FEBS Long-Termfellowship and CC thanks PEDECIBA (Uruguay) and ECOS(France) for partial financial support

References

Alessi DR Saito Y Campbell DG Cohen P Sithanan-dam G Rapp U et al (1994) Identification of the sitesin MAP kinase kinase-1 phosphorylated by p74raf-1EMBO J 13 1610ndash1619

Av-Gay Y Jamil S and Drews SJ (1999) Expression andcharacterization of the Mycobacterium tuberculosis serinethreonine protein kinase PknB Infect Immun 67 5676ndash5682

Barford D Das AK and Egloff MP (1998) The structureand mechanism of protein phosphatases insights intocatalysis and regulation Annu Rev Biophys Biomol Struct27 133ndash164

Bork P Brown NP Hegyi H and Schultz J (1996) Theprotein phosphatase 2C (PP2C) superfamily detection ofbacterial homologues Protein Sci 5 1421ndash1425

Brown CA Draper P and Hart PD (1969) Mycobacteriaand lysosomes a paradox Nature 221 658ndash660

Chaba R Raje M and Chakraborti PK (2002) Evidencethat a eukaryotic-type serinethreonine protein kinase fromMycobacterium tuberculosis regulates morphological

Mycobacterial Serinethreonine kinase and phosphatase 1507

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

changes associated with cell division Eur J Biochem 2691078ndash1085

Cole ST Brosch R Parkhill J Garnier T Churcher CHarris D et al (1998) Deciphering the biology of Myco-bacterium tuberculosis from the complete genomesequence Nature 393 537ndash544

Cowley SC Babakaiff R and Av-Gay Y (2002) Expres-sion and localization of the Mycobacterium tuberculosisprotein tyrosine phosphatase PtpA Res Microbiol 153233ndash241

Cunningham AF and Spreadbury CL (1998) Mycobacte-rial stationary phase induced low oxygen tension Cell wallthickening and localization of the 16-kilodalton alpha-crys-tallin homolog J Bacteriol 180 801ndash808

Dannenberg AM (1999) Pathophysiology basic aspects InTuberculosis and Nontuberculous Mycobacterial Infec-tions Schlossberg D (ed) Philadelphia WB SaundersCompany pp 17ndash47

Das AK Helps NR Cohen PT and Barford D (1996)Crystal structure of the protein serinethreoninephosphatase 2C at 20 Aring resolution EMBO J 15 6798ndash6809

Demaio J Zhang Y Ko C Young DB and Bishai WR(1996) A stationary-phase stress-response sigma factorfrom Mycobacterium tuberculosis Proc Natl Acad Sci USA93 2790ndash2794

Durocher D and Jackson SP (2002) The FHA domainFEBS Lett 513 58ndash66

Durocher D Taylor IA Sarbassova D Haire LF West-cott SL Jackson SP et al (2000) The molecular basisof FHA domain Phosphopeptide binding specificity andimplications for phospho-dependent signaling mecha-nisms Mol Cell 6 1169ndash1182

Dye C Scheele S Dolin P Pathania V and RaviglioneMC (1999) Consensus statement Global burden of tuber-culosis estimated incidence prevalence and mortality bycountry WHO Global Surveillance Monitoring Project J AmMed Assoc 282 677ndash686

Eiglmeier K Parkhill J Honore N Garnier T Tekaia FTelenti A et al (2001) The decaying genome of Myco-bacterium leprae Leprosy Rev 72 387ndash398

Fsihi H De Rossi E Salazar L Cantoni R Labo MRiccardi G et al (1996) Gene arrangement and organi-zation in a approximately 76 kb fragment encompassingthe oriC region of the chromosome of Mycobacterium lep-rae Microbiology 142 3147ndash3161

Gaidenko TA Kim TJ and Price CW (2002) The PrpCserine-threonine phosphatase and PrkC kinase haveopposing physiological roles in stationary-phase Bacillussubtilis cells J Bacteriol 184 6109ndash6114

Global Alliance for TB Drug Development (2001) Scientificblueprint for tuberculosis drug development Tuberculosis81 1ndash52

Gobom J Nordhoff E Mirgorodskaya E Ekman R andRoepstorff P (1999) Sample purification and preparationtechnique based on nano-scale reversed-phase columnsfor the sensitive analysis of complex peptide mixtures bymatrix-assisted laser desorptionionization mass spec-trometry J Mass Spectrom 34 105ndash116

Hanks SK and Hunter T (1995) Protein kinases 6 Theeukaryotic protein kinase superfamily kinase (catalytic)

domain structure and classification FASEB J 9 576ndash596

Hubbard SR and Till JH (2000) Protein tyrosine kinasestructure and function Annu Rev Biochem 69 373ndash398

Hunter T (1995) Protein kinases and phosphatases the yinand yang of protein phosphorylation and signaling Cell 80225ndash236

Hunter T (2000) Signaling ndash 2000 and beyond Cell 100113ndash127

Huse M and Kuriyan J (2002) The conformational plastic-ity of protein kinases Cell 109 275ndash282

Huse M Chen Y-G Massague J and Kuriyan J (1999)Crystal structure of the cytoplasmic domain of the type ITGFbeta receptor in complex with FKBP12 Cell 96 425ndash436

Johnson LN Noble ME and Owen DJ (1996) Activeand inactive protein kinases structural basis for regulationCell 85 149ndash158

Kaufmann R Spengler B and Luetzenkirchen F (1993)Mass spectrometric sequencing of linear peptides by prod-uct-ion analysis in a reflectron time-of-flight mass spec-trometer using matrix-assisted laser desorption ionizationRapid Commun Mass Spectrom 7 902ndash910

Kennelly PJ (2002) Protein kinases and protein phos-phatases in prokaryotes a genomic perspective FEMSMicrobiol Lett 206 1ndash8

Knighton DR Zheng JH Ten Eyck LF Ashford VAXuong NH Taylor SS and Sowadski JM (1991) Crys-tal structure of the catalytic subunit of cyclic adenosinemonophosphate-dependent protein kinase Science 253407ndash414

Koul A Choidas A Treder M Tyagi AK Drlica KSingh Y and Ullrich A (2000) Cloning and characteriza-tion of secretory tyrosine phosphatases of Mycobacteriumtuberculosis J Bacteriol 182 5425ndash5432

Koul A Choidas A Tyagi AK Drlica K Singh Y andUllrich A (2001) Serinethreonine protein kinases PknFand PknG of Mycobacterium tuberculosis characterizationand localization Microbiology 147 2307ndash2314

Leonard CJ Aravind L and Koonin EV (1998) Novelfamilies of putative protein kinases in bacteria andarchaea Evolution of the lsquoeukaryoticrsquo protein kinase super-family Genome Res 8 1038ndash1047

Madec E Laszkiewicz A Iwanicki A Obuchowski Mand Seror S (2002) Characterization of a membrane-linked SerThr protein kinase in Bacillus subtilis implicatedin developmental processes Mol Microbiol 46 571ndash586

Manabe YC and Bishai WR (2000) Latent Mycobacte-rium tuberculosis-persistence patience and winning bywaiting Nature Med 6 1327ndash1329

Matsuhashi M (1994) Utilization of lipid-linked precursorsand the formation of peptidoglycan in the process of cellgrowth and division In Bacterial Cell Wall Ghuysen J-Mand Hakenbeck R (eds) Amsterdam-London Elsevier

Motley ST and Lory S (1999) Functional characterizationof a serinethreonine protein kinase of Pseudomonasaeruginosa Infect Immun 67 5386ndash5394

Muntildeoz-Dorado J Inouye S and Inouye M (1991) A geneencoding a protein serine-threonine kinase is required fornormal development of Myxococcus xanthus a Gram-neg-ative bacterium Cell 67 995ndash1006

1508 B Boitel et al

copy 2003 Blackwell Publishing Ltd Molecular Microbiology 49 1493ndash1508

Ortiz-Lombardiacutea M Pompeo F Boitel B and Alzari PM(2003) Crystal structure of the catalytic domain of the PknBserinethreonine kinase from Mycobacterium tuberculosisJ Biol Chem 278 13094ndash13100

Parkinson JS (1993) Signal transduction schemes of bac-teria Cell 73 857ndash871

Parrish NM Dick JD and Bishai WR (1998) Mecha-nisms of latency in Mycobacterium tuberculosis TIBS 6107ndash112

Peirs P De Wit L Braibant M Huygen K and ContentJ (1997) A serinethreonine protein kinase from Mycobac-terium tuberculosis Eur J Biochem 244 604ndash612

Resing KA Johnson RS and Walsh KA (1995) Massspectrometric analysis of 21 phosphorylation sites in theinternal repeat of rat profilaggrin precursor of an interme-diate filament associated protein Biochemistry 34 9477ndash9487

Robbins DJ Zhen E Owaki H Vanderbilt CA EbertD Geppert TD and Cobb MH (1993) Regulation andproperties of extracellular signal-regulated protein kinases1 and 2 in vitro J Biol Chem 268 5097ndash5106

Russell DG (2001) Mycobacterium tuberculosis here todayand here tomorrow Nature Rev Mol Cell Biol 2 569ndash577

Schlessinger J (2000) Cell signaling by receptor tyrosinekinases Cell 103 211ndash225

Shi L Potts M and Kennelly PJ (1998) The serinethreonine andor tyrosine-specific protein kinases and pro-tein phosphatases of prokaryotic organisms a family por-trait FEMS Microbiol Rev 22 229ndash253

Steinberg RA Cauthron RD Symcox MM andShuntoh H (1993) Autoactivation of catalytic (C-alpha)subunit of cyclic AMP-dependent protein kinase by phos-phorylation of threonine 197 Mol Cell Biol 13 2332ndash2341

Stock AM Robinson VL and Goudreau PN (2000) Two-component signal transduction Annu Rev Biochem 69183ndash215

Sturgill-Koszycki S Schaible UE and Russell DG(1996) Mycobacterium-containing phagosomes areaccessible to early endosomes and reflect a transitionalstate in normal phagosome biogenesis EMBO J 156960ndash6968

Wurgler-Murphy SM and Saito H (1997) Two-componentsignal transducers and MAPK cascades TIBS 22 172ndash176

Wybenga-Groot LE Baskin B Ong SH Tong J Paw-son T and Sicheri F (2001) Structural basis for autoin-hibition of the EphB2 receptor tyrosine kinase by theunphosphorylated juxtamembrane region Cell 106 745ndash757

Yeats C Finn RD and Bateman A (2002) The PASTAdomain a (-lactam-binding domain TIBS 27 438ndash440

Young TA Delagoutte B Endrizzi JA Falick AM andAlber T (2003) Structure of Mycobaterium tuberculosisPknB supports a universal activation mechanism for SerThr protein kinases Nature Struct Biol 10 168ndash174

Zhang CC (1996) Bacterial signalling involving eukaryotic-type protein kinases Mol Microbiol 20 9ndash15