Báo cáo khoa học: Loss of kinase activity in Mycobacterium tuberculosis multidomain protein Rv1364c pot

tuberculosis Rv1364c resulted in the prediction of various domains, namely a phosphatase RsbU domain, an anti-SigF RsbW domain, and an anti-anti-SigF RsbV domain.. Both the Rv1364c RsbW

multidomain protein Rv1364c

Preeti Sachdeva1,2, Azeet Narayan1, Richa Misra1, Vani Brahmachari2and Yogendra Singh1

1 Institute of Genomics and Integrative Biology (CSIR), Delhi, India

2 Dr B R Ambedkar Center for Biomedical Research, University of Delhi, India

Despite global efforts to control tuberculosis, it

remains an epidemic, with one-third of the world’s

population being infected by its etiologic agent,

Myco-bacterium tuberculosis, and over 1.5 million people

dying from the disease each year The notorious

suc-cess of M tuberculosis as a highly adapted pathogen

rests upon its ability to establish a persistent infection

in the hostile environment of the host cell through

mechanisms involving transcriptional reprogramming, ensuring metabolic slowdown and upregulation of vir-ulence and stress response pathways [1] Switching of alternative sigma factors is known to regulate global gene expression to cope with the numerous environ-mental conditions encountered during the establish-ment of a successful infection [2] The M tuberculosis genome encodes 13 sigma factors, including 10

alter-Keywords

kinase; Mycobacterium; RsbW; Rv1364c;

SigF

Correspondence

Y Singh, Institute of Genomics and

Integrative Biology, Mall Road, Delhi

110007, India

Fax: +11 2766 7471

Tel: +11 2766 6156

E-mail: ysingh@igib.res.in

(Received 7 September 2008, revised 16

October 2008, accepted 22 October 2008)

doi:10.1111/j.1742-4658.2008.06753.x

The alternative sigma factors are regulated by a phosphorylation-mediated signal transduction cascade involving anti-sigma factors and anti-anti-sigma factors The proteins regulating Mycobacterium tuberculosis sigma factor F (SigF), anti-SigF and anti-anti-SigF have been identified, but the factors catalyzing phosphorylation–dephosphorylation have not been well estab-lished We identified a distinct pathogenic species-specific multidomain pro-tein, Rv1364c, in which the components of the entire signal transduction cascade for SigF regulation appear to be encoded in a single polypeptide Sequence analysis of M tuberculosis Rv1364c resulted in the prediction of various domains, namely a phosphatase (RsbU) domain, an anti-SigF (RsbW) domain, and an anti-anti-SigF (RsbV) domain We report that the RsbU domain of Rv1364c bears all the conserved features of the PP2C-type serine⁄ threonine phosphatase family, whereas its RsbW domain has certain substitutions and deletions in regions important for ATP binding Another anti-SigF protein in M tuberculosis, UsfX (Rv3287c), shows even more unfavorable substitutions in the kinase domain Biochemical assay with the purified RsbW domain of Rv1364c and UsfX showed the loss of ability of autophosphorylation and phosphotransfer to cognate anti-anti-SigF proteins or artificial substrates Both the Rv1364c RsbW domain and UsfX protein display very weak binding with fluorescent ATP analogs, despite showing functional interactions characteristic of anti-SigF proteins

In view of conservation of specific interactions with cognate sigma and anti-anti-sigma factor, the loss of kinase activity of Rv1364c and UsfX appears to form a missing link in the phosphorylation-dependent interac-tion involved in SigF regulainterac-tion in Mycobacterium

Abbreviations

GST, glutathione S-transferase; MBP, myelin basic protein; MursiF, multidomain regulator of sigma factor F; PAC, PAS domain-associated C-terminus; PAS domain, Per-Arnt-Sim domain; pNP, p-nitrophenol; pNPP, p-nitrophenyl phosphate; PP, protein phosphatase; SigA, sigma factor A; SigB, sigma factor B; SigF, sigma factor F; TNP-ATP, 2,4,6-trinitrophenyl ATP; UPD, RsbU ⁄ phosphatase domain; VSD, RsbV ⁄ substrate domain; WKD, RsbW ⁄ kinase domain.

native sigma factors [3] One of the alternative sigma

factors of M tuberculosis, sigma factor F (SigF) is

responsible for transcription of gene products of

importance to infection and dormancy processes,

including genes involved in the biosynthesis of the cell

envelope and sigma factor C (SigC) [4] An M

tuber-culosisSigF-deleted strain grows to a three-fold higher

density in stationary phase than the wild-type strain,

and is attenuated for virulence in a mouse model of

infection [5]

The regulation of expression and activity of sigma

factors is brought about by phosphorylation and

pro-tein–protein interaction events in a partner-switching

mechanism involving anti-sigma factors and

anti-anti-sigma factors M tuberculosis SigF is related to anti-anti-sigma

factor B (SigB), a stress-response specific sigma factor

of Bacillus subtilis, and SigF of B subtilis, a

sporula-tion-specific sigma factor [6] In both the SigB and

SigF regulatory pathways of B subtilis, the activity of

the sigma factor is negatively regulated by the cognate

anti-sigma factor, RsbW, and SpoIIAB, respectively,

which hold the transcription factor in an inactive

com-plex Release of SigB and SigF from the complex is

mediated by the anti-anti-sigma factors, RsbV and

SpoIIAA, respectively The action of the

anti-anti-sigma factors is counteracted by the kinase activity of

the dual-function RsbW and SpoIIAB proteins, which

phosphorylate and thereby inactivate their respective

anti-anti-sigma factors Finally, the phosphorylated

RsbV and SpoIIAA proteins are reactivated by the

action of phosphatases, RsbU and SpoIIE respectively

Other proteins, such as RsbT, act upstream by binding

and activating the phosphatase activity of RsbU in

B subtilis [7–9] Similarly, the M tuberculosis genome

has been shown to possess a bona fide anti-SigF

protein, UsfX⁄ RsbW, as well as two anti-anti-SigF

proteins, RsfA and RsfB RsfB is regulated by

phos-phorylation, as a mutation that is believed to mimic

phosphorylation renders it nonfunctional However,

RsfB is not phosphorylated by UsfX in an in vitro

kinase reaction [10] The upstream molecules in this

event, i.e a kinase and a phosphatase regulating

UsfX–RsfB interaction, remain to be elucidated

The M tuberculosis genome analysis reveals another

potential SigF regulatory gene, Rv1364c encoding a

protein with multidomain architecture, in which the

entire regulatory cascade akin to the RsbU–RsbW–

RsbV system in B subtilis appears to be present within

a single polypeptide with an additional sensor domain,

the Per-Arnt-Sim (PAS) domain [11] Rv1364c is

upregulated during nutrient starvation in M

tuber-culosis [12], whereas its Mycobacterium bovis ortholog

is upregulated in response to environmental changes

encountered within the macrophages [11] In a yeast two-hybrid-based study, Rv1364c has been reported to interact with SigF as well as UsfX, and interdomain interactions between its RsbW and RsbV domains also occur [13] The role and mechanism of regulation of the multidomain protein Rv1364c are intriguing, and underline the need to study this component of the

M tuberculosis SigF regulation network The present study focuses on the functional characterization and role of Rv1364c in phosphorylation–dephosphorylation

of anti-anti-sigma factors, which are known to be important for regulation of SigF in M tuberculosis

Results and Discussion

Domain and genomic organization of Rv1364c orthologs

The gene product encoded by M tuberculosis Rv1364c, annotated as rsbU [3], represents a multidomain pro-tein comprising fused units that occur as independent stand-alone proteins in the same and other distant bac-terial species As the Rv1364c domain composition represents a previously unexemplified unique fusion of the sensor–RsbU–RsbV–RsbW module for SigF regu-lation, we have renamed the protein as putative multi-domain regulator of SigF (MursiF) (Fig 1A) We performed a comparative genomic analysis of sequenced mycobacterial genomes to identify MursiF orthologs across the Mycobacterium genus, using full-length M tuberculosis H37Rv MursiF as well as its individual domains as query sequences The analysis revealed the conservation of MursiF orthologs across all sequenced M tuberculosis strains, namely M tuber-culosis H37Ra, M tuberculosis F11, M tuberculosis CDC1551, M tuberculosis C and all other sequenced slow-growing pathogenic Mycobacterium spp., namely

M bovis AF2122⁄ 97, Mycobacterium avium 104,

M avium paratuberculosis K-10, Mycobacterium mari-num and Mycobacterium ulcerans (Fig 1A) MursiF ortholog is, however, absent in Mycobacterium leprae, where the gene encoding SigF itself is known to be a pseudogene [14]

On the other hand, nonpathogenic fast-growing Mycobacterium spp., such as Mycobacterium smegma-tis, Mycobacterium gilvum, Mycobacterium vanbaalenii, Mycobacterium sp MCS, Mycobacterium sp KMS, and Mycobacterium sp JLS, were found to have no sequence homologous to the complete multidomain module of MursiF Search of the M smegmatis gen-ome using the RsbU domain of MursiF as a query sequence revealed a protein annotated as response reg-ulator receiver protein (MSMEG_6131) This protein

comprises two domains, a phosphatase domain with

44% similarity to the MursiF RsbU domain, and a

receiver domain analogous to the phosphoacceptor

protein of histidine kinases (Fig 1B) Unlike M

tuber-culosis mursiF, the MSMEG_6131 gene possibly forms

an operon (intergenic distance 3 bp) with a gene

pair comprising a two-component system sensor

kinase (MSMEG_6130) and a response regulator

(MSMEG_6128) (Fig 1B) Interestingly, adjacent and

in opposite orientation to this putative operon in

M smegmatis, we identified two genes in tandem,

MSMEG_6129 and MSMEG_6127 encoding protein

sequences with 44 and 58% similarity to the RsbW

domain and the RsbV domain of M tuberculosis

MursiF, respectively (Fig 1B) The analysis of genomes

of other nonpathogenic Mycobacterium spp namely

M gilvum, M vanbaalenii, Mycobacterium sp MCS,

Mycobacteriumsp KMS, and Mycobacterium sp JLS,

also revealed the conservation of sequences

ortholo-gous to the M smegmatis response regulator receiver

protein (MSMEG_6131) and two-component system

(MSMEG_6130, MSMEG_6128) (Fig 1B) The genes encoding proteins homologous to the RsbW and RsbV domains of MursiF (Fig 1A) in these species were, however, found to be located far apart from response regulator receiver protein orthologs The rsbW gene is spaced precisely 25 genes away in the closely related strains Mycobacterium sp MCS, Mycobacterium sp KMS, and Mycobacterium sp JLS, and 15 genes away

in the closely related species M gilvum and M vanba-alenii, from the gene encoding the response regulator receiver protein (Fig 1B) These observations, very interestingly, suggest that stand-alone RsbU, RsbW and RsbV encoding genes in nonpathogenic mycobac-terial species have converged to form a gene encoding fused mutidomain RsbU–W–V module, specifically in pathogenic Mycobacterium spp The selective advan-tage of domain fusion lies in the increased efficiency of coupling and coregulation of the corresponding bio-chemical reaction or signal transduction step as well as expression of the fused domains [15] It is likely that radically fused genes, which may emerge as a result of

Membrane protein

Rv1364c RsfA

Mb1399 Membrane

protein

RsfA

MAV_1619 Acyl-CoA

dehydrogenase

Hypothetical protein

MAP2361 Hypothetical

protein

Acyl-CoA dehydrogenase

Acyl-CoA dehydrogenase PE-PGRS

family pseudogene

MUL_3853

Acyl-CoA dehydrogenase

PE family protein

MM3991

M bovis

M tuberculosis

H 37 Rv

M marinum

M avium

paratuberculosis

M ulcerans

M avium

M smegmatis

M sp KMS

M gilvum

M sp MCS

M sp JLS

M vanabaalenii

MSMEG_6131 Glucarate

dehydratase Sensor

Kinase Response regulator RsbW RsbV

PAS domain RsbU/Phosphatase domain (UPD) domain (WKD) RsbW/Kinase RsbV/Substrate domain (VSD)

Coiled coil

domain

RsbU/Phosphatase domain

1

Hypothetical protein

Response regulator Sensor Kinase Mmcs_2688 Hypothetical

protein RsbW RsbV

Response regulator Sensor

Kinase

Acyl-transferase Mflv_3268

Phospho-diesterase

RsbW RsbV

Response regulator Sensor

Kinase

Acyl-transferase Mvan_2987

Phospho-diesterase RsbW RsbV

Hypothetical protein

Response regulator Sensor Kinase Mjls_2718 Hypothetical

protein RsbW RsbV

Response regulator Sensor Kinase Hypothetical

protein Mkms_2732 Hypothetical

protein RsbW RsbV Signal receiver domain

Fig 1 Schematic representation of domain architecture and genomic organization of Rv1364c and its orthologs in pathogenic Mycobacte-rium spp (A) and response regulator receiver protein orthologs in nonpathogenic MycobacteMycobacte-rium spp (B) M tuberculosis Rv1364c and its orthologs are shown as ( ) arrows, and M smegmatis response regulator receiver protein and its orthologs are shown as ( ) arrows The numbers below the domain architecture diagram (shown as boxes) refer to amino acids defining boundaries of each of the domains The two proteins share a common domain, RsbU ⁄ phosphatase domain ( ) The sequences homologous to other two domains, the RsbW domain ( ) and the RsbV domain ( ), of Rv1364c (A) exist as independent genes in nonpathogenic Mycobacterium spp (B) represents the region containing a large number of genes separating the response regulator receiver protein and RsbW genes; 25 genes in Mycobacterium sp MCS, Mycobacterium sp KMS and Mycobacterium sp JLS, and 15 genes in M gilvum and M vanbaalenii.

genome rearrangements, fix in a population because

they have a novel function that is advantageous to the

organism [16] Fusion of domains associated with a

reg-ulatory pathway for stress adaptation to form a

contig-uous polypeptide may be advantageous in the

evolutionary optimization of the genome of pathogenic

Mycobacterium spp The unique pathogen-specific

domain architecture and its upregulation in

Mycobacte-riumresiding in a macrophage environment [11] makes

MursiF a protein of utmost importance

Sequence analysis of MursiF

In silico analysis was performed to determine the

nat-ure and domain featnat-ures of M tuberculosis H37Rv

MursiF This 653 amino acid protein has an estimated

molecular mass of 69.523 Da and a pI value of 4.763

MursiF is predicted to be a soluble protein by analysis

programs MursiF has a PAS sensor domain at its

N-terminus, with an adjacent PAC (PAS

domain-asso-ciated C-terminus) region (Fig 1A) PAS domains are

sensory modules that undergo conformational changes

in the presence of various physical stimuli and ligand

molecules [17] PAS domains show conservation in

three-dimensional fold and dynamic properties rather

than in amino acid sequences [18] The MursiF PAS

domain has 38% similarity to that of B subtilis RsbP,

an energy stress-dependent serine phosphatase [19]

Interestingly, in addition to a PAS sensor domain and

RsbU–W–V module, MursiF has a coiled coil motif

spanning residues 135–165, as determined by the

multicoil [20] and smart [21] programs (Fig 1A)

The coiled coil motif is a structural motif involved in

oligomerization [22], and may play an important role

in protein–protein interactions [23], which form the

fundamental basis of the SigF regulation mechanism

The RsbU⁄ phosphatase domain (UPD) of MursiF is

homologous to protein phosphatase PP2C-class

ser-ine⁄ threonine phosphatase, and shows 43% similarity

with B subtilis RsbU Multiple sequence alignment of

MursiF and other similar bacterial PP2C family

mem-bers shows the conservation of all critical residues of

11 characteristic motifs, including those predicted to

be involved in divalent metal ion coordination, Asp211

(subdomain II) and Asp328 (subdomain VIII) (Fig 2)

However, the sequence lacks the Va- and Vb-boxes of

the PP2C-type catalytic domain, as reported for the

RsbU, RsbX and SpoIIE family of phosphatases [24]

MursiF UPD possesses a PAS sensor domain at its

N-terminus (Fig 1A) in place of a motif for interaction

with an activator, RsbT, that is present in Bacillus and

Staphylococcushomologs During environmental stress,

RsbT positively regulates RsbU phosphatase activity

[25] No sequences homologous to genes encoding the RsbRST module are present in the M tuberculosis genome In this scenario, recruitment of a signaling domain, together with loss of a domain that mediates stress-induced interaction with an activator, suggests a direct sensing mechanism for stress signals by MursiF The putative anti-SigF domain of MursiF, the RsbW⁄ kinase domain (WKD), shows 40% similarity

to B subtilis RsbW, a serine kinase belonging to the GHKL family of kinases This family of kinases is defined by the presence of an ATP-binding fold called the ‘Bergerat fold’, comprising three motifs, the N-, G1- and G2-boxes, which have been found to be conserved in histidine kinases and ATPases [26] Anal-ysis of the MursiF WKD sequence revealed conserva-tion of most of the residues characteristic of the N-, G1- and G2-boxes; however, some significant vari-ations were observed Careful comparative analysis of MursiF WKD sequences across all sequenced Myco-bacterium spp and functionally characterized RsbW proteins from other genera was therefore carried out using multiple sequence alignment (Fig 3A) We noted that a region speculated to be a part of the ATP lid in functional RsbW sequences of other genera is deleted

in the WKD sequences of all MursiF orthologs identi-fied across Mycobacterium spp (Fig 3A) The ATP lid changes its conformation on nucleotide binding, and is presumed to couple ATP binding to function-specific interdomain associations [27] Mutagenesis of the pro-posed hinge of the ATP lid in a GHKL family kinase, EnvZ, has shown that this region is essential for kinase activity [28] Furthermore, a conserved lysine residue close to the N-box of histidine kinases, which has been shown to be important for nucleotide binding as well

as catalysis [29], is substituted in the WKD sequences

of all MursiF orthologs (Fig 3A) The novel genes formed as a result of fusion of two or more genes are believed to experience a burst of rapid adaptive substi-tution shortly after they are formed, followed by a slowing of evolution, which is consistent with increased evolutionary constraint [16] Less than 50% similarity

of MursiF UPD and WKD and the aforementioned divergence of MursiF WKD sequences from their orthologous sequences in other bacteria may be attrib-uted to adaptive evolution

The M tuberculosis genome encodes another func-tional anti-SigF protein, UsfX (Rv3287c) [10], which has insignificant similarity to the MursiF WKD sequence UsfX has been shown to catalyze phospho-transfer to an artificial substrate, myelin basic protein (MBP) [30], but not to any of the anti-anti-sigma fac-tor proteins [10,30] We carried out a comparative analysis of UsfX sequences across all sequenced

Myco-bacterium spp and functionally characterized RsbW

proteins from other genera (Fig 3B) We found that

M tuberculosis UsfX, surprisingly, has substantially

divergent motifs (45% similarity to B subtilis RsbW),

and completely lacks the G1-box consensus sequence

(Fig 3B) Furthermore, the conserved amino acids in

the signature sequences of the N- and G2-boxes have

been substituted with other less similar residues

(Fig 3B), and the possibility of functional competence

of these substitutions remains to be studied

Impor-tantly, a Bordetella BtrW (RsbW ortholog) N-box

mutant has been reported to be incapable of

phosphor-ylating its substrate, BtrV (RsbV homolog), whereas its BtrW G1-box mutant phosphorylated BtrV to a les-ser extent than its wild-type counterpart and is also defective in the ability to form a stable complex with the phosphorylatable form of BtrV [31] Similar to MursiF WKD, the UsfX sequence also shows a dele-tion of the ATP lid region as well as an absence of the lysine residue close to the N-box of histidine kinases (Fig 3A,B) However, in view of the presence of sev-eral divergent motifs in the UsfX sequence, it seems to have acquired a large number of deleterious mutations

in an independent evolutionary event In view of the

Fig 2 Comparison of the MursiF RsbU ⁄ phosphatase domain with PP2C family serine ⁄ threonine phosphatases of similar classes from other organisms MursiF of M tuberculosis was aligned with the PP2C domains of RsbU of B subtilis, SpoIIE of B subtilis and IcfG of Synecho-cystis sp using T-COFFEE Identical amino acids are indicated by asterisks, high similarity is indicated by double dots, and lower similarity is indicated by single dots The gaps are introduced to optimize the alignment and are indicated by the dashes Various motifs described in the text are marked and shown in bold Two conserved aspartate residues, Asp211 and Asp328, involved in binding divalent cations and mutated in the study are indicated as shaded residues.

B

Fig 3 Comparison of mycobacterial MursiF WKD sequences (A) and mycobacterial UsfX sequences (B) with functionally characterized RsbW sequences from other genera Alignments were done using T-COFFEE Identical amino acids are indicated by asterisks, high similarity is indicated by double dots, and lower similarity is indicated by single dots The gaps are introduced to optimize the alignment and are indicated

by the dashes Various motifs described in the text are marked and shown in bold.

aforementioned deletions and mutations, the ability of

MursiF WKD and UsfX protein to act as functional

kinases in the SigF regulation cascade is questionable

and needs to be addressed

The putative anti-anti-SigF domain of MursiF, the

RsbV⁄ substrate domain (VSD), has 50% similarity to

B subtilisRsbV The domain has been shown to

inter-act with MursiF WKD [13], and has all the conserved

features of the anti-anti-sigma factor sequence

(Fig S1) A serine residue, Ser600, at the predicted

phosphorylation site is conserved in MursiF VSD

(Fig S1)

Protein expression and purification

The above-mentioned in silico analysis necessitates

bio-chemical analysis of MursiF and UsfX In order to

functionally characterize full-length MursiF as well as

each of its domains, the full-length M tuberculosis

H37Rv mursiFgene and its individual domains, namely

upd, wkd, and vsd, were cloned and expressed as

His⁄ glutathione S-transferase (GST)-tagged

recombi-nant proteins in Escherichia coli (Fig 1A, Table 1) In

addition, tagged M tuberculosis SigF and

His-tagged and GST-His-tagged M tuberculosis UsfX were

also overexpressed in E coli All the proteins were

purified as hexa-His (H) or GST-fusion (G) proteins

(H-MursiF, H-UPD, H-WKD, G-WKD, H-VSD,

G-VSD, H-UsfX, G-UsfX, SigF) using affinity

chro-matography MursiF and its phosphatase domain

(UPD) mutant proteins carrying aspartate residue

(D211A, D328A) mutations (H-MursiF-D211A,

H-UPD-D211A, H-MursiF-D328A, H-UPD-D328A)

were subsequently purified using the same strategy On electrophoretic analysis of the purified proteins, H-MursiF (and its mutants), H-UPD (and its mutants), H-WKD, G-WKD, H-VSD, G-VSD, H-UsfX and G-UsfX were detected as 73, 47, 18, 44,

14, 37, 22 and 45 kDa proteins, respectively (Fig 4) These protein sizes were consistent with predicted mole-cular masses along with appropriate hexa-His (3 kDa)

or GST (26 kDa) affinity tags, except for UsfX, which migrated at a size slightly higher than the expected molecular mass The identity of purified proteins was also confirmed by western blot using monoclonal anti-bodies against penta-His and GST (data not shown)

MursiF UPD characterization The phosphatase activity of purified full-length MursiF (H-MursiF) and MursiF UPD was determined by its ability to dephosphorylate a small molecule substrate, p-nitrophenyl phosphate (pNPP), thereby forming a p-nitrophenolate ion, which is detected by measuring absorbance at 405 nm (Fig 5A) Further characteriza-tion revealed that MursiF phosphatase activity has a

pH optima of 8.5 and an optimum temperature of

37C (data not shown) The activity of H-MursiF was found to be strictly Mn2+-dependent (Fig 5B), being maximal at 2.5 mm Mn2+ (data not shown) Other cations such as Ca2+, Ba2+, Zn2+, Sr2+, Co2+ and

Ni2+ failed to substitute for Mn2+, whereas Mg2+ was found to inhibit pNPP hydrolysis by H-MursiF (Fig 5B) Two conserved aspartate residues, Asp211 and Asp328, at positions known to be involved in metal ion coordination were mutated (Fig 2) The two

Table 1 Summary of expression vectors used in the study.

mutant proteins, D211A and

H-MursiF-D328A, showed  10-fold lower activity than

wild-type MursiF (Fig 6) This further emphasizes the

divalent cation dependence of MursiF activity and the

crucial role of Asp211 and Asp328 in Mn2+

coordina-tion In an attempt to assign MursiF to a specific class

of phosphatases, the effect of various class-specific phosphatase inhibitors, such as sodium orthovanadate, calyculin A, cyclosporine, and okadaic acid, was stud-ied Insensitivity to okadaic acid, a potent inhibitor of the PP2A and PP2B family of phosphatases, is one of the unique characteristics of the PP2C subfamily [32] None of the PP1, PP2A or PP2B class-specific inhibi-tors, including okadaic acid, had any effect on the phosphatase activity of MursiF Only sodium fluoride,

a nonspecific phosphatase inhibitor, inhibited pNPP hydrolysis by H-MursiF by 30% (Fig 7) MursiF is therefore an Mn2+-dependent PP2C-type alkaline phosphatase that is most active in the physiological temperature range The purified phosphatase domain (H-UPD) of MursiF was also studied, and was found

to have similar characteristics (data not shown)

P ro

t ei n

m ar

k er

220

116

97

66

31

21

14.4

220

116

97

66

31

21

45

H- M

u rs

i F

H- M

u rs

i F D2

1 1A

H- M

u rs

i F D3

2 8A

H - UP

D H- WKD G - WKD G - V

S D H- V S

D

P ro

t ei n

m ar

k er

H- Us

f X

G - Us

f X H- S F

Fig 4 Analysis of recombinant proteins by SDS ⁄ PAGE Affinity-purified His-tagged (H-)

or GST-tagged (G-) fusion proteins were subjected to 13% SDS ⁄ PAGE and stained with Coomassie brilliant blue The numbers

on the left indicate sizes of bands of molec-ular mass marker The proteins analyzed are indicated at the top of each lane, and details are listed in Table 1.

0

2

4

6

8

A

B

Time (min)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

No

ion

Mg Mn Ca Ba Ni Zn Co Sr

Divalent cation (5 mM)

A405

Fig 5 Biochemical characterization of MursiF phosphatase activity.

(A) Time kinetics of phosphatase activity of MursiF were

deter-mined by the hydrolysis of a low molecular weight substrate,

pNPP, as measured by absorbance at 405 nm The activity was

expressed as micromoles of pNP liberated ⁄ lg of protein (B)

Puri-fied H-MursiF (5 lg) was incubated with pNPP in reaction buffer

containing different divalent cations (Mg 2+ , Ca 2+ , Ba 2+ , Zn 2+ , Sr 2+ ,

Co 2+ , and Ni 2+ ) A405 nmfor each of the reactions at 100 min is

plot-ted Each value is the average of three individual reactions and is

given as mean ± SD.

0 0.2 0.4 0.6 0.8

1 1.2

MursiF-

WT

MursiF- D211A

MursiF- D328A

A405

Fig 6 Comparative analysis of phosphatase activities of wild-type and mutant MursiF MursiF-WT, unaltered ⁄ wild-type H-MursiF; MursiFD211A, H-MursiF carrying the D211A mutation; Murs-iFD328A, H-MursiF carrying the D328A mutation (Table 1) Con-served aspartate residues (as marked in Fig 2), predicted to be involved in metal ion coordination, were mutated, and proteins were assayed for pNPP hydrolysis A 405 nm for wild-type protein is normalized as 1, and relative values for mutant proteins are plotted Each value is the average of three individual reactions and is given

as mean ± SD.

To study protein dephosphorylation by MursiF,

[32P]serine⁄ threonine-phosphorylated or [32

P]tyrosine-phosphorylated artificial substrates such as casein,

histone and MBP were incubated with H-MursiF and

H-UPD, but no dephosphorylation of any of the

artifi-cial substrates was seen The reaction conditions were

varied over a large number of biophysical and

bio-chemical parameters In all cases, the level of free

inor-ganic phosphate released from 32P-phosphorylated

substrates on incubation with MursiF was found to be

insignificant and almost comparable to that seen in the

presence of H-MursiF-D211A or H-MursiF-D328A

(data not shown) In this regard, MursiF behaves

dif-ferently from other PP2C-type phosphatases, but its

refractory behavior to artificial substrates is in

agree-ment with that reported for the RsbU homolog in

Synechocystissp IcfG (Slr1860) [33]

Similarly, Bacillus RsbU and RsbX protein

phos-phatases display strict specificities for a single

homolo-gous phosphoprotein, RsbV and RsbS, respectively

[34] Another member, SpoIIE, does not

dephosphory-late its physiological substrate protein, SpoIIAA,

following replacement of the phospho-serine residue at

its phosphorylation site by a phospho-threonine [35]

The lack of Va- and Vb-boxes in the PP2C-type

cata-lytic domain is a feature shared by MursiF UPD with

RsbU, RsbX and SpoIIE phosphatases, which are

known to be divergent PP2C-type phosphatases [24]

(Fig 2) The residues in the Va- and Vb-boxes of the

PP2C phosphatase catalytic domain have not been

characterized for their role in catalysis, but their

selec-tive absence in RsbU-like phosphatases may possibly

be relevant for the stringent specificity of this group of phosphatases

MursiF WKD characterization

To examine whether MursiF retains kinase activity in spite of the loss of critical residues involved in ATP binding, as observed in our in silico analysis of MursiF WKD (Fig 3A), purified H-MursiF as well as the H-WKD and G-WKD proteins were incubated with [32P]ATP[cP] in appropriate buffer conditions No autophosphorylation signal was observed for any of the proteins (data not shown) As expected, H-MursiF

or domain alone (H-WKD and G-WKD) did not phosphorylate its purified RsbV⁄ substrate domain (H-VSD, G-VSD), despite the presence of the conserved phosphorylation site Ser600 in VSD (Fig S1) Also,

no phosphotransfer was seen on the artificial substrates MBP and histone in the presence of H-MursiF or H-WKD, despite attempts to standardize reaction con-ditions and a longer exposure of the autoradiogram Purified M tuberculosis UsfX also does not display autophosphorylation or phosphotransfer to VSD Also, MursiF WKD and UsfX failed to phosphorylate certain other anti-anti-SigF proteins, G-RsfB and G-Rv2638 (Table 1) (data not shown) In the auto-radiogram with both GST-tagged UsfX and WKD preparations, an autophosphorylating kinase signal was seen at a size lower than that of UsfX and WKD (Fig S2) This autophosphorylation signal and a low-intensity signal for phosphotransfer on MBP seen in Fig S2 are not unique to the presence of UsfX or WKD, as they are observed in the absence of these proteins but with GST protein and other recombinant nonkinase proteins, similarly purified from E coli (Fig S2) Most notably, autophosphorylation as well

as phosphotransfer signals diminished after stringent and extensive washing of resin-bound G-UsfX and G-WKD with high salt (1 m NaCl) containing buffers before protein elution (Fig S2) The activity was there-fore attributed to copurifying contaminating kinase(s) from E coli Two independent groups have already reported that M tuberculosis UsfX, unexpectedly, is impaired in its ability to phosphorylate its natural sub-strates, anti-anti-sigma factors such as RsfA [10] and Rv0516c [30]

MursiF WKD and UsfX were tested for their ability

to bind ATP by using a fluorescent ATP analog, 2,4,6-trinitrophenyl ATP (TNP-ATP), which shows an increase in fluorescence emission intensity accompanied

by a blue shift upon protein binding [36] The fluores-cence emission spectrum of TNP-ATP in the presence

of MursiF WKD and UsfX showed a small change in

0

1

2

3

No inhibitor

Sod.orthovanadate

Cyclosporine Calyculin A Sod flouride Okadaic acid

A405

Fig 7 Effect of various phosphatase inhibitors on MursiF

phospha-tase activity Purified H-MursiF was preincubated with inhibitor

(sodium orthovanadate, 200 l M ; sodium fluoride, 50 m M ; okadaic

acid, 100 l M ; cyclosporine, 5 m M; calyculin A, 1 l M ) for 20 min at

25 C in assay buffer prior to addition of pNPP A 405 nm for each of

the reactions at 100 min is plotted Each value is the average of

three individual reactions and is given as mean ± SD.

intensity as well as in wavelength of emission, whereas

the positive control, M tuberculosis PknB, showed a

remarkable two-fold enhancement of the emission

intensity and a blue shift (from 556 to 545 nm)

(Fig 8) M tuberculosis MursiF WKD and UsfX

therefore show very weak ATP-binding ability Given this relatively low affinity of MursiF WKD and UsfX, low intracellular ATP concentrations in normal physi-ological conditions, and even lower ATP levels in the hypoxic and nonreplicating state in mycobacteria [37], their association with ATP in the cellular milieu for kinase function is questionable

To rule out loss of activity of purified MursiF WKD and UsfX during purification or handling, and to check for the presence of other features characteristic of anti-sigma factors, we carried out interaction studies with MursiF VSD and SigF using sandwich ELISA and an

Ni2+-nitrilotriacetic acid resin pull-down assay, respec-tively (Fig 9A,B) MursiF WKD was found to interact with both MursiF VSD (Fig 9A) and SigF (Fig 9B),

in agreement with the results of Parida et al [13] The interaction between MursiF WKD and SigF is specific,

as no interaction could be observed with the principal

M tuberculosis sigma factor, sigma factor A (SigA) (Fig S3) Both MursiF WKD and UsfX show inter-molecular interactions with themselves as well as with each other (Fig 9A) Similarly, UsfX interacts with

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

A

B

A490

Added

Coated

resin

G-UsfX G-WKD GST G-UsfX G-WKD GST

H-SF

Fig 9 Interactions of MursiF domains with each other and with UsfX (A) and SigF (B) (A) In a sandwich ELISA-based assay, the GST-tagged proteins (Table 1) were coated (Coated) on ELISA plates, and incubated with different concentrations of His-GST-tagged interacting pro-teins (Added) The bound propro-teins were probed with antibody against penta-His and developed using ortho-phenylenediamine GST served

as negative control in the reactions Each value is the average of three individual reactions and is given as mean ± SD (B) Ni 2+ –Nitrilotriace-tic acid resin-bound H-SF was incubated with purified GST, G-WKD and G-UsfX in separate reactions, washed extensively, and analyzed by SDS ⁄ PAGE Input lanes represent 40% of actual input protein added to H-SF-bound resin, and output lanes represent resin-bound proteins obtained in the interaction assay.

0

500 000

1 000 000

1 500 000

2 000 000

2 500 000

Emission wavelength (nm)

1 2 3 4

Fig 8 Fluorescence spectra of TNP-ATP in the presence and

absence of MursiF Details of the experiment are given in

Experi-mental procedures Curve 1: spectrum of TNP-ATP (8 l M ) in the

presence of buffer alone Curve 2: TNP-ATP (8 l M ) in the presence

of H-WKD (1 l M ) Curve 3: TNP-ATP (8 l M ) in the presence of

H-UsfX (1 l M ) Curve 4: TNP-ATP (8 l M ) in the presence of PknB

(1 l M ).