Global Protein-Protein Interaction Network in the Human Pathogen

Mycobacterium tuberculosis H37Rv

Yi Wang,† Tao Cui,† Cong Zhang, Min Yang, Yuanxia Huang, Weihui Li, Lei Zhang,
Chunhui Gao, Yang He, Yuqing Li, Feng Huang, Jumei Zeng, Cheng Huang, Qiong Yang,
Yuxi Tian, Chunchao Zhao, Huanchun Chen, Hua Zhang, and Zheng-Guo He*

National Key Laboratory of Agricultural Microbiology, Center for Proteomics Research, College of Life Science
and Technology, Huazhong Agricultural University, Wuhan 430070, China

Received August 06, 2010

Analysis of the protein-protein interaction network of a pathogen is a powerful approach for dissecting
gene function, potential signal transduction, and virulence pathways. This study looks at the construction
of a global protein-protein interaction (PPI) network for the human pathogen Mycobacterium
tuberculosis H37Rv, based on a high-throughput bacterial two-hybrid method. Almost the entire
ORFeome was cloned, and more than 8000 novel interactions were identified. The overall quality of
the PPI network was validated through two independent methods, and a high success rate of more
than 60% was obtained. The parameters of PPI networks were calculated. The average shortest path
length was 4.31. The topological coefficient of the M. tuberculosis B2H network perfectly followed a
power law distribution (correlation ) 0.999; R-squared ) 0.999) and represented the best fit in all
currently available PPI networks. A cross-species PPI network comparison revealed 94 conserved
subnetworks between M. tuberculosis and several prokaryotic organism PPI networks. The global
network was linked to the protein secretion pathway. Two WhiB-like regulators were found to be highly
connected proteins in the global network. This is the first systematic noncomputational PPI data for
the human pathogen, and it provides a useful resource for studies of infection mechanisms, new
signaling pathways, and novel antituberculosis drug development.

Keywords: Mycobacterium tuberculosis • PPI network • virulence • secretion • WhiB3

Introduction pylori,4 Saccharomyces cerevisiae,5-7 Caenorhabditis elegans,8

Mycobacterium tuberculosis, one of the most devastating
human pathogens, is responsible annually for about 2 million
deaths worldwide. One-third of the world’s population is
currently infected with the tubercle bacillus.1 The seriousness
of tuberculosis infection has become even more significant with
the emergence of multidrug-resistant strains of M. tuberculosis,
promoting a thorough investigation of this unique pathogen.2
Understanding the virulence and infection mechanisms of M.
tuberculosis will require systematic identification of signal
transduction systems utilized by this pathogen. One approach
to the study of biological regulation at this level is through the
construction of a global protein-protein interaction network.
Protein-protein interactions (PPIs) are important in many
biological processes, such as in the protein secretion or signal
transduction. However, interaction data based on high-
throughput experimental techniques are currently available for
only a very few model organisms.3 Global PPI networks have
been investigated with yeast two-hybrid (Y2H) systems in a
number of models, including organisms such as Helicobacter
* To whom correspondence should be addressed. E-mail:
he.zhengguo@hotmail.com or hezhengguo@mail.hzau.edu.cn. Tel.: +86-27-
87284300. Fax: +86-27-87280670.
†
These two authors equally contributed to this work.
Drosophila melanogaster,9 humans,10 Synechocystis sp.,11 Campy-
lobacter jejuni,12 and, recently, Treponema pallidum.13 In
addition, using a Y2H method, Rajagopala et al.14 studied the
motility networks of T. pallidum and C. jejuni. The generation
of large Y2H interaction maps has led to comprehensive
understanding of the biological processes and functions of
many uncharacterized genes. A similar study of protein-protein
interactions in genetically intractable pathogens such as M.
tuberculosis would be expected to improve the dissection of
virulence pathways and to significantly advance our under-
standing of tuberculosis infection.
Computer analysis has been used to model the construction
of global PPI networks in M. tuberculosis.15,16 However, the M.
tuberculosis genome encodes about 4000 ORFs, with one-third
of these ORFs of unknown function. Therefore, previous studies
on PPI networks in M. tuberculosis, based on bioinformatics
and homology protein mapping, have actually covered only a
very small percentage of the genome.17 These studies provide
rather limited information, and one-third of the ORFs encoded
by the bacterial genome remain uncharacterized. The definition
of protein-protein interactions is now recognized as one of
the most efficient approaches for mining these types of
unknown function.

10.1021/pr100808n  2010 American Chemical Society Journal of Proteome Research 2010, 9, 6665–6677 6665
Published on Web 10/25/2010
research articles Wang et al.

Figure 1. Construction and property of the M. tuberculosis PPI network. (A) The process of constructing large-scale M. tuberculosis
protein-protein interaction networks. (B) Schematic representation of bacterial two-hybrid analysis in this study. (C) Global
protein-protein interaction network views of M. tuberculosis. A graph of the pathogenic PPI network involving 2907 proteins linked
via 8042 interactions. Several major protein families are indicated by different colors. (D) The degrees of top 10 highly connected
proteins in the M. tuberculosis B2H PPI network. The numbers of PPIs are indicated on top of the respective columns. The functions
of 10 proteins are shown on the right of the panel.

For the study of PPIs, the yeast two-hybrid system has come
to the forefront as a powerful tool for PPI identification.
However, the development of bacterial two-hybrid systems now
allows proteins to be assayed for interactions under conditions
that closely match the native environment of M. tuberculosis
(Figure 1B).18-20 In the present paper, we show how we
successfully constructed a global M. tuberculosis PPI network.
We cloned almost the entire ORFeome of the pathogen and
identified more than 8000 mostly novel interactions among the
2907 proteins. The overall quality of the PPI network was
validated. Two WhiB-like transcriptional factors were also found
to be highly connected proteins in the global network, indicat-
ing that these genes might be core regulators. The capacity for
comprehensive screening of PPI in M. tuberculosis is expected
to provide further information regarding infection mechanisms
and to aid in identifying novel antitubercular drugs.
Materials and Methods
Cloning of M. tuberculosis ORFeome into the Bacterial
Two-Hybrid Plasmids. Primer pairs for a total of 3989 ORFs
were designed, and each predicted ORF was amplified with a
specific primer pair through PCR. PCR products were digested
by corresponding restriction enzyme pairs or were ligated with
an EcoRI-SmaI adaptor (TakaRa Biotech. Co.) if this type of
enzyme site was lacking within the ORF sequences. The
cleaned ORF fragments were then cloned into a pair of
BacterioMatch@II vectors of pBT and pTRG vector (Stratagene).
We carried out sequencing of the entire clone library to confirm
that the ORFs were correctly amplified and cloned. All pTRG-
ORF plasmids were then pooled, and from there, an ORF library
for M. tuberculosis was constructed.

6666 Journal of Proteome Research • Vol. 9, No. 12, 2010

Protein-Protein Interaction Network in M. tuberculosis H37Rv
@
Bacterial Two-Hybrid Analysis. The BacterioMatch II Two-
Hybrid System (Stratagene) was used to establish protein-protein
interactions between M. tuberculosis proteins. Each pBT re-
combinant MtbORF was used as a bait plasmid to screen the
pTRG-MtbORF library. The selective medium contained 8 mM
3-AT, 8 mg/mL of streptomycin, 15 mg/mL of tetracycline, 34
mg/mL of chloramphenicol, and 50 mg/mL of kanamycin. The
plates were incubated at 30 °C for 2-4 days. All resulting
positive clones were collected and stored for further use. The
ORF in the recombinant plasmid pTRG-MtbORF was se-
quenced. The unredundant positive cotransformants were
spotted onto selective medium to confirm the interactions. The
plates were incubated at 30 °C for 2-3 days. The cotransfor-
mant containing pBT-LGF2 and pTRG-Gal11P (Stratagene) was
used as a positive control for expected growth on the Selective
Screening Medium. A cotransformant containing the empty
vectors pBT and pTRG was also used as a negative control.
SPR Assays. Surface Plasmon Resonance (SPR) analysis on
a Biacore 3000 instrument (GE Healthcare) with NTA sensor
chips was performed according to our previously published
procedures.21 His-tagged M. tuberculosis protein was im-
mobilized onto the nitrilotriacetic acid (NTA) chips. The other
purified GST-tagged protein, to be used as the ligand, was
diluted in the HBS buffer [10 mM Hepes (pH 7.4), 150 mM
NaCl, 50 µM EDTA, 5 mM ATP, 0.005% BIAcore surfactant P20]
at a concentration of <200 nM and injected at 10 µL/min for 5
min at 25 °C. GST protein was substituted for the GST-tagged
protein for a negative control. Each analysis was performed in
triplicate. An overlay plot was produced using BIA evaluation
3.1 software to depict the interaction between a pair of proteins.
Coexpression/Copurification and Western Blotting
Assays of Protein-Protein Interactions. The pair of potential
interaction proteins was cloned into the pHEX-derived and
pGEX-derived vectors as described in an earlier report.22 A pair
of recombinant vectors was cotransformed into competent E.
coli BL21 (DE3). Co-transformants were grown at 37 °C, and
protein expression was induced through the addition of 0.5 mM
IPTG (isopropyl-β-D-thiogalactopyranoside). The cells were
harvested and sonicated at 4 °C. The supernatants were loaded
onto a 15 mL GST column (pre-equilibrated with PBS). The
column was centrifuged at 500g for 2 min, and the supernatant
was discarded. Each column was washed with PBS and eluted
with Buffer B (16 mM Na2HPO4, 4 mM KH2PO4, 20 mM GSH).
Fractions were pooled and stored at -80 °C for further assays.
The above copurified sample was run on SDS-PAGE for
Western blot analysis. The protein bands were transferred to a
nitrocellulose membrane. Western blot analysis was conducted
using primary antibody (antisera to His-tag) (1:1000) and
secondary antibody IgG-HRP (goat anti-Rabbit) (1:10000). The
signal was developed using DAB detection reagents and
recorded through photography to quantify the protein.
Protein Interactions, Sequence, and Network Visualization.
Protein-protein interaction data sets generated from high-
throughput, two-hybrid, and pull-down assays were obtained
from the following sources: E. coli,23-25 H. pylori,4 C. jejuni,12
Synechocystis sp.,11 T. pallidum,13 and M. tuberculosis (this
study). Summary information of these data sets was shown in
Supporting Information Table S1. Computational predicted
protein interactions and functional linkages of M. tuberculosis
were obtained from Prolinks database,26 STRING database
(version 8.2),27 and our previous study.16 All interaction data
sets were preprocessed to undirected protein interaction
networks for further analysis. The corresponding protein
research articles
sequences of E. coli were obtained from GenoBase and others
from NCBI. Protein interaction networks were visualized using
Cytoscape.28 Classes of proteins were colored differently as
indicated.
Network Topology Analysis. The topological parameters of
M. tuberculosis B2H and other reference PPI networks were
analyzed using NetworkAnalysis29 plugin of Cytoscape.28 The
edges in all PPI networks were treated as undirected. The
definition of network topological measure can be found in
NetworkAnalyzer Online Help (http://med.bioinf.mpi-inf.mpg.
de/netanalyzer/help/2.6.1/index.html).
Cross-Species Networks Comparison Analysis. We per-
formed a cross-species protein interaction network comparison
analysis of M. tuberculosis B2H PPI networks with those PPI
data set currently available prokaryotic organisms such as E.
coli,24,25 H. pylori,4 C. jejuni,12 Synechocystis sp.,11 and T.
pallidum13 to identify significant conserved subnetworks using
the NetworkBlast algorithm.30 The Java version of the Net-
workBlast program (version 1.0) was downloaded from the Web
site of the Ideker lab (http://chianti.ucsd.edu/nct/). The stan-
dalone blast software package was downloaded from the NCBI
FTP Web site. Protein homology relationships between M.
tuberculosis and other reference species were then constructed
from all-against-all blast searches using the blastp program with
E-value set to 1e-10. Finally, a conserved subnetwork search
between M. tuberculosis and other species was performed using
NetworkBlast with default parameters, except that a duplicate
threshold was set to 0.5. Therefore, when multiple subnetworks
contained protein homologues that overlapped g50%, only the
subnetworks with the highest density were reserved in the final
result. The images of conserved subnetworks were generated
automatically using the Dual Layout plugin of the Cytoscape
software.
Results
M. tuberculosis H37Rv ORFeome Cloning and
Construction of a Library. According to the NCBI genome,
there are 3989 protein coding genes in the M. tuberculosis
H37Rv genome. Consequently, 3989 pairs of primers were
designed, and the corresponding ORFs were amplified by PCR
to generate recombinant plasmids and an ORF library of M.
tuberculosis H37Rv for screening the PPI. Figure 1A shows that
the produced ORF-DNA fragments were digested by restriction
enzymes and cloned into a pair of BacterioMatch@II vectors of
pBT and pTRG (Stratagene). We then were able to successfully
clone 3896 ORFs corresponding to 99.1% of the predicted
ORFeome without frameshifts or nonsense mutations; this was
confirmed by resequencing. All 3896 pTRG-ORF recombinant
plasmids were then pooled to construct the ORF library for M.
tuberculosis.
Construction and Properties of the M. tuberculosis PPI
Map. All cloned ORFs were subjected to extensive screening
against the M. tuberculosis ORF library, using the bacterial two-
hybrid technique (Stratagene).16 In total, 11 000 pairs of non-
redundant PPIs were isolated and characterized through se-
quencing of both the recombinant pBT and pTRG plasmids,
after the first screening (Figure 1A). All 11 000 cotransformant
strains, together with the cotransformants containing either
pTRG-ORF/pBT or pTRG/pBT-ORF (self-activation controls),
were then spotted onto selective plates for the second screen-
ing. Some representative plates from the first and second
screening were shown (Supporting Information Figure S1).
Ultimately, 8042 PPIs were characterized (which covered 2907
Journal of Proteome Research • Vol. 9, No. 12, 2010 6667
research articles Wang et al.

Figure 2. Interaction distributions in COG functional categories. (A) Interactions between COG functional categories. Numbers and
colors indicate the number of interactions between each pair of COG functional categories in the M. tuberculosis B2H PPI network. (B)
Number of proteins in each COG functional category encoded by the M. tuberculosis genome. (C) Number of interactions in each COG
functional category in the M. tuberculosis B2H PPI network.

genes encoded by the genome of M. tuberculosis), after
eliminating false-positive cotransformants (Figure 1A). A global
network of M. tuberculosis H37Rv, containing 8042 PPIs, was
constructed (Figure 1C), and the network (or graph) based on
these PPI data was visualized by the program Cytoscape29
applying the Dual Layout plugin (Supporting Information Table
S2).
The bacterial proteins formed a highly linked network. As
shown in Figure 1C, a giant network component of 8020
interactions between 2895 (99.59%) proteins and six small
isolated network components of less than three proteins were
included in the constructed global PPI network of M. tubercu-
losis H37Rv. The top 10 proteins (right panel) and their number
of linkages (left panel) in the B2H network are listed in Figure
1D. These include dihydrodipicolinate reductase DapB, two
WhiB-like transcriptional factors (WhiB3 and WhiB7), and
several proteins of unknown function. On the basis of the COG
categories, the B2H data are presented using a heat map, and
interactions between lipid transport and lipid metabolism
proteins are shown to be relatively enriched in the global
network (Figure 2A). The number and map of B2H interactions
show a consistency with that of ORFs encoded by the M.
tuberculosis genome (Figure 2B and 2C).
We further finished the PPI enrichment analysis in the
context of these COG categories (Supporting Information Figure
S2). Proteins from category I (lipid transport and metabolism),
D (cell cycle control, mitosis, and meiosis), and O (posttrans-
lational modification, protein turnover, chaperones) presented
higher connection to the rest of the proteins in the network
(Supporting Information Figure S2A). Proteins within the
category D (cell cycle control, mitosis, and meiosis) were
remarkably highly interconnected (Supporting Information
Figure S2B). Significant enrichment of the interactions between
categories M (cell wall/membrane biogenesis) and U (intra-

6668 Journal of Proteome Research • Vol. 9, No. 12, 2010

Protein-Protein Interaction Network in M. tuberculosis H37Rv research articles

Figure 3. Topological properties of the M. tuberculosis B2H PPI network. (A) Degree distribution. The number of proteins with a given
link (k) in the M. tuberculosis B2H PPI network follows a power law P(k) ) akb (a ) 3039.8, b ) -1.968). R2 ) 0.884 for the power law
fit. (B) Topological coefficients distribution. The topological coefficient was plotted against the number of links. The topological coefficient
with a given link (k) in the M. tuberculosis B2H PPI network follows a power law TC(k) ) akb (a ) 0.976, b ) -0.972, R2 ) 0.999). (C)
Average clustering coefficient distribution. R2 ) 0.234 for power law fit. (D) Closeness centrality distribution. R2 ) 0.383 for power law
fit. (E) Shortest path length distribution. On average, any two proteins in the M. tuberculosis B2H PPI networks are connected though
4.31 links. (F) Stress distribution. The stress of a node n is the number of shortest paths passing through n. The stress distribution
gives the number of nodes with stress s for different values of s. The values for the stress are grouped into bins whose size grows
exponentially by a factor of 10. The bins used for this distribution are {0}; [1, 10); [10, 100); ...

cellular trafficking and secretion) was observed (Supporting
Information Figure S2B). Signal proteins from category T (signal
transduction mechanisms) tended to interact with proteins
from category I (lipid transport and metabolism), C (energy
production and conversion), and J (translation) (Supporting
Information Figure S2B).
Topology of the M. tuberculosis PPI Network. The topologi-
cal parameters of PPI networks were calculated using Network
Analysis29(Figure 3). The average shortest path length was 4.31,
network diameter 10, average number of neighbors 5.509, and
clustering coefficient 0.006. A complete network parameter of
M. tuberculosis B2H PPI and several prokaryotic reference
networks can be found in Supporting Information Table S1 and
Figure S3. The first three topological features were identical to
our current understanding of PPI networks in prokaryotic
organisms4,12,13,23-25 (Supporting Information Table S1). How-
ever, the M. tuberculosis network only had a consistent cluster-
ing coefficient with that of Synechocystis sp.11 and was lower
than the other reference networks (Supporting Information
Table S1). The clustering coefficient of different data sets
generated from the pull-down assay tended to be closer
(0.064-0.079), while the two-hybrid-based screens showed
apparent variance (0.006-0.233).

Journal of Proteome Research • Vol. 9, No. 12, 2010 6669

research articles
Degree distribution follows a power law [P(k) ∼ ak ] (a ) b
3046.1; b ) -1.968; correlation )0.738; R-squared ) 0.884)
(Figure 3A). About 89.1% (2589/2905) of the M. tuberculosis
genes have 1-10 interaction partners (Supporting Information
Figure S3E). The number of proteins declines very quickly until
about 15 degrees, indicating that there are very few proteins
containing more than 15 interaction partners in the network.
In particular, the topological coefficient of proteins in M.
tuberculosis B2H networks perfectly follows a power law
distribution [TC(k) ∼ akb] (a ) 0.976; b ) -0.971; correlation
) 0.999; R-squared ) 0.999) and represents the best fit in
all current available PPI networks from prokaryotic to
eukaryotic organisms (Figure 3B, also see Supporting Infor-
mation Table S3).
Closeness centrality is a measure of how rapidly information
spreads from a given node to other reachable nodes in the
network. The closeness centrality distribution of the M. tuber-
culosis B2H PPI network shows that highly connected proteins
in the network have a pronounced ability to spread information
in the network (Figure 3C). The same tendency also exists in
reference networks (Supporting Information Figure S3). Nodes
with high closeness centrality have potential significance for
responding to external perturbations and for maintaining
network stabilization. This may be part of the explanation why
the highly connected “hub” proteins usually play essential roles
in cell processes.31
Although lower than other networks, the clustering coef-
ficient distribution (Figure 3D) of the M. tuberculosis B2H PPI
network shares a tendency similar to that seen in other
networks, in that it diminishes when the links increase (Sup-
porting Information Figure S3). The shortest path distribution
(Figure 3E) and stress distribution (Figure 3F) of the M.
tuberculosis B2H PPI network also share patterns similar to
those of other reference networks (Supporting Information
Figure S3).
Experimental Verification of Interaction. Two groups of
genes were randomly chosen for validations that would further
evaluate the quality of our PPI data. A total of 241 pairs of
interactions were examined, using a coexpressing system
reported recently,22 and a significant success rate of 147/241
(61%) was obtained (Figure 4A). Another 35 pairs of interactions
were tested using SPR assay, and the verified PPIs had a success
rate of 20/35 (57%). This was a somewhat lower rate than the
coexpression/copurification assay (Figure 4B). The difference
most likely arose because a relatively large fraction of active
proteins could not be successfully purified or some weak
interactions could not be successfully detected with the SPR
technique. A great deal of our bacterial two-hybrid data could
be confirmed (success rate 60.5%) through various independent
methods, which produced data similar to that from previous
yeast two-hybrid studies on other model organisms.10
Protein Function Predictions and Pathway Mapping. In-
teracting proteins often function in the same pathway or
protein complex. The M. tuberculosis interaction map could
therefore be used to predict the functions of hypothetical
proteins or to map protein complexes and new pathways. The
M. tuberculosis genome encodes about 1361 hypothetical
proteins.17 The 1104 hypothetical proteins are included in the
B2H network, which were involved in 5176 PPIs. This is 64.36%
of the global PPI number (5176/8042), indicating that these
uncharacterized proteins may participate in numerous cell
functions in M. tuberculosis. For example, the PPI network was
also linked to function of PE/PPE proteins, a unique protein
6670 Journal of Proteome Research • Vol. 9, No. 12, 2010
Wang et al.
family in mycobacteria, named for the conserved proline (P)
and glutamate (E) residues near the N-terminal region of the
proteins.17 The function of most PE/PPE protein is unknown.
The B2H network included 106 PE/PPE family proteins and 671
PPIs (Supporting Information Figure S4A). There were 102 PPIs
involved with secreted proteins (or antigens), 44 PPIs associated
with membrane proteins, and 40 with lipid transport and
metabolism proteins (Supporting Information Figure S4B). This
implies that PE/PPE family proteins might play roles in protein
secretion, transmembrane transport, and cell envelope forma-
tion pathways.
Another example of the function of hypothetical proteins and
new pathways concerns lipid metabolism. As shown in Sup-
porting Information Figure S5, two hypothetical proteins,
Rv1638A and Rv2102, were grouped into a subnetwork enriched
in lipid metabolism-related genes. Rv1638A has previously
shown to be markedly repressed together with a group of lipid
metabolism genes (cmaA2, mmaA3, mmaA2, fadE10, echA8,
fadD21, inhA, tesB1, fas, fadD29) in sputum, compared to
aerobic growth.32 Compared with the in vitro growth condi-
tions, Rv2102 was consistently down-regulated, together with
several lipid metabolism genes (fadA2, fadD15, kasA, kasB),33
following M. tuberculosis infection of human macrophage-like
THP-1 cells. Of these, four genes, inhA, fas, kasA, and kasB,
appeared in the subnetwork (Supporting Information Figure
S5). In addition, both Rv1638A and Rv2102 interacted with the
same enoyl-CoA hydratase gene of echA6 in the PPI network.
These suggest that Rv1638A and Rv2102 could function in lipid
metabolism.
Rv3312A was previously shown to encode M. tuberculosis
pili.34 In some Gram-positive pathogens, such as Streptococcus
and C. diphtheriae, pilus assembly has been associated with
sortase genes, whose products are involved in covalent linkage
of the pili to the cell wall peptidoglycan.35-37 Unexpectedly,
M. tuberculosis lacks a sortase gene.38 In our PPI network,
Rv3312A was shown to directly interact with several potential
transmembrane or secreted proteins, such as Rv0508, Rv0854,
Rv1583c (annotated as a hypothetical phiRv1 phage protein),
and lipoprotein LpqX (a cell envelope protein)39 (Supporting
Information Figure S5). These proteins were also shown to
interact with the other functionally related secreted antigen or
transmembrane proteins such as MPT53,40,41 Rv0219, Rv2869c,
and the lipoprotein LpqQ (Supporting Information Figure S5).
This suggests that these proteins may form an integrated
pathway for the assembly or localization of pili by situating on
the bacterial surface or its plasma membrane. Therefore, the
B2H data suggest that M. tuberculosis may use a new sortase-
independent pathway, as proposed here, for the assembly of
its pili.
Conserved Subnetworks Across Species. A cross-species PPI
network comparison based on the NetworkBlast algorithm
method30 revealed 94 conserved subnetworks between M.
tuberculosis and several prokaryotic organism PPI networks. A
complete list of the identified subnetworks is shown in the
Supporting Information. These conserved subnetworks con-
sisted of 65 conserved interactions between M. tuberculosis and
E. coli (44 generated from comparison with the data sets of
Arifuzzaman et al.24 and 21 with Hu et al.,25 15 between M.
tuberculosis and C. jejuni, 12 between M. tuberculosis and
Synechocystis sp., and 2 between M. tuberculosis and T. palli-
dum). No significantly conserved subnetworks were identified
between M. tuberculosis and H. pylori. This was not surprising
because in Parrish’s previous research they also failed to
Protein-Protein Interaction Network in M. tuberculosis H37Rv research articles

Figure 4. Coexpression/copurification and SPR assays for verifying bacterial two-hybrid interactions. (A) A scheme of coexpression/
copurification method for PPI verification. (B) A scheme of SPR method for PPI verification. (C) Coexpression/copurification assay. The
pair of potential interaction genes was cloned into the pHEX-derived and pGEX-derived vectors. The E. coli BL21 (DE3) cotransformants
with a pair of recombinant vectors were grown at 37 °C, and protein expression was induced. The cells were sonicated, and the
supernatants were loaded onto a 15 mL glutathione (GST) column for copurification. The copurified sample was run on SDS-PAGE for
Western blot analysis. A Western blot analysis was conducted using primary antibody (antisera to His-tag) (1:1000) and secondary
antibody IgG-HRP (goat anti-Rabbit) (1:10000). Representative data are shown. (D) SPR assay. His-tagged M. tuberculosis protein was
immobilized onto the NTA chips. Another GST-tagged protein, to be used as the ligand, was diluted in the HBS buffer [10 mM Hepes
(pH 7.4), 150 mM NaCl, 50 µM EDTA, 5 mM ATP, 0.005% BIAcore surfactant P20] at a concentration of <200 nM and injected at 10
µL/min for 5 min at 25 °C. GST protein was used as a negative control. Each analysis was performed three times. An overlay plot was
produced, and representative curves are shown.

identify conserved subnetworks between C. jejuni and H. pylori; groups of regulatory proteins that were obviously enriched,
this may be caused by the low interactome coverage of H. pylori such as sigma factors (SigB, SigE, SigH, and SigF), two-
PPI networks.4 Notably, among the 12 conserved subnetworks component signal proteins (MtrA, MtrB, DevR, PhoP, and
between M. tuberculosis and Synechocystis sp. were several KdpD), serine/threonine protein kinases (PknD, PknJ, and

Journal of Proteome Research • Vol. 9, No. 12, 2010 6671

research articles Wang et al.

Figure 5. Representative examples of conserved subnetworks between M. tuberculosis and other reference species. Conserved
subnetworks identified by NetworkBlast. Red nodes represent the proteins from M. tuberculosis, and green nodes represent the proteins
from corresponding reference species. Blue dashed lines indicate cross-species sequence similarity between proteins. (A) Conserved
subnetworks between M. tuberculosis and C. jejuni. (B) Conserved subnetworks between M. tuberculosis and E. coli. (C) Conserved
subnetworks between M. tuberculosis and Synechocystis sp. (D) Conserved subnetworks between M. tuberculosis and T. pallidum. A
complete list of conserved subnetworks identified by NetworkBlast in this study is available in the Supporting Information.

PknK), and cyclic-di-GMP small molecular signal proteins
(Rv1354c and Rv1357c) (Figure 5 and Supporting Information
Figure S4). All of these represent the dominant regulation
mechanisms in prokaryotic organisms. Our results suggest that
these proteins, through their involvement in the conserved
interactions, may function as protein complexes or in the same
pathway.
Linking PPI Networks to Protein Secretion Pathways.
ESAT-6 and CFP-10 were two major secreted antigens of M.
tuberculosis and were also involved in protein secretion.42
ESAT6 and CFP10-like proteins and 408 PPIs were included in
our global network among 22 ESAT6 and CFP10-like gene
clusters in the genome of M. tuberculosis17 (Supporting Infor-
mation Table S4). Of these, 22 proteins formed a subnetwork,
most of which interacted with multiple other proteins, indicat-
ing that different pairs of ESAT6 and CFP10-like proteins may
cross-regulate during the virulence secretion of the pathogen
(Figure 6A). A linkage between two major M. tuberculosis ESX
protein secretion systems, ESX-1 and ESX-5, was also observed
(Figure 6B). Several family proteins in the subnetwork were
found to interact with these two ESX system proteins, including
eight ESAT6 and CFP10-like proteins, three secretion proteins,
one PE/PPE protein, two transmembrane proteins, two lipid
transport and metabolism proteins, one regulatory protein, and
eight hypothetical proteins (Figure 6B).
In addition, 37 reported secretion proteins were included in
a PPI subnetwork, and these directly interacted with ESX
systems or ESAT6 and CFP10-like proteins (Figure 6C). There
were also 20 transmembrane proteins, 6 PE/PPE proteins, 3
lipid transport and metabolism proteins, 4 regulatory proteins,
and 9 hypothetical proteins included (Figure 6C). This indicates
that these related proteins might be involved in protein
secretion through coupling with ESX system proteins, ESAT6,
and CFP10-like proteins in M. tuberculosis. Multiple M. tuber-
culosis ESX systems may coordinate through protein-protein
interactions and cooperatively serve in bacterial protein secre-
tion. This function would be consistent with the previously
observed crosstalk between two different ESX systems in a
mycobacterial species, M. marinum.43
PPI Network Analysis Revealed Two WhiB-Like Core
Regulators. Two WhiB-like regulators, WhiB3 and WhiB7, were
found among those in the top 10 with high presence in the
global PPI network (Figure 1D). In total, 58.2% (96/165) of these
interactions could be validated by the coexpression/copurifi-
cation method (Supporting Information Table S5), which is very
close to the overall success rate of the PPI network (60%).
WhiB3 has been recently characterized to have extensive
involvement in the regulation of lipid metabolism and in
response to environmental stresses.44 In the present study,
WhiB-3 interacted with 113 proteins including 8 lipid transport
and metabolism proteins, 3 oxidoreductase proteins, 4 regula-
tory proteins, and a number of transport-related proteins, such
as 4 membrane transport proteins, 3 PE/PPE proteins, and 15
secretion proteins (Figure 7). In all, 29 hypothetical proteins
and other function proteins were found to link with WhiB3
(Figure 7).
In contrast, WhiB7 is linked to drug resistance and persistent
infection.45 WhiB7 was involved in 52 PPIs that included 2 drug
resistance proteins (TrpG and Rv1260), 4 oxidoreductase
proteins, 3 persistent infection proteins, 4 regulatory proteins,
and 4 protein kinases (Figure 7). Interestingly, both WhiB
proteins in the subnetwork interacted with the same hypotheti-
cal protein, Rv2418c. Although each WhiB-like protein may play
different roles during infection of M. tuberculosis, our data

6672 Journal of Proteome Research • Vol. 9, No. 12, 2010

Protein-Protein Interaction Network in M. tuberculosis H37Rv research articles

Figure 6. PPI network and protein secretion pathway. The local network was constructed based on the data from our bacterial two-
hybrid experiments. Several major protein families are indicated in different colors. (A) The PPI network of 11 pairs of ESAT6/CFP10-
like proteins. (B) The PPI network involved in ESX1 and ESX5 system proteins. (C) A local network of some reported secretion proteins
in direct linkage with ESX systems or ESAT6 and CFP10-like proteins.

Journal of Proteome Research • Vol. 9, No. 12, 2010 6673

research articles
Figure 7. Local network involved in WhiB3 and WhiB7. The local network was constructed based on the data from our bacterial two-
hybrid experiments. Several major protein families are indicated in different colors.
suggest that these two proteins, WhiB3 and WhiB7, could
crosstalk and cooperatively regulate the persistent infection of
the pathogen through a master protein, Rv2418c.
Discussion
This study introduces the systematic bacterial two-hybrid
analysis of the human pathogen proteins. On the basis of
ORFeome cloning and extensive library screening, we were able
to identify 8042 protein-protein interactions connecting the
2907 bacterial proteins, which represented about 74.1% func-
tional proteins encoded by the entire genome. The M. tuber-
culosis B2H PPI network was separated into seven connected
components, with the largest component accounting for 2895
(99.59%) proteins from the whole PPI network.
The yeast two-hybrid system has been used as a powerful
tool for PPI identification.4-13 However, yeast has certain
limitations, especially with respect to the assay of bacterial
protein interactions. The development of bacterial two-hybrid
systems now allows proteins to be assayed for interactions
underconditionsthatcloselymatchtheirnativeenvironment.46-48
In the current study, we successfully constructed a global PPI
network of the pathogenic bacterium, M. tuberculosis H37Rv,
using the bacterial two-hybrid technique. We identified more
than 8000 mostly novel interactions in the pathogen.
In a PPI network, the term degree represents the number of
proteins that interact with another specific protein. The degree
distribution of the M. tuberculosis network proteins slowly
decreased, leading to the generation of a pattern similar to that
found in the other model organisms.49 The calculated average
shortest path length (4.31) within the largest network (Figure
3C) was close to the value presented for other prokaryotic and
6674 Journal of Proteome Research • Vol. 9, No. 12, 2010
Wang et al.
even human PPI networks.4,10,12,13,23-25 This indicates that the
network property might be a general characteristic. Most
interestingly, the topological coefficient of the M. tuberculosis
B2H network perfectly followed a power law distribution
(correlation ) 0.999; R-squared ) 0.999) and represented the
best fit in all currently available PPI networks (Supporting
Information Table S3).
Very few experimentally validated protein-protein interac-
tions have been reported in M. tuberculosis; nevertheless, some
of the only interactions to have been reported were found in
the current global PPI network.15,50-54 However, among 8042
pairs of PPIs, we observed only about 90 conserved interactions
that were also found in other model organisms (Figure 5 and
Supporting Information Figure S6). This is relatively low but is
similar to what has been reported in other previous studies on
PPIs.10,11,25 It also demonstrates that the newly produced PPIs
in this study were mostly novel interactions.
The ORFeome of an organism is useful for various reverse
proteomics studies, including global analysis of protein-protein
interaction and the identification of functional genes. In the
present study, a high coverage rate and a relatively small library
have been achieved by cloning the ORFeome from predicted
genes in the pathogenic genome. This offered a tremendous
advantage for extensive characterization of protein-protein
interactions within the genome. This is the first-ever report on
the ORFeome cloning in M. tuberculosis in a bacterial two-
hybrid vector. A small number of ORFs (<0.9%) could not be
successfully amplified (Supporting Information Table S6). This
is probably due to sequencing errors in the genome project or
simply because the best PCR conditions have not yet been
found.
Protein-Protein Interaction Network in M. tuberculosis H37Rv
Specific protein-protein interactions are crucial for all
biological processes. Compiling protein-protein interaction
networks provides many new insights into protein function and
gives directions for the development of new drugs targeted to
this pathogen. The genome sequencing project for M. tuber-
culosis H37Rv offers much valuable information. One striking
finding is that the pathogenic genome encodes a unique
protein familysPE/PPEsalthough its function remains largely
unknown. Our PPI network covered 106 PE/PPE proteins and
included 671 PPIs (Supporting Information Figure S6). Their
functions have been linked to protein secretion and membrane
transport and even gene regulation and lipid metabolism. This
suggests that the mycobacterial PE/PPE proteins might be
involved in pathogen growth and virulence. Our PPI network
also allows the annotation of a large group of hypothetical
proteins in M. tuberculosis that may be involved in 5176 PPIs
or 64.5% of the global PPI number. This will serve as a unique
resource for further analysis of protein functions.
Evidence shows that M. tuberculosis might use unique
secretion machinery and may encode nonclassical systems to
secrete its virulence determinants during infection.42,55-57
Protein-protein interactions have been reported to be involved
in secretion of a number of cases, including secretion of two
major proteins, ESAT-6 and CFP-10.42 At present, 22 ESAT6 and
CFP10-like gene clusters have been found in the genome of
M. tuberculosis,17 but their contributions to virulence secretion
in the pathogen have not yet been identified. In the current
study, cross-interactions between different pairs of EAST6/
CFP10-like proteins were found based on the local PPI network
(Figure 6A), while a link between two major M. tuberculosis
ESX protein secretion systems (ESX-1 and ESX-5) was observed
(Figure 6B). Although the cooperative mechanism was unclear,
this correlation between two ESX systems has been previously
observed in other species. For example, a deletion of the ESAT-6
and CFP-10 encoding genes in M. marinum resulted in
increased secretion of the ESX-5 substrate PPE41, whereas an
ESX-5 mutant showed increased secretion of the ESX-1 sub-
strate Rv3881c.43 Several protein families in the current study
were found to link the two ESX system proteins. This implied
the existence of potential molecular mechanisms for com-
munication between these proteins. On the other hand, a
significant level of reported secretion proteins were included
in the network and physically interacted with ESX systems or
ESAT6 and CFP10-like genes (Figure 6C). Our findings dem-
onstrate that ESAT-6/CFP-10-like proteins may not only rep-
resent typical antigens but also may function as shuttle
chaperones for the secretion of other virulence factors. In this
context, multiple ESAT-6/CFP-10-like proteins, through pro-
tein-protein interactions, may function as an integrated system
for virulence factor secretion. However, the single pair of ESAT-
6/CFP-10 may not have this function in the nonpathogenic
species because ESX-1 is present in the genomes of most
mycobacterial species that have been investigated to date,
including fast-growing environmental mycobacteria, such as
M. smegmatis and M. flavescens.58
In the current PPI network, dihydrodipicolinate reductase
(DapB) was shown to interact with 153 proteins, which include
10 PE/PPE family proteins, 7 secreted proteins, and 18 mem-
brane proteins. This suggested that DapB may participate in
protein secretion and membrane transport, in addition to
acting as a tetrahydrodipicolinic acid synthesis enzyme. An
unexpected finding was that two WhiB-like regulators, WhiB3
and WhiB7, were found among those in the “top 10 high
research articles
degrees” in the global PPI network (Figure 1D). The most highly
connected proteins are usually the most important16 and are
considered to participate extensively in cellular processes. In
several recent reports, M. tuberculosis WhiB-3 protein was
found to directly regulate lipid metabolism and to respond to
environmental stresses.44 WhiB3 has been proposed as a
metabolic switch that senses environmental signals during
infection.15 Therefore, the 9-fold up-regulation of WhiB3 in a
BCG vaccine strain is particularly notable.43 The present study
shows that WhiB3 interacted with 113 proteins. Aside from its
roles in lipid transport and metabolism, and oxidoreduction,
this protein is also connected to many transport-related
proteins such as membrane transport protein, PE/PPE proteins,
and secretion proteins (Figure 7). These findings support the
previous observation that WhiB3 is a core regulator for the
pathogen.15,44
Conclusion
The current study presents the first global protein-protein
interaction analysis in the human pathogen M. tuberculosis
H37Rv. Almost the entire pathogenic ORFeome was cloned.
More than 8000 novel interactions were identified. The bacterial
proteins formed a highly linked network. The PPI information
was connected to the pathogenic protein secretion pathway
and gene regulation. These results will serve as a unique
resource for further dissection of signal transduction and
infection mechanisms in M. tuberculosis, as well as for the
development of new antituberculosis drugs.
Abbreviations: PPI, protein-protein interaction; BCG, bacille
Calmette-Guerin; B2H, bacterial two-hybrid; Y2H, yeast two-
hybrid; PCR, polymerase chain reaction; IPTG, isopropyl β-D-
1-thiogalactopyranoside; SPR, surface plasmon resonance; ORF,
open reading frame; SNP, single-nucleotide polymorphism.
Acknowledgment. This work was supported by the
National Natural Science Foundation of China (30930003,
31025002) and 973 Program (2006CB504402). This work was
also supported by the National Special Key Project of China
on Major Infectious Diseases (2008ZX10003-005).
Supporting Information Available: A complete list of
the identified subnetworks and representative plates for the
screenings; PPI enrichment analysis, topological analysis and
measures of each PPI network; power-law distribution of node
degree and topological coefficient; the distribution of protein
category and PPI; examples for the function predictions of
hypothetical proteins and new pathways based on the PPI
network; conserved subnetworks; protein-protein interaction
pairs of M. tuberculosis and PPI validation; protein-protein
interaction pairs involved in 22 ESAT6 and CFP10 proteins; no
PCR-amplified products of M. tuberculosis ORFs. This material
is available free of charge via the Internet at http://pubs.acs.org.
References
(1) World Health Organization. Global Tuberculosis Control 2009:
Epidemiology, Strategy, Financing. Nonserial Publication. WHO,
2009.
(2) Johnson, R.; Streicher, E. M.; Louw, G. E.; Warren, R. M.; van
Helden, P. D.; Victor, T. C. Drug resistance in Mycobacterium
tuberculosis. Curr. Issues Mol. Biol. 2006, 8, 97–111.
(3) von Mering, C.; Jensen, L. J.; Snel, B.; Hooper, S. D.; Krupp, M.;
Foglierini, M.; Jouffre, N.; Huynen, M. A.; Bork, P. STRING: known
and predicted protein-protein associations, integrated and trans-
ferred across organisms. Nucleic Acids Res. 2005, 33, D433–D437.
(4) Rain, J. C.; Selig, L.; De Reuse, H.; Battaglia, V.; Reverdy, C.; Simon,
S.; Lenzen, G.; Petel, F.; Wojcik, J.; Schachter, V.; Chemama, Y.;
Journal of Proteome Research • Vol. 9, No. 12, 2010 6675
research articles
Labigne, A.; Legrain, P. The protein-protein interaction map of
Helicobacter pylori. Nature 2001, 409, 211–215.
(5) Ito, T.; Chiba, T.; Ozawa, R.; Yoshida, M.; Hattori, M.; Sakaki, Y. A
comprehensive two-hybrid analysis to explore the yeast protein
interactome. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 4569–4574.
(6) Uetz, P.; Giot, L.; Cagney, G.; Mansfield, T. A.; Judson, R. S.; Knight,
J. R.; Lockshon, D.; Narayan, V.; Srinivasan, M.; Pochart, P.;
Qureshi-Emili, A.; Li, Y.; Godwin, B.; Conover, D.; Kalbfleisch, T.;
Vijayadamodar, G.; Yang, M.; Johnston, M.; Fields, S.; Rothberg,
J. M. A comprehensive analysis of protein-protein interactions in
Saccharomyces cerevisiae. Nature 2000, 403, 623–627.
(7) Yu, H.; Braun, P.; Yildirim, M. A.; Lemmens, I.; Venkatesan, K.;
Sahalie, J.; Hirozane-Kishikawa, T.; Gebreab, F.; Li, N.; Simonis,
N.; Hao, T.; Rual, J. F.; Dricot, A.; Vazquez, A.; Murray, R. R.; Simon,
C.; Tardivo, L.; Tam, S.; Svrzikapa, N.; Fan, C.; de Smet, A. S.; Motyl,
A.; Hudson, M. E.; Park, J.; Xin, X.; Cusick, M. E.; Moore, T.; Boone,
C.; Snyder, M.; Roth, F. P.; Barabasi, A. L.; Tavernier, J.; Hill, D. E.;
Vidal, M. High Quality Binary Protein Interaction Map of the Yeast
Interactome Network. Science 2008, 322, 104–110.
(8) Li, S.; Armstrong, C. M.; Bertin, N.; Ge, H.; Milstein, S.; Boxem,
M.; Vidalain, P. O.; Han, J. D.; Chesneau, A.; Hao, T.; Goldberg,
D. S.; Li, N.; Martinez, M.; Rual, J. F.; Lamesch, P.; Xu, L.; Tewari,
M.; Wong, S. L.; Zhang, L. V.; Berriz, G. F.; Jacotot, L.; Vaglio, P.;
Reboul, J.; Hirozane-Kishikawa, T.; Li, Q.; Gabel, H. W.; Elewa, A.;
Baumgartner, B.; Rose, D. J.; Yu, H.; Bosak, S.; Sequerra, R.; Fraser,
A.; Mango, S. E.; Saxton, W. M.; Strome, S.; Van Den Heuvel, S.;
Piano, F.; Vandenhaute, J.; Sardet, C.; Gerstein, M.; Doucette-
Stamm, L.; Gunsalus, K. C.; Harper, J. W.; Cusick, M. E.; Roth, F. P.;
Hill, D. E.; Vidal, M. A map of the interactome network of the
metazoan C. elegans. Science 2004, 303, 540–543.
(9) Giot, L.; Bader, J. S.; Brouwer, C.; Chaudhuri, A.; Kuang, B.; Li, Y.;
Hao, Y. L.; Ooi, C. E.; Godwin, B.; Vitols, E.; Vijayadamodar, G.;
Pochart, P.; Machineni, H.; Welsh, M.; Kong, Y.; Zerhusen, B.;
Malcolm, R.; Varrone, Z.; Collis, A.; Minto, M.; Burgess, S.;
McDaniel, L.; Stimpson, E.; Spriggs, F.; Williams, J.; Neurath, K.;
Ioime, N.; Agee, M.; Voss, E.; Furtak, K.; Renzulli, R.; Aanensen,
N.; Carrolla, S.; Bickelhaupt, E.; Lazovatsky, Y.; DaSilva, A.; Zhong,
J.; Stanyon, C. A.; Finley, R. L., Jr.; White, K. P.; Braverman, M.;
Jarvie, T.; Gold, S.; Leach, M.; Knight, J.; Shimkets, R. A.; McKenna,
M. P.; Chant, J.; Rothberg, J. M. A protein interaction map of
Drosophila melanogaster. Science 2003, 302, 1727–1736.
(10) Stelzl, U.; Worm, U.; Lalowski, M.; Haenig, C.; Brembeck, F. H.;
Goehler, H.; Stroedicke, M.; Zenkner, M.; Schoenherr, A.; Koeppen,
S.; Timm, J.; Mintzlaff, S.; Abraham, C.; Bock, N.; Kietzmann, S.;
Goedde, A.; Toksoz, E.; Droege, A.; Krobitsch, S.; Korn, B.;
Birchmeier, W.; Lehrach, H.; Wanker, E. E. A human protein-
protein interaction network: a resource for annotating the pro-
teome. Cell 2005, 122, 957–968.
(11) Sato, S.; Shimoda, Y.; Muraki, A.; Kohara, M.; Nakamura, Y.; Tabata,
S. A large-scale protein protein interaction analysis in Synechocystis
sp. PCC6803. DNA Res. 2007, 14, 207–216.
(12) Parrish, J. R.; Yu, J.; Liu, G.; Hines, J. A.; Chan, J. E.; Mangiola,
B. A.; Zhang, H.; Pacifico, S.; Fotouhi, F.; DiRita, V. J.; Ideker, T.;
Andrews, P.; Finley, R. L., Jr. A proteome-wide protein interaction
map for Campylobacter jejuni. Genome Biol. 2007, 8, R130.
(13) Titz, B.; Rajagopala, S. V.; Goll, J.; Häuser, R.; Mckevitt, M. T.;
Palzkill, T.; Uetz, P. The binary protein interactome of Treponema
pallidum-the syphilis spirochete. PLoS One 2008, 3, e2292.
(14) Rajagopala, S. V.; Titz, B.; Goll, J.; Parrish, J. R.; Wohlbold, K.;
McKevitt, M. T.; Palzkill, T.; Mori, H.; Finley, R. L., Jr.; Uetz, P.
The protein network of bacterial motility. Mol. Syst. Biol. 2007, 3,
128.
(15) Raman, K.; Chandra, N. Mycobacterium tuberculosis interactome
analysis unravels potential pathways to drug resistance. BMC
Microbiol. 2008, 8, 234.
(16) Cui, T.; Zhang, L.; Wang, X.; He, Z. G. Uncovering new signaling
proteins and potential drug targets through the interactome
analysis of Mycobacterium tuberculosis. BMC Genomics 2009, 10,
118.
(17) Cole, S. T.; Brosch, R.; Parkhill, J.; Garnier, T.; Churcher, C.; Harris,
D.; Gordon, S. V.; Eiglmeier, K.; Gas, S.; Barry, C. E.; Tekaia, F.;
Badcock, K.; Basham, D.; Brown, D.; Chillingworth, T.; Connor,
R.; Davies, R.; Devlin, K.; Feltwell, T.; Gentles, S.; Hamlin, N.;
Holroyd, S.; Hornsby, T.; Jagels, K.; Krogh, A.; McLean, J.; Moule,
S.; Murphy, L.; Oliver, K.; Osborne, J.; Quail, M. A.; Rajandream,
M. A.; Rogers, J.; Rutter, S.; Seeger, K.; Skelton, J.; Squares, R.;
Squares, S.; Sulston, J. E.; Taylor, K.; Whitehead, S.; Barrell, B. G.
Deciphering the biology of Mycobacterium tuberculosis from the
complete genome sequence. Nature 1998, 393, 537–544.
6676 Journal of Proteome Research • Vol. 9, No. 12, 2010
Wang et al.
(18) Dove, S. L.; Joung, J. K.; Hochschild, A. Activation of prokaryotic
transcription through arbitrary protein-protein contacts. Nature
1997, 386, 627–630.
(19) Dove, S. L.; Hochschild, A. Conversion of the omega subunit of
Escherichia coli RNA polymerase into a transcriptional activator
or an activation target. Genes Dev. 1998, 12, 745–754.
(20) Joung, J. K.; Ramm, E. I.; Pabo, C. O. A bacterial two-hybrid
selection system for studying protein-DNA and protein-protein
interactions. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 7382–7387.
(21) Zhang, L.; Liu, Y.; Yang, S.; Gao, C.; Gong, H.; Feng, Y.; He, Z. G.
Archaeal eukaryote-like Orc1/Cdc6 initiators physically interact
with DNA polymerase B1 and regulate its functions. Proc. Natl.
Acad. Sci. U.S.A. 2009, 106, 7792–7797.
(22) Zeng, J.; Zhang, L.; Li, Y.; Wang, Y.; Wang, M.; Duan, X.; He, Z. G.
Over-producing soluble protein complex and validating protein-
protein interaction through a new bacterial co-expression system.
Protein Expression Purif. 2010, 69, 47–53.
(23) Butland, G.; Peregrı́n-Alvarez, J. M.; Li, J.; Yang, W.; Yang, X.;
Canadien, V.; Starostine, A.; Richards, D.; Beattie, B.; Krogan, N.;
Davey, M.; Parkinson, J.; Greenblatt, J.; Emili, A. Interaction
network containing conserved and essential protein complexes in
Escherichia coli. Nature 2005, 433, 531–537.
(24) Arifuzzaman, M.; Maeda, M.; Itoh, A.; Nishikata, K.; Takita, C.;
Saito, R.; Ara, T.; Nakahigashi, K.; Huang, H. C.; Hirai, A.; Tsuzuki,
K.; Nakamura, S.; Altaf-Ul-Amin, M.; Oshima, T.; Baba, T.; Yama-
moto, N.; Kawamura, T.; Ioka-Nakamichi, T.; Kitagawa, M.; Tomita,
M.; Kanaya, S.; Wada, C.; Mori, H. Large-scale identification of
protein-protein interaction of Escherichia coli K-12. Genome Res.
2006, 16, 686–691.
(25) Hu, P.; Janga, S. C.; Babu, M.; Diaz-Mejia, J. J.; Butland, G.; Yang,
W.; Pogoutse, O.; Guo, X.; Phanse, S.; Wong, P.; Chandran, S.;
Christopoulos, C.; Nazarians-Armavil, A.; Nasseri, N. K.; Musso,
G.; Ali, M.; Nazemof, N.; Eroukova, V.; Golshani, A.; Paccanaro,
A.; Greenblatt, J. F.; Moreno-Hagelsieb, G.; Emili, A. Global
functional atlas of Escherichia coli encompassing previously un-
characterized proteins. PLoS Biol. 2009, 7, e96.
(26) Bowers, P. M.; Pellegrini, M.; Thompson, M. J.; Fierro, J.; Yeates,
T. O.; Eisenberg, D. Prolinks: a database of protein functional
linkages derived from coevolution. Genome Biol. 2004, 5, R35.
(27) Jensen, L. J.; Kuhn, M.; Stark, M.; Chaffron, S.; Creevey, C.; Muller,
J.; Doerks, T.; Julien, P.; Roth, A.; Simonovic, M.; Bork, P.; von
Mering, C. STRING 8-a global view on proteins and their functional
interactions in 630 organisms. Nucleic Acids Res. 2009, 37, D412–
D416.
(28) Shannon, P.; Markiel, A.; Ozier, O.; Baliga, N. S.; Wang, J. T.;
Ramage, D.; Amin, N.; Schwikowski, B.; Ideker, T. Cytoscape: a
software environment for integrated models of biomolecular
interaction networks. Genome Res. 2003, 13, 2498–2504.
(29) Assenov, Y.; Ramı́rez, F.; Schelhorn, S. E.; Lengauer, T.; Albrecht,
M. Computing topological parameters of biological networks.
Bioinformatics 2008, 24, 282–284.
(30) Sharan, R.; Suthram, S.; Kelley, R. M.; Kuhn, T.; McCuine, S.; Uetz,
P.; Sittler, T.; Karp, R. M.; Ideker, T. Conserved patterns of protein
interaction in multiple species. Proc. Natl. Acad. Sci. U.S.A. 2005,
102, 1974–1979.
(31) Jeong, H.; Mason, S. P.; Barabasi, A. L.; Oltvai, Z. N. Lethality and
centrality in protein networks. Nature 2001, 411, 41–42.
(32) Garton, N. J.; Waddell, S. J.; Sherratt, A. L.; Lee, S. M.; Smith, R. J.;
Senner, C.; Hinds, J.; Rajakumar, K.; Adegbola, R. A.; Besra, G. S.;
Butcher, P. D.; Barer, M. R. Cytological and Transcript Analyses
Reveal Fat and Lazy Persister-Like Bacilli in Tuberculous Sputum.
PLoS Med. 2008, 5, e75.
(33) Fontan, P.; Aris, V.; Ghanny, S.; Soteropoulos, P.; Smith, I. Global
transcriptional profile of Mycobacterium tuberculosis during THP-1
human macrophage infection. Infect. Immun. 2008, 76, 717–725.
(34) Alteri, C. J.; Xicohtencatl-Cortes, J.; Hess, S.; Caballero-Olin, G.;
Giron, J. A.; Friedman, R. L. Mycobacterium tuberculosis produces
pili during human infection. Proc. Natl. Acad. Sci. U.S.A. 2007, 104,
5145–5150.
(35) Ton-That, H.; Schneewind, O. Assembly of pili on the surface of
Corynebacterium diphtheriae. Mol. Microbiol. 2003, 50, 1429–1438.
(36) Lauer, P.; Rinaudo, C. D.; Soriani, M.; Margarit, I.; Maione, D.;
Rosini, R.; Taddei, A. R.; Mora, M.; Rappuoli, R.; Grandi, G.; Telford,
J. L. Genome analysis reveals pili in Group B Streptococcus. Science
2005, 309, 105.
(37) Ton-That, H.; Marraffini, L. A.; Schneewind, O. Sortases and pilin
elements involved in pilus assembly of Corynebacterium diphthe-
riae. Mol. Microbiol. 2004, 53, 251–261.
(38) Paterson, G. K.; Mitchell, T. J. The biology of Gram-positive sortase
enzymes. Trends Microbiol. 2004, 12, 89–95.
Protein-Protein Interaction Network in M. tuberculosis H37Rv
(39) Sutcliffe, I. C.; Harrington, D. J. Lipoproteins of Mycobacterium
tuberculosis: an abundant and functionally diverse class of cell
envelope components. FEMS Microbiol. Rev. 2004, 28, 645–659.
(40) Vosloo, W.; Tippoo, P.; Hughes, J. E.; Harriman, N.; Emms, M.;
Beatty, D. W.; Zappe, H.; Steyn, L. M. Characterization of a
lipoprotein in Mycobacterium bovis (BCG) with sequence similarity
to the secreted protein MPB70. Gene 1997, 188, 123–128.
(41) Harboe, M.; Wiker, H. G.; Ulvund, G.; Lund-Pedersen, B.; Andersen,
A. B.; Hewinson, R. G.; Nagai, S. MPB70 and MPB83 as indicators
of protein localization in mycobacterial cells. Infect. Immun. 1998,
66, 289–296.
(42) Champion, P. A.; Stanley, S. A.; Champion, M. M.; Brown, E. J.;
Cox, J. S. C-terminal signal sequence promotes virulence factor
secretion in Mycobacterium tuberculosis. Science 2006, 313, 1632–
1636.
(43) Abdallah, A. M.; Verboom, T.; Hannes, F.; Safi, M.; Strong, M.;
Eisenberg, D.; Musters, R. J.; Vandenbroucke-Grauls, C. M.;
Appelmelk, B. J.; Luirink, J.; Bitter, W. A specific secretion system
mediates PPE41 transport in pathogenic mycobacteria. Mol.
Microbiol. 2006, 62, 667–679.
(44) Guo, M. M.; Feng, H.; Zhang, J.; Wang, W. Q.; Wang, Y.; Li, Y. Q.;
Gao, C. H.; Chen, H. C.; Feng, Y.; He, Z. G. Dissecting transcription
regulatory pathways through a new bacterial one-hybrid reporter
system. Genome Res. 2009, 19, 1301–1308.
(45) Morris, R. P.; Nguyen, L.; Gatfield, J.; Visconti, K.; Nguyen, K.;
Schnappinger, D.; Ehrt, S.; Liu, Y.; Heifets, L.; Pieters, J.; Schoolnik,
G.; Thompson, C. J. Ancestral antibiotic resistance in Mycobacte-
rium tuberculosis. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 12200–
12205.
(46) Dove, S. L.; Joung, J. K.; Hochschild, A. Activation of prokaryotic
transcription through arbitrary protein-protein contacts. Nature
1997, 386, 627–630.
(47) Dove, S. L.; Hochschild, A. Conversion of the omega subunit of
Escherichia coli RNA polymerase into a transcriptional activator
or an activation target. Genes Dev. 1998, 12, 745–754.
(48) Joung, J. K.; Ramm, E. I.; Pabo, C. O. A bacterial two-hybrid
selection system for studying protein-DNA and protein-protein
interactions. Proc. Natl. Acad. Sci. USA. 2000, 97, 7382–7387.
research articles
(49) Strogatz, S. H. Exploring complex networks. Nature 2001, 410, 268–
276.
(50) Molle, V.; Brown, A. K.; Besra, G. S.; Cozzone, A. J.; Kremer, L. The
condensing activities of the Mycobacterium tuberculosis type II
fatty acid synthase are differentially regulated by phosphorylation.
J. Biol. Chem. 2006, 281, 30094–30103.
(51) Veyron-Churlet, R.; Molle, V.; Taylor, R. C.; Brown, A. K.; Besra,
G. S.; Zanella-Cleon, I.; Futterer, K.; Kremer, L. The Mycobacterium
tuberculosis beta-ketoacyl-acyl carrier protein synthase III activity
is inhibited by phosphorylation on a single threonine residue.
J. Biol. Chem. 2009, 284, 6414–6424.
(52) Veyron-Churlet, R.; Guerrini, O.; Mourey, L.; Daffe, M.; Zerbib, D.
Protein-protein interactions within the Fatty Acid Synthase-II
system of Mycobacterium tuberculosis are essential for mycobac-
terial viability. Mol. Microbiol. 2004, 54, 1161–1172.
(53) Handa, P.; Acharya, N.; Varshney, U. Chimeras between single-
stranded DNA-binding proteins from Escherichia coli and Myco-
bacterium tuberculosis reveal that their C-terminal domains
interact with uracil DNA glycosylases. J. Biol. Chem. 2001, 276,
16992–16997.
(54) Parida, B. K.; Douglas, T.; Nino, C.; Dhandayuthapani, S. Interac-
tions of anti-sigma factor antagonists of Mycobacterium tubercu-
losis in the yeast two-hybrid system. Tuberculosis (Edinb.) 2005,
85, 347–355.
(55) Mahairas, G.; Sabo, P.; Hickey, M.; Singh, D.; Stover, C. Molecular
analysis of genetic differences between Mycobacterium bovis BCG
and virulent M. bovis. J. Bacteriol. 1996, 178, 1274–1282.
(56) Pym, A. S.; Brodin, P.; Brosch, R.; Huerre, M.; Cole, S. T. Loss of
RD1 contributed to the attenuation of the live tuberculosis
vaccines Mycobacterium bovis BCG and Mycobacterium microti.
Mol. Microbiol. 2002, 46, 709–717.
(57) Mostowy, S.; Cousins, D.; Behr, M. A. Genomic interrogation of
the dassie bacillus reveals it as a unique RD1 mutant within the
Mycobacterium tuberculosis complex. J. Bacteriol. 2004, 186, 104–
109.
(58) Parish, T.; Brown, A. Mycobacterium: genomics and molecular
biology; Caister Academic Press: U.K., 2009; p 202.
PR100808N
Journal of Proteome Research • Vol. 9, No. 12, 2010 6677