REVIEWS

ESX secretion systems: mycobacterial

evolution to counter host immunity
Matthias I. Gröschel1,2, Fadel Sayes1, Roxane Simeone1, Laleh Majlessi1
and Roland Brosch1
Abstract | Mycobacterium tuberculosis uses sophisticated secretion systems, named 6 kDa early
secretory antigenic target (ESAT6) protein family secretion (ESX) systems (also known as type VII
secretion systems), to export a set of effector proteins that helps the pathogen to resist or evade
the host immune response. Since the discovery of the esx loci during the M. tuberculosis H37Rv
genome project, structural biology, cell biology and evolutionary analyses have advanced our
knowledge of the function of these systems. In this Review, we highlight the intriguing roles that
these studies have revealed for ESX systems in bacterial survival and pathogenicity during
infection with M. tuberculosis. Furthermore, we discuss the diversity of ESX systems that has been
described among mycobacteria and selected non-mycobacterial species. Finally, we consider
how our knowledge of ESX systems might be applied to the development of novel strategies for
the treatment and prevention of disease.

Actinobacteria
A phylum of Gram-positive
bacteria that is characterized
by high GC content.
Mycolic acids
Long-chain (C60–C90) fatty acids
that are specifically found in
the mycobacterial cell
envelope; together with
extractable lipids, mycolic
acids form the mycobacterial
outer membrane (also known
as the mycomembrane).
1
Institut Pasteur, Unit for
Integrated Mycobacterial
Pathogenomics, Paris 75724,
Cedex 15, France.
2
Department of Pulmonary
Diseases & Tuberculosis,
University Medical Center
Groningen, University
of Groningen, 9700 RB
Groningen, The Netherlands.
Correspondence to R.B.
roland.brosch@pasteur.fr
doi:10.1038/nrmicro.2016.131
Published online 26 Sep 2016
ESAT6 secretion (ESX; also known as type VII secre-
tion) systems are bacterial secretion systems that are
named after the first identified effector, the 6 kDa
early secretory antigenic target (ESAT6; also known as
EsxA), of Mycobacterium tuberculosis, the aetiological
agent of human tuberculosis1,2. ESX systems are found
in mycobacteria and various other genera in the phylum
Actinobacteria3,4, such as Streptomyces, Corynebacterium,
Nocardia or Gordonia, and more distantly related ESX-
like systems also exist in Gram-positive bacteria in the
phylum Firmicutes, including in Bacillus anthracis 5,
Bacillus subtilis 6,7, Staphylococcus aureus 8,9 and Listeria
monocytogenes 10 (BOX 1). In mycobacteria, ESX systems
function as specialized secretion systems that enable the
transport of selected substrates across the complex, thick
mycobacterial cell envelope that forms a structural bar-
rier to protein export 11. The thickness and complexity
of the envelope, which provides protection to myco­
bacteria under harsh environmental conditions, are due
to the presence of mycolic acids that are linked, usually
covalently, to an arabinogalactan–peptidoglycan matrix, as
well as various extractable lipids, polysaccharides, lipo­
glycans and proteins that are not covalently attached to
the matrix and that may vary among species12.
Two characteristics unify all esx loci across the differ-
ent phyla. First, the presence of genes that encode small
secreted proteins of approximately 100 amino acids
that have a conserved Trp‑X‑Gly (WXG) motif, which
contributes to the formation of helix–turn–helix struc-
tures13, in the centre of the polypeptide; and, second, the
presence of genes that encode transmembrane proteins
of the FtsK–SpoIIIE-like ATPase family 14. The most well-
known proteins that contain the WXG motif are EsxA
of M. tuberculosis and its adjacently encoded hetero­
dimerization partner EsxB (also known as CFP10)15.
Apart from these core characteristics, ESX systems are
quite diverse, which suggests that they have been shaped
by a long evolutionary process that has involved gene
duplication and diversification3,16,17, as well as horizontal
gene transfer between chromosomes and plasmids of
different bacterial species and genera4,18 (BOX 2).
Of the five ESX systems that have been described in
M. tuberculosis (ESX‑1, ESX‑2, ESX‑3, ESX‑4 and ESX‑5;
FIG. 1a), at least three are required for full viru­lence. The
first ESX system (ESX‑1) was identified in parallel by
different comparative and functional genomic studies
that involved M. tuberculosis and the attenuated vaccine
strains Mycobacterium bovis bacille Calmette–Guérin
(BCG) and Mycobacterium microti 19–24. The vaccine
strains lack EsxA, owing to spontaneous deletions of dif-
ferent sized portions of the esx‑1 locus, each of which is
known as region of difference 1 (RD1) for the respective
strain25,26 (FIG. 1a). ESX‑1 in M. tuberculosis has subse-
quently been shown to be essential for resistance to, and
evasion of, host responses. One of the key functions of
ESX‑1 is its role in the induction of phagosomal rupture,
which releases bacteria and/or bacterial products into
the cytosolic compartment of host phagocytes. The sens-
ing of bacterial products, such as DNA, triggers a com-
plex signalling cascade of the innate immune system,

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 677

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Arabinogalactan– Box 1 | ESX-like systems in Gram-positive bacteria in the Firmicutes phylum

peptidoglycan matrix
An essential constituent of the The genome sequences of several species in the low‑GC a
mycobacterial cell wall, phylum Firmicutes are predicted to encode 6 kDa early Staphylococcus aureus ess locus
consisting of peptidoglycan secretory antigenic target (ESAT6) secretion (ESX)-like
that is covalently attached to systems, based on the presence of genes that are
the heteropolysaccharide predicted to encode an FtsK–SpoIIIE-like ATPase and
xA saA ssAsaB essB sC xC xB D sD
arabinogalactan, which is
associated small proteins that have conserved Trp‑X‑Gly es e e e es es es esx es
linked to the mycolic acid layer.
(WXG) motifs in the centre of a ~100 amino acid b
polypeptide14, which are the two characteristic features EsxB–EsxC EsxB–EsxD
Mycobactin
An iron-binding compound that that are found in all ESX and ESX-like systems. EsxA–EsxC
is synthesized by most For example, ESX-like systems have been identified in EsxA–EsxA
mycobacteria and is necessary Bacillus anthracis , Bacillus subtilis , Listeria
5 6,7

for the recovery of iron, which
is an essential element
for growth.
Diderm
Refers to the presence of
two membranes (an inner
membrane and an outer
membrane) in the cell
envelope. Diderm phyla
encompass both
Gram-negative bacteria
and mycobacteria.
Bootstrap replicates
Statistical confidence values
in phylogenetic trees that are
inferred from phylogenetic
bootstrapping, which is an
informatics-based method
that is based on reconstructing
many trees or replicates
from minor variations of the
input data.
monocytogenes10 and Staphylococcus aureus8,9. The ESAT6 EssA EsaA
secretion system (ESS)8 of S. aureus is probably the Peptidoglycan
best-characterized ESX-like system, owing to its role in
virulence, and can thus be considered as a model for the
study of ESX-like systems in members of the Firmicutes.
The ess locus in S. aureus encodes 11 genes (see the Cytoplasm
figure, part a)8,137, some of which show recognizable, but EssB EssC
EssD
low, similarity to genes from mycobacterial esx loci. For
example, the ess locus encodes two WXG100 proteins, EsaB
EsxA and EsxB, that share 21% and 18% sequence identity
with EsxA and EsxB in Mycobacterium tuberculosis,
Nature Reviews | Microbiology
respectively. Similarly to the mycobacterial proteins, EsxA and EsxB in S. aureus are secreted substrates that are involved in
host–pathogen interactions. However, the heterodimeric complexes that are formed by EsxA and EsxB in mycobacteria are
not observed in S. aureus; instead, a homodimer formed by EsxA was revealed by X‑ray crystallography138. Furthermore,
biochemical evidence shows that the dimeric complexes EsxA–EsxA, EsxC–EsxC, EsxA–EsxC and EsxB–EsxD are all
secreted by the ESX-like system in S. aureus137; similarly to EsxA and EsxB, EsxC and EsxD are bacterial effectors that are
involved in host–pathogen interactions9,137,139.
The secretion of EsxA and EsxB depends on other genes that are located in the ess locus8,140. Although the mechanistic
roles of these other genes are unclear, a model has been proposed in which esaA encodes a protein that is involved in
substrate secretion across the membrane and essA, essB, essC and essD (also known as esaD) encode transmembrane
proteins that together form a complex with EsaA9,137,139,141,142 (see the figure, part b). In support of this model, recent work
has shown that mutants that lack essD are defective in the secretion of EsxA141. essC is another example of an ess gene that
shows sequence similarity to a mycobacterial esx gene, as it resembles, to some extent, eccC of M. tuberculosis; both essC in
S. aureus and eccC in M. tuberculosis encode Ftsk–SpoIIIE-like ATPases8.
ESX-like secretion in S. aureus has an important role in abscess formation and staphylococcal persistence in host tissues,
but it is dispensable for bacterial growth8,9. For example, compared with wild-type S. aureus, mutant strains of S. aureus
that do not secrete EsxA or EsxB were associated with a substantial decrease in bacterial load in the liver and kidneys in
mouse models8, and intravenous inoculation of mice with S. aureus essD-deletion mutants generated fewer abscesses and
a lower bacterial load141. ESX-like secretion was also found to be required for nasal colonization in a murine lung
pneumonia model, both for RN6390 and COL strains of S. aureus139. Finally, EsxA in S. aureus inhibits apoptotic pathways
in human epithelial cells, as shown by the increased rate of apoptosis in cells that were infected with esxA-deletion
mutants compared with cells that were infected with wild-type S. aureus143. Together, these observations suggest that
ESX-like systems in S. aureus fulfil key functions during the infection process.

with a major effect on host–pathogen interactions and
immunity. Although to a lesser extent than ESX‑1, the
functions of ESX‑3 and ESX‑5 have also been studied,
with reports describing ESX‑3 as a secretion system that
is involved in mycobactin-mediated iron acquisition27–29
and ESX‑5 as a secretion system for two families of pro-
teins that are specific to mycobacteria (Pro-Glu (PE) and
Pro-Pro-Glu (PPE) proteins)16,30–33.
As an alternative to ‘ESX systems’, the nomenclature
‘type VII secretion systems’ is commonly used in the
literature34,35, although an initial debate questioned
the semantic appropriateness of a nomenclature derived
from that of the type I–VI secretion systems, which are
found in Gram-negative bacteria36,37. However, although
mycobacteria are considered to be Gram-positive bac-
teria, the presence of a mycolic acid-containing outer
membrane (FIG. 1b) makes them more comparable, for
the purposes of secretion, with Gram-negative bacteria,
which require specialized systems to ensure the secre-
tion of specific substrates across their diderm cell enve-
lope. In contrast to the well-studied type I–VI secretion
systems, for which functions and structures have largely
been deciphered38, high-resolution structural data on
the more recently discovered ESX systems are only now
starting to be generated. As such, the current model of
the structure and function of the ESX apparatus is still
hypothetical and many questions about specific details of
the biology of ESX systems remain (FIG. 1c). As ESX sys-
tems have numerous roles in the physiology, cell envelope
integrity, conjugation and host–pathogen interactions of
mycobacteria, future work on these systems may reveal
new discoveries that have the potential to broaden our
understanding of many tuberculosis-related features of
mycobacterial biology. In this Review, we describe the

678 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Box 2 | Evolution and horizontal gene transfer of ESX components

The discovery of five loci that encode 6 kDa early secretory antigenic target
(ESAT6) secretion (ESX) systems in the genome of Mycobacterium
tuberculosis H37Rv3,14,16,17 raised interesting questions regarding the origin,
distribution and evolution of these systems and the expansion of the Pro-Glu
(PE) and Pro-Pro-Glu (PPE) gene families39. ESX‑4 systems are thought to
represent the simplest and earliest ESX systems, from which others evolved
by gene duplication and diversification3,35. Indeed, ESX‑4 systems have also
been identified in numerous non-mycobacterial Actinobacteria and show
the most similarity to ESX-like systems in the Firmicutes phylum.
Furthermore, a recent mycobacterial pan-genome analysis, which
constructed a phylogeny based on concatenated sequences of EccB, EccC
and MycP using a maximum-likelihood method with 250 bootstrap replicates
(see the figure; branch lengths are drawn to scale and indicate the number of
substitutions per site), identified a subgroup of ESX‑4 systems, ESX‑4‑bis,
that branches deeper in the phylogeny than previously described systems
and that is present in some non-mycobacterial Actinobacteria, such as
Nocardia farcinica and Gordonia bronchialis (see the figure, green boxes), in
addition to a subset of mycobacterial species4. This analysis also identified
large plasmids that encoded novel esx loci (see the figure, red circles) or
putative esx loci (see the figure, red circles with dashed lines) in both
rapid-growing mycobacteria (see the figure; RGM) and slow-growing
mycobacteria (see the figure; SGM), which suggests that plasmids might
have been major facilitators of the radiation and evolution of mycobacterial
ESX systems4, as was also proposed in a second, independent study18. One of
the plasmids that encodes an esx locus, pRAW from Mycobacterium
marinum, is a conjugative plasmid that also encodes proteins that have some
similarity to components of type IV secretion systems144, which generally
only exist in Gram-negative bacteria. The ESX system that is encoded on
pRAW, ESX‑P1, is similar to ESX‑P systems that are encoded on plasmids
from Mycobacterium kansasii145 and Mycobacterium yongonense146, and the
genes that encode all three of these systems form branches at the root of the
phylogeny of chromosomal ESX‑5 systems4 (see the figure). This location in
the ESX‑5 phylogeny suggests that plasmids that encode ESX‑P1 systems
were involved in conjugation-mediated transfer of ESX‑5 precursors to
slow-growing mycobacteria, which is consistent with the high sequence
similarity among the Ecc and MycP proteins from ESX‑P1 and ESX‑5
systems4,144. Other deep-branching ESX‑P‑containing plasmids might have
had similar roles in the transfer of other ESX systems4,18.
ESX systems have also been shown to enable non-plasmid-mediated
horizontal transfer of chromosomal DNA fragments. Laboratory-based
conjugation experiments in the non-pathogenic, rapid-growing species
Mycobacterium smegmatis showed that DNA could be transferred to a
recipient strain that expressed the ESX‑1 system from a donor strain of the
same species, whereby the obtained transconjugants had highly mosaic
genomes that were reminiscent of products of eukaryotic meiosis147.
Similarly, genome analyses of tubercle bacilli from the Mycobacterium
canettii clade, which is proposed to be the clade from which M. tuberculosis
originated, showed numerous traces of putative interstrain recombination
events148–150. These findings were recently confirmed and extended through
the use of mycobacterial mating assays combined with whole-genome
sequencing, which provided evidence of transfer of large chromosomal DNA
fragments between two strains of M. canettii151, which suggests that such
genetic transfer may have had a role in the emergence of M. tuberculosis and
other M. tuberculosis complex (MTBC) strains. Further studies are required to
obtain more comprehensive insights not only into the potential role of ESX
systems in horizontal gene transfer in tubercle bacilli, but also into how this
mechanism of horizontal transfer might lead to the dissemination of ESX
systems and/or components among other mycobacteria, including
non-pathogenic species. Indeed, the wide distribution and the observed
diversity of ESX systems that are encoded by chromosomes and plasmids
suggest that our current knowledge of ESX systems, which has focused on
pathogenicity, probably only represents a very small portion of the biological
functions that might be carried out by these systems in the diverse
rapid-growing and slow-growing mycobacteria in which they are found.

ESX-3
RGM
ESX-P
cluster 4
SGM
100 100
100
100
98 99 100 84 97
RGM 96
100
ESX-1
94
82 99 99
SGM 94

100
100 99
72 97 100 100
95 81 98
98 99 99
99 98
100 RGM
99 98
97 96 97
97 98 99
100
100 74
ESX-2 SGM 100 95 100
100
100 100 93
100 ESX-4
100 100
ESX-P 100
99 100
cluster 2 ESX-P 96
ESX-P cluster 1 83
SGM
cluster 3
ESX-4-bis
SGM
0.2
ESX-5

Nature Reviews | Microbiology

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 679
©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

a
espACD operon
pA pC pD
es es es

ΔRD1MIC ΔRD1BCG
AAA+ ATPase FtsK/SpoIIIE CFP10 ESAT6
ESX-1
pE F 1 G H A 1 ccB 1 a1 b 1 e35 e68 sxBsxA espI cD 1 esp
J
pK pL B cE 1 cP 1
es esp esp esp ecc e cC cC p pp e e ec es es esp ec my
ec ec

ESX-2
cB 2 cC 2 36 69 xD xC G 2 8c D2 cP 2 cE 2 cA 2
ec ec pe ppe es es esp 388 ecc my ec ec
rv

ESX-3
cA 3 cB 3 cC 3 5 4 xH G 3 cD ycP 3 ccE 3
pe ppe esxG
3
ec ec ec es esp ec m e

ESX-4
cB 4 cP 4 cD 4 cC 4 6c U T
ec my ec ec 44 esx esx
rv3

ESX-5
cB 5 cC 5 43 86 25 18 26 7
e2 19 xM xN pG 5 eccD 5 ycP 5 eccE 5 cA 5
ec ec p1 v17 ppe pe ppe pp pe es es es m ec
cy r

b c
EspB
EsxA
EsxB

PPE
PE
EsxA
EsxB Capsule
Es

PPE
Es

PE
pA

pC

Outer membrane
? (mycomembrane)

Outer membrane Arabinogalactan

(mycomembrane)

Inner membrane
Peptidoglycan

MycP EccD EccC

EspB EccB EccB Periplasm

Inner membrane

Cytoplasm
EccE DUF DUF EccE
1 1
EspC
EspA

ATPase
domains 2 2
EccC 3 C C 3 EccA
EsxB
EsxB

PPE
EsxA
EsxA

PE

EspG

Nature Reviews | Microbiology

680 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro
©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

◀ Figure 1 | Genetic and structural architectures of ESX systems. a | Genetic
organization of the five 6 kDa early secretory antigenic target (ESAT6) secretion (esx) loci
and the espACD operon in Mycobacterium tuberculosis H37Rv. The esx‑1 locus can be
viewed as a model locus for ESX systems and includes esx genes that encode the two
secreted effector proteins EsxA and EsxB, ecc genes that encode proteins deemed to be
ESX-conserved components, and esp genes that encode proteins deemed to be ESX
­secretion-associated proteins35. Spontaneous deletions in esx‑1 that are found in the
vaccine strains Mycobacterium bovis bacille Calmette–Guérin (BCG; red shading) and
Mycobacterium microti (MIC; blue shading) are thought to be associated with attenuated
virulence. The deletion in each of these strains encompasses a different, but overlapping,
region of the locus but in both strains is known as region of difference 1 (RD1). The
espACD operon is a distinct locus from the five esx loci but shares sequence homology
(dashed lines) with espE, espF and espH in esx‑1. b | Electron microscopy image showing
the inner and outer membranes of the mycobacterial cell envelope. For the purposes
of secretion, the organization of the mycobacterial cell envelope is fundamentally
different from the cell envelope of most other Gram-positive bacteria, owing to the
presence of an outer membrane (also known as the mycomembrane). The composition
of the mycomembrane is notable for the presence of mycolic acids and a set of
non-covalently bound lipids that may include sulfatides, acyltrehaloses, trehalose
mycolates, phthiocerol dimycocerosates and phosphatidyl-myo-inositol
mannosides12,168,169. c | Model of the secretion apparatus and secreted substrates of ESX
systems. In this model, the conserved components EccB, EccC, EccD and EccE, which
each contain one or more transmembrane domains (an example number is shown), form
the core structure of ESX systems32, which is located in the inner membrane. EccC has a
unique multidomain structure that consists of a two-pass transmembrane domain, a
short domain of unknown function (DUF) and three ATPase domains, as was recently
shown using EccC from the thermophilic actinobacterium Thermomonospora curvata51.
EccC is thought to be a translocase that provides the energy for the secretion of an
effector known as EsxB, with which it interacts at the carboxy‑terminal signal sequence
(denoted as C in the figure)45,51. EsxB is the putative chaperone for EsxA, which is a major
ESX‑1 effector protein that is thought to be translocated through co‑secretion with
EsxB52. EccA is a cytosolic AAA+ ATPase that interacts with the C terminus of EspC61.
The exact role of EspC, which also interacts with EspA, in the ESX secretion process is not
described, but it is known that the secretion of EspC and EspA is co‑dependent, similarly
to the co‑dependent secretion of EsxA and EsxB59,60. Heterodimers that are formed from
a Pro-Glu (PE) family protein and a Pro-Pro-Glu (PPE) family protein are also substrates for
secretion, and the recruitment of these dimers to the core complex involves interaction
with the putative cytosolic chaperone EspG67–69. EspB is an ESX‑1‑secreted protein that
adopts a PE–PPE-like fold and contains a C‑terminal domain that is processed by the
MycP1 protease during secretion57,58. EspB may then oligomerize into a heptameric
donut-shaped complex57,58 (not shown). Once transported across the inner membrane,
substrates may be transported across the outer membrane through a channel, although
whether such a channel exists, and, if so, what the nature of it is, are currently unknown.
Image in part b is republished with permission of the American Society for Microbiology,
from Direct visualization of the outer membrane of mycobacteria and corynebacteria in
their native state, Zuber, B. et al., 190, 16, (2008); permission conveyed through Copyright
Clearance Center, Inc.
five ESX systems that are found in M. tuberculosis and
use select examples to illustrate known and predicted
functions of these systems and their effector molecules,
including the proposed role of ESX‑1 and its EsxA effector
in phagosome rupture and establishing bacterial contact
with the host cytosol. Finally, we discuss the evolution of
these systems, as well as the diversity of ESX and ESX-
like systems that are found among different mycobacterial
species and in certain Gram-positive bacteria.
AAA+ ATPase
A member of a large, Loci that encode ESX systems
functionally diverse protein One of the highlights of the first report of a whole-­ accessory proteins, such as chaperones. Although global
family of ATPases that are genome sequence for M. tuberculosis, which analysed mechanistic insights into the structure and function of
associated with various cellular the well-characterized H37Rv reference strain, was the the core ESX secretory apparatus are still lacking, numer-
activities that involve
energy-dependent remodelling
identification of multigene families, such as those associ- ous studies have now reported details on some aspects
or the translocation of ated with ESX systems1,16. Further analyses confirmed the of these ­systems — notably protein–protein interac-
macromolecules. presence of five paralogous ESX clusters in the genome of tions, secretion motifs and circuits of transcriptional
M. tuberculosis 3,17, each of which encoded a tandem pair
of WXG proteins, an ATPase with an Ftsk–SpoIIIE motif
and several proteins with predicted trans­membrane
domains. Detailed comparisons of the different loci iden-
tified a core set of proteins that were present in most of
the five ESX systems, which were named ESX-conserved
components (Ecc)35. In the locus that encodes the ESX‑1
system of M. tuberculosis, which is often used as a model
for ESX systems, the genes that encode the secreted pro-
teins EsxA and EsxB are flanked directly upstream by
pe35 and ppe68, which encode locus-specific PE and
PPE proteins; further upstream of pe35 and ppe68 are
the genes that encode the two components of the Ftsk–
SpoIIIE-like ATPase (EccCa and EccCb), a membrane
protein that has two trans­membrane domains (EccB)
and an AAA+ ATPase (EccA). Downstream of esxA and
esxB are genes that encode two transmembrane proteins
(EccD, which has at least 11 transmembrane domains,
and EccE, which has two transmembrane domains). The
esx‑1 locus encodes a subtilisin-like protease (MycP) and
several ESX‑1 secretion-­associated proteins (Esp), some
of which are specific to ESX‑1, whereas others (such as
EspG) have homologues in other ESX systems (FIG. 1a).
Genomic analyses of esx loci and the core compo-
nents contained therein have also suggested a possi­ble
evolutionary history for the origin of ESX systems. In
M. tuberculosis, the esx‑4 locus has the smallest num-
ber of genes of the five esx loci and seems to encode
the least complex system (FIG.  1a). Systems that are
orthologous to ESX‑4 are also present in various
mycobacterial and non-mycobacterial species in the
phylum Actinobacteria3,4. Together, these findings led
to the hypothesis that the ancestral ESX system was an
ESX‑4‑like system from which the other ESX systems
emerged by gene duplication, gene diversification and
horizontal gene transfer 3,4,18,39 (BOX 2). The location in
the phylogeny of ESX‑2 provides an example of the
roles of these processes in the evolution of ESX systems.
Although the genomic region that encodes ESX‑2 is
proximal to esx‑1 in M. tuberculosis 16, esx‑2 does not
correspond to an ancient duplication of the M. tuber‑
culosis esx‑1 locus. Instead, phylogenetic analyses sug-
gest that slow-growing mycobacteria acquired ESX‑2
by ­plasmid-mediated horizontal transfer after the emer-
gence of ESX‑1 (REF. 4) (BOX 2). A similar scenario may
have also led to the acquisition of ESX‑5, which, similarly
to ESX‑2 but unlike other ESX systems, is restricted to
slow-growing mycobacteria. The restricted distribution
of ESX‑2 and ESX‑5 suggests that these are the two ESX
systems that arose most recently in evolution (BOX 2).
Mechanisms and functions of ESX systems
The protein repertoires that are encoded by the different
esx loci are thought to consist of various components
of the trans­locating complex, secreted substrates and

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 681

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

a 89 C termini based on X‑ray crystallography 41 (FIG. 2a). Previous stud-

ies in tubercle bacilli have shown that the last 3–12 amino
acids of the carboxyl terminus of EsxA are required for
N terminus PE25 biological function and virulence, but not secretion40,42–44.
By contrast, the last seven amino acids of the C terminus
α1 of EsxB were found to be essential for EsxB secretion,
PPE41
C terminus owing to an interaction with EccCb (REF. 45), although
115 α2 an additional sequence motif (Tyr-XXX-Asp/Glu
EsxB (YXXXD/E)) that is adjacent to these residues is also
required for full ESX secretion46. The YXXXD/E motif
N terminus has been defined as a general secretion motif of ESX
EsxA
α1 pathways and has been postulated to interact with the
WXG motif, which probably promotes protein–­protein
interactions between (or within) ESX system pro-
teins41,47. Intriguingly, PE35 and PPE68 of the ESX‑1
system are predicted to also form a four-helical com-
plex that is similar to the structure that has been solved
for PE25–PPE41, which, in M. tuberculosis, are encoded
EspG5 by genes outside of a defined esx locus48 (FIG. 2a). PE35,
EspB
which, similarly to EsxB, contains a YXXXD/E motif,
b is required for the expression of esxA and esxB49,50. By
contrast, the function of PPE68, which contains a WXG
~110 aa ~180 aa 200–3500+ aa motif, remains unclear: transposon analyses identified
PE PPE MPTR PPE68 as essential for the full virulence of M. tuber‑
culosis 24, whereas truncation of its C terminus did not
~110 aa 170–588 aa ~180 aa 200–400 aa decrease the secretion of EsxA and EsxB or attenuate
PE Unique sequence PPE GXXSVPXXW virulence of recombinant tubercle bacilli49.
To better understand and/or predict the underlying
~110 aa 40–1680 aa 7–314 aa ~180 aa 0–400 aa mechanisms that drive ESX secretion, we believe that it is
PE PGRS PPE Unique sequence helpful to consider the core Ecc proteins separately from
the accessory Esp proteins. The core ESX‑1 components
Figure 2 | Domains and structures of ESX-secreted proteins. a | The structures of that are predicted to transport substrates across the inner
several ESX substrates have recently been solved, includingNature
the EsxA–EsxB
Reviews |heterodimer,
Microbiology membrane include the putative translocase EccC and at
an EspB monomer and the PE25–PPE41 heterodimer in complex with the EspG5 least three other Ecc proteins: EccB, EccD and EccE35
chaperone. b | The domain organization of Pro-Glu (PE) and Pro-Pro-Glu (PPE) proteins, (FIG. 1c). Homologues of these proteins that are encoded
many of which form heterodimers that are substrates for secretion by 6 kDa early by the esx‑5 locus were shown to form a 1,500 kDa pro-
secretory antigenic target (ESAT6) secretion (ESX) systems, comprises a conserved tein complex that could be purified from the cell enve-
amino‑terminal region that contains the PE or PPE motifs, which have approximate sizes
lope fractions of Mycobacterium marinum, which is a
of 120 amino acids and 180 amino acids, respectively, fused to diverse central and
carboxy‑terminal sequences that may contain repeated motifs (such as polymorphic
fish pathogen that is closely related to M. tuberculosis 32.
GC‑rich-repetitive sequences (PGRS)16,170 or major polymorphic tandem repeats (MPTR)), As these core components of ESX‑5 share approximately
an GXXSVPXXW motif or unique sequences. The domain organizations shown here are 30% amino acid identity with ESX‑1 paralogues, the
based on references16,30. aa, amino acids. The left image in part a is reproduced from presumed structure for the ESX‑1 membrane complex
REF. 41. The middle image in part a is reproduced with permission from REF. 58, Elsevier. might resemble that shown for ESX‑5.
The right image in part a is adapted with permission from REF. 69, Wiley. Part b is adapted New insights into the function of ESX core compo-
with permission from REF. 30, Wiley. nents were recently obtained in a study of the secre-
tion system of the thermophilic actinobacterial species
r­ egulation — which has led to a better understanding of Thermomonospora curvata, which is closely homo­
the putative roles of the individual components of ESX logous to other actinobacterial ESX systems and con-
systems and their interplay with one another. tains homologues of EsxA, EsxB, EccC, EccD, EccB and
MycP51. Structural analyses using recombinant proteins
The ESX‑1 system. The most well-known ESX‑1 from this model organism suggested that binding of the
substrates, EsxA and EsxB, have a characteristic C‑terminal signal sequence of the EsxB substrate to an
Tubercle bacilli helix–­­turn–helix hairpin organization that is due to a empty pocket of the third of three ATPase domains stim-
Mycobacteria that can cause
high α-helical content and the presence of a WXG motif. ulates multimerization of EccC51. By contrast, the addi-
tuberculosis; that is, the
M. tuberculosis complex Nuclear magnetic resonance (NMR) studies on recom- tion of EsxA to the T. curvata EccC–EsxB multimeric
(MTBC; for example, binant EsxA and EsxB proteins of M. tuberculosis that complex led to cooperative disassembly and inhibition
M. africanum, M. bovis, were expressed in Escherichia coli have revealed a 1/1 of the ATPase51, which suggests that binding of the
M. microti and M. tuberculosis) EsxA–EsxB four-helical bundle heterodimer in which substrate might modulate the assembly and activity of
and strains of the M. canettii
clade, which represents the
high-affinity binding between the two proteins is medi- ESX secretion machineries. These in vitro experiments
putative progenitor pool from ated by hydrophobic interactions13,40, and these findings assume a role for EsxB homodimers, whereas EsxA and
which the MTBC evolved. were recently confirmed by structural analyses that were EsxB have previously been described as forming a tightly

682 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Type IV secretion coupling
proteins
Proteins that recruit nucleotide
substrates to the mating pore
formation system for
subsequent DNA transfer.
Mycosin 1
(MycP1). One of a family of
subtilisin-like serine proteases
that is found in mycobacteria.
Pathogenicity-associated
genomic island
A mobile segment of the
genome, transferred by
horizontal transfer, that
contributes to rapid changes in
virulence potential.
Response regulator
A transcriptional regulator that
activates transcription of a
specific set of genes in
response to certain stimuli.
Two-component system
A regulator system that
consists of a
membrane-embedded sensor
protein and a cytoplasmic
response regulator.
Two-component systems
govern numerous cellular
activities in diverse species of
bacteria and archaea.
complexed 1/1 heterodimer 13 that might only become
dissociated outside of the bacterial cell (under appro-
priate conditions, such as low pH)52,53. Further studies
are required to decipher the exact mechanisms of EccC-
mediated translocation processes, and these may benefit
from comparisons with type IV secretion coupling proteins,
which phylogenetic analyses have shown to be related to
EccC ATPases51,54.
Another conserved protein of the esx‑1 locus of
M. tuberculosis is mycosin 1 (MycP1; FIG. 1c), which is
a subtilisin-­like serine protease with a domain archi-
tecture that comprises a canonical signal sequence
followed by a protease domain and a C‑terminal trans-
membrane region55. Deletion of MycP1 in M. tuberculo‑
sis led to the loss of ESX‑1 secretion functions, whereas
mutagenesis of the active sites and consequent loss of
protease activity resulted in increased secretion, which
suggests a putative dual role of MycP1 in substrate
processing and the regulation of ESX‑1 secretion56.
One known substrate of MycP1 is EspB, which is also
indispensable for ESX‑1 function56. Structural analyses
of EspB, which contains a YXXXD/E secretion motif,
have revealed PE–PPE-like domains, the organization of
which resembles those of the PE25–PPE41 heterodimer,
despite only limited sequence similarity between these
proteins and EspB (FIG. 2a). The PE–PPE-like domains
mediate oligomerization into what seems to be a sym-
metrical heptameric barrel or donut-shaped complex
with a central pore57,58.
Aside from EspB, numerous other Esp proteins have
been described as ESX substrates35,59–64, including EspA,
EspC and EspD (also known as Rv3616c, Rv3615c and
Rv3614c), which are encoded by an operon more than
260 kb upstream of esx‑1. This operon is of particular
interest as it has been suggested to be a pathogenicity-­
associated genomic island42,65, owing to its distribution
in the genomes of certain slow-growing, opportunis-
tic or obligate mycobacterial pathogens (M. marinum,
Mycobacterium liflandii, Mycobacterium kansasii,
Mycobacterium haemophilum, Mycobacterium leprae,
Mycobacterium lepromatosis, Mycobacterium canettii and
the different members of the M. tuberculosis complex),
in which it is often present in different chromosomal
regions and not in synteny with flanking sequences.
The presence of espACD operons in these genomes
might therefore be the result of independent horizon-
tal gene transfer events that might have contributed
to pathogenicity. EspA, EspC and EspD were the first
proteins that are not encoded by the esx‑1 locus to be
shown to be involved in ESX‑1 function59,60, and these
proteins also provided the first evidence of interdepend-
ence between ESX‑1 substrates, as the secretion of EspA
and EspC requires EsxA and EsxB and vice versa60. By
contrast, although EspD is not co‑secreted with EsxA,
complete depletion of EspD from bacterial cells results
in the loss of EsxA secretion66. EspD also has a role in the
maintenance of cellular levels of EspA that seems to be
independent of its role in EsxA secretion66.
EspA, EspC and EspD have sequence similarity to
three Esp proteins that are encoded by the esx‑1 locus:
EspE, EspF and EspH, respectively. EspF and EspH are
YXXXD/E motif-containing proteins, as are the ESX‑1
proteins EspC and EspJ. Another Esp protein in ESX‑1 is
EspG 1, which is encoded by a gene that is located
between espF and espH at the 5′ end of the esx‑1 core
locus (FIG. 1a). EspG1 shows some resemblance to EspG
proteins of other esx loci and is thought to function as
a chaperone that protects domains from aggregation or
polymerization, similarly to chaperones in other bac-
terial secretion systems11. In ESX‑1, EspG1 binds to the
PE35–PPE68 complex, which seems to also be a feature
that is conserved for the EspG homologue in ESX‑5,
EspG5, which binds to the corresponding PE–PPE
heterodimers that are secreted by ESX‑5 (REFS 67–69)
(FIG. 2a). Although this conserved substrate specificity
suggests that EspG proteins might be responsible for
selecting substrates for secretion by their corresponding
ESX systems, it should be noted that EspG proteins do
not bind to EsxA or EsxB, and therefore the secretion of
Esx proteins does not seem to generally involve selection
by EspG chaperones11. Similarly, EccA, a cytosolic AAA+
ATPase that interacts with the C terminus of EspC61,
might also be a chaperone for EspC (FIG. 1c).
In addition to regulation through co‑­dependent
export with other ESX‑1 substrates, such as the co-­
dependency between EspA or EspC and EsxA or
EsxB, ESX‑1 secretion is tightly regulated at the tran-
scriptional level by WhiB6 (also known as Rv3862c),
which is encoded by a gene that is located at the 5′ end
of the esx‑1 locus. The role of WhiB6 in regulating the
transcription of ESX‑1 components was revealed by
comparing M. tuberculosis H37Rv with clinical iso-
lates, which showed that clinical isolates expressed
ESX‑1 components at substantially higher levels than
H37Rv 70. Transcriptome analysis identified a single-­
nucleotide insertion in the promoter region of whiB6
that results in the downregulation of whiB6 expres-
sion and, consequently, of numerous esx‑1‑encoded
genes70,71. The insertion was located in the binding site
of PhoP, which, as the response regulator of the PhoP–
PhoR two-­component system, has an important role in
the regulation of virulence and immuno­genicity in
M. tuberculosis 72–74. Another key regulator of ESX‑1
that has been shown to be regulated by PhoP–PhoR is
EspR (also known as Rv3849)75,76. Although EspR was
initially described as being a substrate for, in addition to
a regulator of, ESX‑1 secretion75, a later study could not
confirm the secreted nature of EspR and provided com-
pelling evidence that EspR is a nucleotide-­associated
protein that regulates the expression of a wide range
of virulence-­associated transcripts in M. tuberculosis 76.
Indeed, among other regulatory functions, EspR was
shown to regulate transcription of espACD, which, in
turn, controls the secretion of EsxA and EsxB, owing
to the co‑dependent export of these substrates with
EspA and EspC. EspR is not the only transcriptional
regulator of espACD, as the operon has been shown
to be transcriptionally repressed by MprA–MprB (a
two-component transcriptional regulatory system
that is associated with mycobacterial persistence; also
known as Rv0981 and Rv0982), for which binding sites
have been described in the putative espACD promoter

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 683

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Haemin
An iron-containing porphyrin,
corresponding to a crystalline
chloride of haem, which is
obtained when haemoglobin
reacts with acetic acid and
sodium chloride.
684 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro
region77. Interestingly, the requirement for PhoP–PhoR secretion of PE and PPE proteins. Thus, functional anal-
in the cascade that leads to the secretion of EsxA and ysis of ESX‑5 might help to decipher the biological func-
EsxB can be bypassed through the deletion of the tions of PE and PPE proteins, which, at present, remain
genomic region upstream of espACD, which contains mysterious despite the 18 years that have elapsed since
binding sites of the different regulatory systems75,77,78,. their first identification in the M. tuberculosis H37Rv
This regulation escape mechanism was discovered by genome project16.
transcriptome analysis of tubercle bacilli that lacked Genes that encode the ESX‑3 system (rv0282–
region of difference 8 (RD8)79, which is adjacent to rv0292) are conserved among different mycobacteria,
the espACD operon80, and experimental transfer of the which suggests an important role for ESX‑3 in viabil-
espACD allele from RD8‑deletion strains into a ity for environ­mental and pathogenic mycobacterial
M. ­tuberculosis mutant that lacked PhoP–PhoR. The species24,28,92. Expression of esx‑3 genes is regulated by
expression of EspACD and the secretion of EsxA were iron and zinc, and increases during the chronic phase
restored in the recombinant strains79, which explains of infection in the lungs of mice93,94, which suggests a
the secretion of EsxA in lineages of tubercle bacilli that role for ESX‑3 in metal homeostasis27,28. Efforts to define
have fitness-diminishing mutations in PhoP–PhoR but the function of ESX‑3 in M. tuberculosis had, for many
in which RD8 has been deleted (Mycobacterium afri‑ years, been limited by an inability to recover M. tuber‑
canum lineage 6 and animal-adapted strains, such as culosis mutants that were deficient in ESX‑3 compo-
M. bovis)81–84. These findings suggest an evolutionary nents, owing to the essential requirement of some
scenario in which the emergence of fitness-diminishing esx‑3‑­encoded proteins for in vitro growth in standard
mutations in PhoP–PhoR has been compensated by the media, but this limitation was recently overcome by
deletion of RD8 in these lineages (REF. 79), which ena- supplementing the growth medium with various iron
bles the network that is regulated by PhoP–PhoR to be complexes, such as haemin29. Subsequently, the use of
bypassed and thus restores the secretion of EsxA. deletion mutants that lacked PE5–PPE4, EsxG (also
known as Rv0287) or EsxH (also known as Rv0288
Other ESX systems. ESX‑5 is the most recent ESX sys- or TB10.4; FIG. 1a) in proteomic analy­sis screens and
tem to emerge in evolution and is present only in slow-­ mouse infection experiments revealed that PE5–PPE4
growing mycobacteria3 (BOX 2). The eccB5–eccC5 region has an essential role in the ­siderophore-mediated iron-­
of the esx‑5 locus in M. tuberculosis (rv1782–rv1798; acquisition functions of ESX‑3 (REF.  29). EsxH and
FIG. 1a) is essential for viability under standard in vitro EsxG, which have been described as being secreted by
growth conditions85. However, this essential role can be ESX‑3 as a 1/1 hetero­dimer 95, have also been shown
circumvented through the deletion of the gene cluster to be required for iron metabolism; however, this role
that is responsible for the biosynthesis of phthiocerol might be indirect, through promoting the secretion
dimycocerosates, which are lipids that are found in the of PE5–PPE4 (REF. 29). Indeed, EsxG–EsxH seems to
mycobacterial cell envelope of certain slow-growing be a more general regulator of secretion rather than a
mycobacteria, or through the heterologous expression specific regulator of iron metabolism, as shown by its
of MspA-like porins, which are not naturally present in iron-­independent role in virulence that is thought to be
the outer membranes of slow-­growing myco­bacteria, mediated by promoting the ESX‑3-mediated secretion
from Mycobacterium smegmatis 86. Based on these data, of PE15–PPE20 (REF. 29). Furthermore, a role for EsxG–
a model was proposed in which the ESX‑5 system is EsxH in host–­pathogen interactions is also suggested
responsible for the export of cell envelope proteins that by the findings that EsxH strongly induces interferon-γ
are required for nutrient uptake86, as a mechanism (IFNγ) secretion in T cells from patients with tubercu-
that compensates for the lack of MspA-like porins in losis and BCG-vaccinated donors96, and in CD4+ and
slow-growing mycobacteria86. Such a function is also CD8+ T cells from mice infected with M. tuberculosis 97,98,
supported by the finding that ESX‑5 seems to be regu- and that EsxH interacts with, and inhibits, the endo­
lated at a transcriptional level by the Pst–SenX3–RegX3 somal sorting complex required for transport (ESCRT),
system, which comprises a phosphate uptake component which has important roles in membrane trafficking in
and a two-component regulatory system that collectively mammalian host cells99. Finally, EsxH is a potent protec-
control gene expression in response to the availability tive antigen when used as a subunit vaccine fused with
of phosphate87. M. tuberculosis antigen 85B (a recognized T cell anti-
The role of ESX‑5 in the secretion of mycobacterial gen that is encoded by rv1886c) and formulated with
proteins has been studied extensively using the model an appropriate adjuvant 100. Together, these observations
mycobacterium M. marinum88–90. The first experimen- have led to a growing appreciation of the importance
tal evidence of ESX‑5 secretion by M. marinum was of ESX‑3 in the infection process, although the roles of
obtained by cloning the full esx‑5 locus of this species ESX‑1 and ESX‑5 in the pathogenesis of M. tuberculosis
into M. smegmatis, which resulted in the efficient secre- currently remain better characterized.
tion of heterologously expressed PPE41 (REF. 88). Later, in In contrast to ESX‑1, ESX‑5 and ESX‑3, the
M. marinum and M. tuberculosis, ESX‑5 was shown to be functions of ESX‑2 (rv3895c–rv3884c) and ESX‑4
responsible for the export of a large number of PPE and (rv3444c–rv3450c) are wholly unknown. High-density
PE proteins of diverse subtypes (FIG. 2b), many of which transposon screens have demonstrated that neither
were not encoded by the esx‑5 locus33,88,89,91, which sug- of these systems seems to be required for in  vitro
gests that export by ESX‑5 is a major pathway for the growth nor for virulence of M. tuberculosis in mice,
©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.

Pan-genome
The full set of genes of a
defined species of bacteria or
archaea, which comprises all of
the genes that are collectively
found in individual genomes of
the species.
Pathogen-associated
molecular patterns
(PAMPs). Conserved molecular
structures that are produced
by microorganisms and that
are recognized as foreign by
the receptors of cells of the
innate immune system.
NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 685
REVIEWS
and proteomic approaches have not been able to detect and the activation of associated defence mechanisms
active secretion of the putative ESX‑4 substrates EsxU (FIG. 3) . The development and use of a fluorescence
or EsxT, which leaves an uncertainty as to the function- resonance energy transfer (FRET)-based phagosomal
ality of ESX‑4 in M. tuberculosis 24,90,101. Interestingly, rupture assay, which is based on cleavage of coumarin–­
ana­lysis of the genomes of M. leprae and M. lepro‑ cephalosporin–­fluorescein (CCF) by the mycobacterial
matosis, which have been streamlined into minimal β-lactamase42, enabled cytosolic contact of M. tubercu‑
genomes by a process of extensive genome decay 102,103, losis to be demonstrated in various phagocyte types,
showed that these pathogens have lost the esx‑4 locus with contact detectable using a flow cyto­metry readout
and have non-functional ESX‑2 systems103. Together, as early as three hours post-­infection115. This technique
these observations suggest that the functions of ESX‑2 can reliably distinguish between strains that establish
and ESX‑4 in mycobacteria may be non-essential and cytosolic contact and strains that remain enclosed in
not host-oriented. an intact phagosome; however, it cannot provide a
more precise measurement of the subcellular location
ESX systems in pathogenesis of the pathogen42,114. Studies of phagosomal rupture by
Mycobacterial ESX systems have been known to be mycobacteria have also demonstrated a link between
associated with pathogenicity and host–pathogen inter- the ability of a strain to establish ESX‑1‑mediated cyto-
actions since their discovery 19–23,32,33,88,89,104,105, although solic contact and the induction of host cell death42,43.
pan-genome analyses show that ESX systems are also Notably, the absence of most of the esx‑1 locus in the
widely present in non-pathogenic actinobacterial BCG vaccine strain has been shown to underlie
­species4. ESX systems are thus thought to have diverse the consequent inability of this strain to establish cyto-
functions that may include roles in cellular physiology solic contact, which is fundamental to the attenuation
and horizontal gene transfer between strains. These roles of virulence that is observed for BCG compared with
may be linked to survival in different environments, virulent M. bovis or M. tuberculosis 21,42. The role of
and in some cases have been adapted by certain patho­ ESX‑1 in phago­somal rupture seems to be mediated
genic bacteria to survive under the harsh conditions by the secreted substrate EsxA, which has been shown
that they can encounter in the host. The ESX systems to be able to disrupt artificial lipid b ­ ilayers21 and the
that have been best characterized as associated with plasma membranes of red blood cells116. Furthermore,
host–­pathogen interactions are ESX‑1 and ESX‑5. EsxA, but not EsxB, lyses liposomes52. The two helices
of EsxA each measure around 50 Å in length, which
Phagosomal rupture by ESX‑1. ESX‑1 is the most inten­ corresponds approximately to the thickness of biologi-
sively studied ESX system, which is probably owing cal membranes and thus suggests that the helix–turn–
to its important roles in virulence and other host–­ helix motif of EsxA might form insertions that span the
pathogen interactions. The biological implications phagosome membrane117.
of ESX‑1‑mediated host–pathogen interactions are The biological and immunological repercussions of
remarkable. Furthermore, the presence of a functional contact between mycobacteria and the host cell cyto-
ESX‑1 system is essential for virulence in the obligate sol are numerous and represent an exciting emerg-
human pathogens M. tuberculosis and M. leprae, as well ing field of research into host–pathogen interactions.
as in the facultative human pathogens M. marinum and Possible interactions include those between bacterial
M. kansasii. Mutant mycobacterial pathogens that lack pathogen-­associated molecular patterns (PAMPs) and
ESX‑1 have defects that include attenuated pathogenicity cytosolic sensors in the host cell, similarly to the role of
and decreased intracellular replication23,104,106 compared various cytosolic receptors of the host immune system
with wild-type strains. in detecting molecular components (such as DNA) of
Following phagocytosis by a macrophage, patho- other intracellular pathogens. Indeed, three independ-
genic mycobacteria interfere with phagosome matura­ ent studies have described a central role for the cytosolic
tion and phagolysosome biogenesis to facilitate their DNA sensor nucleotidyltransferase cyclic GMP–AMP
survival 107, which M.  tuberculosis has been shown synthase (cGAS) in the detection of mycobacterial
to do by various means108, such as by inhibiting the DNA that has entered the cytosolic compartment 118–120.
­proton–ATPase activity that is responsible for phago- The precise pathways by which DNA is released from
somal acidification109. ESX‑1 is required for subsequent live M. tuberculosis into the cytosol remain unknown
phagosomal rupture, as was first reported in M. mari‑ but may involve the formation of extracellular vesicles
num 110. Later studies that used cryo-electron micro­ that are secreted from the mycobacterial membrane
scopy suggested that, during the later stages of infection, or export by secretion systems, although mechanis-
intact cells of ESX‑1‑positive strains of M. tuberculosis tic insights are still lacking 65,121. The sensing of DNA
and M. leprae may translocate into the cytosol of the by cGAS leads to the synthesis of the second mes-
infected phagocyte111. These observations challenged senger cyclic GMP–AMP (cGAMP), which activates
the long-standing view that pathogenic mycobacte- the endoplasmic reticulum‑associated stimulator of
ria remain exclusively enclosed in the phagosome112, interferon genes (STING) and the downstream serine/
and, as such, are subject to ongoing scientific discus- threonine-­protein kinase (TBK1)–interferon regulatory
sion 113,114. Following phagosomal rupture, contact factor 3 (IRF3)–IFNβ signalling pathway (FIG. 3). This
with bacterial components, or perhaps the bacterium process leads to the expression of type I IFNs, such as
itself, leads to sensing by the innate immune system IFNβ, which is a strategic event in the interactions of
©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

◀ Figure 3 | Downstream effects of phagosomal

EsxA–EsxB membrane rupture by ESX‑1. A fundamental role of
6 kDa early secretory antigenic target (ESAT6) secretion
ESX-1 system 1 (ESX‑1) in the infection of macrophages by
Mycobacterium tuberculosis is the membrane rupture of
M. tuberculosis
the phagosome in which mycobacteria reside following
phagocytosis. Rupture of the phagosomal membrane
leads to contact between mycobacterial components,
or perhaps even intact mycobacteria, and the host cell
Mycobacterium cytosol. Recent research has suggested that membrane
tuberculosis rupture is mediated by the membranolytic action of
EsxA, which is secreted in a heterodimeric complex with
EsxB into the phagosomal lumen. Subsequent detection
Phagosomal of the bacterium and bacterial-derived molecules (such
rupture
by ESX-1 as DNA) by sensors of the host innate immune system
has important consequences for signalling in the
infected cell. DNA is sensed by cyclic GMP–AMP
Mycobacterial DNA synthase (cGAS), which synthesizes the second
Autophagosomal
membrane ? AIM2 messenger cyclic GMP–AMP (cGAMP) from ATP and
LC3-II IL-1β GTP. cGAMP, which can also disseminate to bystander
NLRP3 and/or
IL-18 non-infected cells, activates the endoplasmic reticulum
Ub Inflammasome (ER)‑associated stimulator of interferon genes (STING)
activation
pathway and the downstream serine/threonine-protein
Inflammasome

kinase (TBK1)–interferon regulatory factor 3 (IRF3)

signalling axis, which leads to the production of
interferon-β (IFNβ). Activation of TBK1 and bacterial
Activated
caspase 1 ubiquitylation may initiate selective autophagy of the
cGAS pathogen through recruitment of the autophagosome-­
Pro-caspase 1
Pro-IL-1β associated lipidated isoform of ­microtubule-associated
ATP and/or protein 1 light chain 3 (LC3‑II) to ubiquitylated
cGAMP
GTP pro-IL-18 mycobacteria and/or phagosomal membrane fragments
(BOX 3). Mycobacterial DNA in the host cytosol can also
be sensed by absent in melanoma 2 (AIM2), which
STING
partially contributes to the activation of the NOD-, LRR-
TBK1 and pyrin domain-containing 3 (NLRP3)–inflammasome,
Neighbouring
host cell leading to the activation of caspase 1 and cleavage of
P P ER the precursor of the protective cytokine interleukin‑1β
(IL‑1β) followed by secretion of mature IL‑1β. The
IRF3
mechanisms shown in this figure are based on those
P P described in REFS 118–120,171,172. Similarly, in CD11c+
IFNβ immune cells (such as dendritic cells), NLRP3–caspase 1
activation was also shown to induce the maturation of
pro‑IL‑18 into IL‑18, which induces the production of
Nucleus IFNγ in CD8+ T cells and natural killer (NK) cells44.
Infected host cell Ub, ubiquitin.

M. tuberculosis with the host, as type I

Inflammasome
A multiprotein signalling
platform that controls the
inflammatory response and
coordinates antimicrobial host
defences through the
activation of caspase 1, which
induces the processing of
pro-interleukin‑1ß and
pro-interleukin‑18 into mature
cytokines to be released from
the cell.
Nature IFNs
Reviews are thought
| Microbiology
to be detrimental to the host in the context of mycobac-
terial infection. Indeed, infection experiments in mice
revealed a decrease in bacterial load when type I IFN
signalling was abrogated122,123 and patients with active
tuberculosis have an upregulated type I IFN transcrip-
tional profile that is correlated with disease severity and
that is downregulated following successful treatment 124.
Aside from cGAS, mycobacterial DNA is also sensed by
absent in melanoma 2 (AIM2), which leads to the acti-
vation of NOD-, LRR- and pyrin domain-­containing 3
(NLRP3)–­inflammasome signalling 125–127 and, in turn,
increased activation and secretion of protective
cytokines, such as interleukin‑1β (IL‑1β)126 (FIG. 3). We
note that M. tuberculosis had previously been reported
to prevent inflammasome activation128, but this finding
was probably due to the use of the BCG strain, which
lacks ESX‑1, in most experiments125,126,128.
ESX‑1‑mediated cytosolic contact is also essential
for NLRP3–inflammasome-mediated secretion of IL‑18
by infected CD11c+ immune cells (such as dendritic
cells), which leads to the rapid, non-cognate produc-
tion of IFNγ by M. tuberculosis-antigen-independent
memory CD8+ T cells and natural killer (NK) cells, as
was recently shown in various M. tuberculosis infection
mouse models44 (FIG. 3). Similarly, cytosolic access of
mycobacteria is instrumental for the accumulation of the
lipidated isoform of microtubule-associated protein 1
light chain 3 (LC3‑II) in dendritic cells129, which is
an isoform that is associated with autophagosomes.
However, the exact role of autophagy during infection
with M. tuberculosis is a matter of debate (BOX 3).
Finally, the effects of ESX‑1 on host–pathogen inter-
actions also concern granuloma formation and bacterial
dissemination. For example, granuloma formation that
is induced by M. marinum in zebrafish larvae was found

686 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

Granuloma
Organized immune cell
aggregates that form in
response to selected stimuli
of an infectious or
non-infectious nature.
Caspase-independent cell
death
Programmed host cell death
that is independent of caspase
activation. In the context of
mycobacterial infection,
caspase-independent cell
death suggests the
involvement of
non-endogenous factors, such
as bacterial effectors, in
mediating host cell death.
Severe combined
immunodeficiency
(SCID). A severe genetic
disorder that is characterized
by the absence of functioning T
lymphocytes and B
lymphocytes.
PE_PGRS proteins
Products of a mycobacterial
multigene family that is
characterized by polymorphic
GC-rich repetitive sequence
(PGRS) motifs. PE_PGRS
proteins have a highly
conserved amino-terminal
domain of ~110 amino acids,
which contains the Pro-Glu (PE)
motif, and variable
carboxy-terminal domains that
contain numerous repetitive
sequence motifs.
to depend on a functional ESX‑1 system130,131. In the
same model, ESX‑1 was also required for the induction
of matrix metalloproteinase 9 (MMP9) in epithelial cells
that were proximal to infected macrophages. The induc-
tion of MMP9 enhanced the recruitment of macro­phages
to the granuloma site, which contributed to nascent
granuloma maturation and bacterial growth132. Thus, it
is plausible that ESX‑1‑mediated induction of necrotic
and/or apoptotic host cell death42,133 might contribute to
granuloma formation and/or cell‑to‑cell spread, which
are generally not observed during infection with BCG or
other ESX‑1‑deficient strains.
Evidence for ESX‑5‑mediated pathogenesis. ESX‑5 was
the second ESX system to be shown to have an important
role in host–pathogen interactions. The first evidence for
such a role was obtained from studies on M. marinum,
which showed that ESX‑5 was required for the induction
of caspase-independent cell death and spread of infection,
but was dispensable for intra­cellular growth31,88. ESX‑5
was found to modulate innate immune responses90,134 by
inhibiting induction of the production of key cytokines
in infected type  I macrophages in human blood 90.
Similarly to ESX‑1 (REF. 135), ESX‑5 contributes to the
activation of the inflammasome and the secretion of
IL‑1β31,90. However, unlike ESX‑1 (REFS 42,43), no evi-
dence for a role of ESX‑5 in phagosomal rupture, which
would lead to cytosolic contact of the mycobacterium,
was found31.
Detailed investigation of the roles of the ESX‑5 sys-
tem and its PE and PPE substrates in M. tuberculosis
virulence, immunogenicity and host–pathogen inter-
actions have relied on M. tuberculosis deletion mutants
and complemented strains32,33,105. In vivo infection mod-
els showed that an M. tuberculosis deletion mutant that
lacked five esx‑5‑encoded PE and PPE proteins (PPE25,
PPE26, PPE27, PE18 and PE19) showed attenuated viru-
lence in both a mouse model of severe combined immuno-
deficiency (SCID) and in immune-competent C57BL/6
mice33,105, and an M. tuberculosis deletion mutant that
lacked EccD5 displayed a small colony phenotype and
decreased intracellular growth capacity in cell culture
and mouse models33,105. Importantly, the reduced vir-
ulence of these two strains was not due to any obvi-
ous interference with the production and/or secretion
of the ESX‑1‑associated virulence factor EsxA or
ESX‑3‑associated EsxH105. Therefore, ESX‑5‑associated
PE and PPE proteins and EccD5 seem to have essen-
tial roles in the pathogenic potential of M. tuberculo‑
sis, although the mechanisms by which these roles are
mediated are currently unknown. The ESX‑5 system in
M. tuberculosis has been shown to be involved in the
export of a plethora of PE and PPE proteins that are
encoded outside of the esx‑5 locus33,105, including PE_
PGRS proteins32. For example, the M. tuberculosis deletion
mutant that lacked five esx5‑encoded PE and PPE pro-
teins (see above) remained able to induce T cell immu-
nity against a large number of PE and PPE proteins
that are encoded outside of the esx5 locus. By contrast,
the export of these PE and PPE proteins was abolished
in the M.  tuberculosis eccD 5-deletion mutant 105, in
accordance with reports that described EccD5 as a con-
served component of the ESX‑5 secretion machinery 32.
Analysis of individual ESX‑5‑exported PE and PPE pro-
teins is ongoing and may reveal novel functions, such
as the recent demonstration that PPE10 is essential for
the integrity and stability of the myco­bacterial cap-
sule136. Finally, deletion of the ESX‑5 component EspG5
resulted in hypervirulence of M. marinum in zebrafish,
which suggests that ESX‑5‑mediated secretion might

Box 3 | How important is autophagy following phagosome rupture by ESX‑1?

6 kDa early secretory antigenic target (ESAT6) secretion system 1 (ESX‑1)‑mediated phagosomal rupture is instrumental
for the cytosolic contact that is required for mycobacteria and/or mycobacterial products to recruit the lipidated
isoform of microtubule-­associated protein 1 light chain 3 (LC3‑II)129,152, which is associated with autophagosomes.
Selective autophagy, in which the ubiquitin-mediated targeting of intracellular pathogens to autophagosomes is
followed by delivery to the lysosome, is an evolutionarily conserved process that is used by eukaryotes to sterilize the
cytosol152,153. Ruptured membranes of Mycobacterium tuberculosis-containing phagosomes can usually be labelled with
antibodies against ubiquitin, which suggests that these components might represent a potential substrate for
autophagic processes43. Selective autophagy was first shown to be a mechanism of the innate immune system
for infections with group A Streptococcus153, with later studies suggesting that autophagy has a similar role during
infections with M. tuberculosis154. Different pathogen-derived ligands can trigger antibacterial autophagy, including
sugar molecules, which are recognized by the host protein galectin 8 (REF. 155), or double-stranded DNA, which is
detected by the stimulator of interferon genes (STING) pathway152. Indeed, a study that showed that mycobacterial DNA
may interact with components of the STING pathway found that M. tuberculosis cells were subsequently ubiquitylated
for selective autophagy152. This study also found that mouse cells that lacked the essential autophagy factor autophagy
protein 5 (Atg5) were highly sensitive to infection with M. tuberculosis, which led the authors to conclude that
Atg5‑mediated autophagy might be a crucial innate immune mechanism against mycobacteria152. However, a function
for Atg5 that is essential for cellular immunity to intracellular pathogens but independent of autophagosomes has
previously been reported156. Furthermore, a recent study concluded that the increased susceptibility of Atg5‑deficient
mice to infection with M. tuberculosis could be attributed to altered neutrophil recruitment and associated immune
system pathologies, rather than the impairment of autophagy. This conclusion was based on experiments in which mice
that had substantial autophagy defects, owing to germline deletions in the autophagy pathway (Ulk1, Ulk2, Atg4b or
p62) or cell lineage-specific deletions of other essential autophagy genes (Atg14l, Atg12, Atg16l, Atg7 or Atg3), were not
more sensitive to infection with M. tuberculosis than control mice157. Thus, the results of this study seriously question the
previously postulated importance of autophagy in the control of mycobacterial infections.

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 687

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

Box 4 | ESX systems in therapeutics and disease prevention

Vaccines that are more effective and new therapeutic options will both be required if the long-standing global health goal of
eradicating tuberculosis is to be achieved. Currently, Mycobacterium bovis bacille Calmette–Guérin (BCG) is the only
licensed anti-tuberculosis vaccine. However, although BCG has an excellent safety record and efficiently prevents serious
forms of childhood tuberculosis (that is, meningitis and miliary tuberculosis), it generally confers only poor protection
against the most contagious form of tuberculosis, which is pulmonary tuberculosis, in adolescents and adults. The limitations
of the BCG vaccine are further enhanced by the heterogeneity of BCG strains, each of which has evolved in vitro in different
institutions worldwide158,159. However, one common feature of all BCG strains is the absence of 6 kDa early secretory
antigenic target (ESAT6) secretion system 1 (ESX‑1), owing to the deletion of region of difference 1 (RD1). However, the
presence of a functional ESX‑1 system may improve the efficacy of live-attenuated vaccine strains. For example, vaccination
with a recombinant strain of BCG that expresses ESX‑1 has been shown to increase protection against disseminated
tuberculosis in mouse and guinea pig models20. Although BCG recombinants that have the intact esx‑1 locus of pathogenic
Mycobacterium tuberculosis are more virulent than BCG in the severe combined immunodeficiency (SCID) mouse model of
disease19, and thus are associated with vaccine safety concerns, the generation of BCG recombinant strains that have
mutations in the esxA gene may resolve this limitation160. Similarly, engineering BCG strains for heterologous expression of
ESX‑1 proteins that are from less virulent mycobacteria than M. tuberculosis also has promising potential for increasing
vaccine efficacy without compromising vaccine safety161. Aside from ESX‑1, another ESX system that might beneficially be
applied to vaccine design is ESX‑3. The high-dose intravenous administration of recombinant Mycobacterium smegmatis that
expressed the ESX‑3 system from M. tuberculosis, but not endogenous ESX‑3, was effective in protecting mice from
challenge with M. tuberculosis162. Another putative vaccine candidate is the attenuated M. tuberculosis Δppe25–pe19 strain,
which lacks the ESX‑5‑encoded Pro-Glu (PE) and Pro-Pro-Glu (PPE) family proteins PE18, PE19, PPE25, PPE26 and PPE27
(REF. 33). This strain, which retains normal ESX‑1 functions, shows improved protective capacity against infections with
M. tuberculosis in mice, compared with BCG105, and has recently enabled the characterization of the significant contribution
of PE‑specific and PPE-specific T helper 1 (TH1) effector cells in anti-mycobacterial protective immunity163.
In parallel to improved vaccines, the identification of new therapeutic agents is also urgently required to cope with the
increasing emergence of drug-resistant M. tuberculosis strains. A recent report that showed the extent of acquired drug
resistance, including resistance to two novel drugs for the treatment of tuberculosis (Bedaquiline and Delamanid)164,
underlines the need to complement current therapies that are used to fight tuberculosis with alternative strategies. In this
respect, compounds that target bacterial virulence factors might conceivably provide additional therapeutic possibilities.
For the intracellular pathogen M. tuberculosis, screens that test antimycobacterial drug efficacy inside host cells have the
potential to identify such molecules165–167. One of the compounds that already has been identified, BTP15, indirectly inhibits
ESX‑1 secretion by targeting the two-component regulatory system MprA–MprB166, which regulates the espACD operon77.
Another molecule, BBH7, increases outer membrane permeability, as shown by increased ethidium bromide uptake166. RNA
sequencing showed that infection with M. tuberculosis was associated with the upregulation of genes with functions in
metal-ion metabolism, which results in metal-ion toxicity and upregulation of components of the ESX‑1 secretion system.
However, ESX‑1 secretion is blocked by BBH7 and thus bacteria remain trapped inside phagosomes166.

help to establish a moderate and persistent infection134,
although further studies are required to understand the
­mechanistic basis for this phenotype.
Conclusions and outlook
The nearly 20 years of research that have followed the
discovery of ESX systems in M. tuberculosis have greatly
contributed to our understanding of the biological func-
tions of selected ESX systems in mycobacterial physiology
and, even more importantly, the effects of these systems
on various aspects of host–pathogen interactions. The
original finding that EsxA and EsxB were prototype pro-
teins for a very large family of then unknown, small, WXG
motif-containing proteins opened a new field of research,
which has produced numerous insights into diverse
aspects of bacterial physiology and infection biology.
Notably, the study of these proteins has led to the descrip-
tion of functions for ESX‑1, ESX‑3 and ESX‑5 systems.
Crucially for our understanding of host–pathogen inter-
actions, ESX‑1 systems, in conjunction with horizontally
acquired espACD loci, have been shown to mediate the
rupture of phagosomal membranes inside host phago-
cytes, whereas the function of the widely distributed
ESX‑3 system has been described as metal ion homeo-
stasis and ESX‑5 has been shown to secrete an extremely
large and diverse set of PE and PPE proteins. Recent
pan-genome analy­ses have further elucidated the prob-
able evolutionary history of ESX systems and suggested
an important role for plasmids in the radiation of esx loci
across mycobacterial species. Furthermore, structural
biology studies have started to solve and interpret several
of the protein complexes that are involved in ESX secre-
tion. However, the biology and structure of ESX secretion
systems nevertheless remain far less well understood than
those of the type I–VI secretion systems that are found in
Gram-negative bacteria. Solving the structure of the ESX
complex in the mycobacterial cell envelope will be key to
obtaining a more comprehensive mechanistic understand-
ing of observations such as the interdependence of certain
secreted substrates and questions such as whether secreted
substrates might also function as structural elements of
ESX systems. Such insights are also expected to better
explain the roles of ESX systems in mycobacterial biology.
As the activities of certain ESX systems have been reported
to have profound effects on host–pathogen interactions,
advances in the field may also help to inform improved
vaccine design and alternative therapeutic intervention
opportunities (BOX 4). In summary, we expect that ESX
systems will remain a highly animated field of research,
with many exciting features still to be discovered.

688 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

1. Sorensen, A. L., Nagai, S., Houen, G., Andersen, P. &
Andersen, A. B. Purification and characterization of a
low-molecular-mass T‑cell antigen secreted by
Mycobacterium tuberculosis. Infect. Immun. 63,
1710–1717 (1995).
2. Brodin, P., Rosenkrands, I., Andersen, P., Cole, S. T. &
Brosch, R. ESAT‑6 proteins: protective antigens and
virulence factors? Trends Microbiol. 12, 500–508
(2004).
3. Gey Van Pittius, N. C. et al. The ESAT‑6 gene
cluster of Mycobacterium tuberculosis and other
high G+C Gram-positive bacteria. Genome Biol. 2,
RESEARCH0044 (2001).
4. Dumas, E. et al. Mycobacterial pan-genome
analysis suggests important role of plasmids in
the radiation of type VII secretion systems.
Genome Biol. Evol. 8, 387–402 (2016).
This study provides the first evidence that
plasmids seem to have had important roles in the
evolution of mycobacterial ESX systems and,
consequently, in the pathogenicity of extant
mycobacterial pathogens.
5. Garufi, G., Butler, E. & Missiakas, D. ESAT‑6‑like
protein secretion in Bacillus anthracis.
J. Bacteriol. 190, 7004–7011 (2008).
6. Baptista, C., Barreto, H. C. & Sao-Jose, C. High levels
of DegU‑P activate an Esat‑6‑like secretion system in
Bacillus subtilis. PLoS ONE. 8, e67840 (2013).
7. Huppert, L. A. et al. The ESX system in Bacillus subtilis
mediates protein secretion. PLoS ONE. 9, e96267
(2014).
8. Burts, M. L., Williams, W. A., Debord, K. &
Missiakas, D. M. EsxA and EsxB are secreted by an
ESAT‑6‑like system that is required for the
pathogenesis of Staphylococcus aureus infections.
Proc. Natl Acad. Sci. USA 102, 1169–1174
(2005).
9. Burts, M. L., DeDent, A. C. & Missiakas, D. M.
EsaC substrate for the ESAT‑6 secretion pathway
and its role in persistent infections of Staphylococcus
aureus. Mol. Microbiol. 69, 736–746 (2008).
10. Way, S. S. & Wilson, C. B. The Mycobacterium
tuberculosis ESAT‑6 homologue in Listeria
monocytogenes is dispensable for growth in vitro and
in vivo. Infect. Immun. 73, 6151–6153 (2005).
11. Houben, E. N., Korotkov, K. V. & Bitter, W. Take five —
type VII secretion systems of mycobacteria.
Biochim. Biophys. Acta 1844, 1707–1716 (2014).
12. Daffe, M., Crick, D. C. & Jackson, M. Genetics of
capsular polysaccharides and cell envelope (glyco)
lipids. Microbiol. Spectr. 2, MGM2‑0021‑2013
(2014).
13. Renshaw, P. S. et al. Structure and function of the
complex formed by the tuberculosis virulence factors
CFP‑10 and ESAT‑6. EMBO J. 24, 2491–2498
(2005).
14. Pallen, M. J. The ESAT‑6/WXG100 superfamily — and a
new Gram-positive secretion system? Trends Microbiol.
10, 209–212 (2002).
15. Berthet, F. X., Rasmussen, P. B., Rosenkrands, I.,
Andersen, P. & Gicquel, B. A. Mycobacterium
tuberculosis operon encoding ESAT‑6 and a novel low-
molecular-mass culture filtrate protein (CFP‑10).
Microbiology 144, 3195–3203 (1998).
16. Cole, S. T. et al. Deciphering the biology of
Mycobacterium tuberculosis from the complete
genome sequence. Nature 393, 537–544 (1998).
17. Tekaia, F. et al. Analysis of the proteome of
Mycobacterium tuberculosis in silico. Tuber. Lung Dis.
79, 329–342 (1999).
18. Newton-Foot, M., Warren, R. M., Sampson, S. L., van
Helden, P. D. & Gey van Pittius, N. C. The plasmid-
mediated evolution of the mycobacterial ESX (type VII)
secretion systems. BMC Evol. Biol. 16, 62 (2016).
19. 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. 46,
709–717 (2002).
20. Pym, A. S. et al. Recombinant BCG exporting ESAT‑6
confers enhanced protection against tuberculosis.
Nat. Med. 9, 533–539 (2003).
21. Hsu, T. et al. The primary mechanism of attenuation of
bacillus Calmette–Guérin is a loss of secreted lytic
function required for invasion of lung interstitial tissue.
Proc. Natl Acad. Sci. USA 100, 12420–12425
(2003).
22. Lewis, K. N. et al. Deletion of RD1 from
Mycobacterium tuberculosis mimics bacille Calmette–
Guerin attenuation. J. Infect. Dis. 187, 117–123
(2003).
23. Stanley, S. A., Raghavan, S., Hwang, W. W. & Cox, J. S.
Acute infection and macrophage subversion by
Mycobacterium tuberculosis require a specialized
secretion system. Proc. Natl Acad. Sci. USA 100,
13001–13006 (2003).
24. Sassetti, C. M. & Rubin, E. J. Genetic requirements for
mycobacterial survival during infection. Proc. Natl
Acad. Sci. USA 100, 12989–12994 (2003).
25. Mahairas, G. G., Sabo, P. J., Hickey, M. J., Singh, D. C.
& Stover, C. K. Molecular analysis of genetic
differences between Mycobacterium bovis BCG and
virulent M. bovis. J. Bacteriol. 178, 1274–1282
(1996).
26. Brodin, P. et al. Bacterial artificial chromosome-based
comparative genomic analysis identifies
Mycobacterium microti as a natural ESAT‑6 deletion
mutant. Infect. Immun. 70, 5568–5578 (2002).
27. Serafini, A., Boldrin, F., Palu, G. & Manganelli, R.
Characterization of a Mycobacterium tuberculosis
ESX‑3 conditional mutant: essentiality and rescue by
iron and zinc. J. Bacteriol. 191, 6340–6344 (2009).
28. Siegrist, M. S. et al. Mycobacterial ESX‑3 is required
for mycobactin-mediated iron acquisition. Proc. Natl
Acad. Sci. USA 106, 18792–18797 (2009).
29. Tufariello, J. M. et al. Separable roles for
Mycobacterium tuberculosis ESX‑3 effectors in iron
acquisition and virulence. Proc. Natl Acad. Sci. USA
113, E348–E357 (2016).
30. Bottai, D. & Brosch, R. Mycobacterial PE, PPE and ESX
clusters: novel insights into the secretion of these most
unusual protein families. Mol. Microbiol. 73, 325–328
(2009).
31. Abdallah, A. M. et al. Mycobacterial secretion systems
ESX‑1 and ESX‑5 play distinct roles in host cell death
and inflammasome activation. J. Immunol. 187,
4744–4753 (2011).
32. Houben, E. N. et al. Composition of the type VII
secretion system membrane complex. Mol. Microbiol.
86, 472–484 (2012).
A paper that describes the first insights into the
components that form the ESX secretion apparatus
of ESX systems.
33. Bottai, D. et al. Disruption of the ESX‑5 system of
Mycobacterium tuberculosis causes loss of PPE
protein secretion, reduction of cell wall integrity and
strong attenuation. Mol. Microbiol. 83, 1195–1209
(2012).
34. Abdallah, A. M. et al. Type VII secretion system of
mycobacteria show the way. Nat. Rev. Microbiol. 5,
883–891 (2007).
35. Bitter, W. et al. Systematic genetic nomenclature for
type VII secretion systems. PLoS Pathog. 5, e1000507
(2009).
36. Desvaux, M., Hebraud, M., Talon, R. &
Henderson, I. R. Secretion and subcellular localizations
of bacterial proteins: a semantic awareness issue.
Trends Microbiol. 17, 139–145 (2009).
37. Bitter, W., Houben, E. N., Luirink, J. & Appelmelk, B. J.
Type VII secretion in mycobacteria: classification in line
with cell envelope structure. Trends Microbiol. 17,
337–338 (2009).
38. Costa, T. R. et al. Secretion systems in Gram-negative
bacteria: structural and mechanistic insights. Nat. Rev.
Microbiol. 13, 343–359 (2015).
39. Gey van Pittius, N. C. et al. Evolution and expansion of
the Mycobacterium tuberculosis PE and PPE
multigene families and their association with the
duplication of the ESAT‑6 (esx) gene cluster regions.
BMC Evol. Biol. 6, 95 (2006).
40. Brodin, P. et al. Functional analysis of early secreted
antigenic target‑6, the dominant T‑cell antigen
of Mycobacterium tuberculosis, reveals key
residues involved in secretion, complex formation,
virulence, and immunogenicity. J. Biol. Chem. 280,
33953–33959 (2005).
41. Poulsen, C., Panjikar, S., Holton, S. J., Wilmanns, M.
& Song, Y. H. WXG100 protein superfamily consists of
three subfamilies and exhibits an α-helical C‑terminal
conserved residue pattern. PLoS ONE. 9, e89313
(2014).
42. Simeone, R. et al. Phagosomal rupture by
Mycobacterium tuberculosis results in toxicity and
host cell death. PLoS Pathog. 8, e1002507 (2012).
A paper that confirms and consolidates previous
controversial findings on M. tuberculosis-induced
phagosomal rupture, using human macrophage-like
THP‑1 cells and BCG strains that express the
M. tuberculosis ESX‑1 system.
43. Houben, D. et al. ESX‑1‑mediated translocation to
the cytosol controls virulence of mycobacteria.
Cell. Microbiol. 14, 1287–1298 (2012).
44. Kupz, A. et al. ESAT‑6‑dependent cytosolic pattern
recognition drives noncognate tuberculosis control
in vivo. J. Clin. Invest. 126, 2109–2122 (2016).
A study that reveals the requirement for
phagosomal contact and inflammasome activation
for the induction of IL‑18‑mediated IFNγ production
in CD8+ T cells and NK cells.
45. 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 313, 1632–1636 (2006).
46. Daleke, M. H. et al. General secretion signal for the
mycobacterial type VII secretion pathway. Proc. Natl
Acad. Sci. USA 109, 11342–11347 (2012).
47. Ates, L. S., Houben, E. N. & Bitter, W. Type VII
secretion: a highly versatile secretion system.
Microbiol. Spectr. 4, VMBF‑0011‑2015 (2016).
48. Strong, M. et al. Toward the structural genomics of
complexes: crystal structure of a PE/PPE protein
complex from Mycobacterium tuberculosis. Proc. Natl
Acad. Sci. USA 103, 8060–8065 (2006).
49. Brodin, P. et al. Dissection of ESAT‑6 system 1 of
Mycobacterium tuberculosis and impact on
immunogenicity and virulence. Infect. Immun. 74,
88–98 (2006).
50. Chen, J. M. et al. Mycobacterium tuberculosis EspB
binds phospholipids and mediates EsxA-independent
virulence. Mol. Microbiol. 89, 1154–1166 (2013).
51. Rosenberg, O. S. et al. Substrates control
multimerization and activation of the multi-domain
ATPase motor of type VII secretion. Cell 161,
501–512 (2015).
A report that provides the first structural insights
into the translocation mechanisms in ESX systems,
with a focus on the FtsK–SpoIIIE-like ATPase EccC.
52. de Jonge, M. I. et al. ESAT‑6 from Mycobacterium
tuberculosis dissociates from its putative chaperone
CFP‑10 under acidic conditions and exhibits
membrane-lysing activity. J. Bacteriol. 189,
6028–6034 (2007).
53. Peng, X. & Sun, J. Mechanism of ESAT‑6 membrane
interaction and its roles in pathogenesis of
Mycobacterium tuberculosis. Toxicon 116, 29–34
(2016).
54. Guglielmini, J., de la Cruz, F. & Rocha, E. P. Evolution
of conjugation and type IV secretion systems.
Mol. Biol. Evol. 30, 315–331 (2013).
55. Solomonson, M. et al. Structure of the mycosin‑1
protease from the mycobacterial ESX‑1 protein type
VII secretion system. J. Biol. Chem. 288,
17782–17790 (2013).
56. Ohol, Y. M. et al. Mycobacterium tuberculosis MycP1
protease plays a dual role in regulation of ESX‑1
secretion and virulence. Cell Host Microbe 7,
210–220 (2010).
57. Solomonson, M. et al. Structure of EspB from the
ESX‑1 type VII secretion system and insights into its
export mechanism. Structure 23, 571–583 (2015).
58. Korotkova, N. et al. Structure of EspB, a secreted
substrate of the ESX‑1 secretion system of
Mycobacterium tuberculosis. J. Struct. Biol. 191,
236–244 (2015).
59. MacGurn, J. A., Raghavan, S., Stanley, S. A.
& Cox, J. S. A non‑RD1 gene cluster is required for
Snm secretion in Mycobacterium tuberculosis.
Mol. Microbiol. 57, 1653–1663 (2005).
60. Fortune, S. M. et al. Mutually dependent secretion
of proteins required for mycobacterial virulence. Proc.
Natl Acad. Sci. USA 102, 10676–10681 (2005).
61. Champion, P. A., Champion, M. M., Manzanillo, P.
& Cox, J. S. ESX‑1 secreted virulence factors are
recognized by multiple cytosolic AAA ATPases in
pathogenic mycobacteria. Mol. Microbiol. 73,
950–962 (2009).
62. McLaughlin, B. et al. A mycobacterium ESX‑1‑secreted
virulence factor with unique requirements for export.
PLoS Pathog. 3, e105 (2007).
63. Bottai, D. et al. ESAT‑6 secretion-independent impact
of ESX‑1 genes espF and espG1 on virulence of
Mycobacterium tuberculosis. J. Infect. Dis. 203,
1155–1164 (2011).
64. Zhang, M. et al. EspI regulates the ESX‑1 secretion
system in response to ATP levels in Mycobacterium
tuberculosis. Mol. Microbiol. 93, 1057–1065
(2014).
65. Majlessi, L., Prados-Rosales, R., Casadevall, A.
& Brosch, R. Release of mycobacterial antigens.
Immunol. Rev. 264, 25–45 (2015).
66. Chen, J. M. et al. EspD is critical for the virulence-
mediating ESX‑1 secretion system in Mycobacterium
tuberculosis. J. Bacteriol. 194, 884–893 (2012).

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 689

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS

67. Ekiert, D. C. & Cox, J. S. Structure of a PE–PPE–
EspG complex from Mycobacterium tuberculosis
reveals molecular specificity of ESX protein secretion.
Proc. Natl Acad. Sci. USA 111, 14758–14763
(2014).
68. Daleke, M. H. et al. Specific chaperones for the type
VII protein secretion pathway. J. Biol. Chem. 287,
31939–31947 (2012).
69. Korotkova, N. et al. Structure of the Mycobacterium
tuberculosis type VII secretion system chaperone
EspG5 in complex with PE25–PPE41 dimer. Mol.
Microbiol. 94, 367–382 (2014).
70. Solans, L. et al. A specific polymorphism in
Mycobacterium tuberculosis H37Rv causes
differential ESAT‑6 expression and identifies WhiB6
as a novel ESX‑1 component. Infect. Immun. 82,
3446–3456 (2014).
71. Solans, L. et al. The PhoP-dependent ncRNA Mcr7
modulates the TAT secretion system in
Mycobacterium tuberculosis. PLoS Pathog. 10,
e1004183 (2014).
72. Perez, E. et al. An essential role for phoP in
Mycobacterium tuberculosis virulence.
Mol. Microbiol. 41, 179–187 (2001).
73. Walters, S. B. et al. The Mycobacterium tuberculosis
PhoPR two-component system regulates genes
essential for virulence and complex lipid biosynthesis.
Mol. Microbiol. 60, 312–330 (2006).
74. Frigui, W. et al. Control of M. tuberculosis ESAT‑6
secretion and specific T cell recognition by PhoP.
PLoS Pathog. 4, e33 (2008).
75. Raghavan, S., Manzanillo, P., Chan, K., Dovey, C.
& Cox, J. Secreted transcription factor controls
Mycobacterium tuberculosis virulence. Nature 454,
717–721 (2008).
76. Blasco, B. et al. Virulence regulator EspR of
Mycobacterium tuberculosis is a nucleoid-associated
protein. PLoS Pathog. 8, e1002621 (2012).
A paper that reports the function of EspR to be a
nucleoid-associated protein that regulates a large
network of genes.
77. Pang, X. et al. MprAB regulates the espA operon in
Mycobacterium tuberculosis and modulates ESX‑1
function and host cytokine response. J. Bacteriol.
195, 66–75 (2013).
78. Hunt, D. M. et al. Long-range transcriptional control of
an operon necessary for virulence-critical ESX‑1
secretion in Mycobacterium tuberculosis. J. Bacteriol.
194, 2307–2320 (2012).
79. Gonzalo-Asensio, J. et al. Evolutionary history of
tuberculosis shaped by conserved mutations in the
PhoPR virulence regulator. Proc. Natl Acad. Sci. USA
111, 11491–11496 (2014).
A study that reveals the role of compensatory
mutations in preserving ESX‑1 functions in PhoP–
PhoR-mutated lineages of the M. tuberculosis
complex.
80. Gordon, S. V. et al. Identification of variable regions in
the genomes of tubercle bacilli using bacterial artificial
chromosome arrays. Mol. Microbiol. 32, 643–656
(1999).
81. Brosch, R. et al. A new evolutionary scenario for the
Mycobacterium tuberculosis complex. Proc. Natl
Acad. Sci. USA 99, 3684–3689 (2002).
82. Mostowy, S., Cousins, D., Brinkman, J., Aranaz, A.
& Behr, M. A. Genomic deletions suggest a phylogeny
for the Mycobacterium tuberculosis complex. J. Infect.
Dis. 186, 74–80 (2002).
83. Gagneux, S. & Small, P. M. Global phylogeography of
Mycobacterium tuberculosis and implications for
tuberculosis product development. Lancet Infect. Dis.
7, 328–337 (2007).
84. Boritsch, E. C. et al. A glimpse into the past and
predictions for the future: the molecular evolution of
the tuberculosis agent. Mol. Microbiol. 93, 835–852
(2014).
85. Di Luca, M. et al. The ESX‑5 associated eccB–eccC
locus is essential for Mycobacterium tuberculosis
viability. PLoS ONE 7, e52059 (2012).
86. Ates, L. S. et al. Essential role of the ESX‑5 secretion
system in outer membrane permeability of pathogenic
mycobacteria. PLoS Genet. 11, e1005190
(2015).
87. Elliott, S. R. & Tischler, A. D. Phosphate starvation:
a novel signal that triggers ESX‑5 secretion in
Mycobacterium tuberculosis. Mol. Microbiol. 100,
510–526 (2016).
88. Abdallah, A. M. et al. A specific secretion system
mediates PPE41 transport in pathogenic
mycobacteria. Mol. Microbiol. 62, 667–679
(2006).
89. Abdallah, A. M. et al. PPE and PE_PGRS proteins of
Mycobacterium marinum are transported via the type
VII secretion system ESX‑5. Mol. Microbiol. 73,
329–340 (2009).
90. Abdallah, A. M. et al. The ESX‑5 secretion system of
Mycobacterium marinum modulates the macrophage
response. J. Immunol. 181, 7166–7175 (2008).
91. Daleke, M. H. et al. Conserved Pro-Glu (PE) and Pro-
Pro-Glu (PPE) protein domains target LipY lipases of
pathogenic mycobacteria to the cell surface via the
ESX‑5 pathway. J. Biol. Chem. 286, 19024–19034
(2011).
92. Sassetti, C. M., Boyd, D. H. & Rubin, E. J. Genes
required for mycobacterial growth defined by high
density mutagenesis. Mol. Microbiol. 48, 77–84
(2003).
93. Rodriguez, G. M., Voskuil, M. I., Gold, B.,
Schoolnik, G. K. & Smith, I. ideR, an essential gene
in Mycobacterium tuberculosis: role of IdeR in iron-
dependent gene expression, iron metabolism, and
oxidative stress response. Infect. Immun. 70,
3371–3381 (2002).
94. Dubnau, E., Chan, J., Mohan, V. P. & Smith, I.
Responses of Mycobacterium tuberculosis to growth
in the mouse lung. Infect. Immun. 73, 3754–3757
(2005).
95. Ilghari, D. et al. Solution structure of the
Mycobacterium tuberculosis EsxG–EsxH complex:
functional implications and comparisons with other
M. tuberculosis Esx family complexes. J. Biol. Chem.
286, 29993–30002 (2011).
96. Skjot, R. L. et al. Epitope mapping of the
immunodominant antigen TB10.4 and the two
homologous proteins TB10.3 and TB12.9, which
constitute a subfamily of the esat‑6 gene family.
Infect. Immun. 70, 5446–5453 (2002).
97. Majlessi, L., Rojas, M. J., Brodin, P. & Leclerc, C.
CD8+-T‑cell responses of Mycobacterium-infected
mice to a newly identified major histocompatibility
complex class I‑restricted epitope shared by proteins
of the ESAT‑6 family. Infect. Immun. 71, 7173–7177
(2003).
98. Hervas-Stubbs, S. et al. High frequency of CD4+ T cells
specific for the TB10.4 protein correlates with
protection against Mycobacterium tuberculosis
infection. Infect. Immun. 74, 3396–3407 (2006).
99. Mehra, A. et al. Mycobacterium tuberculosis type VII
secreted effector EsxH targets host ESCRT to impair
trafficking. PLoS Pathog. 9, e1003734 (2013).
100. Aagaard, C. et al. A multistage tuberculosis vaccine
that confers efficient protection before and after
exposure. Nat. Med. 17, 189–194 (2011).
101. Feltcher, M. E., Sullivan, J. T. & Braunstein, M. Protein
export systems of Mycobacterium tuberculosis: novel
targets for drug development? Future Microbiol. 5,
1581–1597 (2010).
102. Cole, S. T. et al. Massive gene decay in the leprosy
bacillus. Nature 409, 1007–1011 (2001).
103. Singh, P. et al. Insight into the evolution and origin of
leprosy bacilli from the genome sequence of
Mycobacterium lepromatosis. Proc. Natl Acad. Sci.
USA 112, 4459–4464 (2015).
104. Guinn, K. M. et al. Individual RD1‑region genes are
required for export of ESAT‑6/CFP‑10 and for virulence
of Mycobacterium tuberculosis. Mol. Microbiol. 51,
359–370 (2004).
105. Sayes, F. et al. Strong immunogenicity and cross-
reactivity of Mycobacterium tuberculosis ESX‑5 type
VII secretion- encoded PE–PPE proteins predicts
vaccine potential. Cell Host Microbe 11, 352–363
(2012).
106. Wards, B. J., de Lisle, G. W. & Collins, D. M. An esat6
knockout mutant of Mycobacterium bovis produced by
homologous recombination will contribute to the
development of a live tuberculosis vaccine. Tuber. Lung
Dis. 80, 185–189 (2000).
107. Lugo-Villarino, G. & Neyrolles, O. Manipulation of the
mononuclear phagocyte system by Mycobacterium
tuberculosis. Cold Spring Harb. Perspect. Med. 4,
a018549 (2014).
108. Deretic, V. et al. Mycobacterium tuberculosis inhibition
of phagolysosome biogenesis and autophagy as a host
defence mechanism. Cell. Microbiol. 8, 719–727
(2006).
109. Sturgill-Koszycki, S. et al. Lack of acidification in
Mycobacterium phagosomes produced by exclusion of
the vesicular proton-ATPase. Science 263, 678–681
(1994).
110. Stamm, L. M. et al. Mycobacterium marinum escapes
from phagosomes and is propelled by actin-based
motility. J. Exp. Med. 198, 1361–1368 (2003).
111. van der Wel, N. et al. M. tuberculosis and M. leprae
translocate from the phagolysosome to the cytosol in
myeloid cells. Cell 129, 1287–1298 (2007).
A paper that questions the long-standing view that
M. tuberculosis is an entirely intraphagosomal
pathogen.
112. Hart, P. D., Young, M. R., Gordon, A. H. &
Sullivan, K. H. Inhibition of phagosome–lysosome
fusion in macrophages by certain mycobacteria can be
explained by inhibition of lysosomal movements
observed after phagocytosis. J. Exp. Med. 166,
933–946 (1987).
113. Russell, D. G. The ins and outs of the Mycobacterium
tuberculosis-containing vacuole. Cell. Microbiol. 18,
1065–1069 (2016).
114. Simeone, R., Majlessi, L., Enninga, J. & Brosch, R.
Perspectives on mycobacterial vacuole-to‑cytosol
translocation: the importance of cytosolic access. Cell.
Microbiol. 18, 1070–1077 (2016).
115. Simeone, R. et al. Cytosolic access of Mycobacterium
tuberculosis: critical impact of phagosomal
acidification control and demonstration of occurrence
in vivo. PLoS Pathog. 11, e1004650 (2015).
116. Smith, J. et al. Evidence for pore formation in host cell
membranes by ESX‑1‑secreted ESAT‑6 and its role in
Mycobacterium marinum escape from the vacuole.
Infect. Immun. 76, 5478–5487 (2008).
117. Ma, Y., Keil, V. & Sun, J. Characterization of
Mycobacterium tuberculosis EsxA membrane
insertion: roles of N- and C‑terminal flexible arms and
central helix–turn–helix motif. J. Biol. Chem. 290,
7314–7322 (2015).
118. Wassermann, R. et al. Mycobacterium tuberculosis
differentially activates cGAS- and inflammasome-
dependent intracellular immune responses through
ESX‑1. Cell Host Microbe. 17, 799–810 (2015).
119. Watson, R. O. et al. The cytosolic sensor cGAS detects
Mycobacterium tuberculosis DNA to induce type I
interferons and activate autophagy. Cell Host Microbe.
17, 811–819 (2015).
120. Collins, A. C. et al. Cyclic GMP–AMP synthase is
an innate immune DNA sensor for Mycobacterium
tuberculosis. Cell Host Microbe 17, 820–828
(2015).
121. Brown, L., Wolf, J. M., Prados-Rosales, R. &
Casadevall, A. Through the wall: extracellular vesicles
in Gram-positive bacteria, mycobacteria and fungi.
Nat. Rev. Microbiol. 13, 620–630 (2015).
122. Manca, C. et al. Virulence of a Mycobacterium
tuberculosis clinical isolate in mice is determined by
failure to induce TH1 type immunity and is associated
with induction of IFN-α /β. Proc. Natl Acad. Sci. USA
98, 5752–5757 (2001).
123. Stanley, S. A., Johndrow, J. E., Manzanillo, P. &
Cox, J. S. The type I IFN response to infection with
Mycobacterium tuberculosis requires ESX‑1‑mediated
secretion and contributes to pathogenesis.
J. Immunol. 178, 3143–3152 (2007).
124. McNab, F., Mayer-Barber, K., Sher, A., Wack, A.
& O’Garra, A. Type I interferons in infectious disease.
Nat. Rev. Immunol. 15, 87–103 (2015).
125. Mishra, B. B. et al. Mycobacterium tuberculosis
protein ESAT‑6 is a potent activator of the NLRP3/ASC
inflammasome. Cell. Microbiol. 12, 1046–1063
(2010).
126. Wong, K. W. & Jacobs, W. R. Jr. Critical role for
NLRP3 in necrotic death triggered by Mycobacterium
tuberculosis. Cell. Microbiol. 13, 1371–1384
(2011).
127. Dorhoi, A. et al. Activation of the NLRP3
inflammasome by Mycobacterium tuberculosis is
uncoupled from susceptibility to active tuberculosis.
Eur. J. Immunol. 42, 374–384 (2012).
128. Master, S. S. et al. Mycobacterium tuberculosis
prevents inflammasome activation. Cell Host Microbe.
3, 224–232 (2008).
129. Romagnoli, A. et al. ESX‑1 dependent impairment of
autophagic flux by Mycobacterium tuberculosis in
human dendritic cells. Autophagy 8, 1357–1370
(2012).
130. Davis, J. M. & Ramakrishnan, L. The role of the
granuloma in expansion and dissemination of early
tuberculous infection. Cell 136, 37–49 (2009).
131. Stoop, E. J. et al. Zebrafish embryo screen for
mycobacterial genes involved in the initiation of
granuloma formation reveals a newly identified ESX‑1
component. Dis. Model. Mech. 4, 526–536 (2011).
132. Volkman, H. E. et al. Tuberculous granuloma
induction via interaction of a bacterial secreted
protein with host epithelium. Science. 327,
466–469 (2010).

690 | NOVEMBER 2016 | VOLUME 14 www.nature.com/nrmicro

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
 REVIEWS

133. Aguilo, J. et al. ESX‑1‑induced apoptosis is involved in
cell‑to‑cell spread of Mycobacterium tuberculosis.
Cell. Microbiol. 15, 1994–2005 (2013).
134. Weerdenburg, E. M. et al. ESX‑5‑deficient
Mycobacterium marinum is hypervirulent in adult
zebrafish. Cell. Microbiol. 14, 728–739 (2012).
135. Koo, I. C. et al. ESX‑1‑dependent cytolysis in lysosome
secretion and inflammasome activation during
mycobacterial infection. Cell. Microbiol. 10,
1866–1878 (2008).
136. Ates, L. S. et al. The ESX‑5 system of pathogenic
mycobacteria is involved in capsule integrity and
virulence through its substrate PPE10. PLoS Pathog.
12, e1005696 (2016).
A recent paper that reveals the role of the ESX‑5
system in mycobacterial capsule production.
137. Anderson, M., Aly, K. A., Chen, Y. H. & Missiakas, D.
Secretion of atypical protein substrates by the ESAT‑6
secretion system of Staphylococcus aureus. Mol.
Microbiol. 90, 734–743 (2013).
138. Sundaramoorthy, R., Fyfe, P. K. & Hunter, W. N.
Structure of Staphylococcus aureus EsxA suggests a
contribution to virulence by action as a transport
chaperone and/or adaptor protein. J. Mol. Biol. 383,
603–614 (2008).
139. Kneuper, H. et al. Heterogeneity in ess transcriptional
organization and variable contribution of the Ess/type
VII protein secretion system to virulence across closely
related Staphylococcus aureus strains. Mol.
Microbiol. 93, 928–943 (2014).
A paper that provides novel insights into the
ESX-like secretion system that is found in strains of
Staphylococcus aureus.
140. Warne, B. et al. The Ess/type VII secretion system of
Staphylococcus aureus shows unexpected genetic
diversity. BMC Genomics. 17, 222 (2016).
141. Anderson, M., Chen, Y. H., Butler, E. K. &
Missiakas, D. M. EsaD, a secretion factor for the Ess
pathway in Staphylococcus aureus. J. Bacteriol. 193,
1583–1589 (2011).
142. Jager, F., Zoltner, M., Kneuper, H., Hunter, W. N. &
Palmer, T. Membrane interactions and self-association
of components of the Ess/type VII secretion system of
Staphylococcus aureus. FEBS Lett. 590, 349–357
(2016).
143. Korea, C. G. et al. Staphylococcal Esx proteins
modulate apoptosis and release of intracellular
Staphylococcus aureus during infection in epithelial
cells. Infect. Immun. 82, 4144–4153 (2014).
144. Ummels, R. et al. Identification of a novel conjugative
plasmid in mycobacteria that requires both type IV
and type VII secretion. mBio. 5, e01744‑14 (2014).
145. Wang, J. et al. Insights on the emergence of
Mycobacterium tuberculosis from the analysis of
Mycobacterium kansasii. Genome Biol. Evol. 7,
856–870 (2015).
146. Kim, B. J. et al. Whole-genome sequence of a novel
species, Mycobacterium yongonense DSM 45126T.
Genome Announc. 1, e00604-13 (2013).
147. Gray, T. A., Krywy, J. A., Harold, J., Palumbo, M. J.
& Derbyshire, K. M. Distributive conjugal transfer in
mycobacteria generates progeny with meiotic-like
genome-wide mosaicism, allowing mapping of a
mating identity locus. PLoS Biol. 11, e1001602
(2013).
148. Supply, P. et al. Genomic analysis of smooth tubercle
bacilli provides insights into ancestry and
pathoadaptation of Mycobacterium tuberculosis.
Nat. Genet. 45, 172–179 (2013).
149. Mortimer, T. D. & Pepperell, C. S. Genomic
signatures of distributive conjugal transfer among
mycobacteria. Genome Biol. Evol. 6, 2489–2500
(2014).
150. Boritsch, E. C. et al. pks5‑recombination-mediated
surface remodelling in Mycobacterium tuberculosis
emergence. Nat. Microbiol. 1, 15019 (2016).
151. Boritsch, E. C. et al. Key experimental evidence
of chromosomal DNA transfer among selected
tuberculosis-causing mycobacteria. Proc. Natl Acad.
Sci. USA 113, 9876–9881 (2016).
152. Watson, R. O., Manzanillo, P. S. & Cox, J. S.
Extracellular M. tuberculosis DNA targets bacteria
for autophagy by activating the host DNA-sensing
pathway. Cell 150, 803–815 (2012).
153. Nakagawa, I. et al. Autophagy defends cells against
invading group A Streptococcus. Science. 306,
1037–1040 (2004).
154. Gutierrez, M. G. et al. Autophagy is a defense
mechanism inhibiting BCG and Mycobacterium
tuberculosis survival in infected macrophages. Cell
119, 753–766 (2004).
155. Thurston, T. L., Wandel, M. P., von Muhlinen, N.,
Foeglein, A. & Randow, F. Galectin 8 targets
damaged vesicles for autophagy to defend cells
against bacterial invasion. Nature. 482, 414–418
(2012).
156. Zhao, Z. et al. Autophagosome-independent essential
function for the autophagy protein Atg5 in cellular
immunity to intracellular pathogens. Cell Host Microbe
4, 458–469 (2008).
157. Kimmey, J. M. et al. Unique role for ATG5 in
neutrophil-mediated immunopathology during
M. tuberculosis infection. Nature 528, 565–569
(2015).
A paper that questions the efficacy of autophagy
as a major innate defence mechanism against
mycobacteria.
158. Brosch, R. et al. Genome plasticity of BCG and impact
on vaccine efficacy. Proc. Natl Acad. Sci. USA 104,
5596–5601 (2007).
159. Zhang, L. et al. Variable virulence and efficacy of
BCG vaccine strains in mice and correlation with
genome polymorphisms. Mol. Ther. 24, 398–405
(2016).
160. Bottai, D. et al. Increased protective efficacy of
recombinant BCG strains expressing virulence-neutral
proteins of the ESX‑1 secretion system. Vaccine. 33,
2710–2718 (2015).
161. Whole Mycobacteria Cell Vaccines for Tuberculosis
Summary Group. Developing whole mycobacteria cell
vaccines for tuberculosis: workshop proceedings,
Max Planck Institute for Infection Biology, Berlin,
Germany, July 9, 2014. Vaccine. 33, 3047–3055
(2015).
162. Sweeney, K. A. et al. A recombinant Mycobacterium
smegmatis induces potent bactericidal immunity
against Mycobacterium tuberculosis. Nat. Med. 17,
1261–1268 (2011).
163. Sayes, F. et al. CD4+ T cells recognizing PE/PPE
antigens directly or via cross reactivity are protective
against pulmonary Mycobacterium tuberculosis
infection. PLoS Pathog. 12, e1005770 (2016).
164. Bloemberg, G. V. et al. Acquired resistance to
Bedaquiline and Delamanid in therapy for
tuberculosis. N. Engl. J. Med. 373, 1986–1988
(2015).
165. Christophe, T. et al. High content screening
identifies decaprenyl-phosphoribose 2′ epimerase as a
target for intracellular antimycobacterial inhibitors.
PLoS Pathog. 5, e1000645 (2009).
166. Rybniker, J. et al. Anticytolytic screen identifies
inhibitors of mycobacterial virulence protein
secretion. Cell Host Microbe 16, 538–548
(2014).
167. VanderVen, B. C. et al. Novel inhibitors of cholesterol
degradation in Mycobacterium tuberculosis reveal
how the bacterium’s metabolism is constrained by the
intracellular environment. PLoS Pathog. 11,
e1004679 (2015).
168. Zuber, B. et al. Direct visualization of the outer
membrane of mycobacteria and corynebacteria in
their native state. J. Bacteriol. 190, 5672–5680
(2008).
169. Hoffmann, C., Leis, A., Niederweis, M., Plitzko, J. M.
& Engelhardt, H. Disclosure of the mycobacterial
outer membrane: cryo-electron tomography and
vitreous sections reveal the lipid bilayer structure.
Proc. Natl Acad. Sci. USA 105, 3963–3967
(2008).
170. Poulet, S. & Cole, S. T. Characterisation of the
polymorphic GC‑rich repetitive sequence (PGRS)
present in Mycobacterium tuberculosis.
Arch. Microbiol. 163, 87–95 (1995).
171. Majlessi, L. & Brosch, R. Mycobacterium tuberculosis
meets the cytosol: the role of cGAS in anti-
mycobacterial immunity. Cell Host Microbe. 17,
733–735 (2015).
172. Tschopp, J. & Schroder, K. NLRP3 inflammasome
activation: the convergence of multiple signalling
pathways on ROS production? Nat. Rev. Immunol. 10,
210–215 (2010).
Acknowledgements
The authors gratefully acknowledge the support of grants
from the European Community (grant H2020‑PHC‑
643381), the Agence Nationale de Recherche (grant
ANR‑14‑JAMR‑001‑02), Institut Pasteur (grant PTR 441) and
the Fondation pour la Recherche Médicale (grant
DEQ20130326471). R.B. is a member of the LabEx
Integrative Biology of Emerging Infectious Diseases (IBEID)
consortium at Institut Pasteur. M.I.G. is supported by an
M.D.–Ph.D. grant from the University of Groningen,
The Netherlands.
Competing interests statement
The authors declare no competing interests.

NATURE REVIEWS | MICROBIOLOGY VOLUME 14 | NOVEMBER 2016 | 691

©
2
0
1
6
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.