Bordetella pertussis autotransporter Vag8 binds human C1 esterase inhibitor and confers serum resistance.

Abstract

Bordetella pertussis employs numerous strategies to evade the immune system, including the ability to resist killing via complement. Previously we have shown that B. pertussis binds a complement regulatory protein, C1 esterase inhibitor (C1inh) to its surface in a Bvg-regulated manner (i.e. during its virulence phase), but the B. pertussis factor was not identified. Here we set out to identify the B. pertussis C1inh-binding factor. Using a serum overlay assay, we found that this factor migrates at approximately 100 kDa on an SDS-PAGE gel. To identify this factor, we isolated proteins of approximately 100 kDa from wild type strain BP338 and from BP347, an isogenic Bvg mutant that does not bind C1inh. Using mass spectrometry and bioinformatics, we identified the autotransporter protein Vag8 as the putative C1inh binding protein. To prove that Vag8 binds C1inh, vag8 was disrupted in two different B. pertussis strains, namely BP338 and 18-323, and the mutants were tested for their ability to bind C1inh in a surface-binding assay. Neither mutant strain was capable of binding C1inh, whereas a complemented strain successfully bound C1inh. In addition, the passenger domain of Vag8 was expressed and purified as a histidine-tagged fusion protein and tested for C1inh-binding in an ELISA assay. Whereas the purified Vag8 passenger bound C1inh, the passenger domain of BrkA (a related autotransporter protein) failed to do so. Finally, serum assays were conducted to compare wild type and vag8 mutants. We determined that vag8 mutants from both strains were more susceptible to killing compared to their isogenic wild type counterparts. In conclusion, we have discovered a novel role for the previously uncharacterized protein Vag8 in the immune evasion of B. pertussis. Vag8 binds C1inh to the surface of the bacterium and confers serum resistance.

Full text

Page 1

Bordetella pertussis Autotransporter Vag8 Binds Human
C1 Esterase Inhibitor and Confers Serum Resistance
Nico Marr¤, Nita R. Shah, Rose Lee, Emma J. Kim¤, Rachel C. Fernandez*
Department of Microbiology & Immunology, University of British Columbia, Vancouver, British Columbia, Canada
Abstract
Bordetella pertussis employs numerous strategies to evade the immune system, including the ability to resist killing via
complement. Previously we have shown that B. pertussis binds a complement regulatory protein, C1 esterase inhibitor
(C1inh) to its surface in a Bvg-regulated manner (i.e. during its virulence phase), but the B. pertussis factor was not identified.
Here we set out to identify the B. pertussis C1inh-binding factor. Using a serum overlay assay, we found that this factor
migrates at approximately 100 kDa on an SDS-PAGE gel. To identify this factor, we isolated proteins of approximately
100 kDa from wild type strain BP338 and from BP347, an isogenic Bvg mutant that does not bind C1inh. Using mass
spectrometry and bioinformatics, we identified the autotransporter protein Vag8 as the putative C1inh binding protein. To
prove that Vag8 binds C1inh, vag8 was disrupted in two different B. pertussis strains, namely BP338 and 18–323, and the
mutants were tested for their ability to bind C1inh in a surface-binding assay. Neither mutant strain was capable of binding
C1inh, whereas a complemented strain successfully bound C1inh. In addition, the passenger domain of Vag8 was expressed
and purified as a histidine-tagged fusion protein and tested for C1inh-binding in an ELISA assay. Whereas the purified Vag8
passenger bound C1inh, the passenger domain of BrkA (a related autotransporter protein) failed to do so. Finally, serum
assays were conducted to compare wild type and vag8 mutants. We determined that vag8 mutants from both strains were
more susceptible to killing compared to their isogenic wild type counterparts. In conclusion, we have discovered a novel
role for the previously uncharacterized protein Vag8 in the immune evasion of B. pertussis. Vag8 binds C1inh to the surface
of the bacterium and confers serum resistance.
Citation: Marr N, Shah NR, Lee R, Kim EJ, Fernandez RC (2011) Bordetella pertussis Autotransporter Vag8 Binds Human C1 Esterase Inhibitor and Confers Serum
Resistance. PLoS ONE 6(6): e20585. doi:10.1371/journal.pone.0020585
Editor: Jane Deng, University of California Los Angeles, United States of America
Received September 22, 2010; Accepted May 5, 2011; Published June 14, 2011
Copyright: ? 2011 Marr et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was funded in part by the Canadian Institutes of Health Research (http://www.cihr-irsc.gc.ca/) grant # DOP-53723, and the Natural Sciences
and Engineering Research Council of Canada (http://www.nserc-crsng.gc.ca/) grant # 194599. The funders had no role in study design, data collection and
analysis, decision to publish, or preparation of the manuscript. No additional external funding was received for this study.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: rachelf@mail.ubc.ca.
¤ Current address: Department of Pediatrics, University of British Columbia, Child & Family Research Institute, Vancouver, British Columbia, Canada
Introduction
Complement is a critical and multifaceted host defense system
comprised of a series of proteins—many of which are zymogens—
found largely in plasma [1] but also on respiratory mucosal
surfaces [2,3]. Three activation pathways of complement are
known today, designated the classical, lectin, and alternative
pathway. Following activation, complement components of all
three pathways act in a defined cascade to effect immune
clearance of microorganisms either directly by assembly of
membrane attack complexes, or indirectly by opsonization and
release of peptide mediators of inflammation which promote the
attraction of and recognition by various immune cells [1]. To
protect the host from the effector functions of complement and
prevent rapid consumption of complement components in
response to trivial stimuli, all three activation pathways of
complement are tightly regulated at different stages by a number
of complement regulatory proteins [4]. Bordetella pertussis, the
causative agent of whooping cough, has been shown to express a
variety of virulence-associated factors that in concert enable the
bacteria to colonize the mucosa of the upper respiratory tract in
humans. Expression of most virulence-associated factors in B.
pertussis is positively regulated by the BvgAS two component
system [5,6]. Apart from adherence of the bacteria to the ciliated
respiratory epithelium, many of these factors help the bacteria to
evade or modulate host immune defenses [7,8]. One of the
immune defenses to overcome is complement.
There is a basic necessity for B. pertussis to prevent complement
activation because the B. pertussis endotoxin lacks a repetitive O
polysaccharide [9] rendering the bacteria particularly vulnerable
to direct, complement-mediated bacterial lysis—unlike many other
Gram-negative bacteria including other Bordetella species where
endotoxin can act as a protective shield [10]. Indeed, B. pertussis is
relatively resistant to killing by complement [11]. We have
previously shown that the B. pertussis autotransporter protein BrkA
mediates resistance to the classical complement pathway [11] and
although this happens at a very early stage in the pathway [12],
the underlying molecular mechanism remains elusive. Interesting-
ly, B. pertussis expresses other factors in its virulence phase that can
interact with complement components or its regulators, either
directly or indirectly. Filamentous hemagglutinin (FHA) has been
shown to bind a complement regulator called C4-binding protein
(C4BP) [13,14], however how this affects serum resistance is not
known since B. pertussis mutants deficient in FHA expression
exhibit a level of complement resistance that is not significantly
different compared to their parental wild type strain [15]. We have
PLoS ONE | www.plosone.org1 June 2011 | Volume 6 | Issue 6 | e20585

Page 2

recently demonstrated that B. pertussis is able to recruit another
complement regulatory protein of humans, C1 esterase inhibitor
(C1inh). This phenotype is associated with resistance to comple-
ment-mediated killing but requires neither the expression of BrkA
nor FHA [16,17]. With the present study we set out to identify the
B. pertussis factor that is responsible for the binding of human
C1inh to the bacterial surface, hence also crucial for serum
resistance. This factor was found to be the autotransporter protein
Vag8.
Results
Identification of the B. pertussis factor that mediates
recruitment of human C1-esterase inhibitor
We have previously shown that B. pertussis is capable of binding
human C1inh to its surface and that this is dependent on signal
transduction by the BvgAS two-component system, the master
virulence regulatory system of Bordetella species. Furthermore, the
ability of B. pertussis to bind human C1inh was found to be
independent of the Bvg-activated serum resistance protein BrkA
[16]. In the present study we aimed to identify the bacterial ligand
that interacts with human C1inh. We first tested a variety of well-
characterized B. pertussis mutants with defects in BvgAS-activated
genes using the C1inh binding assay that we described previously
[16]. These mutants included: BBC9BrkA (a derivative of wild
type W28) that does not express pertactin and BrkA [11,18]; SK34
(a derivative of wild type 18–323) that harbors a TnphoA insertion
in the tcfA locus [19]; BP338 (Tohama I)-derived Tn5lac mutants
BPM3171 and BPM1809 which fail to secrete pertussis toxin [20]
and lack the expression of dermonecrotic toxin [21], respectively;
as well as BP338 (Tohama I)-derived mutants BP348 and BP353
which harbor Tn5 insertions in cyaA (adenylate cyclase toxin) and
fimC (fimbria), respectively. The latter strain also does not express
filamentous haemagglutinin (FHA) due to a polar effect of the
transposon insertion [22,23]. Wild type BP338 and the isogenic
BvgS mutant BP347 (bvgS::Tn5) [22,23] were used as positive and
negative controls, respectively. We found that with the exception
of the BvgS mutant BP347, none of the B. pertussis mutants tested
here was visibly compromised in C1inh binding (data not shown).
Using microarray-based transcriptional profiling of Bvg+and
Bvg2phase-locked mutants, Cummings et al. [5] revealed that
155 genes are Bvg-activated in B. pertussis Tohama I-derived
strains. To narrow down the list of candidate B. pertussis factors
potentially involved in the recruitment of C1inh, we employed a
modification of a so-called Far-Western or gel-overlay strategy
described elsewhere [24], in combination with mass spectrometry.
This technique has been used to study protein-protein interactions
[24] and it enabled us to determine the approximate molecular
mass of the factor expressed by B. pertussis that mediates
recruitment of C1inh. In brief, whole cell lysates of B. pertussis
strains BP338 and BP347 were subjected to SDS-PAGE.
Separated proteins were blotted onto a PVDF membrane and
subsequently renatured. Membranes were then overlaid with 1%
normal human serum, and human C1inh interacting with any B.
pertussis factor on the PVDF membrane was detected by standard
immunoblot analysis. Using this approach, we found human
C1inh to interact with a B. pertussis factor migrating at
approximately 95 to 100 kDa that was expressed in wild type
BP338 but absent in BvgS mutant BP347 (Fig. 1). To validate that
the bacterial factor in question is indeed subject to BvgAS
regulation, we cultivated wild type BP338 in the absence or
presence of 50 mM MgSO4(which turns off the Bvg system) and
employed the Far-Western strategy described above. As shown in
figure 1, the B. pertussis factor capable of binding human C1inh
was absent when the bacteria were grown under BvgAS-
modulating conditions.
To identify BvgAS-regulated bacterial factors migrating at ca.
95 to 100 kDa on SDS-PAGE gels, crude membrane enrichments
of wild type strain BP338 (grown under Bvg-activating conditions)
and BvgS mutant BP347 were separated by SDS-PAGE and the
polyacrylamide gel stained with Coomassie blue. A visible protein
band migrating at the appropriate molecular weight and present in
BP338 but absent in BP347 was excised out of the protein gel.
Corresponding regions from the lane containing membrane
enrichments of BvgS mutant BP347 were similarly excised.
Trypsin-digested peptides of the gel samples from both strains
were subjected to MS/MS analysis, and a search of the obtained
data against the protein sequence database MSBD was performed
using Mascot. Using this approach, we identified five significant
matches of proteins that were present in membrane enrichments of
BP338, but were not detected in BvgS mutant BP347; namely the
gene product of virulence-activated gene 8 (Vag8), Bvg-interme-
diate phase protein A (BipA), pertactin, a pyruvate dehydrogenase
E1 component and DNA gyrase subunit B (Table 1).
Because the latter two hits are components of essential cytosolic
enzymes that were likely present in BvgS mutant BP347 but which
remained undetected, and since the B. pertussis pertactin-BrkA
double mutant BBC9BrkA was found to be capable of binding
C1inh, we decided to focus on Vag8 and BipA. bipA and vag8 were
disrupted using suicide vectors pTEN34 and pEG7-vag8,
respectively and hemolytic B. pertussis mutants with genomic
insertions of pTEN34 or pEG7-vag8 were isolated and designated
as BP338BipA2or BP338Vag82. Correct genomic integration of
the suicide vectors was validated by PCR of the genomic insertion
site using vector-specific primers in combination with a bipA or a
vag8 promoter-specific primer that could not hybridize to pTEN34
or pEG7-vag8, respectively. PCR products of the expected sizes
were found when using genomic DNA of both mutants as
templates, whereas no product was detected with genomic DNA of
wild type BP338 that served as the negative control. The vag8
results are shown in Figure 2(A and B). The PCR products
generated from genomic DNA of mutants BP338BipA2and
Figure 1. Far-Western analysis showing interaction of C1inh
from normal human serum with an unknown B. pertussis factor
migrating at ca. 90 to 100 kDa. B. pertussis strain BP338 (wild type)
and its isogenic mutant BP347 (bvgS::Tn5) were grown without (2) or
with (+) Bvg-modulating agent MgSO4(50 mM) prior to performing the
Far-Western assay using normal human serum as the probe, followed
by detection using anti-C1inh antibody.
doi:10.1371/journal.pone.0020585.g001
B. pertussis Vag8 Binds Human C1-Inhibitor
PLoS ONE | www.plosone.org2June 2011 | Volume 6 | Issue 6 | e20585

Page 3

BP338Vag82were sequenced and found to encode the expected
genomic insertion sites.
Following verification of the mutation in BP338BipA2and
BP338Vag82, the two B. pertussis mutants were then assayed for
their ability to recruit human C1inh from human serum [16] and
to bind to C1inh by use of the Far-Western approach described
above. Wild type strain BP338 and BvgS mutant BP347 served as
positive and negative controls, respectively. Whereas B. pertussis
mutant BP338BipA2was found to be capable of binding human
C1inh using both the surface-binding and the Far Western overlay
assays, BP338Vag82was not (Fig. 3A and B). The lack of C1inh
binding was not due to phase variation of the Bvg system as
demonstrated by the expression of the Bvg-activated BrkA protein
(Fig. 3C), and by hemolysis which is a function of the Bvg-
activated adenylate cyclase toxin. Furthermore, it is unlikely that
the lack of C1inh binding was due to a polar effect of the insertion
mutation as there are no genes in the same orientation
immediately downstream of vag8, and the gene immediately
following vag8 is in the opposite orientation (Fig. 2A). In any event,
to reproduce these findings, independent Vag8 mutants were
generated again in BP338 and in a streptomycin-resistant
derivative of a different B. pertussis strain called 18–323. All of
the Vag8 mutants expressed BrkA and were hemolytic, but were
unable to bind C1inh. These results strongly implicate Vag8 as the
C1inh-binding factor. Finally, we complemented one of the Vag8
mutants originated from B. pertussis BP338 with pVag8, a plasmid
that replicates in bordetellae and encodes a full-length open
reading frame (ORF) of vag8 driven by promoter Pcpn10. In parallel,
we introduced an empty vector control (pBBR1MCS2) or pBrkA
(a similar plasmid encoding brkA instead of vag8) into this Vag8
mutant via conjugation. As shown in Fig. 3D, we were able to
partially restore the ability to bind C1inh in the Vag8 mutant
complemented with the plasmid that encodes a full-length vag8
ORF, whereas no C1inh could be observed in the conjugants
harboring the empty vector control or pBrkA.
Vag8 is an autotransporter protein [25]. As is typical of
autotransporter proteins, it has a surface-exposed passenger
domain that provides the effector function, and a translocator
domain that is involved in secretion and anchoring of the Vag8
passenger to the bacterial surface (Fig. 2C). To further validate
that Vag8 binds C1inh, the passenger domain of Vag8 comprised
of residues Val40to Leu610was fused to a histidine tag and purified
using nickel chromatography. An ELISA was used to examine
whether the purified Vag8 passenger is capable of directly binding
C1inh. Briefly, Immuno 96 MicroWell MaxiSorpTMplates were
coated with different concentrations of the purified Vag8
passenger domain, and after washing and then blocking non-
specific binding, the plates were probed with either purified C1inh
or human serum. Bound C1inh was detected using anti-C1inh
antiserum. Results using this assay confirm direct binding of the
Vag8 passenger domain to purified C1inh (Fig. 4A) or C1inh from
human serum (Fig. 4B) in a dose-dependent manner. In contrast,
the similar-sized (73-kDa) BrkA passenger domain [26] which
shares 27% identity to Vag8 (residues 205–604) failed to bind
C1inh. Similar observations were made using the Far-Western
approach described above (data not shown). No signal was
detected when we performed the ELISA on wells that were coated
with 5 mg/ml of the purified Vag8 passenger domain but not
Table 1. Genes encoding the proteins that were identified by tandem mass spectrometry.
Gene nameGene product
Probability Based Mowse Score
BP338 BP347
BP2315 vag8 autotransporter Vag8 2011not detected
BP2014acnA Putative aconitate hydratase12651251
BP1112bipA putative OM ligand binding protein (BipA) 1185not detected
BP0197gcvP glycine cleavage system P protein not detected238
BP2497putative zinc protease781202
BP0460topBDNA topoisomerase iii 346458
BP0869pepNaminopeptidase N 240621
BP0993aceE pyruvate dehydrogenase E1 component234not detected
BP1836alaS, lovB alanyl-tRNA synthetase 219662
BP0489gyrBDNA gyrase subunit B 116not detected
BP2573pheTphenylalanyl-tRNA synthetase beta chain PheT 76 527
BP0944gyrADNA gyrase subunit A67 273
BP2044leucyl-tRNA synthetase58565
BP3467putative exported protein56 129
BP1054prnpertactin54not detected
BP1549dnaX DNA polymerase III subunit Taunot detected168
BP1566mutSDNA mismatch repair protein mutSnot detected143
BP2184rpoDRNA polymerase sigma factornot detected 130
BP1198clpB, htpMprotease ClpB, ATPase subunitnot detected 67
BP1417glnD[Protein-PII] uridylyltransferasenot detected 66
(MS/MS) analysis and ions search against protein sequence database MSDS using Mascott. All significant hits (P,0.05) with individual ions scores .50 are listed, from
highest score (top) to lowest (bottom).
doi:10.1371/journal.pone.0020585.t001
B. pertussis Vag8 Binds Human C1-Inhibitor
PLoS ONE | www.plosone.org3June 2011 | Volume 6 | Issue 6 | e20585

Page 4

probed with purified C1inh or serum, ruling out non-specific
binding of the antibodies to Vag8. Taken together, our data show
that the Vag8 passenger domain binds directly to human C1inh.
Serum sensitivity in the absence of Vag8 expression
We have previously shown that the ability of B. pertussis to bind
human C1inh is well-correlated with serum resistance [16]. By
using a serum killing assay, we aimed to confirm that expression of
Vag8 by B. pertussis, which we have shown to be essential for
binding human C1inh, is also crucial for serum resistance. We first
assayed BP338Vag82and its parental strain BP338, as well as
BrkA mutant RFBP2152. As shown in figure 5A, mutant
BP338Vag82exhibited similar serum sensitivity levels as the
BrkA mutant RFBP2152; both were significantly more sensitive to
serum-mediated killing than their parental wild type strain BP338.
We next looked at 18–323SmrVag82and its parental strain 18–
323Smr. In comparison to its parental strain, 18–323SmrVag82
was also unable to bind C1inh and was significantly more serum
sensitive (Fig. 5B). Finally, we assessed serum resistance in the
BP338Vag82strain complemented with pVag8. The partial
complementation of C1inh-binding exhibited in this strain was
not sufficient to restore serum resistance (not shown). In this
regard, it should be noted that whereas the BP338 and
BP338Vag82strains grow similarly, the complemented strain
grows very slowly – likely due to the selection pressure imposed by
2 antibiotics, further complicating efforts to compare serum
resistance using this strain.
Discussion
Vag8 is a Bvg-activated 95-kDa autotransporter protein of
B. pertussis [25,27] whose function has remained elusive [28]. In the
present study we showed that disruption of vag8 abolishes binding
of C1inh and that complementing a Vag8 mutant with a plasmid
encoding a full-length vag8 ORF under control of the non-native
promoter Pcpn10partially restores the C1inh-binding phenotype.
Furthermore, we demonstrated that purified Vag8 protein
(passenger domain) but not BrkA binds to C1inh in a dose-
dependent manner. These experiments clearly indicate that Vag8
is the C1inh binding factor. We also showed that disrupting vag8
(done two times independently in strain BP338, and once in 18–
323) significantly diminishes resistance to serum killing, linking the
phenotypes of C1inh binding and serum resistance to Vag8
expression. Taken together, our results show that expression of
Vag8 is essential for the ability of B. pertussis to bind human C1inh
during the virulent phase, and confers serum resistance.
Previously, a TnPhoA Vag8 mutant derived from B. pertussis
strain 18–323 was found to be slightly but not significantly more
serum sensitive than its parental strain which suggested a
negligible role of Vag8 in complement resistance [15]. However,
Figure 2. Generation of vag8 mutants and description of the vag8 genetic locus and predicted protein architecture. A Schematic of
the vag8 locus that is knocked out in B. pertussis mutant BP338Vag82. The 2748-bp coding sequence of vag8 is shown as an open arrow, with the
hatched region indicating the internal vag8 fragment which was cloned into vector pEG7 to generate vector pEG7-Vag8. Dashed lines indicate vector
sequence in the mutant strain. Solid lines indicate genomic sequence of BP338. vag8 is flanked by rRNA genes at the 59 end and locus tag BP2314 at
the 39 end (solid arrows, not drawn to scale). The bar indicates the 810-bp PCR product (shown in B) that was indicative of the correct genomic
integration of pEG7-Vag8 into the vag8 gene. C Architecture of the 95-kDaVag8 autotransporter protein. The predicted signal peptide and the
passenger and translocator domains are indicated. The region spanning residues 610 and 644 comprise a predicted alpha-helical linker region
present in all autotransporter proteins. Unlike many autotransporter proteins, Vag8 is not cleaved to separate its passenger domain from the
translocator. The duplicated fragments of vag8 in BP338Vag82are missing sequences encoding the translocator and signal peptide respectively.
doi:10.1371/journal.pone.0020585.g002
B. pertussis Vag8 Binds Human C1-Inhibitor
PLoS ONE | www.plosone.org4 June 2011 | Volume 6 | Issue 6 | e20585

Page 5

in that study the parental strain was also found to be significantly
more serum sensitive in comparison to BP338 [15], whereas in the
present study, using a streptomycin-resistant derivative of 18–323,
we found the opposite. It is possible that low levels of serum
resistance of B. pertussis wild type strain 18–323 reported in the
previous study [15] were due to second site mutations in this
strain. Barnes and Weiss [29] have shown that growth phase
affects susceptibility to complement so it is also possible that the
Figure 3. A B. pertussis Vag8 mutant fails to bind C1inh. A Standard immunoblot analysis showing surface binding of the 104-kDa C1inh
protein to whole bacteria incubated in serum; B Far-Western analysis testing for the interaction of C1inh with a ca. 95- to 100-kDa factor of B.
pertussis. The left and right panels in A and B show the results with BP338Vag82(vag8::pEG7-Vag8) and BP338BipA2(bipA::pTEN34), respectively. C
Immunoblot probed for the BvgAS-activated autotransporter protein BrkA. D Immunoblot analysis showing surface binding of C1inh to wild type B.
pertussis (BP338) and the BP338Vag82mutant complemented with pVag8. BP338Vag82complemented with control plasmids pBBR1MCS2 (empty
vector) or pBrkA (vector with the brkA gene under control of Pcpn10) did not bind C1inh.
doi:10.1371/journal.pone.0020585.g003
Figure 4. Purified histidine-tagged Vag8 passenger domain binds C1inh. Shown are means 6 SD of a representative ELISA done in
triplicate, where Nunc MaxiSorpH flat-bottom 96 well plates were coated with increasing two-fold concentrations of purified histidine-tagged Vag8
passenger or purified histidine-tagged BrkA passenger, and after blocking non-specific binding, incubated with either 10 mg/ml purified C1inh (A), or
2% normal human serum diluted in 16Veronal buffer (B). For each graph shown, two independent experiments were done.
doi:10.1371/journal.pone.0020585.g004
B. pertussis Vag8 Binds Human C1-Inhibitor
PLoS ONE | www.plosone.org5 June 2011 | Volume 6 | Issue 6 | e20585