Here we report that a 27-kDa outer membrane protein of unknown function is a member of the DsbA family of oxidoreductases. Based on phylogenetic and functional criteria, as well as confirmation of a periplasmic location, we have named this gene
dsbA2. While
L. pneumophila contains two genes encoding DsbA-like proteins,
dsbA1 mutants were essentially wild type for infectivity of amoeba and exhibited no discernable phenotypes
in vitro. To overcome the apparent essentiality of DsbA2, we constructed a P198T mutant that in
E. coli DsbA (P151T) formed stable complexes with substrate proteins (
Kadokura et al., 2004). In
L. pneumophila, expression of the DsbA2 P198T mutant protein produced a dominant negative effect on several virulence phenotypes including: (i) loss of intracellular multiplication in amoeba; (ii) loss of attachment and invasiveness of HeLa cells; (iii) loss of contact (Dot/Icm dependent) lysis of erythrocytes; and (iv) loss of flagella-driven motility. Disruption of the Dot/Icm complex was inferred from the above results and confirmed by capture of DotC, DotG, DotK and IcmX substrates by P198T mutant protein. The abundance and diversity of captured proteins (see ) suggests that
L.
pneumophila is highly dependent on DsbA2 for oxidative folding of periplasmic proteins, including those destined for assembly into macromolecular structures. These studies not only revealed important biological functions for DsbA2 in
L. pneumophila, but also the likelihood that DsbA2 provides similar function in a very broad range on microbes, including many also expressing T4SS.
In retrospect, the original annotation of Com1 as an outer membrane protein was clearly misleading, but not uncommon. In reviewing the history, we note that i
n silico analysis of the primary amino acid sequence of DsbA2 identified a sec-dependent leader sequence typical of DsbA proteins and predicted with high confidence a periplasmic location (score of 9.84 and similar to that predicted for DsbA1). The periplasmic location was confirmed in our studies by cell fractionation and immunogold electron microscopy. In contrast, the cellular location of Com1 could not be predicted by PSORTb analysis, despite sharing 48% identity and 69% similarity with DsbA2. There was ambiguity in the original communication on the cellular location as recombinant Com1 expressed in
E. coli appeared in the Sarkosyl-soluble (inner membrane) fraction, while surface iodination and protinease K treatments of whole
C. burnetii cells indicated an outer membrane association (
Hendrix et al., 1993). Com1 is an immunodominant antigen of
C. burnetii that is useful in the diagnostics of Q fever (
Hendrix, et al., 1993), while DsbA2 is not an immunodominant antigen based on screening of convalescent sera from patients with confirmed Legionnaires’ disease (personal communication, Claressa Lucas, Centers for Disease Control and Prevention).
In addition to the misleading annotation, DsbA2 also differs sufficiently in amino acid sequence from the canonical DsbA that it is now listed by the NCBI as a separate family. Simple BLAST searches with DsbA2 do not capture DsbA, although domain searches with the thioredoxin fold of DsbA2 would have placed it in the thioredoxin super family, but not necessarily in the DsbA family. Most genera expressing DsbA2, like
L. pneumophila, also expressed a DsbA (DsbA1) orthologue (the exception being
C. burnetii), but without exception all lacked an orthologue of DsbC that provides protein disulfide isomerase activity. Ironically, comparative BLAST searches with
E. coli DsbA and DsbA2 (any member of this group) using a cut-off of e
−24, will reveal the DsbA2 group to be far more conserved through evolution than the DsbA group. Much of the amino acid conservation noted for DsbA2 members is due to a 50 – 60 amino acid extension N-terminal to the CXXC motif compared with the shorter 30 −34 amino acids for DsbA members. In this respect, DsbA2 may be more similar to DsbC (107 amino acids to CXXC motif), but unlike DsbC, DsbA2 lacks a well defined dimerization domain. Other similarities with DsbC include a common amino acid residue adjacent to the conserved
cis-proline (
TPA) instead of
VPA found in DsbA (
Ren et al., 2009). The similarly to DsbC isomerases might explain the ability of the free thiol of C89S mutant protein to form complexes with substrates as demonstrated in lane 2. However, the redox status of DsbA2 in the periplasm was more typical of DsbA than DsbC. If DsbA2 was DsbC-like, most of the protein would have existed in the reduced state. Our studies do not exclude the possibility that DsbA2 may be bifunctional or exist in different redox states as a monomer or potential dimer, all of which might vary with growth stage. We are currently investigating the nature of the substrates captured by DsbA2 C89S mutant protein that might support disulfide isomerase activity.
To our knowledge, all of the
dsb genes studied to date are nonessential for viability, though it is not uncommon for Gram-negative bacteria to express several DsbA-like enzymes. For example,
Neisseria meningitidis contains three DsbA-like homologues (
Tinsley, et al., 2004),
Salmonella enterica expresses SeDsbA, SeDsbL, and SeSgrA that have been crystallized and shown to be DsbA-like (
Jarrott et al., 2010), and in uropathogenic
E. coli a second DsbA/B like system, DsbI/L has been identified (
Grimshaw, et al., 2008). In the case of
N. meningitidis, single and double deletion mutants retained full virulence indicating considerable redundancy in function (
Tinsley et al., 2004). It is noteworthy that the triple mutant, while viable, exhibited a temperature sensitive growth phenotype (
Tinsley, et al., 2004). Several studies have shown that DsbA is required for assembling various extra-cytoplasmic virulence factors including components of T2 and T3 secretion systems and toxins (
Heras et al., 2009). Our inability to isolate mutants of
dsbA2 suggests that DsbA2 must provide an as yet undefined essential function(s) in
L. pneumophila and possibly other bacteria expressing a
dsbA2 orthologue.
In addition to
L. pneumophila and
C. burnetii, other human and plant pathogens expressing T4SS include:
B. suis,
Bartonella henslae,
Ehrlichia sp.,
Anaplasma sp.,
R. prowazekii and
Agrobacterium tumefactions (
Llosa et al., 2009). All of these bacteria express an orthologue of DsbA2, suggesting a role in assembling components of the respective T4SS. Bacterial genera not containing T4SS, but expressing DsbA2 include:
Caulobacter, Bradyrhizobium, Azospirillum, Paracoccus, Methylobacterium, and
Magnetospirillum. Two notable exceptions in bacteria that express T4SS systems include
B. pertussis (Ptl T4SS) which expresses DsbA1 and DsbC orthologues and
H. pylori in which an orthologue of DsbA is absent. In the case of
H. pylori, the acidic gastric mucosa, like anaerobic environments, do not favor stability of disulfide bonds. Further phylogenetic analyses of bacteria containing T4SS will likely confirm an evolutionary relationship with DsbA2. The phenotype of the P198T mutant suggests that the conserved
cis-proline mutation first exploited in
E. coli might be applied to other Gram-negative bacteria as a strategy to capture protein substrates associated with virulence or other cellular functions.
In addition to P198T capture of Dot/Icm components, a wide range of substrate proteins were captured (), including enhanced entry proteins (Enh) that are required for entry into amoeba (
Liu et al., 2008). Perhaps the reduced attachment and invasion seen with P198T mutant protein might be due to impairment of Enh function (). Heat shock proteins HtpA and HtpB (GroES and GroEL), previously reported as periplasmic proteins, were captured in this study (
Hoffman et al., 1990;
Garduno et al., 1998a and
1998c) and it is noteworthy that the localization of HtpB onto the cell surface is Dot/Icm dependent (
Garduno et al., 1998c). HtpB contains four cysteine residues (
E. coli GroEL contains two) whose oxidation might be important for function and like DsbA2 is enriched in the hyper virulent cyst form. Ironically, HtpA contains no cysteine residues and raises the possibility of co-capture (with HtpB) of molecular complexes. It is noteworthy that other Dot/Icm proteins that contain no cysteine residues such as DotD, IcmW and IcmY were not captured, nor were cytoplasmic proteins rich in cysteine residues like isocitrate dehydrogenase. However, as noted previously (
Kadokura et al., 2004), despite the addition of excess iodoacetamide to alkylate free thiols to limit capture of cytoplasmic proteins, some cytoplasmic proteins were captured by this procedure. In retrospect, one might limit cytoplasmic contamination by isolating captured complexes by osmotic shock, though these strategies could potentially limit capture of membrane associated complexes. While subunits of OmpS were also captured in these studies, it is unlikely that DsbA2 plays a major role in catalyzing disulfide bond cross-linking of the subunits as one might expect the over expression of the P198T mutant to have resulted in restricted
in vitro growth or changes in the OmpS profile. However, further studies will be required to fully explore the role of DsbA2 in assembly of the OmpS oligomer.
The lack of motility and assembled flagella in the P198T over-expressing mutant described here mirrors what is known in other Gram-negative bacteria with DsbA mutations: Δ
dsbA E. coli produce no T4 pili (BFP) (
Vogt et al., 2010), and
Salmonella lacking the DsbA deriviative SrgA produce no fimbriae (
Bouwman et al., 2003). The absence of flagellin in the periplasmic shockate with P198T mutant protein suggests that DsbA2 either catalyzes a disulfide bond required for flagella biosynthesis or indirectly, as through a defect in FlgI assembly (flagella motor defect) triggers degradation of subunits and or activation of regulatory systems like CpxR that might repress the flagella system (
Ravio, 2005).
In summary, our studies assign biological functions to a family of proteins previously annotated as 27-kDa outer membrane proteins similar to Com1 as disulfide bond oxidoreductases. The DsbA2 lineage is phylogenetically distinct from members of the DsbA family. Remarkably, all members of this group also lack orthologues of DsbC and DsbG, but retain the DsbD reductase, suggesting the possibility that DsbA2 might be bifunctional, providing disulfide-bond oxidizing as well as editing functions (protein disulfide isomerase) required for repairing nonconsecutive disulfide bonds. Finally, the DsbA2 lineage is conserved among those bacterial species that also express a T4SS, suggesting a likely role for DsbA2 in assembly or function of these macromolecular structures.
The publisher's final edited version of this article is available at 
