In this study we established an unbiased, cultivation independent approach to screen for CEACAM-binding bacteria in macaque stool samples. By 454 pyrosequencing of the CEACAM-enriched bacteria and subsequent bioinformatical analyses we identified several members of the genus Prevotella to be potential macaque CEA-binding bacteria. Using pulldown assays we confirmed the interaction of Prevotella species and macaque CEA biochemically. In addition, Prevotella interacted with several human CEACAMs, but not with CEACAMs from more distantly related mammals.
Prevotella species are anaerobic, gram-negative rods belonging to the phylum Bacteroidetes. A recent study, which analyzed the rhesus monkey gut microbiota, identified the Prevotella group as a common genus residing in the macaque gut (McKenna et al. 2008). In humans some Prevotella, namely P. intermedia and P. nigrescens, have been associated with periodontal disease. However, Prevotella are also part of the normal human microflora and were isolated from different parts of the human body e.g. the oral cavity, the upper respiratory tract, the urogenital tract and the gut (Avgustin et al. 2001).
Different individuals display a high variability in their microflora at the bacterial genus level, however, the proportion of the major bacterial phyla between individuals is similar (Eckburg et al. 2005; Arumugam et al. 2011). In line with this, the human’s gut microbiota is mainly composed of two major phyla, namely Firmicutes and Bacteroidetes (Mahowald et al. 2009). Despite the observed variability at the bacterial genus level, the relative abundance of the different clusters of orthologous groups of genes (COG) and categories of metabolic
pathways revealed a quite consistent pattern between individuals (Turnbaugh et al. 2009). This means that different bacterial species can perform similar tasks in the human body. Furthermore, the microbiota differs a lot between habitats at different sites of the body (Costello et al. 2009). Interestingly, this spatial distribution of the microbiota is relatively stable over time and between individuals. Although, the dissimilarity in the microbial composition of different humans is huge, the microbiota of one human is still more similar to the microbiota of other humans than to other mammalian microbiotas. This indicates that a host co-evolves with its microbiota.
The forces which shape a host’s microbiota are not completely understood and different host factors might select for its microbiota (Bevins and Salzman 2011).
In this regard, the influence of a host’s diet in sculpting its microbiota was extensively studied (Turnbaugh et al. 2006; Ley et al. 2008). For instance, it was shown that children from Europe and rural Africa differ in their microbiota and that these differences were connected to the children’s diet (De Filippo et al. 2010). However, since mammals with a similar diet often strongly differ in their microbiota additional factors must contribute to the microbial makeup in mammals. Thus, former studies revealed that host phylogeny is important for a host’s microbial composition (Dethlefsen et al. 2007; Ochman et al. 2010). In line with this, analyses of the distal gut microbial community of the five great ape species, revealed a clear species-specific signature of the host’s microbiota. Interestingly, the pattern of relationship based on the host’s microbiota was congruent with the known relationships from mitochondrial DNA (Ochman et al. 2010). Another study examined the microbiota of two basal metazoan Hydra species, maintained in laboratory culture for more than two decades under identical conditions (Fraune and Bosch 2007). Surprisingly, their microbial composition differed significantly and was still more similar to their respective Hydra species living in the wild. This is a striking proof that genetic factors outweigh the environmental factors in shaping of a host’s microbiota.
Bacteria have evolved different mechanism to efficiently colonize mammals as their ecological niche. For instance, bacterial interactions with receptors expressed on epithelial cells, do not only allow mucosal colonization, but also trigger signal cascades inside the cell. In line with this, Toll-like receptors (TLRs), which are expressed by many different intestinal cells, recognize
MAMPs (microbial-associated molecular patterns) expressed by bacteria and coordinate a cascade of innate and adaptive immune responses which control infections. One example how TLRs influence the structure of the microbial community was provided by a recent study, demonstrating that the prominent gut commensal Bacteroides fragilis utilizes TLR signaling to establish a successful host-microbial symbiosis (Round et al. 2011). Binding of B. fragilis to
TLR2, results in the suppression of the host inflammatory response, allowing B. fragilis to persist on host tissues. In contrast, in the absence of TLR2
signaling B. fragilis could no longer reside within its intestinal niche. This demonstrates that TLR engagement by B. fragilis is important to establish a host-commensal symbiosis.
The pathogen receptor CEA is a GPI-anchored receptor which is most abundantly found on the apical surface of the gastrointestinal epithelium, but also on other mucosal epithelial cells (Thompson 1995). Interestingly, it was shown that CEA-engagement by CEACAM-binding bacteria, promotes bacterial colonization of the mucosa, due to the suppression of epithelial exfoliation (Muenzner et al. 2010). Accordingly, CEACAM-engagement allows bacteria to permanently colonize the human mucosa. Importantly, CEA is not only expressed as a GPI-anchored protein, it is also secreted as a soluble protein (Yamaguchi and Kawai 1983). The epithelial surface of the gut is separated from the intestinal lumen by a thick mucus layer, making it hard for bacteria to interact with epithelial cells. Since CEA is secreted from epithelial cells, it might stick to the mucus and thus promote interactions with intestinal bacteria from the mucus.
In our study, we enriched a macaque stool sample pool for macaque CEA-binding bacteria. Using 454 pyrosequencing, we identified the genus Prevotella to interact with macaque CEA. Although, Prevotella, which are common intestinal bacteria, are known to adhere to and invade epithelial cells, little is known about the detailed mechanism of interaction with the epithelium (Gursoy et al. 2009). Pulldown experiments with three Prevotella strains, originally isolated from humans, and soluble macaque CEA confirmed the Prevotella-CEACAM interaction obtained from the pyrosequencing results. Surprisingly, additional pulldown analyses also revealed an interaction of Prevotella strains with several human CEACAMs. In sharp contrast, none of these Prevotella
strains interacted with CEACAMs from more distantly related mammals. This is in line with a former study, which demonstrated that CEACAM recognition by pathogens is species-specific (Voges et al. 2010). Unexpectedly, we also observed an interaction of Prevotella and human CEACAM8. For CEACAM8 no bacterial ligands have been identified so far, therefore, it was thought to be a negative control in our screen. In total, we identified 48 specific 16S rRNA sequences as potential CEACAM-binding bacteria, including the Prevotella group. These sequences were found to be enriched in macaque CEA samples compared to the input and to human CEACAM8. However these sequences were also enriched, albeit to a lesser extent, in human CEACAM8 samples compared to the input (data not shown). Therefore, it is feasible that Prevotella bound to macaque CEA and in addition, less strongly, to human CEACAM8.
This result was unexpected, since the amino-terminal domain of CEACAM8 displays various amino acid sequence differences compared to the amino-terminal domains of CEACAMs, known to be engaged by bacteria (Bos et al.
1999). Thus, further studies are required to unambiguously proof that the CEACAM-Prevotella interaction we observed, is specific and independent of carbohydrate moieties.
In addition to the identification of CEACAM-binding bacteria, we also aimed to isolate CEACAM-binding bacteria from macaque and human intestinal flora.
Thus, we plated macaque and human stool samples on selective media, transferred the bacterial colonies on a PVDF membrane and probed this membrane with an anti-Prevotella antibody (bacterial colony dot blot). By this method we were able to isolate several Bacteroides species which are closely related to Prevotella. However, pulldown experiments of these Bacteroides and several CEACAMs did not reveal a clear-cut interaction between Bacteroides and CEACAMs. Therefore, further studies should concentrate on the isolation of Prevotella from stool samples and test their CEACAM-binding capacity.
In summary, we hypothesize that CEACAM-engagement could be a general mechanism, not only for pathogens but also for commensals, to species-specific colonize the mammalian mucosa. Therefore, it would be interesting to screen for CEACAM-binding bacteria in other mammals and to analyze whether they also display a species-specific CEACAM interaction. Taken the wide tissue distributions of CEACAMs into account and the fact that they are highly diverse
in different kind of mammals it is feasible that CEACAMs contribute to the specific microbial make-up of mammalian hosts.
4.6 Acknowledgments
We thank T.F. Meyer (MPI for Infection Biology, Berlin, Germany) for the N. gonorrhoeae MS11 strain used in this study. We also thank R. Hohenberger-Bregger and S. Feindler-Boeckh for expert technical assistance. A.R. is a recipient of a fellowship by the RTG 1331 funded by the Deutsche Forschungsgemeinschaft (DFG). This study was supported by funds from the DFG (Ha2568/6-2) to C.R.H.
5 CHAPTER III
Innate recognition by neutrophil granulocytes differs between Neisseria gonorrhoeae strains
causing local or disseminating infections
Alexandra Roth,1 Corinna Mattheis,1 Petra Muenzner,1 Magnus Unemo,2 and Christof R. Hauck1,3
1 Lehrstuhl für Zellbiologie, Universität Konstanz, 78457 Konstanz, Germany
2 WHO Collaborating Centre for Gonorrhoea and other STIs, Department of Laboratory Medicine, Clinical Microbiology, Örebro University Hospital, Sweden
3 Konstanz Research School Chemical Biology, Universität Konstanz, 78457 Konstanz, Germany
Infect Immun. 2013 Apr 29. [Epub ahead of print]
5.1 Abstract
Members of the carcinoembryonic antigen-related cell adhesion molecule (CEACAM) family serve as cellular receptors for Neisseria gonorrhoeae. More specifically, neisserial colony opacity (OpaCEA) proteins bind to epithelial CEACAMs (CEACAM1, CEA, CEACAM6) to promote bacterial colonization of the mucosa. In contrast, recognition by CEACAM3, expressed on human granulocytes, results in uptake and destruction of OpaCEA-expressing bacteria.
Therefore, CEACAM3-mediated uptake might limit the spread of gonococci.
However, some strains can cause disseminating gonococcal infections (DGI) and it is currently unknown, how these strains escape detection by granulocyte CEACAM3. Therefore, the opa gene loci from N. gonorrhoeae strain VP1, which was derived from a patient with disseminated gonococcal disease, were cloned and constitutively expressed in Escherichia coli. Similar to Opa proteins of the non-disseminating strain MS11, the majority of Opa proteins from strain VP1 bound epithelial CEACAMs and promoted CEACAM-initiated responses by epithelial cells. In sharp contrast to strain MS11, the Opa proteins of strain VP1 failed to interact with the human granulocyte receptor CEACAM3. Accordingly, bacteria expressing VP1 Opa proteins were not taken up by primary human granulocytes and did not trigger a strong oxidative burst. Analysis of Opa variants from four additional clinical DGI isolates again demonstrated a lack of
CEACAM3-binding. In summary, our results reveal that particular N. gonorrhoeae strains express an Opa protein repertoire allowing engagement
of epithelial CEACAMs for successful mucosal colonization, while avoiding recognition and elimination via CEACAM3-mediated phagocytosis. A failure of CEACAM3-mediated innate immune detection might be linked to the ability of gonococci to cause disseminated infections.
5.2 Introduction
Gonorrhea is one of the most common sexually transmitted infections on a global scale. The causative agent of gonorrhea is the human-restricted, Gram-negative bacterium Neisseria gonorrhoeae. In a large proportion of infected people, particularly women, the infection is asymptomatic. Furthermore, in most symptomatic patients the bacteria are limited to the lower urogenital tract, with a characteristic purulent discharge containing a high load of granulocytes as a
hallmark of the infection. Whereas this kind of limited local inflammation can resolve even without treatment, ascending infections of the upper genital tract can lead to complicated disease manifestations such as pelvic inflammatory disease, a serious risk factor for female infertility, and ectopic pregnancy.
Moreover, certain gonococcal strains are able to access and survive in human blood, causing systemic or disseminated gonococcal infections (DGI) with often severe consequences including arthritis, bacterial endocarditis, or meningitis.
The ability of certain gonococcal strains to cause disseminated disease has been attributed to specific traits more often associated with strains isolated from DGI patients compared to strains from patients with local infections. One feature frequently reported for DGI-causing strains is the AHU auxotrophy phenotype (Knapp and Holmes 1975). However, this group of strains seems to represent a clonal lineage, which appears to be absent in some parts of the world (Tapsall et al. 1992; Gutjahr et al. 1997). A more general characteristic of DGI strains is the prevalence of a particular outer membrane porin isoform, PorBIA, which differs from the other existing isoform PorBIB found in the majority of isolates from localized infections (Bohnhoff et al. 1986). The PorBIA–defined serogroup is present in virtually all AHU-strains and in about 60% of the non-AHU strains isolated from DGI patients (Sandstrom et al. 1984; Bohnhoff et al. 1986). The ability of PorBIA to mediate serum resistance by binding to complement factor H and C4bBP (Ram et al. 1998; Ram et al. 2001) as well as its ability to promote host cell invasion under low phosphate conditions (van Putten et al. 1998;
Kuhlewein et al. 2006) might contribute to the prevalence of this outer membrane protein variant in DGI strains. Interestingly, a large fraction (35-40%) of the non-AHU gonococcal strains isolated from DGI patients harbor the more prevalent PorBIB isoform (Bohnhoff et al. 1986). Therefore, also alternative traits might predispose gonococci to cause disseminated forms of disease.
In addition to humoral factors, such as the complement system or anti-gonococcal antibodies, cell-based innate immune mechanisms are thought to limit gonococcal infections. Granulocytes are massively recruited to the sites of infection and are able to recognize and eliminate gonococci in an opsonin-independent manner (Rest and Shafer 1989). A key host factor mediating rapid phagocytosis and destruction of gonococci is CEACAM3, a granulocyte membrane glycoprotein (Pils et al. 2008; Buntru et al. 2012). Recognition by
CEACAM3 depends on the presence of particular colony opacity (Opa) protein variants (Schmitter et al. 2004). Up to 12 distinct opa gene loci are present in the genome of a single N. gonorrhoeae strain. Each locus appears to be constitutively transcribed, but expression of individual Opa proteins is independently regulated on the translational level. This phase variation is due to several pentameric repeat units in the 5’ coding region of each opa gene, which determine the reading frame (Stern et al. 1986). Addition or deletion of pentameric repeats during replication of the bacterial chromosome, most likely due to slipped-strand mispairing, corrects or compromises the reading frame and arrests individual Opa proteins in an “on” or an “off” phase (Meyer et al.
1990).
Opa proteins serve as important adhesins, which mediate intimate attachment of N. gonorrhoeae to host epithelial cells (Hauck and Meyer 2003).
Functional analysis of the complete Opa protein repertoire of gonococcal strain MS11 has demonstrated that one out of the eleven Opa proteins can bind to heparansulphate proteoglycans (HSPGs) or, via recruitment of vitronectin and fibronectin, to host cell integrins (van Putten and Paul 1995; Gomez-Duarte et al. 1997; van Putten et al. 1998) . To indicate their particular binding specificity, Opa proteins binding to HSPGs have also been termed OpaHSPG (Hauck and Meyer 2003). The other ten Opa proteins of MS11 associate with one or several members of the CEACAM family, in particular CEACAM1, CEACAM3, CEA, and CEACAM6 (Bos et al. 1997; Gray-Owen et al. 1997) and have been designated OpaCEA accordingly (Hauck and Meyer 2003). In this respect, gonococcal OpaCEA proteins only bind to human CEACAM family members, not to CEACAM orthologues from other species (Voges et al. 2010). By engaging CEACAM1, CEA, or CEACAM6 on the apical surface of epithelial cells (epithelial CEACAMs), gonococci seem to profit during mucosal colonization (Hauck et al. 2012). In particular, attachment of bacteria to epithelial CEACAMs triggers enhanced integrin-mediated host cell adhesion to the extracellular matrix and counteracts the shedding of infected epithelial cells from the tissue (Muenzner et al. 2005; Muenzner et al. 2010). On the other hand, recognition of OpaCEA proteins by CEACAM3 might limit the spread of gonococci due to granulocyte-mediated opsonin-independent phagocytosis. Indeed, strain MS11, which is associated with localized infections (Swanson et al. 1987; Swanson et
al. 1988), encodes several CEACAM3-binding OpaCEA proteins (Bos et al. 1997;
Gray-Owen et al. 1997). It is conceivable that the CEACAM binding profile, and in particular the lack of CEACAM3-binding Opa proteins might contribute to the ability of certain gonococcal strains to evade opsonin-independent recognition by granulocytes and to cause disseminated disease. However, the redundancy of opa genes in the gonococcal genome and the frequent phase variation of Opa protein expression have impaired a comprehensive analysis of CEACAM-binding Opa proteins.
Therefore, the present study has been initiated to determine the CEACAM-binding profile of the Opa protein repertoire of a DGI strain. Accordingly, we cloned all opa genes of strain VP1, a clinical isolate of a patient with disseminated gonorrhea (Makino et al. 1991). The reading frame of all VP1 Opa proteins was arrested in the “on” phase to circumvent phase variation and individual Opa proteins were heterologously expressed in Escherichia coli.
Similar to the Opa proteins of MS11, most VP1 Opa proteins bound to human CEACAMs. In line with its ability to engage epithelial CEACAMs, VP1 OpaCEA
proteins were able to trigger enhanced host cell adhesion to the extracellular matrix. In contrast, VP1 OpaCEA proteins did not bind CEACAM3 and did not trigger the opsonin-independent uptake and destruction by primary human granulocytes. Analysis of Opa variants from four additional clinical DGI isolates further strengthened the hypothesis that DGI strains can evade CEACAM3-dependent recognition and subsequent elimination by innate immune cells and that a lack of CEACAM3-binding Opa proteins might predispose gonococcal strains to cause disseminated infections.