OUP user menu

Opa proteins and CEACAMs: pathways of immune engagement for pathogenic Neisseria

Manish Sadarangani , Andrew J. Pollard , Scott D. Gray-Owen
DOI: http://dx.doi.org/10.1111/j.1574-6976.2010.00260.x 498-514 First published online: 1 May 2011

Abstract

Neisseria meningitidis and Neisseria gonorrhoeae are globally important pathogens, which in part owe their success to their ability to successfully evade human immune responses over long periods. The phase-variable opacity-associated (Opa) adhesin proteins are a major surface component of these organisms, and are responsible for bacterial adherence and entry into host cells and interactions with the immune system. Most immune interactions are mediated via binding to members of the carcinoembryonic antigen cell adhesion molecule (CEACAM) family. These Opa variants are able to bind to different receptors of the CEACAM family on epithelial cells, neutrophils, and T and B lymphocytes, influencing the innate and adaptive immune responses. Increased epithelial cell adhesion creates the potential for prolonged infection, invasion and dissemination. Furthermore, Opa proteins may inhibit T-lymphocyte activation and proliferation, B-cell antibody production, and innate inflammatory responses by infected epithelia, in addition to conferring increased resistance to antibody-dependent, complement-mediated killing. While vaccines containing Opa proteins could induce adhesion-blocking and bactericidal antibodies, the consequence of CEACAM binding by a candidate Opa-containing vaccine requires further investigation. This review summarizes current knowledge of the immunological consequences of the interaction between meningococcal and gonococcal Opa proteins and human CEACAMs, considering the implications for pathogenesis and vaccine development.

Keywords
  • adhesin
  • epithelial cell
  • gonococcal disease
  • meningococcal disease
  • neutrophil
  • T lymphocyte

Introduction

Neisseria meningitidis and Neisseria gonorrhoeae are important Gram-negative bacterial pathogens for which the only reservoir of infection is humans. Neisseria meningitidis causes approximately 500 000 cases of meningitis and septicaemia worldwide, with a mortality rate of approximately 10% (World Health Organization, 1998). In the United Kingdom, it is the leading infectious cause of childhood death (Office for National Statistics, 2005) and the rates of meningococcal infection during epidemics in the African meningitis belt may be as high as 1000 per 100 000 population (Lapeyssonnie, 1963). Gonococcal infection is the second most common reportable bacterial disease in the United States, with an estimated 600 000 new cases annually. Disease can be localized near the site of infection, for example urethritis and cervicitis, or can be disseminated. The situation is unknown in most developing countries, but the prevalence of gonorrhoea among pregnant women in sub-Saharan Africa is 2–15% (Buve et al., 1993; Gerbase et al., 1998).

Both organisms initiate infection by the colonization of mucosal surfaces – the nasopharynx for the meningococcus and the genitourinary tract for the gonococcus. Invasive disease can develop by breach of the epithelial barrier, followed by purulent inflammation and haematogenous spread. Neisseria meningitidis can persist without causing symptoms for several months (Caugant & Maiden, 2009) and the gonococcus can cause prolonged mucosal infection. This ability to persist relies on their adaptability to the host and their capacity to evade the immune system. Opacity-associated (Opa) adhesin proteins located in the neisserial outer membrane facilitate the interaction of bacteria with a number of host cell types, including epithelial cells on mucosal surfaces and various immune cells, indicating a direct effect on the immune response. Receptor tropism of Opa proteins can be broadly divided into two categories – those that bind to members of the carcinoembryonic antigen cell adhesion molecule (CEACAM) family (Chen & Gotschlich, 1996; Virji et al., 1996a, b, 1999; Bos et al., 1997a; Chen et al., 1997; Gray-Owen et al., 1997a, b; Popp et al., 1999; Muenzner et al., 2000) and those that bind to heparin sulphate proteoglycans (HSPGs) (Chen et al., 1995; van Putten & Paul, 1995), with the former being the major receptors. While other adhesins clearly affect neisserial binding to human cells, this review will focus on the impact of Opa binding to CEACAM family receptors on infection and immunity.

Opa proteins

An important role for Opa during infection is suggested by the observation that Opa+ bacteria are recovered during natural gonococcal infection (James & Swanson, 1978) and following inoculation of humans with Opa bacteria (Swanson et al., 1988; Jerse et al., 1994), the only exception being that Opa phase variants predominate during menses (James & Swanson, 1978). Opa-expressing bacteria tend to resist detergent-mediated solubilization of cell surface components and Opa promotes bacterial aggregation within colonies (Swanson et al., 1978), which presumably explains the inherent resistance of opaque colonies to serum-mediated killing (Bos et al., 1997b). Opa binding to CEACAMs also confers the ability to associate with human mucosal, endothelial and leucocytic cells encountered during neisserial infection, and Opa proteins have the capacity to influence bacterial attachment, engulfment and immunological responses.

Expression of opa genes

Meningococci and gonococci typically encode either four or 11 Opa proteins, respectively, with individual loci dispersed throughout the genome (Bhat et al., 1991; Parkhill et al., 2000; Tettelin et al., 2000). Their name is derived from the fact that their expression often correlates with increased opacity of agar-grown bacterial colonies when viewed by oblique substage lighting due to variances in bacterial aggregation (Bhat et al., 1991; Hobbs et al., 1994). opa genes are constitutively transcribed, but Opa protein expression undergoes phase variation at the translational level due to the presence of the pentameric coding repeat (CR) sequence 5′-CTCTT-3′ within the sequence encoding the amino-terminal leader peptide (Stern et al., 1986). Modulation of the number of repeats occurs by slipped-strand mispairing during DNA replication, resulting in regular frameshifting. The frequency of phase variation in gonococci is estimated to be approximately 10−3 per cell per generation, and occurs independent of homologous recombination (Mayer et al., 1982). In addition, promoter strength influences the phase variation of gonococcal opa genes, in addition to regulating Opa expression levels (Belland et al., 1997). As a result, any given bacterial population can include bacteria expressing none, one or multiple Opa proteins, at varying levels of expression. This contrasts with Opc, another major adhesin, which is not present in the gonococcus, and is encoded by a single gene in the meningococcus. Variable expression of Opc is controlled at the transcriptional level by variations in the length of a poly-cytidine tract in the promoter region. The ability of bacteria to exist in the absence of Opa proteins is presumably a key mechanism of immune evasion, as Opa proteins are known to induce antibodies following meningococcal infection and after immunization with vaccines containing outer membrane complexes or vesicles (Mandrell & Zollinger, 1989; Sjursen et al., 1990; Milagres et al., 1998). Moreover, transparent Opa variants of N. gonorrhoeae predominate during menstruation, but represent a minority of bacteria at other times (James & Swanson, 1978).

Structure and function of Opa proteins

Opa proteins are predicted to contain eight antiparallel β strands forming a barrel structure in the bacterial outer membrane, linked by four extracellular loops (Fig. 1) (de Jonge et al., 2002). Loops 1, 2 and 3 contain variable regions (Malorny et al., 1998), and sequence diversity in these regions confers specificities for host receptors. Sequence variation is observed predominantly within loops 2 and 3, which have been termed hypervariable (HV1 and HV2, respectively). Loop 1 contains some sequence similarities between Opa proteins and is termed semi-variable. Receptor tropism of Opa proteins can be broadly divided into two categories. The first, which is represented by a relatively small number of Opa variants, bind to HSPGs (Chen et al., 1995; van Putten & Paul, 1995; Virji et al., 1999) and extracellular matrix (ECM) proteins, such as vitronectin and fibronectin (Duensing & van Putten, 1997; Gomez-Duarte et al., 1997). Most Opa variants instead bind to the human CEACAM family of receptors expressed by a variety of cell types.

1

Predicted three-dimensional structure of Opa proteins. Predicted eight-stranded β-barrel topology, linked by four putatively surface-exposed loop regions and three periplasmic turns. Variable stretches are located in three of the four loop regions: a semi-variable region is located in loop 1 and hypervariable regions (HV1 and HV2) in loops 2 and 3. Alpha helices are represented in red and β strands in yellow. The figure shows the three-dimensional model of the Neisseria gonorrhoeae Opa58 protein, which was generated using 3d-jigsaw software (Bates et al., 2001) using the neisserial NspA protein crystal structure (Vandeputte-Rutten et al., 2003) as a template. This structure is for illustrative purposes only; the transmembrane region is highly conserved in primary sequence and predicted β-barrel structure, while the extracellular loops of other Opa variants may diverge significantly depending on the length and primary sequence of the surface-exposed sequences. (Figure provided courtesy of Michael Brooks, University of Toronto, Canada.)

Population diversity of Opa

There is a high level of diversity of Opa proteins, generated by a number of mechanisms (Fig. 2). Neisseria are naturally highly transformable bacteria, having the ability to take up extracellular DNA efficiently, and then incorporate it into the chromosome by homologous recombination (Catlin, 1960). This mechanism may have evolved to aid in the repair of damaged chromosomes, as bacteria are most likely to encounter DNA from nearby cells of the same species, but transformation is also a powerful mechanism for generating genetic diversity by horizontal gene exchange. In addition, the presence of multiple opa genes allows genetic recombination between loci in the same chromosome, as well as between different organisms. These recombination events can involve exchange of entire or partial opa genes. Finally, mutations can occur within opa loci during bacterial growth as a further source of diversity. In the case of opa genes, intra- and intergenomic recombination events are much more likely to occur than mutations in both N. meningitidis and N. gonorrhoeae (Hobbs et al., 1994, 1998; Bilek et al., 2009). This was quantified recently in a study of gonococci, which estimated that 77% of diversity was due to recombination between genes of the same isolate, a further 16% due to import of genes from other isolates and only 7% due to de novo mutation (Bilek et al., 2009).

2

Mechanisms of genetic variation of neisserial opa genes. Population diversity of opa genes can be generated by recombination or, less frequently, mutation. In this illustration, the initial strain (A) contains four different opa genes depicted in different colours. Intragenomic recombination can result in the duplication and replacement of an entire opa gene (B) or part of a gene (C). Alternatively, intergenic recombination can occur when the bacteria import exogenous DNA from death or damage of bacteria in the local environment. This can also result in total (D) or partial (E) replacement of one of the opa genes. Homologous recombination is more likely at the same locus due to the presence of similar upstream and downstream sequences, but can also occur at a different locus. Finally, new variants can also occur by de novo mutation (F). Although the figure shows Neisseria meningitidis, with four opa genes, the same principles also apply for Neisseria gonorrhoeae, which typically contain 11 opa genes.

One of the difficulties in quantifying opa diversity has been the lack of a standardized nomenclature for describing alleles (Callaghan et al., 2006). The Opa sequence database sited at the University of Oxford was developed in part to standardize Opa classification. There are currently 338 distinct opa alleles in the database (http://www.neisseria.org, accessed 28 September 2010). These include 26 different semi-variable regions, 96 HV1 variants and 127 HV2 sequences. If all possible combinations of these variable regions existed, there would be over 300 000 opa alleles, suggesting restriction in permissible combinations, which may be due to the stability of the resulting protein structures or a survival advantage of certain variants. Interestingly, opa diversity among meningococci appears to be highly structured at the population level, with clustering of a limited number of allelic variants among clonal complexes determined by multilocus sequence typing (Callaghan et al., 2006, 2008a). This has not been found with studies of gonococcal opa genes, and may reflect the increased number of opa loci present in each chromosome, which particularly increases the opportunity for intragenomic recombination.

Opa diversity may reflect different functions of different Opa variants, or could be the result of immune selection pressure, as has been found for other meningococcal outer membrane proteins, such as PorA (Gupta & Maiden, 2001). Analysis of 216 meningococcal carriage isolates from the Czech Republic revealed the presence of discrete, non-overlapping opa repertoires with low within-repertoire diversity, consistent with a model of strong immune selection where identical alleles can occupy different loci (Callaghan et al., 2008b). This also supports a theory that opa diversity is greater in gonococci because there is less immune pressure due to the lower prevalence of the organism in the population. However, there are no similar studies of gonococcal opa diversity and, conversely, it is also possible that opa diversity in gonococci is greater because of greater immune pressure relating to differences in the transmission of the two organisms. Neisseria meningitidis can infect most young children and therefore has a continuously replenishing population of naïve hosts. Sexually transmitted infections, such as gonorrhoea, however, persist by repeated infections in a relatively small ‘core group’ of sexually active or risky individuals (e.g. commercial sex workers). It is therefore critical that they can continue to evade the immune response to survive in the population, requiring increased opa diversity. Further information on opa diversity and immune responses against both Neisseria species, in particular N. gonorrhoeae, would be required to test these hypotheses.

Opa proteins as a serogroup B meningococcal vaccine candidate

The broad diversity of Opa proteins had led to them being discounted as a potential meningococcal vaccine candidate, despite the presence of anti-Opa antibodies following both invasive disease and vaccination. However, demonstration that only a limited number of Opa variants have been associated with hyperinvasive serogroup B meningococci globally over the last 60 years has renewed interest in their potential as a vaccine (Callaghan et al., 2006). Immunization of mice with recombinant or liposomal Opa proteins elicited high levels of meningococci-specific bactericidal antibody, demonstrating proof in principle of this approach (Callaghan et al., 2008c; Arenas et al., 2010). However, Opa variants do not bind mouse CEACAMs, making it impossible to assess the influence of Opa proteins on the developing immune response in this model. Bactericidal activity against Opa is thought to be mediated by epitopes within the HV1 and HV2 regions (Malorny et al., 1998). Examination of the hypervariable regions from a collection of 227 isolates from invasive meningococcal disease in the United Kingdom between 1996 and 2001 (Callaghan et al., 2008d) provides a theoretical coverage of 90% with a vaccine containing the six most common Opa variants.

CEACAM

CEA, the founding member of the family, was first described following the study of human colon cancer tissue in 1965 (Gold & Freedman, 1965). A total of 22 different genes have now been identified in the human CEA family, divided into the CEACAM and the pregnancy-specific glycoprotein subgroups (http://www.carcinoembryonic-antigen.de, accessed 26 September 2010). The CEACAM subgroup, which contains the Opa protein-binding receptors, contains one secreted and 11 cell surface glycoproteins, which have undergone several changes in nomenclature (Fig. 3). Of these, only CEACAM1, CEACAM3, CEA (or CEACAM5) and CEACAM6 have been shown to bind to neisserial Opa proteins (Chen & Gotschlich, 1996; Virji et al., 1996a, b, 1999; Bos et al., 1997a; Chen et al., 1997; Gray-Owen et al., 1997a, b; Popp et al., 1999; Muenzner et al., 2000).

3

Opa protein-binding receptors of the human CEACAM family. CEACAM proteins consist of a 108-amino acid N-terminal domain homologous to the immunoglobulin-variable domain (shown in red), and between zero and six domains homologous to the immunoglobulin-constant domain of the C2 set (shown in blue) (Williams & Barclay, 1988). The IgC2 domains may either be of type A (93 amino acids) or type B (85 amino acids). There are two types of membrane anchorage observed among the CEA subgroup of CEACAM proteins. CEACAM1 and CEACAM3 contain a hydrophobic transmembrane domain, followed by a cytoplasmic domain (Embedded Image). CEA (also known as CEACAM5) and CEACAM6 are attached to the cell surface via a glycosylphosphatidylinositol moiety (Embedded Image) (Hammarstrom et al., 1999). Embedded Image represent the potential glycosylation sites (Yamashita et al., 1987, 1989). CEACAM1 and CEACAM3 occur in different isoforms, derived by alternative mRNA splicing. The most important differences between these splice variants seem to be the cytoplasmic domain sequence, which determines the presence or absence of immunoreceptor tyrosine-based activation (CEACAM3) or inhibition (CEACAM1) motifs. In all forms, the N-domain is retained (Hammarstrom et al., 1999). (Figure adapted with permission from http://www.carcinoembryonic-antigen.de.)

CEACAM proteins are encoded by the CEA gene family, which are clustered on chromosome 19q13.2 within a region of 1.8 Mbp (Zimmermann et al., 1988). They belong to the immunoglobulin (Ig) superfamily, and consist of an N-terminal domain homologous to the immunoglobulin variable domain, and up to six domains homologous to the immunoglobulin constant domain of the C2 set (Fig. 3) (Oikawa et al., 1987; Paxton et al., 1987). CEACAM1, CEACAM3 and CEACAM7 occur in different forms, derived by alternative mRNA splicing. In all forms, the N-domain is retained. While the effect of modifying the number of IgC-like domains remains unclear, there are obvious effects of varying membrane anchorage and/or cytoplasmic domain sequences. The longer cytoplasmic forms contain two tyrosine residues that are part of modified immunoreceptor tyrosine-based activation or inhibition motifs (ITAM or ITIMs). Phosphorylation of ITAMs and ITIMs results in the binding of protein tyrosine kinases and protein tyrosine phosphatases, respectively, leading to the stimulation or the inhibition of cell-signalling pathways. Naturally occurring ‘short’ isoforms lack the tyrosine-containing sequences and their associated function. The balance between these stimulatory and inhibitory pathways is of prime importance when considering the effect of Opa proteins following binding of CEACAM on the cell surface.

CEACAM1 is present as monomers, dimers and oligomer microclusters in the membrane. Binding of CEACAM1 to intracellular messenger molecules, and therefore downstream signalling, is dependent on the balance between these forms. The N-terminal domain of CEACAM1 mediates trans-homophilic binding between different molecules via an allostery-based mechanism, and increases cis-dimerization, which may therefore be vital to transduction of the transmembrane signal and the cellular consequences of binding (Klaile et al., 2009; Muller et al., 2009).

CEACAM1, the evolutionary progenitor of the CEA family, has the broadest distribution in normal tissues, being expressed in a wide range of epithelial cells, neutrophils, lymphocytes and endothelial cells. CEACAM6 also has a broad tissue distribution, and is present in epithelial cells of different organs and neutrophils. In normal tissue, CEA is predominantly located in epithelial cells of the gastrointestinal tract and within the urogenital tract (Prall et al., 1996), while CEACAM7 appears to be restricted to the gastrointestinal tract. CEA can be expressed at high levels in some tumours. In contrast, CEACAM3 and CEACAM8 are exclusively expressed in neutrophils. CEACAM family molecules facilitate intercellular adhesion through both homophilic (CEACAM1, CEACAM5 and CEACAM6) and heterophilic (CEACAM1–CEACAM5, CEACAM5–CEACAM6 and CEACAM6–CEACAM8) binding. Curiously, CEACAM3 appears to be unique among the group in that it lacks any cell–cell binding activity, its only described binding partner being the CEACAM-specific bacterial adhesins (Chen & Gotschlich, 1996; Billker et al., 2002; Schmitter et al., 2004).

One of the difficulties of examining CEACAM-mediated interactions has been the lack of naturally occurring human CEACAM homologues in animal models of meningococcal and gonococcal infection (http://www.carcinoembryonic-antigen.de, accessed 26 September 2010). The majority of experimental work has therefore been carried out using cell lines that have been transfected to expresss CEACAM proteins. The recent development of transgenic animal models that are able to express human CEACAM should provide a wider perspective in the near future (Bhattacharya-Chatterjee et al., 2008; Gu et al., 2010; Muenzner et al., 2010; Sarantis et al., 2010).

Molecular interactions between Opa and CEACAM

CEACAM family members are differentially recognized by Opa protein variants; hence, the distribution of each CEACAM has the potential to influence the cellular tropism of Neisseria expressing different Opa proteins in vivo (Virji et al., 1996a, 1999; Bos et al., 1997a; Chen et al., 1997; Duensing & van Putten, 1997; Gray-Owen et al., 1997b). Furthermore, each CEACAM mediates different cellular processes (Kuijpers et al., 1992; McCaw et al., 2004; Schmitter et al., 2007b), suggesting that the cellular response to neisserial binding will depend on the specific combination of CEACAMs engaged. Opa–CEACAM interactions occur most effectively with acapsulate organisms (Virji et al., 1996a, 1999), although Opa-mediated adhesion of apparently fully encapsulated bacteria to transfected cells expressing high levels of CEACAM1 has been demonstrated (Virji et al., 1996b; Bradley et al., 2005; Rowe et al., 2007). Binding of CEACAM by Opa proteins requires a conformational interaction between both hypervariable regions of Opa, whereas the semi-variable region appears to be dispensable (Bos et al., 2002) (Fig. 1). The deletion of the conserved fourth loop resulted in a severe decrease in Opa expression; hence, any specific effect it may have on CEACAM binding has not been tested. Antibodies directed against the HV2 region of a gonococcal Opa protein inhibited interactions with neutrophils (Elkins & Rest, 1990) and two meningococcal Opa variants differing only in their HV1 regions varied in their interactions with CEACAM3 and CEACAM6 (Virji et al., 1999), consistent with the importance of both HV1 and HV2 regions. Chimeric Opa variants, containing combinations of hypervariable regions derived from different CEACAM-binding Opa proteins, lost most of their receptor-binding activity, however, suggesting that a specific interaction between HV1 and HV2 is necessary for CEACAM binding (Bos et al., 2002).

The high degree of sequence variation in the hypervariable regions of Opa proteins raises the question of how the binding sites for CEACAM are conserved. Neisseria meningitidis strain H44/76 possesses four different Opa proteins, of which OpaA and OpaJ bind to CEACAM1, while OpaB and OpaD bind to CEACAM1 and CEA. A sequence motif involved in binding to CEACAM1 was identified by alanine scanning mutagenesis of conserved amino acid residues within the hypervariable regions of all four Opa proteins. Hybrid Opa variants with different combinations of HV1 and HV2 showed reduced binding to CEACAM1 and CEA, and strongly suggested the existence of a conserved binding site for CEACAM by the interaction of HV1 and HV2 regions. This included a relatively conserved binding motif in HV2, consisting of Gly172, Ile174 and Gln176 (de Jonge et al., 2003). Similar motifs can be found in almost all CEACAM-binding Opa variants, suggesting that the receptor-binding contacts are conserved in spite of the high degree of Opa sequence variation or that these residues are integral to the functional conformation of all Opa variants.

Opa-expressing isolates of meningococci and gonococci bind to the CEACAM N-domain, which is largely conserved between members of the CEACAM family, and binding occurs to a region within the N-terminal 108 amino acids (Virji et al., 1996a, b; Gray-Owen et al., 1997a; Popp et al., 1999). Recognition is mediated by the CEACAM protein backbone and not by glycosylated moieties; hence, differences in glycosylation should not influence Opa–CEACAM-mediated responses (Virji et al., 1996b; Bos et al., 1998). Using chimeric constructs of different N-domains, distinct binding regions have been identified for specific groups of Opa variants, which are therefore likely to dictate the differential recognition of CEACAM by Opa. The binding surface of the CEACAM1 N-domain is the nonglycosylated face composed of the β strands C″, C′, C, F and G, and similar structures exist on the other Opa-binding receptors of the CEACAM family. This is the region of the highest sequence divergence between human and animal CEACAMs (Voges et al., 2010), which may partly explain why Neisseria are obligate human pathogens, and confirms the difficulty of examining these interactions in animal models. Opa proteins bind to exposed residues of the CC'FG face, and binding appears to require Tyr34 and Ile91 (Virji et al., 1999). Further efficient interaction of distinct Opa proteins depends on different nonadjacent amino acids that spatially lie in close proximity to Tyr34 and Ile91, making continuous conformational binding domains. Variation of an amino acid triplet sequence (residues 27–29) between different CEACAMs is responsible for the differential binding of Opa proteins to CEACAM1, CEA and CEACAM6 (Popp et al., 1999).

CEACAM1 and CEACAM3 have different transmembrane and cytoplasmic domains, and CEA and CEACAM6 are glycosylphosphatidylinositol-anchored; hence, it is likely that distinct processes mediate neisserial uptake following Opa protein binding to each of these receptors. CEACAM-dependent bacterial engulfment has been compared using a panel of transfected HeLa epithelial cell lines, each expressing a different CEACAM (McCaw et al., 2004). CEACAM1 and CEACAM3 each contain proteinaceous transmembrane and cytoplasmic domains, but gonococcal uptake mediated by these receptors differed with respect to their susceptibility to tyrosine kinase inhibitors and the actin microfilament-disrupting agent cytochalasin D, indicating the stimulation of different cell signalling pathways. Bacterial uptake mediated by CEA was reduced by cleavage of the glycosylphosphatidylinositol anchor by phosphatidylinositol-specific phospholipase C after bacterial binding, consistent with a single zipper-like mechanism (McCaw et al., 2004). Regardless of the CEACAM expressed by the transfected HeLa cells, internalized gonococci were effectively killed by a microtubule-dependent process that required acidification of the bacterium-containing phagosome. A more recent study utilized a CEACAM-deficient cell line that was transfected with different CEACAMs. Bacterial internalization via CEACAM1 or CEA, which are present within cholesterol-rich membrane microdomains on the epithlelial cell surface, was inhibited by cholesterol-chelating agents, unlike CEACAM3-mediated internalization (Schmitter et al., 2007b). Given the phase-variable nature of Opa proteins, these results indicate that the mechanism of bacterial engulfment and the cellular response to infection depend on the receptor specificities of the Opa variants expressed and the spectrum of CEACAM present on target cells.

Cellular effects of Opa–CEACAM interactions during infection with pathogenic Neisseria

Epithelial and endothelial cells

The mucosal surface in humans constitutes a very large surface area, continuously exposed to innocuous and commensal bacteria, but is also a portal of entry for potential pathogens. The epithelial cell layer is therefore one of the first lines of defence against infection with pathogenic Neisseria. Initial attachment is mediated by pili, and pilus retraction then allows tight secondary binding via an interaction between Opa proteins and apically expressed CEACAMs. CEACAM binding triggers neisserial engulfment and transcellular transcytosis through polarized epithelia, allowing efficient entry into the subepithelial spaces (Wang et al., 1998). While HSPG binding allows effective neisserial uptake into nonpolarized epithelial cells (Chen et al., 1995; van Putten & Paul, 1995; Freissler et al., 2000), HSPG receptors are not exposed on the apical surface of polarized epithelia and HSPG-specific Opa variants do not mediate apical attachment (Wang et al., 1998). As such, it is enticing to consider that HSPG-mediated attachment promotes the passage of the bacteria from the infected tissues back to the mucosal surface.

Exfoliation is the detachment of infected epithelial cells, and is a first-line innate defence mechanism to prevent bacterial colonization. In the absence of Opa–CEACAM binding, infection with N. gonorrhoeae induced epithelial detachment from an ECM substrate in vitro and exfoliation of mucosal epithelia in vivo, while strains that bind CEACAMs failed to induce detachment and instead promoted enhanced adhesion to the ECM via increased expression of CD105 (Muenzner et al., 2005, 2010). Adherent bacteria then have an opportunity to invade; hence, increased adherence provides a potential mechanism to increase the number of bacteria able to subsequently penetrate into deeper tissues. Transfected Chinese hamster ovary cells expressing high levels of CEACAM1 bound threefold more, but internalized 20-fold more Opa+ encapsulated meningococci than those with intermediate expression (Bradley et al., 2005). No overall selection of the acapsulate phenotype was observed in the internalized population, suggesting that capsule may not be an adequate barrier for cellular interactions mediated by CEACAMs. Upregulation of CEACAM, which can occur in response to proinflammatory cytokines, could therefore lead to the translocation of fully encapsulated bacteria across the mucosal epithelium and into the bloodstream in numbers sufficient to cause a rapid onset of disseminated disease. This implies a rationale for the epidemiological observations that individuals with prior mucosal inflammation, such as respiratory tract infections or during the dry windy season in sub-Saharan Africa, carry an increased risk of invasive meningococcal disease (Greenwood et al., 1984; Moore et al., 1990).

Epithelial cells exposed to interferon-γ (IFN-γ), but not tumour necrosis factor-α (TNFα) or interleukin-1β (IL-1β), supported increased meningococcal adhesion and invasion in vitro, which was abrogated by nuclear factor κB (NFκB) inhibitors (Griffiths et al., 2007). The increase was related to the presence of Opa on the bacterial surface, with de novo synthesis of CEACAM1 occurring in epithelial cells exposed to IFN-γ or infected with Opa-expressing bacteria. CEACAM density was again shown to be important as increased CEACAM expression levels correlated with an increase in bacterial invasion (Griffiths et al., 2007; Rowe et al., 2007). Blocking the Opa–CEACAM interaction reduced invasion and adhesion and there appeared to be a synergistic effect between pili and Opa with regard to the invasion of epithelial cells, independent of capsule type (Rowe et al., 2007). More recently, meningococcal Opa expression was shown to correlate with reduced Toll-like receptor 2 (TLR2)-initiated, NFκB-dependent inflammatory responses of primary pulmonary epithelial cells by recombinant Escherichia coli strains, suggesting that Opa binding to CEACAM1 allows an active suppression of normal innate responses to infection (Slevogt et al., 2008).

CEACAM1 expression on human umbilical vein endothelial cells can be stimulated with the proinflammatory cytokine TNFα (Muenzner et al., 2000, 2001), which is found at increased levels during both meningococcal and gonococcal disease (Girardin et al., 1988; van Deuren et al., 1995; Kirsch et al., 1996). This induced expression of CEACAM1 leads to substantially increased Opa-dependent bacterial binding and invasion into the primary endothelia, implying that these interactions may play an important role in the pathogenesis of invasive disease.

These studies highlight the importance of the local cytokine milieu in determining the pathogenesis of invasive disease, even at this initial stage. CEACAM1-4L and CEACAM1-3L splice variants are both induced with a time course similar to that of proinflammatory cytokines, including TNFα (Muenzner et al., 2001, 2002). Neisserial infection will elevate CEACAM expression, with the resulting increase in bacterial attachment and penetration into the tissues. The continual cycle of infection-induced increases in cytokine and, subsequently, receptor expression would thereby accelerate vascular invasion to promote disseminated disease.

Neutrophils

The majority of Opa-expressing Neisseria interact with neutrophils, hence their original name of leucocyte association proteins (Swanson et al., 1975). Opa mediates intimate attachment of Neisseria to neutrophils, as well as subsequent nonopsonic phagocytosis and stimulation of the bactericidal oxidative burst response (King & Swanson, 1978; Virji & Heckels, 1986; Fischer & Rest, 1988; Elkins & Rest, 1990; Belland et al., 1992; Chen & Gotschlich, 1996; Gray-Owen et al., 1997a, b; McNeil & Virji, 1997; Estabrook et al., 1998; Sarantis & Gray-Owen, 2007). Phagocytosis by and stimulation of neutrophils would seem to confer a selective disadvantage to Opa-expressing bacteria, which would be killed by the bactericidal response. This lends credence to the proposal that neutrophils use CEACAM3 as a decoy to capture bacteria expressing Opa adhesins intended to allow the colonization of other cell types (Schmitter et al., 2004; Sarantis & Gray-Owen, 2007).

Interactions between Neisseria and neutrophils are facilitated by bacterial surface sialic acid (on capsule and lipopolysaccharide) downregulation (McNeil & Virji, 1997). Some Opa variants elicit initial adherence, whereas others promote strong phagocytosis. This nonopsonic interaction with phagocytes is favoured by the meningococcal phenotype, which is often isolated from the nasopharynx (asialylated, piliated, Opa+), whereas the phenotype prevalent in the blood (sialylated, piliated, Opa+) retains the ability to adhere to endothelial cells (via pili), but appears to be refractory to interactions with phagocytic cells.

Neutrophils express CEACAM1, CEACAM3 and CEACAM6, although the potently bactericidal cellular response is driven by Opa binding to CEACAM3, which is only expressed in human neutrophils and to which no nonbacterial ligand has been identified as yet (Nagel et al., 1993). Unlike CEACAM1 and CEACAM6, CEACAM3 bears a functional ITAM that mediates nonopsonic phagocytosis of Opa+ bacteria (Chen & Gotschlich, 1996; Bos et al., 1997a; Gray-Owen et al., 1997a; Virji et al., 1999; McCaw et al., 2003). Opa binding triggers ITAM tyrosine phosphorylation by Src-family kinases, such as Hck and Fgr, and downregulation of SHP-1 tyrosine phosphatase activity. This results in the recruitment of Syk (Sarantis & Gray-Owen, 2007) and the initiation of a cascade of downstream signalling involving phospholipase C, the small GTPase Rac and phosphatidylinositol 3-kinase (Fig. 4) (Hauck et al., 1998, 1999; Chen et al., 2001; Billker et al., 2002; Booth et al., 2003; Schmitter et al., 2004, 2007b; Sarantis & Gray-Owen, 2007). This ultimately promotes the efficient uptake of bound bacteria by an actin microfilament-dependent process. Mutation of the ITAM tyrosine residues abolished the ability of CEACAM3 to transduce signals and mediate phagocytosis because Syk is not engaged (Chen et al., 2001; Billker et al., 2002; McCaw et al., 2003; Sarantis & Gray-Owen, 2007).

4

Opa-mediated internalization of Neisseria by neutrophils via binding to CEACAM3. Following contact of neutrophils by Neisseria, CEACAM3-binding Opa variants recruit receptor molecules to the site of adherence. Receptor ligation by Opa (O) results in the phosphorylation of tyrosine residues in the cytoplasmic ITAM motif of the receptor through Src-family kinases, such as Hck and Fgr. Subsequent recruitment of Syk kinase results in a cascade of downstream signalling, involving the activation of phospholipase C (PLC), phosphatidylinositol 3-kinase (PI3K) and the small GTPase Rac. This leads to actin reorganization and phagocytosis of the organism. The CEACAM3 ITAM can also stimulate Rac via a direct association with the guanine nucleotide exchange factor Vav.

CEACAM3 binding by Opa proteins was the first demonstration of phosphorylation of an ITAM-containing receptor being triggered by direct interaction with a bacterial adhesin (McCaw et al., 2003). A basal level of bacterial uptake continued to occur in the presence of kinase inhibitors and when the ITAM-containing cytoplasmic domain was deleted, indicating that CEACAM3 can mediate the uptake of small (≤1μm) particles via either an efficient, ITAM-dependent mechanism that resembles traditional phagocytosis or a less efficient, tyrosine phosphorylation-independent process (McCaw et al., 2003). More recently, it has been observed that the CEACAM3 ITAM sequence can also stimulate Rac via a direct association with the guanine nucleotide exchange factor Vav to promote rapid phagocytosis and elimination of CEACAM-binding bacteria (Schmitter et al., 2007a). The relative contribution of Syk-dependent (Sarantis & Gray-Owen, 2007) and -independent (Schmitter et al., 2007a) signalling to neisserial engulfment remains to be explored; however, neutrophil bactericidal degranulation and oxidative burst responses downstream of CEACAM3 are clearly Syk dependent (Sarantis & Gray-Owen, 2007).

Acid sphingomyelinase, which augments the conversion of sphingomyelin to ceramide, also plays a crucial role in CEACAM-initiated signalling events and internalization of Opa-expressing N. gonorrhoeae into human neutrophils (Hauck et al., 2000). Rapid formation of ceramide has been shown to result in the generation of ceramide-rich membrane microdomains (Holopainen et al., 1998), which seem to influence the distribution of membrane receptors, especially glycosylphosphatidylinositol-anchored proteins, such as CEACAM6, and the Src tyrosine kinases (Casey et al., 1995; Simons & Ikonen, 1997). It is therefore conceivable that the activation of the acid sphingomyelinase establishes specific membrane properties in the vicinity of the bacteria that ensure a favourable microenvironment for CEACAM–Src family kinase interaction. Activation of the acid sphingomyelinase would then be a prerequisite for efficient signal transmission and bacterial internalization, with the enzyme providing the context for CEACAM-initiated signalling events rather than being directly involved in the signalling cascade.

Many Opa-binding CEACAMs are expressed on neutrophils, and different Opa–CEACAM interactions act synergistically to enable effective adhesion, uptake and stimulation of the oxidative burst response. The competence of individual Opa proteins to interact with CEACAM1 has been correlated with their ability to induce an oxidative response in neutrophils (Gray-Owen et al., 1997b), although in contrast to CEACAM3, for CEACAM1, it is the transmembrane domain that directs internalization of bacteria (Muenzner et al., 2008). Opa-dependent engulfment of Neisseria by neutrophils leads to effective bacterial killing. It is also important to consider that the Opa-dependent binding of outer membrane ‘blebs’, which are naturally released during neisserial infection (Namork & Brandtzaeg, 2002), may carry the potential to activate neutrophils and/or divert the immune response. Different CEACAMs can act independently to mediate these outcomes, but their effects are enhanced when engaged in combination (Skubitz et al., 1996; Skubitz & Skubitz, 2008).

T lymphocytes

The response of T lymphocytes, and particularly CD4+ T lymphocytes, is important during infection with pathogenic Neisseria as these cells are involved in directing the magnitude and quality of humoral immune response. There is good evidence that antibodies directed against surface structures of N. meningitidis are important in immunity, but gonococci do not induce a strong, protective antibody response following infection (Hedges et al., 1999). T lymphocytes are also important in the generation of immunological memory and possibly cell-mediated immunity, which is therefore relevant to vaccine development.

Studies comparing the meningococcal proteins PorA, PorB, Opa and Opc showed that Opa induced the strongest T-lymphocyte proliferative responses and was more immunogenic for T lymphocytes than the other outer membrane proteins (Wiertz et al., 1991, 1996). Recently, live meningococci, heat-killed bacteria and meningococcal outer membrane vesicles (OMVs) have been shown to stimulate CD4+ T-lymphocyte proliferation, independent of their Opa phenotype (Youssef et al., 2009). Moreover, several OMV vaccines, containing variable amounts of Opa, have been proven to be immunogenic in humans (Bjune et al., 1991; Sierra et al., 1991; de Moraes et al., 1992; Noronha et al., 1995; Oster et al., 2005). In other studies, when CD4+ T lymphocytes were exposed to OMVs from Opa-expressing meningococci, their activation and proliferation in response to a variety of stimuli were effectively halted via Opa binding to CEACAM1 (Lee et al., 2007). This effect was specific to CEACAM1-binding Opa variants, and was not observed with non-CEACAM1 binding Opa, or Opa-negative preparations, including OMVs from Opa variants of Neisseria lactamica. To further support this, gonococcal Opa proteins on the surface of whole bacteria were able to suppress the activation and proliferation of CD4+ T lymphocytes via binding to CEACAM1 (Boulton & Gray-Owen, 2002). Infection of CD4+ T lymphocytes with gonococci expressing the HSPG-specific Opa50 resulted in increased expression of the activation marker CD69, which did not occur during infection with bacteria containing the CEACAM1-binding Opa variants. This downregulating effect was accompanied by a reduced proliferative response of CD4+ T lymphocytes to activation stimuli such as IL-2 and/or T-cell receptor (CD3) ligation. These effects resulted from a specific arrest in cell division, rather than from infection-induced cytotoxicity (Boulton & Gray-Owen, 2002).

A suggested mechanism by which Opa proteins could inhibit T lymphocytes is attributed to the coinhibitory function of CEACAM1. Activation of the T-cell receptor results in the stimulation of Src kinase, which phosphorylates the cytoplasmic tyrosine residues of cell surface-expressed CEACAM1 that is bound by Opa (Lee et al., 2008). The phosphorylated ITIM recruits activated tyrosine phosphatases SHP-1 and SHP-2, which effectively oppose the activation response (Chen & Shively, 2004; Chen et al., 2004; Lee et al., 2008) by increasing the activation threshold of the T lymphocytes. An overwhelming stimulus can, therefore, overcome the coinhibitory effect of CEACAM1 (Boulton & Gray-Owen, 2002; Youssef et al., 2009). As always, it is the relative levels of activation and inhibitory signals that determine the cellular response (Fig. 5). It is pertinent to note that the phospho-ITIM-dependent recruitment of SHP-1 upon Opa binding to CEACAM1 also explains the bacteria's ability to suppress TLR2-dependent innate responses in epithelial cells (Slevogt et al., 2008), suggesting that both adaptive and innate immune mechanisms may be depressed.

5

Possible mechanism of Opa-mediated inhibition of CD4+ T lymphocytes via binding to CEACAM1. Activation of the T-cell receptor (TCR) by binding to an antigen presented with major histocompatibility complex (MHC) class II by antigen-presenting cells (APC) results in the stimulation of Src kinase. This leads to tyrosine phosphorylation of CEACAM1 as well as activation of downstream signalling pathways, resulting in T-cell activation. Binding of CEACAM1 by Opa (O) allows the recruitment of tyrosine phosphatases SHP-1 and SHP-2 to the ITIM, which dephosphorylate CEACAM1, the TCR and other intracellular proteins, inhibiting T-cell activation.

There may be a number of reasons to explain the contrasting published data. Purified, denatured Opa proteins induced strong proliferation when incubated with freshly isolated peripheral blood mononuclear cells (PBMCs) (Wiertz et al., 1996), which support cytokine-dependent interactions and receptor signalling between cells. However, studies utilizing whole bacteria or OMVs, which contain Opa in its native conformation but also other potentially important neisserial antigens, have yielded conflicting data (Boulton & Gray-Owen, 2002; Lee et al., 2007; Youssef et al., 2009). It is possible that the different observations simply relate to a variation in the behaviour of different Opa proteins or a variation of their expression levels relative to other outer membrane proteins in the various studies. The experiments with inhibitory effects were performed using purified CD4+ T lymphocytes to specifically investigate Opa–CEACAM effects, and cells were incubated with IL-2 and anti-CD3 and anti-CD28 monoclonal antibodies, rather than a standard antigen proliferation assay (Boulton & Gray-Owen, 2002; Lee et al., 2007). The IL-2-induced upregulation of CEACAM1 surface expression, in combination with cross-linked CD3- and CD28-specific antibodies, may have amplified any Opa-mediated effects. Various other methodological differences in the studies may have also influenced the results. It is possible that purified or recombinant protein may have a different effect in comparison with Opa expressed on the bacterial surface with other proteins, and isolated T lymphocytes will behave differently to PBMCs in vitro. Considering that Neisseria have the capacity to randomly and reversibly switch many surface antigens ‘on’ and ‘off’ by a process of phase variation, it is also possible that the differential expression of uncharacterized factors other than Opa may influence T-cell responses during infection.

It is not clear which of the in vitro conditions used reflects the situation that occurs in vivo in infected patients or immunized individuals, and further investigation of this interaction is necessary to determine the utility of Opa as a vaccine candidate. If Opa does indeed downregulate immune responses in vivo, the deletion of Opa from potential vaccine preparations could improve the immunogenicity of OMVs. Previous clinical trials of OMV vaccines, which contained variable amounts of Opa, have already demonstrated that these vaccines are immunogenic, and that most of the antibody response is directed against PorA (Bjune et al., 1991; Sierra et al., 1991; Rosenqvist et al., 1995; Milagres et al., 1998; Oster et al., 2005, 2007; Martin et al., 2006). There is significant variation in anti-Opa antibody responses, both between studies and among individuals within studies (Poolman et al., 1983; Wedege & Froholm, 1986; Mandrell & Zollinger, 1989). Given the heterogeneous nature of OMV vaccines, it is difficult to verify the key determinants of this response, and therefore the significance of Opa proteins contained in the vaccines. Any proposed OMV- or protein-based meningococcal vaccine would need to demonstrate its ability to elicit bactericidal antibodies in clinical trials to be considered for licensure.

B lymphocytes

Repeated infections among gonorrhoea patients are very common, and the ability of gonococci to inhibit the immune response may be in part due to the suppression of antibody production (Hedges et al., 1999). Opa has been demonstrated to inhibit antibody production from human B lymphocytes that have been induced to express CEACAM1 and then infected (Pantelic et al., 2005). The interaction of N. gonorrhoeae and CEACAM1 on human peripheral B lymphocytes also resulted in induction of cell death. The same findings were obtained in the DT40 B cell line. This CEACAM1-promoted cell death pathway did not involve the inhibitory signals or SHP-1 and SHP-2, but were dependent on Bruton's tyrosine kinase. This observation led the authors to suggest that Opa–CEACAM1 interactions could suppress immunoglobulin expression by killing B lymphocytes (Pantelic et al., 2005); however, it is important to consider that CEACAM1 binding is not generally considered to deliver apoptotic signals in all contexts (Yu et al., 2006). It is, therefore, equally plausible that the Opa-dependent association of these bacteria with B cells may contribute to activation-induced cell death rather than death being a specific consequence of CEACAM1-dependent signals.

The role of other neisserial adhesins during infection

While pili, Opa proteins and Opc proteins remain the most-studied neisserial adhesins, other major neisserial adhesins include NadA, NhhA, App, MspA and HrpA (Peak et al., 2000; Serruto et al., 2003; Capecchi et al., 2005; Turner et al., 2006; Schmitt et al., 2007; Franzoso et al., 2008). To date, these have only been described in N. meningitidis, except App, which is also present in N. gonorrhoeae. Pili are critical in mediating the initial attachment between the organism and epithelial cells, and are also involved in the transformation of exogenous DNA and intercellular signalling. During bacterial colonization at mucosal surfaces, these signalling events enable subsequent, more efficient adhesion between bacteria and host cells, the best studied of which are Opa and Opc. Opc facilitates bacterial binding to epithelial and endothelial cells via HSPGs and the ECM proteins vitronectin and fibronectin. Some studies suggest that Opc is important in meningococcal meningitis, assisting in breach of the endothelial blood–brain barrier. NadA has attracted recent attention as a component of a meningococcal vaccine currently undergoing clinical trials. It aids adhesion and invasion into epithelial cells (Capecchi et al., 2005), activates human monocytes/macrophages in vitro and may play a role in stimulating the inflammatory cascade that occurs during invasive disease (Franzoso et al., 2008). Other adhesins have only recently been identified following genome sequencing of neisserial strains, and much less is known regarding their structure, exact function and importance during infection. NhhA and App are autotransporters involved in adhesion to epithelial cells, MspA is a secreted autotransporter that binds to epithelial and endothelial cells when expressed by E. coli and HrpA forms a two-partner secretion pathway to allow the secretion of very large proteins, and may also be involved in bacterial adhesion to epithelial cells. How these contribute to neisserial infection and disease awaits the identification of their host cellular receptors and downstream effects of bacterial binding.

Concluding remarks

Neisseria meningitidis and N. gonorrhoeae have an incredible capacity to avoid the immune response through antigenic variation and to interact with leucocytes in a manner that prevents an effective immune response. This allows ongoing infection to persist and prevents the development of a memory response that would otherwise protect against subsequent exposures. Herein, we have considered the role that neisserial Opa proteins play in this process. The antigenic variability of Opa proteins allows the avoidance of the humoral response, while Opa binding to HSPG and CEACAM confers the ability to colonize the tissues and, by virtue of the coinhibition of CEACAM1, suppress innate and adaptive responses to infection. While these interactions clearly facilitate infection, humans have responded in evolutionary terms by the acquisition of CEACAM3, a molecular mimic decoy receptor that allows effective capture and killing of Neisseria and other CEACAM-binding bacteria by neutrophils. The lack of a relevant animal model makes it difficult to validate observations made through cellular and molecular studies in vivo. Further work is required to investigate how Opa-mediated interactions integrate with those of other virulence factors during the course of natural infection, as well as the significance of the Opa–CEACAM dependent effects on innate and adaptive immune responses in vivo. While Opa is an attractive and highly immunogenic vaccine candidate that could block colonization and provide protective humoral immunity, it is important to consider the significance of immune-inhibitory interactions with CEACAM1 and the neutrophil-activating activity of CEACAM3 in a development strategy.

Acknowledgements

This work was supported by the Oxford Partnership Comprehensive Biomedical Research Centre Programme with funding from the Department of Health's NIHR Biomedical Research Centres funding scheme. The views expressed in this publication are those of the authors and not necessarily those of the Department of Health. M.S. is supported through a Research Training Fellowship (RTF1263) awarded by Action Medical Research, UK. A.J.P. is a Jenner Institute Investigator. S.D.G.-O. is supported by the Canadian Institutes for Health Research (Grant number MOP-15499) and the Alberta Heritage Foundation for Medical Research Interdisciplinary Team on Vaccine Design and Implementation. This publication made use of the N. meningitidis Opa sequence database (http://neisseria.org/nm/typing/opa/), developed by Keith Jolley and Martin Callaghan and sited at the University of Oxford. A.J.P. acts as chief investigator for clinical trials conducted on behalf of Oxford University, sponsored by vaccine manufacturers (Novartis Vaccines, GlaxoSmithKline, Sanofi-Aventis, Sanofi-Pasteur MSD and Wyeth Vaccines). Industry-sourced honoraria for lecturing or consultancy, travel expenses and grants for education and attendance at scientific meetings are paid directly to an educational/administrative fund held by the Department of Paediatrics, University of Oxford. A.J.P. does not receive any personal payments from vaccine manufacturers. A.J.P. is named as an inventor on a patent for the use of Opa proteins in meningococcal vaccines. S.D.G.-O. is named as an inventor on a patent for the use of modified OMV-derived vaccines lacking CEACAM-binding Opa proteins. M.S. has no conflicts of interest to declare.

Footnotes

  • Editor: Christroph Dehio

References

View Abstract