Introduction

Bacterial outer membrane vesicles (OMVs) are spherical, bilayered, membranous structures released from the surfaces of Gram-negative bacteria [1]. Their first description dates back more than 40 years, when Bishop et al. analyzed the culture supernatants of lysine-requiring Escherichia coli mutants [2]. The scientists were perplexed to find that these mutants under lysine-limiting conditions produced an excess of material they at that time called “extracellular lipoglycopeptide.” Further biochemical and ultrastructural studies revealed that this material was in its composition and appearance similar to the bacterial outer membrane and that the OMVs, or blebs, were produced under common laboratory culture conditions [3–6]. It has later been shown that these vesicles are not products of cell lysis but that they are generated during logarithmic cell growth as well as in stationary phase (Fig. 1). OMVs are sized between 50 and 250 nm in diameter and contain componens of the outer membrane of the bacteria, like lipopolysaccharide (LPS), lipids, adhesins, invasins, toxins, other virulence factors, and sometimes even bacterial DNA. They are associated with a broad array of functions ranging from virulence and metabolism to interspecies communication [7, 8].

Fig. 1
figure 1

Moraxella catarrhalis produces outer membrane vesicles (OMVs) or “blebs” both in vitro as well as in vivo. a Large amounts of OMVs can be seen as small blebs protruding from the bacterial cells. b Direct examination of M. catarrhalis shows secretion of OMVs carrying UspA1/A2 as revealed by a gold-conjugated rabbit anti-UspA1/A2 antiserum. c–e Secretion of OMVs during Moraxella infection in a 9-year-old child with sinusitis. In c, a bacterium and OMVs are seen in the vicinity of a polymorph. In e, UspA1/A2 expression in vivo is verified by the gold-labeled antiserum directed against UspA1/A2. All panels represent pictures obtained by transmission electron microscopy. Adapted from [21] ©2007 by the Infectious Diseases Society of America. Reproduced by permission

Full size image

Biogenesis of OMVs

Despite the early discovery and the extensive research on the isolation of OMVs, the molecular mechanisms behind the production and shedding of OMVs are still not completely understood. A recent review depicts in detail this particular aspect of OMVs [9]. Hence, only an introductory overview of OMV biogenesis will be presented in this chapter.

The production of OMVs is not limited to certain growth conditions, and OMVs can be observed with bacteria growing in vitro on solid agar [10], in liquid medium [11–14], within biofilms [15, 16], during intracellular infection [17], as well as in vivo [18–21]. Nevertheless, the amount of OMV production can vary due to environmental changes. In particular, stress factors such as treatment with the antibiotics gentamicin [22] or chloramphenicol [5], limitation of lysine [2, 4], Mg2+ [23], hemin [24] or phage infections [25], as well as envelope stress due to the accumulation of unfolded proteins in the periplasm [26] cause an overproduction of OMVs.

Any structural protein machinery solely dedicated to the generation of OMVs has not been described to date. Studies in this direction are further complicated by the fact that no mutant bacteria exist, which is completely devoid of vesicle production [27, 28]. Generally, it is assumed that detachment of the outer membrane from the underlying peptidoglycan (PG) is the initial step for the formation of a vesicle [29]. This can be favored at regions where the growth of the outer membrane is faster than the cell wall. These so-called insertion sites where newly synthesized material is incorporated into the outer membrane exist along the bacterial cell membrane at distinct foci [30]. In another model, imbalance in the turnover of PG leads to accumulation of low molecular weight muramic acid, which causes a bulging of the outer membrane by the turgor generated [31]. This is supported by the fact that OMVs contain low molecular weight muramic acid [6, 22, 31]. Finally, other studies showed that also the negative charged nature of the outer membrane might influence the OMV production. For example, Pseudomonas aeruginosa possesses a higher blebbing rate when the bacterium has an outer membrane enriched in more negatively charged B-band LPS presumably due to the stronger repellent forces generated [22, 32, 33]. This is also in accordance with previous observations that limitation of Mg2+, a stabilizer of lipid membranes, enhances vesiculation [23, 34].

Although the initiation of OMV formation can be described by different models without really excluding each other, the localized absence of outer membrane proteins covalently or non-covalently bound to PG, the inner membrane, or other membrane proteins is most certainly crucial for the completion of the process. One early observation supporting this was that lipoproteins were underrepresented in OMVs compared with the outer membrane [29, 35]. Subsequent and independent studies in E. coli [28, 36], Porphyromonas gingivalis [37], and Salmonella typhimurium [38] concordantly identified OmpA, LppAB, and TolA/B as major membrane proteins that stabilize the outer membrane via protein–protein or protein–PG interactions. Deletion of these either increased the vesiculation rate or resulted in oversized vesicles. Intriguingly, Neisseria meningitidis, which is known to produce high amounts of OMVs, seems to lack homologues of the aforementioned membrane proteins [38].

Bacteria produce OMVs in vivo during infection

Several examples exist of interactions in vitro between OMVs and both gastric and mucosal epithelium, as well as several parts of the immune system. The relevance of these studies depends on the assumption that OMVs are produced in vivo, which in fact has been shown in a number of studies. Probably, the earliest proof for the in vivo production of OMVs came from an infant suffering from N. meningitidis. Meningococci and their blebs were observed in negatively stained cerebrospinal fluid of this patient and were proposed to contribute to the disease [20]. In another study, LPS in plasma from a lethal shock patient was quantified using gas chromatography and mass spectrometry. Some of the LPS circulated in plasma as complexes with high sedimentation coefficients, almost always in combination with outer membrane fragments, indicating the presence of OMVs in the serum [18]. Also in another case study where high levels of endotoxins were found in the serum of a severe shock patient with multiple organ failure, cultures were positive for N. meningitidis serogroup B and the in vivo release of OMVs could be proven by electron microscopy. Again the authors concluded that LPS-carrying OMVs contributed to the fatal endotoxin levels in the serum of the patient [39]. Furthermore, it was shown that OMVs were produced by clinical Helicobacter pylori isolates in vivo, and that these contained serologically recognizable Lewis antigens on their surface [19]. Another example for the in vivo production of OMVs comes from a study, where nasal swaps of a child suffering from Moraxella sinusitis were analyzed by electron microscopy (Fig. 1c–e) [21]. In this case, the co-existence of OMVs together with extracellular bacteria and host immune cells could clearly be shown. Also the presence of ubiquitous surface proteins (Usp) A1 and A2, two major virulence factors of Moraxella catarrhalis, on these vesicles was demonstrated by using gold-labeled antibodies. In a more recent publication from our laboratory, it was also demonstrated that the superantigen Moraxella IgD-binding protein (MID) exists on OMVs seen in children [40].

OMVs deliver virulence factors and contribute to bacterial colonization

The discovery that bacteria are able to produce considerably large amounts of OMVs was initially puzzling, since this must come along with high rates of lipid and protein biosynthesis, which should cause a homeostatic disadvantage for the bacterium in question. However, the more the biological functions associated with OMVs were unraveled the more it became apparent that OMVs are crucial for the successful establishment of a bacterial species in its anatomical niche. Thus, in the past, much research focused on OMVs of bacterial pathogens and their functions relevant for virulence.

In order to colonize the host, it is essential for the bacteria to adhere to the epithelial cell membrane in the gut or in the respiratory tract in a way that it also resists physical removal. A wide array of bacterial virulence factors is used by the bacteria, which help them to invade cells, compete with the host for iron and other nutrients and evade or counteract the innate immune defense system of the host. But what is the function of OMVs in the context of bacterial infection?

It has been hypothesized that OMVs can act as an alternative secretion system, similar to type III secretion systems (TIIISS) of bacteria, like Salmonella and Yersinia, or type IV (TIVSS) secretion systems of Legionella or Helicobacter, which are utilized by these species to colonize their hosts or establish intracellular infection [41, 42]. Similarly, OMVs can be considered as a secretion system, delivering proteins and virulence factors to target cells. Clearly exemplifying the role of OMVs as a transport system, the OMVs from the Gram-negative pathogen Shigella flexneri was shown to pick up and entrap the antibiotic gentamicin, and subsequently deliver it to the cytoplasm of host epithelial cells [43]. In contrast to TIIISS or TIVSS, OMVs can act over a long distance, allowing the pathogen to hide at a distance from the site of infection, still allowing it to evade or distract the host immune response. By using OMVs, the pathogen is also able to inject virulence factors across host membranes and into compartments of cells, leading to destruction of the target cells.

Another advantage of OMVs is that they may provide means for bacterial proteins to be transported in complex with other bacterial factors while being protected from degradation. In OMVs, bacteria can transport toxins, enzymes, as well as DNA to eukaryotic cells in a protected manner. Several studies have aimed at comparing the protein composition of OMVs and the outer membranes of the pathogen they derive from (Table 1). For example, a proteomic analysis of H. pylori OMVs showed that the adhesins BabA and SabA were more abundant in the outer membrane compared to the OMVs [44]. In contrast, the chaperone and serine protease HtrA was enriched in the vesicles compared to the outer membrane. In another recent study, H. pylori HtrA was shown to be a potentially important virulence factor since this protein cleaves the cell adhesion protein E-cadherin. This leads to a disruption of the barrier functions of the epithelium, allowing the pathogen to access the intercellular space [45].

Table 1 Virulence factors delivered by OMVs
Full size table

An important aspect of OMVs is that they have been found to contain RNA as well as circular or linear DNA [46–49]. Vesicles from P. aeruginosa, Haemophilus influenzae, M. catarrhalis, and Neisseria gonorrhoeae were all found to contain DNA in various forms, either surface-associated or inside the lumen of the OMVs. Vesicles from N. gonorrhoeae can transfer plasmid DNA both from inside the bacteria as well as exogenous DNA taken up from the extracellular space to other gonococci [50]. Renelli et al. in turn showed that in P. aeruginosa OMVs, DNA could be derived from the periplasm of the cell or the extracellular space. However, the authors were not able to transform neither Pseudomonas nor E. coli cells under a variety of transforming conditions [49]. In the case of Moraxella, this species spreads CpG-motif containing bacterial DNA attached on the surface of OMVs [40]. Any DNA cannot, however, be found inside the vesicles that are derived from M. catarrhalis.

OMVs can also be found in biofilms in natural environments and may contribute to the establishment of biofilms by mediating adherence, delivering extracellular material, or utilizing growth factors [1, 15, 16, 51]. The heme-utilization protein HmuY of P. gingivalis was shown to be associated with OMVs and beneficial for the biofilm formation [52]. Kulp and Kuehn hypothesized that OMVs are an important part of the microflora as part of the biofilm [53]. However, these authors also claimed that OMVs lyse surrounding foreign bacteria in the biofilm, in a situation when nutrients are scarce and during hostile growth conditions. Using a peptidoglycan hydrolase, as it is the case for P. aeruginosa, this would give the OMV-producing bacterial strain an important survival advantage at the expense of the surrounding bacteria [53, 54]. On the other hand, OMVs from species such as P. aeruginosa and M. catarrhalis have been established to carry proteins of the β-lactamase family, which break down β-lactam rings such as those found in amoxicillin [55] [Schaar et al., in preparation]. By this mechanism, the bacteria might transfer antibiotic resistance in biofilms, as well as protect other bacteria from antibiotic killing.

OMV-dependent interaction with the mucosa and epithelial cells

Outer membrane vesicles have a protein composition reflecting that of the outer membrane of the parent bacteria. Considering this, OMVs have the potential to bind to epithelial cells in the respiratory tract as well as in the mucosa, anatomical sites where bacteria interact with their host.

H. pylori is an illustrative example of a pathogen that is residing in the mucosa and colonizes the gut of about 50% of the human population. In most patients, this infection is asymptomatic; but in about 20% of patients, the infection leads to chronic gastritis, as well as, an increased risk of developing peptic ulcer disease, gastric lymphoma, or gastric carcinoma. Surprisingly, it has been found that H. pylori are a non-invasive pathogen and even though the species has the ability to adhere to epithelial cells it is often found to remain unattached. However, LPS-enriched OMVs secreted from the bacteria bind to and invade gastric epithelial cells partly through clathrin-mediated endocytosis. By adherence through adhesins BabA and SabA and the vacuolating cytotoxin VacA, OMV can help cause inflammation by the persistent delivery of antigens like proteases and urease to the gastric mucosa [44, 56]. OMVs have been found to modulate epithelial cell proliferation and apoptosis and stimulate secretion of inflammatory cytokines.

Another example where OMVs bind to epithelial cells is found with M. catarrhalis, a respiratory pathogen commonly causing otitis media in children and exacerbations in patients with COPD. Moraxella OMVs bind to TLR2 in lipid rafts of the epithelial cells, which are subdomains of the membrane with distinct protein and lipid compositions and enriched in cholesterol and sphingolipids [57]. The OMVs are subsequently internalized, causing a pro-inflammatory response resulting in increased IL-8 secretion and ICAM-1 expression. OMVs from M. catarrhalis contain among other proteins two of the main virulence factors of Moraxella, the adhesin UspA1 and the superantigen MID [58]. Slevogt et al. have previously shown that UspA1 interacts with the adhesion molecule CEACAM-1 on human cells, thereby interrupting the TLR2-dependent activation of the transcription factor NF-κB, which in turn leads to a pro-inflammatory response by epithelial cells [59]. Supporting these findings, our experiments revealed that OMVs deficient of UspA1 enhance the pro-inflammatory response. UspA1 in OMVs might thus have an important role in helping Moraxella to evade the host immune response [58]. Hence, OMVs can be seen as bacterial tools to interact with and regulate the inflammatory response of epithelial cells at a distant site from the localization of the parent bacteria.

Attachment of Legionella pneumophila OMVs to A549 lung alveolar epithelial cells has been observed, although the molecular mechanisms remain to be solved [17]. Following attachment of OMVs, the production of IL-6, 7, 8, and 13 as well as GM-CSF, IFN-γ, and MCP-1 was observed. Interestingly, IL-7 and the anti-inflammatory IL-13 were specific to the OMVs and were not produced when epithelial cells were incubated with whole bacteria. It was postulated that this could be due to the altered LPS composition of the OMVs, since EDTA and protease treatment did not alter the cytokine profile. OMVs were also observed in the phagosomes of internalized Legionella suggesting that OMVs might play a role for the intracellular survival of the bacteria. Once taken up, L. pneumophila typically circumvents phagolysosomal degradation with the help of virulence factors, which are secreted by the TIVSS into the cytosol of the host cell. However, also OMVs were shown to inhibit the fusion of the Legionella-containing phagosome with lysosomes independent of the TIVSS [60].

OMVs from several species including P. aeruginosa, Borrelia burgdorferi, Vibrio cholerae, and Vibrio vulnificus have all been shown to bind epithelial cells in vitro, delivering bacterial virulence factors to the membrane of host cells [55, 61–64]. Moreover, OMVs from V. vulnificus were found to cause cytotoxicity and induce cell death, mediated by the cytolysin VvhA [63]. In H. pylori, N. meningitidis and P. aeruginosa OMVs were shown to deliver peptidoglycan into epithelial cells, where it is recognized by the intracellular immune receptor NOD1. This in turn initiated an inflammatory response by the epithelial cells in form of IL-8 production (Fig. 2c). In H. pylori, similar to L. pneumophila, vesicles were identified as an alternative to the TIVSS, in this case for delivering peptidoglycan [65].

Fig. 2
figure 2

Bacterial OMVs mount various immune responses. a M. catarrhalis OMVs interact via the IgD-binding protein MID with surface IgD on B cells and are taken up by the additional interaction of lipoproteins with TLR2. This leads to lipid raft formation and subsequent uptake. TLR9, which recognizes OMV-associated DNA, together with TLR2 activates the B cells non-immunogenically resulting in IL-6 secretion. Additionally, B cells start producing non-specific IgM, which are not directed against Moraxella [40]. b Salmonella vesicles can be taken up by macrophages or dendritic cells (DC). This leads to activation and production of TNF-α, IL-12, and nitric oxide (NO). The subsequent endosomal processing of OMVs leads to upregulation of MHC class II and CD86 on the surface of macrophages and DC and triggers cell-mediated immunity [27]. c The uptake of OMVs by epithelial cells is in many bacterial species dependent on lipid raft formation and leads to a general pro-inflammatory response via NF-κB and p38-mediated signaling and IL-8 production. OMVs were shown to promote this response partly by the delivery of peptidoglycan, which is recognized by the intracellular PAMP-receptor NOD1 [65]. For further references see also Table 1

Full size image

Enterotoxigenic E. coli, a major cause of diarrhea and infant mortality in third world countries, was also analyzed for OMV binding to epithelial cells. The heat-labile enterotoxin LT is secreted by the outer membrane of the bacteria as well as OMVs. LT-containing vesicles and OMVs lacking the toxin were compared for their capacity to invade adrenal and intestinal epithelial cells. It was clearly shown that only LT-carrying vesicles have the ability to invade cells in a time-, temperature-, and receptor-dependent manner. The vesicles attached to the cells at 4°C, and at 37°C, the LT was internalized [66]. This was mediated through cholesterol-rich lipid rafts, and OMVs sometimes co-localized with the endocytosis protein caveolin. Kesty et al. concluded that the OMVs are targeted transport vesicles of the enterotoxin.

Further investigating the mechanism behind uptake of OMVs, Furuta et al. characterized in detail how OMVs from P. gingivalis are taken up and sorted by human epithelial cells [67]. They established that OMVs bound to epithelial cell membranes in a fimbria-dependent manner and were endocytosed through actin-dependent lipid raft-mediated pathways, involving caveolin and dynamin. It was established that OMVs intracellularly were transported to lysosomal compartments where they were eventually degraded. However, OMVs persisted inside the cells for up to 24 h, and caused a large induction of acidic cellular compartments, suggesting cellular stress and impairment. In the case of P. gingivalis, this might suggest a possible mechanism for destruction of the peridontal tissue in patients with chronic peridontitis, eventually leading to tooth exfoliation. OMVs from P. gingivalis thus also exemplify a general mechanism whereby bacteria use OMVs as a powerful offensive weapon to survive within and in some cases fight their host.

Interaction of OMVs with the innate and adaptive immune system

Bacterial colonization of the human body is always accompanied by a host immune response. As a principle, bacteria interact with the human immune system in a number of ways and also use different proteins for binding the same protein or receptor, an efficient strategy resulting in several backup systems. Important parts of this immune response are phagocytic cells such as neutrophils and macrophages, activation of cells of the adaptive immune response including B and T cells, as well as activation of the complement system. In turn, bacteria have evolved different ways to survive inside the host in order to initiate colonization. Outer membrane vesicles have an important role in interacting with the innate immune system, and act both as pro- and anti-inflammatory mediators, perhaps depending on the specific strain and environmental circumstances (Fig. 2).

Several studies have shown that OMVs are decoys for the immune system by redirecting the inflammatory response. A specific example of OMVs interacting directly with immune cells has been shown with M. catarrhalis that activates B cells through a T cell independent mechanism. The interaction between the superantigen MID expressed on OMVs and IgD B cell receptors on the cell surface leads to receptor clustering in lipid rafts, and subsequently, uptake of the OMVs [40]. The membrane of the OMVs interacts with TLR2 and TLR9 sensing lipoproteins and DNA, respectively. This MID-dependent cross-linking of the IgD B cell receptor as well as TLR activation results in IL-6 synthesis and polyclonal IgM production (Fig. 2a). Interestingly, Moraxella OMVs containing MID as well as DNA could be found in specimens from a 6-year-old child, also proving this concept in vivo [40].

In the case of Moraxella, we demonstrated that OMVs interact with the alternative pathway of the complement system, a central part of the innate immunity. Three major pathways are recognized when the complement system is activated [68, 69]. These are (a) the classical pathway, activated by bacterial LPS, nucleic acids, and the antigen–antibody complex, (b) the lectin-mediated pathway, activated by binding of mannose-binding lectin to mannose residues on the bacterial surface, and finally, (c) the alternative pathway, activated by deposition of C3b after recognizing pathogen-associated molecular patterns at the bacterial surface and LPS. Pathogen recognition leads to opsonization of the pathogen and subsequently phagocytosis [70, 71]. In addition, all three pathways eventually lead to the common terminal pathway resulting in formation of a membrane-attack complex [69]. We found that Moraxella OMVs contain the virulence factors UspA1/A2. This multifunctional family of proteins in a dose-dependent manner binds and inactivates complement factor C3, one of the major players in the complement system, and hereby, restraining complement activation [21, 69, 72, 73]. Interestingly, we also observed that Moraxella OMVs carrying UspA1/A2 protected the respiratory pathogen H. influenzae from complement-mediated killing by a mechanism comprising inactivation of C3 [21]. These results suggest a novel, OMV-mediated strategy by which bacteria collaborate with each other in order to defeat innate immunity. Considering the importance of the complement system as an essential part of the innate immune system, leading to lysis of pathogens, opsonization, and an increase in the general inflammatory response, the described mechanisms using OMVs as a “first line of attack” give the bacteria an important advantage in evading the early innate immune response.

S. typhimurium produces OMVs that stimulate professional antigen presenting cells, like macrophages and dendritic cells in vitro [27, 74]. These cells showed an increased expression in MHC class II and induced production of inflammatory cytokines like TNF-α and IL-12 as well as the toxic nitric oxide through a TLR4-dependent and independent signaling pathway. The antigens on Salmonella OMVs were also found to have the capacity to prime both B cells and T helper cells (Fig. 2b).

H. pylori represents another way of OMV-dependent interaction with the immune system. In initial immunization studies with H. pylori LPS, the antibodies that were produced were found to be reactive against Lewis antigen structures on human blood cells. These antigens also commonly occur in the gastric mucosa [75]. Hence, H. pylori seems to express a protein similar to the Lewis structures in some cases of gastritis, thereby creating an autoimmune response in the host via molecular mimicry [75]. Later, it was determined that these antigens could also be found on OMVs secreted from the bacteria, specifically in cases of chronic gastritis and gastric cancer [19]. Additionally, Helicobacter OMVs, like shown for other bacteria, can be taken up by epithelial cells via lipid rafts. This leads to a general inflammatory response and the production of IL-8 (Fig. 2c).

Potential of OMVs as non-replicating vaccines

Bacterial infections are still one of the major causes for hospitalization and death worldwide [76–78]. The treatment of bacterial diseases is hampered due to an increased and rapid spreading of antibiotic resistant bacteria. Hence, vaccines are considered as one of the most straightforward strategies of the post-antibiotic era in the treatment of bacterial diseases [79, 80]. Soon after the discovery that OMVs carry antigenetically active virulence factors, their potential as non-replicating vaccines has become a major interest of immunotherapeutic research [81–85]. The number of studies focusing on OMVs as vaccination tools is steadily growing, and examples from many laboratories including the already existing OMV vaccines against N. meningitidis support the feasibility of this approach (Table 2). A closer look at the development of neisserial vaccines gives also an insight into the potentials and challenges of OMV-based vaccines.

Table 2 Experimental evaluation of OMVs as vaccination tools
Full size table

N. meningitidis is an encapsulated Gram-negative diplococcus, which colonizes the nasopharyngeal mucosa of 5–10% of human beings without causing disease. However, N. meningitidis is also the major cause for bacterial meningitis epidemics affecting especially children beyond the neonatal period and young adults up to the age of 29 [86]. Its polysaccharide capsule allows the distinction between 13 different serogroups, of which the groups A, B, C, W135, and Y account for 90% of the cases. Out of these, group A meningococci are mostly associated with broad epidemics especially in the sub-Saharan region. On the other hand, in temperate and industrialized countries of North America and Europe serogroup B meningococci are the main cause for N. meningitidis accounting for 30–80% of the cases followed by serogroup C [86, 87]. Another striking feature of serogroup B epidemics is that they can persist for a decade and thus much longer than the epidemics of other serogroups.

The polysaccharide capsule proved to be a valuable tool for vaccination purposes in case of meningococci except for the serogroup B. Introduction of the polysaccharide conjugate vaccine against serogroup C (MenC) reduced the incidence by 90% [88, 89]. Similarly, a tetravalent vaccine (MCV4), where polysaccharide of the serogroups A, C, W135, and Y were covalently linked to diphtheria toxin, has been successfully applied in the USA for the age group 11–55 years old [90–92]. Hence, conjugate vaccines are effective in controlling meningococcal disease by inducing the production of bactericidal antibodies with a long lasting immunity due to the elicited T cell-dependent immune response [86].

The success of conjugate vaccines could, however, not be repeated in case of serogroup B meningococci. In early attempts, it was shown that polysaccharide of this serogroup fails to boost an immunological response [93]. Also the coupling to tetanus toxoid did not improve the antigenicity [94]. The reason for this lies in the nature of the capsular polysaccharide of serogroup B, which contains homopolymers of α2-8 linked sialic acid and by this closely resembles the oligosaccharide side chains of human brain tissue [95]. Also the approach of using outer membrane proteins as vaccines had to be abandoned due to low immunogenic response [85]. These drawbacks together with the increase in epidemics in Cuba, Norway, and New Zealand catalyzed the development of OMVs from epidemic disease strains as a last resort against serogroup B meningococci [96–99]. This finally led to the approval of the first OMV-based vaccine in Cuba, which achieved vaccination efficacies ranging from 83% to 94% with low reactogenicity and long-term protection [98], followed by trials in Norway and Brazil as well as the introduction of the MeNZB vaccine in New Zealand [100].

The initial OMV vaccine formulations based on the preparation of culture supernatants of the regional epidemic strain, and applied detergent extraction to remove excess LPS [87, 100]. Although this method aimed at reducing the endotoxicity and increasing tolerability during vaccination, it also had the unwanted side effect of removing most of the lipoproteins leaving basically PorA as the predominant antigenic molecule of the vaccine [101]. However, this neisserial porin is highly variable, and therefore, the bactericidal antibody response generated by a wild-type OMV vaccine confers protection only to one epidemic serotype [102, 103], which is currently the biggest obstacle in the development of a globally effective serogroup B vaccine. The next possible approach of mixing OMV preparations of different disease causing serotypes is due to the high variability of PorA not feasible. There are over 600 PorA variants, and, e.g., a vaccine covering 80% of serogroup B disease causing strains in the USA would need to include 20 different variants [104, 105]. Additionally, the protection of children under the age of 4 cannot be achieved with PorA dominant OMV vaccines [86]. Therefore, much effort is put into improving the efficacy of OMV vaccines by several approaches. Given that PorA alone is not enough for mounting a good protection, the search for alternative OMV-associated antigens has led to the identification of the neisserial adhesin A (NadA), the factor H binding protein (fHBP), neisserial heparin binding antigen (NHBA), or the neisserial surface protein A (NspA) [106, 107]. The combination of recombinantly expressed NadA, fHBP, and NHBA with OMVs of the New Zealand epidemic strain proved to be more efficient in the immunization of infants compared to OMVs or recombinant proteins alone [106].

Another recombinant approach aims at improving the isolation process of OMVs so that the usage of detergents for depleting LPS can be avoided, since this removes also most of the lipoproteins and by this, potential antigens. In three recent publications, these efforts towards the generation of native OMV vaccines were presented [101, 108, 109]. Deletion of the LPS-biosynthesis gene lpxL1 led in both cases to the production of shorter LPS with less toxicity, nevertheless, retaining its adjuvant activity. Additional deletion of rmpM increased the OMV yield, since this mutation causes an easier detachment of OMVs from the outer membrane [101, 110]. In a phase 1 of the clinical trial, the safety and immunogenicity of OMVs isolated form a lpxL2 and synX negative mutant epidemic strain 44/76 was assessed, and the tolerability of native vesicles with a due to lpxL2 deletion modified LPS in human volunteers was shown [108]. In the same study, serum bactericidal activity together with antibodies against LPS and the outer membrane protein OpcA could be detected. Deletion of the sialic acid synthesis gene synX was also applied in the third study, where the authors in addition to the aforementioned lpxL1 deleted a further LPS modification gene lgtA [109]. Interfering with the sialic acid synthesis, although it negatively affected the purification process and OMV yield, was introduced in order to prevent the production of human cross-reactive antibodies. Again LPS was the dominant antigen leading to the production of bactericidal antibodies in mice. The authors also went a step further and introduced recombinant fHBP and NadA variants into the tested vaccine strain and could confirm that NadA was also a potent antigen while antibody response towards fHBP was comparably lower [109].

The recent developments in neisserial OMV vaccines support the potential of OMVs as an alternative way of vaccination and also give an insight into the basics for OMV-based vaccinology, which then can be transferred to other bacterial pathogens. It is therefore not surprising that in case of many other pathogens the interest in identifying the vaccine potential of OMVs or OMV delivered antigens is growing.

While the exact immunological mechanisms still need to be elucidated in detail for each tested pathogen, a general observation is that the immune protection is mainly based upon bactericidal antibodies directed against one dominant or multiple antigens delivered by the OMVs (Table 2). This is clearly shown by a study, where S. enterica Typhimurium was genetically modified, so that it produced less amounts of toxic LPS. In addition, a viral epitope of the canine parvovirus (VP2) was introduced. When mice were immunized with these genetically modified OMVs bactericidal antibodies against Salmonella as well as neutralizing antibodies against VP2 could be detected [111]. In case of V. cholerae, several consecutive studies have proven that immunizing mice or rabbits by administering OMVs via several routes conferred long-term protection against the pathogen by the production of bactericidal IgA and IgG1. In this case, the main antigenic molecule was found to be LPS [62, 112–114]. Bordetella pertussis and B. burgdorferi are two other pathogens, where the protective effect of OMVs in challenged mice was experimentally demonstrated [115, 116]. In case of B. burgdorferi, the major protective antigens were OspA and DbpA, two adhesins responsible for colonization. In case of OspA, it could even be shown that the OMV-associated form is more immunoprotective than the purified protein [115]. In Treponema pallidum OMVs had a much higher killing titer compared with serum from rabbits immunized with whole bacteria [117].

Concluding remarks

OMVs of Gram-negative bacteria that were initially observed under peculiar growth conditions are now an inherent part of microbiology. In the last 40 years of research on OMVs, it has become clear that these “magic bullets” are a general phenomenon rather than an exception. The research on OMVs now covers bacterial physiology, pathogenesis, and ecology with OMVs as virulence factors certainly being one of the most interesting properties. OMVs clearly represent an alternative secretion and transport mechanism for many different proteins, especially for those where no already known and described secretion system can be accounted for. OMVs also allow the bacteria to act on and interact with their environment over a longer distance without risking the disadvantages of direct contact. This is particularly true in cases where bacteria might be able to induce unspecific immune reactions via OMVs, and thereby, circumvent a specific host response, or where bacteriocidal factors might be beneficial for the competition in complex polymicrobial niches.

OMVs have evolved from being an exceptional observation to an important tool of immunology by means of alternative vaccination agents against one of the most epidemic bacterial diseases, namely neisserial meningitis. The benefits of OMV-based vaccines for the control of serogroup B meningococci combined with more than 20 years of experience in clinical trials and approved vaccination programs strongly support the potential of these vaccines. Genetic manipulation, which aims at improving vesicle production and antigenicity as well as decreasing adverse effects, already shows first promising results in the case of neisserial vaccines, and points out the plasticity as well as future potential of this approach. OMV-based vaccines will hence certainly be an important strategy in combating further endemic or epidemic diseases, like cholera, where OMVs are shown to be effective immunogenic vehicles.

Proteomics is a great help in characterizing OMV contents of many bacteria and shows that they can be enriched by certain components when compared to the periplasm or even the bacterial outer membrane they originate from. This further suggests that the biogenesis of OMVs is a relatively controlled process. However, the precise mechanisms of OMV biogenesis and specific targeting of proteins to OMVs are still only poorly understood and will certainly constitute an important part of future OMV research.