cholera toxin B subunit
diffuse region of the NALT
LPS-depleted outer membrane vesicle
lower respiratory tract
nasal-associated lymphoid tissue
native outer membrane vesicle
outer membrane vesicle
respiratory syncytial virus
upper respiratory tract
We have previously shown that following intranasal exposure to influenza virus, specific plasma cells are generated in the nasal-associated lymphoid tissue (NALT) and maintained for the life of the animal. However, we also showed that following infection with respiratory syncytial virus (RSV), specific plasma cells are generated in the NALT but wane quickly and are not maintained even after challenge, even though RSV-specific serum antibody responses remain robust. Only infection with influenza virus generated sterilising immunity, implying a role for these long-lived plasma cells in protection. We show here that the RSV-specific IgA NALT plasma cell population and lung antibody levels can be substantially boosted, both at acute and memory time points, by intranasal immunisation with inactivated RSV (iRSV) in combination with bacterial outer membrane vesicles (OMV) compared to live RSV alone. Finally, challenge with live RSV showed that immunisation with iRSV and OMV protect against both virus replication in the lung and the eosinophil infiltrate generated by either live RSV or iRSV alone. These data show that immunisation with iRSV and OMV maintains a NALT RSV-specific plasma cell population and generates an efficient protective immune response following RSV infection.
See accompanying commentary: http://dx.doi.org/10.1002/eji.200636118
Effective intranasal immunisation strategies suggest the prospect of protection from inhaled antigens within the respiratory and nasal mucosa and also throughout other mucosal tissues that together comprise the common mucosal immune system 1, 2. The nasal-associated lymphoid tissue (NALT) is the first organised lymphoid tissue to encounter inhaled antigen and in mice can be found as paired lymphoid organs under the soft palate and is equivalent to the human Waldeyer's Ring 3. We have previously described a long-lived population of virus-specific antibody-secreting cells (ASC) found within the NALT that actively secrete antibody for the life of the animal after just one infection with influenza virus 4. These plasma cells were found in the diffuse region of the NALT (D-NALT) encompassing the cells lining the nasal passages and nasal region, and were only found in the organised NALT (O-NALT) early in the primary response to the virus. We have also described previously that infection with respiratory syncytial virus (RSV) only leads to short-lived populations of the virus-specific plasma cells in the D-NALT and that, despite significant levels of virus-specific antibody in the serum of infected animals, mice were not protected from further infection with RSV 5. This is similar to the clinical situation where multiple re-infections with this pathogen occur both in infancy and in adult life, indicating that natural immunity is not protective, and has made the identification of immune correlates to protection difficult and the subject of much controversy 6, 7. These data imply that successful protection from RSV would require a “better than nature” immune response and suggest that the stimulation and maintenance of local mucosal antibody rather than systemic antibody may be important in generating effective protection. RSV is the single most important cause of hospitalisation amongst infants in developed countries 8, 9. That a vaccine against RSV has yet to be licensed more than 40 years after the virus was first isolated can be explained by several factors. The primary target of a potential vaccine are infants 0–6 months old, and issues such as the immaturity of the neonatal immune system, safety concerns specific to neonates, and the presence of maternal antibody which may inhibit subsequent primary immune responses must be addressed. RSV also has strong immunomodulatory properties which any potential vaccine must avoid. Indeed, pathology observed following primary infection in infants is largely immune mediated 10, 11. Furthermore, there is a risk of serious immunopathology upon RSV infection secondary to a failed vaccination, as demonstrated by clinical trials conducted in the 1960s using formalin-inactivated RSV (FI-RSV) 12, 13. Although the etiology of this enhanced disease has never been fully elucidated, studies in animal models have suggested that FI-RSV leads to a Th2-skewed CD4+ T cell response that not only fails to clear the virus but also triggers the recruitment of eosinophils 14, 15.
Here, we demonstrate that the intranasal delivery of inactivated RSV (iRSV) in combination with a mucosal adjuvant boosts the frequency and duration of the virus-specific D-NALT plasma cell population and that this immunisation regime protects against further challenge with live RSV. Furthermore, we find that the inclusion of the adjuvant with iRSV significantly reduces the Th2 effect observed with iRSV alone after RSV infection.
Long-term RSV-specific IgA response in the NALT after intranasal immunisation with iRSV and outer membrane vesicles
Mice were immunised with iRSV alone or in combination with either native OMV (NOMV) or LPS-depleted OMV (DOMV) three times over a period of 4 wk. To compare the immune response invoked by this immunisation protocol with that induced by a natural infection, a separate group of mice was immunised once with live RSV (see Materials and methods). RSV-specific B cell immunity in the upper respiratory tract (URT) was assessed by ELISPOT assays on cell suspensions obtained from O-NALT, representing the major inductive site of the URT lymphoid system, and from the D-NALT, representing the major effector site of the URT 4.
The acute immune response, assessed 10 days after the last immunisation, is dominated in the O-NALT by an RSV-specific IgA response after live RSV immunisation (Fig. 1A). In contrast, RSV-specific ELISPOTs from mice immunised with iRSV were predominantly of the IgG2b isotype. This difference in isotype could be explained by the fact that mice infected with live RSV received only one immunisation, whereas mice administered with iRSV received three immunisations. The number of RSV-specific ASC in the O-NALT following iRSV immunisation increased when given in combination with OMV. DOMV appeared to boost the ASC frequency more than NOMV.
The acute humoral response in the D-NALT (Fig. 1B) is dominated by the presence of RSV-specific ASC of the IgA isotype, regardless of the immunisation protocol, confirming the role of this lymphoid organ as an effector site of the mucosal response. While immunisation with iRSV fails to induce any significant IgA response, the iRSV + NOMV immunisation protocol boosts the response to a similar level as that obtained following live RSV immunisation. Interestingly, immunisation with iRSV and DOMV generated numbers of RSV-specific IgA-producing ASC up to 1000 per 106 cells, a level more than twice that observed after live RSV immunisation.
The adjuvanticity of OMV on the URT's IgA response was confirmed in mice culled 8 weeks after immunisation. While RSV-specific ASC have virtually disappeared from the O-NALT (Fig. 1C), the D-NALT is still the site of a robust IgA response in mice immunised with iRSV + OMV or live RSV (Fig. 1D). Furthermore, the frequency of ASC in the D-NALT induced by iRSV and OMV remained at a higher level than those induced by live RSV. It appears therefore that the use of OMV as an adjuvant to iRSV allows the development of long-term RSV-specific humoral immunity in the URT that is greater in magnitude than that observed after RSV infection.
OMV are strong adjuvants of the lower respiratory tract humoral response to iRSV immunisation
We have investigated the post-vaccination RSV-specific humoral response in the lower respiratory tract (LRT) using supernatants from lung fragment cultures. Fig. 2 represents RSV-specific antibody titres in supernatants from lungs harvested either 10 days or 8 wk after immunisation. Acutely, we observed that iRSV alone elicits very little RSV-specific lung IgA compared to iRSV combined with either NOMV or DOMV preparations, both of which showed significantly higher levels of IgA than those observed after live RSV infection. After 8 wk, IgA secreted from the lung after iRSV + DOMV immunisation remained significantly higher than after live virus immunisation (p <0.001). Similarly, the levels of RSV-specific lung IgG1 and IgG2b in mice receiving iRSV + DOMV remained consistently higher than that of their live RSV-immunised counterparts (p <0.05). The overall profile of the RSV-specific response observed in the lung therefore mirrors that observed in the URT, with iRSV + DOMV consistently generating the highest antibody levels.
Adjuvanticity of OMV on the systemic antibody response to intranasal iRSV immunisation
The RSV-specific systemic humoral immune response was measured by ELISA on sera taken from mice culled either 10 days or 8 wk after the last immunisation (Fig. 3). The acute response to live RSV immunisation is indicative of a strongly Th1-biased immunity, with high levels of IgG2a and IgG2b and relatively little IgG1 being produced. In comparison, iRSV immunisation raised higher levels of IgG1 (p <0.001) and lower levels of IgG2a and IgG2b (p <0.01), indicating a Th2-biased response. Remarkably, iRSV in combination with OMV resulted in a strongly Th1-biased immune response, reminiscent of that observed after live RSV immunisation. Moreover, levels of RSV-specific IgG2a and IgG2b were significantly increased in mice immunised with iRSV + NOMV or iRSV + DOMV compared to mice infected with live RSV. RSV-specific humoral immunity in mice immunised with iRSV and OMV is therefore maintained 8 wk after immunisation.
iRSV + OMV do not boost the RSV-specific T cell response
To assess the RSV-specific T cell response, we performed IFN-γ ELISPOT assays on total splenocytes from immune mice (Fig. 4). Live RSV immunisation induces high numbers of RSV-specific IFN-γ-producing splenocytes 10 days after immunisation, which are still maintained 8 wk after infection. In contrast, iRSV induces only small numbers of IFN-γ-producing cells. Immunising with a combination of iRSV and either NOMV or DOMV did not show any significant differences in the number of IFN-γ-producing splenocytes compared to iRSV immunisation alone. It therefore appears that, contrary to our observation on the humoral response, neither NOMV nor DOMV display any significant adjuvant effect on the T cell response to iRSV immunisation.
OMV abrogate the eosinophil-mediated lung pathology after live RSV challenge
At 8 wk after immunisation, mice were subjected to an intranasal challenge with live RSV in order to determine the protective effect of iRSV + OMV vaccination against secondary pathology. Mononuclear cells, neutrophils and eosinophils, were enumerated in the bronchoalveolar lavage (BAL) fluid of mice culled 5 days after challenge (Fig. 5). Counts of mononuclear cells in BAL fluid from mice immunised with either live RSV or iRSV + OMV were not statistically different from each other. Fewer mononuclear cells were found in iRSV-treated mice (p <0.001 versus live RSV). Surprisingly, mice immunised with mock virus preparation combined with OMV showed mononuclear infiltration in response to live RSV challenge. This most likely reflects immunity generated against cellular proteins and/or medium components. This reactivity occurred in spite of the fact that the virus used for immunisation and the one used for challenge were grown on distinct cell lines (the human HEp-2 and the simian Vero cell lines, respectively) and of the care taken to lower the FCS content in preparations.
In mice previously immunised with live virus, a modest infiltrate of neutrophils, ranging from 5 to 10% of the total BAL cells was observed. This influx was similar in magnitude to that observed after iRSV or iRSV + DOMV immunisation, indicating that neither iRSV nor DOMV increase the risk of development of neutrophil-related lung pathology. We also observed a weak eosinophil infiltrate in mice following live RSV infection (1 to 3.5% of total cells), while eosinophilia was markedly enhanced in iRSV-immunised animals (7 to 15% of total cells). However, as the cell counts in BAL fluid from iRSV-immunised mice were consistently lower than in RSV-infected animals, absolute eosinophil counts were not statistically different between these two groups. Previous studies have demonstrated that a single intramuscular injection of iRSV combined with alum mediates an eosinophilia of much greater proportions than observed here, both in terms of percentage and absolute cell numbers 14, 15. Our data therefore suggest that iRSV immunisation via the nasal route, even when repeated, is safer in terms of secondary pathology than the parenteral route. Nevertheless, the modest eosinophilia induced here by iRSV, which correlates with an antibody response indicative of a Th2-skewed T cell response, is likely to proscribe the use of iRSV alone as an intranasal vaccine for human use. In contrast, BAL fluid from mice immunised with iRSV combined with NOMV or DOMV contained as little as 0.1–0.7% eosinophils, with absolute cell numbers significantly lower than in iRSV-immunised mice (p <0.05). Furthermore, eosinophil counts in iRSV + OMV-immunised animals were also significantly lower than in RSV-infected animals (p <0.05).
Immunisation with iRSV + OMV protects against RSV replication in the LRT
Mice were challenged intranasally with live RSV 8 wk after the last immunisation. Lungs taken from mice culled either 3 or 5 days after challenge were assessed for the presence of RSV. As virus titres were consistently higher in lungs from mice culled after 5 days, only data from these mice are shown in Fig. 6. We observed that animals previously infected with RSV showed no significant lung RSV replication. Conversely, iRSV immunisation failed to protect mice from RSV replication in the LRT, although virus titres from these animals display a tenfold reduction in PFU compared to naive or mock-immunised mice. Interestingly, no live RSV could be isolated from mice immunised with iRSV + NOMV or DOMV, implying that these animals are completely protected from re-infection.
In a further study, mice that had undergone the immunisation protocol with iRSV + DOMV were depleted in vivo using mAb to either CD4 or CD8, or both. Depletion of either or both of these subsets did not abrogate protection (data not shown) when mice were challenged with RSV, implying that T cells do not appear to play a role in protection.
Our previous studies identified a long-term population of plasma cells in the NALT after influenza virus infection that may have an important role in protection from re-infection 4. However, long-term virus-specific plasma cells do not appear to be induced after RSV infection in the NALT, even though serum antibody was clearly detectable 5, leading us to propose that this NALT population may be important in protection from RSV infection. Our results demonstrate that the mucosal response, as indicated by the levels of IgA in the lung, seems to fall rapidly after RSV infection but is boosted and maintained to higher levels after immunisation with iRSV + DOMV. Furthermore, an increased frequency of RSV-specific plasma cells is apparent within the D-NALT after iRSV + DOMV immunisation. Immunisation with iRSV + OMV also generates complete protection against viral replication after live RSV challenge.
Immunisation with iRSV + DOMV generated few splenic IFN-γ-producing cells, indicating that the MHC class I-restricted CD8+ T cell response evoked by this immunisation protocol is weak compared to animals immunised with live RSV. Not only does this factor not alter protection of mice against viral replication, it might also contribute to the reduced numbers of infiltrating eosinophils observed after RSV challenge. Indeed, RSV-specific CD8+ T cells have been implicated in the pathology linked with both neutrophil and eosinophil infiltrates upon infection 16, 17. These elements are further evidence that, while CTL might be the main factor contributing to the resolution of primary RSV infection, they are dispensable and possibly deleterious in vaccine-induced protection against re-infection and lung pathology.
The need for a vaccine to RSV is highlighted every winter with the substantial number of hospitalisations of young infants and elderly infected with this virus. Unfortunately, research into potential vaccines has been hindered by the legacy of the 1960s vaccine trials using FI-RSV which led to exacerbated disease after natural RSV exposure 13. Several studies have demonstrated the ability of iRSV to prime for a Th2 response, thus leading to an inappropriate response, highlighted by substantial eosinophilia in the lung after RSV challenge at least in animal models. Previous animal studies have delivered the iRSV mainly by the intramuscular route, although other studies have delivered vaccinia virus constructs carrying the RSV G protein by scarification at the base of the tail 14–16. These methods of immunisation prime for a Th2 response in the lung after RSV challenge. Here, we report that delivery of iRSV by the intranasal route combined with a novel adjuvant can abrogate the Th2 priming and lead to successful protection from further RSV infection.
In this study, NOMV and DOMV were standardised according to their protein content and differed mainly, if not solely, in their content in Neisseria meningitidis LPS. One can therefore assume that differences observed in the immune responses to iRSV + NOMV and iRSV + DOMV are due to the immunomodulating effect of LPS. In vitro studies have shown that various types of LPS, including membrane-associated N. meningitidis LPS, can modulate the acquired immune response by inducing the secretion of IL-12 by dendritic cells, leading to strongly Th1-skewed responses 18, 19. Here, we observe that DOMV used as adjuvant not only diminish the secondary neutrophil infiltrate compared with NOMV, but also generate higher levels of mucosal IgA. One could speculate that this difference might be due to excess LPS content in NOMV modulating the interaction of neisserial porins with the mucosa through the receptor TLR2 20, thereby limiting their adjuvant effect. However, the major LPS signalling receptor TLR4 is also present on mucosal tissues and should therefore allow the uptake of NOMV as efficiently as that of DOMV 21. It is more likely that the pro-inflammatory microenvironment generated by excess LPS in the NOMV interferes with the development of mucosal IgA plasma cells. B cell switching to the IgA isotype has been linked with a number of key cytokines such as IL-10 and TGF-β 22. In addition, T helper polarisation has been described as dependent on the antagonist effects of IL-12 and IFN-γ (Th1-skewing) and of IL-10 and IL-4 (Th2-skewing) 23. One can therefore hypothesise that Th1 cytokines induced by NOMV might inhibit IL-10 secretion and provide a less favourable environment for the development of IgA plasma cells than DOMV. However, one cannot exclude the possibility that LPS contained in DOMV plays a major role in their potent mucosal adjuvanticity.
Immunisation strategies using non-invasive antigens such as iRSV or subunit vaccines given by the nasal route have led to contrasting results. Extensive use has been made of the well-defined mucosal adjuvant cholera toxin (CT) and its B subunit (CTB). Two studies using a purified F protein combined with CT or CTB reported that complete lung protection can only be achieved when intranasal immunisation is combined with a parenteral boost 24, 25. Conversely, a third study using G peptides and CT reported complete lung protection associated with serum IgG levels comparable to those evoked by live RSV infection, but failed to demonstrate secretory IgA 26. Use of iRSV encapsulated in immunostimulating complexes (ISCOM) has also been documented and evokes strong mucosal and systemic humoral responses, although a clear comparison with immunity following live RSV immunisation is still lacking 27, 28. Finally, studies using intranasally delivered MHC class I peptides have allowed the development of protective RSV-specific CTL, but those appeared to be either short-lived 29 or linked with enhanced disease upon RSV challenge 30.
Here, we show that iRSV combined with DOMV evokes a mucosal IgA response superior to, and a serum response at least equivalent to, live RSV infection. Furthermore, the eosinophilic infiltrate observed in mice receiving iRSV + DOMV after live RSV challenge was lower than that evoked by a previous encounter with live virus, while protection against virus replication was equally complete. Bearing in mind the limitations of the mouse model, these elements nonetheless provide evidence that a “better than nature” immunisation against RSV is achievable.
Most interestingly, while adjuvants such as CT and ISCOM are yet to be licensed for human use due to safety concerns, OMV are currently used as parenteral vaccines against meningitis in infants as young as 2 months of age 31–33. Although a number of safety concerns have been related to their use as adjuvants, particularly the amount of LPS contained within the vesicles, it is clear that both NOMV and DOMV are safe when given intranasally to adults 34, 35. This study could therefore represent a first step towards the development of a rationally designed dual vaccine against both meningitis and RSV, whereby advantage is taken of the adjuvant properties of one antigen to modulate the immune response to the other.
Finally, one can also speculate on the importance of the presence of the long-lasting population of virus-specific ASC in the D-NALT. What seems clear is that protection from secondary infection within the respiratory tract is demonstrated when a population of virus-specific ASC is present. This study provides further evidence for the identification of this population as a correlate of protection.
Materials and methods
BALB/c mice 5–8 wk old were purchased from Charles River (Margate, UK) and kept under specific pathogen-free conditions. All procedures were performed in accordance with Home Office ethical guidelines.
Virus used for immunisations was grown in HEp-2 cells (ATCC, Manassas, USA), harvested as clarified cell lysate in serum-free DMEM (Life Technologies, Paisley, UK) and inactivated by β-propiolactone treatment (Sigma) as described 36. Mock virus preparations were obtained from uninfected HEp-2 cells. Virus used for live RSV challenge and for immunological assays was grown on Vero cells (ATCC) and used either as clarified cell lysate or further purified on sucrose gradients, as described 37. Virus titrations were carried out as described 5. NOMV containing approximately 25–50% LPS (relative to protein by weight) and DOMV with 5–8% LPS 35 were produced from mutant 44/76Mu-4 of Neisseria meningitidis as described 31, 38. Both OMV preparations were tested for sterility and standardised according to their protein content.
Mice were anaesthetised by i.p. injection of Ketamine/Xylazine (Sigma, Poole, UK), prior to infection. Mice were immunised intranasally on days 0, 7 and 21 with the equivalent of 2 × 105 PFU of β-propiolactone-inactivated RSV-A2 virus in a 50-µL volume (or an equal volume of mock virus), containing either PBS, 10 µg NOMV or 10 µg DOMV per dose. An additional group received one live RSV immunisation intranasally (2 × 105 PFU) on day 21. The final group consisted of naive animals. Mice were culled either 10 days or 8 wk after the last immunisation, to assess the acute or memory vaccination-induced immune responses, respectively. All remaining mice were then challenged intranasally with live RSV (2 × 105 PFU) and culled 5 days later to assess protection and lung pathology.
RSV-specific ELISPOT and ELISA
Lymphoid cells from the O-NALT or D-NALT were obtained as described 39 and pooled from groups of six animals. The ELISPOT and ELISA assays were carried out as described using sucrose-purified RSV-A2 as the coating antigen 5. RSV specificity was ensured by the fact that no plaques were observed for mice immunised with mock virus preparations. Only results obtained from mice immunised with inactivated or live RSV virus are shown.
Lung fragment cultures were established as described 40. Briefly, the left lobe of the lungs of individual mice was harvested, washed thoroughly and cut into four roughly equal pieces. Each fragment was then cultured in 1.0 mL RPMI medium containing 10% FCS and incubated in 24-well plates in an atmosphere of 95% O2 and 5% CO2. After 5 days, the supernatants were harvested and samples from each individual mouse were pooled and stored at –70°C. For ELISA assays, plates containing serum samples were developed using the alkaline phosphatase substrate p-nitrophenyl phosphate (Sigma). Optical densities were read at 405 nm on a SPECTRAmax (Molecular Devices, Sunnyvale, USA) spectrometer. ELISA plates containing lung fragment cultures were amplified using the AmpliQ kit (Dako, Ely, UK) and read at 492 nm with correction from values read at 650 nm. Titres are expressed as reciprocal endpoint titres. Endpoint values obtained either from sera or lung fragment culture supernatants of mice immunised with mock virus preparations were consistently less than 28 and 22, respectively. Therefore, only data obtained from mice immunised with live or iRSV are shown.
Mixed cellulose-ester flat-bottom plates (Millipore) were coated overnight with 5 µg/mL rat anti-mouse IFN-γ antibody (clone R4–6A2; BD Pharmingen, Oxford, UK). Plates were washed with PBS and then blocked with DMEM containing 10% FCS for 1 h at 37°C. Effector cells consisted of total splenocytes from immune mice. Stimulator cells consisted of splenocytes from naive mice, depleted of T cells by incubation with anti-CD4, anti-CD8 and anti-Thy1.2 antibodies (clones GK1.5, 53.6.72 and 30-H-12, respectively, obtained from ATCC) followed by magnetic depletion using anti-rat IgG Dynabeads (Dynal, Wirral, UK). Stimulators were either infected with RSV (0.1 PFU per cell) for 5 h at 33°C or left uninfected, then irradiated at 3000 rad. Effectors were serially diluted and incubated with 5 × 105 infected or uninfected stimulator cells per well, for 24 h at 37°C, in 5% CO2. Plaques were detected with biotin-labelled anti-mouse IFN-γ antibody (clone XMG1.2; BD Pharmingen) followed by alkaline phosphatase-coupled anti-biotin antibody (Vector, Peterborough, UK) and developed with 5-bromo-4-chloro-3-indolyl phosphate. Spots were enumerated by light microscopy. The plaques in wells containing uninfected stimulators were subtracted from those in wells containing infected stimulators.
BAL fluid analysis
BAL samples were obtained from mice by flushing the lungs with four times 1 mL HBSS containing 0.1% BSA. Cells were deposited onto glass slides by cytospin at 104 cells per slide. Slides were fixed with methanol and stained with Wright-Giemsa (Sigma). At least 20 randomly selected foci were counted, representing a minimum of 400 individual cells. Cellularities were extrapolated from the observed proportion of the different cell types multiplied by the total BAL fluid cellularity, for individual mice. Results are expressed as the mean ± SD cellularity per mouse (five mice per group), for each cell type.
In vivo depletion of T cells
Mice were inoculated i.p. with 0.5 mL mAb GK1.5 (ATCC) to deplete CD4+ T cells, or mAb 2.48 (generous gift from Prof. Peter Doherty, St. Jude Children's Research Hospital, Memphis, TN) to deplete CD8+ T cells. Injections were carried out 3 days prior to viral challenge, on the day of challenge, 3 days after challenge and every 2 days afterwards. mAb preparations were produced using the CL1000 system (Integra Biosciences AG, Chur, Switzerland) and dosages used were determined by in vivo titration. In all experiments, greater than 98% depletion of each subset was achieved.
Data were analysed by using the Student's t-test. In all statistically analysed experiments, immunised groups were only compared to the control, live-infected group.
This work was supported by The Edward Jenner Institute for Vaccine Research.