A Recombinant G Protein Plus Cyclosporine A–Based Respiratory Syncytial Virus Vaccine Elicits Humoral and Regulatory T Cell Responses against Infection without Vaccine-Enhanced Disease Free

Respiratory syncytial virus (RSV) infection is a major cause of respiratory tract disease in children <5 years old. It leads to 64 million cases of bronchiolitis and viral pneumonia (1–5) and causes ∼200,000 deaths annually (4, 6). Moreover, there are direct correlations between RSV infection in childhood and development of asthma in adulthood (7). The infections have become a heavy medical and economic burden around the world, particularly in developing countries. During the infection, RSV modulates the host immune system through several mechanisms. These include arresting the cell cycle by downregulating p53 (8), inhibiting TNF-α via small hydrophobic protein (9), and suppressing maturation of dendritic cells (DCs) and some subtypes of T cells via NS1 and NS2 proteins (10–12). Although the virus was identified >60 years ago, no vaccine is available currently. This problem is largely a legacy from the severity of the pathologic responses that were induced by vaccination with formalin-inactivated RSV (FI-RSV).

In the 1960s, the FI-RSV vaccine caused recipient children severe lung injuries, and two infants died in a phenomenon that is now called vaccine-enhanced disease (VED) (13, 14). Pathological analysis showed that the dead infants had an extended peribronchiolitis and alveolitis, with infiltration by neutrophils and a few eosinophils (15–17). This problem can be reproduced in animal models, including mice and monkeys; when FI-RSV–immunized animals were challenged with RSV, there was extensive lung infiltration by neutrophils, lymphocytes, and some eosinophils, causing severe damage (18–21). Increased complement-fixing Ab titers and lymphoproliferative responses were also observed during viral clearance (22–25). These human cases and animal models suggested that the FI-RSV vaccine primes an overreactive and harmful inflammation that can be elicited when primed individuals encounter RSV infection.

Further reports showed that the harmful inflammation was due to the induction of a Th2-type response in lungs and overproduction of Th2 cytokines in a so-called cytokine storm. The Th2-type response was mainly a consequence of recruitment of infiltrating granulocytes, especially eosinophils (20, 25–28). The eosinophils could not only promote naive CD4⁺ T cells to differentiate into Th2 cells by secreting cytokines such as IL-4 and IL-13, but they also could present Ags to naive CD4⁺ T cells directly to promote Th2 polarization. Moreover, chemokines such as eotaxin 1 and CXCL1 secreted by the eosinophils could recruit Th2 effector cells to the infected tissue (29). Based on the understanding that a correct Th1 versus Th2 balance is needed, the various prototype vaccines that have been developed are mainly aimed at inducing a strong Th1-type response and avoiding Th2. Krause et al. (30) found that an adenovirus-based RSV vaccine could increase the presence of plasmacytoid DCs (pDCs) in the lung during the RSV infection and these increased pDCs could promote the Th1 and regulatory T (Treg) cell responses against the RSV infection. Accordingly, an imbalanced pDC to conventional DC ratio was related to the Th2-type response.

The vaccines tested included attenuated RSV vaccines and subunit vaccines based on F or G glycoprotein and N protein. However, most only induced low levels of anti-RSV Abs, and a strong Ab response might be absolutely required to prevent RSV infections, because two commercially available mAbs, palivizumab and motavizumab, have been used to successfully treat children with RSV infections (31, 32).

Recently, a study demonstrated that within the harmful Th2 response elicited in the lung by infection after FI-RSV vaccination there was a low content of the Treg cells that are needed to attenuate such inflammation and tissue damage (33). The Treg subtype of CD4⁺ Th cells (34) uses a number of mechanisms for homeostatic control of overexuberant or incorrect immune responses. These include secretion of anti-inflammatory cytokines such as IL-10 and TGF-β (35, 36), disturbance of T cell metabolism by consuming IL-2 (37), or suppression of T cells by cell contact involving CTLA4 (38). Animals and human beings without Treg cells develop severe inflammation-related diseases, including allergy, asthma, and autoimmune disease. Because the lack of Treg cells can lead to severe tissue damage, it follows that induction of Treg cells might be used to treat such inflammation-related diseases (39–41). Indeed, during some virus infections, Treg cells migrate into inflamed locations and minimize tissue damage by suppressing the excessive immune responses (42). The functional Treg cells express CCR4 and interact with chemokines, including CCL17 (TARC) and CCL22 (MDC), which are important for recruitment of the cells to skin and lung (43, 44). These important findings have been recently reported showing that both CCL17 and CCL22 are critical mediators in recruiting CCR4 Treg cells to suppress VED during RSV infection after the FI-RSV prime (45). Based on these findings, strategies for developing a safer RSV vaccine that circumvents FI-RSV–induced VED should include the induction of both Treg cells in lung and high levels of neutralizing Abs (20, 45, 46).

Among all of the 11 viral proteins of RSV, only G and F glycoproteins contain neutralizing epitopes and induce long-term neutralizing Abs (47–49). For this reason, both have been considered as ideal candidates for RSV vaccine development. As the F protein is considered to induce high affinity of neutralizing Ab and only mild VED, several research groups have focused on this protein (50, 51). However, it is technically difficult to obtain the recombinant F glycoprotein in a stable prefusion form, and the misfolded or postfusion forms of F protein have shown less ability to induce sufficient neutralizing Abs (50, 51). A novel approach to make the prefusion form has been recently achieved (51). The G protein has been selected as an alternative promising candidate, because the immunogenicity is independent of the recombinant protein structure. It functions as an attachment protein during RSV infection by interacting with the CX3C motif of chemokine receptor CX3CR1 at the 182–186 aa position (52). Such interaction alerts CX3CL1 ligand–dependent T cell responses and impairs airway immune responses (53, 54). Another advantage of the G protein as a vaccine candidate is the presence of neutralizing epitopes that are comparatively independent of protein structure. Several approaches, including subunit-, nanoparticle-, peptide-, and bacterial-based vaccines have been tried, and an mAb against G protein has been demonstrated to inhibit the virus infection in animal models (55–60). However, the association of the G protein with VED indicates the need for caution during its development as an RSV vaccine (26).

Cyclosporine A (CSA) is a widely used immunosuppressant in organ transplantation (61–63), or in treatment of children with aplastic anemia (64, 65) and other autoimmune diseases (66–68). CSA binds to the cytosolic protein cyclophilin and inhibits calcineurin of T cells, consequently suppressing transcription of IL-2 (69–73). It also interferes with DC to T cell interaction (74) and can be used to induce Treg cells (75–77). Moreover, the concept of immunizing with Ag together with an immunosuppressant to induce Treg cells and effectively treat autoimmune diseases has previously been proven in animal models (78, 79). Following this lead, we evaluated the ability of various immunosuppressants, including CSA, dexamethasone, rapamycin, and FK506, to induce tolerogenic reactions with human PBMCs in vitro. CSA showed the best ability to promote such reactions (data not shown). Although we found no report that a combination of CSA with RSV Ag could be useful for RSV vaccine development, in this study, we combined an RSV G truncated protein with CSA (G+CSA) and formulated the combination as a novel vaccine. We then evaluated its protective effect against RSV infection and its ability to inhibit lung inflammation. We demonstrated that G protein combined with CSA not only induced a high level of neutralizing anti-RSV Abs that led to reduced viral titers, but it also suppressed the lung inflammation that causes the severe lung injury often seen with FI-RSV vaccination. The suppression was due to induction of Treg cells. Therefore, G+CSA vaccination presents a novel approach to develop an effective and safe RSV vaccine for children.

Female 6- to 10-wk-old BALB/c mice were used for all experiments and purchased from the Shanghai SLAC Laboratory Animal Co. Ltd. (Shanghai, China). Foxp3-DTR/EGFP mice on a BALB/c background were from The Jackson Laboratory (Sacramento, CA). All mice were kept under specific pathogen-free conditions at Fudan University and handled according to the animal welfare guidelines for experimental animals.

Plaque-purified human RSV (type A2 strain from the American Type Culture Collection, Rockville, MD) was grown in HEp-2 cells and concentrated by ultracentrifugation (50,000 × g for 1 h). Mice were infected intranasally (i.n.) with 5 × 10⁷ PFU RSV in 100 μl under anesthesia on day 14 after the second immunization.

FI-RSV vaccine was prepared as described by Kim et al. (14). In brief, 100 ml clarified RSV was incubated with formalin (37–40%) at 4000:1 (v/v) for 72 h at 37°C and then centrifuged at 50,000 × g for 1 h at 4°C. The pellet was diluted in 4 ml MEM and subsequently mixed with Imject alum adjuvant (Thermo Scientific, Rockford, IL) as 10:1 (v/v), suspended in 1 ml serum-free MEM, and stored at 4°C.

Nonglycosylated G protein was expressed in Escherichia coli BL21 (DE3), purified, and lyophilized. CSA (Santai, Taishan, Chain) was dissolved in propylene glycol/PBS (1:1) solution to 100 mg/ml. Lyophilized G protein was dissolved with the CSA solution or propylene glycol/PBS (1:1) solution to 100 mg/ml before injection.

The mice were randomly divided into groups and immunized s.c. with 10 μg G protein, CSA alone, or G+CSA on days 0 and 14. FI-RSV (50 μl) was used to immunize animals i.m. on days 0 and 14.

Mice were sacrificed by pentobarbital i.p. injection. Single-cell suspensions from the spleens and lung draining lymph nodes (dLNs) were collected with pestle and filter. Bronchoalveolar lavage (BAL) was collected by using 1 ml PBS to flush lungs three times before sacrificing animals.

Serum samples were obtained on days 0, 7, 21, and 38 after the first immunization and stored for ELISA analysis. Briefly, 96-well plates were coated with 5 × 10⁶ PFU/ml heat-inactivated RSV (50 mM carbonate-bicarbonate buffer [pH 9.6]) at 37°C and blocked with 5% BSA in PBST (0.05% Tween 20 in PBS) at 37°C. The plates were incubated with serial 2-fold dilutions of serum for 1 h at 37°C. Bound Abs were detected with HRP-conjugated goat anti-mouse IgG (SouthernBiotech, Birmingham, AL), the enzymatic reaction was developed, and ODs were read at 450/620 nm by an ELISA plate reader (Bio-Rad, Hercules, CA)

The neutralization assay was performed as described previously (80). Briefly, sera were serially diluted 5-fold in a total of 100 μl PBS, heat inactivated at 56°C for 30 min, and incubated with 3 × 10³ 50% tissue culture–infective dose virus for 2 h at 4°C. Approximately 5 × 10³ HEp-2 cells in 100 μl DMEM supplemented with 2% FCS were added to each well of a 96-well microtiter plate. The virus/serum mixture was added to the appropriate wells and incubated for 3 d in a 5% CO₂ incubator at 37°C. Plates were then washed three times with PBST and fixed with 80% cold acetone in PBS followed by blocking with 3% blocking buffer. Goat anti-RSV Ab (Meridian Life Science, Saco, ME) was added to the appropriate wells and incubated for 60 min at 37°C. After three washings, bovine anti-goat IgG-HRP (Santa Cruz Biotechnology, Santa Cruz, CA) was added, the enzymatic reaction was developed, and ODs were read at 450/620 nm by an ELISA plate reader. Neutralizing Ab titers were calculated by the method of Reed and Muench (81); the average OD of triplicate wells was extrapolated by the inverse of the serum dilution that resulted in 50% reduction of RSV activity.

The quantity of IgG Ab produced in coculture medium was quantified by ELISA; 96-well plates were coated with IgG capture Ab (SouthernBiotech), and a curve was generated with mouse IgG-BIOT (SouthernBiotech).

On day 14 after the second immunization, single-splenocyte suspensions were collected and labeled with eFluor 670 proliferation dye (eBioscience, San Diego, CA). Cells at 5 × 10⁵ were added to each well in a 96-well flat-bottom culture plate in 100 μl in a 37°C/5% CO₂ incubator. Proliferation of responder T cells was determined after incubation for 4 d by analysis of eFluor 670 proliferation dye dilution.

Cells were stained with the following surface Abs (all from eBioscience unless noted otherwise): CD3 FITC, CD4 FITC or eFluor 450, CD25 allophycocyanin–eFluor 780, CCR4 BV405 (BioLegend, San Diego, CA), CXCR3 PerCP-Cy5.5, CD103 allophycocyanin, IFN-γ PE, IL-17a allophycocyanin, and IL-4 PE. All intracellular staining was done with the Foxp3/transcription factor staining buffer set (eBioscience) for IL-10 PerCP-Cy5.5 (eBioscience), Foxp3 FITC (eBioscience), and TGF-β (BioLegend). The data from BD LSRFortessa flow cytometer (BD Biosciences, San Diego, CA) was analyzed with FlowJo software (Tree Star, Ashland, OR). IL-10 levels in the BAL and cocultured medium were measured by a cytometric bead array flex set (BD Biosciences) following the manufacturer’s recommendations.

Splenocytes (1 × 10⁶) from RSV noninfected mice were labeled with eFluor 670 proliferation dye (eBioscience) and cocultured with 5 × 10⁴ cells of BAL and dLNs obtained from mice at 4 d after RSV (5 × 10⁷ PFU) infection. Mixed cells were stimulated with heat-inactivated RSV at 37°C/5% CO₂ for 3 d, and eFluor 670⁺ spleen T cells were monitored to determine the proliferation. The level of IL-10 in medium was tested by a cytometric bead array flex set (BD Biosciences) following the manufacturer’s recommendations.

Foxp3-DTR/EGFP mice were injected i.p. with 1 μg diphtheria toxin (DT; Sigma-Aldrich, St. Louis, MO) on days 1 and 2 before RSV infection to deplete Foxp3⁺ Treg cells. For neutralization of CCL17 and CCL22, 20 μg anti-CCL17 and anti-CCL22 Abs or IgG isotype control was i.p. injected on day 1 after RSV infection. IL-10 blocking was performed by adding anti–IL-10 Ab to the cell culture medium to the final concentration of 5 μg/ml, and mice were injected with one dose i.p. of 20 μg anti–IL-10 or IgG isotype control on day 1 after RSV infection.

Four days after RSV infection, lung tissues were fixed with 4% neutral buffered paraformaldehyde, and transverse sections (5–7 μm) were stained with H&E for histopathological evaluations. The infiltrated nuclei were counted with Image-Pro Plus (Media Cybernetics).

Statistical analysis was done using GraphPad Prism 6.0 (GraphPad Software, La Jolla, CA) and presented as means ± SEM. An unpaired Student t test and ANOVA were used in one-way or two-way data analyses. A p value <0.05 was considered statistically significant.

The RSV attachment protein without its transmembrane domain (G protein, aa 67–298) was cloned into the pET28a plasmid and expressed in E. coli BL21 (Supplemental Fig. 1A, 1B). The purified recombinant G protein was characterized by SDS-PAGE and identified by Western blot analysis with anti-RSV sera (Supplemental Fig. 1C, 1D).

To test responses to the combination of G protein with CSA, mice were immunized i.m. or s.c. twice with 2-wk intervals. As shown in Fig. 1A, 1B, and Supplemental Fig. 2, the levels of anti-RSV Abs and neutralizing Abs were higher in the group immunized with G+CSA compared with protein G without CSA, and even higher than in the FI-RSV–immunized group. Subcutaneous injections gave rise to significantly higher Ab responses (p < 0.05) than did the i.m. route (Fig. 1A). The optimal ratio of G protein/CSA to reach maximal inhibition of T cell activation while retaining induction of a high level of anti-RSV Abs was at 10 μg G protein plus 10 μg CSA (Fig. 1C). This dose was used throughout this study except where specifically mentioned.

To evaluate protective efficacy of the 10 μg G+CSA formulation, animals were immunized on days 0 and 14 and then challenged i.n. with 5 × 10⁷ PFU RSV (A2 strain) on day 28 (Fig. 2A). The viral loads in the lung were analyzed 4 d after the challenge. As depicted in Fig. 2B, the G+CSA immunization reduced the virus load in lung significantly more than did immunization with G protein (p = 0.0053) or FI-RSV (p = 0.0013). Meanwhile, the changes of body weight were assessed as an indication for morbidity of disease for 10 d after RSV challenge (Fig. 2C). Change in body weight in the G+CSA-immunized group was not significantly different from the noninfected controls. In contrast, early loss of body weight was revealed in the FI-RSV–immunized mice, and late losses of body weight were associated with groups receiving PBS, CSA, or G protein. These results suggested that G+CSA vaccine provided protection not only against infection but also against morbidity.

Because VED is the major obstacle against RSV vaccine development, we analyzed lung pathology including the tissue index and histochemistry of the challenged mice. The lung tissue index was significantly lower in animals that had received G+CSA than in the other groups (Fig. 2E). H&E staining revealed that immunization with G protein or FI-RSV vaccine followed by challenge with RSV resulted in severe pneumonia associated with drastic perivasculitis, peribronchiolitis, and alveolitis. In contrast, G+CSA recipients had mild lung pathology with some peribronchiolitis and nearly normal alveolar morphology (Fig. 2F). This was consistent with the total cell counts in BAL fluid from the corresponding groups (Fig. 2D). These results demonstrated that the G+CSA vaccine, in contrast to the other vaccines, could significantly reduce the lung inflammation and injury that occurred after RSV challenge.

Because Treg cells were reported to be one of the most important factors in limiting VED (45), we examined whether the G+CSA vaccine could induce a high level of Treg cells in the BAL and dLNs of the lungs and spleens in these animals after RSV challenge. We observed that the percentage of CD4⁺ T cells in spleens that were CD25⁺Foxp3⁺ Treg cells had fallen by 4 d after challenge from around the 10% seen in PBS control mice down to 7% when the mice had been vaccinated with FI-RSV, to 3% with G, or to 5% with CSA (Fig. 3A). The Treg cells were purged to an even lower level (<2%) in the dLNs in these groups (Fig. 3B). In contrast, animals that had received G+CSA before challenge with RSV had higher levels of Treg cells in spleens and dLNs when compared with naive animals (Fig. 3A, 3B). Furthermore, in BAL there was a striking increase in Treg cells after challenge in the G+CSA group, both in their absolute numbers and as a percentage of total CD4⁺ T cells (Fig. 3C, 3D), and the increases were much higher than seen in the other groups (p < 0.0001). This demonstration that G+CSA vaccination leads to robust Treg cell responses in challenged lungs is consistent with the notion that Treg cells might abrogate the lymphocyte activation and subsequent excessive inflammatory responses and lung damage during the RSV infection.

To elucidate the biological importance of G+CSA-induced Treg cells, Treg cell depletion or adoptive transfer studies were performed. Foxp3-DTR mice were immunized twice with G+CSA on days 0 and 14, given two injections of DT on days 26 and 27 to delete Treg cells, and then were RSV challenged on day 28 (Fig. 4A). Foxp3⁺ cells remained depleted for a week after the DT injections (Supplemental Fig. 4C). The viral loads in the lung were analyzed 4 d after the challenge. As depicted in Fig. 4B, the virus load of the Treg cell–depleted G+CSA mice was between those of the G+CSA- and G protein–immunized groups. The changes of body weight were assessed as an indication of disease morbidity 10 d after RSV challenge. As shown in Fig. 4C, even when the mice had been immunized with G+CSA, weight loss was severe once Treg cells were depleted. Moreover, the BAL from Treg cell–depleted mice contained more cells than did the BAL from Treg cell–sufficient mice (Fig. 4D), and among these cells, most CD4⁺ T cells expressed IL-4 (Fig. 4E). Lungs were harvested on day 4 after the RSV challenge, fixed, sectioned, and stained with H&E (Fig. 4F). The lung morphologies were found to be unchanged when animals had been immunized with G+CSA without Treg cell depletion, and massive cell infiltration was exhibited in the G+CSA-immunized mice that had been depleted of Treg cells.

We tested whether any functional activities could be transferred with Treg cells or BAL fluid from G+CSA- or PBS-immunized animals. Splenocytes were isolated from Foxp3-EGFP mice and the Treg cells were sorted by FACS based on GFP positivity. G protein–immunized animals received either the BAL fluid or the sorted Treg cells by i.n. transfer on the second day after the RSV challenge (Fig. 4G). The recipients were sacrificed 2 d after the transfer and their lungs were removed for histochemistry analysis and staining with H&E (Fig. 4H). The recipients of either BAL fluid or Treg cells from the G+CSA-immunized animals showed much less lymphocyte infiltration than did the recipients of BAL from PBS-treated controls or recipients of medium only. Because both donor Treg cells and BAL fluid could suppress inflammatory T cell infiltration into lung, they both may play an important role to protect lung from VED upon RSV infection.

Because the expression of CCR4 by Treg cells is required for their movement from the peripheral circulation into target tissues where the chemokines CCL17/22 are released (82, 83), we asked whether it had a role here. We isolated Treg cells from BAL and looked for surface expression of several chemokine receptors. We found no differences between the groups in the expression levels of CD103 and CXCR3 on the Treg cells (Supplemental Fig. 3). However, the expression of CCR4 on the G+CSA-induced Treg cells was significantly higher (p < 0.05) than for Treg cells from the other groups (Fig. 5A).

To demonstrate the functional relevance of CCR4 on the migration of the G+CSA-induced Treg cells, we neutralized CCL17 and CCL22 by i.p. injection of anti-CCL17 and anti-CCL22 Abs into the G+CSA vaccine recipients on 1 d postinfection (dpi) with RSV. Weight changes were tested for 4 d and the anti-CCL17/22–treated animals showed a significant weight loss (p < 0.01) at 4 dpi compared with IgG isotype Ab controls (Fig. 5B). This result was consistent with the severe lung inflammation and lung injuries seen when lung sections were analyzed by histochemistry and H&E staining (Fig. 5C). When CCL17/22 was blocked, the level of the G+CSA-induced Treg cells in the BAL was significantly reduced (p < 0.05) compared with the IgG isotype controls (Fig. 5D). Furthermore, the level of G+CSA-induced Treg cells in dLNs was slightly higher after the blockade compared with prior to the blocking (Fig. 5E), although the difference was not significant (p = 0.083). These findings imply that G+CSA-induced Treg cells could not migrate from LNs into lung due to the Ab-induced low level of CCL17/22 in the tissue. This is consistent with the evidence that CCR4 expression on the G+CSA-induced Treg cells plays an important role in migration from peripheral or dLNs to the lung after RSV infection.

Because IL-10 and TGF-β are both key anti-inflammatory cytokines produced in Treg cells, we sought to determine whether these were responsible for suppressing inflammation during RSV infection. Treg cells were harvested from spleen, dLNs, and BAL fluid in the various immunization groups and analyzed by intracellular staining and FACS. There were no differences in the level of TGF-β among all of the groups (Fig. 6A). However, a significantly higher level of IL-10 in Treg cells of the spleen and dLNs was induced by immunizing animals with G+CSA compared with other groups (p < 0.001; Fig. 6B, 6C, Supplemental Fig. 4A, 4B). Strikingly, the level of IL-10 in the BAL Treg cells of the G+CSA-immunized group was 6-fold higher than in other immunized groups after the RSV infection (Fig. 6D). These data suggested that IL-10 could have the anti-inflammatory role. To further investigate the source of the IL-10, we depleted Treg cells by injecting DT into Foxp3-DTR mice. The mice were immunized with G+CSA at days 0 and 14, and PBS or DT was injected i.p. at days 26 and 27. The mice were then infected with RSV at day 28. The concentration of IL-10 in the BAL was decreased significantly (p < 0.001) as tested 4 d after Treg cell depletion (Fig. 6J).

To assess the role of IL-10, we performed an in vitro coculture inhibition assay with Treg cells isolated after RSV infection from the BAL and dLNs of the animals that had been immunized with G+CSA or G protein. These isolated cells were cocultured with splenocytes from PBS- or G protein–immunized mice, where IL-10 could be secreted by the cells in the presence of inactivated RSV as a stimulant (Fig. 6F). Proliferation of the splenocytes and the quantity of IL-10 in medium were measured 3 d later. As depicted in Fig. 6G, only the cells from the G+CSA-immunized animals suppressed G protein–primed T cell proliferation, whereas the cells from the G protein–immunized animals did not. The suppression was apparently associated with the high level of IL-10 produced (Fig. 6H). To demonstrate the direct involvement of IL-10 in suppression during RSV infection, splenocytes of the infected mice were stimulated with inactivated RSV in vitro with and without the addition of anti–IL-10 Ab or matched isotype IgG. T cell proliferation was measured after 3 d. As shown in Fig. 6I, the inhibition of T cell proliferation could be effectively reversed by addition of the anti–IL-10 Abs, but much less by the isotope control. Furthermore, when anti–IL-10 Ab was injected i.n. into G+CSA-immunized animals at 2 dpi, lymphocyte infiltration into the lung was more severe than in the isotype control (Fig. 6E). Evidently, IL-10 is one of the functional anti-inflammatory cytokines that limits pathogenic T cell activation and proliferation provoked by the RSV infection.

The effects and associated safety issues of RSV vaccines have puzzled researchers for >50 y, various strategies have been tried (84), but there is still no vaccine licensed for clinical use. Our study provides a new concept for RSV vaccine development, that is, that of inducing Treg cells without impairing Ab production.

The critical role of Treg cells in controlling the pathology of VED in RSV infection has been known for many years (47, 85). Openshaw and colleagues (45) showed that FI-RSV vaccination expelled Treg cells from lung and made pulmonary tissue prone to inflammation so that severe lung injury occurred upon RSV infection. Subsequently, manipulating the expelled Treg cells back into the lung was found to eliminate RSV infection–provoked inflammation and lung injury (45). We confirmed this phenomenon by the use of a Treg cell knockout animal model, because lack of Treg cells can be achieved by deletion in Foxp3-DTR transgenic mouse; the Treg-deleted mice showed more severe pneumonia than did the nondeleted mice (Fig. 4). Accordingly, we explored an implied corollary, that induction of Treg cells might be an important requisite for a successful RSV vaccine. Because we had previously shown that vaccination with a protein Ag plus an immunosuppressive agent such as CSA could induce potent immunosuppressive Treg cells, we tested RSV G protein plus CSA. The demonstrated efficacy of this approach may be ascribed, at least in part, to the induction of Treg cells in the dLNs and BAL where they may be most needed to suppress lung inflammation and injury (Fig. 3). This did not occur with the other vaccines, and they caused VED (Fig. 2). Inductions of these Treg cells may be associated with vaccine-induced rebalance of pDCs and conventional DCs as previous reported (30), but this remains to be tested. Although some Treg cell types have been documented to produce TGF-β (36, 86), the G+CSA-induced Treg cells did not (Fig. 6A), but they produced large amounts of IL-10, 6-fold more than the controls. Furthermore, the IL-10 was produced locally in the lung in response to RSV infection (Fig. 6B–D, 6J) and was demonstrated to participate in suppressing Ag-primed T cell activation and proliferation in vitro (Fig. 6G–I). Although IL-10 plays an important anti-inflammatory role, there may be other molecules, such as IL-35, that also contribute to this suppression, and this should be further investigated. Migration of these Treg cells from LNs to lung appeared to be required because it was prevented by inhibition of the CCL17/22–CCR4 axis of chemotaxis, and the Treg cells expressed high levels of CCR4 (Fig. 5).

There are contradictory reports of the effects of CSA upon immune responses. For example, some groups found that CSA reduced the function and level of Treg cells (87–89), but other evidence showed that CSA could induce Treg cells (76, 77, 90, 91). This discrepancy may be due to dose effects because we observed that a lower dose CSA was more effective in inducing Treg cells than was a higher dose (Fig. 1C). This has been also reported previously (75). Also, there have been some studies of the influence of CSA on B cells and the humoral response (92–94). CSA did not suppress Ab production in autoimmune diseases and transplantations (95–97), suggesting that the target mechanism in T cells is different from that in B cells; CSA targets calcium activation in T cells but may not affect that pathway in B cells (93, 98–100). However, we speculate that the mechanism by which CSA enhanced Ab production here may be indirect and via induction of Treg cells. There are contradictory reports of Treg effects on B cell function; that is, there may be suppression of B cells in one circumstance (101–103), but help for B cells in another (104–106). Help for B cells might be a consequence of IL-10 production by induced Treg cells. Although IL-10 is considered to be an anti-inflammatory cytokine, its enhancement of B cell activation and Ab production has been known for decades (97, 107–110). Induced Treg cells produce high levels of IL-10, and this has been suggested to lead to B cell maturation (108, 111, 112). Although it seems likely that the IL-10–producing Treg cell acted in this case as an alternative Th cell to promote B cell Ab production, the mechanism by which G protein+CSA induce high titers of anti-RSV IgG and neutralizing Ab (Fig. 1, Supplemental Fig. 2) needs further investigation.

We chose to work with the RSV G rather than the F glycoprotein because it has the advantages as a vaccine candidate of being easy to produce, having fewer requirements for a particular structure, and having several neutralizing epitopes within its sequence (113). However, the G glycoprotein may potently stimulate host immunity by interacting with CXC3R1, DC-SIGN, and L-SIGN of DCs to modulate the host immune system during the infection (57, 58, 114–116). Such strong stimulation of the immune responses can present a double-edged sword. On the one hand, there can be rapid host response to fend off infection, but on the other hand, the host system can overreact, leading to severe pathology. It appears that by adding CSA to the G protein the potent stimulation of the protective immune response was preserved and the harmful aspect eliminated, and further investigation of this vaccine is warranted.

We thank Dr. Douglas Lowrie (Shanghai Medical College, Fudan University, Shanghai, China) for valuable discussions and suggestions. We thank Xianghua Shi, Yongfu Xu, and Changsong Xu of Beijing Advaccine Biotechnology Co. Ltd. for fermentation and downstream processing, formulation, and analysis.

Chaofan Li, Y.Z., X.Z., Z.H., and B.W. were included as coinventors on the patent “A RSV vaccine preparation and methods of using the same” (application no. 201510082407.8). The other authors have no financial conflicts of interest.