Virally Delivered Cytokines Alter the Immune Response to Future Lung Infections

Abstract

Respiratory syncytial virus (RSV) is an important cause of infant morbidity and mortality worldwide and is increasingly recognized to have a role in the development and exacerbation of chronic lung diseases. There is no effective vaccine, and we reasoned that it might be possible to skew the immune system towards beneficial nonpathogenic responses by selectively priming protective T-cell subsets. We therefore tested recombinant RSV (rRSV) candidates expressing prototypic murine Th1 (gamma interferon [IFN-γ]) or Th2 (interleukin-4 [IL-4]) cytokines, with detailed monitoring of responses to subsequent infections with RSV or (as a control) influenza A virus. Although priming with either recombinant vector reduced viral load during RSV challenge, enhanced weight loss and enhanced pulmonary influx of RSV-specific CD8⁺ T cells were observed after challenge in mice primed with rRSV/IFN-γ. By contrast, rRSV/IL-4-primed mice were protected against weight loss during secondary challenge but showed airway eosinophilia. When rRSV/IL-4-primed mice were challenged with influenza virus, weight loss was attenuated but was again accompanied by marked airway eosinophilia. Thus, immunization directed toward enhancement of Th1 responses reduces viral load but is not necessarily protective against disease. Counter to expectation, Th2-biased responses were more beneficial but also influenced the pathological effects of heterologous viral challenge.

Viral lung infections are an important cause of mortality and morbidity worldwide (18). Respiratory syncytial virus (RSV) is the main cause of infantile bronchiolitis and therefore of infant hospitalization and is associated with asthma and wheezing in later life (28, 29). Studies of humans and animal models emphasize the importance of host responses in the pathogenesis of RSV disease (22).

In humans, severe RSV disease is associated with increased Th2 responses and with mutations associated with increased transcription of the interleukin-4 (IL-4) gene (6). An increased IL-4/gamma interferon (IL-4/IFN-γ) ratio is also seen in the nasal secretions of infants with RSV lower respiratory tract infection compared to those of infants with milder upper respiratory tract infection (16). Another type 2 cytokine, IL-9, has also been recently been found at high levels in the airways of infants with severe bronchiolitis (17).

A number of studies of murine models of RSV disease have defined roles for both Th1 and Th2 responses. IFN-γ receptor^−/− and IFN-γ^−/− mice display increased eosinophilia and airway hyperresponsiveness following RSV infection (24), whereas IL-4^−/− mice clear RSV with minimal pathology (2). Use of anti-IL-5 and anti-IL-4 antibodies during RSV infection also reduces the development of airway hyperresponsiveness after airway sensitization (24), and overexpression of IL-4 has also been shown to reduce cytotoxic T-lymphocyte activity and to reduce viral clearance in some cases (1, 11).

The pattern of response to RSV can also be modified by prior exposure to previous infections or vaccination with RSV antigens. Primary infection induces Th1 responses (14), and vaccination that enhances Th1 responses tends to be associated with reduced disease during infection (30). Expression of RSV G, F, or M2 results in T-helper 2 cell expansion and eosinophilia (31), T-helper 1 cell expansion and neutralizing antibodies (33), or cytotoxic CD8⁺ T-cell expansion, respectively (7). Immunization with formalin-inactivated RSV tends to induce Th2-biased responses and lung eosinophilia and enhanced disease during RSV infection, but this disease has recently been shown to be abrogated by the reduction of carbonyl groups in the formalin-treated vaccine (19).

One possible approach to the development of novel and effective vaccines is the use of recombinant vectors to deliver host cytokines to the site of infection, with the aim of programming protective, nonpathogenic host responses. For example, expression of IFN-γ augments Th1 responses and suppresses Th2 responses whereas that of IL-4 has the opposite effects (21). However, the effects of cytokine expression in viral vectors can lead to an adverse immune profile. For example, infection with recombinant mousepox (ectromelia virus) expressing IL-4 causes enhanced disease and widens lethality to encompass normally nonsusceptible mouse strains. Such recombinants cause fatal illness even in vaccinated hosts (15).

We have developed live recombinant RSV vectors expressing murine Th1 (IFN-γ) (4) or Th2 (IL-4) (5) cytokines. rRSV/IFN-γ shows many characteristics that make it seem a good vaccine candidate, with attenuated growth in vivo and good protection against subsequent viral replication (4). By contrast, the IL-4-expressing virus has been reported to cause delayed and reduced CD8⁺ T-cell responses without affecting viral clearance (5).

To test and compare these recombinants further, we studied the development of virus-specific immune responses to these two vectors in parallel and examined the effects of secondary challenge with RSV or influenza A virus in vaccinated mice. The results were unexpected: priming with rRSV/IFN-γ caused marked weight loss during secondary challenge with wild-type (wt) RSV RSV, with enhanced lymphocytosis and an increase in virus-specific CD8⁺ T cells. By contrast, rRSV/IL-4 did not cause any weight loss but did lead to significant lung eosinophilia during secondary challenge with either RSV or, surprisingly, influenza A virus infection.

MATERIALS AND METHODS

Virus stocks and mouse infection.

cDNA clones of mouse IL-4 and mouse IFN-γ were modified to be flanked by the RSV gene start and gene end transcription signals and XmaI sites. The PCR products were digested with XmaI and inserted into an XmaI site that had been engineered into the G-F intergenic sequence of a cloned cDNA of the RSV antigenome. The recombinant viruses were recovered by cotransfection of the antigenomic plasmid and plasmids expressing the N, P, L, and M2-1 support proteins into HEp-2 cells along with the recombinant modified vaccinia virus Ankara expressing T7 polymerase (4, 5).

Wild-type strain A2 RSV and recombinant RSV expressing murine IFN-γ (rRSV/IFN-γ) or IL-4 (rRSV/IL-4) were grown in HEp-2 cells. Four- to 8-week-old female BALB/c mice (Harlan Ltd., Queen's Park, UK) were maintained in pathogen-free conditions according to institutional and United Kingdom Home Office guidelines. Mice were inoculated with 5 × 10⁵ PFU of virus in 100 μl of medium intranasally (i.n.). Mice were challenged with 10⁶ PFU of wt RSV in 100 μl of medium. Influenza A virus strain X31 was obtained from the National Institute for Medical Research (NIMR). Five hemagglutinin (HA) units of the X31 strain were given i.n. in a volume of 100 μl of medium. After challenge infection, individual body weights were measured daily.

Cell recovery and analysis.

Collection of bronchoalveolar lavage (BAL) for cells and supernatants, harvesting of lung tissues, and staining for flow cytometry were carried out as previously described (8). For visualization of peptide-specific, cytokine-producing CD8⁺ T cells, 2 × 10⁶ lung cells were incubated with the RSV peptide M2_82-90 (SYIGSINNI) for 5 h in the presence of IL-2 (50 U/ml) and 10 μg/ml brefeldin A. Cells were analyzed on a BD LSR flow cytometer, collecting data on at least 50,000 events. Data was analyzed using WinList (Verity).

Quantification of viral RNA.

Total RNA was extracted from the lung by use of RNA stat-60 (AMS Biotech Ltd.), and cDNA was generated with random hexamers by use of Omniscript reverse transcriptase (QIAGEN). Real-time PCR was carried out for the RSV L gene by use of 900 nM forward primer (5′-GAACTCAGTGTAGGTAGAATGTTTGCA-3′), 300 nM reverse primer (5′-TTCAGCTATCATTTTCTCTGCCAAT-3′), and 100 nM probe (5′-6-carboxyfluorescein-TTTGAACCTGTCTGAACAT-6-carboxytetramethylrhodamine-3′) on an ABI Prism 7000 sequence detection system as described previously (9).

Cytokine ELISA.

Cytokine levels were assessed using BAL supernatants as described before (8). Briefly, enzyme-linked immunosorbent assay (ELISA) plates (Nunc) were coated with capture antibody (anti-IL-4 or anti-IFN-γ; BD) overnight at 4°C. Wells were washed and blocked with 1% bovine serum albumin for 1 h at room temperature. Sample or standard (100 μl) was added to blocked wells for 2 h. Bound cytokine was detected by using biotinylated anticytokine antibody, avidin-horseradish peroxidase, and tetramethylbenzidine. Color development was terminated with 2N H₂SO₄, and the optical density was read at 490 nm. The concentrations of cytokine were determined from the standard curve.

Virus-specific antibody ELISA.

Serum antibody was assessed by ELISA as described previously (8). RSV antigen was prepared by infecting HEp-2 cells with RSV at 1 PFU/cell. Microtiter plates were coated overnight with 100 μl of a 1:500 dilution of either RSV or HEp-2 antigen. Purified influenza virus X31 antigen was kindly provided by Alan Douglas (NIMR). The antigen (400 ng/well) was coated to microtiter plates overnight at 4°C in sodium carbonate buffer (pH 9.6). After blocking with 1% bovine serum albumin for 1 h, dilutions of test samples were added for a further 1 h. Bound antibody was detected using peroxidase-conjugated rabbit anti-mouse immunoglobulin (Ig) (Dako) and o-phenylenediamine as a substrate. Color development was blocked with 2 M H₂SO₄, and the optical density was read at 490 nm. RSV-specific antibody levels were determined by subtracting the RSV absorbance value from the HEp-2 absorbance value for the same sample. Specific isotypes were measured following the same protocol, changing the peroxidase-conjugated rabbit anti-mouse Ig secondary antibody for antibodies specific for mouse IgG1 and IgG2a.

Statistical analysis.

Results are expressed as means ± standard errors of the means (SEM). Statistical significance was calculated by analysis of variance followed by Student's t test for P value determination as described in the figure legends by use of GraphPad Prism software.

RESULTS

Primary infection with recombinant RSV expressing IFN-γ or IL-4.

BALB/c mice were infected with wt RSV, rRSV/IFN-γ, or rRSV/IL-4. Using TaqMan PCR to monitor viral load, we found that rRSV/IFN-γ levels were significantly attenuated, showing an approximate 10-fold reduction in peak viral replication on day 4 and day 8 (Fig. 1A), confirming previous findings obtained using viral plaque assay (4). On days 2 and 4, striking increases in the associated cytokine levels were found in the BAL fluid, indicating efficient in vivo expression of IFN-γ or IL-4 (Fig. 1B and C).

FIG. 1.

Primary infection of mice with recombinant RSV expressing murine cytokines. BALB/c mice (4 weeks old) were infected i.n. with wt RSV, rRSV/IFN-γ, or rRSV/IL-4. (A) Viral titer during primary infection, measured by TaqMan quantification of the RSV L gene. (B and C) Bronchoalveolar (airway) lavage cytokine levels after infection measured by ELISA for IFN-γ (B) and IL-4 (C). Symbols represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05 between outlier group and other group values.

Secondary infection with wild-type RSV of mice primed with recombinant RSV expressing IFN-γ or IL-4.

On day 28 after primary infection, mice were challenged with wt RSV. During secondary viral challenge, mice primed with rRSV/IFN-γ showed significantly enhanced and sustained weight loss compared to mice primed with wt RSV or rRSV/IL-4 (Fig. 2A). This weight loss was associated with an early boost in total BAL cell numbers (Fig. 2B and C), particularly those of lymphocytes (Fig. 2D). Mice primed with rRSV/IL-4 showed significant eosinophilia during secondary infection (Fig. 2E) that was not seen in the other groups of mice. However, this eosinophilia was not associated with increased weight loss. No virus was detected in any group on day 4 postchallenge (data not depicted).

FIG. 2.

Effects of primary infection with cytokine-expressing virus on secondary RSV challenge. Mice infected i.n. with wt RSV, rRSV/IFN-γ, or rRSV/IL-4 were challenged i.n. with wt RSV 4 weeks later. (A) Weight loss following second challenge (2°). (B and C) Airway cell numbers 4 days (B) and 7 days (C) after challenge. (D and E) Percent airway lymphocyte (D) and eosinophil (E) cell types 7 days after challenge. Symbols and bars represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05 between outlier group and other group values.

As seen upon examining the lung lymphocyte subsets in more detail, rRSV/IFN-γ primed for enhanced CD8⁺ T-cell responses during secondary challenge (Fig. 3A) but for fewer CD4⁺ cells (Fig. 3B). DX5⁺ (natural killer [NK]) cells tended to be most abundant on day 4 and declined in number on day 7 and day 15 in mice primed with wt RSV or rRSV/IFN-γ. However, mice primed with rRSV/IL-4 showed a reversal of this pattern, with a low initial NK response followed by numbers climbing and remaining sustained to day 15 (Fig. 3C).

FIG. 3.

Lymphocyte subsets in lungs after RSV rechallenge. Mice infected i.n. with wt RSV, rRSV/IFN-γ, or rRSV/IL-4 were challenged i.n. with wt RSV 4 weeks later (2°). Lung cells were analyzed by flow cytometry using CD8⁺ T cells (A), CD4⁺ T cells (B), and NK cells (C). Symbols represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05.

To examine the functional specificity of the CD8⁺ T-cell subset primed with rRSV/IFN-γ, lung cells were restimulated in vitro 7 days after challenge with the M2 peptide (SYIGSINNI), the immunodominant RSV major histocompatibility complex class I epitope for BALB/c mice (derived from amino acids 82 to 90 of the RSV M2-1 protein). CD8⁺ cells expressing IFN-γ in response to this peptide were abundant in mice primed with rRSV/IFN-γ. Expression of IL-4 during primary infection had no significant effect on the abundance of IFN-γ-positive CD8⁺ cells during challenge (Fig. 4A and B). No IL-4-producing CD8⁺ cells were detected (data not shown), and there were no significant differences in the numbers of peptide-specific tumor necrosis factor alpha-producing cells.

FIG. 4.

rRSV/IFN-γ-primed mice show increased levels of RSV-specific IFN-γ-secreting CD8⁺ cells in lungs. Mice infected i.n. with RSV, rRSV/IFN-γ, or rRSV/IL-4 were challenged i.n. with wt RSV 4 weeks later. Lung cells from mice obtained at day 7 after challenge were stimulated ex vivo with RSV M2 peptide. (A) Sample fluorescence-activated cell sorter plots for individual mice. (B) Pooled data. Symbols represent individual mice; lines represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05.

The rate of appearance and levels of RSV-specific immunoglobulin during primary infection were unaffected by the presence of virally expressed IL-4 or IFN-γ (Fig. 5A). However, on day 15 of primary infection, expression of IL-4 significantly boosted IgG1 and depressed IgG2a responses to RSV (Fig. 5B). By contrast, expression of IFN-γ depressed IgG1 responses on day 15 of primary infection (Fig. 5B). During secondary challenge with RSV, mice initially infected with rRSV/IFN-γ had depressed levels of total RSV-specific immunoglobulin (Fig. 5C) whereas expression of IL-4 resulted in a marginal increase of the antibodies. Expression of IL-4 during primary infection tended to boost IgG1 responses during secondary challenge whereas expression of IFN-γ led to some reduction in IgG2a responses (Fig. 5D).

FIG. 5.

Priming with rRSV/IL-4 alters the subtype of RSV-specific serum antibody. RSV-specific antibody levels were measured by ELISA in sera of mice infected i.n. with wt RSV, rRSV/IFN-γ, or rRSV/IL-4. (A and B) Primary infection. Total values for anti-RSV Ig (A) and IgG subtypes IgG2a and IgG1 (B) measured at 15 days postinfection are shown. (C and D) RSV rechallenge. Total values for anti-RSV Ig (C) and IgG subtypes IgG2a and IgG1 (D) 15 days after RSV rechallenge are shown. Symbols and bars represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05. A₄₉₀ values were obtained at 1:400 dilution.

The influence of recombinant RSV infection on challenge with influenza A virus.

To determine whether the effects detected were specific to RSV, we tested the effect of rRSV on challenge with influenza A virus. Mice were infected with wt RSV, rRSV/IFN-γ, or rRSV/IL-4 or left naive; 4 weeks later, mice were challenged i.n. with 5 hemagglutinin units of the X31 strain of influenza A virus, and groups of mice were harvested on days 3, 8, and 16 after challenge.

Mice that had received wt RSV or rRSV/IL-4 showed faster recovery from influenza virus infection-induced weight loss than naïve mice or mice that had been primed with rRSV/IFN-γ (Fig. 6A). Influenza virus titers in all four groups were similar on day 3 and day 8 postchallenge, and the RSV-L gene was not detectable prior to or during challenge (data not depicted).

FIG. 6.

RSV is partially protective against challenge with influenza A virus. BALB/c mice (4 weeks old) were infected i.n. with wt RSV (gray squares), rRSV/IFN-γ (white circles, broken line), or rRSV/IL-4 (black diamonds) or left as naïve mice (white squares). At 4 weeks later the mice were infected with X31 influenza virus i.n. (A) Weight loss following rechallenge. (B to F) Airway cell numbers (B), anti-influenza A virus and anti-RSV total immunoglobulin levels (C), CD4⁺ T-cell numbers (D), CD8⁺ T-cell numbers (E), and airway eosinophilia levels (F) on day 8 after challenge are shown. (G to I) BAL cytokine levels (in picograms per milliliter) for IL-4 (G), IL-5 (H), and IFN-γ (I) on day 8 after challenge are shown. Symbols and bars represent means ± SEM of values obtained with mice (n ≥ 4). *, P < 0.05; **, P < 0.01; ***, P < 0.001. 1°, initial challenge.

BAL was performed on days 3, 8, and 16 postchallenge. Total viable cell counts showed that cellular infiltration was greatest in all groups on day 8. Mice primed with rRSV/IFN-γ showed significantly increased BAL cell counts on days 3 and 8 compared to mice treated in other ways (data not depicted and Fig. 6B). Conversely, mice primed with rRSV/IL-4 showed reduced cell numbers on day 8 postchallenge compared to other mice (Fig. 6B). Primary infection with recombinant RSV had no significant effect on anti-influenza virus or anti-RSV antibody titers (Fig. 6C).

At day 8 postinfection, the majority of BAL cells in naïve mice or mice primed with wt RSV or rRSV/IFN-γ were CD8⁺ T cells (Fig. 6E). CD8⁺ T-cell levels were similar in naïve mice and mice infected with rRSV/IL-4 whereas CD4⁺ T-cell levels were slightly higher in mice primed with RSV/IL-4 (Fig. 6D). However, rRSV/IL-4 priming led to a significant increase in the numbers of eosinophils present in the BAL postchallenge both at day 3 (data not depicted) and at day 8 (Fig. 6F). At day 8 postchallenge, elevated levels of BAL IL-4 (Fig. 6G) and IL-5 (Fig. 6H) were seen in mice primed with rRSV/IL-4 and of IFN-γ in mice primed with rRSV/IFN-γ (Fig. 6I).

DISCUSSION

Our recombinant RSV vectors were highly effective at producing murine cytokines in vivo. Primary infection with wt RSV or cytokine-expressing viruses induced no appreciable weight loss but did induce distinct patterns of host immunity which resulted in reduced viral load during subsequent challenge with nonrecombinant RSV. An unexpected finding was that priming with rRSV/IFN-γ resulted in weight loss during secondary challenge with native RSV. This altered immunopathology was characterized by an influx of lymphocytes into the lungs during challenge, especially of RSV-specific CD8⁺ T cells producing IFN-γ. By contrast, rRSV/IL-4 induced lung eosinophilia during secondary challenge but without evidence of disease augmentation as characterized by weight loss. Remarkably, the downstream effects of RSV-delivered cytokines were not limited to RSV challenge: when influenza virus was used as the challenge virus, we again observed an enhanced CD8⁺ T-cell influx in rRSV/IFN-γ-primed mice and eosinophilia in mice primed with rRSV/IL-4.

Other studies have shown the potential of IL-4 to increase the severity of viral infections (27). Coexpression of IL-4 in ectromelia virus overcame genetic resistance to infection, making the recombinant lethal to resistant mouse strains (15). This suggested that the presence of IL-4 might lead to an excessive Th2 response and increase viral replication; by contrast, we found no such pathogenic effect with rRSV/IL-4. This virus was essentially nonpathogenic and induced protective immunity to secondary RSV challenge. The rRSV/IL-4 virus also dampened the T-cell response during challenge, with fewer CD4 and CD8 cells detectable in the lungs. These differences may have been due to the different natures of the viruses and viral clearance mechanisms. In the case of ectromelia virus, both NK and CD8⁺ T cells are important in lysis of infected cells and reducing the viral load whereas cellular immunity is often pathogenic in RSV disease. This suggestion is supported by the observation that local overexpression of IL-4 in mice infected with influenza A virus does not affect viral clearance or mortality compared to wild-type control results despite a significantly reduced CD8⁺ T-cell response (3).

Altering the secondary outcome by use of different priming regimens not only applies to the whole virus but also to the delivery of individual viral proteins. It has been shown that priming with various RSV proteins causes weight loss, illness, and enhanced lung pathology during RSV challenge (22). In some ways, the effects we observed with rRSV/IL-4 mirror the effects of immunization with vaccinia virus expressing RSV glycoprotein G (which leads to enhanced Th2 responses and lung eosinophilia during challenge). Eosinophilia per se does not appear to cause enhanced weight loss. It may, however, lead to airway remodeling (13) or airway hyperresponsiveness (25), neither of which we studied. The effects of rRSV/IFN-γ parallel those of priming with vaccinia virus expressing the RSV M2 protein (which generates powerful CD8⁺ T-cell responses and weight loss) (22). These results underline the need for caution when developing vaccines designed to skew T-helper responses. In cases of viral infections, there is a general presumption that Th1 responses are safe and antiviral but that Th2 responses are pathogenic, but the present results instead show the opposite. While cytokines can be used as effective adjuvants in some models (12), their use needs to be carefully considered on a case-by-case basis.

Another interesting phenomenon is the nonspecific effect of viral cytokine delivery. It has been shown that previous exposure to pathogenic agents can alter the course of secondary respiratory infections. This effect can be seen either in the same organ (10) or in different organs (34). Our data on heterologous secondary infection show that rRSV/IL-4 can induce enhanced levels of IL-4 during influenza virus challenge and result in eosinophil recruitment. We think this is probably due to the presence of bystander RSV-specific Th2 T cells early during heterologous challenge, thus influencing the phenotype and function of influenza virus-specific T cells. The effect of rRSV/IFN-γ on the recruitment of CD8⁺ T cells during influenza A virus challenge is also of interest. It has been shown that influenza virus-specific CD8⁺ T cells are recruited during RSV infection and can dampen eosinophilia (32) and that RSV-specific T cells remain in the airways for up to 50 days postinfection (23). Thus, it is possible that bystander recruitment of IFN-γ-primed RSV-specific CD8⁺ T cells is responsible for this increase. However, no difference was observed in the percentages of RSV-specific CD8⁺ T cells during influenza virus rechallenge. It is also possible that infection creates inducible bronchus-associated lymphoid tissue which may skew future infections (20).

Another possibility is that use of recombinant RSV led to persistent infection (26) and thus to persistent production of the encoded cytokine. However, we were unable to detect the RSV L gene at any time later than day 12 after primary infection with the recombinants by use of TaqMan real-time PCR. Further, RSV-F and NS2 genes (which have higher transcriptional frequencies) were undetectable prior to or during influenza virus rechallenge (data not depicted). We ascribe the absence of persistence to the use of recombinant RSV vectors instead of the strains used in the previous study and believe that persistence of recombinant virus is not responsible for the altered phenotype observed during secondary challenge. Rather, we propose that innate and acquired immune responses to secondary challenge were modified by local antigenic and cytokine “imprinting” resulting from the primary infection.

This work is of interest because it shows that cytokine delivery by viral vectors has an effect on the target organ not only during the initial primary infection but also during later challenge with the same pathogen. Furthermore, cytokine delivery can skew the immune environment in the lung so that heterologous infections are also affected, and Th1 skewing is not always beneficial to the safe and effective clearance of viral infections.