|Home | About | Journals | Submit | Contact Us | Français|
Secondary exposure to respiratory syncytial virus (RSV) can lead to immunopathology and enhanced disease in vaccinated individuals. Vaccination with individual RSV proteins influences the type of secondary RSV-specific immune response that develops upon challenge RSV infection, as well as the extent of immunopathology. RSV-specific memory CD4 T cells can directly contribute to immunopathology through their cytokine production. Immunization of BALB/c mice with a recombinant vaccinia virus (vv) expressing the attachment (G) protein of RSV results in pulmonary eosinophilia upon RSV challenge, whereas immunization of mice with a vv expressing the fusion (F) protein does not. We analyzed the CD4 T-cell response to an I-Ed-restricted CD4 T-cell epitope within the F protein of RSV corresponding to amino acids 51 to 66 in an effort to better understand the similarities and differences in the immune response elicited by the G versus the F protein. Vaccination with the G protein induces a mixture of RSV G-specific Th1 and Th2 cells with a restricted T-cell receptor repertoire. In contrast, we demonstrate here that immunization with the F protein elicits a broad repertoire of RSV F-specific CD4 T cells that predominantly exhibit a Th1 phenotype. However, in the absence of gamma interferon (IFN-γ), RSV F51-66-specific CD4 T cells secreted interleukin-5, and mice developed pulmonary eosinophilia after RSV challenge. IFN-γ-deficient mice exhibited decreased weight loss compared to wild-type controls, suggesting that IFN-γ exacerbates systemic disease. These data demonstrate that IFN-γ can have both beneficial and detrimental effects during a secondary RSV infection.
Respiratory syncytial virus (RSV) is the leading cause of lower respiratory tract disease in young children (16). In the United States alone, RSV infection accounts for an estimated 100,000 hospitalizations and approximately 1,000 deaths per year (16, 37, 38, 44). RSV also causes serious disease in the elderly and immunocompromised (11, 12). During a series of RSV vaccine trials conducted during the 1960s, children vaccinated with a formalin-inactivated RSV experienced enhanced morbidity and mortality following a natural RSV infection (8, 13, 26, 27). Histological analysis of lung tissue obtained at autopsy revealed extensive mononuclear cell infiltration in the lung with an elevated frequency of eosinophils (27).
The vaccine-enhanced disease exhibited by the children that received the formalin-inactivated RSV vaccine can be mimicked by priming BALB/c mice with a recombinant vaccinia virus (vv) expressing the attachment (G) protein of RSV (1, 14, 29, 50). The G protein is both secreted and expressed on the surface of the virion (17) and allows for attachment of the virus to the host cell plasma membrane, thereby making it an attractive target as a potential vaccine candidate. The immune response to the G protein has been well studied and is known to elicit a CD4 T-cell response (30) and no CD8 T-cell response (32) in BALB/c mice. The G-specific memory CD4 T cells are a mixture of Th1 and Th2 cells that are directed against a single immunodominant epitope (G183-195) (48). The majority of the RSV G183-195-specific memory CD4 T cells express the Vβ14 T-cell receptor (47). Depletion of Vβ14+ cells from vvG-immunized BALB/c mice is sufficient for a significant decrease in pulmonary eosinophilia upon RSV challenge (25, 47). It has also been shown that depletion of cells expressing T1/ST2, a protein found on the surface of mouse Th2 cells, significantly reduces pulmonary eosinophilia (49). Thus, it is known that immunization with the RSV G protein elicits an oligoclonal CD4 T-cell response that is responsible for inducing pulmonary eosinophilia during challenge RSV infection.
The fusion (F) protein of RSV is expressed on the surface of the RSV virion and mediates fusion between the virion envelope and the host cell plasma membrane. Neutralizing monoclonal antibodies (MAbs) directed against the F protein inhibit RSV replication in vivo and, when used as a prophylactic, protect against disease in animal models and high-risk children (2, 9, 21, 42, 43). In contrast to immunization with vvG, mice vaccinated with a recombinant vv expressing the F protein of RSV (vvF) do not develop pulmonary eosinophilia upon challenge RSV infection (19, 29, 40). For these reasons, the F protein represents an attractive target for future RSV vaccines. Due to the enhanced disease observed after vaccination with the G protein of RSV, it is important to gain a better understanding of the CD4 T-cell-mediated immunopathology elicited after infection of individuals vaccinated with the F protein. The F protein of RSV is known to elicit both CD4 and CD8 T-cell responses (19, 30, 32). An Ld-restricted CD8 T-cell epitope has been identified at amino acids 85 to 93 in the F protein (7, 20), whereas the CD4 T-cell epitope(s) within the F protein remain poorly defined (30), making comparisons of F-specific CD4 cells to their G-protein counterparts difficult. For example, it is currently unclear if the F protein fails to elicit a Th2 response or if this response is merely inhibited by the RSV F-specific CD8 T cells, as has been suggested by previous studies (19, 39).
The lack of a precisely defined CD4 T-cell epitope within the F protein has hindered our ability to track and quantify RSV F-specific CD4 T cells after vaccination and/or RSV infection. In order to study the mechanism(s) responsible for preventing the development of pulmonary eosinophilia in vvF-immunized mice, we precisely identified a CD4 T-cell epitope within the F protein of RSV. We show that RSV F51-66-specific memory CD4 T cells are a heterogeneous population that predominantly make gamma interferon (IFN-γ) upon in vitro stimulation. We demonstrate that in the absence of IFN-γ, vvF-immunized BALB/c mice develop pulmonary eosinophilia and RSV F51-66-specific memory CD4 T cells produce IL-5 after challenge RSV infection. In addition, we show that vvF-immunized IFN-γ-deficient mice have increased levels of CCL11, CCL17, and CCL22 protein in the lung and increased levels of CCL17 and CCL22 in the bronchioalveolar lavage (BAL). Our data suggest that IFN-γ production inhibits the secondary expansion of RSV F-specific memory Th2 cells and/or the recruitment of eosinophils to RSV-infected lungs thus preventing the development of pulmonary eosinophilia in vvF-immunized mice undergoing challenge RSV infection. We also show that IFN-γ is necessary for weight loss and clinical signs of systemic disease after RSV challenge of vvF-immunized mice, demonstrating a complex role for IFN-γ in both inhibiting and exacerbating different aspects of RSV vaccine-enhanced disease.
Female BALB/c (BALB/cAnNCr) mice were purchased from the National Cancer Institute (Frederick, MD). Male and female IFN-γ [C.129S7(B6)-Ifngtm1Ts/J]-deficient mice on BALB/c background were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice aged 6 to 8 weeks were used for all experiments. All experimental procedures were approved by the University of Iowa Animal Care and Use Committee.
All peptides were purchased from BioSynthesis (Lewisville, TX). Previously reported peptides used in our studies included RSV F 51-70 (GWYTSVITIELSNIKENKCN) (10), RSV F 146-160 (VSVLTSKVLDLKNYI) (45), and RSV F 184-199 (SAIASGIAVSKVLH) (45). The peptide sequences for all other peptides used in the present study are listed in Fig. Fig.1B1B and and2A2A.
Recombinant vv stocks were gifts from Gail W. Wertz and Thomas J. Braciale (University of Virginia, Charlottesville), and Judy A. Beeler (U.S. Food and Drug Administration, Bethesda, MD). All vv's were grown in BSC-40 cells (American Type Culture Collection [ATCC], Manassas, VA). Mice were immunized with 3 × 106 PFU of a recombinant vv expressing either β-galactosidase (vvβ-gal) or the fusion protein of RSV (vvF) by scarification with a 25-gauge needle at the base of the tail. RSV (A2 strain) was a gift from Barney S. Graham (National Institutes of Health, Bethesda, MD) and was grown in HEp-2 cells (ATCC). After 3 weeks, immunized mice were anesthetized with 30% halothane (Halocarbon Laboratories, River Edge, NJ) in mineral oil (Fisher Scientific, Fair Lawn, NJ) and were given 1 × 106 to 3 × 106 PFU RSV intranasally. The weight of each mouse was determined and recorded daily. Each mouse was given an illness score daily based on the following scale: 0, no apparent illness; 1, slightly ruffled fur; 2, ruffled fur, active; 3, ruffled fur, inactive; 4, ruffled fur, inactive, hunched posture; and 5, moribund or dead. Cells in the BAL were collected by performing three washes of the airspace of the lung, each with 1 ml of phosphate-buffered saline (PBS). BAL cells were then counted, cytospun, and stained with Diff-Quik (Dade Behring, Inc., Newark, DE) to determine the percentage and total number of eosinophils. The lung vascular bed was perfused by injecting 5 ml of RPMI medium (Gibco, Grand Island, NY) supplemented with 10% bovine growth serum (HyClone, Logan, UT), 10 U of penicillin G/ml, 10 μg of streptomycin sulfate/ml, 2 mM l-glutamine (Gibco), 5 × 10−5 M 2-mercaptoethanol, 1 mM sodium pyruvate (Gibco), 0.1 mM nonessential amino acids (Gibco), and 10 mM HEPES (Gibco) into the right ventricle of the heart. The lungs were then dissected, and single-cell suspensions were prepared by pressing the lungs though a wire mesh screen (Bellco Glass, Inc., Vineland, NJ).
Lung cells (2 × 106 cells/ml) were stimulated with 1 μM peptide in the presence of 10 μg of brefeldin A (Sigma, St. Louis, MO)/ml for 5 h at 37°C in RPMI medium supplemented as described above. In some experiments, 10 μg of an anti-I-Ad antibody (clone 34-5-3; BD Pharmingen, San Diego, CA) or an anti-I-Ad/I-Ed antibody (clone M5/114.15.2; eBioscience, San Diego, CA)/ml was included during incubation. After incubation, the cells were blocked with purified anti-FcγRII/III monoclonal antibody (clone 93; eBioscience) and stained with anti-CD4 MAb (allophycocyanin or fluorescein isothiocyanate-conjugated, both from eBioscience [clone RM4-5]). For determination of T-cell receptor (TCR) chain expression, anti-T-cell receptor α- or β-chain antibodies were used. FACS lysing solution (Becton Dickinson, San Jose, CA) was used to fix the lung mononuclear cells and lyse the red blood cells. Cells were then washed with permeabilization buffer (staining buffer containing 0.5% saponin; Sigma) and stained with a phycoerythrin-conjugated, anti-cytokine MAb, or an isotype control. Cells were collected on either a FACSCalibur or FACSCanto (Becton Dickinson). Single color controls were used for compensation. Lymphocytes were gated based on forward-scatter and side-scatter properties and were then analyzed by using FlowJo software (Tree Star, Inc., Ashland, OR). Background staining was determined by using either isotype controls or no peptide controls.
The following phycoerythrin-conjugated Vα-chain antibodies were used: Vα2 (clone B20.1; Caltag, Burlingame, CA), Vα8 (clone CTVA8; Caltag), and Vα8.3 (clone B21.14; Pharmingen). The following fluorescein isothiocyanate-conjugated Vβ-chain MAbs (all were from Pharmingen except as noted) were used: Vβ2 (clone B20.6), Vβ3 (clone KJ25), Vβ4 (clone KT4), Vβ5.1/5.2 (clone MR9-4), Vβ6 (clone R44-7), Vβ7 (clone TR310), Vβ8 (clone F23.1), Vβ8.1/8.2 (clone MR5-2), Vβ8.2/8.3 (clone CT8E; Caltag), Vβ8.3 (clone 1B3.3), Vβ9 (clone MR10-2), Vβ10b (clone B21.5), Vβ11 (clone RR3-15), Vβ12 (clone MR11-1), Vβ13 (clone MR12-3), Vβ14 (clone 14-2), and Vβ17a (clone KJ23). Cytokine antibodies and isotypes (all eBioscience unless otherwise noted) used were as follows: interleukin-4 (IL-4; clone 11B11), IL-5 (clone TRFK5; Pharmingen), IL-6 (clone MP5-20F3), IL-10 (clone JES5-16E3), IFN-γ (clone XMG1.2), rat immunoglobulin G1 (IgG1; clone R3-34), and rat IgG2b (clone KLH/G2b-1).
Wells of a 48-well plate (Falcon, Franklin Lakes, NJ) were coated overnight with 10 μg of anti-CD3 antibody (clone 145-2C11; eBioscience)/ml in PBS. After several washes with PBS, 2 × 106 lung mononuclear cells were incubated with or without 1 μM peptide or placed on anti-CD3 coated or placed on uncoated wells in the presence or absence of 1 μM RSV F51-66 peptide for 48 h. Cell-free supernatant was collected and frozen at −80°C until further use. Nunc-Immuno Plates with MaxiSorp Surface (Nalge Nunc International, Rochester, NY) were coated overnight at 4°C with 2 μg of capture antibody/ml in 0.1 M Na2HPO4 (pH 9.0). After a wash with PBS-0.5% Tween 20 (Sigma), plates were blocked with RPMI media supplemented as described above for at least 2 h at room temperature. Plates were incubated with 100 μl of samples overnight at 4°C. Cytokine was detected by incubation with 0.1 μg of biotinylated anti-cytokine antibody (IL-13)/ml or 1 μg of biotinylated anti-cytokine antibody/ml (all others) for 2 h at room temperature. Avidin-peroxidase (1:400 dilution; Sigma) was added for 30 min before the plates were developed with 3,3′,5,5′-tetramethyl-benzidine dihydrochloride (Sigma). The reaction was stopped after 5 min with 2 N H2SO4 (Ricca Chemical Company, Arlington, TX). Plates were read at 450 nm using an ELx800 plate reader and analyzed by using KC Junior software (both from Bio-Tek Instruments, Winooski, VT).
The following anti-cytokine pairs were used (all from eBioscience unless otherwise noted): IL-4, purified clone 11B11 and biotinylated clone BVD6-24G2; IL-5, purified clone TRFK5 and biotinylated clone TRFK4; IL-13, purified clone 38213.11 and biotinylated goat anti-mouse IL-13 (both from R&D Systems); IFN-γ, purified clone R4-6A2 and biotinylated clone XMG1.2 (both from eBioscience). Recombinant murine IL-4 (eBioscience), IL-5 (Pharmingen), IL-13 (R&D Systems), and IFN-γ (eBioscience) were used to calculate standard curves.
ELISA for chemokines was performed as described above for cytokines. Plates were coated with either 0.4 μg/ml (CCL11 and CCL22) or 2 μg/ml (CCL17) of capture antibody (R&D Systems). Whole lungs were harvested and homogenized by using glass homogenizers (Kontes Glass Co., Vineland, NJ) in 1 ml of RPMI media supplemented as described above and containing a 1:200 dilution of protease inhibitor cocktail (Sigma). BAL and lung cells were centrifuged, and 50 μl of supernatant was incubated on the precoated plates overnight at 4°C. Recombinant murine CCL11, CCL22, and CCL17 (R&D Systems) were diluted in PBS plus 10% fetal calf serum (FCS; Atlanta Biologicals, Norcross, GA) and used to calculate standard curves. Chemokine detection was performed with 0.1 μg biotinylated anti-chemokine antibody (R&D Systems)/ml.
Whole lungs with the heart attached were harvested from vvβ-gal- and vvF-immunized mice 7 days after RSV challenge. Lungs were fixed in 10% neutral buffered formalin (Fisher Scientific) prior to being processed and paraffin embedded at the University of Iowa Comparative Pathology Laboratory. Paraffin blocks were sectioned at a thickness of 5 μm. Sections were hematoxylin and eosin (H&E) stained at the University of Iowa Central Microscopy Core. Eosinophils were detected by immunoperoxidase reaction for the eosinophil-specific major basic protein (MBP) using the avidin-biotin complex method. Sections were deparaffinized in xylenes, rehydrated in graded alcohols, and rinsed with 1× Dako buffer (Dako, Carpinteria, CA). Proteolytic digestion was accomplished by using proteinase K (Dako) for 5 min at room temperature. Endogenous peroxidase activity was quenched by using a 3% hydrogen peroxide solution for 8 min at room temperature. Endogenous biotin staining was blocked by the application of 1.5% normal rabbit serum (Dako) for 30 min at room temperature. Sections were covered with rat anti-mouse monoclonal primary antibody (kindly provided by Nancy and James Lee, Mayo Clinic, Scottsdale, AZ) at a 1:500 dilution in 1.5% rabbit serum, whereas the negative control was rat IgG (Sigma) using a dilution of 1:2,760 in 1.5% rabbit serum, incubated for 60 min at room temperature, and rinsed. Sections were then covered with biotinylated anti-rat secondary antibody (Vector Laboratories, Burlingame, CA), using a dilution of 1:200 in 1.5% rabbit serum for 60 min at room temperature, rinsed, and then covered in avidin-biotin complex for 60 min at room temperature and rinsed. Sections were incubated with diaminobenzidine chromogen for 5 min to demonstrate the signal of the primary antibody. A counterstain of hematoxylin (Surgipath, Richmond, IL) was applied for 30 s at room temperature. The sections were then blued in tap water, dehydrated, cleared, and mounted. Slides were blinded and scored by a board-certified veterinary pathologist (David Meyerholz, University of Iowa). H&E-stained slides were scored by using the following scales according to the specified area of the lung: (i) airway—0, no detectable inflammation; 1, rare to uncommon inflammatory cells or debris; 2, small numbers of airway inflammatory cells or debris; 3, moderate numbers of inflammatory cells or debris; 4, moderate to severe inflammatory cells or debris with uncommon airway plugging; and 5, severe multifocal inflammatory cells or debris with airway plugging; (ii) alveolar—0, no detectable inflammation; 1, uncommon inflammatory cells; 2, small to moderate numbers of inflammatory cells mostly in septa; 3, moderate inflammation mostly in septa with some rare spillover into alveoli; 4, moderate to severe inflammation extending into alveoli; and 5, severe inflammation and consolidation with multifocal to lobar plugging by cells and fluid; (iii) peribronchiolar—0, no detectable inflammation; 1, rare individual inflammatory cells; 2, small localized numbers of inflammatory cells; 3, small aggregates of cellular inflammation; 4, moderate to severe aggregates of cellular inflammation; and 5, severely expanded aggregates of cellular inflammation; and (iv) perivascular—0, no detectable inflammation; 1, rare individual inflammatory cells; 2, small local numbers of inflammatory cells; 3, small to moderate aggregates of cellular inflammation with mild exocytosis; 4, moderate to severe aggregates of cellular inflammation and edema with active exocytosis; and 5, severely expanded aggregates, edema and exocytosis. Anti-MBP-stained slides were scored for aggregates according to the following scale: 0, no to rare perivascular staining; 1, detectable perivascular staining; 2, small clusters of staining; 3, moderately defined aggregates; and 4, well-defined aggregates. Eosinophils in the lung interstitium were quantified by averaging counts from five random high-power fields.
Lungs were harvested from vvβ-gal- and vvF-immunized wild-type and IFN-γ-deficient BALB/c mice 4 days after RSV challenge. Lungs were homogenized in 1 ml of TRIzol (Invitrogen, Carlsbad, CA), and supernatants were collected. RNA was purified by sequential chloroform (200 μl/lung; Fisher Scientific) and isopropyl alcohol (500 μl/lung; Fisher Scientific) extraction. Pellets were washed with 70% ethanol and air dried before resuspension in distilled water at 55°C for 10 min. RNA was cleaned by using the RNeasy Plus minikit (Qiagen, Valencia, CA). cDNA was prepared by using a SuperScript First-Strand synthesis kit for RT-PCR (Invitrogen) and used as a template for real-time PCR. Real-time PCRs to detect the nucleocapsid (N) gene of RSV were performed with TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA) on either an ABI 7000 or 7300 real-time PCR system (Applied Biosystems) using universal thermal cycling parameters. The N-gene primers and probe have been previously described (18). The probe was synthesized to contain 5′ 6-carboxyfluorescein reporter dye and 3′ carboxytetramethylrhodamine quencher dye (Integrated DNA Technologies, Coralville, IA). Samples were compared to known standard dilutions of a plasmid containing the N gene of RSV. The results were analyzed by using sequence detection system analysis software (Applied Biosystems). The number of copies of the N gene per lung was calculated based on number of copies of N gene in the sample and the total RNA isolated from the lung.
Whole lungs were harvested from vvβ-gal- and vvF-immunized wild-type or IFN-γ-deficient mice, weighed, and homogenized by using an ULTRA-TURRAX T 25 basic homogenizer (IKA, Wilmington, NC). Supernatants were collected and stored at −80°C until further use. Vero cells (ATCC) were grown in minimal essential medium (Gibco) supplemented with 10% FCS (Atlanta Biologicals), 10 U of penicillin G/ml, 10 μg of streptomycin sulfate/ml, 2 mM l-glutamine (Gibco), 5 × 10−5 M 2-mercaptoethanol, 1 mM sodium pyruvate (Gibco), and 0.1 mM nonessential amino acids (Gibco) to 90 to 95% confluence in six-well plates (Falcon). Serial dilutions of supernatants were performed in serum-free minimal essential medium supplemented as described above, followed by incubation on Vero cells for 90 min at 37°C. Plates were rocked every 15 min and overlaid with Eagle minimum essential medium (Lonza, Walkersville, MD) supplemented with 10% FCS (Atlanta Biologicals), 10 U of penicillin G/ml, 10 μg of streptomycin sulfate/ml, and 2 mM l-glutamine (Gibco) and mixed 1:1 with 1% SeaKem ME agarose (Cambrex, North Brunswick, NJ). After 5 days, plates were stained with 1:1 mixture of supplemented Eagle minimum essential medium and 1% agarose with the addition of 1% neutral red (Sigma). Plaques were counted after 24 to 48 h with the assistance of a light box.
Spleens were harvested from wild-type and IFN-γ-deficient mice 3 weeks after immunization with vvF, and single-cell suspensions were prepared by pressing the tissue between the ends of two frosted glass slides. Red blood cells were lysed by the addition of 0.84% ammonium chloride to the cell suspension, followed by washing in supplemented RPMI media. Wells of a 96-well filtration plate (Millipore, Billerica, MA) were coated overnight with 2 μg of purified anti-tumor necrosis factor alpha (anti-TNF-α; clone 1F3F3D4; eBioscience) or anti-IL-5 (clone TRFK5; eBioscience) antibody/ml. Plates were washed with PBS and blocked with medium supplemented as described above. Cells were plated in duplicate twofold dilutions starting with 5 × 106 cells. Plates were incubated at 37°C for 48 h. After a wash with PBS-0.5% Tween 20 (Sigma), the plates were incubated with 2 μg of biotinylated anti-TNF-α (clone MP6-XT22/MP6-XT3; eBioscience) or anti-IL-5 (clone TRFK4; eBioscience) antibody/ml for 2 h. After a washing step, a 1:400 dilution of a 1-μg/ml avidin-peroxidase solution (Sigma) was added for 30 min. The plates were washed and developed with a 10-mg/ml solution of 3-amino-9-ethyl-carbazole (Sigma) in N,N-dimethylformamide (Sigma) diluted in 30 ml of 0.1 M phospho-citrate buffer and filtered by using a 0.45-μm-pore-size syringe filter (Whatman, Clifton, NJ). Plates were read on an ImmunoSpot reader (Cellular Technology, Ltd., Cleveland, OH) and counted by using ImmunoSpot Software (Cellular Technology).
All statistical analyses were performed by using GraphPad InStat software (San Diego, CA). A P value of <0.05 was considered significant.
The specificity of the memory CD4 T-cell response directed against the F protein of RSV is not well defined. An F-specific memory CD4 T-cell response has been reported (30), but the optimal epitope(s) has yet to be identified. A previous study using overlapping peptides scanning the entire F protein of the B strain of RSV identified a single region (F51-70) that elicited strong CD4 T-cell proliferation (10). A separate study identified two other potential CD4 T-cell epitopes by scanning the RSV A2 and B strain F protein sequences for regions corresponding to the I-Ed binding motifs [RIA]XX[IVLA][ALI]XX[KR] and [WYF(ILV)]XX[KRI]X[ILVG]XX[KR] (45). (Bold letters represent potential anchor residues.) Based on these initial studies, the RSV F51-70 (10), RSV F146-160 (45), and RSV F184-199 (45) peptides likely contain one or more potential CD4 T-cell epitopes. Therefore, we tested each of the three previously reported peptides for their ability to stimulate IFN-γ production by lung mononuclear cells harvested from vvβ-gal- and vvF-immunized mice 5 days after RSV infection. Figure Figure1A1A demonstrates that stimulation with RSV F51-70 elicited peptide-specific IFN-γ production, indicating that a memory CD4 T-cell response directed against this peptide is generated in vvF-immunized mice during challenge RSV infection. We were unable to consistently measure a RSV F51-70 peptide-specific CD4 T-cell response at this time point in mice immunized with the control recombinant vv expressing β-galactosidase (i.e., vvβ-gal) and thus undergoing a primary RSV infection (Fig. (Fig.1A).1A). In addition, in both vvβ-gal- and vvF-immunized mice undergoing RSV challenge infection, we were unable to detect IFN-γ, IL-4, IL-5, or IL-6 production in response to ex vivo peptide stimulation with either RSV F146-160 or RSV F184-199 (Fig. (Fig.1A1A and data not shown). Because neither RSV F146-160 nor RSV F184-199 stimulated detectable cytokine production above the unstimulated control, even in mice previously immunized with vvF, we chose to concentrate on the RSV F51-70 peptide-specific response. Kinetic analyses examining IFN-γ production by lung mononuclear cells isolated from vvF-immunized mice on days 3, 5, 7, and 9 after RSV challenge identified day 5 as the peak in both the frequency and the total number of IFN-γ-producing CD4 T cells after F51-70 peptide stimulation (data not shown). These data demonstrate that one or more CD4 T-cell epitope(s) with a peak response at day 5 are contained within RSV F51-70.
In order to identify the region within RSV F51-70 that contains the CD4 T-cell epitope(s), we created a series of peptides truncated at either the carboxyl or amino terminus (Fig. (Fig.1B).1B). We tested these peptides for their ability to simulate IFN-γ production by lung mononuclear cells isolated from vvF-immunized mice 5 days after RSV infection. As shown in Fig. Fig.1C,1C, neither stimulation with the RSV F55-70 peptide nor stimulation with the RSV F57-70 peptide resulted in significant levels of IFN-γ production. In contrast, both the RSV F51-68 peptide and the RSV F51-66 peptide were able to elicit a higher frequency of IFN-γ-producing CD4 T cells than was induced by ex vivo stimulation with an equal concentration of RSV F51-70. Because RSV F51-66 represented the minimal stimulatory sequence of the peptides tested, we focused on identifying the CD4 T-cell epitope(s) within this region of the F protein.
In order to determine the major histocompatibility complex (MHC) molecule(s) responsible for presenting the epitope(s) within the RSV F51-66 peptide, we stimulated lung mononuclear cells from vvF-immunized mice undergoing RSV challenge infection with F51-66 peptide in the presence of antibodies that block either I-Ad MHC recognition or both I-Ad and I-Ed MHC recognition (48). The I-Ad and I-Ed MHC blocking antibody was able to significantly (P < 0.0001) inhibit the frequency of CD4 T cells capable of making IFN-γ in response to stimulation with RSV F51-66, whereas an antibody specific for I-Ad alone did not significantly (P > 0.05) inhibit the frequency of IFN-γ-producing cells (Table (Table1)1) . These results demonstrate that the RSV F51-66 peptide is presented by I-Ed. Therefore, we scanned the RSV F51-66 peptide sequence for potential CD4 T-cell epitopes using a published motif for I-Ed anchor residues (33). The three potential nine-amino-acid epitope cores within the F51-66 peptide sequence that fit the I-Ed binding motif [WYF(ILV)]XX[KRI]X[ILVG]XX[KR] are shown in Fig. Fig.1D1D.
In order to determine how many CD4 T-cell epitopes were within RSV F51-66, we synthesized peptides with an alanine substituted in place of individual amino acids that matched the I-Ed binding motif (Fig. (Fig.2A).2A). As shown in Fig. Fig.2B,2B, substitution at each of the predicted MHC anchor residues within RSV F56-64 (amino acids V56, I59, and L61) resulted in a decreased frequency of IFN-γ-producing cells, suggesting that these amino acids are involved in the MHC binding of RSV F51-66. In contrast, substitution of the predicted MHC anchor residues for potential epitopes located at either RSV F52-60 or RSV F57-65 (amino acids W52, I57, and K65) did not result in a decrease in the frequency of IFN-γ-producing cells, suggesting that these amino acids are not essential for MHC binding and that neither of these sequences represent a CD4 T-cell epitope. Taken together, these data demonstrate that there is a single, immunodominant CD4 T-cell epitope within RSV F51-66 with a nine-amino-acid core at F56-64. However, a peptide spanning amino acids 54 to 66 did not stimulate IFN-γ production by lung mononuclear cells from vvF-immunized mice challenged with RSV (data not shown). Consistent with the finding that MHC class II-restricted epitopes can require flanking residues for optimal stimulation of CD4 T cells (6, 35), we found that RSV F51-66 stimulates a much stronger IFN-γ response than RSV F54-66 (Fig. (Fig.1C1C and data not shown).
Substitution of non-I-Ed consensus amino acids within RSV F56-64, such as T58 and E60, also resulted in a decreased frequency of IFN-γ-producing cells. These amino acids are at positions representing potential TCR contact residues, thus raising the possibility that the population of CD4 T cells recognizing the epitope might express a limited TCR repertoire similar to the highly Vβ14-skewed memory CD4 T-cell response we have previously reported for the G protein of RSV (47, 48). To determine the TCR chain usage within the F-specific memory CD4 T-cell population, we stimulated lung mononuclear cells isolated from vvF-immunized mice 5 days after challenge RSV infection with RSV F51-66 and identified the F51-66-specific memory CD4 T cells on the basis of IFN-γ production. Figure Figure3A3A demonstrates that approximately 3% of the RSV F51-66-specific cells (IFN-γ+) express the Vα2 TCR chain with <0.5% expressing the Vα8.3 TCR chain. The RSV F51-66-specific memory CD4 T cells also expressed a diverse Vβ TCR repertoire, with Vβ8.1 being the most highly represented (Fig. (Fig.3B).3B). These results demonstrate that the RSV F51-66-specific memory CD4 T cells represent a polyclonal response.
Th2 cells are known to be necessary for the development of pulmonary eosinophilia in vvG-immunized mice (49). We hypothesized that the majority of the RSV F-specific memory CD4 T cells would exhibit a Th1 phenotype because vvF-immunized mice do not develop pulmonary eosinophilia after RSV challenge (19, 29, 40). To examine the Th1/Th2 cytokine production potential of RSV F51-66-specific memory CD4 T cells, lung mononuclear cells from vvβ-gal- and vvF-immunized mice were isolated 5 days after RSV challenge and were stimulated in vitro with RSV F51-66 for 48 h. Figure Figure4A4A demonstrates that RSV F51-66-specific cells from vvF-immunized mice secrete IFN-γ and undetectable levels of IL-4, IL-5, or IL-13, a finding consistent with a predominant Th1 phenotype. Mice immunized with vvβ-gal also secrete IFN-γ, although at a decreased level compared to vvF-immunized mice, suggesting that there may be a small RSV F51-66-specific response elicited during primary RSV infection. In order to determine the effect of IFN-γ on the phenotype of the RSV F-specific memory CD4 T-cell response, lung mononuclear cells were isolated from vvβ-gal- and vvF-immunized IFN-γ-deficient mice on day 5 after challenge RSV infection and stimulated in vitro with the RSV F51-66 peptide for 48 h. In the absence of IFN-γ, RSV F51-66-specific cells produced detectable levels of IL-5 (Fig. (Fig.4B).4B). These data suggest that IFN-γ regulates the development or expansion of Th2 cells in vvF-immunized wild-type mice undergoing a challenge RSV infection.
IFN-γ is known to influence the contraction of T cells during the resolution of an immune response to an infection resulting in increased frequencies of antigen-specific memory T cells (15, 28). To determine if IL-5 production by lung mononuclear cells from IFN-γ-deficient mice was simply due to an increased number of RSV F-specific memory CD4 T cells generated after immunization with vvF, we performed ELISPOT analysis on spleen cells isolated from vvF-immunized wild-type and IFN-γ-deficient mice. An intracellular cytokine stain (ICS) of lung mononuclear cells isolated from vvF-immunized mice and stimulated with RSV F51-66 demonstrated that the frequency of TNF-α- and IFN-γ-producing RSV F-specific memory CD4 T cells was similar in wild-type mice (Fig. (Fig.4C),4C), thus allowing us to quantify antigen-specific CD4 T cells using TNF-α in the IFN-γ-deficient mice. As demonstrated in Fig. Fig.4D,4D, wild-type and IFN-γ-deficient mice generated approximately equal numbers of both TNF-α- and IL-5-producing memory cells after vvF immunization. These data suggest that IFN-γ may inhibit the expansion of Th2 cells in vvF-immunized mice undergoing challenge RSV infection.
Because Th2 cells are known to be necessary for development of pulmonary eosinophilia in vvG-immunized mice (49), we hypothesized that the increased capacity of RSV F51-66-specific CD4 T cells isolated from vvF-immunized IFN-γ-deficient mice to produce IL-5 would be sufficient for the development of pulmonary eosinophilia in vvF-immunized mice. BAL cells were collected from vvβ-gal- and vvF-immunized wild-type and IFN-γ-deficient mice 5 days after RSV challenge and analyzed for the presence of eosinophils. IFN-γ-deficient mice immunized with vvF had a significantly (P < 0.05) increased percentage and total number of eosinophils in the BAL after RSV infection compared to wild-type vvF-immunized mice (Fig. 5A and B). These data demonstrate that IFN-γ inhibits the development of pulmonary eosinophilia in vvF-immunized wild-type mice.
Due to the striking difference we detected in airspace (i.e., BAL) eosinophilia, we wanted to determine whether eosinophils were also increased in the lung parenchyma of IFN-γ-deficient mice. Whole lungs were harvested from vvβ-gal- and vvF-immunized wild-type and IFN-γ-deficient mice 7 days after RSV challenge. Eosinophils were detected in lung sections by staining with an anti-major basic protein (MBP) antibody. Slides were blinded and scored for perivascular aggregates and interstitial eosinophils. Similar to the increased eosinophilia that was observed in the BAL, the eosinophil aggregates (Fig. (Fig.5C)5C) and number of interstitial eosinophils (Fig. (Fig.5D)5D) detected were significantly (P < 0.01) increased in vvF-immunized IFN-γ-deficient mice. However, in contrast to our BAL data, we show that vvβ-gal-immunized IFN-γ-deficient mice do have significantly (P < 0.01) increased eosinophilia in the lung parenchyma compared to vvβ-gal-immunized wild-type mice. Representative sections showing anti-MBP staining for perivascular eosinophil aggregates and interstitial eosinophils from vvβ-gal- (A and E) or vvF-immunized (B and F) wild-type mice and vvβ-gal- (C and G) or vvF-immunized (D and H) IFN-γ-deficient mice are shown in Fig. Fig.6.6. These results demonstrate that IFN-γ inhibits the entrance and/or accumulation of eosinophils in the lung and BAL during secondary RSV infections of wild-type mice. However, in the case of a primary RSV infection, IFN-γ is only necessary to prevent eosinophilia within the lung parenchyma.
It has been previously shown that Th2 cells are necessary for the development of pulmonary eosinophilia in vvG-immunized BALB/c mice (49). In the present study, we demonstrate the increased production of Th2 cytokines in vvF-immunized IFN-γ-deficient mice that go on to develop pulmonary eosinophilia. To determine the mechanism of recruitment of eosinophils in IFN-γ-deficient mice, we examined the protein levels of the Th2 cell and eosinophil chemotactic factors CCL11, CCL17, and CCL22. We examined the protein levels in the lung and BAL of three chemokines known to be important in the trafficking of Th2 cells and eosinophils. IFN-γ-deficient mice immunized with vvF had significantly (P < 0.05) increased amounts of CCL11, CCL17, and CCL22 protein in the lung compared to vvF-immunized wild-type mice (Fig. (Fig.7).7). Significantly (P < 0.05) increased amounts of CCL17 and CCL22 protein were also detected in the BAL of vvF-immunized IFN-γ-deficient mice. These data suggest that in the absence of IFN-γ, increased CCL11, CCL17, and CCL22 protein levels recruit Th2 cells and eosinophils into the lung, leading to pulmonary eosinophilia.
Because of the difference in eosinophils detected in the lung parenchyma of wild-type and IFN-γ-deficient mice, we wanted to determine the impact of IFN-γ deficiency on the overall pulmonary inflammation using histology. Whole lungs from vvβ-gal- or vvF-immunized wild-type and IFN-γ-deficient mice were harvested 7 days after challenge RSV infection, embedded in paraffin, and sectioned for H&E staining. Slides were blinded and scored for inflammation in the airway (Fig. (Fig.8A),8A), alveolar (Fig. (Fig.8B),8B), peribronchiolar (Fig. (Fig.8C),8C), and perivascular (Fig. (Fig.8D)8D) regions of the lung. No significant differences in inflammation were detected between the vvF-immunized wild-type and IFN-γ-deficient mice, demonstrating that noneosinophilic pulmonary inflammation is not altered by the presence or absence of IFN-γ.
Our data demonstrate that inhibition of eosinophilia by IFN-γ occurs despite no change in pulmonary infiltrate. This suggests that IFN-γ may not affect all aspects of RSV-induced pathology. In order to determine how IFN-γ affects systemic disease induced by RSV infection, we examined weight loss and clinical illness scores. Figure 9A and B show that the absence of IFN-γ prevents vvF-immunized mice from losing weight and developing clinical signs of illness. Because of these notable differences, we sought to determine if viral clearance was also altered in IFN-γ-deficient mice. Figure Figure9C9C demonstrates that wild-type and IFN-γ-deficient mice do not have a significantly (P > 0.05) different number of copies of the nucleocapsid (N) gene in the lung at day 4 postinfection as measured by real-time PCR. These results were verified and extended with plaque assays to measure infectious virus. The plaque assay demonstrated that vvF-immunized IFN-γ-deficient and wild-type mice both cleared RSV by day 7 postinfection (Table (Table2).2). These data demonstrate that IFN-γ promotes weight loss and systemic disease but does not affect viral clearance in vvF-immunized mice undergoing challenge RSV infection.
Although RSV has been studied for over 40 years, there is still no safe and effective vaccine available. One of the roadblocks in vaccine development has been a lack of complete understanding as to why the children in the 1960s vaccine trial experienced enhanced morbidity and mortality upon subsequent RSV exposure (8, 13, 26, 27). Although the underlying mechanisms responsible for initiating RSV vaccine-enhanced disease are not known, many novel vaccination strategies targeting either the G or F protein have been analyzed in small animal models. Mice immunized with vvG develop pulmonary eosinophilia after RSV challenge similar to the vaccinated children in the 1960s (29), whereas vvF-immunized mice do not develop pulmonary eosinophilia (29, 40). Despite the difference in pulmonary eosinophilia, mice vaccinated with either the G or the F protein exhibit enhanced disease (1, 23, 47) (Fig. 9A and B). Immunization with the G protein elicits memory CD4 T cells that have been shown to play a direct role in immunopathology (1, 14, 25, 30, 47, 48, 50). Neutralizing monoclonal antibodies targeting the RSV F protein inhibit virus replication in vivo, making the F protein an attractive vaccine target (9, 21, 42, 43), but little is known about the F-specific CD4 T-cell response. Therefore, in order to gain a better understanding of the CD4 T cell-mediated immunopathology induced by immunization with the F protein, we have analyzed the F-specific CD4 T-cell response.
Our results suggest that RSV F51-66 represents the optimal CD4 T-cell epitope within this region of the F protein. An alanine scan of RSV G183-195, the only other defined CD4 T-cell epitope within the RSV genome, resulted in a decreased frequency of IFN-γ-producing CD4 T cells when nonconsensus amino acids were substituted (48). These results were shown to be due to recognition of the epitope by a CD4 T-cell population predominantly expressing the TCR chain Vβ14. We demonstrate here that CD4 T cells expressing a diverse Vβ TCR repertoire recognize RSV F51-66, suggesting that a decreased response upon substitution of potential TCR contact residues was not due to the inability of conserved Vα or Vβ TCR chains within the population of RSV F51-66-specific CD4 T cells to recognize the altered peptides (Fig. (Fig.3).3). The RSV G183-195-specific Vβ14+ CD4 T cells were also found to have a conserved CDR3 region, which accounts for the specificity of the cells, within the conserved Vβ14 chains (47). It remains possible that the RSV F51-66-specific CD4 T-cell population expresses a variety of TCR Vα and Vβ chains but that the CDR3 regions are conserved, accounting for the decreased response to substitutions at potential TCR contact residues (36). It is also possible that the RSV F51-66-specific CD4 T cells have a diversified repertoire of CDR3 regions as well as Vα and Vβ chains but that the TCRs all share the same specificity for the potential TCR contact residues at positions 3 and 5 within the nine-amino-acid epitope core (34-36).
We demonstrate that RSV F51-66-specific memory cells produce IFN-γ and no detectable IL-4, IL-5, or IL-13 after ex vivo peptide stimulation, suggesting that they are predominantly Th1 cells (Fig. (Fig.4A).4A). We have also shown that in vvF-immunized mice undergoing a challenge RSV infection in the absence of IFN-γ, the F-specific CD4 T cells produce readily detectable levels of IL-5 protein after in vitro stimulation (Fig. (Fig.4B).4B). Our ELISPOT analyses revealed a low but detectable frequency of RSV F51-66-specific IL-5-producing cells in wild-type and IFN-γ-deficient mice after immunization with vvF that was 5- to 10-fold lower than the frequency of TNF-α-producing cells (Fig. (Fig.4D).4D). These data demonstrate that IFN-γ-deficient mice exhibit no detectable lesion in the development of RSV F-specific memory cells after priming.
In support of previous findings with IFN-γ-depleted mice (19) and STAT1-deficient mice (10), we demonstrate that vvF-immunized IFN-γ-deficient BALB/c mice develop pulmonary eosinophilia upon RSV challenge (Fig. 5A and B). Upon examination of parenchymal eosinophilia in the present study using MBP-specific staining, we found that both vvβ-gal- and vvF-immunized IFN-γ-deficient mice have significantly (P < 0.01) increased perivascular aggregates of eosinophils (Fig. (Fig.5C),5C), as well as significantly (P < 0.01) increased numbers of interstitial eosinophils compared to wild-type mice (Fig. (Fig.5D).5D). These data are in contrast to a previous report examining perivascular eosinophilia in vvF-immunized, IFN-γ-deficient mice (40). The previous study found that IFN-γ deficiency did not lead to an increased number of perivascular eosinophils in vvF-immunized mice. One potential explanation for this discrepancy is that the previous study (40) utilized a Leinert-Giemsa stain and relied on cellular morphology to quantitate eosinophils, whereas we identified eosinophils based on staining for the eosinophil-specific major basic protein via immunohistochemistry. Our data demonstrate that IFN-γ inhibits eosinophils from entering the lung parenchyma and the airspace of the lung in vvF-immunized wild-type mice undergoing RSV infection. The presence of eosinophils in the lung parenchyma in vvβ-gal-immunized IFN-γ-deficient mice suggests that IFN-γ is necessary for preventing the recruitment or accumulation of eosinophils in the lung parenchyma during primary RSV infection as well. This further suggests that increased inflammation or cytokines produced by the memory T-cell response occurring within the BAL after vaccination with vvF and subsequent RSV infection of IFN-γ-deficient mice specifically allows eosinophils to move into the BAL.
We examined the amount of CCL11, CCL17, and CCL22 protein as a potential mechanism for the differential recruitment of eosinophils to the BAL and lung parenchyma in vvβ-gal- and vvF-immunized IFN-γ-deficient mice. Figure Figure77 shows that CCL11, CCL17, and CCL22 protein levels are significantly (P < 0.05) increased in the lungs of vvF-immunized IFN-γ-deficient mice but not in vvβ-gal-immunized mice. Similarly, the amount of CCL17 and CCL22 protein is significantly (P < 0.05) increased in the BAL of vvF-immunized IFN-γ-deficient mice. These data suggest that the increase in Th2 and eosinophil-specific chemokines contributes to the recruitment of eosinophils into the lungs and BAL of vvF-immunized IFN-γ-deficient mice. The mechanism of recruitment of eosinophils into the lung parenchyma of vvβ-gal-immunized IFN-γ-deficient mice is still under investigation. Taken together, our data suggest that the inhibition of pulmonary eosinophilia in vvF-immunized wild-type mice undergoing challenge RSV infection is due to the IFN-γ-mediated inhibition of Th2 cell expansion and subsequent trafficking of eosinophils and Th2 cells into the lung due to decreased chemokine levels necessary for the movement of these two cell populations.
In our studies of vaccinated IFN-γ-deficient mice challenged with RSV, we found no difference in the pulmonary inflammation in any region of the lung (Fig. (Fig.8).8). This contrasts with a previous study using a different scoring system that has shown an increase in overall lung infiltrate in IFN-γ-deficient mice undergoing a primary RSV infection (46). We also show that IFN-γ-deficient mice immunized with vvF and challenged with RSV do not lose weight or present with increased illness scores, as is observed with wild-type mice (Fig. 9A and B). Similarly, a previous study examined weight loss in vvM2-immunized IFN-γ-deficient mice and found that vaccinated mice began recovering on day 4, whereas wild-type mice continued to lose weight (31). Thus, IFN-γ plays a key role in controlling the character of the pulmonary inflammation and also exacerbates the systemic disease after RSV infection.
Consistent with previous work (10), we were unable to detect any significant (P > 0.05) difference in the viral load between wild-type and IFN-γ-deficient mice 4 days after challenge RSV infection (Table (Table22 and Fig. Fig.9C).9C). In addition, Table Table22 shows that both vvF-immunized wild-type and IFN-γ-deficient mice cleared the virus by 7 days postchallenge. At either time point, vvF-immunized mice harbored less virus than mice undergoing a primary RSV infection. Our plaque assay data support previous work with IFN-γ receptor-deficient mice on the 129SvEv background showing that viral clearance after primary infection is not dependent upon IFN-γ signaling (5, 24). These data indicate that IFN-γ deficiency does not alter viral clearance and suggest that viral load would be reduced upon inclusion of the F protein in future RSV vaccines.
It is interesting that the G protein of RSV induces a strong IFN-γ response but that this is not sufficient to inhibit the pulmonary eosinophilia induced by RSV challenge of vvG-immunized mice. In fact, the RSV G-specific IFN-γ response is ~10-fold greater than the RSV F-specific IFN-γ response (41, 47, 48) (Fig. (Fig.1A).1A). It is possible that the source of inhibitory IFN-γ is not present in vvG-immunized mice. One potential source of IFN-γ could be CD8 T cells as an F-specific (7, 20), but not a G-specific CD8 T-cell response has been described (32). In addition, the G protein of RSV is expressed in both membrane-bound and secreted forms, whereas the F protein is only membrane bound (17). It has been previously shown that vaccination with the secreted form of the G protein results in increased pulmonary eosinophilia compared to vaccination with either the wild-type or membrane form of the protein (3, 22, 23). However, the lack of a secreted form of the F protein is an unlikely explanation for the difference in eosinophilia observed after challenge RSV infection of vaccinated mice since it has been shown that vaccination with a secreted form of the F protein did not result in significantly increased numbers of eosinophils (4). The difference between F- and G-protein vaccination may also lie in the oligoclonal nature of the G-specific memory response, which might allow the RSV G-specific Th2 cells to evade regulation by IFN-γ, a situation that does not occur in the polyclonal F-specific memory response that we have described here. Together, these data suggest that Th2 cells are necessary for the development of pulmonary eosinophilia in vaccinated mice undergoing a challenge RSV infection. In addition, our data support a model wherein the production of IFN-γ inhibits the expansion of Th2 cells and subsequent production of chemokines after RSV challenge and thus inhibits the recruitment of eosinophils into the lung.
We thank John Harty and Kevin Legge for critically reviewing the manuscript. We thank Elizabeth Field and Todd Rouse for assistance with the use of an ImmunoSpot reader and ImmunoSpot software, as well as Martin Stoltzfus and Colin Exline for assistance with the use of an ABI 7300 Real Time PCR System. We thank Stacey Hartwig for excellent technical assistance.
This study was supported by American Heart Association Pre- Doctoral Fellowship 0615625Z (E.M.C.), NIH grant AI 063520 (S.M.V.), and the University of Iowa Department of Pathology (D.K.M.).
Published ahead of print on 19 December 2007.