0014-2980/02/0808-2117$17.50+.50/0© WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2002

Virus clearance and immunopathology by CD8+

T cells during infection with respiratory syncytialvirus are mediated by IFN- q

Tobias Ostler1, Wendy Davidson2 and Stephan Ehl1

1 Children’s Hospital, University of Freiburg, Freiburg, Germany2 Department of Immunology, American Red Cross, Rockville, USA

CD8+ T cells (CTL) are important effector cells for virus control and immunopathology afterprimary infection with respiratory syncytial virus (RSV). To investigate the effector mecha-nisms involved, we set up an adoptive transfer model, in which effector CTL specific forp82–90 of RSV M2 were generated in vivo, followed by short-term restimulation in vitro andtransfusion into infected recipients. A total of 4×104 donor-derived p82-specific CTL homingto the lung within 4 days after transfusion were sufficient to completely eliminate a virusinoculum of 1.5×106 pfu. This was accompanied by significant lung pathology. Surprisingly,virus control and immunopathology proceeded unimpaired when donor cells lacking perfo-rin, CD95 ligand or TNF were transfused. By contrast, treatment of recipient mice with a neu-tralizing antibody against IFN- + or transfusion of IFN- + -deficient effector CTL largelyabolished virus control and significantly reduced CD8+ T cell-mediated pathology. In IFN- + -deficient mice, high-dose primary infection experiments revealed attenuated immunopathol-ogy, but only slightly delayed virus clearance, suggesting that other cells and molecules canpartly substitute for the effects of CTL-derived IFN- + on virus clearance. These experimentsidentify IFN- + as a key molecule in RSV-induced immunopathology and in CD8+ T cell-mediated control of RSV infection.

Key words: CD8+ T cell / Virus / Cytotoxicity / Lung / Immunopathology

Received 8/4/02Accepted 17/5/02

[I 23112]

Abbreviations: RSV: Respiratory syncytial virus rVV:Recombinant vaccinia virus BAL: Bronchoalveolar lavageNP: Nucleoprotein p82: RSV matrixprotein (M2) peptide82–91 p118: LCMV NP peptide 118–127 p82 tet: MHCclass I H2-Kd tetramers loaded with p82

1 Introduction

Antiviral T cells are crucial for elimination of respiratorysyncytial virus (RSV) both in humans and in experimentalanimal models. Infants with congenital T cell deficiencieshave difficulties eliminating the virus [1, 2] and depletionof T cells leads to persistent infection in BALB/c mice [3].Adoptive transfer experiments with T cell clones haveshown that T cells are not only necessary, but also suffi-cient to clear RSV from infected mice [4]. The extent ofRSV-induced airway disease, however, is only partlydetermined by the kinetics of virus elimination and viralcytopathogenicity. Many observations in infected infantsand animal models strongly suggest that disease isenhanced by the immune response to the viral infection

[3–5]. Again, at least in the mouse model, antiviral T cellsplay a key role. While T cell depletion diminishes RSVinduced immunopathology [3], transfusion of T cellclones into RSV-infected mice aggravates disease,sometimes with lethal outcome [4]. Both CD4+ and CD8+

T cells can eliminate virus and cause immunopathologyindependently, but CD8+ T cells appear to be more effec-tive [3].

The molecular mechanisms involved in CD8+ T cell-mediated virus control and immunopathology remainpoorly defined. RSV was eliminated with unchangedkinetics from perforin-deficient mice [6]. However, sinceboth RSV clearance and immunopathology also occur inCD8-deficient mice [3], the question whether the contri-bution of CD8 cells is mediated by perforin could not beproperly answered in that study. A potential role forCD95L has not yet been examined. In some studies,neutralization of IFN- + or TNF reduced immunopathol-ogy without affecting virus clearance after secondaryRSV infection of mice that had been primed with an RSV-derived CTL epitope peptide in adjuvant [7] or with arecombinant vaccinia virus [8]. However, the implicationsof these findings for primary RSV infection remain un-

Eur. J. Immunol. 2002. 32: 2117–2123 CTL control RSV via IFN- + 2117

Fig. 1. Transfused RSV-specific donor CTL home to the infected recipient lung without significant cell division. BALB/c Thy 1.1mice were infected i.v. with rVV RSV M2 or rVV LCMV NP. Two weeks later, spleen cells were restimulated on BALB/c Thy 1.2stimulator cells pulsed with M2 p82 or NP p118, respectively. After 5 days, the culture was analyzed by flow cytometry for thepresence of CD8+ p82 tet+ T cells (A, C) and for cytotoxicity on target cells labeled with p82 or p118 (B, D). Four days after trans-fusion of 2×107 cultured CFSE-labeled cells into RSV-infected BALB/c Thy 1.2 recipients, lung cells were isolated and analyzedfor the presence of donor CD8+ p82 tet+ T cells (E, F) and the CFSE staining of donor CD8+ T cells (G). Results are representativeof two independent experiments using four mice per group.

clear. Indeed, several other studies have reported oppo-site results [9–11].

Since several cell types contribute to virus control andimmunopathology, the molecular mechanisms involvedmay be redundant and difficult to identify in these experi-mental set-ups. We therefore chose a more focussedapproach using an adoptive transfer model, whichallowed us to exclusively study the effector mechanismsby which CD8+ T cells mediate RSV elimination andimmunopathology. We identified IFN- + as a key moleculeinvolved in both processes, while TNF, perforin or CD95Lseemed to play no role.

2 Results

2.1 Transfused RSV-specific CTL home to theinfected lung without significant cell division

To study CTL-mediated virus clearance and immunopa-thology in RSV infection, we set up an adoptive transfermodel that allowed us to selectively investigate the con-tribution of RSV-specific CD8+ T cells in a well-controlledfashion (Fig. 1). Thy1.1 donor mice were primed withrecombinant vaccinia virus (rVV) RSV M2 or rVV LCMVNP as a control. Two weeks later, splenocytes were resti-

mulated with the RSV-M2-derived CTL epitope peptidep82, or with the LCMV-NP-derived peptide p118. After5 days, cultured cells were transfused into Thy 1.2 micethat had been infected with 1.5×106 pfu RSV 1 h beforetransfer. Mice were monitored for weight loss and 4 dayslater, lungs were harvested for virus titer determination orhistological analysis.

First, we studied the fate of RSV-specific donor cellsafter transfusion into infected Thy 1-congenic recipients.On day 5 of the restimulation culture, 21% of CD8+ Tcells, i.e. about 5% of all cells, bound H-2 Kd tetramersloaded with p82 (p82 tet) (Fig. 1A). The cells showed highRSV-specific CTL activity in vitro (Fig. 1B). Control cellsobtained with the LCMV protocol did not contain p82-specific CD8+ T cells and only showed LCMV-specificcytotoxicity (Fig. 1C, D). Before transfusion, cells werelabeled with the fluorescent dye CFSE, which allowstracing cell division in vivo [12]. In a representativemouse, 4 days after transfusion of 2×107 labeled cul-tured cells containing 1.4×106 p82tet+ cells, we isolated2.6×105 donor CD8+ T cells from the lung. Of these,4.2×104 (i.e. 15%) bound the p82 tetramer (Fig. 1E). Allstained brightly for CFSE, suggesting that they had notundergone cell division (Fig. 1G). Less than 300 p82 tet+

T cells of Thy 1.2 (recipient) origin were detected at thistime point. Donor CD8+ T cells were also detected in

2118 T. Ostler et al. Eur. J. Immunol. 2002. 32: 2117–2123

Fig. 2. Virus control and pulmonary immunopathology medi-ated by transfused RSV-specific CTL. BALB/c mice wereinfected with RSV and transfused with the indicated numberof restimulated spleen cells from mice primed with rVV RSVM2 or rVV LCMV NP. Four days after transfer, lungs of recipi-ent mice were taken for determination of RSV titers (A) or forhistological analysis (C, D). Representative 10× magnifiedH&E sections of peri-bronchiolar areas are shown. Micewere monitored daily for weight loss (B). The experimentwas performed three times with similar results.

Fig. 3. RSV-specific CTL mediate virus control and immuno-pathology independent of perforin, CD95L and TNF. Thebasic experimental set-up is shown in Fig. 1. (A, D) Cytotox-icity of restimulated spleen cells from rVV RSV M2-primedperforin-deficient (PO), gld/gld and BALB/c control mice ontarget cells labeled with p82. (B, E) RSV titers in recipientlungs 4 days after transfusion of 2×107 cultured spleen cellsfrom PO, gld/gld and BALB/c control mice. (G) BALB/c micewere treated with anti-TNF or an isotype control antibodybefore injection of d-galactosamine and LPS and monitoredfor survival. (H) Mice were treated with anti-TNF (open sym-bols) or an isotype control antibody (closed symbols) beforetransfusion of 2×107 cultured spleen cells from BALB/cmice. Four days later, lung virus titers were determined. Allmice were monitored for weight loss during the experiments(C, F, I). All experiments were repeated at least twice withsimilar results.

lungs of RSV-infected animals after transfusion of LCMV-NP-specific CTL (Fig. 1F), but at significantly lower num-bers.

2.2 Virus control and pulmonaryimmunopathology mediated by transfusedRSV-specific CTL

Transfusion of 2×107 rVV RSV M2-primed, p82-restimulated spleen cells completely eliminated RSVfrom the lung within 4 days, while 2×107 rVV LCMV-NP-primed cells had no effect on viral titers (Fig. 2A). Whenlungs of recipient mice were analyzed histologically,abundant peri-bronchiolar and intra-alveolar infiltratescould be observed in mice receiving RSV-specific cells(Fig. 2D), while these changes were less pronounced inmice receiving LCMV-specific cells (Fig. 2C). This immu-nopathology manifested clinically as reduced activity,ruffled fur and hunched posture and could be objectively

monitored by weight loss. Transfer of RSV M2-specificcells caused clinical disease and weight loss in a dose-dependent fashion, while there were hardly any signs ofillness and only transient weight loss in animals trans-fused with NP-specific cells (Fig. 2B). Optimal restimula-tion conditions and exact dosing of virus were instru-mental in these experiments. An intermittent change invirus stocks necessitated careful retitration of virus doseand cell numbers to avoid lethality and achieve full viruscontrol in this setting.

2.3 RSV-specific CTL mediate virus control andimmunopathology independent of perforin,CD95L and TNF

To characterize the effector mechanisms used by CD8+

T cells, we first transfused cells from perforin-deficient(PO) mice. Although RSV-specific CTL from PO miceshowed little cytotoxicity in vitro (Fig. 3A), they elimi-

Eur. J. Immunol. 2002. 32: 2117–2123 CTL control RSV via IFN- + 2119

Fig. 4. CD8+ T cell-mediated virus control and immunopa-thology is dependent on IFN- + . The basic experimental set-up is shown in Fig. 1. (A) Cytotoxicity of restimulated spleencells from rVV RSV M2-primed IFN- + 0/0 and BALB/c controlmice on target cells labeled with p82. (B, C) RSV titers andweight course in recipient lungs 4 days after transfusion of4×107 cultured spleen cells from IFN- + 0/0 or 1×107 spleencells from BALB/c control mice. (D, E) RSV titers and weightcourse in recipient lungs after transfusion of 2×107 BALB/cspleen cells into mice injected i.p with an anti-IFN- + or anisotype control antibody 1 day before transfer. In (E), anadditional control group received 2×107 restimulated spleencells from rVV LCMV NP-primed mice. Experiments wererepeated twice with similar results.

Fig. 5. Other antiviral effector mechanisms can partiallycompensate for the absence of CTL-derived IFN- + after pri-mary RSV infection. IFN- + 0/0 and BALB/c control mice wereinfected with 2.5×106 pfu RSV. (A) After 6 and 8 days, RSVtiters were determined in the lung. (B) Mice were monitoredfor weight loss daily after infection.

nated virus as well as cells from control mice (Fig. 3B).Weight loss was even slightly enhanced after transfusionof PO cells (Fig. 3C). Similar results were obtained usingRSV-specific CTL generated from gld/gld mice whichcarry a loss-of-function mutation in the CD95L gene(Fig. 3D–F). To investigate the contribution of TNF, wetreated the recipient mice with a monoclonal anti-TNFantibody. This antibody prevented LPS-induced, TNF-mediated lethal septic shock in mice sensitized withD-galactosamine (Fig. 3G). However, TNF neutralizationhad no impact on virus control or weight loss induced byRSV-specific CTL in RSV-infected recipients (Fig. 3H, I).

2.4 CD8+ T cell-mediated virus control andimmunopathology are dependent on IFN- +

The role of IFN- + was investigated using RSV-specificCTL from IFN- + 0/0 mice. After rVV M2 priming and in vitrorestimulation, cells from these mice showed roughlythreefold less CTL activity than control cells (Fig. 4A),and threefold less p82tet+ CD8+ T cells could be identi-fied (not shown). To transfuse the same number of p82tet+ T cells, we therefore adapted the cell numbersaccording to the flow cytometry data and transfusedthreefold more IFN- + 0/0 than BALB/c cells. Nevertheless,CD8+ T cell-mediated virus control was completely abol-

ished in the absence of IFN- + (Fig. 4B). Both groups ofmice suffered significant weight loss during the first2 days after RSV infection and cell transfusion, but recip-ients of cells from IFN- + 0/0 mice recovered by day 4(Fig. 4C). When we analyzed the lungs for the number ofp82 tet+ donor T cells 4 days after transfer, we foundsimilar numbers in recipients of BALB/c and IFN- + 0/0

cells, indicating that the observed differences were notdue to a different homing behavior of IFN- + 0/0 cells. Toconfirm these findings, we transfused RSV-infected micewith activated BALB/c cells 1 day after i.p. injection of aneutralizing anti-IFN- + or an isotype control antibody. Invivo blockade of IFN- + partially reduced CTL-mediatedvirus control (Fig. 4D) and enhanced recovery from CTL-mediated weight loss (Fig. 4E).

2.5 Other antiviral effector mechanisms canpartially compensate for the absence ofIFN- q after primary RSV infection

While the adoptive transfer system is well suited to focuson the effects of CD8+ T cells, it may only give a partialanswer about the complex cellular and molecular mech-anisms of virus control and immunopathology during pri-mary RSV infection. When IFN- + 0/0 and BALB/c controlmice were infected with a high dose (2.5×106 pfu) of RSV,virus titers decreased significantly more slowly in IFN- + 0/0

mice, but RSV was eventually eliminated in both strainsby day 8 after infection (Fig. 5A). The overall effect onweight loss was more pronounced, since weight lossand clinical signs of illness were almost absent in IFN- + 0/0

mice after high-dose RSV infection (Fig. 5B).

2120 T. Ostler et al. Eur. J. Immunol. 2002. 32: 2117–2123

3 Discussion

CD8+ T cells play a key role in both virus elimination andinduction of immunopathology after RSV infection. Here,we used an adoptive transfer model to characterize theeffector mechanisms used by CTL to mediate theseeffects. In this model, 4×104 virus-specific CTL homingto the lung within 4 days after infection were sufficient tomediate elimination of a virus inoculum of 1.5×106 pfu.Entry to the infected lung was not restricted to virus-specific cells. Activated T cells from control cultures spe-cific for LCMV also homed to the lung in significant num-bers, but had no measurable effect. The biologicaleffects of the transfused cells were mediated by the CD8T cell population, as both virus control and immunopa-thology could also be achieved after transfer of MACS-purified CD8 cells (data not shown). The adoptive trans-fer system thus allowed a direct comparison of variousCD8 T cell-dependent effector mechanisms in the sameexperimental model.

Contact-dependent perforin-mediated cytolysis, al-though necessary for RSV-specific cytotoxicity in vitro,was not required for virus control or immunopathology invivo. A previous study found that RSV was eliminatedfrom perforin-deficient mice with kinetics similar to thoseseen for control mice [6]. However, RSV elimination alsooccurs in the absence of CD8 cells [3] and the questionwhether the contribution of CD8 cells is mediated by per-forin is difficult to answer in an experimental situation,where CD8 cells are dispensable. Functional expressionof CD95L on RSV-specific pulmonary CTL has beendemonstrated previously [6], but whether this is requiredfor virus clearance was not examined. Our study is,therefore, the first to show that neither perforin norCD95L are needed for CD8 T cell-mediated virus clear-ance and immunopathology in RSV infection. This is sur-prising because both of these molecules are importantfor the clearance of other respiratory viruses, i.e. influ-enza [13].

Neutralization of TNF did not affect the efficacy of RSVclearance in our transfer model, extending previousobservations [10]. By contrast, an important role for TNFhas been postulated in immunopathology induced byrespiratory viruses [8]. However, these findings werebased on experiments with secondary RSV infection andprimary influenza infection. Another report could notconfirm these observations for primary RSV infection[10]. In our experimental set-up that focussed on theeffects of CTL, we could clearly show that anti-TNFtreatment is not sufficient to reduce CTL-mediatedimmunopathology during primary RSV infection.

We identified IFN- + as the key molecule used by CD8+ Tcells for elimination of RSV infection. CTL-mediated viruscontrol and induction of immunopathology was tightlylinked, and depended on the same effector mechanism.Previous studies investigating RSV elimination andimmunopathology after primary and secondary RSVinfection of IFN0/0 or anti-IFN- + -treated mice have pro-duced conflicting results. Some studies have showna slight enhancement of RSV replication and somedecrease in cell recruitment and airway disease [11, 14],similar to the observations during primary infection in thepresent experiments. Other studies could not documentany significant differences [9, 15]. By contrast, a clearrole for IFN- + in RSV clearance was suggested in experi-ments using a different approach. Thus, recombinantRSV expressing IFN- + was significantly attenuated in therespiratory tract of mice compared to recombinant RSVexpressing a control protein [16]. These conflictingresults indicate that there is significant redundancy inimmune cells [3] and their effector mechanisms involvedin RSV elimination. In this study we exclusively concen-trated on the contribution of CD8 T cells using an adop-tive transfer model. In this setting, direct comparison ofthe different CTL-mediated effector mechanisms waspossible and allowed us to document that IFN- + isindeed the limiting effector molecule used by CD8+ Tcells in RSV infection. Again, this is in interesting contrastto the influenza model, where virus elimination andimmunopathology during primary infection and in adop-tive transfer experiments were not changed in theabsence of IFN- + [17, 18]. Non-cytolytic control of viralinfections by CTL via IFN- + has also been described inmouse models of hepatitis B [19, 20] and Sindbis virusinfection [21]. In this context our data add to evidencethat the direct link between CTL-mediated virus controland immunopathology can be of similar significancewhen non-cytolytic effector mechanisms are involved[22].

How then does IFN- + clear virus and cause disease?IFN- + is an important inducer of chemokines. An inter-esting study has presented evidence that, after contactwith CTL, pulmonary alveolar cells may not be lysed, butactivated to produce significant amounts of chemokines,which may then initiate secondary processes ultimatelyleading to tissue injury [23]. In fact, blockade of the che-mokines IFN- + -inducible protein (IP-10) and monokineinduced by IFN- + (MIG) has been shown to reduce thepathogenic potential of CTL in a mouse model of hepati-tis B infection [24], where IFN- + has also been shown tobe the key effector mechanism used by CTL both forvirus elimination and immunopathology. In addition,IFN- + -induced production of NO synthase (NOS) is anobvious mechanism that could be involved in both pro-cesses [25, 26]. Both NOS and chemokines can also be

Eur. J. Immunol. 2002. 32: 2117–2123 CTL control RSV via IFN- + 2121

induced in the absence of IFN- + [27]. This may explainwhy IFN- + is less important during primary infection thanin the adoptive transfer experiments focussed on thecontribution of CTL. Nevertheless, the clear phenotypesobserved in this model should help to further character-ize mechanisms of virus control and immunopathology inRSV infection. This may be relevant for the developmentof immunotherapeutic approaches for this sometimeslethal respiratory infection.

4 Materials and methods

4.1 Mice

Specific pathogen-free BALB/c mice were obtained fromCharles River (Sulzfeld, Germany), and used at 6–10 weeksof age. BALB/c backcrosses for perforin-deficient mice[28] (6 generations), gld/gld mice (15 generations) IFN- + -deficient mice [29] (9 generations) and Thy 1.1 congenicmice (10 generations) were kindly provided by Barney Gra-ham (Vanderbilt University, Nashville, TN), Wendy Davidson(American Red Cross, Rockville, MD), Uwe Klemm (MPIfor Infection Biology, Berlin, Germany) and Charlie Surh(Scripps Institute, La Jolla, CA), respectively. Mice were keptin a venti-rack.

4.2 Viruses

The RSV A2 strain was grown on HEp-2 cells and kept in liq-uid nitrogen. BALB/c mice were infected intranasally undermethoxyflurane anesthesia with 1.5×106 pfu in 80 ? l RPMI.rVV expressing the RSV M2 protein or the nucleoproteinof lymphocytic choriomeningitis virus (LCMV NP) weregrown on BSC-40 cells. Mice were infected intravenouslywith 2×106 pfu. Lung RSV titers were determined as de-scribed [4].

4.3 Cytotoxicity assay and cell transfer

Single-cell spleen suspensions were prepared from rVV-infected mice and restimulated on stimulator spleen cellslabeled with the RSV M2 peptide p82–90 [30] or the LCMVNP peptide p118–126 [31] in the presence of 2% rat Con Asupernatant. On day 5 the culture was harvested, washedand some cells were tested in a 5-h 51Cr-release assay onP815 target cells labeled with p82 or with np118 (NeosystemLaboratoire, Strasbourg, France). For transfer studies, thecultured cells were resuspended in a volume of 500 ? l PBSand injected intravenously into syngeneic mice within 1 h ofRSV infection. For some experiments cells were labeled withthe fluorescent dye CFSE (Molecular Probes, Eugene, OR).

4.4 Antibody treatment

Mice were depleted of cytokines by a single i.p. injection of500 ? g anti-IFN- + (clone R4–6A2, ATCC, Manassas, VA),500 ? g rat/murine chimeric anti-TNF (cV1q) or 500 ? g of anisotype-matched anti-human CD4 (M-T412) control anti-body (both kindly provided by David Shealy, Centocor, Mal-vern, PA) 1 day before infection with RSV and cell trans-fer. The efficacy of TNF blockade was tested in a septicshock model, where mice were injected i.v. with 20 mgD-galactosamine and 0.1 ? g LPS in a volume of 200 ? lpyrogen-free PBS and monitored for survival.

4.5 Flow cytometry

For isolation of lung cells, lungs were prepared from exsan-guinated mice and strained in a 100- ? m cell strainer (BectonDickinson, product number 2360) with a syringe plunger.Surface stainings were performed using anti-CD8a (clone53–6.7) (PharMingen, San Diego, CA). H-2Kd tetramersloaded with RSV p82–90 were provided by the NIAID Tetra-mer Facility. Tetramer staining was performed prior to CD8staining by incubation for 30 min at room temperature. Cellswere analyzed on a FACScan cytometer using CellQuestsoftware. Calculations of absolute cell numbers were basedon live cells as determined by FSC/SSC analysis.

4.6 Histology

Lung sections stained with hematoxylin and eosin were pre-pared after dissection of lung tissue from animals that hadbeen instilled with 1 ml formalin intratracheally. Sectionswere scored in a blinded fashion by a single experiencedobserver.

Acknowledgements: We thank the NIAID Tetramer Facilityfor generously providing H-2Kd p82 tetramers and M. Freu-denberg (MPI for Immunobiology, Freiburg) for providingLPS and advice with the septic shock model. We thank B.Graham, U. Klemm and C. Surh for providing backcrossedknockout, mutant and congenic mouse strains. We aregrateful to M. Brandis (Children’s Hospital, Freiburg) and H.Pircher (Department of Immunology, Institute for MedicalMicrobiology and Hygiene, Freiburg) for continuous supportand discussion, N. Wehrle for excellent technical assistanceand D. Neumann-Haefelin and O. Haller (Deparment ofVirology, Freiburg) for critical reading of the manuscript.This work was supported by the Deutsche Forschungsge-meinschaft (EH 145–2/2) and the Landesförderungs-Schwerpunktprogramm Baden-Württemberg (S. Ehl).

2122 T. Ostler et al. Eur. J. Immunol. 2002. 32: 2117–2123

References

1 Fishaut, M., Tubergen, D. and McIntosh, K., Cellular responseto respiratory viruses with particular reference to children withdisorders of cell-mediated immunity. J. Pediatr. 1980. 96:179–186.

2 Hall, C. B., Powell, K. R., MacDonald, N. E., Gala, C. L.,Menegus, M. E., Suffin, S. C. and Cohen, H. J., Respiratorysyncytial viral infection in children with compromised immunefunction. N. Engl. J. Med. 1986. 315: 77–81.

3 Graham, B. S., Bunton, L. A., Wright, P. F. and Karzon, D. T.,Role of T lymphocyte subsets in the pathogenesis of primaryinfection and rechallenge with respiratory syncytial virus in mice.J. Clin. Invest. 1991. 88: 1026–1033.

4 Cannon, M. J., Openshaw, P. J. and Askonas, B. A., CytotoxicT cells clear virus but augment lung pathology in mice infectedwith respiratory syncytial virus. J. Exp. Med. 1988. 168:1163–1168.

5 Openshaw, P. J., Immunity and immunopathology to respiratorysyncytial virus. The mouse model. Am. J. Respir. Crit. Care Med.1995. 152: S59–S62.

6 Aung, S., Rutigliano, J. A. and Graham, B. S., Alternative mech-anisms of respiratory syncytial virus clearance in perforin knock-out mice lead to enhanced disease. J. Virol. 2001. 75:9918–9924.

7 Simmons, C. P., Hussell, T., Sparer, T., Walzl, G., Openshaw, P.and Dougan, G., Mucosal delivery of a respiratory syncytial virusCTL peptide with enterotoxin-based adjuvants elicits protective,immunopathogenic, and immunoregulatory antiviral CD8+ T cellresponses. J. Immunol. 2001. 166: 1106–1113.

8 Hussell, T., Pennycook, A. and Openshaw, P. J., Inhibition oftumor necrosis factor reduces the severity of virus- specific lungimmunopathology. Eur. J. Immunol. 2001. 31: 2566–2573.

9 Durbin, J. E., Johnson, T. R., Durbin, R. K., Mertz, S. E.,Morotti, R. A., Peebles, R. S. and Graham, B. S., The role of IFNin respiratory syncytial virus pathogenesis. J. Immunol. 2002.168: 2944–2952.

10 Neuzil, K. M., Tang, Y. W. and Graham, B. S., Protective role ofTNF-alpha in respiratory syncytial virus infection in vitro and invivo. Am. J. Med. Sci. 1996. 311: 201–204.

11 van Schaik, S. M., Obot, N., Enhorning, G., Hintz, K., Gross,K., Hancock, G. E., Stack, A. M. and Welliver, R. C., Role ofinterferon gamma in the pathogenesis of primary respiratory syn-cytial virus infection in BALB/c mice. J. Med. Virol. 2000. 62:257–266.

12 Weston, S. A. and Parish, C. R., New fluorescent dyes for lym-phocyte migration studies. Analysis by flow cytometry and fluo-rescence microscopy. J. Immunol. Methods 1990. 133: 87–97.

13 Topham, D. J., Tripp, R. A. and Doherty, P. C., CD8+ T cells clearinfluenza virus by perforin or Fas-dependent processes.J. Immunol. 1997. 159: 5197–5200.

14 Hussell, T., Baldwin, C. J., O’Garra, A. and Openshaw, P. J.,CD8+ T cells control Th2-driven pathology during pulmonaryrespiratory syncytial virus infection. Eur. J. Immunol. 1997. 27:3341–3349.

15 Srikiatkhachorn, A. and Braciale, T. J., Virus-specific memoryand effector T lymphocytes exhibit different cytokine responsesto antigens during experimental murine respiratory syncytial virusinfection. J. Virol. 1997. 71: 678–685.

16 Bukreyev, A., Whitehead, S. S., Bukreyeva, N., Murphy, B. R.and Collins, P. L., Interferon gamma expressed by a recombinantrespiratory syncytial virus attenuates virus replication in micewithout compromising immunogenicity. Proc. Natl. Acad. Sci.USA 1999. 96: 2367–2372.

17 Wiley, J. A., Cerwenka, A., Harkema, J. R., Dutton, R. W. andHarmsen, A. G., Production of interferon-gamma by influenzahemagglutinin-specific CD8 effector T cells influences the devel-opment of pulmonary immunopathology. Am. J. Pathol. 2001.158: 119–130.

18 Graham, M. B., Dalton, D. K., Giltinan, D., Braciale, V. L., Ste-wart, T. A. and Braciale, T. J., Response to influenza infection inmice with a targeted disruption in the interferon gamma gene.J. Exp. Med. 1993. 178: 1725–1732.

19 McClary, H., Koch, R., Chisari, F. V. and Guidotti, L. G., Relativesensitivity of hepatitis B virus and other hepatotropic viruses tothe antiviral effects of cytokines. J. Virol. 2000. 74: 2255–2264.

20 Guidotti, L. G., Ishikawa, T., Hobbs, M. V., Matzke, B., Schrei-ber, R. and Chisari, F. V., Intracellular inactivation of the hepatitisB virus by cytotoxic T lymphocytes. Immunity 1996. 4: 25–36.

21 Binder, G. K. and Griffin, D. E., Interferon-gamma-mediatedsite-specific clearance of alphavirus from CNS neurons. Science2001. 293: 303–306.

22 Zinkernagel, R. M., Immunology taught by viruses. Science1996. 271: 173–178.

23 Small, B. A., Dressel, S. A., Lawrence, C. W., Drake, D. R., 3rd,Stoler, M. H., Enelow, R. I. and Braciale, T. J., CD8(+) T cell-mediated injury in vivo progresses in the absence of effector Tcells. J. Exp. Med. 2001. 194: 1835–1846.

24 Kakimi, K., Lane, T. E., Wieland, S., Asensio, V. C., Campbell, I.L., Chisari, F. V. and Guidotti, L. G., Blocking chemokine respon-sive to gamma-2/interferon (IFN)-gamma inducible protein andmonokine induced by IFN-gamma activity in vivo reduces the path-ogenetic but not the antiviral potential of hepatitis B virus-specificcytotoxic T lymphocytes. J. Exp. Med. 2001. 194: 1755–1766.

25 Bogdan, C., Nitric oxide and the immune response. Nat. Immu-nol. 2001. 2: 907–916.

26 Karupiah, G., Xie, Q. W., Buller, R. M., Nathan, C., Duarte, C.and MacMicking, J. D., Inhibition of viral replication byinterferon-gamma-induced nitric oxide synthase. Science 1993.261: 1445–1448.

27 Mahalingam, S., Chaudhri, G., Tan, C. L., John, A., Foster, P. S.and Karupiah, G., Transcription of the interferon gamma (IFN-gamma)-inducible chemokine MIG in IFN-gamma-deficient mice.J. Biol. Chem. 2001. 276: 7568–7574.

28 Walsh, C. M., Matloubian, M., Liu, C. C., Ueda, R., Kurahara,C. G., Christensen, J. L., Huang, M. T., Young, J. D., Ahmed, R.and Clark, W. R., Immune function in mice lacking the perforingene. Proc. Natl. Acad. Sci. USA 1994. 91: 10854–10858.

29 Dalton, D. K., Pitts-Meek, S., Keshav, S., Figari, I. S., Bradley,A. and Stewart, T. A., Multiple defects of immune cell function inmice with disrupted interferon-gamma genes. Science 1993.259: 1739–1742.

30 Kulkarni, A. B., Collins, P. L., Bacik, I., Yewdell, J. W., Bennink,J. R., Crowe, J. E., Jr. and Murphy, B. R., Cytotoxic T cells spe-cific for a single peptide on the M2 protein of respiratory syncytialvirus are the sole mediators of resistance induced by immuniza-tion with M2 encoded by a recombinant vaccinia virus. J. Virol.1995. 69: 1261–1264.

31 Schulz, M., Aichele, P., Vollenweider, M., Bobe, F. W., Cardi-naux, F., Hengartner, H. and Zinkernagel, R. M., Major histo-compatibility complex-dependent T cell epitopes of lymphocyticchoriomeningitis virus nucleoprotein and their protective capacityagainst viral disease. Eur. J. Immunol. 1989. 19: 1657–1667.

Correspondence: Stephan Ehl, Children‘s Hospital, Univer-sity of Freiburg, Mathildenstr. 1, D-79106 Freiburg, GermanyFax: +49-761-270-4481e-mail: ehl — kikli.ukl.uni-freiburg.de

Eur. J. Immunol. 2002. 32: 2117–2123 CTL control RSV via IFN- + 2123