cyclosporine a drives a th17- and th2-mediated posttransplant obliterative airway disease

American Journal of Transplantation 2013; 13: 611–620Wiley Periodicals Inc.

C© Copyright 2013 The American Society of Transplantationand the American Society of Transplant Surgeons

doi: 10.1111/ajt.12067

Cyclosporine A Drives a Th17- and Th2-MediatedPosttransplant Obliterative Airway Disease

P. H. Lemaıtrea,∗, B. Vokaera, L.-M. Charbonniera,

Y. Iwakurab, K. A. Fieldc, M. Estenned,

M. Goldmane, O. Leoa, M. Remmelinkd

and A. Le Moinea,d

aInstitute for Medical Immunology, Universite Libre deBruxelles, Gosselies, BelgiumbCenter of Experimental Medicine and Systems Biology,Institute of Medical Science, University of Tokyo, Tokyo,JapancBucknell University, Lewisburg, PAdErasme Hospital, Universite Libre de Bruxelles, Brussels,BelgiumeInnovative Medicines Initiative, Brussels, Belgium∗Corresponding author: Philippe H. Lemaıtre,[email protected]

Calcineurin-inhibitor refractory bronchiolitis obliterans(BO) represents the leading cause of late graft failureafter lung transplantation. T helper (Th)2 and Th17lymphocytes have been associated with BO devel-opment. Taking advantage of a fully allogeneic tra-chea transplantation model in mice, we addressed thepathogenicity of Th cells in obliterative airway disease(OAD) occurring in cyclosporine A (CsA)-treated recip-ients. We found that CsA prevented CD8+ T cell infil-tration into the graft and downregulated the Th1 re-sponse but affected neither Th2 nor Th17 responsesin vivo. In secondary mixed lymphocyte cultures, CsAdramatically decreased donor-specific IFN-c produc-tion, enhanced IL-17 production and did not affect IL-13. As CD4+ depletion efficiently prevented OAD inCsA-treated recipients, we further explored the roleof Th2 and Th17 immunity in vivo. Although IL-4 andIL-17 deficient untreated mice developed an OAD com-parable to wild-type recipients, a single cytokine de-ficiency afforded significant protection in CsA-treatedrecipients. In conclusion, CsA treatment unbalances Thelper alloreactivity and favors Th2 and Th17 as coex-isting pathways mediating chronic rejection of hetero-topic tracheal allografts.

Key words: Bronchiolitis obliterans, cyclosporine A,obliterative airway disease, T cells, transplantation andimmunology

Abbreviations: BO, bronchiolitis obliterans; CNI, cal-cineurin inhibitor; CsA, cyclosporine A; GIL, graft in-filtrating lymphocyte; HTT, heterotopic trachea trans-plantation; NF-AT, nuclear factor of activated T cells;NF-jB, nuclear factor-kappa B; OAD, obliterative air-way disease.

Received 15 June 2012, revised 16 November 2012 andaccepted for publication 17 November 2012

Introduction

Introduction of calcineurin inhibitors (CNIs) has significantlyreduced acute rejection rates and was anticipated to haveimplications for long-term prognosis (1). However, currentdata indicate that, contrary to expectations, long-term sur-vival rates poorly reflect short-term improvements. Lungtransplantation (LTx), which remains the only therapeuticapproach for end-stage lung failure, is no exception. The1-year survival rate for LTx is >80% whereas the 10-yearsurvival rate does not exceed 23% (1). Bronchiolitis oblit-erans (BO) is the leading cause of late failure after lungtransplantation. Although immune mechanisms includingT cell-mediated alloreactivity and autoimmunity appear tobe central to the pathology of BO, detailed mechanismsremain poorly understood. Recent reports in rodents andin humans suggest a role for the involvement of Th2 (2)and Th17 (3,4) in the pathogenesis of BO, although thelatter may still be controversial (5). Whereas the largemajority of lung transplant recipients receive CNI-basedimmunosuppressive regimens, more than 50% developBO (6). This suggests that mechanisms underlying BO re-sist treatment by CNIs or could even be promoted by thesedrugs.

Cyclosporine A (CsA) represents the prototypic calcineurininhibitor (CNI). When bound to the intracellular cyclophilin,CsA inhibits the enzymatic activity of calcineurin. As cal-cineurin permits the nuclear translocation of the nuclearfactor of activated T cells (NFAT), CNIs are classicallyknown to downregulate IL-2 synthesis, therefore ulti-mately preventing Th1, CD8 and NK cell functions and IFN-c synthesis. However, IL-2−/− and IFN-c −/− mice have con-served immune capacities and reject cardiac allografts atrates comparable to WT animals (7) and IL-2−/− or IL-2R−/−

mice experience deregulated T cell activation and general-ized autoimmune disease (8,9). In contrast, the suppres-sive impact of CNIs on Th2 and Th17 pathways is controver-sial (10–16). Several studies have shown that CNIs affectpathways other than NFAT, such as nuclear factor-kappaB (NF-jB) or ETS-like transcription factor 1 (ELK-1) (17,18).The current view is that the immunosuppressive propertiesof CNIs result from the inhibition of calcineurin but down-stream mechanisms are not known (19).

611

Lemaıtre et al.

In the heterotopic trachea transplantation (HTT) model de-scribed by Hertz (20), fully allogeneic tracheas develop anobliterative airway disease (OAD) that is considered anexperimental model of BO. In this setting, T cells havebeen shown to be essential, as attested by OAD preven-tion in lymphopenic recipients (21). Among T cell subsets,CD4+ dominate CD8+ as evidenced by the fact that OADis unaffected in CD8−/− recipients and CD4−/− recipientsshow a significant delay of rejection (22). In this study, weused the HTT model to investigate the pathogenicity ofT helper cells in experimental BO in the context of CsAtreatment.

Materials and Methods

Mice

Wild-type C57BL/6 (B6) and BALB/C (B/C) mice were purchased from Har-lan, Netherlands. IL-17A−/− B6 mice were kindly provided by Dr Iwakura(University of Tokyo, Tokyo, Japan). IL-4−/− B6 and IFN-c −/− B/C mice werepurchased from the Jackson Laboratory (Bar Harbor, ME, USA). Eight- to12-week old animals were used and animals were bred in the Institutefor Medical Immunology pathogen-free animal facility. All animals receivedhumane care in compliance with the Principles of Laboratory Animal Careformulated by the National Institute of Health (Guide for the Care and Useof Laboratory Animals, Eighth Edition, National Research Council, 2010) andprotocols were approved by the local committee for animal welfare.

Heterotopic trachea transplantation

HTT was performed according to an adaptation of the method of Hertz (20).Briefly, donor mice were euthanized in 100% CO2. Donor hearts and lungswere then exposed via a midline incision through the skin and peritoneumextending through the rib cage and sternal notch. Thymus tissue was dis-sected away. The trachea was separated from the esophagus by bluntdissection, excised from the first tracheal ring to the main bronchi andplaced in 0.9% sodium chloride until transplantation. Recipient mice wereanaesthetized with a mixture of xylazine (Rompun) 5% and ketamine 10%in phosphate-buffered saline (PBS). After shaving a surface of 0.5 cm ×0.5 cm over the left scapula, a 3 mm incision was made through the dermisand a 1.5 cm × 0.5 cm pouch was created by blunt dissection over theposterior upper back area. One trachea was placed in each pouch. Skin wasthen closed with 5/0 silk suture. The time between recovering and trans-plantation never exceeded 15 min. Recipient mice were monitored until fullrecovery and then every day until killed. At the time of recovering, recipientmice were killed by cervical dislocation and grafts were removed by bluntdissection.

In vivo treatments

When specified, animals were treated daily with 25 mg/kg CsA (Sandim-mun, Novartis Pharma). Treatment always started on the day of transplan-tation. CsA trough levels at 24 and 48 h were 1185 ± 128 ng/mL and 645 ±240 ng/mL, respectively (n = 5 mice/group). Analyses were performedusing a liquid chromatography tandem-mass-spectrometry technique (LC-MS/MS) with the Mass Tox R© ONE minute test (Chromsystems GMBH,Grafelfing, Germany). Chromatographic separation was carried out using a1260 Infinity HPLC system, (Agilent Technologies, Diegem, Belgium). TheMS/MS detection was performed using an Agilent Technologies 6460 TripleQuad LC-MS/MS with a Jet Stream electrospray source ionization.

When specified, 500 lg of a depleting anti-CD4 cocktail (YTS191 +YTA3.1.2) or isotype control (YCATE, both provided by Stephen Cobbold, Sir

Dunn School of Pathology, Oxford University, UK) was intraperitoneally in-jected on days 0, 2, 4, 7, 14 and 21 after trachea transplantation. Depletionefficiency of >90% was measured by flow cytometry on draining lymphnode cells at the time of recover.

Histopathology and CK14 immunostaining

For pathological assessment, grafts were recovered 28 days after trans-plantation. Cross-sectional specimens were fixed in 4% formaldehyde andembedded in paraffin. For pathological assessment, 5 lm sections werecut and stained with hematoxylin-eosin, Masson’s trichrome and Periodicacid-Schiff (PAS). All specimens were examined in blind fashion and scoredas previously described (23). Briefly, all qualitative histological changes werenoted, and four easily identifiable pathological items were scored: (a) air-way lining epithelial loss, (b) luminal obliteration due to granulation tissueformation and/or fibrosis, (c) deposition of extracellular matrix (ECM) and (d)leukocyte infiltration. For (a) and (b), the score ranged from 0 to 4: 0 = nochange, 1 = mild (<25%), 2 = moderate (25–50%), 3 = severe (50–75%)and 4 = very severe damage (>75%). For the ECM deposition, the scorewas 0 = no change, 1 = mild, 2 = moderate, 3 = dense and 4 = very densecollagen deposition evaluated after Masson’s trichrome staining. For theleukocyte infiltration, the score was 0 = no change, 1 = mild perivascularinfiltrates, 2 = moderate perivascular changes, 3 = severe diffuse infiltra-tion and 4 = very severe diffuse infiltration with subepithelial and epithelialcomponents.

For CK14 immunostaining, 5 lm paraffin sections were cut, deparaffinizedand rehydrated. Endogenous peroxidase activity was first quenched byH2O2 peroxidase blocking reagent (DakoCytomation). Sections were thenincubated with 1/100 diluted anti-CK14 antibody (clone LL002, Novocastra,UK) for 20 min at room temperature. Sections were then washed and in-cubated with 1/500 diluted biotinylated goat antirabbit antibody (JacksonImmunoresearch, West Grove, PA, USA) for 20 min at room temperature.Thereafter, streptavidin-HRP was added and coloration was revealed usingdiaminobenzidine (DAB) with the substrate chromogen system from Dako-Cytomation. The total number of positive cells was considered for eachsection.

Flow cytometry

To isolate graft infiltrating and draining lymph node (DLN) lymphocytes, tra-chea transplants and DLNs were minced and then incubated at 37◦C for2 h with type I collagenase at 2 mg/mL (Sigma) in a phosphate bufferedsolution. IFN-c and IL-17A intracytoplasmic stainings were performed af-ter cell incubation with 50 ng/mL PMA and 500 ng/mL ionomycin for 4h with brefeldin A (10 lg/mL) in the last 2 h. For IL-5 staining, the cellswere incubated in PMA/ionomycin with Golgi Plug (BD Biosciences) for4 h. The cells were then washed with PBS with 0.1% Bovine Serum Al-bumin and 0.01% NaN3, incubated for 10 min with Fc block and thenstained for surface markers for 20 min. After fixation and permeabilizationwith CytoFix/CytoPerm (BD Biosciences), cells were washed in PermWashbuffer (BD Biosciences) and labeled with anticytokine antibodies. Cytom-etry analysis was performed on a CyAn-LX cytometer using Summit 4.1software (DakoCytomation, Golstrup, Denmark). Pacific-Blue (PB) conju-gated antimouse CD8 (clone 53–6.7) or antimouse CD3ξ (clone 500A2),Phycoerythrin (PE) conjugated antimouse CD4 (clone RM4–5), antimouseIL5 (clone TRFK5) or antimouse H-2Kb (clone AF6-88.5), Fluorescein isoth-iocyanate (FITC) conjugated antimouse IFN-c (clone XMG1.2), antimouseCD4 (clone RM4–5), peridinin-chlorophyll-protein complex (PerCP) conju-gated antimouse CD3ξ (clone 145–2C11) and antimouse CD16/CD32 (Fcblock, clone 2.4G2) monoclonal antibodies and isotype controls were pur-chased from BD Pharmingen. APC conjugated antimouse IL-17A (cloneeBio17B7) was purchased from eBiosciences.

612 American Journal of Transplantation 2013; 13: 611–620

CsA Favors Th2 and Th17 Alloreactivity

RNA extraction and real-time RT-PCR

Total RNA was extracted from skin grafts using the MagnaPure LC RNAIsolation Kit III for tissue (Roche Diagnostics). Reverse transcription andreal-time PCR were performed using LightCycler-RNA Master HybridizationProbes (one-step procedure) on a Lightcycler 480 apparatus (RocheDiagnostics). The number of mRNA copies was evaluated by using astandard curve for each gene of interest and was normalized to b-actinas the housekeeping gene. Primer and probe sequences were as follows:b-actin: forward CCGAAGCGGACTACTATGCTA, reverse TTTCTCATA-GATGGCGTTGTTG, probe ATCGGTGGCTCCATCCTGGC; IFN-c : forwardGGATGCATTCATGAGTATTGC, reverse GCTTCCTGAGGCTGGATTC, probeTTTGAGGTCAACAACCCACAGGTCCA; IL-17A: forward GCTCCAGAAGGC-CCTCAG, reverse CTTTCCCTCCGCATTGACA, probe ACCTCAACCGTTC-CACGTCACCCTG; IL-13: forward GACCTGAGCAACATCACACAA, reverseGCCAGGTCCACACTCCATAC, probe CCCTGTGCAACGGCAGCATG.

MLC and cytokine production

Cells isolated from spleens of naive, control or CsA-treated trans-planted mice were used as responders (8 × 106 cells/mL) andstimulated with 1 × 107 cells/mL syngeneic B6 or allogeneic B/C irradi-ated splenocytes (2000 rad) in 96-well U-bottom plates (Cellstar, GreinerBio-One, Belgium). Cultures were incubated at 37◦C in a 5% CO2 atmo-sphere in medium consisting of RPMI 1640 supplemented with 2mM L-glutamine, 1 mM nonessential amino acids, 5% heat-inactivated FCS and 1mM sodium pyruvate. IFN-c , IL-13 and IL-17A production were measuredin culture supernatants after 96 h using commercially available ELISA kits(Duoset; R&D Systems, Minneapolis, MN, USA). The detection thresholdwas <20 pg/mL for all cytokines.

Statistical analyses

Statistical analyses of differences between groups were performed usingthe two-tailed Mann–Whitney nonparametric test except for differencesbetween cell culture triplicates that were determined by an unpaired t-test.p < 0.05 was considered statistically significant.

Results

Posttransplant obliterative airway disease develops

under CsA treatmentHTT was performed as described by Hertz et al. (20).In this context, we looked at the effect of CsA treat-ment on the development of posttransplant OAD. Syn-geneic B6 or fully allogeneic B/C tracheas were graftedinto B6 recipients treated, or not, with 25 mg/kg CsA,starting at day 0 and during the whole posttransplanta-tion period. With this dosage, CsA trough levels were1185 ± 128 ng/mL. Grafts were recovered at day 28posttransplantation and pathologic scores were deter-mined as described in materials and methods. OADdid not develop in the absence of alloantigens, asdemonstrated by the presence of a normal pseudostrat-ified epithelium without collagen deposition in both un-treated and CsA-treated syngeneic groups (Figure 1A–C).In contrast, dense fibro-obliterative lesions developed inthe allografts and these lesions were not controlled byCsA treatment. Indeed, both control and CsA-treated allo-grafts showed a similar pathologic score (Figure 1D). Thelesions comprised a complete destruction of the respira-tory epithelium and included leukocyte infiltration. In ad-dition, dense collagen deposits enlarged the lamina pro-

pria, thickened the basal membrane and obstructed thelumen (Figure 1E and F). A network of new vessels under-lay these fibro-obliterative processes. Therefore, our datademonstrate that CsA treatment is ineffective in prevent-ing chronic rejection of fully allogeneic trachea transplants.

CsA counteracts CD8 but not CD4 T cell-mediated

alloreactivityTo characterize the graft T cell infiltrate under CsA treat-ment, B/C trachea were grafted into B6 mice treated,or not, with 25 mg/kg CsA and recovered 8 days aftertransplantation. Graft infiltrating lymphocytes (GILs) wereisolated after collagen digestion and analyzed by flow cy-tometry. B6 MHC-I H-2Kb-specific staining revealed thatmore than 95% of the graft-infiltrating cells derived fromrecipient origin (Supporting Figure S1).

CsA treatment significantly reduced the number of graft-infiltrating CD4+ and CD8+ T lymphocytes, although eachpopulation was not equally affected (Figure 2A, B, D and E).Indeed, CD4+ count was decreased by twofold whereasCD8+ numbers dropped by fivefold, resulting in a signif-icant increase of the intragraft CD4+/CD8+ ratio, as plot-ted in Figure 2C. In contrast, this treatment did not affectthe recruitment of macrophages to the grafts (data notshown).

To investigate the role of CD4+ T cells in obliterative dis-ease, CsA-treated B6 recipients of B/C tracheas were in-jected with a depleting regimen of anti-CD4 mAbs during28 days and analyzed as previously described. Depletion ofCD4+ T cells abrogated graft lesions (Figure 3A), demon-strated by the normal pseudostratified ciliated columnarepithelium secreting mucus that accumulated in the lu-men of anti-CD4-treated mice (Figure 3C). Notably, no signof collagen deposits could be found in the lamina propria,basal membrane or lumen. Altogether, these experimentsconfirm that the development of obliterative lesions in thecontext of CsA treatment depends on CD4+ T cell allore-activity.

IFN-c producing CD8+ T cells and Th1 but neither Th2

nor Th17 alloreactivity is suppressed by CsATo further characterize T cell reactivity associated withchronic trachea allograft rejection, we first analyzed intra-graft cytokine mRNA profiles typically correlated with var-ious T helper cell subsets. We compared syngeneic B6and allogeneic B/C tracheas recovered from B6 recipientstreated or not with CsA for 8 days. mRNA levels werealmost undetectable in syngeneic transplants (Figure 4A,B and C), similar to native tracheas (data not shown). Incontrast, significant amounts of IFN-c and, to a lesserextent, IL-17 and IL-13 mRNAs were detected in allo-grafts. CsA treatment dramatically reduced intragraft IFN-cmRNA (Figure 4A). In contrast, IL-17 and IL-13 mRNA lev-els increased, although not reaching statistical significance(Figure 4B and C). To confirm these results at the singlecell level, intracytoplasmic staining of GILs was performed

American Journal of Transplantation 2013; 13: 611–620 613

Lemaıtre et al.

Allo

gen

eic

CTRL CsA0

4

8

12

16

Pat

ho

log

ic S

core

CsA CTRL

Syn

gen

eic

CTRL CsA0

4

8

12

16

Pat

ho

log

ic S

core

CsA CTRL

A B C

D E F

Figure 1: Posttransplant obliterative airway disease develops under CsA treatment. Syngeneic B6 or fully allogeneic B/C tracheaswere heterotopically grafted into B6 recipients treated or not with 25 mg/kg CsA and recovered after 28 days. Upper panel, syngeneicgrafts. (A) Pathologic score. (B and C) Histologic analysis of one representative untreated and CsA-treated B6 trachea, respectively.Masson’s trichrome staining of a whole section (magnification × 40 and × 200 for the inset). The empty arrow indicates the normallamina propria, the empty arrowhead shows the normal pseudostratified respiratory epithelium and the empty star identifies the freelumen. Lower panel, allogeneic grafts. (D) Pathologic score. (E and F) Histologic analysis of one representative untreated and CsA-treatedB/C trachea, respectively. Masson’s trichrome staining of a whole section (magnification × 40 and × 200 for the inset). The black arrowsindicate the enlarged lamina propria, the black arrowheads show the basal membrane thickening and adjacent epithelial loss and blackstars identify the fibroproliferative process obstructing the tracheal lumen. The bars represent the mean ± SEM of n = 5 organs insyngeneic groups and 11–15 organs in allogeneic groups. Data are pooled from three independent experiments.

at day 8 posttransplantation. CsA treatment decreasedby one-third the number of IFN-c -producing Th1 cells(Figure 4E, I and J), with no significant consequences onthe accumulation of intragraft Th17 (Figure 4F, I and J) orTh2 cells (Figure 4G). Similarly, the percentage of foxp3-positive regulatory T cells was comparable in both groups(Figure 4H). Finally, the number of IFN-c + CD8+ T cells dra-matically dropped under CsA treatment (Figure 4K, L andM). These results indicate that CsA treatment, althoughefficient in reducing IFN-c , did not prevent Th2 and Th17alloreactivity.

CsA differentially affects T cell alloreactivityWe next assessed whether CsA could modulate T cellpriming. For this purpose, we performed mixed lympho-cyte cultures (MLC) with transplanted animals. Eight daysafter transplantation, spleen cells from either control orCsA-treated recipients were stimulated with syngeneic B6or donor-specific B/C stimulators. Spleen cells from naiveanimals served as controls. Syngeneic stimulation did notinduce detectable cytokine levels (data not shown). B/Cstimulation-induced high amounts of IFN-c in untreatedrecipients that was prevented by in vivo CsA treatment(Figure 5A). As we saw with qPCR experiments, CsA treat-

ment strongly induced IL-17 production and did not affectIL-13 secretion (Figure 5B and C). The downregulation ofIFN-c producing T cells was confirmed by a concomitantintracytoplasmic staining of CD4+ and CD8+ T cells fromdraining lymph nodes (Figure 5D).

IFN-c regulates OADBecause CsA potently controls IFN-c secretion by T cells,we next addressed its role in OAD in the absence ofCsA treatment. For this purpose, B6 tracheas were trans-planted into WT or IFN-c −/− B/C recipients. Preliminaryexperiments have shown comparable OAD in either B/Cinto B6 or B6 into B/C combinations (Supporting Fig-ure S2). Grafts recovered from IFN-c −/− recipients exhib-ited an increased pathologic score compared to WT recip-ients (Figure 6A). Indeed, fibro-obliteration of the lumenwas increased in IFN-c −/− recipients (Figure 6B and C).These results support regulatory functions of IFN-c in OADdevelopment.

IL-4 and IL-17 deficiency independently improve

posttransplant obliterative airway diseaseAs Th17 and Th2 cells were not inhibited by CsA treatment,we assessed their pathogenic role in OAD development.



CD

8

CD4

CTRL CsA

CTRL CsA0

4000

8000 **

Nu

mb

er

of

CD

3+C

D4+

GIL

s

CTRL CsA0

4000

8000

***

Nu

mb

er

of

CD

3+C

D8+

GIL

s

CTRL CsA0

1

2

3

4 ***

Rat

io C

D4+

/CD

8+

amo

ng

CD

3+G

ILs

A B C

D

6245

886

2958

514

4533 660 928 156

E

Figure 2: CsA modulates graft-

infiltrating T cells. Fully allogeneic B/Ctracheas were heterotopically graftedinto B6 recipients treated, or not(CTRL), with 25 mg/kg CsA and recov-ered after 8 days. GILs were extractedand analyzed by flow cytometry. (A)Absolute CD3+CD4+ T cell count pergraft. (B) Absolute CD3+CD8+ T cellcount per graft. (C) Ratio betweenCD4+ and CD8+ T cells among CD3+GILs. (D and E) Representative plotsof CD4+ and CD8+ cells among CD3+GILs in control and CsA-treated groups.The bars represent the mean ± SEMof 14 and 12 individual grafts, respec-tively. Results are pooled from three in-dependent experiments. ∗∗p < 0.005and ∗∗∗p = 0.0001.

We first transplanted B/C tracheas into IL-17A−/− B6 an-imals treated or not with CsA for 28 days. While dis-ease development in untreated IL–17A−/− recipients wassimilar to the pathology that occurred in WT recipients,IL-17 deficiency in the context of CsA treatment ledto a significant improvement of the pathologic score(Figure 7A). Indeed, tracheas from CsA-treated IL-17−/− re-cipients showed almost no luminal fibro-obliteration andfewer collagen deposits (Figure 7B and C). In addition,CK14 immunostaining was performed based on the im-portant role of these stem cells in epithelium regener-ation (24). CK14+ basal epithelial stem cells were pre-served in CsA-treated IL-17−/− recipients (Figure 7D). Inparallel, we evaluated the impact of IL-4 deficiency in IL-4−/− B6 recipients treated or not with CsA. Similar to IL-17 deficiency, OAD fully developed in IL-4−/− recipientsbut was significantly prevented in CsA-treated animals

(Figure 7E). We found epithelial protection and CK14+

stem cell preservation, fewer collagen deposits, andcomplete lumen patency in these organs (Figure 7F–H).Together, these experiments reveal IL-17 and IL-4 as inde-pendent pathogenic cytokines in OAD development underCsA treatment.

Discussion

Bronchiolitis obliterans represents the leading cause ofchronic allograft failure and late death after lung transplan-tation despite the use of potent immunosuppressive drugsincluding CNIs. Herein, we focused on the link betweenCNIs, Th cells and chronic allograft rejection. For that pur-pose we used a model of posttransplant obliterative airwaydisease (OAD) appearing under a dosage of CsA with high

CTRL CD4-0

4

8

12

16

DEPL.

*

Pat

ho

log

ic S

core

A B C

CD4-DEPL. CTRL

Figure 3: Development of obliterative lesions under CsA treatment depends on CD4± T cell alloreactivity. Fully allogeneic B/Ctracheas were grafted into either control or CD4+ T cell-depleted B6 recipients treated with CsA for 28 days. (A) Pathologic score is com-pared. (B and C) Histologic analysis of one representative control and CD4-depleted transplanted trachea, respectively. Periodic acid-Schiffstaining of a whole section (magnification × 40 and × 200 for the inset). The black arrow indicates the enlarged lamina propria, the blackarrowhead shows the basal membrane thickening and adjacent epithelial loss and the black star identifies the fibroproliferative processobstructing the lumen. The empty arrow indicates the normal lamina propria, the empty arrowhead shows the normal pseudostratifiedrespiratory epithelium and the empty star indicates mucus accumulating in the lumen. The bars represent the mean ± SEM of 5 organsin each group. ∗p < 0.05.


Lemaıtre et al.

CTRL CsA0

3

6%

IL-1

7+ C

D4+

CTRL CsA0

5

10

15 ***

% IF

N-

+ C

D4+

CTRL CsA0

10

20

30

40 ***

% IF

N-

+ C

D8+

CTRL CsA

IL-1

7

4.8 ± ±

±

± ±

±

0.6 %

13.9 0.8 %

3.4 0.6 %

8.1 0.5 %

IFN-γ

CTRL CsA

IL-1

7

IFN-γ

30.5 2 % 9.7 1 %

CTRL CsA0

10

20

30

% f

oxp

3+C

D4+

Intr

agra

ft m

RN

A l

evel

s

Gra

ft in

filt

rati

ng

lym

ph

ocy

tes

IFN-γ IL-17A IL-13A B C

D

H

E F G

J K

CTRL CsA0

1

2

3 p = 0.1

% IL

-5+ C

D4+

I

L

CD

4+ T

cel

lsC

D8+

T c

ells

0

100

200

300

400

500 *** ***

Re

lati

ve m

RN

Ae

xpre

ss

ion

0

20

40

60

80 **

Re

lati

ve m

RN

Ae

xpre

ss

ion

0

5

10

15

**

Re

lati

ve m

RN

Ae

xpre

ss

ion

SYNGSYNG + CsA

ALLOALLO + CsA

γ

γ

Figure 4: CsA inhibits CD8 and Th1 but not Th2 and Th17 alloreactivity. B6 (SYNG) or B/C (ALLO) tracheas were grafted into B6recipients treated, or not (CTRL) with 25 mg/kg CsA for 8 days. Upper panel, intragraft mRNA levels. (A) IFN-c mRNA. (B) IL-17A mRNA. (C)IL-13 mRNA. Results are expressed as mean ± SEM of cytokine copy numbers normalized on 107 copies of b-Actin, n = 9 organs pooledfrom three independent experiments. ∗∗p < 0.005 and ∗∗∗p = 0.0001. Lower panels, GIL analysis after membranous and intracellularimmunostaining. IFN-c (D), IL-17A (E), IL-5 (F) and foxp3 (G) expression by CD3+ CD4+ cells in each group. (H and I) Plots represent theintracellular expression of IL-17A and IFN-c among CD3+ CD4+ GILs in a control and CsA-treated graft, respectively. (J) IFN-c expressionin CD3+ CD8+ GILs. (K and L) Plots represent the intracellular expression of IL-17A and IFN-c among CD3+ CD8+ GILs in a control andCsA-treated graft, respectively. The bars represent the mean ± SEM. n = 15 organs pooled from three independent experiments exceptfor IL-5 and foxp3 expression in CD4+ cells: n = 5. ∗∗∗p < 0.001.



0

2000

4000

6000

8000

10000 ** *

pg

/ml

0

200

400

600

800 ***

pg

/ml

0

500

1000

1500

**

pg

/ml

IFN-γ IL-17 IL-13

A

CTRL CsA0

5

10

15 ***

% IF

N-

+ C

D8+

amo

ng

CD

3+

CTRL CsA0.0

0.5

1.0

1.5 *

% IF

N-

+ C

D4+

amo

ng

CD

3+

D

B C

E

NAIVEALLO ALLO + CsAMix

ed ly

mp

ho

cyte

cu

ltu

reD

rain

ing

lym

ph

no

de

cells

γ γ

Figure 5: CsA differentially affects T cell alloreactivity. Upper panel, spleen cells from naıve, control or CsA-treated HTT recipients(8 × 106 cells/mL) were stimulated with irradiated donor-type B/C spleen cells for 90 h and cytokine levels were measured in culturesupernatants by ELISA. (A) IFN-c . (B) IL-17. C, IL-13. The bars represent the mean ± SEM. The values shown were measured in triplicateand are representative of two independent experiments. ∗p < 0.05 and ∗∗p < 0.005. Lower panel, DLN lymphocyte analysis aftermembranous and intracellular immunostaining at day 8 posttransplantation. (A) % IFN-c + CD3+ CD8+ lymphocytes. (B) % IFN-c + CD3+CD4+ lymphocytes. The bars represent the mean ± SEM. n = 10 per group pooled from two independent experiments. ∗p < 0.05 and∗∗∗p < 0.001.

trough levels. Although King et al. found that this dosagereduced lumenal fibrosis after 28 days (25), we did notobserve a protection by CsA. This apparent discrepancymight be related to the different allogeneic strain combina-tion used. Although inefficient in preventing OAD after 28days, we found that CsA treatment significantly decreasedpathologic score 14 days after transplantation, which isin agreement with other observations (1) (SupportingFigure S3). Altogether, the experimental equivalent of BO

in our model gave us the opportunity to study the mech-anisms underlying disease development in the context ofCNI treatment.

Quantitative analysis of graft infiltrating lymphocyte sub-sets showed a dramatic impact of CsA on CD8+ T cells,leaving CD4+ T cells as the main effectors since OADwas completely prevented by CD4+ depletion at day 28.Others have reported an increased calcineurin requirement

WT IFN- -/-0

4

8

12

16 *

Pat

ho

log

ic S

core

A B C

WT IFN-γ-/- γ

Figure 6: IFN-c deficiency enhances OAD. Fully allogeneic B6 tracheas were heterotopically grafted into WT or IFN-c −/− B/C recipientsand recovered after 21 days. (A) Pathologic score. (B and C) Histologic analysis of one representative allogeneic trachea in each group.Masson’s trichrome staining of a whole section (magnification × 40 and inset × 200). The empty arrowhead shows the basementmembrane and adjacent epithelial loss and the empty star identifies the free lumen. The black arrow identifies the enlarged laminapropria, the black arrowhead shows the basal membrane thickening and adjacent epithelial loss and black stars identify the fibroproliferativeprocess obstructing the tracheal lumen. The bars represent the mean ± SEM of 3–6 organs in each group. ∗p < 0.05.


Lemaıtre et al.

IL-1

7-/-

RE

CIP

IEN

TS

IL

-4-/

- R

EC

IPIE

NT

S

CTRL CsA0

4

8

12

16

***P

ath

olo

gic

Sco

re

CTRL CsA0

4

8

12

16*

Pat

ho

log

ic S

core

CsA CTRL

CsA CTRL

CTRL CsA0

50

100

150p = 0.06

CK

14+ c

ell c

ou

nt

A B

E F H G

D C

CTRL CsA0

50

100

150

*

CK

14+ c

ell c

ou

nt

Figure 7: IL-4 and IL-17 deficiency independently prevent posttransplant OAD. B/C tracheas were heterotopically grafted into eitherIL-17A−/− or IL-4−/− B6 recipients treated, or not (CTRL), with 25 mg/kg CsA and recovered after 28 days. (A and E) Pathologic scoresof grafts recovered from IL-17A−/− or IL-4−/− recipients, respectively. (B, C and F, G) Histologic analysis of one representative allogeneictrachea in each group. Masson’s trichrome staining of a whole section (magnification × 40 and inset × 200). The black arrows identifythe enlarged lamina propria, the black arrowheads show the basal membrane thickening and adjacent epithelial loss and black starsindicate the fibroproliferative process obstructing the lumen. The empty arrows identify the normal lamina propria, the empty arrowheadsshow the normal pseudostratified respiratory epithelium and the empty stars indicate mucus accumulating in the free lumen. (D andH) Absolute CK14+ basal cell count per section after specific immunostaining of grafts recovered from IL-17A−/− or IL-4−/− recipients,respectively. The bars represent the mean ± SEM of 5–15 organs in each group. Data for IL-17−/− recipients are pooled from threeindependent experiments. ∗p < 0.05 and ∗∗∗p < 0.001.

of CD8+ T cells compared to CD4+ cells (26). CD8+ T cellscould be perceived as potentially damaging anti-MHC I al-loreactive T cells (27). Nevertheless, they also behave asimportant IFN-c providers (28,29) and IFN-c has knownantifibrotic properties. Specifically, IFN-c suppresses col-lagen synthesis in fibroblasts and promotes the activationof inflammatory macrophages that favor the degradationof extracellular matrix components (ECM) (30). Based onthe powerful antifibrotic effects of IFN-c and our data withIFN-c −/− recipients, the net effect of CD8+ T cell inhibitionby CsA could be equivalent to the deprivation of antifibroticregulatory mechanisms (31). In addition, CD8-derived IFN-c may directly affect Th2 and Th17 differentiation as ob-served in other settings (29,32). Furthermore, other stud-ies have reported a regulatory role for IFN-c in rejectionprocesses (33,34) consistent with the results reported inour study.

Our results provide evidence for the causal relationshipbetween either IL-17 or IL-4 and the OAD process. CsAtreatment failed to control Th17 and Th2 alloreactivity, asattested by IL-17 and IL-13 mRNAs in rejected allograftsand IL-17 and IL-5 expressing CD4+ T cells in GILs. Inter-

estingly, these results mirror others obtained from bron-choalveolar lavage fluid analyses in lung transplant recip-ients in which increased levels of IL-17 or IL-13 mRNAwere observed in BO patients compared to stable re-cipients, the large majority of both receiving CNI treat-ment (2,4). Indeed, although we cannot exclude the possi-bility that other IL-4- or IL-17-producing cells are involved,we provide compelling evidence that both Th2 and Th17alloreactive cells constitute independent and codominantpathways of chronic allograft rejection developing duringCsA treatment. Apparently contrasting with these find-ings, Snell’s study did not associate IL-17 and early BOsyndrome (5). Even, they found increased IL-17 amounts inendobronchial biopsies when CsA levels were at their high-est. In addition to early time points, this study was based ona relatively small number of patients, as mentioned by theauthors.

A possible synergy of the Th2 and Th17 pathways may con-tribute to the results seen in our study, as demonstratedin other experimental models (35). Importantly, recentreports have implicated Th17 autoreactive cells directedagainst type-V collagen in the pathogenesis of both human



and rodent lung obliterative diseases (3,36,37). Althoughsyngeneic transplants did not develop OAD (21), the re-spective role of alloantigen-reactive versus self-reactive Tcells has not been investigated in our model. Mirroringthe difference between healed syngeneic and rejected al-logeneic grafts (Supporting Figure S3), the persistence ofCK14+ basal epithelial stem cells confirmed the protectionafforded in IL-17−/− and IL-4−/− recipients. Indeed, epithe-lial stem cells have shown potent regenerative capacitiesafter injury (24) and the epithelium is known to be the pri-mary target in posttransplant OAD (38). In addition, Th2cytokines (39) and IL-17 (40) have been causally linked tothe development of fibrosis in a variety of chronic inflam-matory diseases. Accordingly, IL-4 and IL-13 have beenshown to induce the proliferation and differentiation of fi-broblasts (41) while IL-17, either directly or through theinduction of IL-6, may promote collagen production by fi-broblasts (42).

Previous studies addressing the impact of CNIs on T cellresponses have reported conflicting results. Although CsAwas shown to inhibit IL-17 and Th2 cytokine productionby in vitro stimulated human peripheral blood mononu-clear cells (PBMCs) (13,15,16), these observations havenot been confirmed by others (10). In our hands, in vivo CsAtreatment significantly inhibited IFN-c producing CD4+ andCD8+ T cells, but was inefficient in controlling IL-17- andIL-5-producing GILs. This was also observed by others in acardiac allograft model (43). Another study, using a rat HTTmodel in the omentum, concluded that CsA treatment in-hibited both Th1 and Th17 pathways (44). This apparentdiscordance might be related to the model and method-ology. In addition, the roles of IL-17 and IFN-c were notaddressed specifically. The paradigm of a downregulationof IFN-c by CsA, biasing the T cell response toward Th2and Th17 is largely supported by our MLC results. In addi-tion, a different impact of CsA on naive and memory CD4+

cells could underlay these mechanisms as a previous re-port showed that memory PBMCs were less sensitive thannaive cells to CsA-mediated inhibition (45).

In conclusion, our animal studies highlight a potential rolefor CsA in promoting Th2 and/or Th17-mediated OAD,possibly through the inhibition of CD4+ and CD8+ T cell-derived IFN-c production. Targeting IL-4 and/or IL-17 in ad-dition to current protocols may represent a valuable strat-egy in clinical transplantation.

Acknowledgments

Financial support: the Institute for Medical Immunology is funded by re-search grants of the Walloon Region, the FNRS-Belgium and GlaxoSmithK-line Biologicals. P.H.L. is a doctoral researcher funded by the FNRS (FondsNational de la Recherche Scientifique, Belgium) and the Fonds Erasme (Uni-versite Libre de Bruxelles, Brussels, Belgium).

The authors would like to thank Dr. P. Horlait, Laurent Depret, ChristopheNotte, Gregory Watherlot and Samuel Vanderbist for outstanding animalcare; Frederic Paulart, Nicolas Passon and Frederic Cotton for technicalassistance; Angelique Francois, Morgane Delanoy and Dr. M. Petein forhistopathology processing and Pr. S. Cobbold for the kind donation of theanti-CD4+ and the control antibodies.

Disclosure

The authors of this manuscript have no conflicts of inter-est to disclose as described by the American Journal ofTransplantation.

References

1. Lodhi SA, Lamb KE, Meier-Kriesche HU. Solid organ allograft sur-vival improvement in the United States: The long-term does notmirror the dramatic short-term success. Am J Transplant 2011; 11:1226–1235.

2. Keane MP, Gomperts BN, Weigt S, et al. IL-13 Is pivotal in thefibro-obliterative process of bronchiolitis obliterans syndrome. JImmunol 2007; 178: 511–519.

3. Burlingham WJ, Love RB, Jankowska-Gan E, et al. IL-17-dependent cellular immunity to collagen type v predisposes toobliterative bronchiolitis in human lung transplants. J Clin Invest2007; 117: 3498–3506.

4. Vanaudenaerde BM, De Vleeschauwer SI, Vos R, et al. The role ofthe IL23/IL17 Axis in bronchiolitis obliterans syndrome after lungtransplantation. Am J Transplant 2008; 8: 1911–1920.

5. Snell GI, Levvey BJ, Zheng L, et al. Interleukin-17 and airway in-flammation: A longitudinal airway biopsy study after lung trans-plantation. J Heart Lung Transplant 2007; 26: 669–674.

6. Christie JD, Edwards LB, Kucheryavaya AY, et al. The Registryof the International Society for Heart and Lung Transplantation:Twenty-eighth adult lung and heart-lung transplant report—2011.J Heart Lung Transplant 2011; 30: 1104–1122.

7. Zand MS, Li Y, Hancock W, et al. Interleukin-2 and interferon-gamma double knockout mice reject heterotopic cardiac allografts.Transplantation 2000; 70: 1378–1381.

8. Sadlack B, Lohler J, Schorle H, Klebb G, et al. Generalized au-toimmune disease in interleukin-2-deficient mice is triggered byan uncontrolled activation and proliferation of CD4+ T cells. Eur JImmunol 1995; 25: 3053–3059.

9. Suzuki H, Kundig TM, Furlonger C, et al. Deregulated T cell activa-tion and autoimmunity in mice lacking interleukin-2 receptor beta.Science 1995; 268: 1472–1476.

10. Abadja F, Videcoq C, Alamartine E, Berthoux F, Mariat C. Differ-ential effect of cyclosporine and mycophenolic acid on the humanregulatory T cells and TH-17 cells balance. Transplant Proc 2009;41: 3367–3370.

11. Abadja F, Atemkeng S, Alamartine E, Berthoux F, Mariat C. Impactof mycophenolic acid and tacrolimus on Th17-related immune re-sponse. Transplantation 2011; 92: 396–403.

12. Jiang H, Yang X, Soriano RN, Fujimura T, Krishnan K, KobayashiM. Distinct patterns of cytokine gene suppression by the equiv-alent effective doses of cyclosporine and tacrolimus in rat heartallografts. Immunobiology 2000; 202: 280–292.


Lemaıtre et al.

13. Pacocha SE, Oriente A, Huang SK, Essayan DM. Regulation ofantigen-induced human T-lymphocyte responses by calcineurin an-tagonists. J Allergy Clin Immunol 1999; 104: 828–835.

14. Shimizu K, Fujii S, Fujimoto K, Kawa K, Yamada A, Kawano F.Tacrolimus (FK506) treatment of CD34+ hematopoietic progenitorcells promote the development of dendritic cells that drive CD4+T cells toward Th2 responses. J Leukoc Biol 2000; 68: 633–640.

15. Tsuda K, Yamanaka K, Kitagawa H, et al. Calcineurin inhibitors sup-press cytokine production from memory T cells and differentiationof naive T cells into cytokine-producing mature T cells. PLoS One2012; 7: e31465.

16. Zhang C, Zhang J, Yang B, Wu C. Cyclosporin A inhibits the produc-tion of IL-17 by memory Th17 cells from healthy individuals andpatients with rheumatoid arthritis. Cytokine 2008; 42: 345–352.

17. Frischbutter S, Gabriel C, Bendfeldt H, Radbruch A, Baumgrass R.Dephosphorylation of Bcl-10 by calcineurin Is essential for canon-ical NF-KappaB activation in Th cells. Eur J Immunol 2011; 41:2349–2357.

18. Sugimoto T, Stewart S, Guan KL. The calcium/calmodulin-dependent protein phosphatase calcineurin is the major Elk-1phosphatase. J Biol Chem 1997; 272: 29415–29418.

19. Halloran PF. Mechanism of action of the calcineurin inhibitors.Transplant Proc 2001; 33: 3067–3069.

20. Hertz MI, Jessurun J, King MB, Savik SK, Murray JJ. Reproductionof the obliterative bronchiolitis lesion after heterotopic transplan-tation of mouse airways. Am J Pathol 1993; 142: 1945–1951.

21. Kelly KE, Hertz MI, Mueller DL. T cell and major histocompatibilitycomplex requirements for obliterative airway disease in hetero-topically transplanted murine tracheas. Transplantation 1998; 66:764–771.

22. Higuchi T, Jaramillo A, Kaleem Z, Patterson GA, MohanakumarT. Different kinetics of obliterative airway disease development inheterotopic murine tracheal allografts induced by CD4+ and CD8+T cells. Transplantation 2002; 74: 646–651.

23. Belperio JA, Keane MP, Burdick MD, et al. Critical role for CXCR3chemokine biology in the pathogenesis of bronchiolitis obliteranssyndrome. J Immunol 2002; 169: 1037–1049.

24. Hong KU, Reynolds SD, Watkins S, Fuchs E, Stripp BR. Basal cellsare a multipotent progenitor capable of renewing the bronchialepithelium. Am J Pathol 2004; 164: 577–588.

25. King MB, Jessurun J, Savik SK, Murray JJ, Hertz MI. Cyclosporinereduces development of obliterative bronchiolitis in a murine het-erotopic airway model. Transplantation 1997; 63: 528–532.

26. Auchincloss H, Jr., Winn HJ. Murine CD8+ T cell helper functionis particularly sensitive to cyclosporine suppression in vivo. J Im-munol 1989; 143: 3940–3943.

27. Richards DM, Dalheimer SL, Hertz MI, Mueller DL. Trachea allo-graft class I molecules directly activate and retain CD8+ T cellsthat cause obliterative airways disease. J Immunol 2003; 171:6919–6928.

28. Le Moine A, Goldman M, Abramowicz D. Multiple pathways toallograft rejection. Transplantation 2002; 73: 1373–1381.

29. Teixeira LK, Fonseca BP, Vieira-de-Abreu A, et al. IFN-gamma pro-duction by CD8+ T cells depends on NFAT1 transcription factor andregulates Th differentiation. J Immunol 2005; 175: 5931–5939.

30. Wynn TA. Integrating mechanisms of pulmonary fibrosis. J ExpMed 2011; 208: 1339–1350.

31. Oldroyd SD, Thomas GL, Gabbiani G, El Nahas AM. Interferon-gamma inhibits experimental renal fibrosis. Kidney Int 1999; 56:2116–2127.

32. Chan SY, DeBruyne LA, Goodman RE, Eichwald EJ, Bishop DK. Invivo depletion of CD8+ T cells results in Th2 cytokine productionand alternate mechanisms of allograft rejection. Transplantation1995; 59: 1155–1161.

33. Mele TS, Kneteman NM, Zhu LF, et al. IFN-gamma is an absoluterequirement for spontaneous acceptance of liver allografts. Am JTransplant 2003; 3: 942–951.

34. Sawitzki B, Kingsley CI, Oliveira V, Karim M, Herber M, Wood KJ.IFN-gamma production by alloantigen-reactive regulatory T cells isimportant for their regulatory function in vivo. J Exp Med 2005;201: 1925–1935.

35. Besnard AG, Togbe D, Couillin I, et al. Inflammasome-IL-1-Th17response in allergic lung inflammation. J Mol Cell Biol 2012; 4:3–10.

36. Fukami N, Ramachandran S, Saini D, et al. Antibodies to MHCclass I induce autoimmunity: Role in the pathogenesis of chronicrejection. J Immunol 2009; 182: 309–318.

37. Yoshida S, Haque A, Mizobuchi T, et al. Anti-type V collagen lym-phocytes that express IL-17 and IL-23 induce rejection pathologyin fresh and well-healed lung transplants. Am J Transplant 2006;6: 724–735.

38. Fernandez FG, Jaramillo A, Chen C, et al. Airway epithelium is theprimary target of allograft rejection in murine obliterative airwaydisease. Am J Transplant 2004; 4: 319–325.

39. Wynn TA. Fibrotic disease and the T(H)1/T(H)2 paradigm. Nat RevImmunol 2004; 4: 583–594.

40. Simonian PL, Roark CL, Born WK, O’Brien RL, Fontenot AP.Gammadelta T cells and Th17 cytokines in hypersensitivitypneumonitis and lung fibrosis. Transl Res 2009; 154: 222–227.

41. Saito A, Okazaki H, Sugawara I, Yamamoto K, Takizawa H. Potentialaction of IL-4 and IL-13 as fibrogenic factors on lung fibroblasts invitro. Int Arch Allergy Immunol 2003; 132: 168–176.

42. Booth AJ, Bishop DK. TGF-beta, IL-6, IL-17 and CTGF direct multi-ple pathologies of chronic cardiac allograft rejection. Immunother-apy 2010; 2: 511–520.

43. Wang H, Hosiawa KA, Min W, et al. Cytokines regulate the patternof rejection and susceptibility to cyclosporine therapy in differentmouse recipient strains after cardiac allografting. J Immunol 2003;171: 3823–3836.

44. Ropponen JO, Syrjala SO, Krebs R, Nykanen A, Tikkanen JM, Lem-strom KB. Innate and adaptive immune responses in obliterativeairway disease in rat tracheal allografts. J Heart Lung Transplant2011; 30: 707–716.

45. Schwinzer R, Siefken R. CD45RA+ and CD45RO+ T cells differ insusceptibility to cyclosporin A mediated inhibition of interleukin-2production. Transpl Immunol 1996; 4: 61–63.

Supporting Information

Additional Supporting Information may be found in the on-line version of this article at the publisher’s web site:

Figure S1: Graft-infiltrating CD4± T cells are recipient

derived.

Figure S2: B/C into B6 and B6 into B/C OAD are com-

parable.

Figure S3: CsA treatment prevents acute lesions of tra-

cheal allografts at day 14.

Figure S4: CK14 counts in natives, syngeneic grafts and

allografts.


cyclosporine a drives a th17- and th2-mediated posttransplant obliterative airway disease

Documents