Bronchiolitis obliterans syndrome is characterized by fibrotic obliteration of small airways which severely impairs graft function and survival after lung transplantation. Bronchial epithelial cells from the transplanted lung can undergo epithelial to mesenchymal transition and this can be accentuated by activated macrophages. Macrophages demonstrate significant plasticity and change phenotype in response to their microenvironment. In this study we aimed to identify secretory products from macrophages that might be therapeutic targets for limiting the inflammatory accentuation of epithelial to mesenchymal transition in bronchiolitis obliterans syndrome. TNFα, IL-1β and IL-8 are elevated in bronchoalveolar lavage from lung transplant patients prior to diagnosis of bronchiolitis obliterans syndrome. Classically activated macrophages secrete more TNFα and IL-1β than alternatively activated macrophages and dramatically accentuate TGF-β1-driven epithelial to mesenchymal transition in bronchial epithelial cells isolated from lung transplant patients. Blocking TNFα, but not IL-1β, inhibits the accentuation of epithelial to mesenchymal transition. In a pilot unblinded therapeutic intervention in five patients with progressive bronchiolitis obliterans syndrome, anti-TNFα treatment improved forced expiratory volume in 1 second and 6-min walk distances in four patients. Our data identify TNFα as a potential new therapeutic target in bronchiolitis obliterans syndrome deserving of a randomized placebo controlled clinical trial.
Over the last 25 years lung transplantation has evolved from an experimental intervention to an accepted therapeutic option for patients with end-stage lung disease. However, long-term survival remains limited to a median of 5 years by the development of bronchiolitis obliterans syndrome (BOS) . BOS is characterized by inflammation and fibrosis in small and medium-sized airways leading to airflow obstruction . Transforming growth factor-β1 (TGF-β1) is critically important in BOS development with elevated levels found in bronchoalveolar lavage (BAL) postlung transplant . It can induce fibroblast proliferation and differentiation of fibroblasts to a myo-fibroblasts phenotype . In addition, an important role for TGF-β1 driven epithelial to mesenchymal transition (EMT) in the pathogenesis of BOS has been identified [5-7].
Several studies have demonstrated an accentuative effect of inflammatory cytokines on TGF-β1-driven EMT [6, 8, 9]. There is growing interest in the role of the macrophage as an effector cell in allograft injury and fibrosis. Airway macrophages from posttransplant patients with histological evidence of acute rejection express increased levels of proinflammatory cytokines compared to posttransplant patients without rejection . Furthermore, in the murine heterotopic tracheal transplant model depletion of recipient macrophages significantly abrogates obliteration of the transplanted airway . Our group has previously shown that lipopolysaccharide-activated monocytic cells can accentuate TGF-β1 driven EMT further highlighting the macrophage as a key immune cell in airway remodeling .
Macrophages demonstrate remarkable plasticity and change their phenotype in response to the microenvironment. Classically activated macrophages (CAMφ) are induced during cell-mediated responses and are a vital component of host defense. They secrete high levels of proinflammatory cytokines and their activation is tightly controlled because the cytokines they produce can lead to tissue damage. Alternatively activated macrophages (AAMφ) are programmed to promote wound healing by secretion of extracellular matrix. However, AAMφ can be detrimental to the host when their matrix-enhancing activity is dysregulated and have been identified in the lungs of mice with experimental asthma .
Interestingly, selective depletion of macrophages in a liver fibrosis model revealed distinct macrophages populations associated with injury and recovery phases of inflammatory scarring . Macrophages in CCR4-deficient mice exhibit an AAMφ phenotype and are protected from airway remodeling in response to bleomycin challenge . In posttransplant patients there is an early elevation in Th1-cytokines in those who developed BOS . Furthermore, CD4+ T cells in patients who developed BOS are of a Th1-phenotype  suggesting that the microenvironment within the lung allograft may skew the immune response toward a Th1/CAMφ phenotype that predisposes to BOS.
We hypothesized that the ability to accentuate TGF-β1-driven EMT might be limited to CAMφ—and their secretory products might be a target for limiting the inflammatory accentuation of EMT in the development of BOS.
Materials and Methods
Primary bronchial epithelial cells (PBEC) and BAL were sampled in accordance with approval from the Newcastle and North Tyneside Local Regional Ethics Committee and informed written consent from all study patients (2001/179).
PBEC from stable lung transplant recipients were isolated as previously described .
Patients who had undergone lung transplantation were followed longitudinally with regular surveillance bronchoscopy (1, 3, 6 and 12 months) and at the time a fall in lung function was identified. BAL was performed in a standardized manner using 180 mL of sterile 0.9% sodium chloride administered into the right middle or lower lobe or the left lower lobe. Between 2007 and 2010, BAL samples were collected from 52 patients, including 26 patients who developed BOS within 3 years of transplant.
Macrophage differentiation and stimulation
THP-1 monocytic cells were differentiated into CAMφ and AAMφ by incubating with phorbol myristate acetate (PMA) (5 ng/mL) + IFNγ (20 ng/mL) or IL-4/IL-13 (both 20 ng/mL) for 24 h. AM were differentiated into CAMφ and AAMφ as above in the absence of PMA.
Pseudomonas aeruginosa whole cell lysates were prepared from clinical isolates from our local repository of posttransplant patients as previously described .
Cytokine levels were measured by ELISA performed with commercially available antibody pairs (R&D Systems). TNFα, IL-1β and IL-8 concentration in BAL was measured using the ultrasensitive human proinflammatory-4 kit (Meso Scale Discovery).
Active TGF-β was measured using MFB-F11 mouse embryonic fibroblasts containing a TGF-β induced secreted alkaline phosphatase reporter gene as previously described .
Total cell lysates (10 μg) were separated on 4–12% bis-Tris gels (Invitrogen) and electrophoretically blotted onto HyBond-P Polyvinylidene difluoride (Amersham). Membranes were incubated with primary antibodies and detected with HRP-labeled IgG conjugates. Antibody complexes were visualized using the SuperSignal West Pico chemiluminescent kit (Perbio). Results are normalized to β-actin.
Cells fixed in 4% paraformaldehyde were incubated with primary antibodies and detected using appropriate flourochrome-labeled IgG conjugates. 4′,6-diamidino-2-phenylindole (DAPI) was used as a nuclear counterstain. Images acquired using a Leica TCS-SP-2UV laser scanning confocal microscope (×63 magnification).
Infliximab treatment of patients
We performed a pilot, uncontrolled proof-of-concept therapeutic intervention of Infliximab therapy in five patients with progressive BOS. All patients met standard criteria for BOS demonstrating progressive loss in forced expiratory volume in 1 second (FEV1) and exercise capacity with the exclusion of other causes on bronchoscopy, transbronchial biopsy and imaging. The proof-of-concept clinical intervention was intended to offer patients with progressive BOS despite maximal therapy the opportunity to receive an additional antiinflammatory therapy and was therefore not a registered clinical trial. Patients were approached to give written informed consent before receiving therapy and were chosen on the basis of demonstrating progressive deterioration in lung function unresponsive to a switch to tacrolimus therapy from cyclosporine, a trial of azithromycin (250 mg three times per day), a lack of regular overt acute lower respiratory infections and a negative quantiferon test. Three of the five patients had previously received total lymphoid irradiation without benefit. All patients were treated with infliximab (3 mg/kg intravenously), repeated at least at 2 and 6 weeks. FEV1 and 6-min walk distance were carried out before and between 2 and 4 weeks following the third dose of Infliximab. All patients have been followed for at least 18 months.
The response of PBEC to a range of treatments was assessed and compared to untreated controls using Student's t-test. Differences in cytokine concentrations between groups of BAL were assessed using a Mann–Whitney U-test. A p-value of < 0.05 was considered statistically significant.
Inflammatory cytokines are elevated in the BAL of lung transplant recipients around the time of diagnosis of BOS
TNFα, IL-8 and IL-1β concentrations were assessed in BAL samples collected longitudinally from lung transplant recipients who developed BOS and a control group of lung transplant recipients who did not develop BOS (table 1). TNFα, IL-8 and IL-1β concentrations were significantly elevated in BAL samples acquired < 3 months prior to BOS diagnosis compared to samples acquired >3 months prior to BOS diagnosis or samples from stable lung transplant recipients (Figure 1A and B). Additionally, the total cell number was increased primarily due to an increase in neutrophils as no significant difference in the number of macrophages, lymphocytes or eosinophils was observed (Figure 1A and C). The percentage of BAL returned was lower in samples acquired < 3 months prior to BOS diagnosis (Figure 1A) as previously described . The 3-month time point was chosen as it represents a biologically plausible timeframe in which cellular events that contribute to BOS development might be detectable.
Table 1. Patient demographics of the study population. BAL samples were collected longitudinally from lung transplant recipients (n = 26) who developed BOS and a control group of lung transplant recipients (n = 26) who did not develop BOS
When inflammatory cytokine concentration was compared to time from BOS diagnosis we saw a trend of increased TNFα, IL-8 and IL-1β concentration at 3 months prior to BOS diagnosis (Figure S1A). This peak in inflammatory cytokines was associated with a small increase in total cell number and lymphocyte number but no significant change in neutrophil or macrophage numbers was seen, although there was a significant increase in neutrophils at the time of BOS diagnosis (Figure S1B).
Our group and others have previously shown that de novo acquisition of Pseudomonas aeruginosa in the transplanted airway is associated with an increased risk of developing BOS [21-23]. Patients who are culture positive (73% were positive for Pseudomonas aeruginosa) have a significantly elevated BAL concentration of TNFα, IL-8 and IL-1β compared to culture-negative patients (Figure S2A and B). The total cell number in BAL was also significantly increased although no significant difference in percentage of BAL returned was observed. The increase in cell numbers was primarily due to an increased neutrophilia as no significant difference in macrophage, lymphocyte and eosinophil numbers was observed (Figure S2A and C).
Macrophage differentiation and cytokine secretion
Our longitudinal data demonstrated a peak in inflammatory cytokines in BAL samples acquired 3 months prior to BOS diagnosis; however no change in the total macrophage number was observed. As macrophages demonstrate remarkable plasticity and change their phenotype in response to the microenvironment  we proceeded to investigate if polarization of macrophage phenotype could be a possible explanation for the increased inflammatory cytokines in BAL prior to BOS diagnosis.
The THP-1 monocytic cell line was differentiated into CAMφ and AAMφ as previously described (Figure 2A) . Treatment of THP-1 cells with PMA + IFNγ generates an adherent population of cells that are uniformly positive for the CAMφ marker iNOS and negative for the AAMφ marker Arginase-1 (Figure 2Bi and C). In contrast, treatment with PMA + IL-4/IL-13 generates a population of adherent cells that are uniformly positive for Arginase-1 and negative for iNOS (Figure 2Bi and C).
CAMφ can be induced by treatment with IFNγ alone or with IFNγ/TNFα. Furthermore, AAMφ can be induced by treatment with IL-4, IL-13 or IL-4/IL-13 . We therefore differentiated THP-1 cells to all CAMφ and AAMφ subsets and investigate the secretion of proinflammatory cytokines in response to stimulation with Pseudomonas aeruginosa. There was a significant increase in cytokine secretion in both CAMφ subsets compared to all AAMφ subsets (p < 0.05, n = 3). However, no significant difference in cytokine secretion was seen within the CAMφ or AAMφ subsets (p > 0.05, n = 3). Consequently, CAMφ differentiated with IFNγ and AAMφ differentiated with IL-4/IL-13 were used in all subsequent experiments (Figure 2D). Interestingly, when we then assessed cytokine secretion over a 48-h time course we found that IL-8 (and IL-1β) secretion increased up to 48 h in both CAMφ and AAMφ; however, TNFα secretion reached a plateaux at 6 h and remained elevated up to 48 h (Figure 2E).
AM isolated from the BAL of lung transplant patients were differentiated to CAMφ (IFNγ) or AAMφ (IL-4/IL-13) and the secretion of proinflammatory cytokines assessed. In agreement with the THP-1 data we found that AM derived CAMφ secreted significantly more TNFα and IL-8 in response to stimulation with Pseudomonas aeruginosa compared to AM derived AAMφ (p < 0.05, n = 5). Significantly the amount of cytokine secreted from CAMφ and AAMφ was similar between those derived from THP-1 cells and AM confirming the suitability of THP-1 cells in our model (Figure 2F).
Investigating the effect of CAMφ and AAMφ on EMT
Several groups have published convincing data suggesting an important role for EMT in the pathogenesis of BOS [5-7]. Furthermore, previous data from our group have shown that lipopolysaccharide activated monocytic cells, via the secretion of soluble factors, can accentuate TGF-β1-driven EMT . Cultures of PBEC were established from six (4M/2F) lung transplant recipients between 7 and 57 months posttransplant. No subjects showed any evidence of acute rejection (grade A2 or above by ISHLT classification), or infection and were BOS stage 0 (>90% baseline FEV1) at the time of sampling.
Conditioned media from CAMφ or AAMφ at baseline or stimulated with Pseudomonas aeruginosa did not drive EMT in PBEC (p > 0.05, n = 6). TGF-β1 is shown as a positive control for induction of EMT, suggesting that CAMφ or AAMφ are not secreting sufficient TGF-β1 to drive EMT (Figure 3A, B and E). We measured active TGF-β in conditioned media and demonstrated that CAMφ secrete < 1 ng/mL of TGF-β1 and it was not significantly increased by stimulation with Pseudomonas aeruginosa (p > 0.05, n = 3) (Figure3C).
We therefore investigated the ability of CAMφ and AAMφ to accentuate TGF-β1 driven EMT. Conditioned media from unstimulated CAMφ and AAMφ had no significant effect on TGF-β1-driven EMT (p > 0.05, n = 6). However conditioned media from Pseudomonas aeruginosa stimulated CAMφ significantly accentuated the downregulation in cytokeratin-19 and E-cadherin (61% ± 6, p < 0.01) expression and the increase in fibronectin (169% ± 41, p < 0.01) and vimentin (357% ± 147, p < 0.05) expression compared to treatment with TGF-β1 alone. Conditioned media from Pseudomonas aeruginosa stimulated AAMφ also appears to accentuate the change in EMT marker expression (E-cadherin 34% ± 8, p = 0.09; fibronectin 53% ± 40%, p = 0.24; vimentin 218% ± 104, p = 0.09); however the results failed to reach significance and the accentuating effect was much less than seen with CAMφ (Figure 3D and E, Figure S3).
Investigating the contribution of TNFα and IL-1β to the inflammatory accentuation of EMT
Our previous work identified a hierarchy of proinflammatory cytokines capable of accentuating TGF-β1-driven EMT, with TNFα and IL-1β being the most potent and IL-8 having no accentuative effect . We therefore blocked TNFα and IL-1β in the conditioned media from Pseudomonas aeruginosa activated macrophages and investigated the effects on TGF-β1-driven EMT. Significantly, a TNFα neutralizing antibody inhibited (E-cadherin 44% ± 5; fibronectin 71% ± 10; vimentin 60% ± 9, all p < 0.05) the accentuating effect of the conditioned media on TGF-β1-driven EMT (Figure 4A and B). Similar results were achieved using Infliximab, a monoclonal TNFα antibody used in clinical practice to treat rheumatoid arthritis (Figure S4). In contrast the IL-1β neutralizing antibody had no effect on EMT marker expression (p > 0.05, n = 6) (Figure 4C and D), suggesting that the majority of the accentuative effect occurs via the secretion of TNFα.
TNFα as a therapeutic target in lung transplant recipients with BOS
We proceeded to test the potential role of TNFα as a therapeutic target in BOS by performing a pilot, uncontrolled proof-of-concept open therapeutic intervention of Infliximab therapy in five patients with progressive BOS. All patients met standard criteria for BOS  including assessment by transbronchial lung biopsy but were not subjected to video-assisted thoracoscopic lung biopsy. None had shown a sustained response following treatment with azithromycin, total lymphoid irradiation or a switch from ciclosporine to tacrolimus and all demonstrated progressive loss in FEV1 and reduced exercise capacity. All patients were treated with Infliximab (3 mg/kg intravenously) at a minimum of 0, 2 and 6 weeks with three patients continuing to receive infliximab at 8 weekly intervals for 6 months and one patient receiving infliximab at 8 weekly intervals for 10 months. FEV1 and 6-min walk distance were carried out before treatment and again between 2 and 4 weeks following the third dose of Infliximab. Four patients demonstrated an absolute improvement in both FEV1 and 6-min walk distances (Table 2). The fifth patient, who had experienced an extremely rapid decline in lung function, stabilized. Although not formally assessed, three of the patients spontaneously reported an improvement in wellbeing following Infliximab therapy. There were no acute adverse events associated with the Infliximab therapy and four of the patients have experienced no medium term complications despite being colonized with Pseudomonas aeruginosa prior to treatment. All patients were on maintenance nebulized Colisitin therapy before and during the therapeutic intervention. One patient who had previously isolated Aspergillus fumigatus in sputum a year prior to Infliximab developed recurrent Pseudomonal infections 2 months after the third dose and developed a cavity in the lung. Repeated investigations for Mycobacterium tuberculosis were negative although Aspergillus fumigatus was cultured and the patient was stabilized on antifungal therapy. No evidence of systemic bacteremia was observed. This patient is living independently at home 16 months later and their FEV1 remains 0.2 L over the pretreatment level. No other infectious complications were reported in the other patients. All patients have now been followed up for at least 18 months and demonstrated stable lung function over this time period. The first patient treated has a longer follow-up and remained stable for 24 months before developing a further decline in lung function leading to death as a result of respiratory failure 3 years following the initial infliximab therapy.
Table 2. Infliximab improves lung function and exercise capacity in patients with BOS. Patient demographics of the study population. Assessment of forced expiatory volume in 1 second (FEV 1) and 6 minute walk distance (MWD) in our study population pre- and postinfliximab treatment. Four of five patients showed an increase in both FEV 1 and 6 MWD following infliximab treatment
AZA = azathioprine; PR = prednisolone; TAC = tacrolimus; MMF = mycophenolate mofetil; AZI = azithromycin; TLI = total lymphoid irradiation; CY = cyclosporine; PA = Pseudomonas aeruginosa; AF = Aspergillus fumigatus.
Therapies to limit or reverse endobronchial fibrosis in the transplanted lung have thus far proved unsuccessful highlighting the need for a better understanding of the mechanisms driving fibrosis and in particular the link between fibrosis and inflammation. This manuscript aims to challenge the paradigm that the myo-fibroblast is the only cell contributing to fibrosis by significantly increasing our understanding of the role of macrophages in the fibrotic response to injury in the lung.
It has been demonstrated that there is an early elevation in Th1-cytokines in lung transplant recipients who develop BOS compared to stable recipients . Furthermore, CD4+ T cells isolated from patients who developed BOS are of a Th1-phenotype  suggesting that a skewed Th-1 immune response in the lung allograft predisposes to the development of BOS. In such a Th1 cytokine dominated environment it is likely that macrophages will favor a CAMφ phenotype, a hypothesis backed up by the observation that airway macrophages from posttransplant patients with histological evidence of acute rejection express increased levels of proinflammatory cytokines, including TNFα, compared to patients without rejection .
In this study we investigate the ability of Pseudomonas aeruginosa stimulated CAMφ and AAMφ to drive/accentuate EMT in PBEC isolated from lung transplant recipients. Our data demonstrates that Pseudomonas aeruginosa stimulated CAMφ strongly accentuates TGF-β1-driven EMT. Interestingly, CAMφ do not accentuate TGF-β1 driven EMT at baseline, confirming that stimulation of the CAMφ with one or more of the pathogen associated molecular patterns within lysates of Pseudomonas aeruginosa is required to elicit an effect. These in vitro findings may help explain the clinical observation that de novo acquisition of Pseudomonas aeruginosa in the transplanted airway is associated with an increased risk of developing BOS [21-23]. Furthermore, CAMφ are unable to drive EMT in the absence of TGF-β1 suggesting a requirement for TGF-β1 in our model. TGF-β1 has been shown to be critically important in the development of fibrosis in all compartments of the lung and is present in elevated levels in patients with BOS, IPF and COPD [3, 27, 28] suggesting that the mechanism described in this study may be relevant in the development of several other lung diseases.
Previously our group and others have shown that the neo-macrolide azithromycin given at sub-Minimum inhibitory concentration for respiratory pathogens can reverse the decline in lung function in some patients with BOS [29, 30]. This lead to a randomized placebo controlled study demonstrating that patients receiving azithromycin after lung transplantation had a lower incidence of BOS in their first 2 years posttransplantation . Neo-macrolides are a group of antibiotics that are bacteriostatic and only bactericidal at high concentrations. Independently of their antimicrobial activity, macrolides possess immunomodulatory properties that may contribute to clinical benefits observed in patients with chronic airway inflammation. The mechanism of action was initially believed to be through a reduction in airway neutrophilia and IL-8 . However, macrolides can also penetrate macrophages making them more accessible at the site of inflammation; azithromycin concentration in macrophages reaches up to 110 times the external concentration . Recently, azithromycin has been shown to modulate inflammation by shifting macrophage polarization toward an AAMφ phenotype identifying another possible mechanism of action . However the mechanism of action is likely to be indirect as azithromycin did not directly downregulate TNFα release from CAMφ in response to LPS stimulation (see Figure S5).
Longitudinal analysis of BAL samples from lung transplant recipients who develop BOS demonstrated a significant elevation in TNFα, IL-1β and IL-8 BAL levels 3 months prior to BOS diagnosis. Previous work by our group and others has shown that TNFα and IL-1β, but not IL-8, can accentuate TGF-β1-driven EMT [6, 8, 35]. Significantly blocking TNFα, but not IL-1β, in conditioned media from Pseudomonas aeruginosa activated macrophages significantly inhibits the accentuation of TGF-β1-driven EMT suggesting that the majority of the accentuative effect is elicited through TNFα. Interestingly, the development of obliterative bronchiolitis (OB) in a mouse model is associated with a significant elevation in TNFα levels  and neutralizing antibodies to TNFα prevents the development of . Similar beneficial effects of neutralizing TNFα have been reported in obliterative airway disease in rat tracheal allografts  and in a heterotopic porcine bronchial transplantation model .
Double-blind randomized studies investigating the treatment of established BOS are extremely limited and urgently required. Current reported uncontrolled studies have investigated the role of changing immunosuppression including switches from cyclosporine to tacrolimus , azathioprine to mycophenolate  or the use of methotrexate , cyclophosphamide , mammalian inhibitors of rapamycin (m-Tor) inhibitors , photophoresis  and total lymphoid irradiation . In general, most studies demonstrate reduction or arrest in loss in FEV 1 in some patients. However, there have been no reported trials of using biological agents targeting TNFα in BOS postlung transplantation, although there is a single case report indicating improvement in wellbeing and FEV1 after Infliximab therapy in a child who developed BOS following bone marrow transplantation . Additionally, treatment with a combination of corticosteroids and Etanercept, a decoy receptor that binds to and inhibits the activity of TNFα, has been shown to improve lung function in patients with idiopathic pneumonia syndrome .
Accordingly we carried out a pilot, uncontrolled proof-of-concept therapeutic intervention of Infliximab therapy in five patients with progressive BOS following lung transplantation. The data demonstrated a rapid and sustained elevation in FEV 1 and an increased 6-min walk distance following Infliximab treatment in four patients and stabilization of disease in the other. The magnitude of the changes in FEV1 and 6-min walk distance would be considered clinically significant but due to the small sample size were not statistically assessed in this small proof-of-concept therapeutic intervention. Follow-up of the patients demonstrated stable lung function in all up to 18 months following infliximab therapy although the first patient treated deteriorated after 24 months and died 36 months following initial treatment from respiratory failure. All other patients remain alive between 18 and 30 months following initial therapy. We fully recognize that the experience reported here does not constitute a formal clinical trial but we believe that the data support the potential role of TNFα as an important proinflammatory molecule that contributes to airway remodelling in BOS and identifies biological agents targeting TNFα as a therapeutic option worthy of careful evaluation in a randomized controlled clinical trial in patients with progressive BOS.
However we acknowledge that the mode of action of Infliximab in the five patients reported here may be multifactorial and not limited exclusively to the novel mechanism identified in this study. For example TNFα has also been shown to stimulate proliferation , chemotaxis  and TGF-β1 gene transcription  in fibroblasts. Inhibition of these critical mechanisms may also contribute to the beneficial effect of Infliximab on patients with progressive BOS. However, regardless of the precise mechanism of action, we believe this to be encouraging data and suggest that a randomized blinded study is now warranted.
This work was supported by a research grant from the Medical Research Council UK (G0700861). LAB is supported by a Marie Curie fellowship. ADS is supported by a HEFCE Senior Lectureship. DAM is supported by the Wellcome Trust (WT086755MA). AJF is supported by a GlaxoSmithKline clinical fellowship award.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.