INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In 1984 Burke and coworkers reported a distinctive clinical syndrome of progressive dyspnea and airflow obstruction in five of 14 long-term survivors of heart-lung transplantation, which he called obliterative bronchiolitis (OB) (1). OB is considered to be analogous to late graft loss after transplantation of other organs, consisting of a fibroproliferative process leading to obliteration of tubular structures in the organ, i.e., airways in the lung (2). The cumulative risk of OB may reach 60 to 80% between 5 and 10 yr after transplantation (2, 3). OB occurs unpredictably, is currently undetectable in a preclinical state, and in general treatment is unsuccessful. Therefore OB is the biggest threat to health and long-term survival of lung and heart-lung transplant recipients. Despite a classic clinical presentation and physiological deterioration, transbronchial biopsies (TBB) may remain negative for the classic histologic appearance.

OB is thought to result from an alloimmune mechanism involving chronic presentation of allograft antigens to recipient T cells (2, 3). Dendritic cells (DC), the most potent antigen presenting cells (APC), present allograft antigens to T cells in conjunction with major histocompatibility complex (MHC) Class II molecules. DC strongly express Class II MHC molecules. For optimal T-cell activation costimulatory pathway activation is necessary via CD80 and CD86 molecules on the APC interacting with CD28 on T cells (4). In the context of OB, bronchial inflammation develops in parallel to disease in the small airways, perhaps representing alloresponsive inflammation triggered by bronchial epithelial expression of donor-specific MHC molecules (2).

Yousem and coworkers showed that increased DC were present in chronically rejecting lung allografts in 1990 (5). However, despite a potential important role for DC in the development of OB, there are few studies on the role of DC in human lung allografts since Yousem's work (5). Recently, Milne and coworkers looked at mononuclear phagocyte populations in the transplanted human lung, using unused donor lungs, open lung biopsies from transplanted lungs, and TBB from transplanted lungs (6). They found a marked depletion of CD1a⁺ DC in lung allografts compared with unused donor lungs. CD1a defines potent APC of DC phenotype and a meshlike network of CD1a⁺ DC exists in normal airway epithelium (7).

Milne and coworkers explain that their data seem to contradict Yousem's 1990 study, which showed increased numbers of DC in chronically rejecting lung allografts, and argue that this may be a failure of migration of CD1a⁺ DC (6). But they admit their finding may also reflect a decreased expression of the CD1a marker, perhaps owing to immunosuppressive medication. We propose that the latter explanation is most plausible. Even though Milne and coworkers found minimal numbers of DC in transplant patients, they did not look at endobronchial biopsies (EBB); and further, the DC marker used was CD1a which is expressed on only 30% of DC, and other phenotypes of DC are found in the lung (8, 9). Also, Yousem and coworkers used a different marker of DC, the S-100 monoclonal antibody (5).

Cells of the DC/macrophage lineage are strikingly heterogeneous with some populations with phagocytic and antigen presenting functions and others with the capacity to suppress T-cell function (10). The monoclonal antibodies (mAb) RFD1 and RFD7 define three distinct macrophage/APC populations (10). RFD1 is a mAb that sees an epitope within the MHC Class II complex with restricted expression to cells known to be potent stimulators of T cells, i.e., DC. The RFD7 mAb recognizes a predominantly cytoplasmic antigen that is present in mature tissue phagocytes but absent on inductive APC such as veiled cells, Langerhans cells, and interdigitating cells. Double-positive (RFD1⁺RFD7⁺) macrophages have been shown in vitro to suppress T-cell proliferation and have been referred to as "suppressor" macrophages; these suppressor macrophages have been shown to account for 46% of the resident macrophage population in airways of healthy subjects (10).

We hypothesized that, despite Milne's findings, DC are important in lung allograft chronic rejection. The aims of the present study were to investigate the role of lung macrophages in chronic rejection of lung allografts by (1) staining lung allograft biopsies for DC using not just CD1a, but also RFD1 and MHC Class II staining to decrease the chances of missing any DC as a result of low levels or absent expression of any one DC marker; (2) looking at EBB as well as TBB, as TBB may not be the best site for sampling early inflammatory changes associated with OB as discussed previously; (3) looking for "suppressor" macrophages using RFD1 and RFD7 monoclonal antibodies; and (4) staining for the costimulatory molecules CD80 and CD86 and then assessing for correlation with clinical course.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

All patients gave informed consent on a protocol consent form approved by the Stanford University Medical Center institutional review board. All lung or heart-lung transplant patients in the Stanford University Medical Center Lung and Heart-Lung Transplantation Program undergoing routine surveillance bronchoscopy or bronchoscopy indicated for clinical suspicion of infection or rejection were requested to enter the study. This study was prospective and involved 22 patients who were status post single-lung, double-lung, or heart- lung transplantation at Stanford University Medical Center. Table 1 summarizes patients' demographic details. All patients had induction immunosuppression with rabbit anti-thymocyte globulin (RATG) postoperatively, followed by a triple immunosuppression regimen of cyclosporin A or tacrolimus, azathioprine or mycophenolate mofetil, and prednisone.

Bronchoscopy and Biopsy

Bronchoscopy was performed using an Olympus bronchoscope (Olympus America, Inc., Long Beach, CA). Sedation was achieved with intravenous morphine and midazolam, topical airway anesthesia with 2% lidocaine, and a nasal route was used where possible. Each bronchoscopy was performed in a standard fashion, first a bronchoalveolar lavage being taken from right middle lobe or lingula. Then 2 to 3 EBB were taken from the transplanted lung, if right lung from the take-off of the right upper lobe, using an Olympus fenestrated spike forceps (FB-34C-1; Olympus America, Inc., Long Beach, CA), and the samples flash frozen in optimal cutting temperature (OCT) compound using liquid nitrogen. Then TBB, usually 10 in number were taken from different subsegments of the transplanted lung under fluoroscopy guidance, the majority being used for histologic diagnostic purposes and 2 to 3 being flash frozen in liquid nitrogen for subsequent DC analysis. Sections for DC staining were stored in liquid nitrogen at 70° C until being cut into 7-µm sections using a Leica Cryostat (Leica Microsystems Inc., Deerfield, IL).

Immunofluorescent Staining

Cryostat sections (7 µm) were thawed to room temperature, ringed with a hydrophobic marker, washed with phosphate-buffered saline (PBS), and then incubated with 5% fetal calf serum (FCS) and 2% bovine serum albumin (BSA) for 1 h. The sections were washed 3 times in PBS and then incubated with either anti-CD80 (mouse IgM, 5 µg/ ml; Pharmingen, San Diego, CA) or anti-CD86 (mouse IgG2b, 5 µg/ ml; Pharmingen) mAbs, or the appropriate isotype control immunoglobulin for 1 h at 4° C. Then the sections were washed in PBS and consecutively incubated with Texas Red (TR)-labeled anti-mouse IgG (Caltag Laboratories, Burlingame, CA), or TR-labeled anti-mouse IgM (Cortex Biochem, San Leandro, CA), for 1 h. After the sections were washed three times, they were incubated for 1 h with 2.5 µ/ml of the fluorescein isothiocyanate (FITC)-labeled anti-CD1a mAb (IgG2b; Research Diagnostics, Inc., NJ) or FITC-labeled mouse IgG (DAKO A/S, Glostrup, Denmark) (negative control). Sections were then washed three times in PBS, a drop of 2.5% diazabicyclooctane (DABCO) applied and then a coverslip was placed over each section. Using a fluorescent microscope, CD1a-positive DC were counted per high-power field (HPF) (×40) and the percentage of CD1a⁺ cells expressing CD80 or CD86 quantified. A minimum of 10 HPF were counted per patient.

For quantifying the percentage of MHC Class II DC expressing CD80 or CD86, the technique was exactly the same as previously described except that an anti-MHC Class II antibody conjugated to FITC (IgG1; DAKO A/S) was used in the same way that anti-CD1a conjugated to FITC was used above.

For RFD1 and RFD7 antibodies the procedure was similar. Sections were thawed to room temperature, ringed with a hydrophobic marker, washed in PBS, then incubated for 1 h with the primary layer antibodies to RFD1 (mouse IgM) and RFD7 (mouse IgG1) (Serotec, Raleigh, NC), both antibodies at a 1:250 dilution. The sections were washed three times in PBS. Then sections were incubated for 1 h with two immunoglobulin class-specific second layer reagents, conjugated respectively to FITC (goat anti-mouse IgM-FITC; Cortex Biochem, San Leandro, CA) and TR (goat anti-mouse IgG-TR; Caltag Laboratories, Burlingame, CA). Finally the sections were washed three times in PBS, a drop of 2.5% DABCO applied and then a coverslip was placed over each section. The relative proportions of RFD1⁺ stimulatory cells, RFD7⁺ phagocytes, and RFD1⁺RFD7⁺ "suppressor" cells were then counted per HPF (×40).

Immunoperoxidase Staining

Sections were thawed to room temperature, ringed with a hydrophobic marker, washed with PBS, then incubated with 5% FCS and 2% BSA for 1 h. The sections were then washed three times in PBS and incubated for 1 h either with a mAb mix against human leukocyte-associated antigen-DP (HLA-DP), -DQ, -DR (mouse IgG, clone CR3/43; DAKO A/S) or a nonspecific control mouse IgG (DAKO A/S). Sections were then washed in PBS and incubated with a peroxidase-coated goat anti-mouse immunoglobulin (DAKO A/S). After 1 h sections were washed in PBS, and incubated with the development solution which contained 50 mM Tris, pH 7.5, diaminobenzidine, and 30% hydrogen peroxide. After 5 min sections were washed in PBS, covered with a drop of glycerol and a coverslip. DC were counted as numbers of high Class II MHC-staining cells with dendritic morphology/HPF (Figure 1).

View larger version (109K): [in this window] [in a new window]   Figure 1.   Immunoperoxidase staining of dendritic cells using Class II MHC expression. Immunoperoxidase staining of airway dendritic cells by Class II MHC expression. Sections stained with control monoclonal antibody were negative except for very rare cells displaying endogenous peroxidase activity.

Statistical Methods

As data were distributed in a non-Gaussian fashion, nonparametric analysis was used (14). The Mann-Whitney U test was used for comparing independent groups. When comparing paired samples, the Wilcoxon matched pairs test was used. Correlations were tested using Spearman's method. Computer software used was STATISTICA (StatSoft, Tulsa, OK).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For the 22 patients we had a total of 38 bronchoscopies (EBB and TBB at each time point where possible). Of 22 patients, eight met the criteria for chronic rejection, two with biopsy-proven OB and six with bronchiolitis obliterans syndrome (BOS). The remaining 14 patients had stable pulmonary function (five were within 90 d of their transplant and nine were more than 6 mo post-transplant).

Dendritic Cells as Defined by CD1a Expression

When numbers of CD1a⁺ DC/HPF in biopsies (TBB and EBB) of the eight OB/BOS patients (median 0.25, range 0 to 4.8) were compared with the 14 stable patients (median 0, range 0 to 2.4), there were significantly more CD1a⁺ cells in the OB/BOS group, p < 0.02 (Figure 2A).

View larger version (29K): [in this window] [in a new window]   Figure 2.   Dendritic cells in lung allografts. (A) CD1a+ dendritic cell numbers. (B) MHC Class II dendritic cell numbers. Dendritic cell numbers per HPF as defined by CD1a expression (Figure 1A) and MHC Class II expression (Figure 1B) in 22 patients, including two with OB, six with BOS, and 14 patients with stable lung function.

The presence of acute rejection (AR) at the procedure time (7 episodes of Grade A2 or greater in our 22 patients) was not associated with increased CD1a⁺ DC numbers in either TBB or EBB compared with those biopsies with no evidence of rejection (Mann-Whitney U test, p = 0.4). We also tested whether the occurrence of AR episodes before the study bronchoscopies (9 episodes of Grade A2 or greater) correlated with CD1a⁺ DC numbers and found no correlation (Spearman r = 0.04, p = 0.7). The presence of cytomegalovirus (CMV) pneumonitis at the procedure time (3 episodes) or prior episodes of CMV pneumonitis (4 episodes) also did not correlate with DC numbers seen in biopsies (Spearman r = 0.15, p = 0.24).

Dendritic Cells Expressed as High MHC Class II Staining with Dendritic Morphology

We saw significantly greater numbers of DC/HPF when we defined them by high Class II MHC expression and dendritic morphology (mean 3.94, median 3, range 0 to 17) compared with DC as defined by CD1a positivity (mean 0.52, median 0, range 0 to 4.8), p < 0.000001 (Figure 2B).

When numbers of Class II MHC DC in the eight patients with BOS/OB (median 6.4, range 0 to 11.1, n = 8) are compared with stable patients (mean 2.8, median 2.75, range 0 to 6.7, n = 14), the difference is statistically significant (p = 0.042).

There was no correlation between numbers of Class II MHC DC in biopsies and presence of AR at the time of the biopsy (7 episodes of Grade A2 or greater in 22 patients) or number of prior episodes of AR (9 episodes of Grade A2 or greater) (Spearman r = 0.07, p > 0.5). There was no correlation between the presence of CMV pneumonitis at the time of biopsy (3 episodes) or between the occurrence of prior episodes of CMV pneumonitis (4 episodes) and numbers of Class II MHC-high DC found on biopsy.

Costimulatory Molecule Expression

Taking all the biopsies, 50% of CD1a⁺ cells expressed CD86 and 20% of CD1a⁺ cells expressed CD80. When comparing the percentage of CD1a⁺ cells expressing CD86 in the eight patients with BOS (median 32.5, range 0 to 100) with the other 14 stable patients (median 50, range 0 to 100), there was no significant difference, p > 0.5. A comparison of the percentage of CD1a⁺ cells expressing CD80 in eight BOS/OB patients (median 8.25, range 0 to 100) with the 14 stable patients (median 0, range 0 to 50) showed a trend to increased CD80 staining in chronic rejection patients but this did not reach statistical significance (p > 0.05).

Similar percentages of MHC Class II DC expressed CD80 and CD86, and as with CD1a⁺ DC there was no difference in costimulatory molecule expression between the eight patients with chronic rejection and the 14 stable patients.

Comparison of Dendritic Cell Numbers in Endobronchial and Transbronchial Biopsies

Comparing EBB with TBB (Figure 3) showed that the number of CD1a⁺ DC/HPF in EBB (median 0.2, range 0 to 4.8) tended to be greater than that in TBB (median 0, range 0 to 2.9) but this difference did not reach statistical significance (p = 0.4). Nine patients had greater numbers of CD1a⁺ cells in their EBB than their TBB, whereas four patients had greater numbers of CD1a⁺ cells in their TBB than their EBB. There were significantly more MHC Class II DC in EBB (median 4.5, range 0.25 to 17) compared with TBB (median 2, range 0 to 7), p < 0.003. 

View larger version (19K): [in this window] [in a new window]   Figure 3.   Dendritic cell numbers in endobronchial versus transbronchial biopsies. MHC Class II+ dendritic cell numbers per HPF in EBB and TBB. The horizontal bars represent median values. There were significantly more dendritic cells in EBB than TBB (p < 0.003).

Macrophage Phenotype Staining Using RFD1 and RFD7

The RFD1 staining correlated with CD1a staining in EBB and TBB (i.e., similar numbers of CD1a⁺ cells and RFD1⁺ cells in all the biopsies), with Wilcoxon's matched pairs test showing no difference, p = 0.8, Spearman correlation r = 0.57, p = 0.003. We saw minimal numbers of single RFD7-staining cells, and likewise for double RFD1⁺RFD7⁺ staining (< 0.1% of cells staining for RFD1 also stained for RFD7) in either TBB or EBB of the 22 patients. This was not a technical problem as we observed RFD7 staining in normal lung tissue (positive control).

Serial Analysis of Dendritic Cells in Lung Transplant Patients

To examine DC numbers over time in EBB and TBB we analyzed repeat biopsies in five of eight OB/BOS patients and nine of 14 stable patients (Figure 4). At the first biopsy the OB/BOS patients have significantly more MHC Class II DC in their EBB and TBB than the stable patients (p > 0.05). At the second bronchoscopy, this trend continued with the higher numbers of dendritic cells in the OB/BOS patients than in the stable patients but it fell short of statistical significance (p > 0.05). This may in part reflect that only five of eight OB/BOS patients and nine of 14 stable patients had follow-up biopsies for analysis.

View larger version (15K): [in this window] [in a new window]   Figure 4.   Serial dendritic cell numbers in EBB and TBB. Mean numbers of dendritic cells from endobronchial and transbronchial biopsies in patients with chronic rejection (OB/BOS, n = 8), and patients with stable lung function (Stable, n = 14) on initial study biopsy and on follow-up biopsy for five of eight OB/BOS patients and nine of 14 stable patients. *Significant difference, p < 0.05, between the OB/BOS patients and stable patients at the initial biopsy.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

These results warrant discussion in the context of the recent study by Milne and coworkers who showed a marked depletion of CD1a⁺ DC in lung allografts exposed to a standard immunosuppressive regime when compared with unused donor lungs, as discussed in detail in the introduction (6). Our results suggest that Milne and coworkers may have underestimated the number of DC by not using another marker of DC such as high Class II MHC expression. This is suggested by our data which show markedly greater numbers of DC when using Class II MHC expression and dendritic morphology.

The other reason for Milne and coworkers' finding of minimal DC may be a sampling problem in that endobronchial tissue was not sampled (6). It is well accepted that chronic rejection is a patchy process which spares respiratory bronchioles, alveolar ducts, alveoli, and the lung interstitium (2, 3). It is thought that inflammatory changes in the larger airways reflect events in the smaller airways during chronic rejection (2, 3, 15). TBB, as mentioned previously, is the standard way of defining acute and chronic lung rejection. Disadvantages have been noted by others, with large numbers of specimens required, complications in up to 10%, and occasional deaths (16). Others have noted that the yield of bronchiolar tissue from TBB is variable, and it has been shown that bronchiolar material was detected in only 78% of surveillance TBB procedures performed according to current guidelines, and these investigators argue that the small samples of bronchiolar tissue may not reflect airway inflammatory changes (18). We found markedly greater numbers of DC in EBB compared with TBB.

We propose, notwithstanding the downregulation in CD1a expression observed with immunosuppression, that DC are increased in chronic rejection and that analysis of DC numbers in lung allografts using EBB in addition to TBB may be helpful to document the emergence of chronic rejection. We found relatively small numbers of CD1a⁺ DC in our biopsies, and less in TBB than EBB. However, in the eight patients with chronic rejection there were significantly more DC than in stable patients, presumably a reflection of the upregulation of the expression of foreign allograft antigen in chronic rejection. There was variability in DC numbers with low numbers of DC seen in some patients with chronic rejection and relatively high numbers in some stable patients (Figure 2B). We expect variability in airway DC numbers and do not suggest that there is an absolute number of DC above which a patient must have chronic rejection, but rather that for any given individual changes in their DC numbers may predict the development of OB. In Figure 2B there are two BOS patients with relatively small numbers of DC. These two BOS patients had early BOS compared with the other chronic rejection patients. Also in Figure 2B there were three stable patients with > 4.0 MHC Class II DC/HPF and all three of these had prior episodes of CMV pneumonitis. It has been shown that CMV infection increases MHC Class II expression and this is one possible explanation. It is also possible that over time these patients will go on to develop chronic rejection. It is worth reiterating that in the overall analysis CMV pneumonitis, either current or prior, did not correlate with DC numbers in biopsies.

A proportion of DC that were present in biopsies expressed costimulatory molecules, albeit CD86 more than CD80, and thus are capable of optimal T-cell stimulation. Costimulatory molecules have been extensively looked at in vitro and in animal models of transplantation but we are not aware of previous reports of CD80 or CD86 expression in human lung allografts. The interaction of CD86 on APC with CD28 on T cells has been shown to have a number of effects both in vitro and in animal models. For example, this costimulatory pathway regulates T-cell cytokine production, apoptosis, and proliferation (19, 20). Lu and coworkers showed that DC expressing large numbers of CD86 induce very low levels of T-cell apoptosis (21). When CD86 was blocked a marked enhancement of T-cell apoptosis was seen. In a mouse model of heart transplantation, administration of anti-CD86 and anti-CD80 mAb prolonged allograft survival (22). Anti-CD86 mAb suppressed intragraft interleukin-4 (IL-4), IL-10, IL-12 p40, and IL-15 messenger RNA (mRNA) expression. These investigators concluded that anti-CD86 and/or anti-CD80 mAb are potent immunosuppressants in prolonging allograft survival. Blockade of CD86, in comparison to CD80, had the greatest immunosuppressant effect (22). Our finding of increased DC in chronic lung allograft rejection and that 20% of CD1a⁺ cells express CD80 and 50% express CD86 supports a rationale for costimulatory blockade in human lung transplants to prolong allograft survival.

We found no difference in DC numbers with RFD1 staining, another DC marker, compared with CD1a staining. RFD7 expression by cells of the macrophage lineage seems to be downregulated by a standard immunosuppressive regimen. We found minimal numbers of cells expressing both RFD1 and RFD7, a phenotype that has been shown to have T-cell-suppressive capability (10). These "suppressor" macrophages have been shown to account for 46% of the resident macrophage population in airways of healthy subjects (13). Asthmatic airways, however, which are also subject to chronic T-cell-mediated inflammation, appear to have more of the potent antigen presenting/DC type of macrophages at the expense of "suppressor" macrophages (23). With treatment asthmatic airways show a decrease in the DC-type cells (23). It has also been shown that: (1) IL-10 is capable of pushing differentiating monocytes toward a "suppressor" phenotype; (2) asthmatics are deficient in IL-10 production and "suppressor" macrophages; (3) corticosteroid treatment increases IL-10 in asthmatics (26). It is possible that a similar process exists in lung transplants. Whether the lack of "suppressor" macrophages contributes to the lung rejection process is currently unknown.

Recent interest has focused on persistent donor-derived DC in organ transplants and their potential role in tolerance induction (29). It would be of interest to know whether the DC seen were of donor or recipient origin, but this was beyond the scope of this current study.

In summary, most studies that have looked at markers of the chronic rejection process have used TBB. We propose that serial changes in immunopathology on EBB are more sensitive than TBB, as the chronic rejection process is patchy and may spare more distal respiratory tissue. Our data also suggest that studies looking at DC in lung allografts should also not rely on CD1a as the sole marker of DC. MHC Class II staining should also be examined, otherwise DC numbers in allografts will be underestimated. We show that DC staining is most prominent in chronic rejection, and prospective study will continue to follow serial changes in DC numbers in lung allografts to determine if this is a useful early warning sign of chronic rejection. A proportion of DC found in lung allografts express costimulatory molecules of the B7 family, especially CD86, and therefore may be capable of inducing optimal T-cell activation. CD86 blockade may therefore be clinically useful in preventing lung allograft rejection. Also, a relative lack of "suppressor" macrophages may increase the lung allograft's susceptibility to the rejection process.