IL17-dependent autoimmunity to collagen type V (Col V) has been associated with lung transplant obliterative bronchiolitis. Unlike the T helper 1 (Th1)-dependent immune responses to Tetanus Toxoid (TT), the Th17 response to Col V in lung transplant patients and its Th1/17 variant observed in coronary artery disease patients requires IL-1β, tumor necrosis factor α and CD14+ cells. Given the involvement of the P2X7 receptor (P2X7R) in monocyte IL-1β responses, we investigated its role in Th17-, Th1/17- and Th1-mediated proinflammatory responses. Transfer of antigen-pulsed peripheral blood mononucleated cells (PBMCs) from Col V-reactive patients into SCID mouse footpads along with P2X7R antagonists revealed a selective inhibition of Col V-, but not TT-specific swelling responses. P2X7R inhibitors blocked IL-1β induction from monocytes, including both Col V-α1 peptide-induced (T-dependent), as well as native Col V-induced (T-independent) responses. Significantly higher P2X7R expression was found on CXCR3negCCR4+/6+ CD4+ [Th17] versus CXCR3+CCR4/6neg CD4+ [Th1] subsets in PBMCs, suggesting that the paradigm of selective dependence on P2X7R might extend beyond Col V autoimmunity. Indeed, P2X7R inhibitors suppressed not only anti-Col V, but also Th1/17-mediated alloimmunity, in a heart transplant patient without affecting anti-viral Epstein–Barr virus responses. These results suggest that agents targeting the P2X7R might effectively treat Th17-related transplant pathologies, while maintaining Th1-immunity to infection.
antigen presenting cells
coronary artery disease
cryopyrin-associated periodic syndrome
cardiac allograft vasculopathy
- Col V
collagen type V
hematoxylin and eosin
intracellular cytokine staining
median fluorescence intensity
peripheral blood mononucleated cell
phosphate buffered saline
tumor necrosis factor
trans vivo delayed-type hypersensitivity
The P2X7 receptor (P2X7R) has emerged as a potential site of regulation in a number of inflammatory states, including graft-versus-host disease , islet allograft rejection , chronic heart rejection , rheumatoid arthritis [4, 5] and psoriasis . Functioning as an adenosine triphosphate (ATP) gated ion channel [7, 8], the P2X7R allows cation passage through the cell, leading to downstream activation of inflammasomes and production of IL-1β, which is obligatory for T helper 17 (Th17) development [9, 10]. The loss of tolerance to the minor fibrillar collagen type V (Col V), a sequestered self-antigen , was a reported common characteristic of patients listed for lung or heart transplantation, such as patients suffering from idiopathic pulmonary fibrosis , late-stage coronary artery disease (CAD) [13, 14] and patients developing bronchiolitis obliterans syndrome after lung transplantation [13, 15]. Investigation into the loss of tolerance to Col V in these patient groups has revealed that the cellular immune response to Col V is Th17 mediated, as it was dependent on IL-17 and CD4 T cells, but also required IL-1β, tumor necrosis factor (TNF)α and monocytes . Similar to the reported IL-17-mediated response to soluble donor antigen (allo) in kidney transplant patients , the IL-17 requirement was also associated with IL-1β dependence . Unlike Col V, the cellular immune response to Tetanus Toxoid (TT) or Epstein–Barr virus (EBV), is Th1 mediated, as it depends on interferon (IFN)γ and not IL-17 [14, 16]. Knowing that the Th17-mediated Col V response in lung transplant patients differs from the cellular immune response to TT in the requirements for IL-1β and monocytes, we tested the hypothesis that P2X7R function (required for inflammasome activation and IL-1β production in certain innate immune contexts) was required for the Th17 cellular immune response to Col V. To test this hypothesis, we used pharmacological inhibitors of the P2X7R (Suramin, AZD9056 and periodate-oxidized ATP [oATP]) to evaluate P2X7R involvement in Col V-specific cellular immune responses.
Immunologic monitoring was performed on cryo-preserved blood samples from Col V-reactive patients with end-stage CAD (n = 3), lung pathology due to primary ciliary dyskinesia (n = 1), lung transplant (n = 1) or heart transplant (n = 1). Both transplant recipients were sampled 7–8 years posttransplantation, the lung patient at a time of normal graft function, the heart patient during an episode of acute rejection combined with cardiac allograft vasculopathy (CAV). Subject consent was obtained using human subjects committee-approved, written, informed consent procedures at the University of Wisconsin Hospital and Clinics. Blood was collected and peripheral blood mononucleated cells (PBMCs) were processed as described previously . Where applicable, human PBMCs were incubated with CD14 (120-001-146; Miltenyi Biotec, San Diego, CA) or pan T cell (120-008-788; Miltenyi Biotec) microbeads and separated using an autoMACS (Miltenyi Biotec) as per the manufacturer's protocol.
YOPRO uptake assay
The YOPRO dye uptake assay and the generation of transfected HEK293 cells expressing a human P2X7R with normal (P2X7-wt) or loss of function (P2X7-496) activity was described previously [18, 19]. Briefly, 5 × 104 transfected HEK cells were stimulated for 20 min with various concentrations of AZD9056 and Suramin in the presence of 100 μM 2′-3′-O-(4-benzoylbenzoyl) adenosine 5′triphosphate (BzATP, Sigma Chemical, St. Louis, MO) with 10 μM YO-PRO-1 (Invitrogen, Carlsbad, CA).
Trans vivo delayed-type hypersensitivity assay
The trans vivo delayed-type hypersensitivity (TV-DTH) assay was performed as described previously [14, 20]. Inactivated TT was purchased from Sanofi-Aventis Pasteur (Swiftwater, PA). Human Col V was obtained from D. Brand (University of Tennessee, Memphis, TN), and was prepared from placenta as described elsewhere . Human Col I was purchased from BD Pharmingen (San Jose, CA). The Col V-HLA-DR15-restricted α1 peptides, p1049 and p1439, were identified and synthesized as previously described . AZD9056 was obtained from Astrazeneca Pharmaceuticals (Wilmington, DE). Suramin and oATP were both purchased from Sigma Chemical. Neutralizing antibodies to IL-17 (16-7178-85; eBioscience, San Diego, CA), IL-1β (16-7018-85; eBioscience) or IFNγ (16-7318-85; eBioscience) were used along with normal rabbit IgG (AB-105-C; R&D Systems, Minneapolis, MN) where applicable.
Footpads from SCID mice were harvested and submitted for routine histology and immunohistochemical staining as described previously .
Intracellular cytokine staining
Intracellular cytokine staining (ICCS) experiments for IL-1β and TNFα were carried out as described previously . All samples were acquired on a FACScaliber flow cytometer and populations of IL-1β and/or TNFα positive cells among CD3−/CD14+ subsets were determined with FlowJo analysis software (Treestar, Ashland, OR).
P2X7R flow cytometry staining
PBMCs (1 × 106) from patients were stained for surface markers CD3 (344814; Biolegend, San Diego, CA), CD14 (325614; Biolegend), CXCR3 (550967; BD Biosciences, San Jose, CA), CCR4 (561034; BD Biosciences) and CCR6 (551773; BD Biosciences). Following surface staining, cells were fixed, permeabilized (Lyse/Fix Buffer; BD Biosciences) and subsequently incubated with 1 μg of P2X7R antibody (OASA05733; Aviva BioSystems, San Diego, CA) for 45 min on ice, followed by incubation with secondary antibody conjugated to Alexa488 (A11006; Invitrogen) for 30 min. Cells were washed three times and fixed in 2% paraformaldehyde. All samples were acquired on an LSR-II flow cytometer (BD Biosciences) and frequencies of T cell or monocyte populations positive for P2X7R, or corrected median fluorescence intensity (MFI) [MFI of sample signal − MFI of isotype control] of P2X7R signal were determined with FlowJo analysis software (Treestar).
Multiple replicates of each sample within sample groups were averaged across each assay and data were combined and analyzed by Mann–Whitney U and analysis of variance or paired t-test performed with Prism 5.0 (GraphPad Prism Software, La Jolla, CA).
The cellular immune response to Col V, but not TT, in Col V-reactive patients requires activation of the P2X7R
To examine the role of P2X7R in the cellular immune response to Col V and TT, we tested two pharmacological antagonists of P2X7R function, Suramin and AZD9056. We first found the maximum effective concentration of both P2X7R antagonists (Suramin [30 μM], and AZD9056 [300 nM]) at inhibiting P2X7R pore function (Figure 1). Next we analyzed TV-DTH responses to Col I, Col V and TT by PBMCs of six Col V-reactive patients in the presence or absence of Suramin or AZD9056 (Figure 2A). Col V elicited a significant increase in footpad swelling over phosphate buffered saline (PBS) and Col I controls, consistent with previous reports [13, 14]. Both Suramin and AZD9056 treatment of PBMCs significantly inhibited Col V-mediated TV-DTH responses. This effect was not due to toxicity, as Suramin and AZD9056 failed to inhibit the response to TT. Of note, similar levels of P2X7R inhibition of Col V responses were observed in all three groups of Col V-reactive patients examined (CAD, lung pathology, heart transplant) (Figure S1). Figure 2B illustrates the alteration in footpad histology of Col V versus TT-stimulated PBMCs in the presence of a P2X7R antagonist, Suramin. Hematoxylin and eosin (H&E) staining of Col V- or TT-stimulated, PBMC-injected, mouse footpads revealed a robust infiltration by mononuclear and polymorphonuclear cells extending into the skin and skeletal muscle. The Col V-stimulated proinflammatory response was dramatically decreased in the presence of both Suramin (Figure 2B) and AZD9056 (data not shown). Human cells were limited to a highly focused area in the sub-dermal fatty tissue, with few Ly6G+ mouse neutrophils present. In contrast, the TT response exhibited a diffuse infiltration pattern of CD45+ hematopoietic cells, human CD68+ macrophages and Ly6G+ mouse neutrophils both in the presence and in the absence of P2X7R inhibitors, This histological observation was consistent with the TV-DTH response we observed for each sample.
HLA-DR-restricted Col V peptide epitopes stimulate Col V-specific cellular immunity in a P2X7R-dependent manner
We recently identified two peptides, p1049 and p1439, located within the immunogenic α1(V) chain of the Col V molecule that bind to DR15 and induce a Th17 response in PBMCs of Col V-reactive DR15+ patients, and in a DR15neg recipient of a DR15+ lung transplant . We hypothesized that these short peptides, lacking α-helical structure or posttranslational modifications of the parent molecule, will stimulate Th17 cells solely as p/MHC complexes on antigen presenting cells (APCs) while circumventing any direct activation of the monocyte. To address whether or not P2X7R antagonists could inhibit TV-DTH responses stimulated by Col V DR-specific peptides, we tested PBMCs from three DR15+ Col V-reactive patients. Figure 3A shows that TV-DTH responses to Col V peptides p1049 and p1439 recapitulated that of whole Col V. Like the response to Col V, both p1049- and p1439-induced swelling was significantly inhibited by Suramin and AZD9056. In Figure 3B, representative H&E staining of footpads 24 h after injection of p1439- or p1049-stimulated PBMCs shows that both Suramin and AZD9056 cause a dramatic decrease in the DR-restricted peptide-induced swelling.
T cell–dependent and –independent production of IL-1β and TNFα by monocytes in response to Col V occurs in a P2X7R-sensitive manner
We have previously shown that Col V or Col V-α1 synthetic peptides can stimulate IL-1β production by monocytes cultured overnight in the presence of responder Th17 cells . Therefore, we examined whether P2X7R inhibition blocks the production of IL-1β or TNFα from monocytes in Col V- and α1-peptide-stimulated whole PBMCs, and as a control for possible T-independent effects, in cultures of isolated monocytes. Furthermore, since the Col V molecule containing numerous posttranslational modifications  could potentially activate the T cell as well as the APC for Col V, the monocyte, through damage-associated molecular patterns interactions, we investigated whether Col V and DR-restricted peptides could stimulate monocytes directly by separating monocytes from PBMCs and stimulating isolated monocytes. Figure 4A shows representative flow plots of Col V-, p1049- and p1439-stimulated PBMCs, in the presence or absence of Suramin or AZD9056. Overnight stimulation of PBMCs with Col V, p1049 or p1439 caused a significant increase in the percentage of IL-1β and TNFα from CD3−/CD14+ populations. Furthermore, the Col V-, p1049- and p1439-stimulated production of IL-1β (Figure 4B, top) and TNFα (Figure 4B, bottom) in PBMC overnight cultures was significantly inhibited by P2X7R antagonism. Unlike the Col V or peptide response, lipopolysaccharides (LPS)-stimulated IL-1β and TNFα production was unaffected by either Suramin or AZD9056 (Figure S2). To investigate whether Col V or DR-restricted peptides could directly stimulate isolated monocytes to produce IL-1β or TNFα, we performed ICCS assays on 2 × 105 MACS separated monocytes. In comparison to monocytes in whole PBMCs exposed to Col V (Figure 4C, black bars), in the absence of T cells, Col V stimulated a lower but significant increase in IL-1β and TNFα from isolated monocytes (gray bars, left and right), a response that was sensitive to P2X7R antagonism (Figure S3). Unlike the response of isolated monocytes to whole Col V, both p1049 and p1439 failed to increase IL-1β or TNFα production in isolated monocytes.
Inhibition of P2X7R function on both monocytes and T cells is required to block Th17 cellular immune responses in Col V-reactive patients
To examine whether P2X7R function on the monocyte or T cell defines the Col V response in our patient cohort, we used another P2X7R antagonist, oATP. oATP inhibits P2X7R function by covalently binding to the P2X7R [3, 24]. This inhibitor, unlike Suramin or AZD9056, irreversibly inhibits P2X7R function on treated cells alleviating concerns regarding potential drug wash-off or release. Isolated monocytes and T cells from Col V-reactive patients were loaded with oATP. Figure 5 shows that oATP-treated whole PBMCs exhibit a significant decrease in Col V-stimulated footpad swelling, similar to that observed with Suramin and AZD9056. When isolated monocytes were treated with oATP and added to untreated autologous T cells, there was no significant decrease in the Col V response. Similarly, when oATP-treated T cells were added to untreated autologous monocytes, there was no significant reduction in the Col V response. However, when both monocytes and T cells were separately treated with oATP, and added together and stimulated with Col V, there was a significant reduction in the Col V-stimulated swelling response. This result indicates that inhibition of P2X7R function on both T cells and monocytes is necessary for inhibiting the pathogenic cellular immune response in Col V-reactive patients.
Chemokine receptor defined T cell subsets express different levels of P2X7R
The use of chemokine receptor expression profiles has proven reliable for identifying subsets of IFNγ- or IL-17-producing T cells [25, 26]. Most notably, Th1-IFNγ-producing T cells can be identified by CXCR3 expression [25, 27, 28], whereas IL-17-producing cells can be identified by the expression of CCR4 and CCR6 [29-32]. With a clear effect of P2X7R antagonists on inhibiting cellular immunity to Col V (IL-17 dependent) but not TT (IFNγ dependent) we decided to investigate whether IL-17-producing T cell subsets have higher expression of P2X7R as opposed to their IFNγ-producing counterparts. In addition, we also investigated monocyte expression levels of P2X7R. In Figure 6A, representative flow plots from a Col V-reactive patient show P2X7R expression is highest on monocytes (CD14+/CD3−), followed by CXCR3−/CCR4+/CCR6+ T cells, and finally CXCR3+/CCR4−/CCR6− T cells, as assessed by MFI of P2X7R staining. This trend of expression was relatively consistent within our patient cohort. Corrected MFI data for each Col V-reactive patient were compiled and show that monocytes have significantly higher expression of P2X7R than CXCR3+/CCR4−/CCR6− T cells and trend toward higher P2X7R expression than CXCR3−/CCR4+/CCR6+ T cells. Furthermore, CXCR3−/CCR4+/CCR6+ T cells have significantly higher expression of P2X7R than CXCR3+/CCR4−/CCR6− T cells (Figure 6B, top). Investigation of the frequency of P2X7R positive cells within monocyte or T cell subset populations illustrates that there are significantly higher frequencies of P2X7R positive cells in monocytes and Th17 populations as compared to Th1 (Figure 6B, bottom).
Differential requirement for P2X7R function in viral versus allo-immune TV-DTH responses in transplant recipients
The finding of differential P2X7R expression on circulating Th1 versus Th17 cells suggested that we broaden the scope of our inquiry into P2X7R dependency of T cell–mediated immunity in transplant patients. To do this, we looked at the P2X7R requirement of TV-DTH responses to allo-antigen and EBV viral proteins, in addition to Col V and TT in a heart transplant patient with mixed chronic and acute rejection, as well as in a well-established cohort of chronic kidney rejectors. Figure 7A shows TV-DTH responses to Col V and allo-antigens. A biopsy taken at the time of leukapheresis revealed active acute rejection level 3A and CAV as indicated by luminal occlusion of arterioles (Figure S4). Both Col V and allo-TV-DTH responses were inhibited by Suramin and AZD9056. In addition, IL-17 and IL-1β neutralization dramatically inhibited TV-DTH responses to both Col V and the allo-indirect pathway. Of note, IFNγ neutralizing antibody also inhibits the cellular immune response to both Col V and the allo-indirect pathway, indicating that this patient exhibits a Th1/17 response in the face of Col V or allo-challenge. Figure 7B shows the P2X7R and cytokine dependency of both TT (open) and EBV (checkered). Both TT and EBV show no decrease in TV-DTH responses in the presence of Suramin and AZD9056. Similarly, IL-17 and IL-1β neutralization do not decrease their cellular immune responses. Only neutralization of IFNγ reduced the response to TT and EBV. To further investigate P2X7R sensitivity of the allo-indirect response versus anti-viral responses to EBV, we used a group of chronic kidney rejectors in the TV-DTH assay in conjunction with P2X7R antagonists and antibody neutralization strategies. Figure 7C shows TV-DTH responses to allo (left) and EBV (right) antigens. AZD9056 and Suramin both significantly inhibit Allo-induced swelling, while exhibiting no effect on the EBV response. Furthermore, the allo-indirect response in this chronic kidney rejector group was significantly inhibited by neutralization of IFNγ, IL-17, IL-1β and TNFα (Figure 7D).
We have previously shown that IL-1β and TNFα are required for the IL-17-mediated cellular immune response to Col V [13, 14]. More recent reports have shown that innate derived signals, including IL-1β, can release IL-17-producing T cells from quiescence and into active IL-17-producing cells . While P2X7R inhibition on monocytes can be associated with the loss of IL-1β production , the effect of P2X7R inhibition on T cells is not so clear. What is clear from our study (Figure 6) is that both monocytes and CXCR3−/CCR4+/CCR6+ T cells, a subset of T cells associated with IL-17 production, have higher P2X7R expression than the IFNγ-associated T cell subset, CXCR3+/CCR4−/CCR6−. Furthermore, monocytes, consistent with the work of Gu et al , have the highest expression of P2X7R among blood mononuclear cells. While the function of the T cell P2X7R was not directly investigated, there is clearly a nonredundant role for this as only when both the monocyte and T cell P2X7R (Figure 5) were blocked by oATP, is the cellular immune response to Col V lost. One way in which the function of P2X7Rs on T cells can be explained in the context of the Th17-mediated Col V response is through an observation made by Schenk et al . In this report, ATP signaling through T cell P2X7Rs led to sustained activation of ERK 1/2, causing altered phosphorylation of a family of T cell–related transcription factors, the nuclear factor of activated T cell. In this study, inhibition of T cell P2X7Rs with oATP blocked ERK 1/2 activation, leading to T cell anergy . Interestingly, we found that an inhibitor of ERK 1/2 activation (U0126) blocked the Col V-TV-DTH response to similar levels observed with P2X7R antagonists (Figure S5). Consistent with Th17 but not Th1 sensitivity to P2X7R inhibition, U0126 did not inhibit the TT response. It is possible that the role of P2X7Rs on T cells involves releasing Th17 cells from the constraint of intrinsic regulation of TCR signaling. It has been suggested that CD161+ T cells, which comprise the bulk of CD4 and CD8 IL-17-producing T cells, have an intrinsic regulation of TCR signal upon contact with APCs . The P2X7R-ERK cascade may unlock the intrinsic regulation of the TCR signal, allowing rapid secretion of IL-17. This cascade may function independently of the IL-1 receptor signaling pathway through which monocyte-derived IL-1β promotes IL-17 release .
It is now clear that the Col V-α1-derived 15mer peptides are sufficient to elicit a response in the TV-DTH assay equivalent to that of the whole protein . The response to peptides of the α1(V) chain, which represents the immunogenic portion of the Col V [α1α2] heterotrimer , was also exquisitely sensitive to P2X7R inhibition, both in vivo (Figure 3) and in vitro (Figure 4). Since the peptides can only induce cellular immune responses via the TCR interaction with pMHC-II complexes on the monocytes, this finding was consistent with the critical importance of P2X7R immune signaling for optimal TCR function in Th17 cells.
The observation that Col V also stimulated IL-1β and TNFα in isolated monocytes in a P2X7R-dependent manner is consistent with the idea that Col V, a heavily posttranscriptionally modified protein , can activate leukocytes directly as previously reported for lung epithelial cells in culture . How this occurs, and whether pattern recognition receptors are involved, is unclear. The downstream signaling from activated P2X7Rs can lead to inflammasome activation [38, 39]. This activated complex of proteins is responsible for the processing of pro IL-1β/IL-18 into the mature IL-1β and IL-18 proteins [18, 40, 41]. While multiple mechanisms for the secretion of IL-1β in response to LPS exist , Col V activation of monocyte IL-1β would appear to utilize a prototypical P2X7R-dependent Nlrp3 inflammasome pathway, consistent with that reported by Jankovic et al as a potential regulator of acute graft-versus-host disease . In this study, the authors found that removal or deletion of IL-1β or Nlrp3 inflammasome components increased allo-reactive T cells and decreased IL-17-producing T cells. Furthermore, a recent report from Lasiglie et al  underscored the importance of IL-1β/inflammasome functioning in the regulation of Th17 responses in patients with cryopyrin-associated periodic syndrome (CAPS). They reported a naturally occurring mutation in NLRP3, an important member of the inflammasome complex, within the CAPS patient population, caused enhanced IL-1β production from monocytes, and subsequent increased IL-17 serum levels and production in vitro . While this report deals with mutations in the inflammasome pathway, the evidence strongly supports a role for IL-1β driving Th17 effector functions, similar to what we observe in our Col V-reactive patients.
Our data suggest that there are multiple ways that monocytes can be activated during immune responses to Col V, including: (1) through direct interaction with intact Col V protein, a component of “proinflammatory matrix” [44, 45]; (2) as a result of TCR interaction with DR-bound α1(V) peptide on the monocyte APC; and (3) by interaction with IL-17 (its receptor being abundantly expressed by monocytes ). Consistent with the TV-DTH results, we also observed that the Col V-specific peptides increased IL-1β production in whole PBMCs, albeit at a lower level than observed with whole Col V, and in a P2X7R-dependent fashion. Since the peptides are only working through TCR-MHC interactions (Figure 4; i.e. peptides failed to activate monocytes directly), the monocyte would not receive the same amount of stimulation with Col V-peptides as it would with the whole Col V molecule. The coordinated signaling through P2X7Rs on both the monocyte and T cell may release Th17 effector function previously held in check.
We have previously shown that kidney graft chronic rejectors made anti-donor antigen responses in the TV-DTH assay that were inhibited by neutralizing both IFNγ and IL-17 . Herein, we report that allo-indirect responses of chronic kidney rejectors are also sensitive to IL-1β and TNFα neutralization as well as P2X7R inhibitors. One important conclusion from this study is that P2X7R inhibition in a heart transplant patient or a group of chronic kidney rejectors does not just selectively block the cellular immune response to sequestered self-antigens like Col V, but also inhibits the allo-response to donor antigen, while not affecting the viral (EBV) or bacterial (TT) response (Figure 7). In both groups of chronic rejectors, both the Col V and indirect allo-response were dependent on both IL-17 and IFNγ, indicating a Th1/Th17 response, whereas the EBV and TT responses were purely IFNγ dependent. Since P2X7R inhibitors appear to selectively inhibit the self- and allo-responses while maintaining viral and bacterial immune responses, the use of specific inhibitors to P2X7R may be ideal for anti-rejection drugs as viral and bacterial immunity would be unaffected while immune responses to allo and self would be attenuated.
The authors wish to thank Matt Pestrak for aiding in the completion of the experiments, Dr. Jose Torrealba for reviewing the histology data and the laboratory of Sam Gellman for the production of Col V (α1) peptides. The authors would like to further thank Dr. Richard Derks and Dr. Jenny Gumperz for their expert consultation and guidance throughout this project. This work was supported by NIH grant #PO1AI084853 (JAS, EJ-G, DSG, DAW and WJB), by the EU-sponsored One Study (WJB), and by R01# HL118115 (LCD and LS).
J. A. Sullivan designed the research, performed the experiments, analyzed the data and wrote the paper. E. Jankowska-Gan performed experiments, analyzed data and helped write the paper. L. Shi performed experiments and analyzed data. D. Roenneburg performed experiments and helped analyze data. S. Hegde helped design experiments. D. S. Greenspan provided access to cells from CAD patients enrolled in his atherosclerosis study, and helped design experiments. D. S. Wilkes helped design experiments and provided helpful comments on the manuscript. L. C. Denlinger helped design experiments, provided vital reagents and analyzed data. W. J. Burlingham designed experiments, analyzed data and helped write the manuscript.
The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation. David S. Wilkes is a co-founder of ImmuneWorks, Inc., a biotechnology company involved in designing therapeutics for various forms of lung diseases. All other authors have no conflicting interests.