CD4⁺CTLs in Fibrosing Mediastinitis Linked to Histoplasma capsulatum

This article is featured in Top Reads, p.453

Fibrosis occurs in the context of autoimmune, allergic, and infectious diseases and is also prominent in some cancers. In most chronic fibrotic inflammatory diseases, including fibrosing mediastinitis (FM), the underlying mechanism of fibrosis is poorly understood (1–4). FM histologically resembles IgG4-related disease (IgG4-RD), a chronic inflammatory disorder with characteristic tumescent lesions that exhibit storiform fibrosis and are infiltrated by activated T and B lymphocytes with an abundance of IgG4-expressing plasma cells and/or plasmablasts (5). In some of our recent studies on IgG4-RD, we have established that CD4⁺CTLs are clonally expanded in the blood, enter tissue sites where they secrete profibrotic cytokines, including IL-1β and TGF-β, and may contribute to apoptosis followed by overexuberant repair processes (6–8). T_H2 cells are relatively sparse and are not clonally expanded in active IgG4-RD (6). In addition, we have also observed clonal expansions of activated B cells, including plasmablasts, that infiltrate disease tissues, interact with CD4⁺ T cells and secrete profibrotic products (9–11). More recently we have established that in systemic sclerosis tissues as well, T_H2 cells are relatively sparse and CD4⁺CTLs are the dominant CD4⁺ T cells in tissue lesions, suggesting that broadly similar pathogenic mechanisms may drive these two distinct autoimmune fibrotic diseases (12–14). In systemic sclerosis, prominent apoptosis of endothelial cells and subsequent overexuberant tissue remodeling likely result in fibrosis, but the specific antigenic peptides that are recognized by CD4⁺CTLs remain to be identified (12). It is therefore likely that the process of inflammatory fibrosis in IgG4-RD, systemic sclerosis, and possibly other autoimmune and infection-linked fibrotic disorders may be linked to the expansion of Ag-specific CD4⁺CTLs that induce apoptosis and secrete profibrotic products. Although many descriptions of the underlying basis of fibrosis invariably mention T_H2 cells, the role of T_H2 cells in fibrosis may be restricted to a relatively limited spectrum of disorders.

A number of autoantibodies have been described in IgG4-RD (15–18), but whereas the Ag or Ags recognized by CD4⁺CTLs in this disease are likely to be self-antigens, no specific Ag that triggers these expanded T cells has so far been identified in patients with IgG4-RD. Similarly, whereas specific autoantibodies have long been described in systemic sclerosis, Ags that drive T cells in this disease have not been characterized (13). Confounding this issue, CD4⁺CTL expansions have been associated with ubiquitous human viral infections, such as EBV and CMV; one of the reasons for our studying FM is to ask whether indeed CD4⁺CTLs are abundant in this disorder and if fungal Ags can activate CD4⁺CTLs in this disease.

Despite being the rarest manifestation of Histoplasma capsulatum infection, FM is arguably its most severe presentation. Mortality is very high in the 20% of FM patients in whom vessels of both lungs are affected (19). H. capsulatum–associated FM represents the vast majority of FM cases in North America (20) and presumably results from an excessive fibroinflammatory response to H. capsulatum Ag in the vicinity of mediastinal lymph nodes and surrounding mediastinal tissues. As dictated by the geographic distribution of this dimorphic fungus, the original infection is acquired in the central portions of the United States, but progression of fibrosis is slow and silent for a long period (5–10 years) before vascular or airway stenosis reaches the critical levels that cause symptoms, which happen wherever the patient resides at that time (3).

The cellular and molecular mechanisms leading to FM development are currently poorly understood. H. capsulatum generally replicates in macrophages, typically induces a T_H1-type response in immunocompetent individuals and is cleared. Dominant T_H2 cell and regulatory T cell expansion are known to impede clearance (2). It is understood that, following H. capsulatum inhalation in most individuals, this dimorphic fungus disseminates to the mediastinal lymph nodes and then throughout the body, as commonly evidenced by multiple splenic microcalcifications. H. capsulatum is unique, as remnants of cell wall persist indefinitely in the lungs and mediastinal lymph nodes in most persons postinfection, and it is the only organism shown to cause FM in the US, so the hypothesis is that seepage of fungal Ags into the mediastinum trigger a hypersensitivity reaction coupled to a local inflammatory response and exuberant fibrosis in genetically predisposed individuals (21). Collagen deposition by activated fibroblasts that invade inside critical vessels or airways leads to total obstruction, with reduced blood flow or airflow that is the basis for the various clinical manifestations of FM (20).

Despite the strong association with H. capsulatum, the fungus itself is rarely grown from FM tissues, so the organism is likely dead. In endemic areas, the fungal link is established by serologic testing and pathognomonic clinical features. Adaptive immune involvement in the pathophysiology of FM has been suggested by the abundance of CD20⁺ B cells infiltrating the lesion (3). Those B cells surround areas of fibrosis or form poorly organized secondary lymphoid structures without germinal centers. The lesions are also marked by an abundance of both CD3⁺CD8⁺ and CD3⁺CD8⁻ T cells, the latter most likely representing CD4⁺ T cells. Overall, the similarities to IgG4-RD (and systemic sclerosis) are quite striking. FM linked to H. capsulatum may therefore provide us with an infectious “model” with defined Ags that has the potential to further our understanding of idiopathic inflammatory fibrotic disorders such as IgG4-RD and systemic sclerosis.

In this study, we have demonstrated that CD4⁺CTLs are expanded in the blood of patients with FM and that these cells also infiltrate disease lesions. The CD4⁺CD45RO⁺CD27⁻SLAMF7⁺CD28^loCD57^hi effector population of CD4⁺CTLs in the blood is clonally expanded. Purified effector CD4⁺CTLs can be reactivated in vitro following exposure to H. capsulatum Ags, suggesting that fungal-specific CD4⁺CTLs likely contribute to the pathogenesis of FM.

Patients with FM were evaluated and diagnosed at Vanderbilt University, and anticoagulated blood was shipped to Boston, where PBMCs were isolated by Ficoll and preserved in gas-phase liquid nitrogen until the time of use. Mediastinum biopsies were performed at Vanderbilt University and paraformaldehyde-fixed, paraffin-embedded tissue slides were shipped to Boston for staining and analysis. Blood from healthy donors was obtained at Massachusetts General Hospital. Healthy donors were defined as lacking any current or prior history of malignancy, autoimmune disease, or recurrent/chronic infections. Clinical and demographic data on all subjects used for these studies is displayed in Supplemental Fig. 2 and Supplemental Table I. Data on sex, age, ethnicity, clinical manifestations, and H. capsulatum serology were extracted from the medical records of all patients.

These studies were approved by the Institutional Review Boards at the Massachusetts General Hospital and Vanderbilt University School of Medicine. All patients provided written informed consent prior to inclusion in the study.

Aliquots of frozen PBMCs were rapidly thawed and washed twice in complete DMEM. To avoid nonspecific binding of the Abs used in the staining procedure, Fc receptors were blocked using Human TruStain FcX (422302; BioLegend) following manufacturer recommendations. Cells were then stained for 30 min on ice and protected from light to minimize any photobleaching. T cell immunophenotype was assessed by staining (10 million cells/ml) using the following Ab panel (manufacturer, clone, concentration used): anti-human CD3-BUV395 (clone SK7, 1:40; BD Biosciences), anti-human CD45RO- allophycocyanin-Cy7 (clone UCHL1, 1:300; BioLegend), CD27-BV510 (clone M-T271, 1:20; BioLegend), CD4-BUV805 (clone SK3, 1:80; BD Biosciences), anti-human SLAMF7-AF648 (clone 235614, 1:10; BD Biosciences), CD28-PerCP-Cy5,5 (clone CD28.2, 1:20; BioLegend), and CD57-PE (clone NK-1, 1:100; BioLegend). After 30 min in the staining buffer at 4°C, cells were washed three times with cold 1% BSA in PBS and resuspended in 500 μl for analysis. Immediately prior to flow cytometry, cells were further stained with SYTOX AADvanced Dead Cell Stain (S10274; Thermo Fisher Scientific) to exclude any dead cell. Flow cytometry was then performed on a BD LSRFortessa (BD Biosciences, San Jose, CA). Rainbow calibration particles (eight peaks) were used to monitor instrument performances and ensure consistency between analysis. The flow cytometry standard (fcs) files generated were analyzed using FlowJo software (version 10.6).

Tissue samples were fixed in formalin, embedded in paraffin, and sectioned. These specimens were incubated with the following Abs: CD4 (clone CM153A; Biocare Medical), GATA3 (clone CM405A; Biocare Medical), and granzyme A (GZMA) (clone LS-C312742; Lifespan Biosciences) followed by incubation with secondary Ab using an Opal Multiplex IHC Kit (PerkinElmer). The samples were mounted with ProLong Diamond Antifade mountant containing DAPI (Invitrogen).

Images of the tissue specimens were acquired using the TissueFAXS platform (TissueGnostics). For quantitative analysis, the entire area of the tissue was acquired as a digital grayscale image in five channels with filter settings for FITC, Cy3, and Cy5, in addition to DAPI. Cells of a given phenotype were identified and quantitated using the TissueQuest software (TissueGnostics), with cut-off values determined relative to the positive controls. This microscopy-based multicolor tissue cytometry software permits multicolor analysis of single cells within tissue sections similar to flow cytometry. The principle of the method and the algorithms used have been described in detail elsewhere (22).

In TCR repertoire studies, PBMC were handled and stained exactly as for the flow cytometry experiments above. CD4⁺CTLs were defined using the following markers: DUMP(CD20-CD8)-CD3⁺CD4⁺CD45RO⁺CD27⁻SLAMF7⁺. CD4⁺CTLs were further divided using CD28 and CD57 to delineate “effector” and a “memory-like” phenotype within the CD4⁺CTLs. Our previous studies have indeed shown that CD28^loCD57^hi CD4⁺CTLs exhibit increased markers of cytotoxicity, whereas CD28^hiCD57^loCD4⁺CTLs harbor a phenotype resembling memory CD4⁺ T cells (23). Cells were sorted directly into 350 μl of RLT-BME lysis buffer (Qiagen), vortexed briefly, and flash-frozen on dry ice and kept at −80C until RNA extraction. Cells were sorted on a BD FACSAria II operating with BD FACSDiva version 7. RNA isolation was performed from lysates using RNeasy Plus Micro Kits [74034 processed through 5′ rapid amplification of cDNA ends RT-PCR; QIAGEN, as previously described (24)]. For gene-specific amplification of the TCRβ gene, the following reverse primer was used: 5′-TGCTTCTGATGGCTCAAACACAGCGACCT-3′.

The Ag used for this study was a cell wall/cell membrane extract of the G217B strain of H. capsulatum. Preparation of this extract has been described elsewhere (25). For in vitro stimulation, the final concentration in a well was 1 μg/ml.

Fresh PBMCs from an FM patient and healthy controls were plated in a 96-well plate at a density of 1 million cells per ml. Cells were incubated in the presence of CD40L blocking Ab and either vehicle, anti-CD3/CD28 beads (1:500) or the H. Capsulatum Ag mixture at a dilution of 1 μg/ml for 24 h. Cells were then washed and stained to assay for the induction of various activation markers. Briefly, to detect activation of CD4⁺CTLs, the flow cytometry panel was performed with the following additional markers: CD69-BV421 (clone FN50, 1:20; BioLegend), CD25-BV515 (clone 2A3, 1:20; BD Pharmingen); CD40L-Pe-Dazzle 594 (clone 24–31, 1:20; BioLegend), and OX40-BV711 (clone ACT35, 1:20; BD Pharmingen). Results were analyzed using fluorescence-minus-one control for every activation marker and a healthy control.

For flow cytometry and clinical correlations, GraphPad Prism version 8 was used for statistical analysis, curve fitting, and linear regression. A two-tailed Mann–Whitney U test was used to calculate p values for continuous, nonparametric variables, and a p value <0.05 was considered significant.

Based on our earlier studies on IgG4-RD, we quantified circulating CD4⁺CTLs by gating on SLAMF7⁺ CD4⁺ T effector/memory cells as defined by CD27⁻CD45RO⁺ in PBMCs from both FM patients (n = 39) and healthy controls (n = 24). Subsets of CD4⁺CTLs were additionally determined based on surface expression of CD28 and CD57 (Fig. 1A) (23, 26). We observed expansions of total CD4⁺ T effector/memory cells, CD4⁺CTLs, and specifically of CD28^lo CD57^hiCD4⁺CTLs in the blood of FM patients (Fig. 1B). The CD28l^hiCD57^lo CD4⁺CTL memory subset was decreased in the blood of FM patients, whereas the intermediate (CD28^loCD57lo) and unspecified (CD28^hiCD57^hi) CD4⁺CTLs did not vary significantly (Supplemental Fig. 1).

Next-generation sequencing of the rearranged TCR β-chain gene was undertaken on the four different subsets of CD4⁺ SLAMF7⁺ T cells, based on the expression of CD28 and CD57 (Fig. 2A). We identified large clonal expansions among the effector CD28^loCD57^hi CD4⁺CTL subset in FM patients, indicating a dominant response of these T cells to some defined Ag. Moreover, effector CD4⁺CTLs showed a high degree of clonal connectivity with the remaining CD4⁺CTLs (CD28^loCD57^lo, CD28^hiCD57^hi, and CD28^hiCD57^lo) but with markedly increased clonal expansion, with the top five clones accounting for more than 90% of the total clonal populations (Fig. 2B). Among the five FM patients whose CD4⁺CTLs were studied by TCR repertoire, we observed no shared Vβ gene usage in the dominantly expanded clones across these patients. In contrast to the CD28^loCD57^hi effector CD4⁺CTLs, the other subsets of CD4⁺SLAMF7⁺ T cells demonstrated a low degree of clonal expansion. TCRβ sequence and distribution from a representative patient are provided in Supplemental Table III.

Our previous studies have suggested that the fibrosis in IgG4-RD and systemic sclerosis occurs in a non–T_H2-driven inflammatory rather than allergic context and that CD4⁺CTLs are a prominent component of the inflammatory infiltrate but T_H2 cells are not. We used antigenicity restoration and multicolor immunofluorescence on formalin-fixed paraffin-embedded tissue sections from four FM patients and quantitated CD4⁺GATA3⁺ T_H2 cells and GZMA⁺ CD4⁺CTLs in each patient. An example of the staining performed is depicted in Fig. 3A.

Paralleling our results in IgG4-RD and systemic sclerosis, in FM tissues, CD4⁺CTLs were more prominent than T_H2 cells (31.86 ± 18.97% versus 6.33 ± 4.9% of total T cells) (Fig. 3B, 3C).

PBMCs from healthy controls and from patients with FM were either left untreated or exposed to anti-CD3 and anti-CD28 beads (as a positive control) or exposed to H. capsulatum Ags at a concentration of 1 μg/ml for 24 h. CD4⁺CTLs were gated on and levels of makers of recent activation (CD69, CD25, CD40L, and OX40L) were subsequently compared. As seen in Fig. 4A, 4B, a large proportion of the CD4⁺CTLs in FM patients responded to H. capsulatum Ags and were activated. The response was minimal in healthy controls, although their CD4⁺CTLs were responsive to the nonspecific antiCD3 and antiCD28 stimulation. Response to H. capsulatum was further objectively measured in five patients by assessing the variation of the median fluorescence intensity difference between CD4⁺CTLs exposed to vehicle or H. capsulatum Ag (Fig. 4B). A significant increase in the expression of the early activation markers CD69 (p = 0.0046), CD25 (p = 0.0157), CD40L (p = 0.0201), and OX40L (p = 0.0065) were observed.

Convincing evidence exists for the role of T_H2 type immune responses in fibrosis caused by nematodes, such as Schistosoma mansoni (27), and clearly, T_H2 cells may be relevant to some infections that result in fibrosis. However, there are numerous infectious diseases that result in fibrosis, and these include many diseases, such as tuberculosis, amebiasis, and some fungal infections in which type 2 immune responses are not known to be prominent. Our studies on IgG4-RD and systemic sclerosis have provided support for non–type 2 adaptive immune responses that drive autoimmune fibrosis (6, 23, 28). In many infections, as in autoimmune fibrotic conditions, induction of cell death at sites of infections by CD4⁺CTLs, possibly sustained by activated pathogenic B cells, may result in apoptosis, overexuberant tissue remodeling, and organ damage resulting from the ensuing fibrosis. CD28⁻ CD4⁺CTLs are considered especially active among CD4⁺CTLs subsets, as they exhibit increased markers of cytotoxicity (GZMA and perforin) (6, 23). This report therefore raises two important issues relevant to inflammatory fibrotic diseases. It suggests that many fibrotic diseases of both autoimmune and infectious origin may be driven by CD4⁺CTLs and not by T_H2 cells. It also provides evidence for Ag-specific CD4⁺CTLs in the context of any fibrotic inflammatory disease.

Very little is known about the pathogenesis of FM beyond the strong link to. H capsulatum. FM is an important clinical problem because it has high mortality, especially when both lungs are affected. Based on our current understanding of IgG4-RD and systemic sclerosis, we surmise that FM may be driven by H. capsulatum–specific CD4⁺CTLs (and possibly CD8⁺CTLs as well) that recognize H. capsulatum Ags, possibly on macrophages and inflamed stromal cells. These activated T cells may induce apoptosis in the mediastinal space as well and also contribute to overexuberant tissue remodeling and activation of macrophages and myofibroblasts. Activated B cells likely also play some role in reactivating CD4⁺CTLs at tissue sites, as they could display peptides on HLA class II molecules (6, 13) and may themselves secrete profibrogenic molecules like LOXL2 and PDGF (11, 13, 14).

The studies described in this article have focused on CD4⁺CTLs in FM and have established that at least some of these CD4⁺CTLs were induced by H. Capsulatum Ags. Given that there is a known linkage of this disease to HLA-DQβ1*0402 (29), future studies may focus on identifying the specific H. capsulatum antigenic peptide(s) that are presented on this MHC class II molecule to potentially generate and activate CD4⁺CTLs. However, this HLA association was modest and based on a relatively small sample size, so it likely that in any population susceptibility may be linked to a number of HLA class II alleles. We will also seek to more thoroughly dissect all infiltrating adaptive and innate immune cells and to explore the contributions of CD8⁺ T cells and especially B cells because, as for IgG4-RD, there is hope that B cell depletion might contribute to clinical improvement in FM (30).

The authors have no financial conflicts of interest.