Diffuse Large B-Cell Lymphoma Remodels the Fibroblastic Reticular Network That Acquires Aberrant Immunosuppressive Capabilities; Implications for the Regulation of Anti-Tumor Immunity in the Immuno-Oncology Era

Abstract

Immunotherapy has demonstrated potential to reactivate or transfer T cell immunity and regress tumors, offering hope to relapsed or refractory diffuse large B-cell lymphoma (DLBCL) patients. However, many DLBCL patients do not experience therapeutic benefit, likely owing to a lack of pre-existing anti-tumor immunity and/or poorly understood immunosuppressive mechanisms in the tumor microenvironment (TME). Understanding the different obstacles that cytotoxic T cells face in the DLBCL TME will help the development of novel therapeutic approaches to overcome them and optimize immunotherapy. Stroma-associated gene signatures reflecting fibroblast and immune cell activity as well as angiogenesis have been associated with outcome in DLBCL but the biology underlying these signatures has been understudied. Here we have examined beyond tumor 'effacement' and hypothesized that, rather than being sparse bystanders, lymph node stromal cells may be important players in driving immune suppression in lymphoma. Multiplex immunofluorescence (IF) confocal microscopy analysis of the major stromal cell subsets revealed a marked expansion and remodeling of podoplanin, PDPN+ fibroblastic reticular cells (FRCs) in DLBCL lymph node tissue TME biopsies across GCB and ABC subtypes (n=40) compared to reactive control tissues (n=10). FRC myofibroblasts were similarly remodeled in tumors from the transgenic mouse model of DLBCL (Iμ-HABCL6, n=10) compared to wild type littermates (n=5). These altered PDPNhigh, αSMAhigh FRC networks were interspersed within effaced lymph node tissues in close proximity to DLBCL tumor cells. To model the interactions between tumor cells and FRCs, we established 2D and 3D co-culture platforms that combined DLBCL cells (or non-malignant control B-cells) and PDPN+ FRCs derived from human or murine lymph node tissues. These quantitative assays have shown that tumor cells activate FRCs promoting their proliferation, increased expression of PDPN, marked elongation/stretching and subsequent reduced ability to contract 3D collagen matrix (non-contractile) (P<.01). Exposure to tumor cells caused uncoupling of PDPN from RHOA signaling and redistribution of PDPN into lipid rafts, permitting FRC stretching. FRCs purified from both human and murine tumor lymph nodes showed the same activated morphology and phenotype, demonstrating that our co-culture systems recapitulate in vivo findings. Screening experiments have shown that lymphoma-expressed membrane and soluble lymphotoxins (LTα1β2, LTα3) as well as TNFα significantly contribute to the remodeling of FRCs (P<.01). Co-culture assays have revealed evidence for mutualistic interaction as FRCs, that express BAFF, promote the survival of DLBCL cells in 3D matrix gels (P<.01). Expanded lymphoma PDPNhigh FRCs in situ co-expressed BAFF compared to a more restricted expression profile (marginal zone FRCs) in reactive lymph node tissues. Flow cytometry revealed that lymphoma FRCs exhibit a cancer-associated fibroblast (CAF)-like immunophenotype including upregulation of fibroblast activation protein (FAP) and αSMA, as well as immunomodulatory MHC class I, PD-L1 and PD-L2 molecules compared to control FRCs (P<.01). An important function of FRCs in regulating immunity is attracting and maintaining T cells by secreting chemokines and promoting their migration along the network. Functional assays revealed that T cells show significantly reduced chemotaxis as well motility (quantitative time-lapse movies) across 2D and 3D lymphoma FRC networks compared to control FRCs (P<.01). Multiplex IF analysis revealed reduced CCL19 and CCL21-expressing FRCs in DLBCL compared to the reactive control FRC network as well as low numbers of tumor-infiltrated CD8+ T cells, which localized closely with remodeled FRCs. We next determined whether lymphoma FRCs (CAFs) could negatively regulate T cell function. Triple culture autologous assays (murine and human) have shown that prior exposure of tumor-infiltrated CD8+ T cells to FRCs significantly decreased their cytolytic killing activity against tumor cells (P<.01). In conclusion our data indicate that DLBCL tumor cells convert FRCs into immunosuppressive CAFs, which exhibit altered immumomodulatory activities at different levels that we believe has important implications for the regulation of anti-tumor immunity as well as response to immunotherapy. Disclosures Vardi: Gilead: Research Funding; Janssen: Honoraria. Ramsay:Roche Glycart AG: Research Funding; MedImmune: Research Funding; Celgene Corporation: Research Funding.

Developing a 3D B Cell Lymphoma Culture System to Model Antibody Therapy

“…Further evidence that the 3D spheroid co-culture environment featured supportive interactions between its constituents was provided by the preservation of podoplanin and ICAM-1 expression on ADSC-derived lymphoid fibroblasts recovered from 3D DLBCL/ADSC/MDM spheroid co-cultures following cytokine removal ( Supplementary Figure 6A ). This ability to compensate for the loss of extrinsic cytokine-derived signals is supported by recent data showing that co-culture of DLBCL cells with human lymph node-derived FRC increased their expression of podoplanin, a process linked to lymphoma secreted lymphotoxins and TNF-α ( 14 ).…”

“…The viability of DLBCL cells recovered from 3D collagen spheroid co-cultures and 2D co-cultures ran in parallel, was similar suggesting that the 3D co-culture system generated an environment compatible for DLBCL cell survival ( Figures 7C, D ). Interestingly, recent data indicate that BAFF-expressing FRC promote the survival of DLBCL cells in 3D matrix gel co-cultures ( 14 ), highlighting a key support role for lymphoid fibroblasts in the DLBCL TME. However, a role for MDM in the improved survival of DLBCL in our system cannot be excluded, given the reported supportive role of monocytes in DLBCL survival and proliferation ( 57 ).…”

“…In the case of FL, cross talk between tumor cells and cells of the local FRN drives their differentiation into tumor-supporting lymphoid stroma ( 13 ). Similarly in DLBCL, malignant cells and non-malignant TME components have been shown to induce cells of the FRN, specifically Fibroblastic Reticular cells (FRC), to adopt a CAF-like phenotype ( 14 , 15 ). Furthermore, CAF can promote survival of primary lymphoma cells in-vitro ( 14 , 16 ), further highlighting the intimate inter-relationship between CAF and tumor cells in the DLBCL TME.…”

“…Similarly in DLBCL, malignant cells and non-malignant TME components have been shown to induce cells of the FRN, specifically Fibroblastic Reticular cells (FRC), to adopt a CAF-like phenotype ( 14 , 15 ). Furthermore, CAF can promote survival of primary lymphoma cells in-vitro ( 14 , 16 ), further highlighting the intimate inter-relationship between CAF and tumor cells in the DLBCL TME. Similarities between CAF and normal lymphoid fibroblasts have also been reported ( 17 ), with human tonsil derived primary stromal cell cultures shown to support the survival and proliferation of DLBCL cell lines ( 18 ).…”

The Tumor Microenvironment of DLBCL in the Computational Era

“…For instance, genetically unstable DLBCL cells display reduced surface expression of MHC and CD58 molecules, thus lowering T cell and NK infiltration and cytotoxicity (51). Conversely, DLBCL-released lymphotoxins and TNF-alpha were (9) reported to promote the proliferative attitude of podoplanin-, PD-L1/L2-positive fibroblasts, while lowering their ability to contract collagen fibers and attract cytotoxic T cells (52). Overall, it is conceivable that the local extent of constitutional and reactive processes of both stromal and inflammatory nature shapes the final cellular composition of the affected lymph nodes, forming specialized contextures with topographical and functional identity (Figure 1).…”