IFNγ Primed MSCs Suppress Stimulated T Cells By Inducing Perturbations of Redox Homeostasis

Abstract

Acute Graft versus host disease (GVHD) remains the primary cause of morbidity and nonrelapse mortality after allogeneic hematopoietic cell transplantation (HCT). Despite the current standard prophylaxis, approximately 50% of patients will develop acute GVHD. the outcome of patients with steroid-refractory and/or severe GVHD is poor; hence, more effective prophylaxis reducing the risk of these scenarios is greatly needed. In recent years, considerable attention has been directed toward mesenchymal stromal cells (MSCs) as prophylaxis and therapy for acute GVHD. Even though the immunosuppressive and immunomodulatory properties of MSCs are enhanced by proinflammatory cytokines such as interferon γ (γMSCs), relatively less attention has been focused on these cells. Our laboratory has shown human γMSCs are effective in GVHD prophylaxis in a fully mismatched murine HCT/GVHD model. When infused IV on day +1 after HCT, γMSCs traffic to the spleen, liver, and gut-associated lymphoid tissue (GALT) which is essential for effective prophylaxis. This may be a key element of the mechanism of prophylaxis as the gut and GALT are thought to be paramount for the development of GVHD. We now sought to understand the mechanism(s) of γMSC suppression of alloreactive T cells suppressive activity in this specific clinical and biological setting. In vivo, control mice transplanted with a fully MHC mismatched graft developed GVHD in 6-8 days (0% GVHD-free survival) while human γMSCs administration at day +1 completely prevented GVHD in the first 17 days (100% GVHD survival) but the animals developed acute GVHD in 21-28 days. Hence, a single dose of γMSCs is 100% effective early but is insufficient for overall GVHD prevention. The mechanism is highly effective but of relatively short duration, and seemingly reversible. In vitro, the reversible T cell suppressive activity was entirely contact-independent, while cell-to-cell contact could induce T cell death. Moreover, γMSC IDO1 expression, alone, was sufficient to suppress T cells. After 24 hours of T cell activation or γMSC suppression in vitro, we found the T cells were arrested in G0 inconsistent with most cell cycle regulation which would arrest cells in late G1 at the so-called restriction point. In scRNA Seq, we identified 16,478 expressed genes in both groups but only 98 genes were significantly differentially expressed (≥20% FC, Padj ≤ 10 -4). ROS response pathways were significantly enriched in the suppressed cells suggesting oxidative stress. Paradoxically, flow cytometric analysis of CM-H2DCFDA stained T cells revealed less ROS in suppressed cells compared to controls. Additionally, total NAD is elevated and the NAD/NADH ratio was 10-fold less in suppressed T cells compared to activated cells indicating an excess of reducing equivalents in the suppressed cells. Collectively these data define reductive stress. On this note, the upregulation of ROS response genes is also consistent with reductive stress. IDO1 activity seems to have induced reductive stress. The IDO1 catalytic activity converts tryptophan (Trp) to kynurenine (Kyn). We then found the addition of Trp (10 µM) to γMSC CM abolished suppression; supplementing fresh media with 15-fold excess Kyn had no effect. Moreover, RPMI without Trp and with dialyzed FBS also conferred suppression to activated T cells. These data suggest that among the many metabolites consumed or generated in γMSC CM, the lack of Trp induces T cell suppression. Trp supplementation of γMSC CM showed the expected level of ROS in activated T cells suggesting the Trp deficiency caused the reduced ROS. Supplementing γMSC CM with a graded concentration of hydrogen peroxide, we found that 100 nM hydrogen peroxide was not toxic to control cells, had ROS similar to activated T cells, and abolished suppression indicating the reduced ROS directly caused suppression. Kyn, nutritional deficiency, or an integrated stress response did not seem to play a measurable role in this setting. Collectively, our data suggest that IDO1-mediated Trp deficiency perturbs redox homeostasis (reductive stress) which suppresses activated T cells. We postulate that once redox homeostasis is restored, T cell activation proceeds. Future efforts will be directed toward identifying the link between Trp deficiency and redox homeostasis in this setting, and the genomic changes that directly mediate the lack of T cell proliferation.