Heterogeneity of PD-1+regulatory T Cells after Low-Dose IL-2 Therapy for Patients with Chronic Graft-Versus-Host Disease

Background. After allogeneic hematopoietic cell transplant (allo-HCT), impaired recovery of CD4 regulatory T cells (Treg) is associated with the development of chronic GVHD (cGVHD). Previous clinical trials have shown that daily administration of low-dose (LD) interleukin-2 (IL-2) selectively expands circulating CD4Treg in all patients with steroid-refractory cGVHD, with approximately 50% of adult patients having clinical improvement. Although some predictors of clinical response have been identified, CD4Treg-intrinsic factors associated with clinical improvement still need to be elucidated.

PD-1 is a coinhibitory receptor that attenuates the activation and function of effector T cells. However, the role of PD-1 on CD4Treg is not fully understood. LD IL-2 therapy induces expression of PD-1 on CD4Treg, and PD-1 has a critical role in maintaining CD4Treg expansion during LD IL-2 treatment (Asano T et al, Blood 2017). This prompted us to conduct a detailed analysis of PD-1⁺CD4Treg and their relevance to clinical outcomes after LD IL-2 therapy.

Methods. To identify intrinsic Treg factors associated with clinical outcome, we analyzed serial cryopreserved PBMC samples from 21 patients with steroid-refractory cGVHD who were enrolled in a clinical trial combining extracorporeal photopheresis (ECP) and LD IL-2 therapy. Phenotypic changes in PBMC during treatment were monitored by mass cytometry. To elucidate CD4Treg heterogeneity, we also used single-cell RNA sequencing to characterize CD4Treg (CD3⁺CD4⁺CD25⁺CD127^low) derived from cryopreserved PBMC of responders (n = 4) and non-responders (n = 6) four weeks after initiation of LD IL-2.

Results. ECP therapy alone had minimal impact on CD4Treg, but additional treatment with LD IL-2 markedly altered multiple CD4Treg populations, including a PD-1⁺CD4Treg cluster characterized as highly activated, immune suppressive, and proapoptotic. This effect was also observed in CD4Treg from healthy individuals stimulated with LD IL-2 alone in vitro. After in vitro LD IL-2 stimulation, PD-1⁺CD4Treg suppressed the proliferation of effector T cells more effectively than PD-1^-CD4Treg. Although PD-1⁺CD4Treg clusters increased from baseline in clinical responders, the increase of PD-1⁺CD4Treg clusters was not different between responders and non-responders.

Unsupervised clustering of single-cell RNAseq data identified six distinct groups of CD4Treg, including a naïve-like subset (PDCD1-low with high expression of CCR7, TCF7, and LEF1), an exhausted subset (PDCD1⁺ with high expression of LAG3, CTLA4, and TIGIT), and an effector subset (PDCD1⁺ with high expression of CCR4 and IKZF2). The exhausted CD4Treg subset also featured expression of cytotoxicity markers (NKG7, GZMA, and GZMK). When comparing LD IL-2 responders to non-responders, we found that the ratio of effector/exhausted Treg was significantly higher in patients with clinical response (p = 0.019).

To further explore differences within CD4Treg associated with clinical improvement of cGVHD, we performed pseudo-bulk differential gene expression analyses. We found that there were many individual genes whose expression was significantly associated with clinical response, particularly in an effector CD4Treg subset in non-responders. Gene set enrichment analysis revealed that pathways related to cytokine signaling, including interferon (STAT1, IFNGR2, IRF1, SOCS1, and SOCS3), were enriched in non-responders in multiple CD4Treg subsets. In responders, IFITM3 was upregulated among the effector and exhausted CD4Treg cells; this gene was recently reported as a negative feedback regulator of the IFNg-induced STAT1 pathway in CD4Treg. These results suggest that activation of the IFNg-induced STAT1 pathway in CD4Treg may have promoted expansion of more exhausted CD4Treg in clinical non-responders.

Conclusions. In summary, LD IL-2 therapy in patients with cGVHD induced multiple CD4Treg populations, including PD-1⁺CD4Tregs. However, PD-1⁺CD4Tregs are heterogeneous, including both exhausted and effector subsets, and the balance between these two CD4Treg subsets after IL-2 therapy appears to correlate with clinical improvement after LD IL-2 therapy. Additionally, activation of the STAT1 pathway is one of the putative CD4Treg-intrinsic factors associated with lack of clinical response to LD IL-2 therapy. Validation studies and functional assays are ongoing.