Single Cell Transcriptomic Analysis of the Bone Marrow Microenvironment Uncovers a Role for Antigen Presentation Processes in MDS/AML Relapse after Allogeneic Hematopoietic Stem Cell Transplantation

Background: Bone marrow (BM) diseases such as myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) are characterized by a poor overall survival with allogeneic hematopoietic stem cell transplantation (allo-HSCT) often representing the only curative treatment option. However, patient mortality even after allo-HSCT remains high, mainly due to relapse. Previous studies showed involvement of immune escape processes in relapsed patients via downregulation of human leukocyte antigen class II (HLA-II) genes on leukemic blasts, or T cell exhaustion. Nevertheless, other components of the BM microenvironment (BMME) including mesenchymal stromal cells (MSC), endothelial cells and immune cells from the innate and adaptive immune system might play a role in relapse after allo-HSCT. This role remains unclear and needs further characterization. Therefore, we performed single cell transcriptomic analysis of different subpopulations of the BMME in MDS/AML patients who either achieved a long-term complete remission (CR) or relapsed after allo-HSCT.

Methods: Patient BM aspirates were obtained before and after 3-6 months post allo-HSCT after written informed consent, and mononuclear cells were stored in liquid nitrogen. To reduce confounding factors (i.e., pre-conditioning treatment, age, sex, diagnosis, graft versus host disease), "best-matched" patient pairs in CR or relapse were established. Subsequently, 2 pairs of matched CR/relapse patients were chosen for further evaluation (Table 1). From each patient, 6 different BM subpopulations were FACS-sorted to enrich for rare cell populations: hematopoietic stem and progenitor cells (HSPC, CD45^dim/CD34⁺), MSC (CD45^-/CD271⁺), endothelial cells (CD45^-/CD31⁺), monocytes (CD45^dim/CD14⁺), granulocytes (CD45^dim/CD11b⁺), T cells (SSC^low/CD45^hi/CD3⁺) and NK cells (SSC^low/CD45^hi/CD3^-/CD56⁺/CD16⁺). A maximum number of 5x10³ cells from each of these 6 cell populations from a single patient and time point were pooled back together in one lane of a 10X Genomics chip for single cell RNA sequencing library preparation. Data analysis was performed in R.

Results: After UMAP clustering, 15 groups of cells could be identified by their conserved expression patterns, recapitulating the different input sorted subpopulations. Gene set enrichment analysis comparing cell types from patients in CR and relapse revealed enrichment in immune related pathways in many clusters from relapsing patients both before and after allo-HSCT. In contrast, this pathway enrichment was only observed in classical monocytes from patients in CR after allo-HSCT.

Further gene expression analysis showed that within these pathways, the most significantly differentially expressed genes were HLA class I (HLA-I) and HLA-II genes, as well as the transcription factor STAT1. Of note, STAT1 expression was increased 1.2 to 2.3 fold in several cell types from patients in relapse compared to patients in CR. Similarly, the known STAT1 target genes HLA-A and HLA-C showed a 1.2 to 7.5 fold higher expression in several cell types from relapsed patients. Importantly, upregulations were observed in progenitor cells as well as in lymphocytes both before and after allo-HSCT, suggesting that STAT1 upregulation might trigger HLA-I genes expression and play a role in and/or serve as predictive marker of post-transplant relapse.

Interestingly, HLA-II gene enrichment patterns were more heterogeneous. Indeed, before transplantation, genes such as HLA-DRA, HLA-DPB1 or CD74 were upregulated 1.4 to 3.0 fold in most cell types of patients in relapse, while after allo-HSCT a downregulation of 1.5 to 3.5 fold was observed in progenitor cells from patients in relapse, and was even more prominent in classical monocytes showing a 2 to 8 fold downregulation (Figure 1). This observation suggests that by preventing T cell activation via HLA-II antigen presentation, monocytes might also play a role in relapse mechanisms after allo-HSCT.

Conclusions: Altogether, our results suggest that antigen presentation processes in the BMME via HLA-I and HLA-II could be involved in relapse mechanisms. Their expression level even before transplantation might be used as a predictor of relapse after allo-HSCT. Further analyses are ongoing to confirm our preliminary findings including functional analyses linking relapse with those observed differences in gene expression.

Bassermann:BMS: Honoraria. Goetze:Abbvie: Honoraria; Servier: Honoraria; BMS: Honoraria.