Cellular heterogeneity of AITL compared to PTCL-NOS.
The PTCL cases were identified primarily by morphological assessment and the expression of WHO-defined IHC markers. The pathological diagnosis of PTCL and interpretation of the bone marrow biopsy (BMB) histology results were made on tissue specimens by an experienced hematopathologist. We collected 11 fresh bone marrow aspirates samples from the patients with BMI, including 8 samples of AITL and 3 samples of PTCL-NOS. We also collected 2 samples from the healthy donors as control.
Based on the scRNA-seq profiles, we clustered totally 60170 cells of different type across samples by using Seurat software (v3.0.2) and visualized them in two-dimensional space (Fig. 1A and B). The cell clusters were annotated with SingleR software (v1.0.1) and refined with expression of canonical marker genes (Fig. 1C and D). Major cell types comprising the bone marrow cells were well captured for scRNA-seq, including T cells (CD3D, CD3E, CD3G), natural killer (NK) cells (NCAM1 or CD56, KLRB1, NKG7), B cells (CD79A, CD79B, MS4A1), myeloid cells (CD1C) and plasmacytoid dendritic cells (pDCs) (LILRA4). There were also some residual erythrocytes (HBB) and megakaryocytes (PPBP) mixed in the bone marrow (Supplementary Fig. 1). We pooled the bone marrow cells by conditions and found that, compared with healthy controls, the AITL patients showed a decreased percentage of B cells, while an increased percentage of myeloid cells and NK cells (Fig. 1B). These observations were in line with PTCL-NOS, suggesting that scRNA-seq recovered key immune alterations of PTCLs (Supplementary Fig. 2). However, the pro-inflammatory response in AITL was more active than that in PTCL-NOS and heathy control (Fig. 1E, both p value < 10− 16). As lymphoma has been associated with cytotoxic lymphocyte exhaustion, we profiled the expression of genes encoding canonical exhaustion markers by T and NK cells. The significant exhaustion of T cells was only found in patients with AITL (Fig. 1F, both p value < 10− 16). However, the NK cells from all samples with AITL or PTCL-NOS appeared exhausted based on expression of LAG3, PDCD1, TIGIT and HAVCR2 (Fig. 1G).
In addition to these common features in PTCLs compared with healthy controls, there are some unique characteristics for AITL or PTCL-NOS. The immune alterations between AITL and PTCL-NOS are mainly concentrated in lymphocyte and myeloid cells. We separated these cells for a more detailed and comprehensive analysis. In the cluster of 28463 lymphocytes, there were 11 sub-clusters named with CD4 + memory, CD4 + naive, CD4 + Treg, CD8 + effector GZMB+, CD8 + effector GZMK+, CD8 + effector Memory, CD8 + effector CMC1+, CD8 + naive, Cycling, Cycling PCLAF + and precursor-exhausted cells (Fig. 2A-D). We identified the proliferative lymphocytes cells that appeared to be increased in the samples of AITL (10%) and PTCL-NOS (21%) compare with healthy control. The proportion of CD4 + Naive and CD8 + Naive cells were dramatically decreased (AITL: CD4 + 38–10%, CD8 + 41–11%; PTCL-NOS: CD4 + 38–17%, CD8 + 41–8%) in the patients compared with healthy control. These naive cells developed into other effector cells, regulator cells and memory cells. The CD8 + effector cells (GZMB + and GZMK+) and CD4 + regulator cells (FOXP3+) were enriched in the patients of AITL compared with PTCL-NOS containing high proportion of CD8 + CMC1 + effector cells (28%). There was no significant evidence of CD4 + T cell and CD8 + T cell exhaustion in patients. Moreover, we observed an increased level of precursor exhausted T (Texp) cells enriched in AITL patients specifically (Fig. 1D,).
Based on the differentially expressed gene (DEG) signatures, we identified relatively specific genes that were highly expressed in AITL patients were enriched in T cell activation and regulation panels (Supplementary Fig. 3). The top enriched GO terms of biologic process also included the response to interferon and pro-inflammatory cytokines (Fig. 2E). Conversely, the relatively specific genes that were highly expressed in PTCL-NOS patients were mainly enriched in the neutrophil activation (Fig. 2F), suggesting the cross-talk between T cells and neutrophil cells in PTCL-NOS.
In the cluster of 21645 myeloid cells, there were 11 sub-clusters named with CD14 + monocyte, VCAN + CD14 + monocyte, IL1B + CD14 + monocyte, CD16 + monocyte, FCGR3B + neutrophil, DEFA3 + neutrophil, IFITM2 + neutrophil, myeloid dendritic cells (mDC) and proliferative myeloid cells (MKI67) (Fig. 3A-C and Supplementary Fig. 4). We identified more activated monocyte cells enriched in the patients of AITL and more neutrophil cells enriched in the patients of PTCL-NOS (Fig. 3D). Theses neutrophil cells may be activated by the T cells with highly expressed genes enriched in the GO term of neutrophil activation in the patients of PTCL-NOS. For the patients of AITL, the relatively specific genes that were highly expressed in myeloid cells were mainly enriched in the Toll-like receptor, NF-kappa B, or TNF signaling pathways (Fig. 3E and F), which were the canonical pathway enriched in AITL compared with other PTCL (Fig. 3E and F). In addition, the HLA class II upregulation in the monocyte cells were enriched in the patients of AITL (Fig. 3E).
Immune heterogeneity of bone marrow involvement in AITL under different conditions
The immune heterogeneity is not only related to the development of different cell types with lymphoma, but also could affect the progresses of disease. After treatment, the progresses of AITL patients collected were in different conditions. We divided these patients into four groups based on the prognosis after the treatment: the response group under Anti-CD30 antibody-based therapy (Anti-CD30, n = 3); the response group under Chidamide treatment (Chidamide, n = 2); the progressed group developed into the diffuse large B cell lymphoma (DLBCL, n = 2); the progressed group developed into the lymphoma-associated hemophagocytic syndrome (LAHS, n = 1). Further analyses showed the immune heterogeneity associated with these different prognoses after treatment. For the lymphocytes, the proportion of different subtypes varies greatly (Fig. 4A-C). In the response group, we found more CD8 + effector cells were enriched in the Anti-CD30 group, and more CD4 + regulatory and memory cells were enriched in the Chidamide group. In the progressed group, more plasma cells and CD4 cells with the expression of CD79, a maker gene of B cells, were enriched in the DLBCL group, indicating the early signs of secondary B-cell lymphoma (supplementary Fig. 5). Moreover, the precursor exhausted cells were found enriched in the progressed group of LAHS (Fig. 4D).
Based on the DEG signatures, we identified relatively specific genes that were highly expressed in the response group of Anti-CD30 were enriched in T cell activation and cytotoxicity. In the response group of Chidamide, the relatively specific genes highly expressed were enriched in T cell regulation and proliferation. Compared with the progressed group, the immune response was more active in the response group (Fig. 4E).
For the myeloid cells, more neutrophil cells and proliferative cells were enriched in the response group of Anti-CD30 (Fig. 5A, B, C and D), while more monocyte cells and IFITM2 + neutrophil cells were enriched in the response group of Chidamide (Fig. 5D). Compared with other progressed group, the IL1B + CD14 + monocyte cells in the group of Chidamide and the DEFA3 + neutrophil cells in the group of Anti-CD30 were more prominent in the proportion. Compared with the progressed group, the immune response of myeloid cells were more active in the response group. However, the pro-inflammatory response caused by the myeloid cells were enriched in the progressed group of LAHS (Fig. 5E).
Genetic variation associated with the BMI progress of AITL after the treatment
Besides the immune heterogeneity, the genetic variations associated the progress of AITL were found in the bone marrow involvement. In the group Anti-CD30, the RHOA mutation was found enriched (Fig. 6A). We performed whole exome sequencing (WES) and single-cell T cell receptor sequencing (scTCR-seq) for confirming the RHOA mutation in these samples. The HLA typing and neoantigen prediction were conducted based on these sequence data (supplementary Fig. 6). We found that there was indeed the neoantigen of RHOA mutation peptide as expected (Fig. 6B). The mutation-associated neoantigens (MANA) may cause the enrichment of specific TCR-T cells. Based on this hypothesis, the single-cell immune profiling of these samples were obtained. We found the enrichment of specific TCR-T cells in the group of Anti-CD30 compared with other AITLs, which is beneficial for the immunotherapy (Fig. 6C). For the response group under Chidamide treatment, we found the clonal expansion in CD4 T cells with the same TCR sequence.
However, not all genetic variants leading to the better treatment outcome. Based on the scRNA-seq data, we found that the copy number variation happened in the group LAHS with chromosome 5 (chr5) gain and enriched in the precursor exhausted cells (Fig. 6D and E). This group is characterized by lymphoma-associated hemophagocytic syndrome with aggressive clinical behavior. The examination of chr5 genes in cases with GEP showed that 95 genes were significantly upregulated in the cases with chr5 gain, including IL4, IL13, and MAPK9, which affect cell cycle regulation and T-cell differentiation 43. Many of these genes related to cell cycle and T-cell activation were also significantly up-regulated in cases with chr5 gain, suggesting the potential connection with the lymphoma-associated hemophagocytic syndrome.