Integrative analysis of bulk and single-cell gene expression profiles to identify bone marrow mesenchymal cell heterogeneity and prognostic significance in multiple myeloma

Abstract

Background: Multiple myeloma is a hematologic malignancy characterized by complex interactions within the tumor microenvironment, where mesenchymal stem cells (MSCs) contribute significantly to disease progression, immune suppression, and drug resistance.

Methods: This study investigated the heterogeneity of MSCs in multiple myeloma using single-cell RNA sequencing (10X) and bulk transcriptomic data. Further analysis was performed by Seurat, SCENIC, CellChat. GSE4581 and GSE136337 were used as training set and validation set to construct a newly described prognostic model through COX and LASSO.

Results: By analyzing bone marrow samples from healthy donors and multiple myeloma patients at different Revised International Staging System (R-ISS) stages, this study identified distinct MSC subpopulations, including osteogenic, angiogenic, immune regulatory, and multipotent clusters, each of which plays unique roles in the tumor microenvironment. Interestingly, we found a unique subclone with upregulated expression of high mobility group proteins, these MSC exert a strong regulatory effect, which was defined as "HMGhMSC".

Conclusions: Our findings reveal that the proportion of osteogenic MSCs, which are crucial for bone health, decreases as the disease progresses, which is correlated with the bone lysis commonly observed in advanced multiple myeloma. Additionally, immune regulatory MSCs contribute to the formation of an immunosuppressive microenvironment, promoting tumor immune evasion. A prognostic model based on HMGhMSC subpopulations was developed, which demonstrated that these cells have significant potential as therapeutic targets for improving the prognosis and developing treatments for bone disease in multiple myeloma patients.

Keywords: High mobility group proteins; Immunosuppressive microenvironment; Mesenchymal stem cells; Multiple myeloma; Osteogenesis; Prognostic model; Single-cell RNA sequencing.

Fig. 1

Single-cell RNA-seq reveals heterogeneity of Mesenchymal Stem Cells in normal bone marrow and multiple myeloma BM tissues. A Illustration of workflow of scRNA-seq in human normal bone marrow and multiple myeloma samples. B UMAP visualization of color-coded clustering of 31,766 cells reveals 6 cellular clusters. The general identity of each cell cluster is shown on the right. C The proportion of cell lineages in multiple myeloma BMSC and normal BMSC

Fig. 2

Single-cell landscape of iregMSCs in multiple myeloma at different stages. A Kaplan–Meier survival analysis of iregMSCs in multiple myeloma (MM) based on the GSE4581 dataset. B UMAP plot showing iregMSC subpopulations. C Dot plot showing the expression of representative genes in iregMSC subpopulations. D The Gene Ontology (GO) analysis of iregMSC subpopulations: iregCO, iregC1, and HMGhMSC (left to right). E Transcription factor (TF) analysis of iregMSC subpopulations and FeaturePlot of HMGN2 expression. F Kaplan–Meier survival analysis of the HMGh genes (GSE4581) and its high expression in MM (GSE24870). G Kaplan–Meier survival analysis of the HMGhMSC subpopulation (GSE4581) and its high infiltration proportion in MM (GSE24870)

Fig. 3

HMGhMSC subpopulation in multiple myeloma at different stages. A DEGs of HMGhMSC subpopulations in MM and HD. B Heatmap showing the expression of representative prognostic DEGs in iregMSC subpopulations during MM progression and HD. C Kaplan–Meier curve of overall survival of TXN, PPP4 C, and NEK7 in MMbased on the GSE4581. D Dot plot showing the downregulation of membrane proteins involved in Wnt signaling and cell communication in MM-HMGhMSCs compared to HD

Fig. 4

Cell communication network between HMGhMSCs and other subpopulations in the bone marrow. A Cell–cell communication between iregMSC and ostMSC in HD (left) and MM patients (right). B Cell–cell communication between iregMSC and ostMSC in MM patients. C Cell–cell communication between single-subpopulation iregMSC and ostMSC in MM patients. D The GAS signaling pathway network in MM-I patients. Left and right portions show the autocrine and paracrine signaling, respectively. E Heatmap showing the correlation between HMGhMSC and other subpopulations

Fig. 5

Establishment and Evaluation of a Prognostic Model Based on the HMGhMSC subpopulations. A Confidence interval under each lambda in LASSO regression. B Change in trajectory of the LASSO regression independent variable. C, D The performance of th.e model in different cohorts. The survival curve of patients in high- and low-risk groups (c) and 1-, 3-, and 5-year time-dependent ROC curves of models from the GSE136337, and the GSE4581 cohorts, respectively (d)

Fig. 6

Single-cell landscape of osteogenic MSCs in multiple myeloma at different stages. A Kaplan–Meier survival analysis of osteogenic MSCs in multiple myeloma (MM) based on the GSE4581 dataset. B UMAP plot showing osteogenic MSC subpopulations. C Dot plot showing the expression of representative genes in osteogenic MSC subpopulations. D The numbers of osteogenic MSC subpopulations in multiple myeloma progression. E Transcription factor (TF) analysis of osteogenic MSC subpopulations and FeaturePlot of JUND expression. F The Gene Ontology (GO) analysis of osteogenic MSC subpopulations: ostC0 and ostC1. G DEGs of osteogenic MSC subpopulations in MM-III vs other patients and MM-I vs HD. H Kaplan–Meier survival analysis of the CTNNB1 (GSE4581) and its high expression in MM (GSE24870)