Integrating single-cell biophysical and transcriptomic features to resolve functional heterogeneity in mantle cell lymphoma

Abstract

Intratumor heterogeneity impacts disease progression and therapeutic resistance but remains poorly characterized by conventional histologic, immunophenotypic, and molecular approaches. Single-cell biophysical properties distinguish functional phenotypes complementary to these approaches, providing additional insight into cellular diversity. Here, we link both buoyant mass and stiffness to gene expression to identify clinically relevant phenotypes within primary mantle cell lymphoma (MCL) cells, using MCL as a model of biological and clinical diversity in human cancer. Linked measurements reveal that buoyant mass and stiffness characterize B cell development states from naïve to plasma cell and correlate with expression of oncogenic B cell receptor signaling genes such as BLK and CD79A. In addition, changes in cell buoyant mass within primary patient specimens ex vivo correlate with sensitivity to Bruton's tyrosine kinase inhibitors in vivo in MCL and chronic lymphocytic leukemia, another B cell malignancy. These findings highlight the value of biophysical properties as biomarkers of response in pursuit of future precision therapeutic strategies.

Fig. 1.. Comparative analysis of biophysical properties, morphological attributes, and transcriptomic features in primary human MCL cells.

(A) Schematic workflow for isolating tumor cells from MCL PDX models for biophysical profiling, immunophenotyping, histopathologic imaging, and bulk RNA/DNA sequencing. Created in BioRender. M. Zhang (2025); https://BioRender.com/695y46e. (B) Histogram distributions of single-cell buoyant mass and stiffness measurements using the SMR, along with representative hematoxylin and eosin (H&E) stain and p53 IHC images (40×; scale bars, 15 μm) of human-enriched cells from the spleen of three PDX models: DFBL-96069, DFBL-39435, and DFBL-91438. Mass and stiffness profiles are from one representative replicate of three repeats. A box plot representing the median and interquartile range (IQR) is shown above each histogram, with the median mass and stiffness of DFBL-96069 indicated by a dotted red line for reference. The technical IQR of mass was determined by repeatedly measuring the same 10-μm polystyrene bead using the SMR and that of stiffness by repeatedly measuring the same mouse lymphocytic leukemia cell line L1210 (fig. S3, A to E). The TP53 mutational status based on targeted sequencing (table S1) and the relative percentage of p53 and Ki67 positive cells based on IHC stains (fig. S2B) are described below the H&E and IHC images. (C) Gene set signature enrichment evaluation of bulk RNA sequencing of MCL cells isolated from the spleen of the three PDXs. WT, wild type; a.u., arbitrary unit; MAPK, mitogen-activated protein kinase.

Fig. 2.. Cell mass and stiffness are strongly associated with the expression of genes annotated for cell division, cell cycle, and B cell activation.

(A) Schematic workflow linking the biophysical measurements of individual MCL cells from different tissues of three PDX models to downstream scRNA-seq. Created in BioRender. M. Zhang (2025); https://BioRender.com/s2zyoz3. (B) UMAP analysis of linked single-cell mass, stiffness, and scRNA-seq data of human tumor cells enriched from DFBL-96069, DFBL-39435, and DFBL-91438 isolated from different tissues (SP, spleen; PB, peripheral blood; BM, bone marrow; LIV, liver). Biological replicates, each from a different mouse, were included per tissue, as detailed in fig. S4. (C) GSEA of the top genes correlated with mass (left) and stiffness (right) measurements among the combined PDX cells. The color represents the P values, and the size of the spots represents the gene ratio, which is defined as the number of genes in the overlap divided by the total size of the gene set. (D) Biophysical correlation with gene expression in MCL PDXs. The correlations were analyzed using data from 500 MCL cells across 9442 genes. Representative genes are color-coded as follows: red for genes positively correlated with both mass and stiffness, purple for genes positively correlated with mass only, green for genes positively correlated with stiffness only, dark blue for genes negatively correlated with mass and stiffness, light blue for genes negatively correlated with stiffness only, and yellow for genes negatively correlated with mass only. (E) UMAP plots showing the expression of nine representative genes positively correlated with MCL cell mass and stiffness, marked as red in Fig. 2D.

Fig. 3.. Mature B cells undergo substantial biophysical change after activation.

(A) Schematic workflow showing the isolation of naïve B cells from PBMCs of three healthy donors, followed by ex vivo stimulation for immunophenotyping and biophysical profiling. Created in BioRender. M. Zhang (2025); https://BioRender.com/mjg6i9w. (B) Summary of the flow gating strategy used to characterize the different B cell populations. Naïve B cells and stimulated B cells at various differentiation stages were identified by immunophenotyping and sorted by fluorescence-activated cell sorting (FACS) on D0, D3, D7, and D14. The full panel is shown in fig. S2. (C) Mean proportion of B cells at different differentiation stages following ex vivo stimulation (n = 3). (D) Representative histogram distributions of single-cell buoyant mass (left) and stiffness (right) of each mature B cell population measured by the SMR, from one experiment out of three biological replicates using cells from three healthy donors. The number of single cells measured in each condition is indicated on the right. Technical IQRs were measured by trapping single cells (naïve B cell or L1210 for naïve activated D3) on the SMR (fig. S3, F to G). (E) Changes in cells’ buoyant mass (above) and stiffness (below) across three independent ex vivo simulations using human PBMCs from three different donors. *P < 0.05 and **P < 0.01 as compared between indicated groups (Student’s t test). (F) Correlation plot of cells’ buoyant mass and stiffness at each stage of differentiation. Data are presented as means ± SD of the three biological replicates depicted in (E). HLA, human leukocyte antigen.

Fig. 4.. Single-cell mass as a biomarker of susceptibility to pharmacologic inhibition of BCR signaling pathway activity in primary patient specimens.

(A) Workflow to assess the impact of BTK inhibition on the buoyant mass of MCL cells from patients before and after 4 weeks of acalabrutinib treatment. MCL cells were enriched from PBMC by immunomagnetic depletion. Created in BioRender. M. Zhang (2025); https://BioRender.com/rk6gi2x. (B) Single-cell mass distribution of more than 1000 cells measured by the SMR of two acalabrutinib-sensitive MCL primary samples (left) and two less-sensitive MCL primary samples (right) before and after 4 weeks of treatment with the BTKi acalabrutinib as outlined in workflow (A). The samples were categorized as sensitive or less sensitive based on changes in their white blood cell count following BTKi treatment (fig. S10). (C) Schematic workflow to examine the impact of BTK inhibition on the buoyant mass of MCL or CLL cells pre- and postacalabrutinib in vivo treatment, followed by 24 hours of ex vivo drug treatment with DMSO or acalabrutinib. Created with BioRender. One sample of DMSO-treated cells serves as the “reference” distribution. Mass response signals are calculated using Earth Mover’s Distance (EMD) analysis, comparing the reference to a second replicate of DMSO-treated cells (“Ref versus DMSO”) and to acalabrutinib-treated cells (“ref versus acalabrutinib”), respectively. A P value is determined by comparing the difference between these two mass response signals against a decision threshold. (D) Mass response signals of two acalabrutinib-sensitive and two less-sensitive MCL primary samples obtained (bottom), which were subjected to 24 hours of ex vivo drug treatment in duplicate with DMSO or acalabrutinib, as outlined in workflow (C). (E) Mass response signals of three CLL primary samples at pretreatment (top) and progression (bottom) time points, followed by 24 hours of ex vivo drug treatment in duplicate with DMSO or acalabrutinib as detailed in workflow (C). ns, nonsignificant, *P < 0.05 and **P < 0.01.