Paper Details

ABSTRACT

Background: Myeloma plasma cells (PCs) are fragile and can be a small minority of bone marrow mononuclear cells (BMMCs). Conventional single-cell RNA sequencing involves fluorescence-activated cell sorting or magnetic-bead isolation followed by transfer to a separate barcoding platform, introducing opportunities for sample loss and technical noise at each step. Selective singulation consolidates marker-directed isolation and barcoding into one step using a novel bead chemistry, with target recognition, transcript capture, and barcoding on one substrate. 

Methods: Diagnostic aspirates from patients with newly diagnosed or relapsed multiple myeloma were processed to BMMCs by Ficoll gradient centrifugation and cryopreserved in liquid nitrogen, then singulated against BCMA or CD138 with CD45/EpCAM non-target and healthy-donor PBMC negative controls. Barcoded transcripts were processed to Illumina libraries and analyzed in replicate at pseudobulk and single-cell level. Using separate aliquots from the same bone marrow aspirates, fresh CD138? plasma cells were enriched by magnetic-bead selection and subjected to bulk RNA and whole-exome sequencing (WES).

Results: Ten libraries were generated from 3 myeloma patients and a healthy-donor PBMC control. Aspirate PC frequency by clinical flow cytometry ranged 1-15%. Libraries yielded 1,203-4,480 median UMIs and 571-890 median genes per cell, with 1.87% median mitochondrial reads.

Target cells were successfully recovered from all samples, including an aspirate containing only 1% PCs, achieving recovery rates between 15% and 46%. Direct comparisons using matched inputs demonstrated that BCMA-directed singulation recovered 2 to 3-fold more PCs than CD138 singulation, while exhibiting superior selectivity. Nearly 100% of BCMA-singulated cells were annotated as PCs by CellTypist automated cell annotation software, whereas non-target singulation (CD45/EpCAM) yielded only 0–6% of the initial input. Furthermore, transcriptional profiles were highly consistent across technical replicates and between BCMA- and CD138-singulated cells from the same patient, indicating that the recovered molecular signatures are independent of the specific singulation target. To validate the genomic fidelity of the captured populations, copy number variations inferred from the scRNA-seq data were cross-referenced against matched bulk WES. Single-cell inferred profiles demonstrated up to 52% gene-level concordance with the bulk WES, confirming that the isolated cells accurately reflect the underlying tumor genetics.

Intratumoral architecture differed across patients: one resolved into discrete subpopulations distinguished by cell-cycle arrest, secretory and adhesion-remodeling programs; a second formed a continuum from a secretory core toward cytoskeletal-remodeling and low-proliferation extremes. Structure was replicate- and depth-independent.

Conclusions: Consolidating isolation and barcoding into one bead-chemistry step recovered highly pure transcriptionally intact myeloma PCs from limited bone marrow input, and head-to-head comparison identified BCMA as the preferred capture target. Discrete subclonal structure and continuous plasticity have each been described in myeloma; detecting both in identically assayed patients indicates the method resolves architecture inaccessible to bulk profiling. Because profiles recovered from as little as 1% aspirate PC content carry cytogenetic, copy-number, and program-level information, we anticipate the approach will support clinical interpretation where material is limiting, such as profiling myeloma precursor conditions or minimal residual disease, subtype assessment, and detection of minor treatment-relevant subpopulations. Concordance with clinical annotation is being assessed in the ongoing cohort.