AUTHORS
Arghya Ray, PhD, Sindhu Pillai, PhD2, Kenneth Wen, BS3, William Love, BSc4, Vincent van Hoef, PhD5, Simon Fredriksson, PhD6, Ji Hyun Lee, PhD7, Kenneth Anderson, MD8, Paul G. Richardson, MD9
AFFILIATIONS
1 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States
2 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States
3 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States
4 Pixelgen Technologies,, Stockholm, Stockholm, Stockholm County 113 68, Sweden
5 Pixelgen Technologies,, Stockholm, Stockholm, Stockholm County 113 68, Sweden
6 Pixelgen Technologies,, Stockholm, Stockholm, Stockholm County 113 68, Sweden
7 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States, Dong-A University College of Medicine, Busan, 49201, Republic of Korea, Busan,, Korea, Rep. of South
8 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States
9 Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts, United States
ABSTRACT
Introduction: Despite significant recent progress with novel therapies, multiple myeloma (MM) remains incurable, highlighting the need for continued study of factors mediating disease progression. In the bone marrow (BM) tumor microenvironment (TME), interactions between clonal plasma cells (PCs) and immune cells are complex. The spatial organization of cell surface proteins governs cell-to-cell interactions and triggers different immune signaling cascades contributing to MM pathogenesis. We utilized the Proximity Network Assay (PNA) (Karlsson et al. 2025. bioRxiv doi:10.1101/2025.06.19.660329) to characterize the spatial organization of membrane proteins on the immune cells of the MM BM-TME and how cell surface architecture correlates with disease progression.
Methods: BM samples from MM patients with relapsed/refractory disease (RRMM: n=3) and newly diagnosed disease (NDMM: n=2) were collected at Dana-Farber Cancer Institute (DFCI) under DFCI protocol 07-150. Normal BM samples (n=3) were obtained from Stem Express (MA) as controls. BM mononuclear cells were analyzed with PNA (Pixelgen Technologies AB, Sweden). PNA is an optics-free DNA-based method that constructs 3D nanoscale maps of 155 surface proteins in single cells: DNA-barcoded antibodies are added to single cells, and an in situ rolling circle amplification generates >40,000 spatial nodes per cell, subsequently linked by proximity-dependent gap-fill ligation. After DNA sequencing, single-cell proximity networks were reconstructed using Pixelator.
Results: PNA captured protein abundance, colocalization and self-clustering at ~50 nm resolution for 155 surface proteins in 10,846 single cells (mean detected molecules/cell: normal 30,669, ND 33,515, RR 36,499). Abundance-based clustering identified 13 populations: PCs, conventional B, CD4+T, CD8+T, NK, NKT, monocytes, pDCs, neutrophils, platelets, other B and T cells, and other immune cells. The PCs were distinguished from non-PCs by the high abundance of CD319/SLAMF7. PC proportions were highest in RR (12.2% vs ND 5.1% vs normal 0.25%; P=.0152, ND vs normal). Differential abundance analysis in PCs showed immunosuppressive CD200/OX2 (1.8-fold), CD274/PD-L1 (1.42-fold), CD229/Ly9 (1.24-fold), and several SLAMF proteins with high abundance in RR vs ND (adj. P<.05). CD47, the tumor cell's "do not eat me" signal, showed a high abundance in RR vs ND (adj. P<.05). CD8+T showed high abundance of KLRG1 and CD38 in ND vs Normal (adj. P<.05), indicative of an exhausted phenotype. At the same time, low CD27 abundance in CD8+T in ND vs normal indicated poor prognosis. CD4+T cells low CD45RA and high CD45RO abundance in RR vs ND suggested a switch from naive to memory phenotype indicative of potentially dysregulated immune responses (adj. P<.05). NK inhibitory receptors CD159a and CD328/Siglec-7 abundance was in this order, RR>ND>normal (adj. P<.05). High Siglec-7 also correlates with MM hypersialylation. A low abundance of CD16+NK in RR vs ND suggests impaired immune function.
The Proximity Score (PS) metric revealed substantial changes in self-clustering and pairwise colocalization across most cell types. RR vs ND PCs show a decreased colocalization of VLA-4 complex, tetraspanins, adhesion molecules (ICAM-1/3), and GPI-anchored proteins suggesting increasingly disorganized membranes during disease progression. RR vs ND CD8+T cells show loss of colocalized lipid raft proteins and tetraspanin-enriched microdomains, indicating dysfunctional membrane organization. In addition, CD38 shows reduced self-clustering suggesting changes to the immunological synapse that governs cell-cell interactions. The complement receptors CD35/CR1, CD21/CR2 colocalized less in RR vs ND B cells, implicating a switch from functional cooperativity in the complement cascade to progressively poor innate and adaptive immunity in MM. Cell-cell interaction analysis suggested CD8+T/PC doublets in RRMM (4% vs normal 0.8%; adj. P<.05).
Conclusions: PNA resolved abundance and nanoscale spatial organization of 155 surface proteins across 13 BM populations, showing higher immunosuppressive protein abundance on RR PCs and progressively disorganized membranes in RR PCs, CD8+T and B cells. This preliminary evaluation suggests PNA can better characterize the MM TME and so potentially help development of immunotherapies and novel agents. Analysis continues in an expanded cohort (n=20) across both settings to further elucidate these observations.
DISCLOSURES
Arghya Ray, PhD: No relevant conflicts of interest to declare.