The research team, led by Gersende Caron and Thierry Fest of the University of Rennes, established a laboratory system that coaxes purified B cells from human tonsils down the full path to antibody secretion. The culture proceeds in two stages. First, B cells sit on stromal cells engineered to display CD40L, bathed in a cocktail of the cytokines BAFF, IL-2, IL-4, and IL-21, which mimics the help that T cells provide inside germinal centers. After six days of proliferation, the cells are transferred onto a different feeder layer with IL-6, pushing survivors toward terminal differentiation. Flow cytometry confirmed that the system genuinely reproduces physiology, generating CD20-low, CD38-high, CD138-positive cells with the hallmarks of mature plasma cells.

To watch the transition unfold molecule by molecule, the researchers performed single-cell RNA sequencing at days zero, six, eight, and ten, profiling 17,220 cells from three donors. Unsupervised clustering resolved twelve transcriptionally distinct populations spanning naive B cells, germinal center states, proliferating activated cells, and antibody-secreting cells. When the team reconstructed developmental trajectories with pseudotime analysis, a striking pattern emerged: day eight marked a fork in the road. One branch led to plasma cells, a second to memory B cells, and a third to a germinal-center-like state. Crucially, the cells sitting at the branch point had not yet turned on the canonical plasma cell regulators PRDM1 and IRF4, meaning they were transcriptionally primed for the secretory fate before committing to it.

To capture the coordinated gene programs behind this priming, the investigators applied high-dimensional weighted gene co-expression network analysis, which grouped thousands of genes into thirteen modules based on their correlated expression across single cells. One module, designated M3, stood out. It sat between germinal-center and mature plasma cell modules in the network, overlapped heavily with genes defining the plasma cell trajectory, and was rich in targets of XBP1, the master transcription factor of the unfolded protein response that drives endoplasmic reticulum expansion. Seven of its ten hub genes matched the pseudotime-defined plasma cell fate, and one of the most highly expressed members of the module was CAV1, the gene encoding Caveolin-1.

Chromatin immunoprecipitation experiments confirmed that this was no coincidence. XBP1, the spliced form activated during ER stress, bound directly to a site roughly 450 base pairs upstream of the CAV1 transcription start site in differentiating cells, with enrichment comparable to that seen at HSPA5, a well-established XBP1 target. CAV1 transcripts, undetectable in resting B cells, rose sharply along the plasma cell trajectory, reaching peak levels in both proliferating plasmablasts and mature plasma cells. Independent single-cell datasets from memory-to-plasma-cell differentiation reproduced the same induction, and surveys of public atlases showed CAV1 expression in tonsillar plasmablasts and bone marrow plasma cells, even though it is essentially absent from immune cells elsewhere.

The story took an intriguing turn when the team examined protein isoforms. Western blotting revealed that in vitro-generated plasma cells exclusively expressed the beta isoform of Caveolin-1, a roughly 19-kilodalton splice variant lacking the tyrosine-14 phosphorylation site that features prominently in classical caveolin signaling. Ex vivo, plasmablasts expressed only the beta isoform, while mature bone marrow plasma cells added low levels of the alpha transcript. Strikingly, mouse plasma cells show no Cav1 expression at all, making this a human-specific feature of antibody-secreting cells. Immunofluorescence microscopy showed the protein sitting predominantly inside the cell, co-localizing with the Golgi marker GM130 rather than with the endoplasmic reticulum, mitochondria, or lysosomes, and neither activated B cells nor plasmablasts expressed the Cavin proteins and EHD2 needed to build true caveolae. Human plasma cells, in other words, deploy a non-caveolar Caveolin.

Single-cell and bulk RNA-seq data revealed that CAV1 expression in B cells is actually biphasic. Within twelve to twenty-four hours of activation, a transient pulse of CAV1 appears, driven primarily by CD40L signaling through the transcription factor Ikaros. When the researchers degraded Ikaros pharmacologically with the molecular glue Mezigdomide, the early pulse collapsed. This initial wave is functionally silent as far as the team could detect: silencing CAV1 did not alter BCR-proximal phosphorylation cascades, STAT and AKT signaling, T cell-B cell conjugate formation, proliferation, or plasmablast differentiation. The second wave, sustained and XBP1-dependent, accompanies plasma cell commitment and persists through terminal differentiation.

Functional perturbation showed that Caveolin-1β is a fine-tuner rather than a master switch. Silencing CAV1 in pre-plasmablasts slightly but reproducibly increased the yield of CD138-positive plasma cells without changing total immunoglobulin secretion. Transcriptomic profiling of the knockdown cells revealed only modest shifts, but among them were meaningful signals: integrin alpha-6, encoded by ITGA6 and implicated in retaining plasma cells within the bone marrow niche, went up, while ID3, whose repression marks plasma cell commitment, went down. Co-immunoprecipitation detected an association between Cav-1 and the autophagy protein Atg5, though autophagic flux assays detected no functional change, and XBP1s levels were unaffected by the knockdown even under drug-induced ER stress.

The most revealing results came from proteomics. Affinity purification followed by mass spectrometry in a multiple myeloma cell line identified 348 proteins physically associated with Cav-1β, enriched for components of the endoplasmic reticulum, Golgi, mitochondria, lysosomes, and secretory machinery. Among the interactors were twenty-eight proteins encoded by XBP1 target genes from module M3, including the oligosaccharyltransferase complex and the SEC61 translocon, placing Cav-1β squarely within the secretory remodeling program. Even more striking was the overlap with nutrient-sensing machinery: the RAGulator complex, Rag GTPases, V-ATPase subunits, the amino acid transporter CD98/LAT1, and the XBP1-regulated chaperone HYOU1, which helps activate mTORC1. Cells expressing CAV1 were indeed enriched for mTORC1 transcriptional signatures, yet the interactions did not translate into regulation. Depleting Cav-1 left acute amino-acid-induced mTORC1 activation, RAGulator assembly, and RAPTOR recruitment intact, arguing that Cav-1β functions as a spatial scaffold rather than a signaling catalyst.

The authors propose a model in which Cav-1β acts as a structural organizer of the expanding plasma cell cytoplasm, tethering ER, Golgi, lysosomal, and metabolic networks into a coordinated architecture that balances secretory output with nutrient sensing. Whether this scaffolding matters for the long-term survival of the bone marrow plasma cells that maintain antibody immunity for decades remains an open question, as does its relevance to multiple myeloma and autoimmune diseases driven by pathological antibody-secreting cells. The mouse-human divergence is itself a caution: the tools of murine immunology may simply not see this layer of plasma cell biology. For now, the study adds a previously hidden membrane scaffold to the blueprint of the human antibody factory and demonstrates, once again, how single-cell maps can expose molecular players invisible to bulk approaches.

Subject of Research: Single-cell analysis of human B cell differentiation identifying a commitment checkpoint and XBP1-driven Caveolin-1β expression associated with mTORC1 machinery in plasma cells

Article Title: Single-cell mapping of human plasma cell fate reveals a commitment checkpoint and Cav-1β expression associated with mTORC1 machinery

Article References: Caron, G., Desmots, F., Chabaud, M., Lemarié, M., Olivier, V., Chatonnet, F., Nonn, C., Jadeau, C., Léonard, S., Monvoisin, C., Santamaria, K., Blouin, C. M., Attaf, N., Gharsalli, T., Milpied, P., Lamaze, C., Pangault, C., Haas, M., & Fest, T. (2026). Single-cell mapping of human plasma cell fate reveals a commitment checkpoint and Cav-1β expression associated with mTORC1 machinery. iScience, 29(11), Article 117865. https://doi.org/10.1016/j.isci.2026.117865

Image Credits: AI Generated

DOI: Not provided

Keywords: plasma cells, B cell differentiation, Caveolin-1, XBP1, mTORC1, single-cell RNA sequencing, unfolded protein response, endoplasmic reticulum, germinal center, RAGulator, antibody secretion, proteomics