AUTHORS
Martin Bablok, Ella Yee2, Shami Chakrabarti, PhD3, Lingting Shi, PhD4, Christian A Gordillo, MSc5, David W. Harle6, Thomas McMahon-Skates7, Ajna Uzuni, MS8, Ximi K. Wang, MD9, Caroline Kam10, Elham Azizi11, Ran Reshef, MD, MSc12
AFFILIATIONS
1 Columbia University Irving Medical Center, New York, New York, United States, Institute of Anatomy, Medical Faculty, Ruhr University Bochum, Bochum, Germany
2 Columbia University, New York, New York, United States
3 Columbia University Irving Medical Center, New York, New York, United States
4 Columbia University, New York, New York, United States, Columbia University, New York, New York, United States
5 Columbia University Irving Medical Center, New York, New York, United States
6 Columbia University Irving Medical Center, New York, New York, United States
7 Columbia University Irving Medical Center, New York, New York, United States
8 Columbia University Irving Medical Center, New York, New York, United States
9 Columbia University Irving Medical Center, New York, New York, United States
10 Columbia University, New York, New York, United States
11 Columbia University, New York, New York, United States, Columbia University, New York, New York, United States
12 Columbia University Irving Medical Center, New York, New York, United States, Columbia University Irving Medical Center, New York, New York, United States
ABSTRACT
Gastrointestinal graft-versus-host disease (GI GVHD) is typically viewed as a T-cell–mediated epithelial injury syndrome. We previously characterized donor- and recipient-derived T-cell states in human GVHD biopsies (Shi et al., Nat Immunology, in press), but the contributions of other immune populations remain unclear. Donor plasma-cell reconstitution occurs late after transplant, whereas recipient plasma cells often survive conditioning. We investigated whether intestinal plasma cells undergo transcriptional, metabolic, and clonal remodeling after allogeneic hematopoietic cell transplantation (allo-HCT).
We performed single-cell RNA sequencing of 28 upper- and lower-GI biopsies from 14 patients with varying GVHD severity sampled on days 27–444 after transplantation and 2 normal donors. Clustering, differential expression, gene-set enrichment, and generalized linear models defined plasma-cell states and temporal changes. TRUST4 BCR reconstruction, trajectory, cell–cell communication, and spatial transcriptomic analyses assessed clonal relationships and cellular interactions.
B cells were nearly absent after allo-HCT, including in biopsies obtained over 1 year post-transplant despite recovery of peripheral blood B-cell counts, suggesting tissue-specific failure of donor B-cell reconstitution in the gut. Clustering identified three predominantly recipient-derived plasma-cell states: homeostatic secretory IgA+ cells (C0), early IgA+ cells (C1), and an aberrant stressed population (C2). Within the plasma-cell compartment, the relative abundance of C0 and C1 progressively declined over time post-transplant (C0: Spearman ρ=-0.877, p=0.0001; C1: ρ=-0.731, p=0.0045), whereas C2 accumulated over time (ρ=0.941, p<0.0001) in both upper and lower GI samples. C2 was nearly absent in healthy controls and was characterized by proteotoxic and endoplasmic-reticulum stress, unfolded protein response activation, and AP-1/NF-κB transcription factors (FOSB, JUN, NFKBIZ). Crucially, C2 exhibited a dual cytoprotective and immune-evasive program, upregulating the pro-survival factor MCL1, the caspase-8 inhibitor CFLAR (c-FLIP), and the granzyme B inhibitor SERPINB9, while simultaneously downregulating HLA Class I antigen-presentation machinery (HLA-A/B/C, B2M) and mitochondrial bioenergetics. Longitudinally, C0/C1 progressively lost electron-transport programs, whereas C2 showed declining apoptotic and antigen-processing pathways over time. Together, these adaptations define C2 as a stress-adapted, antigenically cloaked recipient plasma-cell state capable of persisting under chronic tissue injury and actively resisting alloreactive cytotoxicity by donor T cells.
BCR analysis showed clone sharing among plasma-cell states. C2 shared more clones with C0/C1 in severe GVHD than in no or mild GVHD (median 12.1% vs 4.0%; p=0.028), and trajectory analysis suggested that the transition of homeostatic IgA+ plasma cells into the stressed C2 state is a potential hallmark of severe disease. Severe GVHD was also associated with increased clonal expansion (Gini index 0.158 vs 0.077 in severe vs ND/no/mild; p=0.013), independent of time after transplantation, indicating that the plasma-cell repertoire became increasingly dominated by expanded clones.
Cell–cell communication analysis with MultiNicheNet identified increased plasma-cell interactions with myeloid and T-cell populations in severe GVHD. C2 uniquely expressed IL15 and showed interactions involving IL-15 with T cells, myeloid cells, fibroblasts and endothelial cells. Notably, interactions between C2 and cytotoxic unconventional CD8+ T cells were significantly enriched in severe GVHD, suggesting that C2 may contribute to persistent tissue injury by sustaining inflammatory crosstalk. Preliminary lower-GI spatial transcriptomics showed closer plasma cell–stem-cell hub proximity in severe GVHD and increased interactions in crypt-loss regions, linking plasma cells to epithelial injury and regeneration.
In summary, GI GVHD is associated with a progressive shift of persistent recipient intestinal plasma cells from homeostatic IgA+ states toward an aberrant, stress-adapted, clonally skewed population. Its transcriptional programs and immune–stromal interactions suggest resistance to donor T-cell–mediated removal and a role in sustaining intestinal injury and remodeling, identifying plasma-cell reprogramming as an underrecognized component of the GI GVHD microenvironment.
DISCLOSURES
Martin Bablok: No relevant conflicts of interest to declare.