Does RhoA Y42C drive immune evasion in diffuse gastric cancer?

RhoA Y42C usually appears without CDH1 loss

The question matters because diffuse gastric cancer (DGC) is an aggressive subtype with poor prognosis, and RHOA Y42C recurs in it while its ability to act on its own was unclear. The study first checked whether the mutation depends on the other classic DGC driver, CDH1 loss, which removes E-cadherin. Sequencing 431 DGC cases showed that most RHOA mutations occurred without CDH1 alterations: 81.0% (47/58) of RHOA-mutant DGC cases had no CDH1 change, and for Y42C specifically the figure was 83.3% (10/12). Across all gastric cancers in the cohort, 83.2% (84/101) of RHOA-mutant cases and 89.5% (17/19) of Y42C cases lacked CDH1 mutation.

That co-occurrence pattern is an association, not proof of independence. The authors used it to justify a functional test: build gastric organoids carrying Rhoa Y42C with Kras G12D and Trp53 loss (RKP) but with intact Cdh1, and ask whether tumors form. The comparison line is EKP, which carries Cdh1 loss plus Kras G12D and Trp53 loss and represents E-cadherin-loss-driven DGC.

RKP organoids form mucinous DGC-like tumors

Genetically engineered murine gastric organoids were transplanted into immunocompetent C57BL/6 mice. RKP tumors developed without E-cadherin inactivation and showed signet-ring-cell-like morphology with mucin production, features the paper links to human DGC. Both EKP and RKP tumors mixed intestinal- and diffuse-type morphology, but immune cells were relatively rare in RKP tumors while EKP tumors showed substantial immune infiltration.

Growth speed is a separate measure and should not be read as RKP superiority: EKP tumors grew faster than RKP both in vitro and in vivo. The authors explicitly note that higher proliferation markers in RKP single-cell data do not mean faster overall growth, because samples were collected at different time points; the signal instead reflected lineage-specific proliferation. Human datasets also showed increased AQP5, MUC1 and MUC4 expression in RHOA Y42C-mutant tumors compared with wild-type cases, supporting the resemblance rather than establishing it causally.

An Aqp5-to-Muc1/Muc4 lineage is RKP-specific

Single-cell RNA sequencing and Xenium in situ spatial transcriptomics were combined to trace where tumor cells come from. RNA-velocity analysis placed the primary RKP lineage at Aqp5+ cells that differentiate into Muc1/Muc4+ cells, whereas EKP lineages stemmed from Runx1+ cells and generated Zeb2/Arg1+ cells. RKP tumors also contained a less prominent Runx1+-derived lineage shared with EKP.

The paper interprets Aqp5+ cells as more plastic: on a Waddington-landscape representation they sat at a higher valley-ridge score than Runx1+ cells and showed stronger proliferation-marker expression (Top2a, Mki67) with higher S or G2/M scores. Spatial analysis added direction: EKP Runx1+ cells ran cortex-to-core while Krt8+ cells ran core-to-cortex, whereas the RKP Aqp5+-derived trajectory spread from a small region outward without a cortex-core pattern. These are inferred trajectories from expression and spatial data, not direct lineage tracing.

RhoA Y42C and YAP1 raise AREG to exhaust T cells

The mechanism proposed runs from RhoA Y42C to YAP1 to AREG to T cell exhaustion. RKP-specific Muc1/4+ clusters were enriched for Hippo signaling, and Yap1 expression and Hippo pathway scores were highest in the Muc1/4+ lineage relative to the Zeb2/Arg1+ lineage. Because RhoA-ROCK signaling can activate HIPPO/YAP and RhoA Y42C hyperactivates RhoA signaling, the authors tested whether YAP1 transactivates AREG, an EGFR ligand.

Functional evidence supports the chain: Areg expression was elevated in RKP tumor cells, AREG immunostaining was quantified relative to DAPI+ cells (EKP n=6, RKP n=4 biological replicates), and CUT&RUN sequencing showed peaks on the Areg locus. Recombinant AREG induced Pdcd1 (PD-1) in exhausted OT-I T cells and in CTLL-2 cells, and an AREG-neutralizing antibody was used in co-culture experiments. In human DGC samples, RHOA Y42C correlated with PD-1 upregulation. The causal wording is the paper's; the human correlation alone does not establish the mechanism in patients.

PD-1 blockade works in RKP but not EKP-like tumors

The therapeutic comparison is the paper's most decision-relevant result. RKP tumors were highly susceptible to PD-1 blockade, whereas E-cadherin-loss-associated EKP tumors were not. This separates two DGC scenarios that share a diagnosis but differ in driver: the R subtype, E-cadherin-positive with activated RhoA signaling, versus the E subtype with E-cadherin inactivation and EZH2 activation.

The authors address a conflicting prior report of RhoA-mutant tumor resistance to PD-1 blockade and sensitivity to CTLA4 inhibition. They attribute the discrepancy to model differences: the earlier work used the MC38 colon adenocarcinoma cell line, while this study used gastric organoids with minimal genetic manipulation. That is the authors' explanation, not an independently tested resolution, so the susceptibility claim should be tied to the RKP organoid-derived model rather than all RhoA-mutant cancers.

Where the RKP conclusion stops

Several design features bound the result. Tumors were grown subcutaneously, and the authors note that immune landscapes in the stomach may differ. Before transplantation, organoid cells were expanded in 2D culture to obtain sufficient cell numbers, and the 3D-to-2D transition could affect signaling, drug response, cell-cell interactions or secretory profiles; the authors call for organoid-based transplantation or genetically modified mouse models to confirm. T cell analyses were also limited by low cell numbers in single-cell data and limited gene coverage in spatial transcriptomics, although RKP T cells resembled the human DGC-R subtype dataset.

The practical priority follows from the internal comparison rather than from DGC as a whole: within this study, the RKP genotype, not E-cadherin loss, marked the tumors that responded to PD-1 blockade, and the proposed biomarker logic rests on RHOA Y42C status plus the AREG-YAP1-PD-1 axis. Testing PD-1 inhibition in patients selected by RHOA Y42C remains a hypothesis generated by mouse organoid models, not a demonstrated clinical benefit.

Sources used in this answer