henhaobio
Sign in

Nature CommunicationsImmunologyPeer-reviewed journal article

A Paradoxical Tumor Antigen Specific Response in the Liver

Rajiv TrehaniD, Patrick HuangiD, Xiao Bin Zhu, Xin WangiD, Marlaine Soliman, Dillon StrepayiD, Amran Nur, Noémi KedeiiD, Martin ArhiniD, Shadin GhabraiD, Francisco Rodríguez-MatosiD, Mohamed-Reda BenmebarekiD, Chi MaiD, Firouzeh KorangyiD, Tim F. GreteniD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Functional tumor-specific CD8+ T cells are essential for an effective anti-tumor immune response and the efficacy of immune checkpoint inhibitor therapy. In comparison to other organ sites, we found higher numbers of tumor-specific CD8+ T cells in primary, metastatic liver tumors in murine tumor models. Despite their abundance, CD8+ T cells in the liver displayed an exhausted phenotype. Depletion of CD8+ T cells showed that liver tumor-reactive CD8+ T failed to control liver tumors but was effective against subcutaneous tumors. Similarly, analysis of single-cell RNA sequencing data from patients showed a higher frequency of exhausted tumor-reactive CD8+ T cells in liver metastasis compared to paired primary colon cancer. High-dimensional, multi-omic analysis combining proteomic CODEX and scRNA-seq data revealed enriched interaction of SPP1+ macrophages and CD8+ tumor-reactive T cells in profibrotic, alpha-SMA rich regions in the liver. Liver tumors grew less in Spp1-/- mice and the tumor-specific CD8+ T cells were less exhausted. Differential pseudotime trajectory inference analysis revealed extrahepatic signaling promoting an intermediate cell (IC) population in the liver, characterized by co-expression of VISG4, CSF1R, CD163, TGF-{beta}R, IL-6R, SPP1. scRNA-seq of a third data set of premetastatic adenocarcinoma showed that enrichment of this population may predict liver metastasis. Our data suggests a mechanism by which extrahepatic tumors facilitate the formation of liver metastasis by promoting an IC population inhibiting tumor-reactive CD8+ T cell function.

In-depth summary

Claude Sonnet 4.6 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (24 Sept 2024) · CC0 1.0

Liver tumors in mice and patients hold many tumor-specific CD8+ T cells that are exhausted; SPP1+ macrophages and a liver-specific intermediate macrophage population are linked to this dysfunction.

Patients with liver metastases fare worse on immune checkpoint inhibitors than those without, and the paper notes that SPP1 macrophages have been proposed bioinformatically to differ from M1/M2 macrophages but lack mechanistic study. It links SPP1+ macrophages to tumor-reactive CD8+ T cell dysfunction in the liver through a mechanism it presents as independent of M2 macrophages.

Key results

  1. ShownMouse CT26 tumors in the liver carried the highest frequency of tumor-antigen-specific (AH1 tetramer+) CD8+ T cells, yet grew faster than subcutaneous tumors.

    Day 14: higher AH1+ frequency in intrahepatic vs subcutaneous TIL (p = 0.0021) with similar tumor weights (p = 0.3792). Day 10: 12.45% intrahepatic vs 3.75% subcutaneous TIL. Day 21 weights: 1.2495 g intrahepatic vs 0.28186 g subcutaneous (p=0.0452). B16F10 with a TRP2 tetramer gave similar results. · A paradoxical high frequency of…; Figure legends

  2. SuggestedLiver-resident tumor-specific CD8+ T cells in mice were exhausted and could not control tumors in the liver, whereas CD8+ T cells did control subcutaneous tumors.

    CD39+ among intrahepatic TAS cells was higher than in subcutaneous TIL at day 14 (p = 0.0287) and day 21 (p = 0.013); 89.7% of intrahepatic TILs were CD39+ at day 21. Anti-CD8 depletion left intrahepatic tumors unchanged but enlarged subcutaneous tumors (p=0.0011). · Phenotype and function of TAS CD8+ T cells; Figure legends

  3. ShownIn public scRNA-seq from paired colorectal cancer patients, potentially tumor-reactive CD8+ T cells were most frequent in liver metastases and mostly exhausted; CellChat predicted stronger SPP1/FN1 macrophage signalling to them than in primary tumors.

    304,816 cells from 10 patients; 1,310 cells passed the stringent scGSEA cutoff; 66.58% of intrahepatic pTRT were in the exhausted LAYN cluster. SPP1 and FN1 ligands to CD44, ITGA4_ITGB1 and ITGA4_ITGB7 were enriched in liver metastasis vs primary tumor. · SPP1+ macrophages interact with pTRT cells in…; Figure legends

  4. ShownSpp1 knockout mice had less exhausted tumor-specific CD8+ T cells and smaller liver tumors.

    B16F10 intrahepatic tumors: lower CD39 and TIM-3 on TRP2+ but not TRP2- CD8+ T cells in Spp1 KO (n=6) vs WT (n=7). RIL-175 HCC tumor weights were significantly lower in KO vs WT (n=5 each). · SPP1+ macrophages cause pTRT cell dysfunction; Figure legends

  5. SuggestedPseudotime analysis of liver macrophages identified an intermediate cell (IC) population between Kupffer cells and SPP1+ macrophages; its score was enriched in liver biopsies from pancreatic cancer patients who later developed liver metastasis.

    4,833 liver macrophage/monocyte cells; SPP1 Moran's I 0.72 in liver vs 0.29 in colon (1,330 cells). Pancreatic data: 26,627 cells, 2,345 CD68+ cells. Individually VSIG4, CSF1R and SPP1 were not highly predictive. · Pseudotemporal Analysis Reveals a Liver-Specific…; Figure legends

Methods, data and reproducibility

Combined flow cytometry in mouse tumor models (including Spp1 knockout mice), CD8 depletion, and computational analysis of three human scRNA-seq datasets (scGSEA, CellChat, pseudotime) with CODEX imaging of human HCC.

  • Mouse tumor models: CT26 colorectal (AH1 tetramer), B16F10 melanoma (TRP2 tetramer), RIL-175 HCC; intrahepatic, subcutaneous and tail-vein (lung) injection · mouse · n = 4 to 9 mice per group in most experiments
  • Spp1 knockout vs wild-type mice with intrahepatic B16F10 or RIL-175 tumors · mouse · n = B16F10: 6 KO vs 7 WT; RIL-175: 5 vs 5
  • Public processed scRNA-seq of primary colorectal tumor, adjacent colon, liver metastasis, adjacent liver, lymph node and PBMC · human · n = 10 patients; 304,816 cells in 78 clusters
  • Authors' prior HCC CODEX (37-plex) imaging and scRNA-seq · human · n = 15 regions in proximity analyses
  • Published scRNA-seq of non-tumor liver biopsies from pancreatic adenocarcinoma patients · human · n = 26,627 cells after QC; 2,345 CD68+ cells
  • Bulk TCGA data scored with ssGSEA for IC signature vs survival (HCC and colon cancer) · human (TCGA)
Methods & tools
CellChat, scGSEA, ssGSEA, pseudotime trajectory analysis (Moran's I), CODEX multiplexed imaging, MHC class I tetramers (AH1, TRP2), flow cytometry, RT-qPCR, TCGA
Reproducibility gaps
Data availability statement says only that data are in the main text or supplement; no code repository, accession or software versions given.
Limitations (5)
  • Human scRNA-seq sample sizes were modest, and survival stratification was not possible because such data were not available. · stated by the authors
  • The effect of immune checkpoint blockade on SPP1+ macrophage and CD8+ T cell interactions was not tested. · stated by the authors
  • Mechanistic data linking the profibrotic proteins to liver fibrosis were limited. · stated by the authors
  • The subcutaneous anti-CD8 group is listed as n=1 in the legend, so the depletion contrast rests on very little data for that arm. · noted in this summary
  • Links from primary-tumor cytokines to liver SPP1/FN1 and the Kupffer cell to IC to SPP1 ordering rest on correlations across 10 patients and pseudotime inference, without direct causal tests. · noted in this summary

Read this if you study CD8+ T cell exhaustion in liver metastasis or want an example linking SPP1+ macrophages to T cell dysfunction using mouse models and public scRNA-seq.

Citations
81
FWCI
34.41
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISKazi '26Madani '23Lee '25Rahim '22Wan '25Prawira '24
76% semantic matchCancer Biology
Amira Z Kazi, Yingying Cao, Chirayu Mohindroo+8 more·bioRxiv·Jul 31, 2026