. 2025 Dec 23;44(12):116701.

doi: 10.1016/j.celrep.2025.116701. Epub 2025 Dec 15.

A cell-state axis underlying colonization in carcinomas with implications for metastasis risk prediction and interception

Affiliations

¹ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Sidney Kimmel Comprehensive Cancer Center, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA. Electronic address: jesse.stone.handler@emory.edu.
² Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, MD, USA.
³ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
⁴ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
⁵ Department of Biochemistry & Biophysics and Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, San Francisco, CA, USA; Arc Institute, Palo Alto, CA, USA.
⁶ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Sidney Kimmel Comprehensive Cancer Center, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Convergence Institute, Johns Hopkins Data Science and AI Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA; Department of Applied Mathematics and Statistics, Johns Hopkins University, Baltimore, MD, USA.
⁷ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Molecular Biology and Genetics, Department of Medicine, and Department of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA. Electronic address: kalhor@jhu.edu.

PMID: 41401069
PMCID: PMC12831586
DOI: 10.1016/j.celrep.2025.116701

A cell-state axis underlying colonization in carcinomas with implications for metastasis risk prediction and interception

Jesse S Handler et al. Cell Rep. 2025.

. 2025 Dec 23;44(12):116701.

doi: 10.1016/j.celrep.2025.116701. Epub 2025 Dec 15.

Authors

Affiliations

¹ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Sidney Kimmel Comprehensive Cancer Center, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA. Electronic address: jesse.stone.handler@emory.edu.
² Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, MD, USA.
³ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
⁴ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD, USA.
⁵ Department of Biochemistry & Biophysics and Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, San Francisco, CA, USA; Arc Institute, Palo Alto, CA, USA.
⁶ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Sidney Kimmel Comprehensive Cancer Center, Department of Oncology, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Convergence Institute, Johns Hopkins Data Science and AI Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21287, USA; Department of Applied Mathematics and Statistics, Johns Hopkins University, Baltimore, MD, USA.
⁷ Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Center for Epigenetics, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD, USA; Department of Molecular Biology and Genetics, Department of Medicine, and Department of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD, USA. Electronic address: kalhor@jhu.edu.

PMID: 41401069
PMCID: PMC12831586
DOI: 10.1016/j.celrep.2025.116701

Abstract

Metastasis to the liver drives mortality in pancreatic ductal adenocarcinoma (PDAC), yet mechanisms of colonization remain unclear. Using genomic barcoding, we developed a clonal competition model under immune surveillance, isolating murine PDAC subclones with high or low liver-colonization potential. Combined transcriptome and chromatin-accessibility analyses revealed a distinct "metastatic-potential axis," separate from the normal-to-PDAC and classical-basal axes. We established "MetScore" as a biomarker of this axis. MetScore distinguishes metastases from primary PDAC tumors in patients, predicts outcomes beyond classical-basal classifications, and generalizes across carcinoma subtypes, suggesting conserved colonization mechanisms. High-MetScore PDAC cells preferentially occupy immune cell-enriched niches, suggesting they remodel the metastatic microenvironment. Functional screening identified c-Fos as a positive mediator of colonization and a candidate anti-metastatic target. Collectively, we identify a cell-state axis underpinning PDAC liver colonization, introduce MetScore as a broadly applicable biomarker, and nominate actionable targets for peri-operative therapeutic intervention.

Keywords: CP: Cancer; PDAC metastasis; c-Fos; carcinoma metastasis pathways; intratumoral heterogeneity; lineage barcoding; metastasis biomarker; metastatic potential; pancreatic cancer; tumor cell fate; tumor microenvironment.

PubMed Disclaimer

Conflict of interest statement

Declaration of interests J.S.H. and R.K. are listed as co-inventors on a patent application on the use of this study’s findings for clinical metastasis risk prediction. E.J.F. serves on the Scientific Advisory Board of Resistance Bio, as a consultant for Mestag Therapeutics and Merck, and receives research funding from AbbVie Inc. and Roche/Genentech outside the scope of this work. H.G. has outside interest as a co-founder of Exai Bio, Tahoe Therapeutics, and Therna Therapeutics; serves on the board of directors at Exai Bio; and is a scientific advisory board member for Verge Genomics and Deep Forest Biosciences, which are outside the scope of this work.

Figures

**Figure 1.. Isolation of primary PDAC subclones with high and low liver-colonization potential**
(A) Schematic overview. (B) Representative gross pathology and H&E-stained formalin-fixed paraffin-embedded (FFPE) sections. Scale bars: 500 μm and 50 μm (insets). (C and D) (C) Histograms of tumor clonalities—the number of unique subclones per tumor. (D) Detection frequency of each subclone. Points, individual mice; bars, group means ± SEM. Sample sizes: KPC-1 liver, 27 metastases (three mice); KPC-1 peritoneum, 24 metastases (five mice); KPC-2 liver, 46 metastases (five mice); KPC-2 peritoneum, 60 metastases (five mice). (E) *In vitro* growth curves. Points represent the mean average of four technical replicates, with error bars representing the SEM. (F) Fraction of three 10-cm dishes in which each subclone was observed after 28 days of passaging.

**Figure 2.. Identification of genes defining met-high and met-low states**
(A) Averaged ATAC-seq signal for peaks with significantly greater accessibility in met-low (left) or met-high (right) subclones (FDR <0.05). Each line represents a subclone, colored by metastatic potential (red = high, blue = low). (B and C) Heatmaps showing scaled, log-transformed normalized (B) accessibility or (C) expression for differentially accessible peaks or differentially expressed genes, respectively (FDR < 0.05). Subclones clustered by Pearson correlation. (D) Scatterplot showing the relationship between differential chromatin accessibility (x axis) and differential gene expression (y axis) for each significant peak. Points are colored based on whether the nearest gene is differentially expressed. (E and F) Dot plots of enriched pathways among met-low (E) or met-high (F) genes (FDR < 0.05). (G) Heatmap of scaled, log-transformed, normalized expression for module genes across subclones. (H) Rank-ordered plot of differential binding scores (DBS) for significant TF motifs (n = 499; two-sided t test, Bonferroni-adjusted p < 0.05), with positive values indicating increased binding in met-high.

**Figure 3.. The metastatic-potential axis is independent of normal-to-PDAC and classical-basal axes**
(A) PCA of normalized ATAC-seq signal across a consensus peak set. (B) Predicted basal-like subtype probabilities based on PurIST, with values <0.5 indicating likely classical subtype. (C) Bar plots showing ssGSEA scores (left) and average transcripts per million (TPM) (right) for classical and basal marker genes across met-high and met-low subclones. Bars, means ± SEM. NS, not significant (p > 0.05; two-sided Wilcoxon rank-sum tests). (D) Heatmap of scaled, log-transformed, normalized expression for the marker genes used in (C) across subclones.

**Figure 4.. The metastatic-potential axis is conserved between mouse and human PDAC**
(A) Boxplots showing MetScore distributions in primary and metastatic tumors across PDAC cohorts. Points represent individual tumors. Two-sided Wilcoxon rank-sum tests; *p < 0.05, ***p < 0.001. (B) Dot plot of MetScores for matched primary and metastatic tumors from rapid autopsy patients in the PACA-US cohort. Lines connect samples from the same patient. Linear mixed-effects model; ***p < 0.001. (C) As in (A) but stratified by molecular subtype as annotated in Moffitt et al. (D) Violin plot showing MetScore distributions across tumor cells derived from primary tumors and metastases. Generalized linear model; ***p < 0.001. (E) Scatterplot showing the relationship between MetScore and scB/scC across tumor cells. Gray line, linear regression fit. Pearson’s r indicated. (F) Left: dot plot showing accuracy (median ± interquartile range) across 10-fold cross-validation for logistic-regression models predicting donor type (primary vs. metastasis) using either MetScore or scB/scC. One-sided one-sample proportion tests comparing model accuracy to the no-information rate; ***p < 0.001. NS, not significant. Right: PR curves for models trained on MetScore or scB/scC. (G) Top: UMAP plots of tumor cells colored as indicated. For MetScore and scB/scC, values above the 90th percentile and below the 10th percentile were capped. Bottom: bar plot showing the proportion of primary and metastasis-derived cells within each cluster. (H) Forest plot showing hazard ratios and 95% confidence intervals for high MetScore (top 50%) and basal subtype (PurIST predicted probability >0.5) with respect to overall survival in Cox proportional hazards models adjusted for both variables. *p < 0.05, **p < 0.01, ***p < 0.001. (I) Kaplan-Meier curve showing overall survival of patients pooled from cohorts in (H) stratified by MetScore and classical-basal subtype. p value from log rank test.

**Figure 5.. A shared cell-state axis underlies metastatic potential across multiple human carcinoma subtypes**
(A and B) Boxplots showing MetScore distributions in primary and metastatic tumors across the indicated carcinoma cohorts. Points represent individual tumors. Two-sided Wilcoxon rank-sum tests; *p < 0.05, ***p < 0.001. (C and D) Kaplan-Meier curves showing overall survival stratified by MetScore (high = top 50%, low = bottom 50%) in each cohort. p values from log rank tests. (E) Kaplan-Meier curves for overall survival in patients with stage II/III pMMR CMS2 COAD in the TCGA and CIT cohorts, stratified by receipt of post-operative therapy (yes vs. no) within low-MetScore (bottom 50%; left) and high-MetScore (top 50%; right) groups. p values from log rank tests.

**Figure 6.. Met-high PDAC cells enrich immune cells in their microenvironments**
(A) Scatterplots showing the relationship between average neoplastic cell MetScore and the abundance of stromal populations across tumors in the human PDAC scRNA-seq atlas. (B) Correlation between bulk tumor MetScore and the immune-stromal score, calculated as the difference between scaled immune and stromal scores from the ESTIMATE algorithm. (C) The same analysis restricted to metastases. (D) Scatterplot showing correlation between single-cell MetScore in tumor-enriched epithelial cells in the PDAC scRNA-seq atlas and a proliferation score. (E) UMAP plots of tumor cells colored as indicated. For MetScore and *MKI67*, values above the 90th percentile and below the 10th percentile were capped. (F) Scatterplot showing correlation between single-cell MetScore in tumor-enriched epithelial cells in the PDAC scRNA-seq atlas and a survival score. In (A–D) and (F), linear regression lines are shown, with Spearman’s rho and p values (FDR adjusted for A–C).

**Figure 7.. c-Fos is a positive functional mediator of PDAC liver colonization**
(A) Experimental design for (B). (B) Volcano plot showing mean log₂ fold change in shRNA abundance between primary tumor and liver metastasis conditions (averaged across all shRNAs per gene) and −log₁₀ weighted combined p value from a linear model. Sample sizes: primary (n = 6 tumors, 6 mice); liver metastasis (n = 36 tumors, 4 mice). (C) RT-qPCR quantification of *Fos* mRNA in KPC-2_HiB cells expressing shRNAs targeting *Fos* or a non-targeting control. *Rps29*, housekeeping gene. Bars, mean ± SEM from three technical replicates. One-sided Wilcoxon rank-sum test; *p < 0.05. (D) Histograms showing the distribution of shFos representation (shFos/(shFos + shControl)) in liver metastases following intrasplenic injection of mixed populations. Dashed lines indicate pre-injection (blue) and post-injection (red) medians. Left, shFos #1 (n = 32 tumors, 3 mice); right, shFos #2 (n = 56 tumors, 5 mice). p values from one-sided one-sample Wilcoxon signed-rank tests. (E) *In vitro* proliferation of KPC-2_HiB cells expressing *Fos*-targeting or control shRNAs. Points represent mean ± SEM from technical replicates (n = 4). Linear mixed-effects model; NS, not significant. (F) Scatterplot depicting the relationship between *FOS* expression and a proliferation gene signature across tumor cells in the human PDAC scRNA-seq atlas. A regression line and Spearman’s rho (0.10) are shown, indicating negligible correlation. (G) Volcano plot of differential expression between *Fos-*knockdown and control conditions in KPC-2_HiB cells. Two independent *Fos*-targeting shRNAs and one non-targeting control shRNA were used, each with three technical replicates. Genes colored by significance (red, upregulated; blue, downregulated; black, unchanged) with FDR < 0.05. (H) Heatmap of scaled, log-transformed, normalized expression for genes highlighted in (G) across technical replicates. (I) Kaplan-Meier curve for overall survival stratified by *IL1A* expression (high = top 50%; low = bottom 50%) in the PACA-US cohort. p value from log rank test. (J and K) CUT&RUN and ATAC-seq tracks in KPC-2_HiB cells at the *Il1a* (J) and *Il1rn* (K) loci. c-Fos CUT&RUN had two technical replicates; no-antibody CUT&RUN and ATAC-seq had one.

See this image and copyright information in PMC

Update of

Identifying a gene signature of metastatic potential by linking pre-metastatic state to ultimate metastatic fate.
Handler JS, Li Z, Dveirin RK, Fang W, Goodarzi H, Fertig EJ, Kalhor R. Handler JS, et al. bioRxiv [Preprint]. 2024 Aug 17:2024.08.14.607813. doi: 10.1101/2024.08.14.607813. bioRxiv. 2024. Update in: Cell Rep. 2025 Dec 23;44(12):116701. doi: 10.1016/j.celrep.2025.116701. PMID: 39185156 Free PMC article. Updated. Preprint.

References

1. Groot VP, Rezaee N, Wu W, Cameron JL, Fishman EK, Hruban RH, Weiss MJ, Zheng L, Wolfgang CL, and He J (2018). Patterns, Timing, and Predictors of Recurrence Following Pancreatectomy for Pancreatic Ductal Adenocarcinoma. Ann. Surg 267, 936–945. 10.1097/SLA.0000000000002234. - DOI - PubMed
1. Liu KH, Hung CY, Hsueh SW, Chang PH, Chen YY, Lu CH, Chen PT, Yeh KY, Huang PW, Tsang NM, et al. (2019). Lung Metastases in Patients with Stage IV Pancreatic Cancer: Prevalence, Risk Factors, and Survival Impact. J. Clin. Med 8, 1402. 10.3390/jcm8091402. - DOI - PMC - PubMed
1. Conroy T, Hammel P, Hebbar M, Ben Abdelghani M, Wei AC, Raoul JL, Choné L, Francois E, Artru P, Biagi JJ, et al. (2018). FOLFIRINOX or Gemcitabine as Adjuvant Therapy for Pancreatic Cancer. N. Engl. J. Med 379, 2395–2406. 10.1056/NEJMoa1809775. - DOI - PubMed
1. Bailey P, Chang DK, Nones K, Johns AL, Patch AM, Gingras MC, Miller DK, Christ AN, Bruxner TJC, Quinn MC, et al. (2016). Genomic analyses identify molecular subtypes of pancreatic cancer. Nature 531, 47–52. 10.1038/nature16965. - DOI - PubMed
1. Moffitt RA, Marayati R, Flate EL, Volmar KE, Loeza SGH, Hoadley KA, Rashid NU, Williams LA, Eaton SC, Chung AH, et al. (2015). Virtual microdissection identifies distinct tumor- and stroma-specific subtypes of pancreatic ductal adenocarcinoma. Nat. Genet 47, 1168–1178. 10.1038/ng.3398. - DOI - PMC - PubMed

MeSH terms

Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions

Save citation to file

Email citation

Add to Collections

Add to My Bibliography

Your saved search

Create a file for external citation management software

Your RSS Feed

A cell-state axis underlying colonization in carcinomas with implications for metastasis risk prediction and interception

Affiliations

A cell-state axis underlying colonization in carcinomas with implications for metastasis risk prediction and interception

Authors

Affiliations

Abstract

Conflict of interest statement

Figures

Update of

References

MeSH terms

Substances

Grants and funding

LinkOut - more resources

Full Text Sources

Medical

Research Materials

Miscellaneous