Skip to main navigation Skip to main content

Clin Mol Hepatol : Clinical and Molecular Hepatology

OPEN ACCESS
Search
Advanced Search
ABOUT
BROWSE ARTICLES
FOR CONTRIBUTORS

Articles

Page Path

Original Article

COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling

Jianxing Zhang1,2,*, Liang Qiao1,2,*, Zongfeng Wu1,2,*, Dinglan Zuo1,2, Shanshan Huang1,2, Shaoru Liu1,2, Zhenkun Huang1,2, Yi Zeng1,2, Yu Li1,2, Yichuan Yuan1,2, Chenwei Wang1,2, Wei He1,2, Jiliang Qiu1,2, Yunfei Yuan1,2orcid, Yi Niu1,2orcid, Binkui Li1,2orcid
Clinical and Molecular Hepatology 2026;32(2):772-786.
Published online: January 20, 2026

1State Key Laboratory of Oncology in South China, Guangdong Provincial Clinical Research for Cancer, Sun Yat-sen University Cancer Center, Guangzhou, China

2Department of Liver Surgery, Sun Yat-sen University Cancer Center, Guangzhou, China

Corresponding author : Binkui Li Department of Liver Surgery, Sun Yat-sen University Cancer Center, 651 Dongfeng Road E, Guangzhou 510060, China Tel: +86-20-8734-3951, Fax: +86-20-8734-3951, E-mail: libk@sysucc.org.cn
Yi Niu State Key Laboratory of Oncology in South China, Guangdong Provincial Clinical Research for Cancer, Sun Yat-sen University Cancer Center, 651 Dongfeng Road E, Guangzhou 510060, China Tel: +86-20-8734-2651, Fax: +86-20-8734-2651, E-mail: niuyi@sysucc.org.cn
Yunfei Yuan Department of Liver Surgery, Sun Yat-sen University Cancer Center, 651 Dongfeng Road E, Guangzhou 510060, China Tel: +86-20-8734-3118, Fax: +86-20-8734-3118, E-mail: yuanyf@mail.sysu.edu.cn

These authors contributed equally.


Editor: Ji Won Han, The Catholic University of Korea, Korea

• Received: September 26, 2025   • Revised: January 4, 2026   • Accepted: January 13, 2026

Copyright © 2026 by The Korean Association for the Study of the Liver

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

  • 3,464 Views
  • 297 Download
  • 1 Web of Science
  • 1 Crossref
  • 1 Scopus
prev next

Full Article

Download PDF Download PDF
See the commentary-article "COLEC12high tumor-associated macrophages: A novel paracrine axis driving lenvatinib resistance in hepatocellular carcinoma: Editorial on “COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling”" in Volume 32 on page 1463.
  • Background/Aims
    Lenvatinib resistance remains a critical barrier in advanced hepatocellular carcinoma (HCC) therapy. However, the underlying mechanisms and strategies for reversing resistance remain incompletely understood.
  • Methods
    Integrated transcriptomics of lenvatinib-resistant patient tumors and an acquired-resistance murine model identified a novel macrophage subpopulation. Functional validation employed CRISPR-SAM screening, conditioned medium (CM) assays, subcutaneous/orthotopic xenografts, patient-derived organoids (PDOs), and patient-derived xenografts (PDXs). Mechanistic studies included ChIP-qPCR, co-immunoprecipitation, and pharmacologic targeting. Clinical relevance was assessed in a retrospective cohort.
  • Results
    Resistant HCC exhibited significant enrichment of a COLEC12high TAM subset, which correlated with poor survival and treatment response. These TAMs secreted neuregulin-1 (NRG1), activating HER2/HER3-AKT signaling in tumor cells to drive cancer stemness and lenvatinib resistance. Mechanistically, in TAMs COLEC12 sequestered STAT1 in the cytoplasm, preventing its phosphorylation, and thereby derepressing STAT3-mediated NRG1 transcription. Depletion of NRG1 reversed the stemness phenotypes and resensitized tumors to lenvatinib both in vitro and in vivo. Clinically, high NRG1 expression predicted an inferior lenvatinib response and shorter survival. Crucially, the bispecific anti-HER2/HER3 antibody zenocutuzumab restored lenvatinib efficacy in PDOs, PDXs, and murine models.
  • Conclusions
    Our work establishes the COLEC12high TAM/NRG1 axis as a master regulator of therapeutic resistance and identifies NRG1 as a predictive biomarker, providing a clinically actionable strategy to overcome lenvatinib resistance in HCC.
  • Keywords: Hepatocellular carcinoma; Lenvatinib resistance; COLEC12; NRG1; Cancer stemness
• A COLEC12high TAM subset is enriched in lenvatinib-resistant hepatocellular carcinoma.
• COLEC12high TAM-derived NRG1 activates tumor HER2/HER3–AKT signaling to drive cancer stemness and lenvatinib resistance.
• COLEC12 promotes STAT3-dependent NRG1 transcription by sequestering STAT1 and suppressing its phosphorylation.
• NRG1 levels correlate with poor lenvatinib response and survival; adding zenocutuzumab restores sensitivity across organoid and xenograft models.
Graphical Abstract
Hepatocellular carcinoma (HCC) is a leading cause of cancer mortality worldwide, with systemic therapies central to advanced disease management [1,2]. Lenvatinib, a multityrosine kinase inhibitor of VEGFRs, FGFRs, PDGFRα, RET, and KIT, is a standard first-line treatment that is noninferior to sorafenib [3]. Despite initial efficacy, rapid intrinsic and acquired resistance limits benefit, with most patients progressing within months [4]. Tumor cell-intrinsic mechanisms, such as kinase rewiring (e.g., EGFR hyperactivation) and oncogenic mutations [5-8], have been studied, but the immunosuppressive and plasticity-inducing tumor microenvironment (TME), particularly tumor-associated macrophages (TAMs), plays a critical but incompletely understood role in therapeutic resistance.
TAMs are a dominant, heterogeneous TME component with remarkable plasticity [9-11]. They promote progression, angiogenesis, immune evasion, and resistance to therapy, including sorafenib [12,13]. Spatial and single-cell analyses revealed distinct TAM states and spatial patterns linked to tumor aggressiveness and poor prognosis [11] Notably, specific subsets, such as TREM2+ TAMs, occupy immunosuppressive niches that exclude CD8+ T cells and are strongly associated with immunotherapy resistance and metastasis [14,15]. These findings highlight how TAM subsets and spatial dynamics shape resistance. However, the specific TAM subsets responsible for lenvatinib resistance and their molecular mediators have not yet been fully characterized.
To explore macrophage-derived mediators potentially involved in lenvatinib resistance, we focused on Collectin Subfamily Member 12 (COLEC12), a pattern recognition receptor involved in innate immunity and phagocytosis that is expressed on specific macrophage subsets [16]. However, its functional role in cancer, and particularly in mediating therapeutic resistance, is undefined. Similarly, Neuregulin 1 (NRG1) is a potent ligand for ERBB2 (HER2) and ERBB3 (HER3) receptors [17]. NRG1 binding induces HER3–HER2 heterodimerization, activating the PI3K-AKT and MAPK cascades that are central to survival, proliferation, and stemness [17-20]. While NRG1 gene fusions are oncogenic drivers in other cancers (e.g., lung and pancreas) and are targeted by agents such as bispecific anti-HER2/HER3 antibody zenocutuzumab (MCLA-128) [21,22], the significance of NRG1 in HCC and its contribution to TKI resistance are unknown.
Here, using integrated transcriptomic analysis of clinical samples and an acquired resistance model, we identified a COLEC12high TAM subset enriched in lenvatinib-resistant HCC. These TAMs are the main source of secreted NRG1, which activates tumor cell-intrinsic HER2/HER3 signaling to drive stemness and lenvatinib resistance. COLEC12 promotes STAT3-dependent NRG1 transcription by sequestering STAT1 and inhibiting its phosphorylation. We further validated NRG1 as a predictive biomarker for lenvatinib response and survival, and showed that zenocutuzumab, which targets the NRG1–HER2/HER3 axis, overcomes resistance in preclinical patients, including patient-derived systems. This study identifies the COLEC12high TAM/NRG1 axis as a potential driver of lenvatinib resistance and proposes a dual therapeutic strategy targeting this circuit.
Ethics statement, patients and tissues
This study was approved by the Ethics Committee of Sun Yat-sen University Cancer Center (SYSUCC) (Approval No.: G2023‑184‑01). This retrospective study involved fifty HCC tissue samples collected from patients who received preoperative lenvatinib treatment at the SYSUCC between 2018 and 2023. Treatment response—classified into sensitivity and resistance groups—was evaluated based on RECIST v1.1 criteria. Sample processing followed standard pathological procedures, and all human specimens were anonymized in compliance with the principles of the Declaration of Helsinki (1975, as revised).
Animal experiments
All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University Cancer Center (Ethics Approval ID: L102012024050F). BALB/C nude mice (male, 4–5 weeks old, ~30 g) were maintained in a specific pathogen-free (SPF) facility under standard conditions (12-h light/dark cycle, 20–22°C, 40–70% humidity).
Statistics and reproducibility
Analyses were performed in GraphPad Prism 8.0 with two-sided tests and significance at P<0.05. Survival was estimated using Kaplan–Meier method and compared using the log-rank test. Two-group comparisons were performed using Student’s t-test or Welch’s t-test for parametric data and Mann–Whitney U for nonparametric data; comparisons among >2 groups used one-way ANOVA or the Kruskal–Wallis test. Normality was assessed using the Shapiro–Wilk test, and variance homogeneity was assessed using Levene’s test. The quantitative data are presented as the mean±standard error of the mean. The numbers of biological replicates, animals, or independent experiments are given in the figure legends. Immunoblotting (IB), immunohistochemical (IHC), and immunofluorescent (IF) images are representative of ≥3 independent replicates with similar results.
Detailed descriptions of cell lines, macrophage differentiation, conditioned medium preparation, functional assays, molecular biology experiments, and preclinical models are provided in the Supplementary Materials and Methods.
COLEC12high TAMs promote lenvatinib resistance in patients with HCC
To identify the key TME component mediating lenvatinib resistance in HCC, we integrated transcriptomic and CIBERSORT analyses of data from lenvatinib-treated patients (n=6; 3 resistant, 3 sensitive). Macrophages were the most markedly increased immune subset in resistant tumors (Fig. 1A). Using a MycOETrp53KO spontaneous HCC model with chronic lenvatinib exposure, we profiled macrophages from resistant tumors and patient samples, and identified a conserved 58-gene resistance-associated macrophage signature (Fig. 1B, 1C). A CRISPR-SAM gainof-function screen of THP-1-derived macrophages revealed that COLEC12 overexpression strongly increased HCC cell survival under lenvatinib treatment (Fig. 1D, 1E; Supplementary Fig. 1A).
Clinically, COLEC12 expression was restricted to macrophages in single-cell datasets (GSE151530 and GSE149614) and colocalized with CD68 in HCC tissues (Fig. 1F; Supplementary Fig. 1B, 1C). COLEC12 expression correlated with CD68 expression in the TCGA-LIHC (Supplementary Fig. 1D), and tumors enriched with COLEC12high TAMs had shorter PFS and OS and lower objective response rates to lenvatinib (Fig. 1F; Supplementary Fig. 1E, 1F).
Collectively, these data reveal that COLEC12high macrophages constitute an enriched TME subpopulation associated with lenvatinib resistance.
COLEC12high TAMs drive lenvatinib resistance through paracrine-mediated stemness reprogramming
Based on the results of transwell coculture experiments showing physical separation of macrophages and HCC cells, we hypothesized that a paracrine mechanism is involved. Huh7 and Hep3B cells were exposed to conditioned medium (CM) from COLEC12-overexpressing (COLEC12OE) macrophages, including cells derived from THP-1 cells, monocyte-derived macrophages (MDMs) generated from patient blood, and TAMs isolated from HCC resections, with each population verified by flow cytometry (CD45+CD11b+CD68+; Fig. 2A; Supplementary Fig. 2A, 2B, 3A). COLEC12OE CM increased cell viability and clonogenicity, reduced apoptosis and increased the IC50 of lenvatinib (Fig. 2B; Supplementary Fig. 2C–2E, 3B).
This paracrine effect was further validated in vivo. Subcutaneous xenografts exposed to COLEC12OE CM showed accelerated growth, reduced lenvatinib sensitivity, and increased tumor volume/weight. In contrast, orthotopic models similarly failed to respond to systemic lenvatinib, as reflected by increased liver-to-body weight ratios (Fig. 2C).
Mechanistically, COLEC12OE CM enriched hematopoietic stem cell differentiation pathways and upregulated stemness markers (SOX2/OCT4/CD44/CD133) at the mRNA and protein levels in vitro and in xenografts (Fig. 2D; Supplementary Fig. 2G). Sphere and limiting dilution assays demonstrated enhanced self-renewal and stem cell frequency (Fig. 2D; Supplementary Fig. 2F, 2H), and CM-preconditioned cells initiated tumors more efficiently at lower inocula in vivo (Fig. 2E).
Collectively, COLEC12high macrophages drive lenvatinib resistance in HCC through paracrine secretion activating cancer stemness reprogramming.
NRG1 mediates COLEC12high TAM-induced stemness and therapeutic resistance
We integrated COLEC12OE macrophage transcriptomic data with the secretome dataset from The Human Protein Atlas (THPA) to identify potential mediators of COLEC12high macrophage–induced stemness in HCC. These findings highlight NRG1 as a key candidate related to stemness and therapeutic resistance [19,23]. Validation confirmed that NRG1 upregulation at the transcriptional, protein, and secretion levels in COLEC12high macrophages (Fig. 3A). Consistently, single-cell RNA-seq datasets (GSE151530 and GSE149614) revealed that NRG1 expression is predominantly confined to macrophage/monocyte clusters (Supplementary Fig. 4A).
To functionally validate NRG1, we treated HCC cells in parallel with rNRG1 and COLEC12OE-CM. At an optimal dose (Supplementary Fig. 4B), rNRG1 increased proliferation (Supplementary Fig. 4C), upregulated stemness markers (SOX2/OCT4/CD44/CD133; Supplementary Fig. 4D), promoted sphere formation (Supplementary Fig. 4E) and reduced lenvatinib sensitivity (colony formation; Supplementary Fig. 4F), with effects comparable to those seen with COLEC12OE CM.
NRG1 knockdown in COLEC12OE macrophages reversed resistance, restoring IC50, normalizing proliferation, reducing colony formation, and increasing apoptosis (Fig. 3B; Supplementary Fig. 4G–4K). In subcutaneous and orthotopic models, COLEC12OE CM promoted tumor growth, liver burden, and lenvatinib resistance, all of which were abolished by silencing NRG1, establishing secreted NRG1 as the main resistance driver (Fig. 3C).
Mechanistically, NRG1 was required for stemness: its knockdown blocked stemness marker induction, impaired sphere formation, and reduced the tumor-initiating cell frequency both in vitro and in vivo (Fig. 3D, 3E; Supplementary Fig. 4L–4N).
Collectively, these data demonstrate that NRG1 is the master effector through which COLEC12high macrophages orchestrate stemness-driven therapeutic resistance.
COLEC12 drives NRG1 transcription via STAT1 sequestration-mediated STAT3 activation
Our data revealed that COLEC12 overexpression upregulated NRG1 mRNA, suggesting transcriptional regulation. Since COLEC12 lacks intrinsic transcriptional activity, we screened NRG1 regulators across hTFtarget, KnockTF, and FIMO-JASPAR, identifying STAT3, JUN, RELA, and KLF4; siRNA in COLEC12OE macrophages showed only STAT3 knockdown reduced NRG1, pinpointing STAT3 as the key effector, with JASPAR predicting a STAT3 site near the NRG1 TSS (Fig. 4A). ChIP–qPCR confirmed STAT3 occupancy at this site, and luciferase assays demonstrated that STAT3 activated wild-type but not mutant promoters (Fig. 4B).
To investigate its regulation, IP–MS of COLEC12-Flag complexes revealed STAT1 as a major binding partner (Fig. 4C). As phosphorylated STAT1 inhibits STAT3 via sequestration [24], we hypothesized that COLEC12 binding blocks STAT1 phosphorylation, alleviating STAT3 repression. Co-IP confirmed endogenous STAT1–COLEC12 interaction across THP-1–derived macrophages, MDMs, and TAMs, immunofluorescence showed cytoplasmic co-localization, and docking predicted direct binding of COLEC12’s intracellular domain to STAT1 SH2 (Fig. 4C).
Docking predicted direct binding between the COLEC12 intracellular domain and STAT1, and COLEC12 lacking this domain failed to interact with STAT1 (Fig. 4D). Full-length COLEC12 reduced STAT1 phosphorylation while increasing STAT3 phosphorylation in THP-1, MDM, or TAM macrophages whereas the mutant had no effect (Fig. 4E). In addition, the IF results revealed that full-length COLEC12 increased the nuclear localization of pSTAT3 (Fig. 4F).
Thus, COLEC12 promotes STAT3-driven NRG1 transcription by binding and inhibiting STAT1 phosphorylation.
NRG1 activates cancer cell-intrinsic HER2/HER3 signaling to drive lenvatinib resistance and stemness in HCC
Previous studies have established that NRG1-driven oncogenesis occurs mainly through HER2/HER3 activation in tumor cells [17]. Immunoblots revealed that compared with vector CM, COLEC12OE CM increased HER3, HER2, and AKT phosphorylation, and these effects were attenuated by NRG1 knockdown and by zenocutuzumab (a clinically validated HER2/HER3 bispecific antibody) in Huh7 and Hep3B cells exposed to CM (Fig. 5A; Supplementary Fig. 5A) [21]. Treatment with zenocutuzumab markedly lowered the lenvatinib IC50 in HCC cells exposed to COLEC12OE CM and restored drug sensitivity, as further supported by enhanced growth inhibition in proliferation assays (Fig. 5B; Supplementary Fig. 5B, 5C). COLEC12OE CM also enhanced colony formation and tumorsphere growth, both of which were abolished by zenocutuzumab, indicating effective suppression of COLEC12–NRG1–driven stemness (Fig. 5C; Supplementary Fig. 5D, 5E).
Zenocutuzumab suppressed COLEC12OE CM–driven tumor growth and the combination of zenocutuzumab and lenvatinib produced the most potent suppression in both subcutaneous and orthotopic Huh7 HCC models (Fig. 5D; Supplementary Fig. 5F). Furthermore, CDX-3379, an anti-HER3 antibody [25], enhanced lenvatinib sensitivity in the immunocompetent C57BL/6 Hepa1-6 subcutaneous HCC model, and the combination therapy of CDX-3379 and lenvatinib yielded the strongest tumor control, characterized by increased intratumoral CD8+ T-cell infiltration and reduced TAM abundance. Consistent with the proposed mechanism, IHC showed that the combination therapies across all models significantly reduced p-HER3, p-HER2, and p-AKT (Fig. 5E; Supplementary Fig. 5G–5I; Supplementary Fig. 6A, 6B). In patient-derived organoids (PDOs) and patient-derived xenografts (PDXs), the lenvatinib–zenocutuzumab combination achieved the strongest growth inhibition compared with monotherapies (Fig. 5F; Supplementary Fig. 4J). Collectively, these findings demonstrate that COLEC12high macrophage-derived NRG1 drives HER2/HER3 signaling to promote lenvatinib resistance and cancer stemness in HCC, suggesting a combinatorial strategy to overcome this microenvironment-mediated resistance mechanism.
NRG1 serves as a biomarker and therapeutic target for lenvatinib resistance in HCC
To assess the role of NRG1 in lenvatinib resistance, we analyzed residual tumors from 50 SYSUCC HCC patients treated preoperatively with lenvatinib. IHC stratified patients into high-NRG1 and low-NRG1 groups (Fig. 6A). MRI and waterfall plots showed significantly less tumor regression in NRG1-high patients under RECIST v1.1 (Fig. 6B). The objective response rate was higher in NRG1-low patients (55.2% vs. 14.3%; P=0.003), and ROC analysis indicated that NRG1 predicted the lenvatinib response (AUC=0.704; Fig. 6C). Elevated NRG1 by IHC or serum ELISA correlated with shorter PFS and OS (Fig. 6D, 6E). Thus, NRG1 in tumors and circulation serves as a biomarker of poor response and prognosis in lenvatinib-treated HCC. Mechanistically, COLEC12high macrophages secrete NRG1 to activate HER2/HER3 signaling and cancer stemness, and COLEC12 increases STAT3-dependent NRG1 transcription by binding STAT1, blocking its phosphorylation, and relieving STAT1-mediated suppression of STAT3 (Fig. 6F).
This study suggests a previously underappreciated, TME-driven mechanism of lenvatinib resistance in HCC, centered on a specific COLEC12high TAM subpopulation that orchestrates therapeutic resistance through paracrine NRG1-HER2/HER3 signaling. Our integrated approach combining clinical cohorts, murine resistance models, and functional validation demonstrates that COLEC12high TAMs are transcriptionally reprogrammed to secrete NRG1, activating cancer stemness pathways in tumor cells and conferring therapeutic resistance. Crucially, we identified NRG1 as a clinically actionable biomarker and validated the pharmacological blockade of the NRG1-HER2/HER3 axis as an effective strategy to overcome this microenvironment-mediated resistance.
Several key findings distinguish our work from prior studies on TKI resistance. First, in addition to the reported role of TAMs in sorafenib resistance (e.g., HGF/MET and CXCR2–ERK) [13,26], our discovery of the COLEC12high TAM subset specifically driving lenvatinib resistance reveals previously unrecognized functional heterogeneity within macrophage populations. The conserved 58-gene signature identified through cross-species integration provides a molecular fingerprint for resistance-associated TAMs. Second, our mechanistic analysis revealed a noncanonical signaling axis in which COLEC12, traditionally recognized as a pattern recognition receptor in innate immunity [16,27,28], functionally repurposes its intracellular domain to sequester STAT1, relieving the repression of STAT3 and enabling the transcription of NRG1—which is distinct from cytokine-driven STAT3 activation in TAMs [29,30]. Third, we established that TAM-derived (rather than tumor cell-autonomous) NRG1 is the dominant driver of stemness reprogramming in this context, resolving conflicting reports on the sources of NRG1 in HCC.
Given compensatory ERBB signaling in HCC, prior work emphasized EGFR-mediated lenvatinib resistance (PAK2–ERK5; STAT3–ABCB1) [31,32], we identified a distinct HER2/HER3-dependent escape pathway. NRG1 elevation in resistant tumors is correlated with poor response and survival, which is consistent with the established role of ERBB signaling in HCC progression [33-35]. Our data demonstrate that pharmacologic HER2/HER3 blockade with zenocutuzumab abrogates AKT activation and resensitizes tumors to lenvatinib. This finding suggests that combinatorial ERBB targeting—potentially dual inhibition of EGFR and HER2/HER3—may be needed to overcome adaptive resistance, especially in NRG1-high contexts.
The clinical relevance of this axis is supported by several findings: COLEC12high TAM infiltration is correlated with poor survival and reduced lenvatinib response; NRG1 expression predicts therapeutic resistance and worse survival; and the bispecific HER2/HER3 antibody zenocutuzumab reverses resistance in preclinical models, including patient-derived models. Together, these data suggest that combining lenvatinib with HER2/HER3 inhibitors could restore efficacy in patients with high-NRG1 HCC. The proven clinical activity of zenocutuzumab in NRG1 fusion–driven cancers further supports its potential for repurposing in this setting.
Our findings align with emerging recognition that TME-mediated resistance operates independently of tumor cell-intrinsicadaptations [36,37]. While prior studies focused on TAMs in immunotherapy resistance [38], our work expands their role to targeted kinase inhibitor failure. The spatial organization of COLEC12high TAMs within macrophage-enriched niches may create localized “resistant microenvironments” in which high NRG1 concentrations drive stemness reprogramming—a hypothesis that warrants future investigation using spatial transcriptomics.
Our small cohort limits power and subgroup generalizability. The Hepa1-6 syngeneic model complements nude/Huh7 cells but reflects murine, not human, immunity; thus orthotopic and humanized models are needed. Although CDX-3379 potentiated lenvatinib with immune modulation, single-model inferences warrant caution. The drivers of lenvatinib-induced COLEC12high expression remain unknown, and mediators other than NRG1 may contribute. Prospective validation of serum NRG1 levels and in vivo investigation of the COLEC12–NRG1 axis in orthotopic/humanized HCC are warranted.
Clinically, NRG1 (tissue or serum) could stratify lenvatinib responsiveness; trials combining lenvatinib with HER2/HER3 inhibitors (e.g., zenocutuzumab) should prioritize for NRG1-high HCC; and targeting COLEC12high TAMs (e.g., COLEC12-directed or polarization strategies) may extinguish resistance at its source. Future work should test TAM plasticity during therapy and prospectively validate NRG1 as a predictive biomarker.
In conclusion, the COLEC12high TAM/NRG1 axis may act as a central regulator of lenvatinib resistance in HCC. By linking macrophage-specific COLEC12 to STAT1 sequestration, STAT3-dependent NRG1 transcription, and tumor stemness, we provide mechanistic hypotheses and potential therapeutic strategies to overcome microenvironment-mediated resistance.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Authors’ contributions

Binkui Li and Yi Niu conceived and supervised the study. Jianxing Zhang led data curation, formal analysis, methodology, software, visualization, and project administration, and contributed equally to the original draft. Liang Qiao and Zongfeng Wu contributed equally to the formal analysis and methodology; and Zongfeng Wu contributed equally to the validation and visualization. Dinglan Zuo contributed to methodology and supervision. Yunfei Yuan provided project administration and senior supervision. Yi Niu and Binkui Li led conceptualization and manuscript writing, with Binkui Li also securing funding. Shanshan Huang and Zhenkun Huang assisted with validation and data curation. Shaoru Liu, Yi Zeng, and Yu Li supported the methodology and software. Chenwei Wang contributed to the formal analysis, and Wei He contributed to the original draft. All the authors contributed to manuscript review/editing, discussed the results, and approved the final manuscript.

Acknowledgements

This work was supported by grants from the National Natural Science Foundation of China (No. 82373405 to B.L., 82272887 to Y.Y. and 82473037 to Y.N.) and the Young Talents Program of Sun Yat-sen University Cancer Center (YTP-SYSUCC-0088).

Generative AI and AI-assisted technologies were NOT used in the preparation of this work.

Conflicts of Interest

The authors have no conflicts to disclose.

Supplementary material is available at Clinical and Molecular Hepatology website (http://www.e-cmh.org).
Supplementary Materials and Methods.
cmh-2025-1059-Supplementary-Materials-and-Methods.pdf
Supplementary Figure 1.
COLEC12high TAMs promote lenvatinib resistance in HCC. (A) Heatmap showing the qPCR validation of 58 sgRNAs targeting genes in THP-1 cells after CRISPR/dCas9-mediated transcriptional activation (n=3). (B) Left: UMAP projection of integrated single-cell RNA-seq datasets from HCC patients (GSE151530 and GSE149614), annotated by cell type. Right: Feature plot showing expression of COLEC12 across cell clusters, with color intensity representing normalized expression levels. (C) Spatial co-localization of COLEC12 and CD68 signals based on gray value distribution profiles across tumor regions. (D) Correlation analysis of COLEC12 and CD68 mRNA expression levels in the TCGA-LIHC cohort. (E) Kaplan-Meier survival curve based on the level of macrophages COLEC12 in the TCGA-LIHC dataset. (F) Waterfall plot (left) showing the percentage change in tumor size from baseline for patients with COLEC12high (bottom) macrophage infiltration, as classified by RECIST v1.1 criteria. Bar chart (right) illustrating the distribution of lenvatinib treatment response (PR, SD, PD) in COLEC12high and COLEC12low macrophage groups. Data are presented as mean±standard error of the mean. Statistical significance was assessed using Student’s t-test. Survival data were analyzed by the log-rank test. COLEC12, Collectin Subfamily Member 12; TAMs, tumor-associated macrophages. Source data are provided as a Source Data file.
cmh-2025-1059-Supplementary-Figure-1.pdf
Supplementary Figure 2.
COLEC12high TAMs drive lenvatinib resistance through paracrine-mediated stemness reprogramming. (A) qRT-PCR analysis confirming successful overexpression of COLEC12 in THP-1-derived macrophages, MDMs and TAMs. Western blot analysis verifying COLEC12 expression in THP-1-derived macrophages following lentiviral transduction (n=3). (B) Flow cytometric identification of CD11b+CD68+ macrophages in MDMs and TAMs. Representative dot plots showing CD11b versus CD68 staining in control cells (left panels) and in (top right) MDMs and (bottom right) TAMs. (C) IC50 values of lenvatinib in Huh7 cells and Hep3B cells treated with COLEC12OE CM or Vector CM (n=3). (D) Cell proliferation was assessed by CCK-8 assay in the presence of 2.5 μM lenvatinib, and clonogenic potential was evaluated through colony formation assays with varying concentrations of lenvatinib (0, 2.5, and 5 μM). (E) Representative flow cytometry plots and quantification by flow cytometry in Huh7 and Hep3B cells treated with 2.5 μM lenvatinib (n=3). (F) Sphere formation assay of Hep3B cells cultured with COLEC12OE CM or Vector CM (n=3). Scale bar, 100 μm. (G) qRT-PCR analysis of stemness markers SOX2, OCT4, CD44, and CD133 in Hep3B cells after exposure to COLEC12OE CM or Vector CM (n=3). (H) In vitro limiting dilution assay of Hep3B cells treated with COLEC12OE CM or Vector CM. Data are presented as stem cell frequency in chart form (n=3). Data are presented as mean±standard error of the mean. Statistical significance was determined using Student’s t-test for two-group comparisons or one-way ANOVA for multiple groups. Kruskal–Wallis test for tumor initiation assays (M, N). COLEC12, Collectin Subfamily Member 12; TAMs, tumor-associated macrophages; MDMs, monocyte-derived macrophages; COLEC12OE, COLEC12-overexpressing; CM, conditioned medium. *P<0.05, **P<0.01, ***P<0.001. Source data are provided as a Source Data file.
cmh-2025-1059-Supplementary-Figure-2.pdf
Supplementary Figure 3.
Flow-cytometric gating strategy for TAM isolation and apoptosis assessment of HCC cells. (A) Flow-cytometric gating strategy for human macrophage analysis. From left to right: (All events) mononuclear cells were first gated based on FSC-A versus SSC-A (P1). (P1) Singlets were selected using SSC-H versus SSC-A (P2). (P2) Live leukocytes were identified as Dye⁻CD45+ cells (P3). (P3) Macrophages were finally defined as CD11b+CD68+ cells (P4), with the percentage of cells within each gate indicated. (B) Representative Annexin V–APC and 7-AAD staining of HCC cells treated with conditioned media from COLEC12-overexpressing or control macrophages under lenvatinib, illustrating the gating strategy used to quantify viable and apoptotic cell populations. TAM, tumor-associated macrophage; HCC, hepatocellular carcinoma.
cmh-2025-1059-Supplementary-Figure-3.pdf
Supplementary Figure 4.
RG1 mediates COLEC12high TAMs-induced stemness and therapeutic resistance. (A) Single-cell RNA-seq analysis of NRG1 expression across HCC tumor microenvironmental cell types. (B) Proliferation response of Huh7 and Hep3B cells to recombinant human NRG1 (rNRG1; 0–50 ng/mL), assessed by CCK-8 assay (n=6). (C) Cell proliferation curves for Huh7 and Hep3B exposed to rNRG1 (10 ng/mL) or COLEC12OE CM±lenvatinib (2.5 μM), measured via CCK-8 assay (n=3). (D) qRT-PCR of stemness markers in Huh7 and Hep3B cells treated with rNRG1 (10 ng/mL) or COLEC12OE CM (n=3). (E) Tumorsphere formation assays demonstrating increased sphere-forming capacity after rNRG1 (10 ng/mL) or COLEC12OE CM exposure (n=3). (F) Colony formation assay in Huh7 and Hep3B cells treated with rNRG1 (10 ng/mL) or COLEC12OE CM in lenvatinib treatment (0, 2.5, and 5 μM) (n=3). (G) Validation of NRG1 knockdown efficiency in COLEC12OE macrophages via qRT-PCR and Western blot (n=3). (H) IC50 values of lenvatinib in Hep3B cells treated with Vector CM, COLEC12OE CM, COLEC12OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM (n=3). (I) Proliferation of Hep3B cells in response to Vector CM, COLEC12OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM, with or without lenvatinib (2.5 µM), measured by CCK-8 assay (n=3). (J) Colony formation of Hep3B cells treated with Vector CM, COLEC12OE CM, COLEC12OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM (n=3). (K) Left: Representative flow cytometry plots of Huh7 and Hep3B cells treated with Vector CM, COLEC12OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM with 2.5 µM lenvatinib. Right: Quantification of apoptosis in Hep3B cells by flow cytometry after 2.5 µM lenvatinib treatment in the corresponding Vector CM, COLEC12 OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM (n=3). (L) qRT-PCR analysis of stemness markers (SOX2, OCT4, CD44, and CD133) in Hep3B cells after treatment with Vector-CM, COLEC12OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12 OE+shNRG1#2 CM (n=3). (M) Tumorsphere formation assay of Hep3B cells treated with Vector CM, COLEC12OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM (n=3). (N) In vitro limiting dilution assay demonstrating altered stem cell frequency in Hep3B cells treated with Vector-CM, COLEC12OE CM, COLEC12 OE+shNRG1#1 CM, or COLEC12OE+shNRG1#2 CM (n=3). Data are presented as stem cell frequency in chart form. Data are presented as mean±standard error of the mean. Statistical significance was assessed by Student’s t-test for comparisons between two groups and one-way ANOVA for comparisons among multiple groups. Tumor initiation assays were analyzed using the Kruskal–Wallis test. *P<0.05, **P<0.01, ***P<0.001. Source data are provided as a Source Data file. NRG1, Neuregulin 1; COLEC12, Collectin Subfamily Member 12; TAMs, tumor-associated macrophages; HCC, hepatocellular carcinoma; COLEC12OE, COLEC12-overexpressing; CM, conditioned medium.
cmh-2025-1059-Supplementary-Figure-4.pdf
Supplementary Figure 5.
NRG1 activates cancer cell-intrinsic HER2/HER3 signaling to drive lenvatinib resistance and stemness in HCC. (A) Immunoblots of phosphorylated and total HER3, HER2, and AKT in Hep3B cells cultured with Vector CM or COLEC12OE CM with or without zenocutuzumab, or CM from COLEC12OE macrophages transduced with shNRG1 (n=3). (B) IC50 values of Hep3B cells cultured with Vector CM, COLEC12OE CM, or COLEC12OE CM plus zenocutuzumab (0.01 μM) and treated with increasing concentrations of lenvatinib (n=3). (C) Proliferation of Hep3B cells cultured with CM from THP-1-derived macrophages, with or without zenocutuzumab (0.01 μM), in the absence or presence of lenvatinib (2.5 μM), as assessed by CCK-8 assay (n=3). (D) Representative images of colony formation assays in Hep3B cells treated with Vector CM, COLEC12OE CM, or COLEC12OE CM plus zenocutuzumab (0.01 μM), in the presence of lenvatinib (0, 2.5, or 5 μM). (E) Representative brightfield images (left) and quantification (right) of tumorspheres (>100 μm) formed by Hep3B cells in the indicated CM. Sphere numbers were significantly reduced by zenocutuzumab (0.01 μM) (treatment n=3). Scale bar, 100 μm. (F) Tumor growth curves of Huh7 subcutaneous xenografts treated with vehicle, lenvatinib, zenocutuzumab, or the combination (n=6). (G) Tumor growth curves of Hepa1-6 subcutaneous xenografts treated with vehicle, lenvatinib, zenocutuzumab, or the combination (n=6). (H) Flow cytometry analysis of intratumoral immune cells from Hepa1-6 tumors in C57BL/6 mice treated with vehicle, CDX-3379, lenvatinib, or the combination. The combination therapy markedly increased CD3+CD8+ T-cell infiltration (top) and reduced CD11b+F4/80+ tumor-associated macrophages (bottom). (I) Representative images of haematoxylin and eosin (H&E), p-AKT, p-HER2 and p-HER3 staining in Huh7 and Hepa1-6 xenografts. Scale bars, 50 μm. (I) Growth of patient-derived xenograft (PDX) tumors under the same treatment regimens (n=6). Data are presented as mean±standard error of the mean. Statistical significance was assessed by two-tailed unpaired Student’s t-test or one-way ANOVA with multiple comparisons. *P<0.05, **P<0.01, ***P<0.001. NRG1, Neuregulin 1; HCC, hepatocellular carcinoma; CM, conditioned medium; COLEC12, Collectin Subfamily Member 12; COLEC12OE, COLEC12-overexpressing; PDX, patient-derived xenograft. Source data are provided as a Source Data file.
cmh-2025-1059-Supplementary-Figure-5.pdf
Supplementary Figure 6.
Flow-cytometric gating strategy for CD8+ T cells and tumor-associated macrophages. (A) Gating strategy for identification of CD8+ T cells in dissociated tumor suspensions: live single cells were selected using FSC-A/SSC-A and FSC-A/FSC-H, followed by exclusion of dead cells (L/D), gating on CD45+ leukocytes, and subsequent gating on CD3+CD8+ T cells. (B) Gating strategy for tumor-associated macrophages (TAMs): live single cells were selected as in (A), dead cells were excluded, CD45+ leukocytes were gated, and TAMs were defined as CD11b+F4/80+ cells.
cmh-2025-1059-Supplementary-Figure-6.pdf
Supplementary Table 1.
Antibodies used for flow cytometry
cmh-2025-1059-Supplementary-Table-1.pdf
Supplementary Table 2.
Cell line authentication reports
cmh-2025-1059-Supplementary-Table-2.pdf
Supplementary Table 3.
List of sgRNA sequences
cmh-2025-1059-Supplementary-Table-3.pdf
Supplementary Table 4.
Sequences of RNA oligonucleotides
cmh-2025-1059-Supplementary-Table-4.pdf
Supplementary Table 5.
Antibodies included in the study
cmh-2025-1059-Supplementary-Table-5.pdf
Supplementary Table 6.
The primers used in present study
cmh-2025-1059-Supplementary-Table-6.pdf
Figure 1.
COLEC12high TAMs promote lenvatinib resistance in HCC. (A) Upper: MRI scans of lenvatinib-sensitive (Patients #1–3) and lenvatinib-resistant (Patients #4–6) tumors; dashed lines denote tumor margins. Lower: CIBERSORT analysis of bulk RNA-seq data from the same tumors (red, lenvatinib-sensitive; green, lenvatinib-resistant) (n=3). (B) Spontaneous HCC via hydrodynamic injection (Myc-OE, Trp53-sg, SB); bioluminescence shows resistance (no treatment=3; lenvatinib-induced therapy resistance=3). (C) Venn diagram of genes upregulated in resistant patient tumors (n=3) and resistant-mouse TAMs (n=3) (FC>1.5, P<0.05). (D) CRISPR/dCas9-SAM activation in THP-1 macrophages; cocultured with Huh7; Calcein-AM/PI viability assay under 2.5 μM lenvatinib. (E) Representative Calcein-AM/PI images and quantification after 48 h of coculture with lenvatinib (2.5 μM). (F) Immunofluorescence of CD68 (green) and COLEC12 (red) in macrophages; nuclei DAPI; scale bar, 25 μm. Kaplan–Meier plots of PFS/OS by TAM COLEC12 expression (log-rank test). Data are presented mean±standard error of the mean. Two-group: Student’s t-test; multiple groups: one-way ANOVA; survival: log-rank. HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; COLEC12, Collectin Subfamily Member 12; ns, not significant. *P<0.05. Source data available.
cmh-2025-1059f1.jpg
Figure 2.
COLEC12high TAMs drive lenvatinib resistance through paracrine-mediated stemness reprogramming. (A) Workflow schematic: THP-1–derived macrophages, MDMs, or TAMs expressing COLEC12 (COLEC12OE) or vector control produce CM, which is applied to Huh7 and Hep3B cells. (B) Lenvatinib sensitivity in Huh7 cells (n=3): up, proliferation by CCK-8 under 2.5 µM lenvatinib; down, colony formation at 0, 2.5, and 5 µM. (C) Subcutaneous xenograft (n=6): tumor images, growth curves, final weights with intratumoral CM+lenvatinib or PBS. Orthotopic liver model: Liver-to-body weight ratios with intraperitoneal CM+lenvatinib or PBS. (D) Bulk RNA-seq of Huh7 cells treated with COLEC12OE CM or Vector CM, followed by pathway enrichment. Stemness assays (n=3): qRT–PCR for SOX2/OCT4/CD44/CD133; limiting dilution; immunofluorescence of tumors (scale bar, 100 µm); sphere formation (scale bar, 100 µm). (E) In vivo limiting dilution assay in subcutaneous models (n=6), with tumor-initiating cell frequencies shown in the chart and table. Data are presented mean±standard error of the mean. Two-group: Student’s t-test; multiple groups: one-way ANOVA; limiting dilution: Kruskal–Wallis. COLEC12, Collectin Subfamily Member 12; TAMs, tumor-associated macrophages; MDMs, monocyte-derived macrophages; COLEC12OE, COLEC12-overexpressing; CM, conditioned medium. *P<0.05, **P<0.01, ***P<0.001. Source data are available.
cmh-2025-1059f2.jpg
Figure 3.
NRG1 mediates COLEC12high TAM-induced stemness and therapeutic resistance. (A) Overlap of genes upregulated in COLEC12OE macrophages with the THPA secretome. Heatmap of the top 20 induced genes. qRT-PCR, immunoblot, and ELISA confirm increased NRG1 expression and secretion in COLEC12OE THP-1 macrophages (n=3). (B) COLEC12OE CM increases lenvatinib resistance in Huh7 cells, whereas NRG1 knockdown reverses this effect, as shown by IC50, CCK-8 proliferation, colony formation, and apoptosis assays (n=3). (C) Subcutaneous Huh7 xenografts (n=6) and orthotopic liver models were treated with intratumoral or intraperitoneal injections of Vector-CM, COLEC12OE-CM, or COLEC12OE+shNRG1#1-CM with/without lenvatinib, showing tumor images, volumes, weights, and liver/body weight ratios (n=6). (D) qRT-PCR of SOX2/OCT4/CD44/CD133 in Huh7 under indicated CM (n=3). Tumorsphere formation under indicated CM (n=3). (E) Limiting dilution assays in vitro (M, n=3) and in vivo (N, n=6 per group), showing stem cell or tumor-initiating frequencies. Data are presented mean±standard error of the mean. Two group comparisons used Student’s t-test; multiple groups used one-way ANOVA; tumor initiation used Kruskal–Wallis. NRG1, Neuregulin 1; COLEC12, Collectin Subfamily Member 12; THPA, The Human Protein Atlas; CM, conditioned medium; COLEC12OE, COLEC12-overexpressing. *P<0.05, **P<0.01, ***P<0.001. Source data available.
cmh-2025-1059f3.jpg
Figure 4.
COLEC12 drives NRG1 transcription via STAT1 sequestration-mediated STAT3 activation. (A) Predicted NRG1 transcription factors from hTFtarget, KnockTF, and JASPAR, highlighting shared TFs (STAT3/JUN/RELA/KLF4). qRT-PCR validation of TF knockdown and its effect on NRG1 expression in COLEC12-overexpressing THP-1 macrophages (n=3). STAT3 motif in the NRG1 promoter predicted by JASPAR. (B) STAT3 binds the NRG1 promoter as shown by ChIP-PCR and ChIP-qPCR in THP-1 macrophages (n=3). Luciferase assays demonstrate that STAT3 activates the WT NRG1 promoter but not the binding-site–deleted mutant (n=6). (C) Mass spectrometry identified STAT1 among COLEC12 interactors. Co-IP confirmed endogenous COLEC12–STAT1 binding in THP-1, MDM, and TAM macrophages (n=3). Immunofluorescence showed cytoplasmic colocalization of COLEC12 and STAT1; scale bar, 10 μm. (D) Docking predicts a direct COLEC12–STAT1 interface. Deletion of the COLEC12 intracellular domain abolishes its interaction with STAT1, as shown by co-IP (n=3). (E) Immunoblots showing that COLEC12-WT decreases p-STAT1 and increases p-STAT3 and NRG1 across THP-1, MDM, and TAM macrophages, whereas the intracellular-domain mutant shows no effect (n=3). (F) Immunofluorescence staining of pSTAT3 in THP-1–derived macrophages expressing control vector, COLEC12-WT, or COLEC12-Mut, with DAPI used for nuclear counterstaining. Scale bar, 10 μm. (G) Data are presented mean±standard error of the mean. Two group comparisons used Student’s t-test; multiple groups used one-way ANOVA; tumor initiation used Kruskal–Wallis. COLEC12, Collectin Subfamily Member 12; NRG1, Neuregulin 1; MDM, monocyte-derived macrophage; TAM, tumor-associated macrophage. *P<0.05, **P<0.01, ***P<0.001. Source data available.
cmh-2025-1059f4.jpg
Figure 5.
NRG1 activates cancer cell-intrinsic HER2/HER3 signaling to drive lenvatinib resistance and stemness in HCC. (A) Immunoblots showing p-HER3, p-HER2, and p-AKT in Huh7 cells exposed to Vector CM, COLEC12OE CM, COLEC12OE CM+zenocutuzumab, or COLEC12OE CM from shNRG1 macrophages (n=3). (B) Lenvatinib IC50, proliferation (±2.5 μM lenvatinib and 0.01 μM zenocutuzumab), and time-course CCK-8 assays under the indicated CM conditions (n=3). (C) Colony formation under CM with 0–5 μM lenvatinib and tumorsphere assays (>100 μm), showing zenocutuzumab abolishes COLEC12OE CM–driven sphere growth (n=3; scale bar, 100 μm). (D) Huh7 subcutaneous and orthotopic xenografts treated with lenvatinib, zenocutuzumab, or the combination, showing maximal tumor suppression in combination groups (n=6). (E) Tumor growth and flow cytometry analysis in an immunocompetent C57BL/6 Hepa1-6 model treated with vehicle, lenvatinib, CDX-3379, or the combination, assessing intratumoral CD8+ T cells and TAMs, and IHC quantification showed p-HER3, p-HER2, and p-AKT levels (n=6). Scale bar, 50 μm. (F) Patient-derived HCC organoids and xenografts showed maximal growth inhibition with the lenvatinib–zenocutuzumab combination (n=6; scale bar, 200 μm). Data are presented mean±standard error of the mean. Two-tailed unpaired Student’s t-test or one-way ANOVA with multiple comparisons. NRG1, Neuregulin 1; HCC, hepatocellular carcinoma; CM, conditioned medium; COLEC12, Collectin Subfamily Member 12; COLEC12OE, COLEC12-overexpressing; IHC, immunohistochemistry. *P<0.05, **P<0.01, ***P<0.001. Source data available.
cmh-2025-1059f5.jpg
Figure 6.
NRG1 serves as a biomarker and therapeutic target for lenvatinib resistance in HCC. (A) IHC of NRG1 in post-treatment tumors from SYSUCC HCC patients with low or high NRG1, with paired pre- and post-lenvatinib MRI; scale bar 25 μm. (B) Waterfall plots of maximal target-lesion change by RECIST v1.1 for NRG1 high (n=21) and NRG1 low (n=29); right panels show response distribution. (C) ROC for NRG1 predicting lenvatinib response by RECIST v1.1 (AUC=0.704). (D) Kaplan–Meier survival by tumor NRG1 level (IHC) in patients receiving preoperative lenvatinib; log-rank test. (E) Kaplan–Meier survival by serum NRG1 level (ELISA, median split) in patients receiving preoperative lenvatinib; log-rank test. (F) Model: COLEC12 overexpressing macrophages activate STAT3 and induce NRG1, which activates HER2/HER3 in HCC cells to drive stemness and resistance; zenocutuzumab blocks HER2/HER3 heterodimerization. NRG1, Neuregulin 1; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; MRI, magnetic resonance imaging. (PD, progressive disease; SD, stable disease; PR, partial response) based on RECIST v1.1.
cmh-2025-1059f6.jpg
cmh-2025-1059f7.jpg

CM

conditioned medium

COLEC12

Collectin Subfamily Member 12

COLEC12OE

COLEC12-overexpressing

HCC

hepatocellular carcinoma

IF

immunofluorescent

IHC

immunohistochemical

MDMs

monocyte-derived macrophages

MRI

magnetic resonance imaging

NRG1

Neuregulin 1

OS

overall survival

PDOs

patient-derived organoids

PDX

patient-derived xenografts

PFS

progression-free survival

TAMs

tumor-associated macrophages

THPA

The Human Protein Atlas

TME

tumor microenvironment
  • 1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-263.
  • 2. Singal AG, Kanwal F, Llovet JM. Global trends in hepatocellular carcinoma epidemiology: implications for screening, prevention and therapy. Nat Rev Clin Oncol 2023;20:864-884.
  • 3. Kudo M, Finn RS, Qin S, Han KH, Ikeda K, Piscaglia F, et al. Lenvatinib versus sorafenib in first-line treatment of patients with unresectable hepatocellular carcinoma: a randomised phase 3 non-inferiority trial. Lancet 2018;391:1163-1173.
  • 4. Ladd AD, Duarte S, Sahin I, Zarrinpar A. Mechanisms of drug resistance in HCC. Hepatology 2024;79:926-940.
  • 5. Wan TR, Wei L, Cheng LH, Chin WC, Shen J, Chan FF, et al. Genome-wide CRISPR screening identifies NFκB and c-MET as druggable targets to sensitize lenvatinib treatment in hepatocellular carcinoma. Cell Mol Gastroenterol Hepatol 2025;19:101502.
  • 6. Cao P, Chen H, Zhang Y, Zhang Q, Shi M, Han H, et al. Genomic amplification of TBC1D31 promotes hepatocellular carcinoma through reducing the Rab22A-mediated endolysosomal trafficking and degradation of EGFR. Adv Sci (Weinh) 2024;11:e2405459.
  • 7. Jin H, Shi Y, Lv Y, Yuan S, Ramirez CFA, Lieftink C, et al. EGFR activation limits the response of liver cancer to lenvatinib. Nature 2021;595:730-734.
  • 8. Huang Q, Wang W, Ge Q, Xu D, Dong R, Chang R, et al. NEK7-induced phosphorylation of EGFR on serine 1070 drives the acquired lenvatinib resistance in hepatocellular carcinoma. Hepatology 2025 July 22;doi: 10.1097/HEP.0000000000001474.
  • 9. Yin Y, Feng W, Chen J, Chen X, Wang G, Wang S, et al. Immunosuppressive tumor microenvironment in the progression, metastasis, and therapy of hepatocellular carcinoma: from bench to bedside. Exp Hematol Oncol 2024;13:72.
  • 10. Cheng K, Cai N, Zhu J, Yang X, Liang H, Zhang W. Tumorassociated macrophages in liver cancer: From mechanisms to therapy. Cancer Commun (Lond) 2022;42:1112-1140.
  • 11. Zheng H, Peng X, Yang S, Li X, Huang M, Wei S, et al. Targeting tumor-associated macrophages in hepatocellular carcinoma: biology, strategy, and immunotherapy. Cell Death Discov 2023;9:65.
  • 12. Chu X, Tian Y, Lv C. Decoding the spatiotemporal heterogeneity of tumor-associated macrophages. Mol Cancer 2024;23:150.
  • 13. Dong N, Shi X, Wang S, Gao Y, Kuang Z, Xie Q, et al. M2 macrophages mediate sorafenib resistance by secreting HGF in a feed-forward manner in hepatocellular carcinoma. Br J Cancer 2019;121:22-33.
  • 14. Chu T, Zhu G, Tang Z, Qu W, Yang R, Pan H, et al. Metabolism archetype cancer cells induce protumor TREM2+ macrophages via oxLDL-mediated metabolic interplay in hepatocellular carcinoma. Nat Commun 2025;16:6770.
  • 15. Tan J, Fan W, Liu T, Zhu B, Liu Y, Wang S, et al. TREM2+ macrophages suppress CD8+ T-cell infiltration after transarterial chemoembolisation in hepatocellular carcinoma. J Hepatol 2023;79:126-140.
  • 16. Ohtani K, Suzuki Y, Eda S, Kawai T, Kase T, Keshi H, et al. The membrane-type collectin CL-P1 is a scavenger receptor on vascular endothelial cells. J Biol Chem 2001;276:44222-44228.
  • 17. Lee CY, Lin Y, Bratman SV, Feng W, Kuo AH, Scheeren FA, et al. Neuregulin autocrine signaling promotes self-renewal of breast tumor-initiating cells by triggering HER2/HER3 activation. Cancer Res 2014;74:341-352.
  • 18. Zhang Z, Karthaus WR, Lee YS, Gao VR, Wu C, Russo JW, et al. Tumor Microenvironment-Derived NRG1 Promotes Antiandrogen Resistance in Prostate Cancer. Cancer Cell 2020;38:279-296.e9.
  • 19. Jardé T, Chan WH, Rossello FJ, Kaur Kahlon T, Theocharous M, Kurian Arackal T, et al. Mesenchymal niche-derived neuregulin-1 drives intestinal stem cell proliferation and regeneration of damaged epithelium. Cell Stem Cell 2020;27:646-662.e7.
  • 20. Hosni S, Kilian V, Klümper N, Gabbia D, Sieckmann K, Corvino D, et al. Adipocyte Precursor-Derived NRG1 Promotes Resistance to FGFR Inhibition in Urothelial Carcinoma. Cancer Res 2024;84:725-740.
  • 21. Schram AM, Odintsov I, Espinosa-Cotton M, Khodos I, Sisso WJ, Mattar MS, et al. Zenocutuzumab, a HER2xHER3 bispecific antibody, is effective therapy for tumors driven by NRG1 gene rearrangements. Cancer Discov 2022;12:1233-1247.
  • 22. Xiang S, Zheng Y, Wang M, Liu X, Zhang X, Chen D, et al. Comprehensive identification of NRG1 fusions in 25,203 patients with solid tumors. NPJ Precis Oncol 2025;9:262.
  • 23. De Boeck A, Pauwels P, Hensen K, Rummens JL, Westbroek W, Hendrix A, et al. Bone marrow-derived mesenchymal stem cells promote colorectal cancer progression through paracrine neuregulin 1/HER3 signalling. Gut 2013;62:550-560.
  • 24. Ou Q, Qiao X, Li Z, Niu L, Lei F, Cheng R, et al. Apoptosis releases hydrogen sulfide to inhibit Th17 cell differentiation. Cell Metab 2024;36:78-89.e5.
  • 25. Wang Z, Goto Y, Allevato MM, Wu VH, Saddawi-Konefka R, Gilardi M, et al. Disruption of the HER3-PI3K-mTOR oncogenic signaling axis and PD-1 blockade as a multimodal precision immunotherapy in head and neck cancer. Nat Commun 2021;12:2383.
  • 26. Wang HC, Haung LY, Wang CJ, Chao YJ, Hou YC, Yen CJ, et al. Tumor-associated macrophages promote resistance of hepatocellular carcinoma cells against sorafenib by activating CXCR2 signaling. J Biomed Sci 2022;29:99.
  • 27. Jang S, Ohtani K, Fukuoh A, Yoshizaki T, Fukuda M, Motomura W, et al. Scavenger receptor collectin placenta 1 (CLP1) predominantly mediates zymosan phagocytosis by human vascular endothelial cells. J Biol Chem 2009;284:3956-3965.
  • 28. Zhang X, Sun Y, Ma Y, Gao C, Zhang Y, Yang X, et al. Tumorassociated M2 macrophages in the immune microenvironment influence the progression of renal clear cell carcinoma by regulating M2 macrophage-associated genes. Front Oncol 2023;13:1157861.
  • 29. Su YL, Banerjee S, White SV, Kortylewski M. STAT3 in Tumor-Associated Myeloid Cells: Multitasking to Disrupt Immunity. Int J Mol Sci 2018;19:1803.
  • 30. Hu Z, Sui Q, Jin X, Shan G, Huang Y, Yi Y, et al. IL6-STAT3-C/EBPβ-IL6 positive feedback loop in tumor-associated macrophages promotes the EMT and metastasis of lung adenocarcinoma. J Exp Clin Cancer Res 2024;43:63.
  • 31. Hu B, Zou T, Qin W, Shen X, Su Y, Li J, et al. Inhibition of EGFR Overcomes Acquired Lenvatinib Resistance Driven by STAT3-ABCB1 Signaling in Hepatocellular Carcinoma. Cancer Res 2022;82:3845-3857.
  • 32. Guo Y, Zhou Y, Qin W. A CRISPR-Cas9 screen shows the combination efficacy of lenvatinib plus epidermal growth factor receptor inhibitors for treatment of liver cancer. Cancer Biol Med 2021;19:136-139.
  • 33. Lim M, Franses JW, Imperial R, Majeed U, Tsai J, Hsiehchen D. EGFR/ERBB2 Amplifications and Alterations Associated With Resistance to Lenvatinib in Hepatocellular Carcinoma. Gastroenterology 2023;164:1006-1008.e3.
  • 34. Ni Q, Chen Z, Zheng Q, Xie D, Li JJ, Cheng S, et al. Epithelial V-like antigen 1 promotes hepatocellular carcinoma growth and metastasis via the ERBB-PI3K-AKT pathway. Cancer Sci 2020;111:1500-1513.
  • 35. Liu DL, Lu LL, Dong LL, Liu Y, Bian XY, Lian BF, et al. MiR-17-5p and miR-20a-5p suppress postoperative metastasis of hepatocellular carcinoma via blocking HGF/ERBB3-NF-κB positive feedback loop. Theranostics 2020;10:3668-3683.
  • 36. de Visser KE, Joyce JA. The evolving tumor microenvironment: from cancer initiation to metastatic outgrowth. Cancer Cell 2023;41:374-403.
  • 37. Qu Y, Dou B, Tan H, Feng Y, Wang N, Wang D. Tumor microenvironment-driven non-cell-autonomous resistance to antineoplastic treatment. Mol Cancer 2019;18:69.
  • 38. Liu F, Li X, Zhang Y, Ge S, Shi Z, Liu Q, et al. Targeting tumorassociated macrophages to overcome immune checkpoint inhibitor resistance in hepatocellular carcinoma. J Exp Clin Cancer Res 2025;44:227.

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

Include:

COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling
Clin Mol Hepatol. 2026;32(2):772-786.   Published online January 20, 2026
DOI: https://doi.org/10.3350/cmh.2025.1059
Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.
Format:
Include:
COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling
Clin Mol Hepatol. 2026;32(2):772-786.   Published online January 20, 2026
DOI: https://doi.org/10.3350/cmh.2025.1059
Close

Figure

  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6

Related articles

COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling
Image Image Image Image Image Image Image
Figure 1. COLEC12high TAMs promote lenvatinib resistance in HCC. (A) Upper: MRI scans of lenvatinib-sensitive (Patients #1–3) and lenvatinib-resistant (Patients #4–6) tumors; dashed lines denote tumor margins. Lower: CIBERSORT analysis of bulk RNA-seq data from the same tumors (red, lenvatinib-sensitive; green, lenvatinib-resistant) (n=3). (B) Spontaneous HCC via hydrodynamic injection (Myc-OE, Trp53-sg, SB); bioluminescence shows resistance (no treatment=3; lenvatinib-induced therapy resistance=3). (C) Venn diagram of genes upregulated in resistant patient tumors (n=3) and resistant-mouse TAMs (n=3) (FC>1.5, P<0.05). (D) CRISPR/dCas9-SAM activation in THP-1 macrophages; cocultured with Huh7; Calcein-AM/PI viability assay under 2.5 μM lenvatinib. (E) Representative Calcein-AM/PI images and quantification after 48 h of coculture with lenvatinib (2.5 μM). (F) Immunofluorescence of CD68 (green) and COLEC12 (red) in macrophages; nuclei DAPI; scale bar, 25 μm. Kaplan–Meier plots of PFS/OS by TAM COLEC12 expression (log-rank test). Data are presented mean±standard error of the mean. Two-group: Student’s t-test; multiple groups: one-way ANOVA; survival: log-rank. HCC, hepatocellular carcinoma; MRI, magnetic resonance imaging; COLEC12, Collectin Subfamily Member 12; ns, not significant. *P<0.05. Source data available.
Figure 2. COLEC12high TAMs drive lenvatinib resistance through paracrine-mediated stemness reprogramming. (A) Workflow schematic: THP-1–derived macrophages, MDMs, or TAMs expressing COLEC12 (COLEC12OE) or vector control produce CM, which is applied to Huh7 and Hep3B cells. (B) Lenvatinib sensitivity in Huh7 cells (n=3): up, proliferation by CCK-8 under 2.5 µM lenvatinib; down, colony formation at 0, 2.5, and 5 µM. (C) Subcutaneous xenograft (n=6): tumor images, growth curves, final weights with intratumoral CM+lenvatinib or PBS. Orthotopic liver model: Liver-to-body weight ratios with intraperitoneal CM+lenvatinib or PBS. (D) Bulk RNA-seq of Huh7 cells treated with COLEC12OE CM or Vector CM, followed by pathway enrichment. Stemness assays (n=3): qRT–PCR for SOX2/OCT4/CD44/CD133; limiting dilution; immunofluorescence of tumors (scale bar, 100 µm); sphere formation (scale bar, 100 µm). (E) In vivo limiting dilution assay in subcutaneous models (n=6), with tumor-initiating cell frequencies shown in the chart and table. Data are presented mean±standard error of the mean. Two-group: Student’s t-test; multiple groups: one-way ANOVA; limiting dilution: Kruskal–Wallis. COLEC12, Collectin Subfamily Member 12; TAMs, tumor-associated macrophages; MDMs, monocyte-derived macrophages; COLEC12OE, COLEC12-overexpressing; CM, conditioned medium. *P<0.05, **P<0.01, ***P<0.001. Source data are available.
Figure 3. NRG1 mediates COLEC12high TAM-induced stemness and therapeutic resistance. (A) Overlap of genes upregulated in COLEC12OE macrophages with the THPA secretome. Heatmap of the top 20 induced genes. qRT-PCR, immunoblot, and ELISA confirm increased NRG1 expression and secretion in COLEC12OE THP-1 macrophages (n=3). (B) COLEC12OE CM increases lenvatinib resistance in Huh7 cells, whereas NRG1 knockdown reverses this effect, as shown by IC50, CCK-8 proliferation, colony formation, and apoptosis assays (n=3). (C) Subcutaneous Huh7 xenografts (n=6) and orthotopic liver models were treated with intratumoral or intraperitoneal injections of Vector-CM, COLEC12OE-CM, or COLEC12OE+shNRG1#1-CM with/without lenvatinib, showing tumor images, volumes, weights, and liver/body weight ratios (n=6). (D) qRT-PCR of SOX2/OCT4/CD44/CD133 in Huh7 under indicated CM (n=3). Tumorsphere formation under indicated CM (n=3). (E) Limiting dilution assays in vitro (M, n=3) and in vivo (N, n=6 per group), showing stem cell or tumor-initiating frequencies. Data are presented mean±standard error of the mean. Two group comparisons used Student’s t-test; multiple groups used one-way ANOVA; tumor initiation used Kruskal–Wallis. NRG1, Neuregulin 1; COLEC12, Collectin Subfamily Member 12; THPA, The Human Protein Atlas; CM, conditioned medium; COLEC12OE, COLEC12-overexpressing. *P<0.05, **P<0.01, ***P<0.001. Source data available.
Figure 4. COLEC12 drives NRG1 transcription via STAT1 sequestration-mediated STAT3 activation. (A) Predicted NRG1 transcription factors from hTFtarget, KnockTF, and JASPAR, highlighting shared TFs (STAT3/JUN/RELA/KLF4). qRT-PCR validation of TF knockdown and its effect on NRG1 expression in COLEC12-overexpressing THP-1 macrophages (n=3). STAT3 motif in the NRG1 promoter predicted by JASPAR. (B) STAT3 binds the NRG1 promoter as shown by ChIP-PCR and ChIP-qPCR in THP-1 macrophages (n=3). Luciferase assays demonstrate that STAT3 activates the WT NRG1 promoter but not the binding-site–deleted mutant (n=6). (C) Mass spectrometry identified STAT1 among COLEC12 interactors. Co-IP confirmed endogenous COLEC12–STAT1 binding in THP-1, MDM, and TAM macrophages (n=3). Immunofluorescence showed cytoplasmic colocalization of COLEC12 and STAT1; scale bar, 10 μm. (D) Docking predicts a direct COLEC12–STAT1 interface. Deletion of the COLEC12 intracellular domain abolishes its interaction with STAT1, as shown by co-IP (n=3). (E) Immunoblots showing that COLEC12-WT decreases p-STAT1 and increases p-STAT3 and NRG1 across THP-1, MDM, and TAM macrophages, whereas the intracellular-domain mutant shows no effect (n=3). (F) Immunofluorescence staining of pSTAT3 in THP-1–derived macrophages expressing control vector, COLEC12-WT, or COLEC12-Mut, with DAPI used for nuclear counterstaining. Scale bar, 10 μm. (G) Data are presented mean±standard error of the mean. Two group comparisons used Student’s t-test; multiple groups used one-way ANOVA; tumor initiation used Kruskal–Wallis. COLEC12, Collectin Subfamily Member 12; NRG1, Neuregulin 1; MDM, monocyte-derived macrophage; TAM, tumor-associated macrophage. *P<0.05, **P<0.01, ***P<0.001. Source data available.
Figure 5. NRG1 activates cancer cell-intrinsic HER2/HER3 signaling to drive lenvatinib resistance and stemness in HCC. (A) Immunoblots showing p-HER3, p-HER2, and p-AKT in Huh7 cells exposed to Vector CM, COLEC12OE CM, COLEC12OE CM+zenocutuzumab, or COLEC12OE CM from shNRG1 macrophages (n=3). (B) Lenvatinib IC50, proliferation (±2.5 μM lenvatinib and 0.01 μM zenocutuzumab), and time-course CCK-8 assays under the indicated CM conditions (n=3). (C) Colony formation under CM with 0–5 μM lenvatinib and tumorsphere assays (>100 μm), showing zenocutuzumab abolishes COLEC12OE CM–driven sphere growth (n=3; scale bar, 100 μm). (D) Huh7 subcutaneous and orthotopic xenografts treated with lenvatinib, zenocutuzumab, or the combination, showing maximal tumor suppression in combination groups (n=6). (E) Tumor growth and flow cytometry analysis in an immunocompetent C57BL/6 Hepa1-6 model treated with vehicle, lenvatinib, CDX-3379, or the combination, assessing intratumoral CD8+ T cells and TAMs, and IHC quantification showed p-HER3, p-HER2, and p-AKT levels (n=6). Scale bar, 50 μm. (F) Patient-derived HCC organoids and xenografts showed maximal growth inhibition with the lenvatinib–zenocutuzumab combination (n=6; scale bar, 200 μm). Data are presented mean±standard error of the mean. Two-tailed unpaired Student’s t-test or one-way ANOVA with multiple comparisons. NRG1, Neuregulin 1; HCC, hepatocellular carcinoma; CM, conditioned medium; COLEC12, Collectin Subfamily Member 12; COLEC12OE, COLEC12-overexpressing; IHC, immunohistochemistry. *P<0.05, **P<0.01, ***P<0.001. Source data available.
Figure 6. NRG1 serves as a biomarker and therapeutic target for lenvatinib resistance in HCC. (A) IHC of NRG1 in post-treatment tumors from SYSUCC HCC patients with low or high NRG1, with paired pre- and post-lenvatinib MRI; scale bar 25 μm. (B) Waterfall plots of maximal target-lesion change by RECIST v1.1 for NRG1 high (n=21) and NRG1 low (n=29); right panels show response distribution. (C) ROC for NRG1 predicting lenvatinib response by RECIST v1.1 (AUC=0.704). (D) Kaplan–Meier survival by tumor NRG1 level (IHC) in patients receiving preoperative lenvatinib; log-rank test. (E) Kaplan–Meier survival by serum NRG1 level (ELISA, median split) in patients receiving preoperative lenvatinib; log-rank test. (F) Model: COLEC12 overexpressing macrophages activate STAT3 and induce NRG1, which activates HER2/HER3 in HCC cells to drive stemness and resistance; zenocutuzumab blocks HER2/HER3 heterodimerization. NRG1, Neuregulin 1; HCC, hepatocellular carcinoma; IHC, immunohistochemistry; MRI, magnetic resonance imaging. (PD, progressive disease; SD, stable disease; PR, partial response) based on RECIST v1.1.
Graphical abstract

Figure 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Graphical abstract

COLEC12high tumor-associated macrophages orchestrate lenvatinib resistance and cancer stemness in hepatocellular carcinoma via paracrine NRG1-HER2/HER3 signaling
Clin Mol Hepatol : Clinical and Molecular Hepatology
Twitter

ABOUT

BROWSE ARTICLES

FOR CONTRIBUTORS

© The Korean Association for the Study of the Liver.