CCL11 promotes hepatocellular carcinoma recurrence after surgery by potentiating immunosuppressive CCR5 + CD206 + M2-like macrophages and promoting tumor invasiveness
Jiaqi Wang, Oscar Wai-Ho Yeung, Wenqi Qiu, Li Pang, Jiang Liu, Xinxiang Yang, Shinuan Zeng, Tao Ding, Zhe Wang, Zhenhua Hu, Tan To Cheung, Kwan Man, Kevin Tak-Pan Ng
TL;DR
This study shows that CCL11 contributes to liver cancer recurrence after surgery by suppressing the immune system and making cancer cells more invasive.
Contribution
The study identifies CCL11 as a key driver of hepatocellular carcinoma recurrence through dual mechanisms involving immunosuppressive macrophages and tumor invasiveness.
Findings
Postoperative CCL11 levels correlate with HCC recurrence and poor survival.
CCL11 recruits immunosuppressive M2-like macrophages and promotes tumor cell invasion.
Anti-CCL11 therapy reduces HCC recurrence in mouse models.
Abstract
Liver resection is the primary curative treatment for early-stage hepatocellular carcinoma (HCC); however, high recurrence rates remain a major challenge in the absence of effective prognostic and preventive strategies. Here, we identified surgery-induced C-C motif chemokine ligand 11 (CCL11) as a pivotal driver of HCC recurrence through dual mechanisms of immunosuppression and tumor invasiveness. Elevated postoperative circulating CCL11 levels correlated strongly with HCC recurrence and poorer survival, and their integration with clinical parameters enhanced the predictive accuracy of HCC recurrence. Mechanistically, hepatic injury-induced CCL11 recruited immunosuppressive CCR5+CD206+ M2-like macrophages into the residual liver. These macrophages exhibited enhanced PD-L1 expression via activation of the CCL11/IKK/IκB/NF-κB1 axis and promoted regulatory T cell (Treg) induction from…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7- —https://doi.org/10.13039/501100010877Shenzhen Science and Technology Innovation Commission
- —The Research Grants Council, Hong Kong (Grant Reference Number: T12-703/19R)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsImmune cells in cancer · Chemokine receptors and signaling · Hepatocellular Carcinoma Treatment and Prognosis
Introduction
Hepatocellular carcinoma (HCC), the fifth leading cause of cancer-related mortality globally, is one of the most aggressive cancers [1]. Liver resection remains the primary curative treatment for early-stage HCC; however, post-operative HCC recurrence rates are alarmingly high, reaching up to 70%. HCC recurrence is the leading cause of mortality in patients following liver surgery, largely due to the lack of standardized and effective treatments for recurrent HCC [2]. Consequently, elucidating novel mechanisms and developing effective therapeutic strategies to prevent HCC recurrence after surgery are critical for improving the prognosis of HCC patients.
Liver resection is the first-line surgical treatment for early-stage HCC, involving the excision of tumor-affected hepatic tissue while ensuring the preservation of a viable liver remnant capable of regeneration and maintenance of physiological hepatic functions. A growing body of evidence has indicated that an intensified inflammatory microenvironment within the liver remnant—fueled by the predisposed chronic inflammation and acute inflammation triggered by surgical stress—plays a critical role in driving postoperative HCC recurrence [3–7]. However, the precise inflammatory mediators that drive postoperative HCC recurrence remain incompletely characterized. Consequently, elucidating the mechanisms through which inflammatory factors shape the post-resection hepatic milieu—both in modulating the immune microenvironment and influencing malignant cell behavior—is essential for developing preventive and targeted therapeutic strategies to reduce HCC recurrence after liver resection.
Chemokines are critical mediators that initiate and regulate inflammatory responses. Emerging evidence has demonstrated that dysregulated chemokine signaling pathways contribute substantially to tumor progression and remodel the hepatic immune microenvironment [8]. Within the tumor microenvironment, chemokines are actively secreted by neoplastic cells, immune cells, and stromal cells, exerting direct regulatory effects on both tumor progression and antitumor immunity. To date, approximately 50 chemokines and 20 chemokine receptors have been characterized in humans [9]. Nevertheless, the precise biological functions and mechanistic contributions of many chemokine family members in remodeling the hepatic microenvironment and driving HCC recurrence after liver resection remain incompletely elucidated.
In this study, we profiled 10 inflammatory cytokines in post-resection peripheral blood and identified CCL11 as the most significantly upregulated cytokine in patients with HCC recurrence compared to non-recurrent patients. CCL11 is a member of the CC chemokine family, exerting a key mediator in various inflammation-related diseases [10]. The functional roles and therapeutic potentials of CCL11 in various human cancers and liver diseases are increasingly being characterized [11]. Notably, studies have also demonstrated that activated hepatic stellate cells (HSCs) can generate a “CCL11 storm” which subsequently triggers liver injury and fibrogenesis [12, 13]. Furthermore, CCL11 has emerged as a potential therapeutic target in non-alcoholic fatty liver disease [14]. Interestingly, our team’s previous study in liver transplantation of HCC patients revealed that elevated CCL11 level following liver transplantation is one of the significant risk factors for HCC recurrence after liver transplantation [15]. In addition, CCL11 expression is markedly upregulated in small-for-size liver grafts, which are associated with increased post-transplant hepatic injury and higher HCC recurrence rates [15]. These observations suggest CCL11 may play a mechanistic role in connecting post-surgical hepatic injury with HCC recurrence. While CCL11 has been implicated in various disease processes, its specific roles and mechanistic contributions to HCC progression and post-surgical recurrence remain poorly elucidated. The present study systematically examined CCL11-mediated regulation of both the hepatic immune microenvironment and malignant cell behavior, elucidating key molecular pathways involved. These findings potentially establish a mechanistic framework for developing predictive and targeted therapeutic interventions to prevent HCC recurrence after liver surgery, both for liver resection and liver transplantation in HCC patients.
Materials and methods
Patient specimens
Serum samples from the first cohort were collected on post-resection day 7 from 85 HCC patients treated at Queen Mary Hospital between 2005 and 2017 (Supplementary Table 1). The second cohort included paired tumor and non-tumor liver tissues from 110 HCC patients who underwent liver resection between 1998 and 2014 at the same institution (Supplementary Table 2). Available and qualified clinical samples from adult (age >=18) HCC patients were provided by the Surgical Tissue Bank, Department of Surgery, the University of Hong Kong, which was approved by Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB, Ref. No.: UW 05-359T/1022). Informed consent was obtained from all patients prior to sample collection.
Antibodies and reagents
Antibodies used for immunohistochemistry (IHC) and immunofluorescence (IF) included: anti-human CCL11 (#MAB320-SP) and anti-human/mouse CCR5 (#MAB182-SP) from R&D Systems; anti-human/mouse CD206 (#24595) from Cell Signaling Technology; and anti-human Mafk (#PA5-115534) and anti-human MMP13 (#MA5-43722) from Invitrogen. Antibodies used for flow cytometry analysis were purchased from Biolegend including anti-human CCR5 (AF700, #359115), anti-human CD68 (BV421, #333827), anti-human CD206 (PE, #321105), anti-human CD86 (APC, #374207), anti-mouse CD45 (BV510, #103137), anti-mouse CD11b (APC-CY7, #101225), anti-mouse F4/80 (AF700, #123130), anti-mouse CD206 (FITC, #141703), anti-mouse CCR5 (APC, #107011), anti-mouse PDL1 (BV650, #124336), anti-mouse CD3 (APC-CY7, #100221), anti-mouse CD4 (FITC, #116003), anti-mouse CD25 (AF700, #102024), anti-mouse Foxp3 (AF647, #126407), and anti-mouse CD8 (APC, #100711). Antibodies used for Western blot analysis anti-human/mouse p-IKKα/β (#2078T), anti-human/mouse p-IkBα (#2859T), anti-human/mouse NFKB1-p50 (#13586S), anti-human/mouse p-AKT (#9271S), and anti-human/mouse AKT (#9272S) from Cell Signaling Technology; anti-human/mouse PDL1 (#14-5983-82), anti-human/mouse p-PI3K (#PA5-104853), and anti-human/mouse Mafk (#PA5-115534) from Thermo Fisher Scientific; anti-human/mouse PI3K (#sc-423) and anti-human GAPDH (#sc47724) from Santa Cruz Biotechnology; and anti-human/mouse MMP13 (#18165-1-AP) and anti-mouse GAPDH (#60004-1-Ig) from Proteintech Group.
Cell culture
Human HCC cell lines Hep3B and Huh7, human monocyte cell line THP-1, and murine HCC cell line Hepa1-6 were purchased from American Type Culture Collection (ATCC, USA). Hep3B, Huh7, and Hepa1-6 cells were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen). THP-1 was cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 100 μg/mL streptomycin, and 100 U/mL penicillin (Sigma). Short tandem repeat (STR) profiling was verified for Hep3B and Huh7 authentication(Supplementary Materials). All cells were tested negative for mycoplasma contamination.
Luminex multiplex assay
The concentrations of 10 cytokines/chemokines were simultaneously quantified by the Human cytokine/chemokine magnetic bead panel kit (#HCYTMAG-60K, Millipore) [15, 16]. The 10 targeted cytokines/chemokines included CCL11, granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon alpha-2 (IFN-α2), interferon-gamma (IFN-γ), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-10 (IL-10), interferon gamma-induced protein 10 (IP10), monocyte chemoattractant protein-1 (MCP-1), and tumor necrosis factor-alpha (TNF-α). A volume of 25 μL of each serum sample was used to quantify 10 cytokines in a 96-well plate format according to the manufacturer’s instructions (Millipore). The fluorescent intensities of the beads were read by the MAGPIX system (Millipore). The Milliplex Analyst software was used for data analysis according to the manufacturer’s instructions (Luminex).
Quantitative real-time RT-PCR
Total RNA was isolated from HCC tumors, non-tumor liver tissues, and HCC cell lines using Trizol Reagent (Invitrogen). The extracted RNA was reverse transcribed into cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). mRNA expression levels were quantified by real-time PCR using SYBR Green Master Mix (Roche Diagnostics) on an ABI Prism ViiA7 Real-Time PCR System. The relative expression levels of targeted genes were determined based on our previous studies [17]. Primer sequences are listed in Supplementary Table 3.
Immunostaining
Paraffin-embedded tissue sections (4–5 µm thick) were mounted on glass slides, baked at 60 °C for 1 h, deparaffinized with xylene, and rehydrated through graded ethanol (99%, 95%, and 70%). For immunohistochemistry (IHC) staining, sections were stained with a mouse anti-human CCL11 antibody and visualized using a ready-to-use IHC detection kit. Immunofluorescence (IF) staining was performed with primary antibodies, followed by a 30-min incubation with secondary fluorescence antibodies. Nuclei were counterstained with DAPI, and sections were mounted with anti-fade medium. Fluorescence signals were captured using a fluorescence microscope and analyzed with Adobe Photoshop CS. Details of primary antibodies used are provided in Supplementary Table 4.
Generation of cell-conditioned media and measurement of CCL11 protein levels
For preparation of conditioned media (CM), different cell types, including fibroblast, macrophage, endothelial cell, epithelial cell and normal liver cell, were cultured in FBS-free media for 48 h. The collected supernatants were centrifuged at 3000 rpm for 10 min at 4 °C to remove dead cells and debris, followed by filtration through a 0.22 µm filter. For human serum CCL11, ELISA was performed using the Human CCL11/Eotaxin ELISA Kit (Proteintech) according to the manufacturer’s instructions. Similarly, mouse serum CCL11 levels were measured with the Mouse CCL11/Eotaxin ELISA Kit (Proteintech). Briefly, a CCL11 standard at 1000 pg/ml was serially diluted to generate a standard curve with concentrations of 500, 250, 125, 62.5, 31.25, and 15.6 pg/ml, using the zero standard (0 pg/ml) as the diluent. For each well, 100 μL of either serum sample or standard was added and incubated at room temperature for 2 h. After three washes, 100 μL of detection antibody solution was added and incubated for 1 hour at room temperature. Following another three washes, 100 μL of HRP-conjugated antibody solution was added and incubated in the dark for 40 min at room temperature. The reaction was completed by adding 100 μL of TMB working solution, followed by 100 μL of stop solution. Optical density was measured at 450 nm, with readings at 540 nm used for background correction.
Flow cytometry
Flow cytometry was performed following standard protocols [18]. Cells were pretreated with anti-mouse or anti-human Fc Block for 10 min at room temperature and subsequently stained with fluorophore-conjugated antibodies for 30 min at 4 °C in FACS buffer (PBS containing 2% bovine serum albumin). Dead cells were excluded using Zombie dye (Biolegend). After two washes with FACS buffer, the stained cells were analyzed on a CytoFLEX S flow cytometer (Beckman Coulter), and data were processed using FlowJo software (FlowJo). Details of the antibodies used for FACS are listed in Supplementary Table 5.
Macrophage polarization from THP-1
M2 macrophage polarization from THP-1 cells was performed as described by Yeung et al. [19]. Briefly, cells were stimulated with 320 nM PMA for 6 h, followed by treatment with 20 ng/mL recombinant interleukin-4 (IL-4, Invitrogen) and recombinant interleukin-13 (IL-13, R&D Systems) to induce M2 polarization. After 24 h of stimulation at 37 °C, cells were washed three times with PBS and cultured in RPMI 1640 medium containing 1% FBS before further experiments. Recombinant CCL11 protein was added to the polarized M2 macrophages for 48 h, while the control group was treated with PBS. After 48 h of stimulation, the cells were analyzed using flow cytometry and RT-qPCR.
CCR5+CD206+ macrophages, naïve CD4+ and CD8+ T cell isolation
Mouse CCR5^+^CD206^+^ macrophages were isolated from the livers of 6- to 8-week-old C57BL/6 mice (Jackson Laboratory) via a two-step liver perfusion process. The liver was first perfused with liver perfusion buffer (Ca^2+^ and Mg^2+^ free HBSS containing 0.5 mM EDTA), followed by liver digestion with liver digest solution (DMEM medium containing Collagenase IV). Non-parenchymal cells (NPCs) were obtained using a 100% Percoll density gradient (GE Healthcare Life Sciences, Cat#17544501) for single-cell isolation. Naïve CD4^+^ and CD8^+^ T cells were extracted from the spleens of 6- to 8-week-old C57BL/6 mice. Spleens were washed in Hank's solution, minced into small fragments, and ground using a nylon mesh and syringe plunger to obtain a cell suspension. Red blood cells were lysed using red blood cell lysis solution, and lymphocytes were isolated by centrifugation. The pellet was resuspended in Hank's solution or PBS for further analysis. Both NPCs and lymphocytes were stained with fluorophore-conjugated antibodies for 30 min at 4 °C in FACS buffer (PBS containing 2% bovine serum albumin). Dead cells were excluded using Zombie dye (Biolegend). After staining and two washes with FACS buffer, CCR5^+^CD206^+^ macrophages and naïve CD4^+^ and CD8^+^ T cells were isolated using flow cytometry sorting.
Treg differentiation and CD8+T cell proliferation assay
The methods of Treg differentiation and CD8^+^T cell proliferation assay were adopted from Pang et al. [20]. For the ex vitro experiment, 2 × 10^5^/ml naïve CD4^+^T cells + 2 × 10^5^/ml CCR5^+^CD206^+^ macrophage treated with 50 ng/ml recombinant CCL11 protein or PBS were cocultured and stimulated with 5 μg/ml anti-CD3/CD28 beads in 96-well plates for 3 days. The functional markers of Treg cells were analyzed by flow cytometry. The CD8^+^ T cells were labeled with 2 μM CFSE in PBS for 15 min at 37 °C and then washed three times with PBS. Then, 5 × 10^6^/ml CFSE-labeled CD8^+^ T cells + 2 × 10^5^/ml CCR5^+^CD206^+^ macrophage treated with 50 ng/ml recombinant CCL11 protein or PBS were also stimulated with 5 μg/ml anti-CD3/CD28 beads in 96-well plates for 3 days. The proliferation of cytotoxic CD8^+^ T cells will be analyzed by flow cytometry.
Cytokine antibody array analysis
The expression profile of 42 human cytokines (ENA-78, GCSF, GM-CSF, GRO, GRO-alpha, I-309, IL-1alpha, IL-1beta, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 p40/p70, IL-13, IL-15, IFN-gamma, MCP-1, MCP-2, MCP-3, MCSF, MDC, MIG, MIP-1delta, RANTES, SCF, SDF-1, TARC, TGF-beta1, TNF-alpha, TNF-beta, EGF, IGF-I, Angiogenin, Oncostatin M, Thrombopoietin, VEGF-A, PDGF BB, Leptin) in two culture (PBS-treated CCR5^+^M2-like macrophages and CCL11-treated CCR5^+^M2-like macrophages) conditioned media were determined using human cytokine antibody array (abcam). The protein array consisted of 42 distinct antibodies spotted in duplicates across four membranes, each corresponding to different conditioned media. All procedures were performed following the manufacturer’s instructions. The membranes were first blocked in blocking buffer for 30 min and then incubated at room temperature for 2 h with 1.5 mL of conditioned media collected from PBS-treated CCR5^+^M2-like macrophages or CCL11-treated CCR5^+^M2-like macrophages. After washing, the membranes were incubated with a diluted cocktail of biotinylated antibodies, followed by additional washes. Sandwiched antigens were detected by incubation with a peroxidase-labeled streptavidin solution for 2 h. Protein signals were visualized using enhanced chemiluminescence and captured on LAS-1000 X-ray film (FijiFilm). Positive and negative controls were included for signal normalization.
Western blot assay
Cultured cells were lysed using 1× cell lysis buffer (Cell Signaling, Cat#9803) supplemented with protease inhibitors. The lysates were incubated on ice for 15 min and then centrifuged at 20,000 × g for 15 min at 4 °C. The supernatant was collected, and total protein concentration was determined using the Bradford protein assay (Bio-Rad). For Western blot analysis, 20–30 µg of protein was loaded onto 10% SDS-polyacrylamide gels (SDS-PAGE) and resolved by electrophoresis. The proteins were then transferred onto PVDF membranes. Membranes were blocked with 5% skimmed milk in TBST buffer (TBS with 0.1% Tween-20) for 1 h at room temperature, followed by overnight incubation at 4 °C with primary antibodies. After washing, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies for 1 h at room temperature. Protein bands were detected using an enhanced chemiluminescence (ECL) reagent (GE Healthcare) and visualized on X-ray film. Details of the primary antibodies used are provided in Supplementary Table 6.
MTT proliferation assay
For the MTT assay, 1 × 10⁴ Hep3B cells were seeded per well in 96-well plates. Cells were treated with 50 ng/ml or 100 ng/ml recombinant CCL11 protein as experimental groups, while untreated cells served as controls. After treatment, the medium was discarded, and the cells were washed once with warm PBS. Next, 200 μl of 1× MTT working solution (prepared by adding 20 μl of 5 mg/ml MTT stock to 180 μl DMEM) was added to each well, followed by incubation at 37 °C for 3 h. Three wells containing only MTT solution were used as background controls. After incubation, the MTT solution was carefully removed, and 200 μl of DMSO was added to each well. Plates were shaken at room temperature for 5 min. Absorbance was measured at 570 nm and 650 nm, and O.D. values were calculated as (abs.570 nm–abs. 650 nm). The relative proliferation ratio was determined using the formula: Relative proliferation ratio = (Sample O.D. - Background O.D.)/(Control O.D. - Background O.D.). All samples were tested in triplicate, and the experiment was repeated at least three times.
Wound-healing assay
Cells were cultured for two days with DMEM with 10% FBS to form a tight cell monolayer, and then the serum was starved for 16 h. After serum starvation, the cell monolayer was wounded with a 10-µL plastic pipette tip. The remaining cells were washed twice with PBS to remove cell debris, and photos were taken on day 0. There were two different studies using a wound-healing assay. For the first study, Hep3B cells and Huh7 cells were separated into three groups and treated with 50 ng/ml recombinant CCL11 protein and 100 ng/ml recombinant CCL11 protein as experiment groups, while cells were treated without treatment as the control group in DMEM with free FBS. For the second study, cells were separated into five groups, treated with 50 ng/ml recombinant CCL11 protein + IgG, CCL11 protein + anti-CCR2, CCL11 protein + anti-CCR3, CCL11 protein + anti-CCR5 as experiment groups, while cells were treated without treatment as the control group in DMEM with 1% FBS. Then, cells were incubated at 37 °C for two days. At the indicated times, migrating cells at the wound front were photographed using an inverted microscope (Leica). Image-J software measured the wound closure percentage of the cleared area at each time point compared with time zero.
Transwell assay
Transwell assay without matrigel was carried out to detect the migration of Hep3B and Huh7 cells, while transwell assay with matrigel was carried out to detect the invasion of HCC cells. The density of 5 × 10^5^ cells/well was planted into the upper chamber of a transwell, combined with adding anti-CCR2, anti-CCR3, anti-CCR5, and LY294002 (inhibitor-PI3K) in DMEM with 1% FBS. After 6 h, recombinant human 50 ng/ml CCL11 protein or PBS in DMEM with 1% FBS was added into the lower chamber to attract Hep3B and Huh7 cells. Twenty-four hours later, migrated or invasive cancer cells were stained with crystal violet, and then the number of migrated cancer cells was photographed and counted.
RT2 assay
Hep3B cells were separated into three groups, treated with 50 ng/ml recombinant CCL11 protein + IgG, and 50 ng/ml recombinant CCL11 protein + anti-CCR3 as experiment groups, while cells were treated without treatment as the control group in DMEM with 1% FBS. Total RNA was extracted from Hep3B cells of these three groups using Trizol Reagent (Invitrogen). Total RNA was reverse transcribed with a high-capacity cDNA Reverse Transcription Kit (Applied Biosystems). The expression of 84 essential genes (APC, CD44, CDH1, CDH11, CDH6, FAT1, FXYD5, ITGA7, PNN, SYK, VEGFA, ITGB3, RPSA, CTNNA1, FN1, MCAM, MGAT5, MTSS1, MMP10, MMP11, MMP13, MMP2, MMP3, MMP7, MMP9, TIMP2, TIMP3, TIMP4, COL4A2, HPSE, SERPINE1, BRMS1, CDKN2A, HRAS, IL1B, KRAS, NF2, NME1, PTEN, RB1, TGFB1, TP53, MYC, CTBP1, GNRH1, MDM2, SSTR2, IGF1, IL18, TSHR, HGF, CCL7, CXCL12, TNFSF10, CXCR2, CXCR4, EPHB2, FGFR4, FLT4, KISS1R, MET, NR4A3, PLAUR, RORB, DENR, EWSR1, SET, SRC, TRPM1, HTATIP2, CHD4, ETV4, MTA1, MYCL, SMAD2, SMAD4, TCF20, CST7, CTSK, CTSL, CD82, KISS1, METAP2, NME4) involved in human tumor metastasis was profiled by RT² Profiler PCR Array (Qiagen, PAHS-028ZC) according to instructions provided by the manufacturer.
Mouse hepatic ischemia and reperfusion injury model
Male C57BL/6 mice (6–8 weeks old) underwent 50% nonlethal hepatic ischemia and reperfusion (I/R). The procedure involved clamping the right lateral and triangular lobes, along with half of the median lobe, using two microvascular clamps to occlude the right and middle branches of the portal vein just above the bifurcation for 90 min. Reperfusion was initiated by removing the clamps. At 2 h, 6 h, and 2 days post-reperfusion, mice were anesthetized, and tissue and blood samples were collected to measure CCL11 levels. The procedure was referred to the animal model of Pang et al. [18].
Mouse orthotopic HCC model
Firstly, Hepa1-6 cells were digested with 0.25% trypsin in the logarithmic phase for subcutaneous tumor implantation. A total of 3 × 10^6^ Hepa1-6 cells were suspended in 100 μl PBS and injected into the right inguinal fat pad of 5–7-week-old male C57BL6 mice. After 18 days, for orthotopic HCC implantation, the subcutaneous tumor of Hepa1-6 cells was minced into 1mm^3^ fragments and implanted into the left liver lobes of 5–7-week-old male C57BL6 mice. After tumor implantation, recombinant CCL11 protein (20 ug/kg, N = 5) or PBS (N = 5) was injected into the portal vein of each C57BL6 mouse to evaluate the tumor-promoting effect of CCL11 signaling on tumor-promoting and immune microenvironment effects. For detecting tumor progression, mice with implanted tumors were imaged every three days. The in vivo tumor growth in the liver was reflected by the intensity of the bioluminescent signals by injecting D-luciferin before imaging (100 mg/kg; volume 100 μL diluted in sterile water; Intraperitoneal injection), which were visualized and assessed using a Xenogen IVIS 100-cooled CCD camera. After 2 weeks, mice were anesthetized and collected for their tissue and blood for measuring the dynamic levels of CCL11 and other inflammatory cells and markers, respectively. The procedure was referred to the animal model of Pang et al. [20]. No animals were excluded from the analysis in the mouse orthotopic HCC model.
Surgical orthotopic HCC resection model
Firstly, for orthotopic HCC implantation, the subcutaneous tumor of Hepa1-6 cells was minced into 1 mm^3^ fragments and implanted into the left liver lobes of 5–7-week-old male C57BL6 mice. Mice bearing Hepa1-6 cells orthotopic tumors were anesthetized for tumor resection 18 days after implantation. Incisions were made in the upper middle abdomen, and the implanted tumors were surgically excised. Mice with successfully removed primary tumors, as determined by luciferase expression acquired by Xenogen IVIS Lumina system, were randomly divided into two groups by a masked technician. The mice were subsequently treated with PBS (N = 6) or with the anti-CCL11 neutralization antibody at 25 ug/kg of body weight (N = 6) for one time by intraperitoneal injection, once every three days for 30 days. The Xenogen IVIS Lumina system detected post-surgical recurrence of tumors every ten days. The survival and vital signs of every mouse were closely monitored after surgery. On day 50 after tumor resection, observation ended, and all alive mice were euthanized and collected for tissue and blood. The model of surgical orthotopic HCC resection was adopted from Li et al. [21]. Data from animals in the surgical orthotopic HCC resection model that died due to surgical complications and tumors that had not been cut cleanly after liver resection were excluded from the analysis. All subsequent post-operative care, drug administration (e.g., the person administering the anti-CCL11 or control was unaware of the group identity), and all outcome assessments, including in vivo imaging, quantification of tumor nodules, metastatic foci counting, and histological scoring, were performed by investigators blinded to the group allocation. All animal studies were approved by the Committee on the Use of Live Animals in Teaching and Research (CULATR 23-326) at the University of Hong Kong and conducted in accordance with the Animals Ordinance of Hong Kong.
Prognostic predictive nomogram construction for post-resection HCC recurrence
A receiver operating characteristic (ROC) analysis was performed to evaluate the accuracy of CCL11 and clinical variables in predicting HCC recurrence after liver resection. The optimal cut-off value for continuous variables was determined using the K-S statistics method of the ROC analysis. Univariate logistic regression analysis was conducted to assess the predictive significance of the variables, and those with p < 0.05 were included in the multivariate logistic regression analysis (Method = LR) to identify the optimal combination of predictors for the prediction model. The model was developed using a logistic regression algorithm and internally validated via 500 bootstrapping replications using the rms package in R (version 4.3.3) executed in RStudio (version 2023.6.0). A prognostic nomogram for the prediction model was constructed using the rms package. The associations between the prediction model and post-operative overall survival and disease-free survival were analyzed using Kaplan-Meier analysis and Cox proportional hazard regression analysis. The model’s performance was assessed by validated C-statistic, ROC curve analysis, and overall model quality metrics, as referenced in Ng et al. [10]. Statistical significance was considered at p < 0.05.
Data acquisition and analysis
To investigate the potential mechanisms of CCL11-treated CCR5^+^CD206^+^ macrophages-mediated immunosuppressive T-cell immunity, RNA-sequencing analysis was performed, comparing CCL11-treated CCR5^+^CD206^+^ macrophages with PBS-treated CCR5^+^CD206^+^ macrophages (NCBI SRA Accession number: SRR34743402 to SRR34743407). Total RNA was extracted, and cDNA libraries were constructed and sequenced by BGI Genomics (China) using an Illumina platform. The sequence reads were mapped to the human genome (version hg19) and analyzed. Differentially expressed genes (DEGs) were identified using thresholds of |log2(fold-change)| > 1 and p-value < 0.05. Hierarchical clustering and KEGG pathway analysis were conducted on these DEGs to compare gene expression profiles between the two groups. The highly upregulated DEGs and enriched KEGG pathways were used to guide mechanistic studies, providing insights into the immunosuppressive effects mediated by CCL11-treated CCR5^+^CD206^+^ macrophages. For Gene Set Enrichment Analysis (GSEA), two cohorts of upregulated metastasis-related genes—one from the CCL11-high group and the other from the CCR3-high group—were retrieved from TCGA data for liver hepatocellular carcinoma (https://www.cancer.gov/tcga). GSEA was performed using the GSEA software to analyze metastasis-associated pathway genes, utilizing gene sets from the Molecular Signatures Database, available from the Broad Institute (http://software.broadinstitute.org/gsea/index.jsp). Subsequently, KEGG pathway analysis was applied to compare these two cohorts of upregulated metastasis-related genes. This analysis identified highly significant enrichment pathways associated with metastasis in the CCL11/CCR3-high group, which represent the potential downstream pathways of CCL11/CCR3 signaling involved in tumor invasiveness. The correlation analysis and Kaplan-Meier analysis of TCGA Tumor in hepatocellular carcinoma were analyzed using the TIMER online tool (https://cistrome.shinyapps.io/timer/). For in vitro experiments, each experiment was independently repeated at least three times. Continuous variables were compared using Student’s t test and the Mann–Whitney U test, as applicable. Statistical significance was set at P < 0.05. All statistical analyses were performed using SPSS 29.0 or GraphPad Prism 10.0. Data were presented as mean ± s.d. in the figures.
Ethics statement
The use of clinical samples in this study was approved by the Surgical Tissue Bank. The collection and storage of clinical specimens for the Surgical Tissue Bank had gained consent from the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB, Ref. number: UW 05–359 T/1022). Informed consent was obtained from each patient before collecting tissue samples and associated data. Animal study was approved by the Committee on the Use of Live Animals in Teaching and Research (CULATR), The University of Hong Kong. All methods were performed in accordance with the relevant guidelines and regulations.
Results
Post-resection serum CCL11 protein was a significant prognostic factor for HCC patients after liver resection
To identify HCC recurrence-associated post-resection circulating cytokines/chemokines, a multiplexing assay was employed to compare the concentrations of 10 inflammation-related cytokines (GM-CSF, CCL11, IFN-α2, IFN-γ, IL-10, IL-6, IL-8, IP-10, MCP-1, and TNF-α) in the post-resection 1-week peripheral sera between HCC patients with and without post-resection HCC recurrence. Interestingly, among all the examined cytokines/chemokines, CCL11 was found to be the most significantly upregulated chemokine in patients with early HCC recurrence compared patients without early HCC recurrence (P = 0.008; Fig. 1A). Further validation of post-resection serum CCL11 protein in 85 HCC patients showed that post-operative serum CCL11 protein was significantly higher in HCC patients with HCC recurrence after liver resection (5-year and cumulative recurrence) compared to patients without HCC recurrence (Fig. 1B). ROC analysis showed that post-resection serum CCL11 could significantly distinguish patients with 5-year (AUC = 0.664; P < 0.001) and cumulative HCC recurrence (AUC = 0.643; P = 0.032) after liver resection (Fig. 1C). Moreover, Kaplan-Meier analysis indicated that a high level of serum CCL11 was significantly associated with poor disease-free survival of patients after liver resection (Log Rank = 4.259; P = 0.039; Fig. 1D). The univariate logistic regression analysis showed that post-operative circulating CCL11 was a significant predictor of 5-year HCC recurrence (OR = 4.295; P = 0.003; Table 1). Other significant predictors included tumor size, venous infiltration, PreOT-WBC, PreOT Albumin, PreOT AST, and PreOT ALT. Multivariate logistic regression analysis showed that post-operative circulating CCL11 was the significant covariate with the highest odds ratio (OR = 5.809; P = 0.005) along with venous infiltration (OR = 3.836; P = 0.044) and a low PreOT albumin (OR = 3.501; P = 0.037) in predicting 5-year HCC recurrence after liver (Table 1).Fig. 1. Upregulated CCL11 was significantly associated with HCC recurrence and poor survival after liver resection.A Multiplexing analysis of 10 post-operative circulating cytokines in HCC patients who received liver resection. Non-Recur, non-recurrence (N = 22); Recur, Recurrence (N = 26). B Comparison of post-resection serum CCL11 protein between patients with and without HCC recurrence after hepatectomy. C ROC analysis of post-resection serum CCL11 protein in distinguishing patients with HCC recurrence after hepatectomy. D Kaplan-Meier analysis of post-resection serum CCL11 protein (High group vs Low group) in association with patients’ disease-free survival. E Nomogram of CCL11-integrated prediction model in predicting 5-year HCC recurrence after hepatectomy. F Comparison of ROC curves of CCL11-integrated predictive model with reference model (without CCL11) in predicting 5-year HCC recurrence after hepatectomy. G The expression level of CCL11mRNA in tumor and non-tumor liver tissues of HCC patients as well as in healthy liver tissues. H Comparison of the expression level of non-tumoral CCL11 (ntCCL11) mRNA between patients with and without HCC recurrence. I ROC curves of ntCCL11 mRNA levels in predicting 5-Year HCC recurrence after hepatectomy. J Kaplan-Meier analysis of ntCCL11 mRNA in predicting disease-free survival. , P < 0.05; **, P < 0.01; **, P < 0.001; ns not significant.Table 1. Univariate and multivariate logistic regression analysis of variables for predicting 5-year HCC recurrence after liver resection.VariablesUnivariate logistic regressionPMultivariate logistic regressionPOdd ratio (95% CI)Odd ratio(95% CI)Circulating CCL11 (pg/ml) (≥83.75 vs <83.75)4.30(1.63–11.29)0.0035.81(1.70–19.84)0.005****Sex (Male vs Female)0.55(0.19–1.63)0.284Age (year)1.01(0.97–1.05)0.695Tumor size (cm) (≥4.45 vs <4.45)2.88(1.13–7.37)0.0272.72(0.70–10.50)0.147Venous infiltration (Yes vs No)4.68(1.73–12.71)0.0023.84(1.04–14.17)0.044Tumor number** (Multiple vs Single)1.66(0.59–4.65)0.338Differentiation (Poor vs Well/moderate)2.43(0.46–12.81)0.296PreOT AFP (ng/ml) (≥11.5 vs <11.5)2.15(0.90–5.18)0.087PreOT WBC (10^9^/l) (≤ 5.05 vs >5.05)3.10(1.18–8.13)0.0223.13(0.88–11.19)0.079PreOT Albumin (g/l) (≤ 42.5 vs >42.5)3.59(1.44–8.94)0.0063.50(1.08–11.36)0.037PreOT AST (u/l)** (≥37.5 vs <37.5)4.89(1.92–12.49)0.0012.48(0.71–8.66)0.155PreOT ALT (u/l)** (≥45 vs <45)3.38(1.24–9.17)0.0172.41(0.62–9.40)0.206**p < 0.05, **p < 0.01, ***p < 0.001.
Furthermore, multivariate logistic regression analysis applying the method of Backward Stepwise Likelihood Ratio (LR) was utilized to determine an optimal combination of predictors in predicting 5-year HCC recurrence after liver resection. A model consisting of 5 predictors, including PostOT serum CCL11, PreOT WBC, PreOT Albumin, PreOT AST, and venous infiltration, was determined (Supplementary Table 7). Therein, a 5-predictor nomogram was established to assess the probability of 5-year HCC recurrence after liver resection (Fig. 1E). Internal validation of the model, conducted through 500-repeat Bootstrapping, demonstrated its ability to predict the post-resection 5-year HCC recurrence with a concordance statistic (c-statistic) of 0.818. ROC analysis showed that the CCL11-integrated predictive model significantly predicted 5-year recurrence with an AUC value of 0.855 (P < 0.001), which achieved a better accuracy than the reference model (AUC = 0.811) without incorporating CCL11 (Supplementary Table 8 and Fig. 1F).
High levels of non-tumoral CCL11 were significantly associated with HCC recurrence and poor prognosis after liver resection
The expression level of CCL11 mRNA in 110 pairs of tumor (T) and non-tumor (NT) tissues of HCC patients undergoing liver resection, and 15 normal liver tissues, was examined. The expression level of CCL11 mRNA in non-tumoral liver tissues (ntCCL11) of HCC patients was significantly higher than in the tumor tissues and normal liver tissues (Fig. 1G). Moreover, the expression level of ntCCL11 mRNA in patients with post-resection HCC recurrence was significantly higher than in patients without HCC recurrence (Fig. 1H). ROC analysis showed that the ntCCL11 mRNA level could significantly distinguish patients with 5-year HCC recurrence after liver resection (AUC = 0.0653; P = 0.006； Fig. 1I). Notably, the high level of non-tumoral CCL11 was significantly associated with poor disease-free survival (Fig. 1J). T-tests and correlation analyses showed that a higher level of non-tumoral CCL11 (ntCCL11) was significantly associated with AFP, liver cirrhosis, advanced TNM stage, distant metastasis, and HCC recurrence (Table 2). In contrast, the expression level of CCL11 mRNA in tumor tissue (tCCL11) was not significantly different from the normal liver tissue (Fig. 1G), and its level showed no significant difference between HCC patients with and without HCC recurrence after liver resection (Supplementary Fig. 1A). The expression level of tCCL11 mRNA was not significantly associated with post-resection 5-year HCC recurrence (Supplementary Fig. 1B) and survival of HCC patients (Supplementary Fig. 1C). Unlike ntCCL11, the expression level of tCCL11 mRNA was not significantly correlated with all the clinicopathologics of HCC patients (Table 2).Table 2. Correlation between non-tumoral CCL11 (ntCCL11) and tumoral-CCL11 (tCCL11) with HCC clinicopathologic indexes.VariablesntCCL11PtCCL11PAge(year)0.1690.176 ≤55 (N = 60)8.04(3.53-13.88)8.71(0.27-13.68) >55 (N = 50)8.66(5.01-14.43)7.24(0.22-13.24)Sex0.7870.910 Male (N = 84)8.42(3.53-13.56)7.95(0.22-13.24) Female (N = 26)7.89(3.76-14.43)8.23(1.83-13.68)HBsAg0.2840.447 Negative (N = 16)8.82(3.76-14.43)9.87(1.83-13.30) Positive (N = 94)8.12(3.53-13.88)7.74(0.22-13.68)AFP (ng/ml)0.0460.416 ≤400 (N = 72)7.92(3.53-13.88)7.93(0.22-13.68) >400 (N = 37)8.67(3.82-14.43)7.97(0.27-12.12)Liver cirrhosis0.0400.474 No (N = 54)7.74(3.53-14.43)7.30(0.22-13.68) Yes (N = 56)8.70(4.29-13.88)8.06(0.27-13.24)Microvascular infiltration0.2260.937 No (N = 39)8.11(3.76-13.88)6.84(0.63-13.68) Yes (N = 71)8.35(3.53-14.43)8.43(0.22-13.24)Tumor differentiation0.3790.753 Well/moderate (N = 90)8.16(3.53-13.88)7.88(0.22-13.68) Poor (N = 20)8.51(3.99-14.43)8.81(0.95-13.24)7th AJCC TNM stage0.0030.873 I, II, IIIA (N = 91)7.98(3.53-14.43)7.82(0.22-13.68) IIIB/C, IV (N = 19)9.85(5.55-13.56)7.97(0.27-13.24)Distant metastasis0.0005*0.180 No (N = 101)7.97(3.53-14.43)8.07(0.22-13.68) Yes (N = 9)11.13(8.35-13.56)5.42(0.27-10.78)HCC recurrence0.0090.361 No (N = 28)7.26(3.53-12.48)8.72(0.27-13.68) Yes (N = 82)8.65(3.99-14.43)7.55(0.22-13.24)Total bilirubin (μmol/L) (N = 110)R = 0.090.346R = 0.0590.540Tumor size(cm)** (N = 110)R = 0.1050.273R = –1.010.294**p* < 0.05, **p < 0.01, ***p < 0.001.
Additionally, ischemia-reperfusion injury (IRI) is an inevitable hepatic injury produced by the general procedure of liver resection. To investigate whether CCL11 is induced after liver resection, an IRI mouse model was established to mimic the hepatic condition after liver resection. The results showed that the CCL11 protein and mRNA were significantly upregulated after IR at 2 h and reached the highest level at 6 h. The expression levels of CCL11 protein and mRNA at day 2 dropped but remained significantly higher than those of the sham control group (Supplementary Fig. 2A, B). To further investigate which cells would be responsible for upregulating CCL11 expression under these conditions, the expression level of CCL11 mRNA was examined in five different types of cells, including myofibroblast, macrophage, HUVEC, HIBEC, and normal hepatocyte-MIHA under hypoxic or normoxic conditions. CCL11 protein was upregulated in these 5 types of cells under hypoxia conditions compared to normoxic conditions, with a significant upregulation in macrophage and HUVEC cells. Among these cells, CCL11 was found to be expressed at the highest level by myofibroblasts in the normoxic or hypoxia conditions (Supplementary Fig. 2C).
Elevated CCL11 was positively associated with enhanced immunosuppressive CCR5+M2-like macrophages in the hepatic immune microenvironment
To identify the target immune cells in responding to CCL11, we first analyzed the expression correlations of CCL11 with its receptors CCR2, CCR3, and CCR5 in the hepatic microenvironment of HCC patients. The expression level of ntCCL11 mRNA exhibited the highest correlation with non-tumoral CCR5 (ntCCR5) mRNA (R = 0.271, P = 0.004) compared to ntCCR2 (R = 0.148, P = 0.12) and ntCCR3 (R = 0.194, P = 0.04, Fig. 2A), implying that CCL11 might be prone to interact with CCR5^+^ cells in hepatic tissues. To further identify the CCL11-interactive CCR5^+^immune cells in the hepatic immune microenvironment, public single-cell Sequencing datasets, including series GES 149614 and GES 202642, were retrieved and analyzed for the profiles of CCR5^+^ immune cells. Comparing to the low CCL11 expression (CCL11-low) group, the CCL11-high group were found to significantly amplify one of the subtypes of CCR5^+^immune cells (Fig. 2B), which was identified to be the CCR5^+^macrophages (Fig. 2C). We validated that the expression level of ntCCL11 mRNA in non-tumoral tissues was significantly associated with CD206 mRNA (M2 macrophage marker) but not significantly correlated with iNOS mRNA (M1 macrophage marker) (Fig. 2D). Furthermore, the results from immunohistochemical and immunofluorescent staining experiments demonstrated a positive correlation between CCL11 protein and the number of CCR5^+^M2-like macrophages in non-tumor liver tissue (Fig. 2E and Supplementary Fig. 3A, B). Importantly, Kaplan-Meier analysis showed that the higher infiltration number of CCR5^+^M2-like macrophages in the non-tumoral tissues of HCC patients was significantly associated with poor overall survival and poor disease-free survival after liver resection (Fig. 2F, G).Fig. 2. Elevated CCL11 was positively associated with CCR5 + CD206 + M2-like macrophages in hepatic immune microenvironment.A Correlation analyses of CCL11 levels with its receptors CCR2, CCR3 and CCR5 in non-tumoral liver tissue. B UMAP plot of CCR5+immune cells in CCL11-low group and CCL11-high group. C UMAP plot of all immune cell types. D Correlations of CCL11 levels with iNOS and CD206 in non-tumor liver tissues. E Correlation of CCL11 with CCR5 + M2-like macrophages. Kaplan-Meier survival curve of (F) overall survival and G disease-free survival based on CCR5 + M2-like macrophages. *, P < 0.05; **, P < 0.01; ***, P < 0.001.
CCL11 could activate IKK-IkB-NFkB1 pathway and PD-L1 expression of CCR5+M2-like macrophages
To further explore the functions of CCL11 in regulating CCR5^+^M2-like macrophages, an in vitro monocyte-to-macrophage polarization model (using the THP-1 monocyte) was employed. The administration of recombinant CCL11 protein could significantly promote the activation of CCR5^+^CD206^+^M2-like macrophages (Fig. 3A) but not the CCR5^+^CD86^+^M1-like macrophages (Supplementary Fig. 4A), from M0-like macrophages. Agreed with the above finding, CCL11 could significantly increase the expression of several immunosuppressive markers in the CCR5^+^M2-like macrophages, including PD-L1, TIM-3, and IL-10 (Fig. 3B). Moreover, CCL11-enriched medium could stimulate the transcription and secretion of MCP-1 and IL-8 in CCR5^+^M2-like macrophages compared with PBS-treated CCR5^+^M2-like macrophages (Fig. 3C-D). Functionally, CCL11-treated CCR5^+^M2-like macrophages could significantly stimulate the induction of Tregs from naïve CD4^+^ T (Fig. 3E). In contrast, CCL11-treated CCR5^+^M2-like macrophages could not affect the proliferation of CD8^+^ T cells (Supplementary Fig. 4B). These data suggested that CCL11-treated CCR5^+^M2-like macrophages could promote the induction of immunosuppressive Tregs cells without affecting CD8^+^T cells.Fig. 3CCL11 enhanced the activation of immunosuppressive CCR5 + M2-like macrophages.A Flow cytometry quantification of the number of monocyte-differentiated CCR5 + CD206 + M2-like macrophages after administration of recombinant human CCL11 (rhCCL11) protein. Control group: PBS treatment. B The expression Levels of M2-macrophage-related immunosuppressive markers and cytokines in CCL11-treated and PBS-treated CCR5 + M2-like macrophages. C Cytokine array analysis of secreted cytokines in CCL11-treated and PBS-treated CCR5 + M2-like macrophages. Red dot boxes indicate two highly upregulated cytokines: 1, MCP1; 2, IL-8. D mRNA Levels of IL-8 and MCP-1 in CCL11-treated and PBS-treated CCR5 + M2-like macrophages. E CCL11-treated CCR5 + M2-like macrophages enhanced the induction of CD4+ Foxp3+Tregs from naive CD4 + T compared to the PBS-treated control group. F Top 10 enriched KEGG pathways from differentially expressed genes in the CCL11-treated CCR5 + M2-like macrophages. G Gene set enrichment analysis of Toll-like receptor signaling pathway. H Levels of p-IKKα/β, p-IkBα, NFkB1-P50, and PD-L1 expression via western blot analysis. I Schematic diagram of the molecular mechanism of CCL11 in regulating CCR5 + M2-like macrophages. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns not significant.
For exploring the molecular mechanisms of CCL11 in regulating the functions of CCR5^+^M2-like macrophages, RNA-sequencing analysis was carried out to compare the transcription profile between CCL11-treated and PBS-treated CCR5^+^M2-like macrophages. The volcano plot illustrated that CCL11 could alter the expressions of more than 6039 upregulated and 7584 downregulated transcript genes in CCR5^+^M2-like macrophages (Supplementary Fig. 5A). Among the top 10 enriched pathways, the “lymphocyte chemotaxis” and “chemokine-mediated signaling pathway” were the most remarkably enriched functions in CCL11-treated CCR5^+^M2-like macrophages (Supplementary Fig. 5B). In addition, among the top 10 enriched KEGG pathways, the pathway of the “PD-L1 expression and PD-1 checkpoint pathway in cancer” was one of the most remarkably enriched pathways in CCL11-treated CCR5^+^M2-like macrophages (Fig. 3F). Within the PD-L1 expression pathway, the Toll-like receptor signaling pathway was one of the influencing pathways in regulating the expression of PD-L1 (Supplementary Fig. 5C). From the further GSEA, NFkB1 was identified to be the highest enriched gene in the Toll-like receptor signaling pathway, showing a higher Normalized Enrichment Score (NES) in CCL11-treated CCR5^+^M2-like macrophages (Fig. 3G). Western blot analysis verified that CCL11-treated CCR5^+^M2-like macrophages exhibited higher levels of p-IKKα/β, p-IkBα, NFkB1-P50, and PD-L1 protein compared to the PBS-treated control (Fig. 3H). Moreover, administration of DHE (IKK inhibitor) on CCL11-treated CCR5^+^M2-like macrophages could suppress the expression levels of p-IKKα/β, p-IkBα, NFkB1-P50, and PD-L1 proteins (Fig. 3H), indicating an important mechanism of CCL11 in promoting the expression of PD-L1 of CCR5^+^M2-like macrophages via IKK/IkB/NFkB1 axis (Fig. 3I).
CCL11 promoted the migration and invasion of HCC cells in a CCR3 receptor-dependent manner
The expression level of CCL11 mRNA could not be detected in the normal liver cell line and HCC cell lines (Supplementary Fig. 6A). Different HCC cell lines could express different levels of CCR2, CCR3, and CCR5 mRNAs, with the dominant levels of CCR3 and CCR5 compared to the normal liver cell line (Supplementary Fig. 6B). Among all the HCC cell lines, Hep3B and Huh7 expressed relatively higher levels of these receptors (Supplementary Fig. 6B). Functionally, the administration of recombinant CCL11 protein did not alter the proliferation rate of HCC cells (Supplementary Fig. 6C). Administration of recombinant CCL11 protein could significantly enhance the wound-healing ability of Hep3B (Fig. 4A) and Huh7 cell lines in a dose-dependent manner (Supplementary Fig. 7A), suggesting a migration-promoting effect of CCL11 on HCC. Notably, blockage of CCR3 could cause the most inhibitory effect on CCL11-promoting migration on Hep3B (Fig. 4B) and Huh7 cell lines (Supplementary Fig. 7B), compared to the inhibitory effect of CCR2 or CCR5. Additional experiments validated that blockage of CCR3 could exhibit a more inhibitory effect on CCL11-promoting migration on Hep3B (Fig. 4C) and Huh7 cell lines (Supplementary Fig. 7C) than the blockage of CCR2 or CCR5. Moreover, blockage of CCR3 significantly inhibited CCL11-induced invasion in Hep3B (Fig. 4D) and Huh7 (Supplementary Fig. 7D) cell lines, and the inhibitory effect was higher than that of the inhibition of CCR2 or CCR5. Kaplan–Meier survival analysis on our clinical samples verified that a high level of CCR3 mRNA was significantly associated with poor overall survival (Fig. 4E) and disease-free survival (Fig. 4F) of HCC patients after liver resection.Fig. 4. Blockage of receptor CCR3 caused the most inhibitory effect on CCL11-promoting migration and invasion of HCC cells.A Wound-healing assay to evaluate the migration of Hep3B with different concentrations of recombinant CCL11 protein (0, 12.5, 25, 50 ng/ml). B Wound-healing assay of Hep3B when blockage of CCL11-receptor CCR2, CCR3 and CCR5 in CCL11-enriched condition; C Transwell migration assay of Hep3B when blockage of CCL11-receptor CCR2, CCR3 and CCR5 in CCL11-enriched condition. D Transwell invasion assay of Hep3B when blockage of CCL11-receptor CCR2, CCR3 and CCR5 in CCL11-enriched condition. Kaplan-Meier survival curve of (E) Overall survival and F disease-free survival based on the expression level of CCR3 mRNA in HCC patients. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns not significant.
CCL11/CCR3 axis promoted HCC metastasis through activating PI3K/AKT/MafK pathway and upregulating MMP13 expression
To identify the molecular mechanisms of CCL11 in regulating HCC invasion and metastasis, the RT^2^ PCR Array consisting of 84 genes associated with tumor metastasis was analyzed in HCC cells upon the stimulation of CCL11. We found that MMP13 was the most up-regulated gene in the CCL11-treated HCC cells, compared to the control cells (Fig. 5A). Meanwhile, MMP13 was also the most down-regulated gene when CCR3 was blocked by anti-CCR3 antibody (Fig. 5B), indicating that MMP13 was the potential downstream effector protein of the CCL11-CCR3 axis in HCC. To investigate CCL11-CCR3 downstream signaling in HCC, TCGA data were retrieved and analyzed. Results revealed 122 upregulated metastasis-related genes in ntCCL11-high compared to ntCCL11-low groups, and 138 in tCCR3-high relative to tCCR3-low groups (Supplementary Fig. 8A). Among them, 92 metastasis-associated genes were identified to be co-upregulated in both the ntCCL11-high and tCCR3-high groups (Fig. 5C). Within the top 20 KEGG pathways, Focal adhesion and PI3K-Akt signaling were the most significant enrichment pathways in the ntCCL11-high and tCCR3-high groups (Fig. 5D). KEGG analysis demonstrated PI3K-Akt signaling’s involvement in focal adhesion (Supplementary Fig. 8B). Notably, MMP13 was identified as a downstream effector of PI3K-Akt (Supplementary Fig. 8C), suggesting a potential mechanistic explanation for how PI3K-Akt signaling may mediate the upregulation of MMP13 in HCC following CCL11-CCR3 activation. Validation experiments indicated that blockage of PI3K could reverse the CCL11-induced MMP13 upregulation (Fig. 5E) and functionally decrease the CCL11-induced migration (Fig. 5F and Supplementary Fig. 9A) and invasion (Fig. 5G and Supplementary Fig. 9B) abilities of HCC cells. Therefore, these results verified that the PI3K-Akt pathway would be essential for CCL11-CCR3 to promote MMP13 expression and the metastatic ability of HCC. We next explored Akt-associated transcription factors (TFs) potentially mediating MMP13 upregulation. Bioinformatic analysis revealed nine MMP13-promoter-binding TFs and thirteen Akt-downstream TFs, with Mafk as the sole overlap (Fig. 6A). JASPAR analysis identified two Mafk-binding motifs in the MMP13 promoter (Fig. 6B), implicating the PI3K/Akt/Mafk axis in CCL11-CCR3-driven MMP13 induction. rCCL11 protein promoted Mafk nuclear translocation, reversible by PI3K inhibition (Fig. 6C). Western blot confirmed rCCL11-activated PI3K/Akt signaling pathway and -upregulated Mafk/MMP13, all suppressed by PI3K inhibition (Fig. 6D). TIMER database analysis showed positive correlation between CCR3 and MMP13 mRNA in HCC (Supplementary Fig. 10A), with high MMP13 linked to poor survival (Supplementary Fig. 10B).Fig. 5CCL11/CCR3 axis promoted MMP13 expression and metastasis through activating PI3K/Akt pathway.A RT2 PCR array analysis of 84 metastasis-associated genes in CCL11-treated and PBS-treated (Ctrl) HCC cells. B RT2 PCR array analysis of 84 metastasis-associated genes in CCL11 treatment (50 ng/ml) with anti-CCR3, compared to CCL11 single treatment. C Identification of upregulated genes in CCL11-high and CCR3-high in HCC from TCGA database. D Top 20 KEGG pathways associated with tumor metastasis among the 92 upregulated genes in CCL11-high & CCR3-high group. E The expression level of MMP13 mRNA CCL11-treated HCC cells followed by administration of LY294002 (PI3K inhibitor), compared with DMSO-treated control cells. F Migration assay and G Invasion assay of CCL11-treated HCC cells with or without adding LY294002. *, P < 0.05; **, P < 0.01; ***, P < 0.001.Fig. 6CCL11 could enhance the metastatic ability of HCC.A The bioinformatic analysis of MMP13-promoter binding transcript factor and AKT-activating transcript factor. B The promoter region of MMP13 gene contains 2 binding motifs for the AKT downstream transcription factor Mafk, located at nucleotide sequences -232 to -218 and -57 to -43. C Inhibition of PI3K significantly reduced the CCL11-induced Mafk transloation from cytoplasm to nuclear in HCC cells. D Western blot analysis of p-PI3K, PI3K, p-Akt, Akt, Mafk, and MMP13 protein in CCL11-treated HCC cells with and without PI3K inhibition. E Schematic diagram of mouse orthotopic HCC tumor model, followed by injection of rCCL11 Protein or PBS via portal vein. F IVIS evaluations of orthotopic HCC tumor in rCCL11 Protein group or PBS group. G Liver organs dissected from rCCL11 group or PBS group. H Western blot analysis of PI3K/Akt/MafK/MMP13 axis in rCCL11 protein group and PBS group. I Schematic diagram of the molecular mechanism of CCL11 in promoting HCC progression. *, P < 0.05; **, P < 0.01.
Moreover, the effect of upregulation of CCL11 on HCC was validated in a mouse orthotopic HCC tumor model (Fig. 6E). The in vivo result showed that liver tumor luciferase intensity and tumor volume in the rCCL11-injected group were higher than the PBS-injected group (Fig. 6F). In addition, two out of five mice (40%) developed extrahepatic metastasis among the rCCL11-injected mice, while there was no metastatic nodule found in the PBS-injected group (Fig. 6G). Western blot analysis of the samples showed that the expression levels of PI3K, phospho-PI3K, Akt, phospho-Akt, MafK, and MMP13 proteins were up-regulated in the rCCL11-injected group, compared to the PBS-injected group (Fig. 6H). Therefore, these results indicated the molecular mechanism of CCL11 in promoting HCC growth and metastasis by activating the CCR3/PI3K/Akt/MafK/MMP13 axis (Fig. 6I).
Targeted inhibition of CCL11 after tumor resection could prevent liver cancer recurrence and prolong the survival of mice
To investigate the therapeutic potential of targeted inhibition of CCL11 in preventing HCC recurrence after liver resection, an in vivo resection model on the orthotopic HCC tumor was developed in mice. CCL11-neutralizgin antibody was administrated at the 2 days after hepatectomy for the tumor (Fig. 7A). At 17 days after tumor resection, recurrent tumors were detected in 3 out of 5 mice (60%) were detected to have larger recurred tumor in the control group, while only one mouse (16.7%) was detected to have a smaller recurred tumor in the anti-CCL11 therapy group (Fig. 7B, C). At day 47 after tumor resection, all mice in the control group were observed with tumor recurrence and 80% (4 out of 5) of mice died of tumor recurrence. In contrast, the recurrence rate in the anti-CCL11 therapy group at the experimental endpoint remained 16.7% (Fig. 7B, C). The overall survival rate of mice was significantly improved from 20 to 80% after anti-CCL11 therapy (Fig. 7D). The body weight during treatment between the control and anti-CCL11 therapy groups had no significant difference (Supplementary Fig. 11).Fig. 7. Inhibition of CCL11 could prevent HCC recurrence after tumor resection.A Schematic diagram of orthotopic HCC resection model, followed by anti-CCL11 therapy. B IVIS evaluations of intrahepatic and extrahepatic tumors in the control group and anti-CCL11 treatment group. C Recurred rates of recurred tumors in the control group and anti-CCL11treatment group. D Kaplan-Meier analysis for overall survival of mice after treatment. E Summary of molecular mechanisms of CCL11 in promoting HCC recurrence after liver resection. *, P < 0.05.
Discussion
HCC recurrence is the major cause of mortality in patients following liver surgery. Delineation of the underlying mechanisms leading to HCC recurrence after liver surgery is critical for preventing HCC recurrence and improving the prognosis of HCC patients. Tumor relapse after surgery involves very complicated processes involving dynamic interactions between tumor cells and the inflammatory hepatic microenvironment resulting from both chronic and acute inflammations [22]. The inflammatory microenvironment within the liver comprises a heterogeneous population of inflammatory cells and a diverse spectrum of inflammatory mediators, such as chemokines, cytokines, and prostaglandins [23]. Chemokines have gained significant recognition as pivotal mediators, playing a critical role not only in facilitating the migration of cancer cells to metastatic sites but also in orchestrating the recruitment of diverse cell populations to both tumor and non-malignant microenvironments. This underscores the fact that dysregulation of specific chemokines has been established as a significant risk factor for postoperative tumor recurrence. With over 50 identified chemokines, the functional roles of many in tumor recurrence remain poorly characterized. CCL11 has been identified as an inflammatory peptide secreted by eosinophils. Nevertheless, its specific roles in tumor recurrence, as well as its interactions with infiltrating immune cells and tumor cells, particularly in HCC, remain largely uncharacterized.
In our study, we identified that post-operative serum CCL11 protein was the most significantly upregulated chemokine in patients with HCC recurrence compared to patients without HCC recurrence. We further validated that elevated levels of post-resection CCL11 was a significant risk factor for HCC recurrence. Moreover, increased post-operative serum CCL11 level was predictive of HCC recurrence and correlated with poorer disease-free survival outcomes following liver resection. Importantly, post-operative serum CCL11 level could be an independent prognostic factor of 5-year HCC recurrence following liver resection. Existing literature has established associations between elevated CCL11 levels and adverse oncological outcomes, including tumor recurrence and poor prognosis, across multiple malignancies such as renal cell carcinoma [24], breast cancer [25], and esophageal cancer [26]. Although CCL11 plays crucial roles in liver diseases [11], its roles in HCC pathogenesis and recurrence are still elusive. Our previous study has demonstrated that upregulation of post-operative CCL11 was a significant risk factor of HCC recurrence after liver transplantation [15]. Taken together, our current findings provide compelling clinical evidence that elevated postoperative CCL11 levels represent a significant risk factor for HCC recurrence across hepatic surgical interventions, including both partial hepatectomy and liver transplantation.
Over the past decade, numerous predictors have been identified to evaluate the risk of tumor recurrence [27]. However, existing risk assessment models have predominantly focused on pre-surgical factors, such as tumor stage, size, other tumor characteristics, and serum AFP level, while post-operative risk factors remain largely unexplored. In this study, we integrated post-operative circulating CCL11 levels with pre-operative predictors to develop a clinically applicable prediction model for HCC recurrence following liver resection. This model, structured as a 5-predictor nomogram, incorporates post-resection serum CCL11, PreOT WBC, PreOT albumin, PreOT SGOT, and venous infiltration. The CCL11-integrated model achieved an accuracy of 0.85 in predicting HCC recurrence after liver resection, outperforming the reference model constructed solely on pre-operative predictors. Therefore, these findings establish CCL11 as a novel postoperative prognostic biomarker and suggest that its combination with preoperative factors could significantly enhance the prediction accuracy for HCC recurrence following resection. Notably, this study developed a clinically practical, user-friendly nomogram to assess and stratify patients at high risk of HCC recurrence after resection.
To date, numerous studies have focused extensively on the relationship between the tumor microenvironment and tumor recurrence [28, 29], often overlooking the critical role of the non-tumoral hepatic microenvironment in influencing tumor progression and recurrence. Our study revealed that CCL11 expression in non-tumor liver tissues of HCC patients was significantly higher than in tumor tissues or normal liver tissues. Furthermore, an elevated level of non-tumoral CCL11, rather than tumoral CCL11, was significantly associated with tumor recurrence and poor prognosis following liver resection. Overexpression of non-tumoral CCL11 was also significantly linked to advanced TNM stage and distant metastasis. The microenvironment of non-tumor liver tissue during hepatectomy reflects the microenvironment of the post-resection liver remnant. Growing evidence highlights the importance of peritumoral (or non-tumoral) tissues in determining the risk of HCC recurrence after hepatectomy. One study identified a gene expression profile in adjacent non-tumor tissue that correlates with late tumor recurrence following resection of small HCC [30]. Our previous work also demonstrated that infiltration of M2 macrophages in adjacent non-tumor tissue was associated with early recurrence and poor survival in HCC patients [19]. Similarly, in gliomas, aggressive invasion and metastasis predominantly occur within brain tissue, often leading to post-resection recurrence [31]. Winkler et al. suggested that peritumor angiogenesis plays a pivotal role in facilitating the invasion and progression of primary gliomas [32]. Thus, our study offers the first clinical evidence that CCL11 in non-tumor liver tissue (liver remnant) may significantly influence HCC recurrence risk post-resection.
CCL11 is upregulated in various liver injuries such as liver cirrhosis, non-alcoholic fatty liver disease, and drug-induced liver injury [12–14, 33]. CCL11 can be secreted by stromal cells and immune cells under inflammatory conditions like hypoxia [34–36]. Given that hepatic ischemia-reperfusion injury (IRI) is an unavoidable hepatic injury during the operation of liver resection to induce cytokine release [18], we assessed its effect on CCL11. We found that IRI significantly elevated CCL11 in mice. Moreover, the in vitro model showed that CCL11 protein was upregulated in different cell types under hypoxic conditions compared to normoxic conditions, particularly in myofibroblasts with the highest levels. This finding is consistent with the known role of myofibroblasts in secreting inflammatory chemokines and cytokines in response to environmental stimuli [37, 38]. Thus, our results indicated that myofibroblasts might be a key cellular source of CCL11 in hepatic injury. The specificity and molecular mechanisms of myofibroblasts in CCL11 secretion after liver resection need to be further characterized.
The “Seed and Soil” hypothesis in oncology proposes that the dissemination and progression of cancer depend on the dynamic interplay between cancer cells (the “Seed”) and the specific microenvironment of the target organ (the “Soil”) [7]. This concept suggests that cancer cells are more likely to thrive and proliferate in specific tissues with a conducive environment, underscoring the critical roles of both tumor cell characteristics and the properties of the tissue microenvironment in determining the outcomes of tumor metastasis and recurrence. CCL11, a chemotactic cytokine, has the potential to influence both tumor cells and non-tumor cells through ligand-receptor interactions. Given that both tumor and non-tumor cells express CCL11 receptors, we hypothesized that upregulation of CCL11 in the hepatic environment would impact HCC cells (“Seed”) and non-tumor cells within the hepatic microenvironment (“Soil”) in a receptor-dependent manner.
A growing body of research has highlighted the influential roles of specific cytokines and chemokines in modulating the hepatic immune microenvironment to promote tumor progression and metastasis. [21, 39, 40]. Our previous studies have shown that upregulation of the CXCL10/CXCR3 signaling pathway during the early phase following liver transplantation can induce Tregs and endothelial progenitor cells, thereby fostering an immunosuppressive hepatic microenvironment [3, 41]. We have also demonstrated the immunosuppressive effects mediated by regulatory B cells through the CD40/CD40L signaling pathway and by myeloid-derived suppressor cells (MDSCs) via the LTR4/CXCR3 axis, both of which contribute to the promotion of HCC progression [42, 43]. In this study, we observed that upregulated CCL11 expression was positively correlated with the presence of immunosuppressive CCR5^+^CD206^+^M2-like macrophages within the hepatic microenvironment. Furthermore, a higher infiltration rate of CCR5^+^CD206^+^M2-like macrophages in non-tumoral tissue was associated with a poorer prognosis following liver resection. This finding aligns with the observation that elevated levels of hepatic and circulating CCL11 were positively linked to a higher incidence of post-surgical HCC recurrence. Functionally, CCL11 could promote the activation of immunosuppressive CCR5^+^M2-like macrophages from monocytes. Moreover, CCL11-activated CCR5^+^M2-like macrophages facilitated the induction of Tregs from naïve CD4^+^ T cells, potentially fostering an immunosuppressive hepatic microenvironment conducive to tumor progression and recurrence. Additionally, CCL11-activated CCR5^+^M2-like macrophages enhanced the production of immunoregulatory cytokines, including MCP-1 and IL-8. MCP-1 has been demonstrated to promote the recruitment and polarization of macrophages toward the immunosuppressive M2 phenotype within the liver tumor microenvironment [44, 45], creating a more immunosuppressive milieu. Consequently, upregulated secretion of MCP-1 induced by CCL11 may further amplify the generation of CCR5^+^M2-like macrophages. Meanwhile, IL-8 has been demonstrated to play a role in recruiting CD4^+^ T cells to the liver tumor microenvironment and enhancing the differentiation of immunosuppressive Tregs from CD4^+^ T cells [46, 47]. This suggests that CCL11-induced IL-8 upregulation may contribute a more immunosuppressive hepatic microenvironment. Collectively, our study unveils a novel mechanism by which CCL11 modulates CCR5^+^M2-like macrophages to foster an immunosuppressive hepatic microenvironment (“Soil”), thereby promoting HCC recurrence after liver resection.
Furthermore, our study revealed that enrichment of CCL11 could activate the IKK-IκB-NF-κB1 signaling pathway in CCR5^+^M2-like macrophages resulting in the upregulation of PD-L1. PD-L1, a critical immune checkpoint regulator, plays a pivotal role in establishing an immunosuppressive tumor microenvironment, enabling tumors to evade immune surveillance and promote progression [48]. By binding to its receptor PD-1 on T cells, PD-L1 suppresses T-cell activation and proliferation, dampening anti-tumor immune responses [49]. Additionally, PD-L1 enhances the differentiation of naïve CD4^+^ T cells into regulatory T cells (Tregs) [50], which further enhances immunosuppression. In the presence of PD-L1, Tregs expand and maintain their suppressive functions more effectively, reinforcing tumor immune evasion [51]. Thus, our study elucidates the molecular mechanisms by which CCL11 enhances the immunosuppressive function of CCR5^+^M2-like macrophages via PD-L1 upregulation. These findings provide critical insights into how surgery-induced inflammation fosters a hepatic immunosuppressive microenvironment, promoting HCC recurrence after liver resection.
The role of CCL11 in HCC remains poorly understood. Although previous studies have implicated CCL11 in promoting tumor progression and invasiveness in other cancers [25, 52, 53], its mechanisms in HCC are distinct. For example, CCL11 has been reported to drive CCR3^+^ tumor cell proliferation in renal cell carcinoma [24] and promote tumor metastasis through the MDSCs-CCL11-ERK/AKT-Epithelial–mesenchymal transition (EMT) axis in non-small cell lung cancer [54]. In contrast, our findings demonstrated that CCL11 did not affect HCC cell proliferation. It significantly enhanced the invasiveness of HCC cells. CCL11 signals through three cognate receptors: CCR2, CCR3, and CCR5 [55]. Our functional studies demonstrated that CCR3 blockade most effectively attenuated CCL11-induced invasion in HCC cells, identifying CCR3 as the principal receptor mediating CCL11’s pro-invasive effects. Mechanistically, we found that CCL11-CCR3 engagement activates the PI3K/Akt/MafK axis, leading to upregulation of MMP13. MMP12 is a metalloproteinase that promotes metastasis through extracellular matrix degradation and liberation of pro-angiogenic factors [56, 57]. These results delineate a signaling cascade (CCL11-CCR3/PI3K/Akt/MafK/MMP13) driving HCC invasion. Importantly, our findings provide mechanistic insight into how surgery-induced CCL11 elevation in remnant liver tissue may enhance the invasiveness of residual neoplastic cells (the “Seed”), thereby increasing the risk of HCC recurrence post-resection.
We used an orthotopic HCC model in immunocompetent mice to faithfully recapitulate the complex post-surgical tumor microenvironment, where immune and tumor cells coexist and interact. CCL11 can chemoattract both immune and tumor cells via receptor-mediated interaction [11]. Consequently, the enhancement of tumor growth and metastasis by CCL11 was attributed to its dual effects in promoting an immunosuppressive axis in immune cells and a pro-invasiveness axis in tumor cells. Future studies employing cell-specific CCR3 or CCR5 knockout models, or immunodeficient mice reconstituted with specific immune populations, would help delineate the individual weight of these two parallel mechanisms in driving tumor recurrence after liver surgery.
Chemokines and their receptors have emerged as important therapeutic targets for different human diseases [21, 58–60]. Targeted inhibition of CCL11 is an effective therapeutic strategy in treating inflammation-associated diseases, such as Allergic disorders, asthma, and ulcerative colitis [58–61]. In this study, we developed an orthotopic HCC mouse model with liver resection to investigate the therapeutic potential of CCL11 inhibition. Strikingly, anti-CCL11 treatment significantly reduced post-resection tumor recurrence (17% vs 100% in controls) and improved survival outcomes. To our knowledge, this represents the first experimental evidence identifying CCL11 as a promising therapeutic target for preventing post-operative HCC recurrence. While preventing recurrence is recognized as crucial for improving HCC patient survival [62, 63], current clinical practice lacks effective standardized treatments for this purpose. With the addition of our previous finding of post-transplant CCL11 in HCC recurrence after liver transplantation [15], the present study therefore provides both mechanistic insights and a potential therapeutic strategy that could translate to improved outcomes for HCC patients undergoing either liver resection or liver transplantation.
Conclusion
In conclusion, our study has elucidated two novel mechanisms by which CCL11 drives post-resection HCC recurrence (Fig. 7E). First, it activates the CCR3/PI3K/Akt/MafK/MMP13 axis to enhance the invasiveness of residual HCC cells (“Seed”). Second, it potentiates immunosuppressive CCR5^+^CD206^+^M2-like macrophages via IKK/IκB/NF-κB1/PD-L1 signaling, shaping a pro-tumorigenic hepatic microenvironment (“Soil”). These dual effects synergistically promote immune evasion and tumor recurrence. Our work provides mechanistic insight into how post-operative ischemia-reperfusion injury fuels recurrence through inflammatory pathways, offering a scientific foundation for developing targeted therapies to prevent HCC recurrence after liver resection.
Supplementary information
Supplementary Tables and Figures Original Western Blot Images Cell Line Authentication Test Checklist
The reference list from the paper itself. Each links out to its DOI / PubMed record.