Deciphering hepatoma cell resistance to tyrosine kinase inhibitors: insights from a Liver-on-a-Chip model unveiling tumor endothelial cell mechanisms

Abstract

Liver cancer represents a significant global burden in terms of cancer-related mortality, with resistance to anti-angiogenic drugs such as Sorafenib and Lenvatinib presenting a formidable challenge. Tumor angiogenesis, characterized by the formation of new blood vessels within tumors, plays a pivotal role in cancer progression and metastasis. Tumor endothelial cells, specialized endothelial cells lining tumor blood vessels, exhibit unique phenotypic and functional traits that drive aberrant vessel formation and contribute to therapy resistance. CD105, a cell-surface glycoprotein that is highly expressed on endothelial cells during angiogenesis, including tumor endothelial cells, regulates endothelial cell proliferation, migration, and vessel formation by modulating transforming growth factor-beta (TGF-β) signaling pathways. Elevated CD105 expression on tumor endothelial cells correlates with increased angiogenic activity and poor prognosis in cancer patients. Targeting CD105 with antibodies presents a promising strategy to inhibit tumor angiogenesis and disrupt tumor vasculature, offering potential therapeutic benefits by interfering with the tumor microenvironment and inhibiting its progression. This study investigates tumor angiogenesis through a three-dimensional (3D) microfluidic co-culture system incorporating endothelial cells and hepatocellular carcinoma (HCC) cells. The primary focus is on the role of CD105 expression within the liver tumor microenvironment and its contribution to increased chemoresistance. Additionally, this research examines the influence of CD105 expression on the efficacy of tyrosine kinase inhibitors (TKIs) and its pivotal function in facilitating angiogenesis in liver tumors. The proposed microfluidic chip model investigates liver cancer cell interactions within a microfluidic chip model designed to simulate aspects of liver tumor angiogenesis.

---

Introduction

Hepatocellular carcinoma (HCC) is the sixth most commonly diagnosed cancer and is distinguished by its heightened vascularization.¹ It exhibits a notable predisposition toward vascular invasion, and its angiogenic activity is directly associated with the likelihood of vascular invasion and carries prognostic significance.² Enhancing the efficacy of anti-angiogenic therapy and addressing resistance to therapeutic drugs are paramount objectives in advancing non-surgical treatments for HCC. Its incidence statistics are primarily attributed to etiological factors, including chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infections, alcohol consumption, and non-alcoholic fatty liver disease (NAFLD).^3–9

Angiogenesis, a foundational hallmark of cancer,¹⁰ significantly influences the genesis and progression of HCC.^11,12 Hepatic tumor growth necessitates neovascularization,¹³ leading to pronounced neoangiogenesis in HCC development, characterizing it as a hypervascular tumor.¹⁴ Tumor expansion relies on angiogenesis, with the tumor vasculature serving vital roles in nutrient/oxygen supply and waste disposal in tumor tissue, and acting as a gatekeeper for tumor-cell metastasis to distant organs.^15,16

Sorafenib, a multi-target kinase inhibitor, impedes tumor-cell proliferation by targeting Raf1, B-Raf, and kinases in the Ras/Raf/MEK/ERK pathway. It also inhibits angiogenesis by antagonizing receptors such as c-Kit, FLT-3, VEGFR-2, VEGFR-3, PDGFR-β, and other tyrosine kinases.^17,18 Lenvatinib, approved for first-line HCC treatment, inhibits receptors involved in angiogenesis and tumor growth, which include VEGFRs 1–3, FGFRs 1–4, PDGFRα, RET, and c-KIT. In addition to its strong anti-angiogenic effects, Lenvatinib is expected to act as a unique immunomodulator by blocking multiple kinases, including FGFRs.¹⁹ Sorafenib primarily exerts its pro-apoptotic, anti-angiogenic, and antitumor effects by inhibiting VEGFRs, unlike Lenvatinib, which also significantly impacts FGFRs.²⁰ Endoglin (CD105), a transforming growth factor β receptor, is linked to cell proliferation and induced by hypoxia. It is more abundantly expressed in angiogenic endothelial cells than traditional markers like CD31 and vWF. Studies highlight CD105's role in liver fibrosis and hepatocellular carcinoma (HCC) progression.^21–33 CD105 marks activated endothelial cells (ECs) and is primarily expressed in the angiogenic endothelium of tumors like HCC. Its expression in HCC is associated with changes in CD105-positive ECs in and around the tumor.³⁴ CD105 occurs in a soluble form and has been detected in the serum of both healthy individuals and patients with various pathologies, including hepatocellular carcinoma.^29,30 Elevated levels of soluble CD105 have been documented in patients with colorectal and breast carcinomas. Evidence indicates that soluble CD105, when fused to an Fc domain, exhibits anti-angiogenic properties, reducing both in vitro and in vivo angiogenesis and inhibiting tumor growth in colorectal carcinoma. Soluble CD105 plays a dual role in cancer, inhibiting tumor-associated angiogenesis while also promoting a malignant phenotype in myeloma and breast cancer cells.^35–39 CD105-expressing tumor endothelial cells are more resistant to apoptosis, exhibit enhanced motility, and have an increased ability to adhere to HCC. A study demonstrated that tumor endothelial cells expressing CD105 were more resistant to Sorafenib compared to normal endothelial cells like HUVECs. Sorafenib, targeting Raf, VEGFR2, PDGFR, and other receptors that are involved in angiogenesis, inhibits cell proliferation and survival by blocking MAPK, STAT3, and Akt signaling pathways.^40–42 Sorafenib inhibited STAT3, Akt, and MAPK pathways in all endothelial cells, but p-STAT3 and p-Akt levels were higher in CD105-expressing tumor endothelial cells compared to normal endothelial cells, both before and after treatment.⁴³ Another investigation discovered that therapy involving endoglin expression, when combined with Sorafenib, exhibited increased effectiveness against tumors in a preclinical mouse model. Additionally, a phase I clinical trial revealed promising results when the humanized anti-endoglin monoclonal antibody, TRC105, was combined with Sorafenib in patients with advanced HCC. The trial demonstrated evidence of efficacy, with objective and relatively durable responses observed in a portion of patients.⁴⁴

The microfluidic chip model used herein is an effective tool for studying liver cancer cell interactions by simulating key aspects of liver tumor angiogenesis. The chip's design incorporates separate chambers for HepG2 cancer cells, HUVEC endothelial cells, and their co-culture. It allows for precise manipulation and dynamic flow conditions within a controlled microenvironment, which is ideal for examining cell–cell interactions. The chip supports on-chip cell culture, enabling detailed investigations into the influence of endothelial cells on the cancer cell environment and their role in drug-resistance mechanisms. Additionally, the chip's reproducibility, standardization, and accessibility do not compromise its functionality, making it a valuable resource for robustly exploring the mechanisms of drug resistance. This reliability ensures consistent results across multiple experiments, facilitating comparative studies and accelerating the development of new therapeutic strategies. Our findings highlight the critical role of angiogenesis in the progression of HCC and its resistance to the drugs Sorafenib and Lenvatinib. By incorporating recombinant CD105 into our study, we have been able to observe significant endothelial-cell phenotypic transitions that are closely associated with therapeutic resistance. These observations are critical as they provide deeper insights into the dynamic behavior of liver cancer, particularly how it adapts and resists current treatment options.

Materials and methods

Microdevice design and fabrication

The microfluidic chip was intricately designed using AutoCAD software (Autodesk, San Francisco, CA, USA), and resembled a leaf shape with three distinctive chambers: a co-culture chamber and two individual monoculture chambers for human endothelial and liver cancer cells. This design incorporated pillars to augment perfusion of medium, cell viability, and gelatin methacryloyl (GelMA) retention. The chip exhibited a simple layout featuring a single inlet and three separate outlets for monocultures and co-culture. For fabrication, PDMS (polydimethylsiloxane) was employed. Using PDMS for bio-chip fabrication offers key advantages, including gas permeability for effective oxygen and carbon dioxide exchange, which is crucial for cellular respiration. It supports stable in vitro conditions mimicking the physiological environment, with a pH value of 7.2 to 7.5, temperature of 37 °C, and 95% relative humidity. Additionally, PDMS maintains a controlled 5% CO₂ concentration, ensuring constant pH levels, thus enhancing the accuracy and relevance of biological experiments.⁴⁵ The PDMS microchannel (SYLGARD 184, Dow Corning, Midland, MI, USA) with a 100 μm height was produced using photolithography. A silicon wafer was coated with SU-8 negative photoresist (SU-8 2100, Microchem, Westborough, MA, USA), and precise microchannel dimensions were defined through photoresist patterning. A mixture of the PDMS elastomeric base and a hardening agent in a 10 : 1 ratio was poured onto the surface of the silicon mold. Following a one-hour bake at 65 °C, the PDMS layer was delicately separated from the silicon wafer. Holes, each with a 1 mm diameter, were meticulously drilled into the PDMS layer to facilitate fluidic connections with external tubing. In order to ensure a strong bond between the PDMS layer and the glass bottom, oxygen plasma treatment was thoroughly administered.

Cell culture and cell viability assay in a 2D setting

In this research, we utilized HepG2 (HB-8065, ATCC, Manassas, VA, USA) cells to represent hepatocellular carcinoma and HUVEC (CRL-1658, ATCC) cells as endothelial cells. The culture medium for both cell types comprised Dulbecco's modified Eagle's medium (DMEM) (11965092, Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin. Cells were maintained under standard culture conditions in a humidified 37 °C incubator with 5% CO₂.

For drug sensitivity analysis, varying concentrations of Sorafenib (S7397, Selleckchem, Houston, TX, USA) and Lenvatinib (SML3017, Sigma-Aldrich) were tested (ranging from 100 to 0.1 μM) to establish their respective IC₅₀ values. Liver cancer and endothelial cells were seeded at a density of 5 × 10³ cells per well in a 96-well plate with a complete growth medium. 24 hours later, the medium was replaced, and CCK-8 (CK04, Dojindo Laboratories, Kumamoto, Japan) was administered according to the manufacturer's instructions. Subsequently, drug concentrations equivalent to the IC₅₀ values were applied to the co-culture chamber of the chip to evaluate drug resistance.

Biocompatible gelatin methacryloyl (GelMA)

The 3D bio-scaffolds applied in this investigation were crafted from GelMA (gelatin methacryloyl). We integrated GelMA into our bio-chip to enhance cell viability and function. GelMA supports cell adhesion and proliferation and creates a 3D environment like native tissue. Its permeability maintains proper nutrient and oxygen levels, ensuring cellular metabolism. GelMA's compatibility with PDMS allows precise cell distribution, improving experimental reproducibility and insights into cellular responses. The GelMA solution was formulated by dissolving a 10% (wt/vol) gelatin solution derived from porcine skin (G2500, Sigma-Aldrich, Burlington, MA, USA) in 100 ml of Dulbecco's phosphate-buffered saline (DPBS) (D1408, Sigma-Aldrich) at 40 °C. Subsequently, 1 g of methacrylic anhydride (MA) solution (276685, Sigma Aldrich) was gradually introduced to the gelatin solution while it was magnetically stirred at 300 rpm for 3 hours, ensuring a uniform gelatin–MA reaction. Post-reaction, 200 ml of DPBS at 40 °C was incorporated. The resultant mixture was then placed into a dialysis membrane in a 5 L container filled with deionized water, maintained at 40 °C, and the water changed twice daily for 7 days. This step aimed to eliminate any unreacted MA. Following dialysis, the pH of the GelMA was adjusted to 7.4 using a 1 M sodium hydroxide solution, and then it was filtered through a 0.22 μm sterile syringe filter with a PES membrane (SLGVR33RB, Merck Millipore, Darmstadt, Germany). The dialyzed GelMA solutions were subsequently frozen at −80 °C, lyophilized, and stored at 4 °C.

For utilization, the freeze-dried 5% (wt/vol) GelMA and a 0.5% (wt/vol) photoinitiator solution containing the photoinitiator 2-hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone (or Irgacure 2959, I-2959) (410 896, Sigma-Aldrich) were dissolved in PBS at 37 °C, creating the GelMA prepolymer solution. Photo-crosslinking was achieved by exposing the GelMA prepolymer to UV light with an intensity of 8.6 mW cm⁻¹ within the wavelength range of 320–500 nm for 30 seconds at room temperature. This process employed an OmniCure S1500 UV lamp (Lumen Dynamics, San Francisco, CA, USA). The experimental procedures described herein were conducted in accordance with the protocol established by Y.-C. Chen et al., with minor modifications.⁴⁶ In our experimental approach, GelMA underwent crosslinking via UV light exposure. Notably, prior investigations⁴⁷ have demonstrated that UV exposure during GelMA crosslinking does not compromise cell viability. This assurance supports our confidence in the efficacy of UV crosslinking to preserve cell viability and uphold the integrity of our experimental outcomes.

Cell culture in the bio-chip

The chip is designed to create distinct culture chambers for HepG2 cancer cells, HUVEC endothelial cells, and a co-culture of HepG2 and HUVEC cells. HepG2 cancer cells mixed with GelMA are initially introduced into the chip through a common inlet. These cells navigate the chip, directed towards the co-culture and monoculture cancer chambers, with any excess cells exiting through the outlets. At this stage, only the outlet for the HepG2 monoculture is open, while the other two are blocked. An exposure mask is then used to cross-link the GelMA selectively, defining the HepG2 culture chamber. UV light exposure, precisely calibrated, ensures the formation of these defined chambers. After thoroughly washing with PBS to remove excess cells, only the HepG2 cells remain in their designated chamber.

Next, HUVEC endothelial cells are introduced using a similar method. An exposure mask is again used to create culture chambers for the HUVEC cells. Finally, the co-culture chamber is prepared by mixing both cell types in a 1 : 1 ratio in GelMA and loading them through the common inlet, following the same procedure. After washing off excess cells, the chip is ready for incubation and experimentation. The cell-loaded chip is maintained in a humidified 37 °C incubator with 5% CO₂. This process establishes controlled co-culture and monoculture environments, providing a versatile platform for various biomedical applications. The comprehensive operation of the chip is detailed step-by-step in Fig. S2 (ESI†).

Fluorescence staining in bio-chip

To visualize the spatial arrangement of mono-cultured cancer cells, endothelial cells, and their co-culture within the bio-chip, we utilized CellTracker Red CMTPX dye (C34552, Invitrogen), CellTracker Green CMFDA (C2925, Invitrogen), and/or CellTracker Blue CMAC dye (C2110, Invitrogen) in accordance with the manufacturer's guidelines. To evaluate cell viability within the bio-chip, we performed a Live/Dead assay (L3224, Invitrogen) following the manufacturer's protocol.

To perform an immunofluorescent assay (IFA) for specific protein targets using fluorescence, a bio-chip was fixed with a 4% paraformaldehyde solution for 20 minutes at room temperature. Subsequently, a series of washes with 1× PBS were conducted. To enable permeabilization, a solution containing 0.1% Triton X-100 in 1× PBS was applied at room temperature for 30 minutes, followed by further PBS washes. Blocking of the chip was carried out using a 2% bovine serum albumin (BSA) solution for 60 minutes at room temperature. To distinguish between HepG2 and HUVEC cells in co-culture, we performed counter-labeling using liver-specific and endothelial-specific markers. Immunostaining utilized three biomarkers, anti-CD105 (ab114052, Abcam), anti-CD31 (ab76533, Abcam) and albumin (ab207327, Abcam) antibodies, at a 1 : 200 concentration. The primary antibodies were allowed to incubate with the chip for 3 hours at room temperature, followed by thorough washing. For the secondary antibody, a goat anti-rabbit IgG antibody labeled with DyLight488 (GTX213110-04, GeneTex, Alton Pkwy Irvine, CA, USA) was used at a 1 : 200 concentration. The chip underwent a one-hour incubation with the secondary antibody and was extensively washed with 1× PBS to remove unbound antibodies. Finally, a mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI) was applied to facilitate nuclear staining and preserve the sample.

Investigating the role of CD105 in mediating endothelial-cell-induced drug resistance to Sorafenib and Lenvatinib in liver cancer cells

Several studies have reported that CD105 expression in endothelial cells (ECs) of tumor tissues and serum levels of soluble endoglin correlate positively with advanced clinical stages and poor prognosis.²⁶ Elevated soluble-CD105 levels in patients with metastatic melanoma further suggest a link between these levels and cancer progression. These findings indicate that CD105 can allow identification of patients with advanced disease and prediction of metastasis risk, and show changes in response to chemotherapy. It may also help assess treatment response, particularly to anti-angiogenic therapies, and allow monitoring for disease recurrence.³⁷

To explore the effect of CD105 on the viability of liver cancer cells (HepG2) and endothelial cells (HUVEC) when treated with Sorafenib and Lenvatinib, we introduced recombinant CD105 (Ab54338, Abcam, Woburn, MA, USA) into the culture medium of our microfluidic chip system. This approach aimed to more closely analyze the tumor microenvironment and how CD105 influences cellular responses to these tyrosine kinase inhibitors. The rationale for including recombinant CD105 proteins in the medium is based on the established role of CD105 in modulating tumor biology. CD105 is overexpressed in the tumor endothelium and is associated with advanced cancer and adverse clinical outcomes.^48–50

By adding recombinant CD105, we aimed to investigate the potential effects of CD105 on drug-resistance mechanisms. Studies have demonstrated that CD105 can decrease drug sensitivity, promote invasion and migration, and inhibit apoptosis in cancer cells. Hence, its presence in the medium might alter the interaction between the drugs and the cells by enhancing these processes. This setup enabled continuous monitoring of changes in cell behavior in response to treatment with Sorafenib and Lenvatinib, both in the presence and absence of recombinant CD105. This method provides valuable insights into how CD105 contributes to liver cancer treatment resistance and supports its potential role as a prognostic marker and therapeutic target.

Results and discussion

Liver lobule

The liver serves as a vital organ, comprising functional units known as liver lobules. Each fulfills crucial tasks essential for overall health and metabolism. With a unique architecture measuring approximately 1.1 mm in diameter and 1.7 mm in length,^51,52 these microscopically structured lobules consist of hepatic cells, including hepatocytes and endothelial cells, which are organized in a specific pattern to facilitate liver functions such as metabolism, detoxification, and bile production.³¹ Endothelial cells are essential components of blood vessels, and angiogenesis involves the proliferation and migration of these cells to form new blood vessels. In the context of liver cancer, the angiogenic liver encompasses diverse cell types, including endothelial cells, cancer stem cells, cancer cells, pericytes, progenitor cells, inflammatory cells, and endothelial cells, while the microfluidic chip model consists of an inlet and outlets for flow of medium, and culture chambers (Fig. 1a and b). Fig. 1c outlines the fabrication process of the chip. Constructed from PDMS and standing at a height of 100 microns, the chip features one inlet and three outlets. Subsequently, two distinct cell types were introduced into the chip through separate inlet entries with GelMA. After cell seeding, the GelMA was solidified under ultraviolet (UV) light exposure. Following this, the culture medium was consistently supplied to the chip through the inlet.

Fig. 1 Overview of liver microfluidic bio-chip. (a) Tumor microenvironment and microfluidic bio-chip design. The left panel illustrates the complex tumor microenvironment of liver cancer, encompassing cancer stem cells, cancer cells, fibroblasts, macrophages, immune cells, endothelial cells, and tumor endothelial cells. The right panel presents the schematic design of the microfluidic bio-chip, showing the inlet and outlets, and how only one inlet could be used for the whole chip as a cell-, drug- and medium-loading inlet. (b) A top view of the three culture chambers representing different cells. (c) A description of the sequential fabrication process for the microfluidic bio-chip along with a fully assembled bio-chip, displaying the separate inlet and outlets for loading distinct cell types. The PDMS chip is bonded on glass using oxygen plasma, as shown in the figure.

Assessing the biocompatibility and competency of the liver bio-chip for studying liver cancer–endothelial interaction

Considering potential concerns surrounding GelMA solidification under UV light, which may lead to cellular damage such as DNA damage, induction of apoptosis, cellular stress, and/or disruption of cellular processes, we conducted a Live/Dead staining assay to evaluate cell viability within the chip.

Fig. 2a shows the cell viability percentages after 24 and 48 hours under distinct conditions: co-culture, monoculture with cancer cells, and monoculture with endothelial cells. The co-culture setting displays 98% and 81% viability rates at 24 and 48 hours, respectively. Similarly, the mono-cancer model presents 95% and 81% viability percentages, while the mono-endothelial model demonstrates rates of 85% and 80% at the corresponding time points.

Fig. 2 (a) Live/Dead staining assay results showing cell viability, with fluorescence intensities converted to viability percentages. Comparative analyses of cell viabilities across different chambers and culture durations are shown in (b).

These findings indicate that the chip provides a favorable environment for cell viability and growth – a crucial factor for assessing its biocompatibility during culture periods. Consistent viability percentages across the various scenarios emphasize the stability of the chip in supporting cellular functions and interactions. Fig. 2b shows the corresponding graph.

Assessment of CD105, CD31 and albumin expression in co-culture of liver cancer cells and endothelial cells

To determine the mechanism of interaction between liver cancer cells and endothelial cells, we examined the activation of endothelial cells by assessing endothelial activation after co-culture using the biomarker CD105. As shown in Fig. 3, distinct fluorescence intensities were observed under different culture conditions. The co-culture of endothelial and liver cancer cells exhibited the highest fluorescence intensity at 27.2 arbitrary units (a.u.). In contrast, the endothelial monoculture showed a lower intensity of 12.1 a.u., and the cancer monoculture displayed the lowest intensity at 2.6 a.u.

Fig. 3 (a), (c) and (e) CD105, albumin and CD31 expression levels after 24 hours under endothelial monoculture, co-culture, and cancer monoculture conditions. (b), (d) and (f) Comparative analysis of fluorescence intensities among chambers for CD105, albumin and CD31, respectively, presented in the form of graphs. P values were determined using a two-tailed paired Student's t-test. ***, P < 0.001. The scale bar is 100 μm.

We also performed counter-labeling of both cell types using albumin and CD31 markers. Liver cancer cells labeled with albumin showed the highest fluorescence intensity at 12.5 a.u. In comparison, the co-culture exhibited a lower intensity of 8.0 a.u., and the endothelial monoculture displayed the lowest intensity at 2.5 a.u. Similarly, endothelial cells labeled with CD31 showed the highest fluorescence intensity at 7.9 a.u., whereas the co-culture exhibited a lower intensity of 4.1 a.u., and the cancer monoculture displayed the lowest intensity at 1.9 a.u.

These findings strongly indicate that tumor endothelial cell expression is predominantly induced in endothelial cells co-cultured with liver cancer cells, highlighting the interaction between liver cancer cells and tumor endothelial cells (TECs). These outcomes emphasize the significant role of TECs and their interplay with cancer cells in influencing their expression within the microenvironment of the 3D bio-chip.

Development of drug resistance to Sorafenib and Lenvatinib in liver cancer cells after co-culture with endothelial cells

To ascertain the potential utility of our bio-chip in investigating activated drug resistance of tumor endothelial cells to Sorafenib and Lenvatinib, we commenced by evaluating the IC₅₀ of each drug within a 2D context. As depicted in Fig. S1 (ESI†), a large spectrum of drug concentrations spanning from 100 to 0.1 μM was tested. After 24 hours of incubation, the determined IC₅₀ values for Sorafenib and Lenvatinib were 10.4 and 20.7 μM, respectively. A Live/Dead assay was employed to examine cell viability following drug perfusion with IC₅₀ concentrations.

For Sorafenib (Fig. 4a and b), the co-culture initially exhibited enhanced cell viability compared to the monocultures after 24 hours, though this advantage diminished by 48 hours. Despite this, the co-culture still demonstrated better viability than the monocultures, indicating reduced Sorafenib sensitivity initially due to co-culture. Ultimately, all cells perished within 48 hours of drug treatment. The results suggest that endothelial cells augment drug resistance through intricate interactions with cancer cells, likely via enhanced endothelial cell (EC) differentiation and angiogenesis facilitated by direct cell–cell interactions. This physical support from HepG2 cells promotes EC morphogenesis and tubule formation, which are critical for sustaining the tumor microenvironment.

Fig. 4 Investigating Sorafenib and Lenvatinib resistance using the fabricated bio-chip. (a) Representative images of co-culture cell viability after 24 and 48 hours of Sorafenib treatment; fluorescence intensities are converted to assess cell viability. A comparative analysis of cell viability among chambers and culture durations is shown in (b). (c) Representative images of co-culture cell viability after 24 and 48 hours of Lenvatinib treatment; fluorescence intensities are converted to assess cell viability. A comparative analysis of cell viability among chambers and culture durations is shown in (d). The scale bar is 100 μm.

Similarly, co-culture with endothelial cells resulted in elevated cell viability with Lenvatinib (Fig. 4c and d) after 24 hours, surpassing monoculture viability. By 48 hours, no substantial distinction was observed between co-culture and monoculture viabilities. This reinforces the pivotal role of the co-culture microenvironment, composed of liver cancer cells and endothelial cells, in driving drug resistance to Sorafenib and Lenvatinib within the 3D microfluidic bio-chip. The integrin proteins on the surface of endothelial cells are key players in the development of drug resistance, and interfering with these pathways using FAK inhibitors can break down the physical support that helps the cancer and endothelial cells resist treatment. These changes lead to a more robust angiogenic environment, supporting tumor growth and survival, thereby contributing to drug resistance.⁵³ Additionally, antiangiogenic TKI-induced hypoxia increases the expression of proteins involved in drug resistance, further contributing to the ineffectiveness of these treatments in patients with advanced HCC.⁵⁴

Functional test of CD105 in the regulation of endothelial-cell-induced drug resistance to Sorafenib and Lenvatinib in liver cancer cells

The introduction of recombinant CD105 was carried out with the aim to investigate whether the extracellular presence of CD105 can enhance drug resistance. To elucidate the connection between elevated CD105 levels and drug resistance in liver cancer cells toward Sorafenib and Lenvatinib, we conducted a comprehensive 3D experimental study. Recombinant CD105 was directly introduced into the culture medium, leading to a notable increase in cell viability under Sorafenib and Lenvatinib exposure. These on-chip experiments, featuring continuous perfusion of medium, were performed both with and without CD105 proteins.

Following a 24-hour drug treatment period, Live/Dead cell staining revealed a significant enhancement in cell viability with Sorafenib and CD105 as compared to the cell viability with only Sora (Fig. 5a). In co-cultured endothelial–liver cancer cell arrangements, viability increased from 45.0% to 62.7%. In cancer cell monocultures, viability rose from 22.2% to 31.2%, while in endothelial monocultures, it increased from 40.6% to nearly 60.0% following treatment with Sorafenib and CD105 (Fig. 5b). After 48 hours of drug treatment, a further increase in viability was observed with Sorafenib and CD105 (Fig. 5c and d). In co-cultured endothelial–liver cancer cell configurations, viability increased from 41.0% to 60.7%. Cancer cell monocultures showed a modest increase from 19.0% to 21.2%, and endothelial monocultures exhibited a rise from 39.6% to approximately 45%.

Fig. 5 Cell viability tests were conducted on liver cancer cells with and without treatment with recombinant CD105 in the presence of Sorafenib. (a and c) Representative images of co-culture cell viability using Live/Dead staining assay after (a) 24- and (c) 48-hour incubation, comparing cells in each chamber with and without treatment with recombinant CD105, and with and without Sorafenib. A comparative analysis of cell viability among chambers and culture durations is shown in (b and d). The scale bar is 100 μm.

A similar trend was observed with Lenvatinib (Fig. 6a–d). Co-culture viability increased from 45.0% to 63.7% when treated with Lenvatinib and CD105. Monocultures of cancer cells showed an increase from 21.3% to 31.0%, while endothelial monocultures saw a rise from 39.3% to 55.7% upon exposure to Lenvatinib and CD105. After 48 hours of treatment, co-cultured endothelial–liver cancer cell viability increased from 40.0% to 59.7%. In monoculture setups, cancer cell viability increased from 19.0% to 21.2%, and endothelial monocultures exhibited a notable rise from 31.6% to nearly 52% following treatment with Lenvatinib and CD105.

Fig. 6 Cell viability tests were conducted on liver cancer cells with and without recombinant CD105 in the presence of Lenvatinib. (a and c) Representative images of co-culture cell viability using Live/Dead staining assay after (a) 24- and (c) 48-hours incubation, comparing cells in each chamber with and without treatment with recombinant CD105, with and without Lenvatinib. A comparative analysis of cell viability among chambers and culture durations is shown in (b and d). The scale bar is 100 μm.

These findings strongly suggest that recombinant CD105 enhances drug resistance in both co-cultured and mono-cultured liver cancer and endothelial cells when treated with Sorafenib and Lenvatinib, highlighting the significant role of CD105 in the tumor microenvironment.

Conclusions

In conclusion, this study demonstrates a multifaceted exploration into the interplay between liver cancer cells and tumor endothelial cells, designed to study liver tumor angiogenesis. This microfluidic chip model provides a comprehensive platform for investigating interactions between liver cancer cells. This innovative chip design layout with a single inlet and three separate outlets enables the efficient culture of monocultures and co-cultures, facilitating simultaneous drug testing and assessment of potential synergistic effects. The chip's controlled, reproducible environment is ideal for modelling liver tumor angiogenesis and precisely examining drug effects on cell viability, proliferation, and function. Its ability to co-culture multiple cell types provides a valuable platform for studying intercellular communication and drug response interactions. Our findings demonstrate the biocompatibility and competency of the fabricated bio-chip for studying liver cancer–tumor endothelial cell interactions. In hepatocellular carcinoma, angiogenesis emerges as a crucial mechanism driving tumor progression, where aberrant vascular network formation fosters tumor growth, invasion, and metastasis, underscoring its intricate involvement in liver cancer pathogenesis and therapeutic resistance. Through this platform, we delved into the development of drug resistance to Sorafenib and Lenvatinib, key chemotherapeutic agents for liver cancer treatment. Our investigations began with a 3D microfluidic chip, consistently revealing the fostering of drug resistance within the co-culture setting. Utilizing this bio-chip model and recombinant CD105, we attempted to replicate the relative abundance of secreted CD105 and explored its potential influence on drug resistance mechanisms in hepatocellular carcinoma (HCC). Significantly, our findings revealed that the integration of CD105 into a co-culture system comprising HepG2 hepatoma cells and HUVEC endothelial cells promotes the phenotypic transition of HUVEC cells towards a tumor-associated endothelial cell phenotype, which exhibits increased resistance to therapeutic interventions. This model facilitates the investigation of cellular cross-talk between endothelial cells and tumor cells within the tumor microenvironment. We can assess how endothelial cells influence tumor behavior and response to therapies, including potential mechanisms of drug resistance. Collectively, this study leverages innovative microfluidic technology to advance our understanding of liver cancer–TEC interactions and drug resistance mechanisms, opening new avenues for therapeutic intervention.

Author contributions

M. S. P: Writing – original draft, conceptualization, visualization, methodology, investigation, formal analysis, data curation. Y.-D. C.: writing – review, methodology, resources. C.-T. Y.: conceptualization, resources, supervision, funding acquisition. C.-H. L.: writing – review & editing, supervision, resources, project administration, methodology, funding acquisition, conceptualization.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was financially supported by the Chang Memorial Hospital and National Tsing Hua University Joint Research Program Funds (Grant No. CORPG3M0022 to CTY/Grant No. 111F7MDNE1 and 112F7MCNE1 to CHL) and the National Science and Technology Council, Taiwan, under the grants MOST 111-2223-E-007-009 and NSTC 112-2223-E-007-011. Part of the research fund also came from the Interdisciplinary Integration of Advanced Medical Technology (Medical+X) Program (Grant No. 113QF003E1). The fabrication facilities were supported by the Centre for Nanotechnology, Materials Science, and Microsystems (CNMM) of National Tsing Hua University and the National Nano Device Laboratory (NDL).

References

1. F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre and A. Jemal, Ca-Cancer J. Clin., 2018, 68(6), 394–424.
2. L. C. Ong, I. C. Song, Y. Jin, I. H. C. Kee, E. Siew, S. Yu, C. H. Thing, H. Huynh and P. K. H. Chow, Mol. Imaging Biol., 2009, 11(5), 334–342.
3. L. Kulik and H. B. E. Serag, Gastroenterology, 2019, 156(2), 477–491.
4. J. P. Perz, G. L. Armstrong, L. A. Farrington, Y. J. F. Hutin and B. P. Bell, J. Hepatol., 2006, 45(4), 529–538.
5. J. W. Park, M. Chen, M. Colombo, L. R. Roberts, M. Schwartz, P. J. Chen, M. Kudo, P. Johnson, S. Wagner, L. S. Orsini and M. Sherman, Liver Int., 2015, 35(9), 2155–2166.
6. Y. H. Lin, Y. C. Liu, C. Y. Chen, H. C. Chi, M. H. Wu, P. S. Huang, C. C. Chang, T. K. Lin, C. T. Yeh and K. H. Lin, Cells, 2022, 11(16), 2610.
7. J. L. Petrick, A. A. Florio, A. Znaor, D. Ruggieri, M. Laversanne, C. S. Alvarez, J. Ferlay, P. C. Valery, F. Bray and K. A. McGlynn, Int. J. Cancer, 2020, 147(2), 317–330.
8. N. Chalasani, Z. Younossi, J. E. Lavine, M. Charlton, K. Cusi, M. Rinella, S. A. Harrison, A. M. Brunt and A. J. Sanyal, Hepatology, 2018, 67(1), 328–357.
9. European Association for the Study of the Liver (EASL), European Association for the Study of Diabetes (EASD) and European Association for the Study of Obesity (EASO), J. Hepatol., 2016, 64(6), 1388–1402.
10. M. L. Cuestas, J. R. Oubiña and V. L. Mathet, World J. Pharmacol., 2015, 4(1), 96–116.
11. A. Smith, K. Baumgartner and C. Bositis, Am. Fam. Physician, 2019, 100(12), 759–770.
12. R. Bataller and D. Brenner, J. Clin. Invest., 2005, 115(2), 209–218.
13. J. Baglieri, D. A. Brenner and T. Kisseleva, The Role of Fibrosis and Liver-Associated Fibroblasts in the Pathogenesis of Hepatocellular Carcinoma, Int. J. Mol. Sci., 2019, 20, 1723.
14. A. Smith, K. Baumgartner and C. Bositis, Am. Fam. Physician, 2019, 100(12), 1723.
15. F. R. Bruniera, F. M. Ferreira, L. R. M. Saviolli, M. R. Bacci, D. Feder, M. L. G. Pedreira, M. A. S. Peterlini, L. A. Azzalis and V. B. C. Junqueira, Riv. Eur. Sci. Med. Farmacol., 2015, 19(4), 694–700.
16. D. Öhlund, E. Elyada and D. Tuveson, J. Exp. Med., 2014, 211(8), 1503–1523.
17. S. M. Wilhelm, C. Carter, L. Tang, D. Wilkie, A. McNabola and H. Rong, et al. , Cancer Res., 2004, 64(19), 7099–7109.
18. S. Tanaka and S. Arii, Semin. Oncol., 2012, 39(4), 486–492.
19. R. S. Finn, et al. , J. Clin. Oncol., 2020, 38(26), 2960–2970.
20. J. Hasskarl, Europe PMC, 2010, vol. 184, pp. 61–70.
21. S. E. Duff, C. Li, J. M. Garland and S. Kumar, FASEB J., 2003, 17(9), 984–992.
22. N. A. Dallas, S. Samuel and L. Xia, et al. , Clin. Cancer Res., 2008, 14(7), 1931–1937.
23. J. W. Ho, R. T. Poon, C. K. Sun, W. C. Xue and S. T. Fan, World J. Gastroenterol., 2005, 11(2), 176–181.
24. F. Tanaka, Y. Otake and K. Yanagihara, et al. , Clin. Cancer Res., 2001, 7(11), 3410–3415.
25. J. P. Paschoal, V. Bernardo, N. H. Canedo, O. D. Ribeiro, A. Caroli-Bottino and V. L. Pannain, BMC Cancer, 2014, 14, 72.
26. L. Y. Yang, W. Q. Lu, G. W. Huang and W. Wang, BMC Cancer, 2006, 6, 110.
27. Y. Yao, Y. Pan, J. Chen, X. Sun, Y. Qiu and Y. Ding, Ann. Clin. Lab. Sci., 2007, 37, 39–48.
28. Y. Li, Z. Zhai, D. Liu, X. Zhong, X. Meng, Q. Yang, J. Liu and H. Li, Tumor Biol., 2015, 36, 737–745.
29. E. Yagmur, M. Rizk, S. Stanzel, C. Hellerbrand, F. Lammert, C. Trautwein, H. E. Wasmuth and A. M. Gressner, Eur. J. Gastroenterol. Hepatol., 2007, 19, 755–761.
30. D. M. Elnemr, H. A. Abdel-Azeez, H. A. Labib and F. M. Abo-Taleb, Clin. Lab., 2012, 58, 1023–1028.
31. S. Teama, A. Fawzy, A. Helal, A. D. Drwish, T. Elbaz and E. Desouky, Asian Pac. J. Cancer Prev., 2016, 17, 2429–2434.
32. D. Yu, L. Zhuang, X. Sun, J. Chen, Y. Yao, K. Meng and Y. Ding, BMC Cancer, 2007, 7, 122.
33. Y. Wang, X. H. Zhang, P. Guo, L. N. Yan and D. He, J. Sichuan Univ., 2010, 41, 818–821.
34. L. S. Rosen, H. I. Hurwitz and M. K. Wong, et al. , Clin. Cancer Res., 2012, 18(17), 4820–4829.
35. R. Castonguay, E. D. Werner, R. G. Matthews, E. Presman, A. W. Mulivor, N. Solban, D. Sako, R. S. Pearsall, K. W. Underwood and J. Seehra, et al. , J. Biol. Chem., 2011, 286, 30034–30046.
36. L. J. Hawinkels, P. Kuiper, E. Wiercinska, H. W. Verspaget, Z. Liu, E. Pardali and C. F. Sier, Cancer Res., 2010, 70(10), 4141–4150.
37. C. Li, B. Guo and P. B. Wilson, et al. , Int. J. Cancer, 2000, 89, 122–126.
38. C. Li, R. Gardy and B. K. Seon, et al. , Br. J. Cancer, 2003, 88, 1424–1431.
39. N. Takahashi, R. Kawanishi-Tabata and A. Haba, et al. , Clin. Cancer Res., 2001, 7, 524–532.
40. A. Kasprzak and A. Adamek, Int. J. Mol. Sci., 2018, 19(12), 3887.
41. J. M. Llovet and J. Bruix, Hepatology, 2008, 48(4), 1312–1327.
42. F. Yang, T. E. Van Meter and R. Buettner, et al. , Mol. Cancer Ther., 2008, 7, 3519–3526.
43. Y.-Q. Xiong, H.-C. Sun, W. Zhang, X.-D. Zhu, P.-Y. Zhuang and J.-B. Zhang, et al. , Clin. Cancer Res., 2009, 15(15), 4838–4846.
44. A. G. Duffy and C. Ma, et al. , Clin. Cancer Res., 2017, 23(16), 4633–4641.
45. J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. Schueller and G. M. Whitesides, Electrophoresis, 2000, 21(1), 27–40.
46. Y.-C. Chen, K.-Y. Lee, H.-J. Liao, W.-L. Sun, W.-C. Huang, Y.-C. Wang, W.-C. Chang and C.-H. Liu, Sens. Actuators, B, 2024, 406, 135409.
47. T. Liu, W. Weng, Y. Zhang, X. Sun and H. Yang, Molecules, 2020, 25(22), 5305.
48. A. Saroufim, Y. Messai, M. Hasmim, N. Rioux, R. Iacovelli, G. Verhoest, K. Bensalah, J. J. Patard, L. Albiges, B. Azzarone, B. Escudier and S. Chouaib, et al. , Br. J. Cancer, 2014, 110(7), 1778–1784.
49. N. I. Nikiteas, N. Tzanakis, G. Theodoropoulos, V. Atsaves, Z. Christoni, P. Karakitsos, A. C. Lazaris, A. Papachristodoulou, C. Klonaris and M. Gazouli, et al. , Gastric Cancer, 2007, 10(1), 12–17.
50. A. Nogues, E. Gallardo-Vara, M. P. Zafra, P. Mate, J. L. Marijuan and A. Alonso, et al. , World J. Surg. Oncol., 2020, 18(1), 99.
51. D. Yu, L. Zhuang, X. Sun, J. Chen, Y. Yao, K. Meng and Y. Ding, BMC Cancer, 2007, 7, 122.
52. Y. Wang, X. H. Zhang, P. Guo, L. N. Yan and D. He, J. Sichuan Univ., 2010, 41, 818–821.
53. G. G. Y. Chiew, A. Fu, K. P. Low and K. Q. Luo, Sci. Rep., 2015, 5, 10801.
54. H. Suzuki, et al. , Cancer Commun., 2023, 43(4), 415–434.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4lc00238e

---