Microfluidic organoid-slice-on-a-chip system for studying anti-cholangiocarcinoma drug efficacy and hepatorenal toxicity

Abstract

Organ-chip technology, in contrast to cell culture and animal models, offers a promising platform for accelerating drug development. However, current chip designs simulate human organ functions and there is a lack of multi-organ chip designs that can simultaneously study drug efficacy and hepatorenal toxicity. Here, we developed a novel microfluidic multi-organ chip that integrated cholangiocarcinoma organoids (CCOs) with recellularized liver slices (RLS) and recellularized kidney slices (RKS), to simultaneously assess anti-cholangiocarcinoma drug efficacy and hepatorenal toxicity. Co-culture of patient-derived CCOs with RLS and RKS was successfully achieved for 7 days under flow conditions with enhanced liver and renal cell functions. Furthermore, an in vitro biomimetic model showed IC₅₀ values of trastuzumab emtansine (T-DM1) of around 6.42 ± 7.34 μg mL⁻¹ in four clinical cases, with one outlier of 77.77 μg mL⁻¹ due to patient variability. Post-treatment, RLS and RKS cell viability remained high at 75.67% and 81.03%, respectively, suggesting low hepatorenal toxicity of T-DM1 for treating cholangiocarcinoma. Our study demonstrates the use of an organoid-slice-on-a-chip (OSOC) platform for personalized drug efficacy and toxicity assessment, particularly aiming at leveraging anticancer drugs for off-label use to save patient lives.

---

1. Introduction

New drug development is under increasing pressure, as it typically takes more than a decade and billions of dollars to develop a new drug, and 90% of candidate drugs fail in clinical trials.^1–3 Before entering clinical trials, the new drugs should be accompanied by comprehensive studies, verifying the therapy's efficacy for the targeted disease and ensuring the absence of unforeseen adverse effects in patients due to drug interactions, especially in the field of cancer therapy.^4,5 Therefore, effective drug testing models are essential to assess the therapeutic potential and toxicity of anticancer drugs while reducing the costs and time associated with drug development. In view of the above, numerous methods have been developed, with in vitro cell models (2D monolayer cultures or 3D cell cultures) and in vivo animal models being the most prevalent.¹ However, traditional cell culture models and animal models still have shown limitations in predicting drug therapeutic efficacy and toxicity.⁶ Cellular models compensate for the shortcomings of interspecies differences in animal models, but they lack physiological relevance and fail to encompass complex intercellular interactions and systemic physiological processes, which may lead to inaccurate results.⁷ Animal models are subject to interspecies physiological and metabolic variations, which may limit their ability to comprehensively predict the potential toxicity and efficacy of drugs in humans.^8,9 Hence, organ-on-a-chip technologies have emerged in response to the limitations of traditional models.¹⁰

Currently, organ chips are increasingly used as a valuable tool for effectively and accurately assessing drug efficacy and toxicity to facilitate drug development.^11–13 For example, numerous single-organ chips such as liver chips,¹⁴ kidney chips,¹⁵ heart chips,¹⁶ and tumor chips,¹⁷ are used to study disease progression,^18,19 to screen for drug efficacy,²⁰ and to analyze adverse drug reactions,²¹ providing valuable references for clinical trials. While single-organ chips focus on simulating the function of a single organ, multi-organ chips can better simulate whole-body physiology to study complex inter-organ interactions and pharmacokinetic characteristics by fluidically coupling two or more organ chips.^12,22 For instance, physiological pharmacokinetic modelling coupling gut, liver, and kidney chips was used to assess drug absorption, metabolism, and elimination, predicting pharmacokinetic responses in humans.²³ Similarly, a multi-organ chip coupling heart, liver, bone and skin chips by mimicking circulating blood vessels replicated organ interactions and was employed to investigate multi-organ toxicity.²⁴ Furthermore, recent efforts have integrated organoid technology with organ chips to enhance physiological relevance.²⁵ By combining the biomimetic nature of organoids with the dynamic control of organ chips, this system enables precise modeling of drug responses, bridging the gap between in vitro models and human physiology.^26,27 Despite the technology advances in multi-organ chips, new drug development is still a time-consuming and costly process. Thus, expanding indications of existing drugs are gaining prominence to enhance options for cancer treatment while effectively easing the pressure on new drug development.^28–30 To this end, a biomimetic multi-organ chip is highly needed to simultaneously evaluate drug efficacy and toxicity such as hepatotoxicity and nephrotoxicity.^31–33

Here, we designed a novel microfluidic organoid-slice-on-a-chip (OSOC) combining tumor organoids and recellularized tissue slices, aiming to assess trastuzumab emtansine (T-DM1) repurposing in cholangiocarcinoma (CCA) by testing the efficacy and hepatic renal toxicity of T-DM1. In brief, we constructed cholangiocarcinoma organoids (CCOs) as a tumor model and sliced the recellularized decellularized scaffolds to construct human liver and kidney models with enhanced liver cell functions in recellularized liver slices (RLS) and renal cell functions in recellularized kidney slices (RKS) under flow conditions. The biomimetic OSOC model showed that the average IC₅₀ value for T-DM1 was approximately 6.42 ± 7.34 μg mL⁻¹ in four clinical cases. Following treatment with T-DM1, the viability of RLS and RKS cells remained high at 75.67% and 81.03%, respectively, indicating low hepatorenal toxicity during cholangiocarcinoma treatment. Therefore, our findings suggest that the OSOC can be used to simultaneously assess drug efficacy and toxicity in vitro for rapidly expanding indications of existing anticancer drugs to save the lives of patients.

2. Materials and methods

2.1. Human samples

Primary cholangiocarcinoma (pCCA) samples were obtained after surgery at West China Hospital of Sichuan University. Sample collection required informed consent from the patients or their families and approval of the Ethics Committee on Biomedical Research West China Hospital of Sichuan University (2023 Review No. 108).

2.2. Fabrication of the microfluidic device

The device was mainly made of polydimethylsiloxane (PDMS) and methyl methacrylate (PMMA). The design of the two PMMA layers and one PDMS layer was created using the CorelDraw® software. The top PMMA layer contained one inlet, one outlet and 3 microculture wells. To prepare the middle PDMS film, a 10 : 1 mixture of PDMS precursor and coagulant was prepared. Residual air bubbles from the mixture were removed by degassing for 30 min prior to curing at 60 °C for 4 h. Once the curing process was completed, the PDMS film was gently extracted from the mold. These two layers of PMMA and the middle PDMS layer were assembled into a microfluidic device using screws. Tubing was connected to the inlet and the outlet to allow for medium perfusion and a peristaltic pump connected to the tubing facilitated medium circulation and controlled the flow rate.

2.3. Preparation and characterization of decellularized liver matrix (DLM) and decellularized kidney matrix (DKM)

Rat livers and kidneys were prepared and characterized as previously described.^17,18 The use of Sprague Dawley (SD) rats and the harvesting of their livers and kidneys were in accordance with the Guidelines for Care and Use of Laboratory Animals of Sichuan University and approved by West China Hospital of Sichuan University's Animal Care Ethics Committee (No. 20220714003). Livers and kidneys were obtained from SD rats weighing approximately 300 g at 8 weeks of age. Prior to decellularization, livers and kidneys were frozen at −80 °C for 24 h. After the livers and kidneys were thawed at room temperature, the needle of a 1 mL syringe was inserted into the hepatic portal vein or renal vein. Then the livers and kidneys were perfused sequentially through a peristaltic pump at a flow rate of 15 mL min⁻¹ with 0.1% SDS for 6 h, 1% Triton X-100 for 2 h, 80 U mL⁻¹ DNase and 5 U mL⁻¹ RNase for 30 min, and PBS containing 2% penicillin/streptomycin and amphotericin B for 2 h. Prior to scanning electron microscopy (SEM) characterization, the native livers and native kidneys as well as DLM and DKM were dehydrated using ethanol at concentrations of 30%, 50%, 70%, 80%, 90%, 95%, and finally 100%, followed by drying using a critical point dryer (Leica, Nussloch, Germany). The native livers, native kidneys, DLM, and DKM were observed and analyzed using SEM (Hitachi, Tokyo, Japan). To detect residual DNA in the decellularized matrix, the total DNA from six native liver, native kidney, DLM, and DKM samples were quantified, respectively. DNA was extracted from these samples using a QIAamp® DNA Mini kit (QIAGEN, Hilden, Germany), and then quantified using a NanoDrop spectrophotometer (DeNovix, Wilmington, DE, USA). The DNA content relative to the weight of the lyophilized samples, including the native livers, native kidneys, DLM, and DKM, was calculated.

Hematoxylin and eosin (H&E) and immunofluorescence (IF) staining were performed on the native livers, native kidneys, DLM, and DKM, respectively. The samples were sequentially paraffin embedded, sliced, and dehydrated. Then, the sample slices were treated with rabbit polyclonal antibody targeting collagen I, collagen IV, fibronectin, and laminin (Abcam, Cambridge, UK), and incubated at 4 °C overnight. Next, the sliced samples were incubated using a secondary goat anti-rabbit antibody Alexa Fluor® 594 (Abcam, Cambridge, UK) for 50 min at room temperature in the dark. Finally, the nuclei were stained using 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) (G1012, Servicebio, China) for 10 min at room temperature in the dark. The fluorescence-stained samples were imaged using a fluorescence microscope (Zeiss, Oberkochen, Germany).

2.4. Preparation of recellularized slices

Human liver cell line (HepG2) and human kidney-2 (HK-2) cells were provided by Wuhan Pricella Biotechnology Co., Ltd. A 5% (w/v) GelMA solution was prepared by adding 1 g of GelMA (Suzhou Institute of Intelligent Manufacturing, Suzhou, China) to 20 mL of 0.25% (w/v) LAP and mixing thoroughly. Then, HepG2 and HK-2 cells were resuspended uniformly in 1 mL of 5% (w/v) GelMA solution, and their cell density concentration was maintained at 1.5 × 10⁷ cells per mL. The HepG2 cell mixture was subsequently injected into the DLM and the HK-2 cell mixture was injected into the DKM using a 1 mL syringe. Immediately after the injection was completed, the GelMA was exposed to ultraviolet (UV) light at 405 nm for 20 s for gel solidification. Finally, the recellularized liver and kidney were then cut into slices with a thickness of 250 μm using a vibratome (VT1200s, Leica, Nussloch, Germany).

2.5. Isolation and culture of CCOs

Clinical cancer tissue samples were washed several times with cold DPBS (Gibco, CA, USA). The samples were then cut into pieces of 0.5–1.0 cm³ with scissors and digested at 37 °C with tissue dissociation medium I (Aimingtmed, Hangzhou, China). Once cell clusters with a diameter of around 200 μm were observed under a microscope, enzymatic digestion was terminated. The samples were subjected to centrifugation at 300g for a duration of 5 min, and the supernatant was then carefully discarded. The obtained cell clusters were further digested using tissue dissociation medium II (Aimingtmed, Hangzhou, China). The digestion was terminated once the size of the cell clusters reached approximately 50 μm. After digestion, the mixture was filtered through a 100 μm cell strainer and the resultant cell clusters were then washed using culture medium. Subsequently, the cell pellet was treated with erythrocyte lysis buffer (Servicebio, China) for 5 min to remove erythrocytes. After washing with culture medium, the obtained cells and cell clusters were resuspended in Matrigel (Corning, NY, USA), and 25 μL of the cell mixture was dispensed in each well in a 48-well plate. Once the Matrigel solidified, 300 μL bile duct cancer organoid medium (Aimingtmed, Hangzhou, China) was added. When the size of the CCOs reached roughly 300 μm, the CCOs were passaged after a 5-min incubation in TrypLE (Gibco, CA, USA).

2.6. Characterization of organoids and their source tissues

Organoids and tumor tissues were stained with H&E and immunohistochemistry (IHC). For IHC staining, organoids and tumor tissues were paraffin-embedded and sliced, and all slices were deparaffinized and hydrated. Sample slices were antigenically repaired with sodium ethylenediaminetetraacetic acid (EDTA) and then were immersed in 3% hydrogen peroxide solution for 25 min at room temperature in the dark to inhibit endogenous peroxidase activity. To reduce background, sample slices were covered with 3% bovine serum albumin for 30 min at room temperature. Sample slices were then stained with the first antibodies, including anti-CK7 antibody, anti-CK19 antibody, anti-EpCAM antibody, and anti-Ki67 antibody (Abcam, Cambridge, UK), and incubated at 4 °C overnight. Finally, after incubation with the HRP-conjugated 2nd antibody (Servicebio, China) for 50 min at room temperature, sample slices were color developed by dropwise addition of 3,3′-diaminobenzidine (DAB). Stained slices were observed under a fluorescence microscope (Zeiss, Oberkochen, Germany).

2.7. Albumin (ALB) and glucose uptake assay

The procedure was carried out following previously established guidelines.³⁴ Briefly, RKS were first cultured for 2 days in the microfluidic chip with a flow rate of 10 μL min⁻¹. After 24 h of serum starvation, RKS were exposed to medium containing 50 μg mL⁻¹ FITC-ALB (A9771, Sigma, St. Louis, MO, USA) for 3 h and kept perfused. For static controls, experiments were performed in 24-well cell culture dishes. For glucose uptake experiment, the glucose uptake was assayed by 2-NBDG (Cayman Chemical, MI, USA). According to the aforementioned method, RKS were cultured for 2 days first, starved in serum for 24 h, and then exposed to culture medium containing 200 μg mL⁻¹ 2-NBDG for 2 h. Finally, after the exposure period, cells were washed with cold PBS and fluorescence was visualized using a fluorescence microscope.

2.8. Measurement of urea and ALB

Quantification of urea and ALB secretion over 7 days was performed to assess liver cell function. The RLS were divided into two groups, one cultured in the microfluidic chip and the other cultured in static incubation. The supernatants were collected on days 1, 4, and 7 to quantify the levels of urea and ALB using an ELISA kit (Meiman, Jiangsu, China). Measurements were taken using a multimode microplate reader (Synergy H1, Thermo Fisher, USA).

2.9. Long-term co-culture in the OSOC

The RLS, RKS, and CCOs were added into the chip and incubated by perfusion at a flow rate of 10 μL min⁻¹. The cells were characterized as alive or dead using a live/dead assay kit (DOJINDO, Japan) staining reagents on days 1, 4, and 7 of culture. Calcein acetoxymethyl ester (calcein AM) (10 μL) and propidium iodide (PI) (15 μL) were dissolved in 5 mL PBS and mixed. 1 mL of dye solution was spread on the culture wells and incubated at 37 °C in the dark for 30 min. Following color development, the samples were then imaged using a fluorescence microscope.

2.10. Drug efficacy and toxicity testing based on the OSOC

The RLS, RKS and CCOs were added into the microfluidic chip and cultured with a flow rate of 10 μL min⁻¹. The next day, the culture medium was replaced with culture medium containing different drug concentrations and incubated for another five days. At the end of each experiment, organoids and recellularized slices were stained with a live/dead assay kit (DOJINDO, Japan) and imaged using a fluorescence microscope (Zeiss, Oberkochen, Germany). The viability of CCOs was assessed using a CellTiter-Glo 3D cell viability assay (Promega, USA), and the viability of RLS and RKS was assessed using a cell counting kit-8 (MedChemExpress, NJ, USA). LDH and KIM-1 were measured using an LDH cytotoxicity assay kit (Beyotime, Shanghai, China) and an ELISA kit (Meimian, Jiangsu, China), respectively. In addition, AUC is the area under the dose–response curve of a drug, and AUC index is the AUC value normalized to compare the sensitivity of different CCOs to the drug. The formula for calculating the AUC index is as follows:

2.11. Data analysis

The results were analyzed using GraphPad Prism Version 9.0.0. All data were presented as mean ± SD. Multiple group comparisons were conducted using one-way ANOVA with p < 0.05 to indicate statistical significance.

3. Results

3.1. Establishment of the microfluidic chip

To better understand the microfluidic OSOC platform for drug testing, we have diagrammed the compositional structure of recellularized slices and the overall design of the chip platform (Fig. 1A and B). The chip was mainly divided into three layers (Fig. 1C and D); the top and bottom layer were made of PMMA and the middle layer was made of PDMS. After the three-layer structure was assembled, the microfluidic device was secured with screws and flowed with culture medium with the flow rate controlled using a peristaltic pump. Then, CCOs, RLS, and RKS were added into the culture wells; soft plugs were used to seal the holes to prevent leakage and contamination. During the culture period, the flow rate was maintained at 10 μL min⁻¹.

Fig. 1 Schematic of the OSOC and its application in assessing drug efficacy and hepatorenal toxicity. (A) Diagram of the integration of (I) recellularized liver by injecting HepG2 cells mixed with GelMA into the DLM and (II) recellularized kidney by injecting HK-2 cells mixed with GelMA into the DKM. (B) Schematic of recellularized liver slices, recellularized kidney slices and organoids in a microfluidic device. The recellularized slices and organoids were added into the culture wells of the chip, and then all three wells were sealed with plugs; samples were subsequently used for anticancer drug testing. (C) 3D diagram of the microfluidic chip assemblies (top, intermediate and bottom layers, top and intermediate microchannels, plugs, screws, inlets, and outlets) and (D) physical image.

3.2. Characterization of DLM and DKM

The decellularized matrices were obtained according to a previous protocol.^18,35 After a 9 h of decellularization, the rat liver and kidney became translucent and the distribution of blood vessels could be clearly observed (Fig. 2A). H&E staining showed that the decellularized matrices retained pink eosinophilic staining without basophilic staining, indicating that four collagens remained intact while cells were absent (Fig. 2B). SEM imaging showed an acellular, mesh-like structure of DLM and DKM, whereas cells were observed in native liver and kidney (Fig. 2C). The results of DNA quantification showed that the remaining DNA within decellularized rat liver and kidney was 24.82 ± 0.90 ng mg⁻¹ and 31.30 ± 1.08 ng mg⁻¹, respectively. Both DNA concentrations were less than 50 ng mg⁻¹, which was the standard DNA concentration for decellularization.³⁶ In contrast, the total DNA within native liver and kidney prior to decellularization was 5546 ± 189.50 ng mg⁻¹ and 6759 ± 68.74 ng mg⁻¹, respectively (Fig. 2D). Furthermore, the results of immunofluorescence showed that four key extracellular matrix (ECM) proteins (i.e., collagen I, collagen IV, fibronectin, and laminin) were preserved in the decellularized liver and kidney matrices (Fig. 2E). Moreover, there were no nuclei stained blue by DAPI in the decellularized matrix. These results demonstrated that decellularization of the rat liver and kidney was successfully completed and four major ECM proteins were preserved.

Fig. 2 Characterization of rat native and decellularized liver and kidney. (A) Gross images of rat liver and DLM and rat kidney and DKM. (B) H&E staining of native and decellularized liver and kidney. (C) SEM staining of native and decellularized liver and kidney. (D) DNA quantification of native and decellularized liver and kidney (n = 6). (E) IF staining of collagen I (red), collagen IV (red), fibronectin (red), and laminin (red) in native liver, DLM, native kidney and DKM. The presence of cells in rat native liver and kidney is indicated by DAPI staining (blue). Scale bars: 100 μm (B and E), 10 μm (C).

3.3. Long-term slice culture on the OSOC

To construct recellularized slices, a mixture of cells and 5% (w/v) GelMA was injected into the decellularized matrix with a final cell density of 1.5 × 10⁷ cells per mL, and then the recellularized matrix was sliced to a thickness of 250 μm using a vibratome (Fig. 3A). Live/dead staining and data analysis on days 1, 4, and 7 showed that the cell viability of recellularized slices remained stable (more than 90%) when perfused at a rate of 10 μL min⁻¹ in the chip (Fig. 3B and C). Basophilic staining of nuclei was visible in the recellularized slices, with the nuclei adhering to the decellularized matrix (Fig. 3D). In addition, the levels of urea (Fig. 3E-I) and ALB (Fig. 3E-II) secreted by RLS in the perfusion culture were consistently higher than those in static culture throughout the entire 7-day culture period. Moreover, the uptake of both ALB and 2-NBDG (a fluorescently labeled glucose analog) by RKS was increased more than 2-fold under flow conditions compared to static conditions (Fig. 3F and G). Taken together, these results clearly indicated that recellularization was successfully achieved in the decellularized matrix, and the recellularized slices maintained good viability in the microfluidic chip for 7 days. Compared to static culture, RLS in dynamic culture exhibited enhanced urea and ALB secretion functions, and RKS showed improved ALB and glucose uptake capabilities.

Fig. 3 Construction of recellularized liver and kidney slices and the evaluation of liver and kidney functions on chip. (A) The preparation process of recellularized liver slices. After mixing cells with GelMA, they are injected into the decellularized scaffold and then cut into slices using a vibratome. (B) Live/dead staining images of RLS and RKS on days 1, 4, and 7 of culture. (C) Cell viability of RLS and RKS on days 1, 4, and 7 (n = 3). (D) Gross images and H&E staining of RLS and RKS. (E) Measurements of urea (I) and ALB (II) of RLS under static versus fluidic cultures (n = 3). (F) ALB uptake by RKS after 3 h of incubation with 50 μg mL⁻¹ FITC-ALB (green) added and 2-NBDG uptake by RKS after 2 h of incubation with 200 μg mL⁻¹ 2-NBDG (green) added under static versus fluidic culture conditions. (G) Relative FITC-ALB (I) and 2-NBDG (II) intensity in RKS under static versus fluidic culture conditions (n = 4). (mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001). Scale bars: 100 μm (B and F), 50 μm (C).

3.4. Characterization of CCOs

To construct the tumor model, we established CCOs from patient CCA samples. After being sheared and digested into single cells and cell clusters, CCA samples were mixed with Matrigel and cultured in medium for several days, leading to the outgrowth of CCOs (Fig. 4A). CCOs appeared cystic under bright-field microscopy, with their size gradually increasing as culture time extended after passaging (Fig. 4B). To evaluate the histological characterization of CCOs, H&E staining and immunohistochemistry (IHC) (Fig. 4C) were performed. CCOs showed irregular, cyst-like formations and glandular configurations, echoing the H&E staining outcomes observed in the original CCA tissues. The expression of biomarker proteins, including CK7, CK19, EpCAM, and Ki67, was consistent with that observed in CCOs and original CCA tissues. Live/dead staining and data analysis on days 1, 4, and 7 showed that CCOs maintained good viability (more than 92%) and proliferation when perfused at 10 μL min⁻¹ in the chip (Fig. 4D and E). Thus, we successfully established CCOs from CCA patient samples and patient-derived CCOs maintained the histological characteristics of the original CCA tissues.

Fig. 4 Patient-derived organoids encompass the characteristics of the primary tumor. (A) The preparation process of tumoral organoids. Fresh CCA samples were obtained from patients and processed as outlined in the Materials and methods section. (B) Bright-field staining of CCOs on days 1, 3, 5 and 7 of culture. (C) IHC and H&E staining of the CCA biomarkers (CK7, CK19, EpCAM, and Ki67) on CCOs and original CCA tissues. (D) Live/dead staining of CCOs on days 1, 4, and 7 of culture. (E) Cell viability of CCOs on days 1, 4, and 7 (n = 3). Scale bars: 100 μm (B and D), 50 μm (C).

3.5. Efficacy of T-DM1 on the OSOC

CCOs from 5 patients were used to assess the effectiveness of T-DM1 in the treatment of CCA on the OSOC platform. After treatment with different concentrations of T-DM1 (0, 0.1, 1, 5, 10, 50, and 100 μg mL⁻¹), the live/dead staining and bright-field images of CCOs indicated that as drug concentration increased, both the size and number of CCOs exhibited a decreasing trend, and the inhibition of growth and proliferation increased gradually (Fig. 5A). The drug sensitivity profiles of CCOs showed significant variability among patients in their reactions to individual chemotherapeutic agents, as assessed through dose–response curves and the associated AUC index (Fig. 5B). Furthermore, the IC₅₀ values of the CCOs from 5 patients were 5.93 μg mL⁻¹, 16.93 μg mL⁻¹, 1.70 μg mL⁻¹, 77.77 μg mL⁻¹, and 1.08 μg mL⁻¹, respectively (Fig. 5C). As a positive control for T-DM1, the viability of CCOs treated with 20 μM gemcitabine fell below 10% (Fig. 5D), which was consistent with the findings reported in the literature.³⁷

Fig. 5 Validation of the anti-cholangiocarcinoma biologic activity of the antibody–drug conjugate (ADC) T-DM1 on the chip. (A) The live/dead and bright-field images of CCOs after receiving different treatments as indicated. (B) The in vitro chemosensitivity reactions of CCOs from 5 patients to T-DM1 are presented, with the findings illustrated as dose–response curves (n = 3). AUC index derived from the original dose–response was normalized and represented in a violin plot format. The sensitivity of the drug was represented by the AUC index. (C) The IC₅₀ values changed in different concentrations of T-DM1. (D) The cell viability of CCOs from 5 patients under the action of the positive drug gemcitabine (20 μM) (n = 3). Scale bar: 200 μm (A).

3.6. Hepatorenal toxicity of T-DM1 on the OSOC

To investigate the hepatorenal toxicity of T-DM1 during treatment for CCA, RLS and RKS were used as human liver and kidney models in the drug testing. Hepatorenal toxicity responses were tested for 7 different concentrations of T-DM1. Live/dead staining showed that cell activity of both RLS and RKS decreased with increasing drug concentration during CCA treatment (Fig. 6A). However, even at 100 μg mL⁻¹, RLS and RKS maintained high viability at 75.67 ± 3.64% and 81.03 ± 5.69%, respectively (Fig. 6B and C). Additionally, the release of both LDH and KIM-1 (a marker of proximal tubular cell damage) increased with the rising drug concentrations, but the extent of this rise remained modest, indicating limited damage to RLS and RKS within drug concentrations below 100 μg mL⁻¹ (Fig. 6D and E). The results showed that T-DM1 exhibited low hepatorenal toxicity at concentrations below 100 μg mL⁻¹ while demonstrating good therapeutic efficacy on CCA within this range.

Fig. 6 Characterization of T-DM1's toxicological response to the liver and kidney on the chip. (A) The live/dead imaging of RLS and RKS after receiving different treatments as indicated. (B and C) Analysis of the cell viability of RLS (B) and RKS (C) on day 1, 4, and 7 (n = 3). (D and E) Measurement of LDH (D) and KIM-1 (E) secreted from the OSOC at different drug concentrations (n = 3). Mean ± SD, NS indicates p > 0.05, * indicates p < 0.05.

4. Discussion

In this study, we constructed a novel microfluidic multi-organ chip by co-culturing patient-derived CCOs and RLS and RKS to simultaneously study the efficacy and hepatorenal toxicity of a repurposed anticancer drug (i.e., T-DM1) for CCA treatment (Fig. 1). We cultured HepG2 cells and HK-2 cells mixed with GelMA in 3D biomimetic DLM and DKM to construct the human liver and kidney models. The decellularized matrices, rich in collagen I, collagen IV, fibronectin, and laminin (Fig. 2) along with GelMA serve as a 3D scaffold with a biologically active component, contributing to enhanced cell adhesion, proliferation, and recellularization on these scaffolds.^38–40 Moreover, to promote more efficient diffusion of nutrients and oxygen between cells in the DLM and DKM, the recellularized liver and kidney were cut into slices (4 to 5 mm in diameter and 250 μm thick), and the thickness has been successfully used in precision-cut slices for drug toxicity testing.^41,42 In addition, compared to traditional CCA cell models, this patient-derived CCO can better mimic tumor characteristics and tumor microenvironment in individual patients for drug efficacy testing.⁴³ Recellularized slices and CCOs were cultured in the chip with a flow of fluid (10 μL min⁻¹), which provided the fluid shear stress required by the cells and promoted nutrient supply and removal of metabolic wastes, making our OSOC more bionic for repurposed drug testing.

CCOs, RLS, and RKS retained high long-term cell viability and mimicked biological functions in the microfluidic chip, indicating that they can be used as reliable and applicable experimental models. Compared to static culture, the RLS in dynamic culture for 7 days consistently secreted higher levels of ALB and urea (Fig. 3E), and the RKS showed a twofold increase in both ALB and glucose uptake (Fig. 3F and G). Our microfluidic device improved the metabolism of liver cells and uptake function of renal cells, which is also observed in some human liver cell models^17,19 and some 2D renal tubular epithelial cell models.^44,45 Unlike animal models and animal-derived cell models, our recellularized slices were constructed using human-derived cells, which more closely mimic the human physiological state, thereby offering more accurate insights into drug responses. In addition, the patient-derived CCOs and original CCA tissues co-expressed the biomarkers CK7, CK19, EpCAM, and Ki67 (Fig. 4C). The result showed that CCOs retained the same expression pattern of the original tumor, which is consistent with previous studies.^37,46 On the other hand, the recellularized slices and CCOs remained highly viable (more than 90%) in the microfluidic chip for at least 7 days, which provides a more accurate and reliable experimental platform for long-term drug testing. Therefore, these results suggested that our CCO, RLS and RKS can mimic CCA, human liver and kidney for testing the efficacy and hepatorenal toxicity of anticancer drugs.

The OSOC can predict the therapeutic efficacy and hepatorenal toxicity of anticancer drugs, particularly for drug repurposing. The live/dead staining confocal images showed that increasing concentrations of T-DM1 increasingly restricted the growth and proliferation of CCOs (Fig. 5A). The average IC₅₀ value for T-DM1 was approximately 6.42 ± 7.34 μg mL⁻¹ in four clinical cases, whereas one was an outlier of 77.77 μg mL⁻¹ (Fig. 5D). This variation might be due to T-DM1 targeting HER2, while some CCOs do not express HER2.⁴⁷ This observation aligns with the finding that IBI315, another targeted therapy, exhibited potent anti-tumor efficacy in HER2-positive gastric cancer organoids.⁴⁸ Notably, even at a concentration of 100 μg mL⁻¹ T-DM1, the viability of RLS and RKS remained high (75.67% ± 3.64% and 81.03% ± 5.69%), with a modest increase in LDH and KIM-1 release as drug concentration rose (Fig. 6). The results indicated that T-DM1 had low hepatorenal toxicity, which is consistent with the precision therapy and low toxicity characteristics of ADCs. ADCs, a novel type of targeted anticancer therapies, utilize antibody specificity to deliver toxic drugs directly to tumor cells, minimizing harm to healthy cells, compared to conventional chemotherapeutic drugs.^49–51 However, there are currently no ADCs specifically for CCA treatment. Previous CCOs studies mostly focused on classical chemotherapeutic agents (e.g., gemcitabine and cisplatin),^37,52 whereas our work highlighted the potential of ADCs like T-DM1. T-DM1, commonly used for treating breast cancer, has shown efficacy against CCA, with ongoing preclinical and clinical trials related to T-DM1 in CCA.^47,53 Therefore, our study may help to discover new therapeutic strategies and provide more treatment options for CCA patients.

Although our organoid-slice combination multi-organ chip is effective for drug assessment, there are still limitations to our study. First, only 5 patient-derived CCOs were used in this study to test T-DM1 for CCA, but provided a valuable starting point for future investigations. Thus, the response of CCA to T-DM requires further validation with an expanded set of clinical data. Second, we used rat liver and kidney organs rather than human liver and kidney organs to prepare the decellularized matrices. Although previous studies have shown significant similarities between rat and human decellularized matrices, attributed to the shared presence of crucial growth factors and primary collagens,⁴⁶ biological differences between humans and rodents remain.⁵⁴ Lastly, this study focused on exploring the more common hepatorenal toxicity of drugs while omitting the exploration of cardiac toxicity, pulmonary toxicity and other potential toxicities. However, our recellularized slice technique can be customized to meet specific research needs by constructing models with specific cell types for toxicity study or disease research in future work. Collectively, our multi-organ chip constructed through a combination of organoid and recellularized slices provides a biomimetic and efficient platform for the simultaneous in vitro study of the therapeutic efficacy and hepatorenal toxicity of anticancer drugs, promoting the study of anticancer drug repurposing.

5. Conclusion

In this study, we developed a multi-organ chip combining tumor organoid and recellularized tissue slices to mimic a CCA model along with human liver and kidney models, enabling us to assess the therapeutic efficacy and hepatorenal toxicity of T-DM1 for drug repurposing. Perfusion-based culture systems offered continuous biomechanical cues, while human cell-derived models provided better physiological relevance, closely mimicking in vivo conditions and improving predictions of drug efficacy and toxicity. Results of T-DM1 testing on this platform demonstrated its effectiveness in treating CCA with low hepatorenal toxicity in line with the expected ADC profile. Therefore, this microfluidic chip enables the simultaneous in vitro assessment of drug efficacy and toxicity, particularly with strong potential for accelerating the repurposing of existing anticancer drugs. Although our study provides a valuable basis for the future repurposing of T-DM1 in CCA research, further clinical trials are needed to rigorously validate these findings before they can be translated into clinical practice.

Ethical approval

The study was conducted in accordance with the Declaration of Helsinki and approved by the ethical committees of the Ethics Committee on Biomedical Research West China Hospital of Sichuan University (2023 Review (No. 108)). Informed consent was obtained from all patients for being included in the study. All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Sichuan University and approved by West China Hospital of Sichuan University's Animal Care Ethics Committee (No. 20220714003).

Data availability

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

Author contributions

JL generated the concept, performed the experiments, data collection and data analysis, and was involved in writing the manuscript. GW, DW, and CS contributed to the conception of the study and helped revise the data analysis and manuscript with constructive discussions. LW, WZ, QD, QL, WH, HM, ZL, and GL contributed to the conception of the study and helped revise the manuscript with constructive discussions. SW, BH and HH contributed to conceptualization, writing and funding acquisition.

Conflicts of interest

The authors declare that they have no conflict of interest.

Acknowledgements

The authors would like to acknowledge the support from the National Key Research and Development Program of China (2022YFB3804700, 2022YFA1105200), the Organ-on-a-Chip Innovation Base, Central Government-Guided Special Project for Local Scientific and Technological Development in Sichuan Province (2023ZYD0166), the Chengdu City “Unveiling and Commanding” Science and Technology Project (2024-JB00-00018-GX), the Sichuan Science and Technology Program (2024YFFK0384), National Natural Science Foundation of China (82272188), and the Science and Technology project of the Health Planning Committee of Sichuan (24QNMP031).

References

1. Y. Wang, Y. Gao, Y. Pan, D. Zhou, Y. Liu, Y. Yin, J. Yang, Y. Wang and Y. Song, Acta Pharm. Sin. B, 2023, 13, 2483–2509.
2. O. J. Wouters, M. McKee and J. Luyten, JAMA, J. Am. Med. Assoc., 2020, 323, 844–853.
3. D. Sun, W. Gao, H. Hu and S. Zhou, Acta Pharm. Sin. B, 2022, 12, 3049–3062.
4. C.-P. Day, G. Merlino and T. Van Dyke, Cell, 2015, 163, 39–53.
5. K.-J. Jang, M. A. Otieno, J. Ronxhi, H.-K. Lim, L. Ewart, K. R. Kodella, D. B. Petropolis, G. Kulkarni, J. E. Rubins, D. Conegliano, J. Nawroth, D. Simic, W. Lam, M. Singer, E. Barale, B. Singh, M. Sonee, A. J. Streeter, C. Manthey, B. Jones, A. Srivastava, L. C. Andersson, D. Williams, H. Park, R. Barrile, J. Sliz, A. H. Section, S. Haney, K. Karalis, D. E. Ingber and G. A. Hamilton, Sci. Transl. Med., 2019, 11, eaax5516.
6. B. Zhang, A. Korolj, B. F. L. Lai and M. Radisic, Nat. Rev. Mater., 2018, 3, 257–278.
7. S. N. BhatiaandD and E. Ingber, Nat. Biotechnol., 2014, 32, 760–772.
8. J. D. Caplin, N. G. Granados, M. R. James, R. Montazami and N. Hashemi, Adv. Healthcare Mater., 2015, 4, 1426–1450.
9. N. B. Robinson, K. Krieger, F. M. Khan, W. Huffman, M. Chang, A. Naik, R. Yongle, I. Hameed, K. Krieger, L. N. Girardi and M. Gaudino, Int. J. Surg., 2019, 72, 9–13.
10. P. Jarvinen, A. Bonabi, V. Jokinen and T. Sikanen, Adv. Funct. Mater., 2020, 30, 2000479.
11. Q. Liu, L. S. S. Mille, C. Villalobos, I. Anaya, M. Vostatek, S. Yi, W. Li, J. Liao, H. Wu, Y. Song, L. Xiong and Y. S. Zhang, Bio-Des. Manuf., 2023, 6, 373–389.
12. S. H. Lee and J. H. Sung, Adv. Healthcare Mater., 2018, 7, 1700419.
13. C. Ma, Y. Peng, H. Li and W. Chen, Trends Pharmacol. Sci., 2021, 42, 119–133.
14. Y.-S. Weng, S.-F. Chang, M.-C. Shih, S.-H. Tseng and C.-H. Lai, Adv. Mater., 2017, 29, 1701545.
15. M. G. Valverde, L. S. Mille, K. P. Figler, E. Cervantes, V. Y. Li, J. V. Bonventre, R. Masereeuw and Y. S. Zhang, Nat. Rev. Nephrol., 2022, 18, 241–257.
16. O. Mourad, R. Yee, M. Li and S. S. Nunes, Circ. Res., 2023, 132, 483–497.
17. S. Lu, F. Cuzzucoli, J. Jiang, L.-G. Liang, Y. Wang, M. Kong, X. Zhao, W. Cui, J. Li and S. Wang, Lab Chip, 2018, 18, 3379–3392.
18. D. Wu, J. Wu, H. Liu, S. Shi, L. Wang, Y. Huang, X. Yu, Z. Lei, T. Ouyang, J. Shen, G. Wu and S. Wang, Lab Chip, 2023, 23, 4708–4725.
19. Y. Wang, D. Wu, G. Wu, J. Wu, S. Lu, J. Lo, Y. He, C. Zhao, X. Zhao, H. Zhang and S. Wang, Theranostics, 2020, 10, 300–311.
20. R. Booth and H. Kim, Lab Chip, 2012, 12, 1784–1792.
21. X. Ai, L. Zhao, Y. Lu, Y. Hou, T. Lv, Y. Jiang, P. Tu and X. Guo, Anal. Chem., 2020, 92, 11696–11704.
22. C. Strelez, H. Y. Jiang and S. M. Mumenthaler, Trends Biotechnol., 2023, 41, 278–280.
23. A. Herland, B. M. Maoz, D. Das, M. R. Somayaji, R. Prantil-Baun, R. Novak, M. Cronce, T. Huffstater, S. S. F. Jeanty, M. Ingram, A. Chalkiadaki, D. B. Chou, S. Marquez, A. Delahanty, S. Jalili-Firoozinezhad, Y. Milton, A. Sontheimer-Phelps, B. Swenor, O. Levy, K. K. Parker, A. Przekwas and D. E. Ingber, Nat. Biomed. Eng., 2020, 421–436.
24. K. Ronaldson-Bouchard, D. Teles, K. Yeager, D. N. Tavakol, Y. Zhao, A. Chramiec, S. Tagore, M. Summers, S. Stylianos, M. Tamargo, B. M. Lee, S. P. Halligan, E. H. Abaci, Z. Guo, J. Jackow, A. Pappalardo, J. Shih, R. K. Soni, S. Sonar, C. German, A. M. Christiano, A. Califano, K. K. Hirschi, C. S. Chen, A. Przekwas and G. Vunjak-Novakovic, Nat. Biomed. Eng., 2022, 6, 351–371.
25. H. Wang, W. Zhu, C. Xu, W. Su and Z. Li, Metab., Clin. Exp., 2025, 162, 156065.
26. H. Wang, X. Ning, F. Zhao, H. Zhao and D. Li, Theranostics, 2024, 14, 788–818.
27. C. Hu, S. Yang, T. Zhang, Y. Ge, Z. Chen, J. Zhang, Y. Pu and G. Liang, Environ. Int., 2024, 184, 108415.
28. Y. Xia, M. Sun, H. Huang and W.-L. Jin, Signal Transduction Targeted Ther., 2024, 9, 92.
29. L. Sleire, H. E. Forde-Tislevoll, I. A. Netland, L. Leiss, B. S. Skeie and P. O. Enger, Pharmacol. Res., 2017, 124, 74–91.
30. Z. Zhang, L. Zhou, N. Xie, E. C. Nice, T. Zhang, Y. Cui and C. Huang, Signal Transduction Targeted Ther., 2020, 5, 113.
31. M. A. Perazella, Kidney Int., 2012, 81, 1172–1178.
32. A. Reuben, D. G. Koch, W. M. Lee and Acute Liver Failure Study, Hepatology, 2010, 52, 2065–2076.
33. C. J. Zhang, S. R. Meyer, M. J. O'Meara, S. Huang, M. M. Capeling, D. Ferrer-Torres, C. J. Childs, J. R. Spence, R. J. Fontana and J. Z. Sexton, J. Hepatol., 2023, 78, 998–1006.
34. J. Theobald, A. Ghanem, P. Wallisch, A. A. Banaeiyan, M. A. Andrade-Navarro, K. Taskova, M. Haltmeier, A. Kurtz, H. Becker, S. Reuter, R. Mrowka, X. Cheng and S. Woelfl, ACS Biomater. Sci. Eng., 2018, 4, 78–89.
35. G. Wu, D. Wu, J. Lo, Y. Wang, J. Wu, S. Lu, H. Xu, X. Zhao, Y. He, J. Li, U. Demirci and S. Wang, Biomater. Sci., 2020, 8, 2814–2824.
36. P. M. Crapo, T. W. Gilbert and S. F. Badylak, Biomaterials, 2011, 32, 3233–3243.
37. X. Ren, M. Huang, W. Weng, Y. Xie, Y. Wu, S. Zhu, Y. Zhang, D. Li, J. Lai, S. Shen, J. Lin, M. Kuang, X. Li, J. Yu and L. Xu, Cell Rep. Med., 2023, 4, 101277.
38. Y. L. Yu, Y. K. Shao, Y. Q. Ding, K. Z. Lin, B. Chen, H. Z. Zhang, L. N. Zhao, Z. B. Wang, J. S. Zhang, M. L. Tang and J. Mei, Biomaterials, 2014, 35, 6822–6828.
39. K. H. Hussein, B. Ahmadzada, J. C. Correa, A. Sultan, S. Wilken, B. Amiot and S. L. Nyberg, Bioact. Mater., 2024, 40, 280–305.
40. X. Chen, D. Zhang, X. Wang, Z. Liu, H. Kang, C. Liu and F. Chen, Chem. Eng. J., 2023, 477, 146444.
41. S. O. Goksoyr, F. Yadetie, C. T. Johansen, R. G. Jacobsen, R. Lille-Langoy, A. Goksoyr and O. A. Karlsen, Environ. Sci. Technol., 2024, 58, 14098–14109.
42. M. van de Bovenkamp, G. M. M. Groothuis, D. K. F. Meijer and P. Olinga, J. Hepatol., 2006, 45, 696–703.
43. V. Veninga and E. E. Voest, Cancer Cell, 2021, 39, 1190–1201.
44. N. Ferrell, K. B. Ricci, J. Groszek, J. T. Marmerstein and W. H. Fissell, Biotechnol. Bioeng., 2012, 109, 797–803.
45. K.-J. Jang, A. P. Mehr, G. A. Hamilton, L. A. McPartlin, S. Chung, K.-Y. Suh and D. E. Ingber, Integr. Biol., 2013, 5, 1119–1129.
46. L. Li, H. Knutsdottir, K. Hui, M. J. Weiss, J. He, B. Philosophe, A. M. Cameron, C. L. Wolfgang, T. M. Pawlik, G. Ghiaur, A. J. Ewald, E. Mezey, L. S. Bader and F. M. Selaru, Jci Insight, 2019, 4, e121490.
47. Y. Yamashita-Kashima, Y. Yoshimura, T. Fujimura, S. Shu, M. Yanagisawa, K. Yorozu, K. Furugaki, R. Higuchi, J. Shoda and N. Harada, Cancer Chemother. Pharmacol., 2019, 83, 659–671.
48. W. Lin, Y. Zhang, Y. Yang, B. Lin, M. Zhu, J. Xu, Y. Chen, W. Wu, B. Chen, X. Chen, J. Liu, H. Wang, F. Teng, X. Yu, H. Wang, J. Lu, Q. Zhou and L. Teng, Adv. Sci., 2023, 10, 2303908.
49. C. H. Chau, P. S. Steeg and W. D. Figg, Lancet, 2019, 394, 793–804.
50. U. Hafeez, S. Parakh, H. K. Gan and A. M. Scott, Molecules, 2020, 25, 4764.
51. N. Ashman, J. D. Bargh and D. R. Spring, Chem. Soc. Rev., 2022, 51, 9182–9202.
52. H. S. Lee, D. H. Han, K. Cho, S. B. Park, C. Kim, G. Leem, D. E. Jung, S. S. Kwon, C. H. Kim, J. H. Jo, H. W. Lee, S. Y. Song and J. Y. Park, Nat. Commun., 2023, 14, 237.
53. E. G. E. de Vries, J. Rueschoff, M. Lolkema, J. Tabernero, L. Gianni, E. Voest, D. J. A. de Groot, D. Castellano, G. Erb, J. Naab, M. Donica, R. Deurloo, M. S. van Der Heijden and G. Viale, Cancer Med., 2023, 12, 12071–12083.
54. J. L. Balestrini, A. L. Gard, K. A. Gerhold, E. C. Wilcox, A. Liu, J. Schwan, A. V. Le, P. Baevova, S. Dimitrievska, L. Zhao, S. Sundaram, H. Sun, L. Rittie, R. Dyal, T. J. Broekelmann, R. P. Mecham, M. A. Schwartz, L. E. Niklason and E. S. White, Biomaterials, 2016, 102, 220–230.

---