Establishing a Microfluidic Tumor Slice Culture Platform to Study Drug Response

Accurate models for tumor biology and prediction of drug responses of individual tumors require novel technology to grow tumor tissue ex vivo to maintain tumor growth characteristics in situ. Models containing only tumor cells, without the stromal components of the tumor, are suboptimal for many purposes and are generally problematic because the cells are passed through extensive culture and selection. Therefore, direct culture of (human) tumors is of considerable interest for basic tumor biology and diagnostic purposes. Microfluidic technologies have been proposed to accurately mimic physiological conditions for tissue growth. Most published systems build tissues from individual cell types in so‐called Organ‐on‐Chip (OoC) cultures. We here describe a novel OoC device for growing tumor specimens. Thin tumor slices are grown in a microfluidic ‘chip’ that allows precisely controlled in vitro culture conditions. The performance of the OoC device was extensively validated for predicting therapeutic responses in human breast cancer patient‐derived xenograft (PDX) tumor material. The system is amenable to primary tumor material from surgery or biopsies. In addition to using the model to predict and evaluate therapeutic responses, the model can also be used for mechanistic studies of human cancers, such as clonal evolution or immune responses, or to validate new or repurposed (cancer) drugs. The Bi/ond Cancer‐on‐Chip (CoC) device is designed to culture tumor slices and investigate aspects of tumor growth and drug responses. Here, we describe the step‐by‐step process of setting up tumor slice cultures using a Bi/ond CoC device and performing in vitro drug response evaluation. © 2023 The Authors. Current Protocols published by Wiley Periodicals LLC.


INTRODUCTION
Organ-on-chip (OoC) platforms are microfluidic devices that provide a way to culture biological materials in a controlled environment that can mimic the in vivo functionalities of cells, tissues, and organs. Optimal OoC platforms combine 1) a microfluidic chip, 2) cells or a tissue specimen, 3) physiological levels of required stimuli for tissue maturation, and 4) sensors or optics for results readout. The capability of these models to effectively mimic the tumor microenvironment is expected to revolutionize oncology research and anticancer drug screening.
Most OoC systems are constructed as closed devices in which cells are inserted with the regular medium flow. This works well for cell lines or primary cells but cannot be used for larger pieces of tissue. Thin slices can be generated from tumor specimens and cultured for several days ex vivo in the presence or absence of chemotherapeutic drugs, antibodies, or radiation treatment (Ladan et al., 2022;Naipal et al., 2016). Such tissue slices can also be grown in OoC platforms, but only if the system can be closed after inserting the tissue specimen. Compared to standalone tumor slices, the slices on a chip more closely mimic the in vivo conditions with a continuous circulation of nutrients, oxygen gradients, and waste removal systems, making them attractive models to study chemosensitivity. For example, a reproducible microfluidic model was designed to culture precision-sliced thyroid tissue for up to 4 days to evaluate radioiodine sensitivity in real-time (Riley et al., 2019). Similarly, colorectal cancer tumor slices cultured in a microfluidic device demonstrated the potential utility of the platform for multiplexed drug testing, assays with various types of fluorescent cell death indicators, and real-time measurements . A microfluidic device was developed for culturing tumor explants to investigate the responses of HNSCC tissue biopsies to chemotherapeutic drugs (Hattersley et al., 2012).
All these devices were custom-made and are not commercially available. Recently, we described a Cancer-on-Chip device with a 6-well plate design incorporating silicon-based microfluidics, which has application in the parallel culture of multiple tumor slices and diagnostic assays using primary tumor material . The CoC device is commercially available from BIOND Solutions B.V. (Bi/ond) (Delft, The Netherlands, www.gobiond.com), enabling other researchers to use the system for their own purposes. This device has a unique design that 1) facilitates culture of relatively large (tumor) tissue, 2) allows the tissue to receive media and nutrients through top and bottom flow, 3) combines polydimethylsiloxane (PDMS) and silicon-based components, and 4) has a relatively simple design that mimics 6-well culture plates, making it suitable for assessing drug response using patient biopsy material in a clinical setting. Earlier studies describing similar OoC devices had only a single flow (top or bottom) or a perfusion system for media flow (Astolfi et al., 2016;Horowitz et al., 2020;Kennedy et al., 2019;Rodriguez et al., 2020;Shirure and George, 2017). In the Bi/ond CoC device, the tumor biopsy material receives medium via top and bottom flow, which minimizes drug gradients throughout the culture period. The CoC device is connected to fluidics for constant replenishment with media and gas, minimizing culture-induced stress responses in the tumor material .
Earlier studies have revealed the limitations of PDMS-based OoC devices with hydrophobic drugs (Shirure and George, 2017). New materials have been proposed to modify or replace PDMS, such as polyethylene terephthalate (PET) and polystyrene (PS), thermoplastics that mitigate the inherent absorption limitations of PDMS materials (Campbell et al., 2021). However, these materials are less easily modified in a laboratory setting. One of the advantages of the Bi/ond CoC device is that a PDMS membrane is used to support the tumor tissue (enabling easy adaptation of the membrane properties). The rest of the microfluidic components are made of silicon, minimizing the risk of small molecule absorption. The device provides capacity for six simultaneous independent tumor specimens, making it amenable to a broad spectrum of experimental designs. Additionally, the silicon material in the microfluidic chip can be connected to sensors for real-time monitoring of gas concentrations, metabolites, and pH. A convenient thermoreversible hydrogel attaches the tumor tissue to the PDMS membrane, further reducing the effect of shear forces on the tissue as media flow above and below, sustaining optimal growth conditions.
We have demonstrated that the Bi/ond microfluidic CoC device can maintain breast tumor tissue slice cultures for up to 14 days with minimum alteration in cell viability, apoptosis, and tissue architecture. The application of the CoC device was validated for cisplatin response in a breast cancer PDX model and apalutamide (anti-androgen drug) response in prostate cancer PDX tumor slices . Here, we describe the Bi/ond CoC device, the first commercially available microfluidic platform for tumor slice culture, for its application in chemotherapeutic drug testing.
Basic Protocol 1 outlines the materials and methods for breast cancer PDX tumor slice preparation, establishing the CoC device setup for tumor slice culture, and cisplatin drug treatment.
Basic Protocol 2 outlines the materials and methods for preparing the breast cancer PDX tumor slices for histological and immunohistochemical analysis, evaluation of tumor tissue architecture, tumor cell proliferation, DNA damage response, and cell death in treatment-naive and cisplatin-treated samples.

NOTE:
Appropriate informed consent is necessary for obtaining and use of human study material.

ESTABLISHMENT OF BREAST CANCER TUMOR SLICE CULTURE USING A MICROFLUIDIC CANCER-ON-CHIP PLATFORM FOR CHEMOTHERAPY TESTING EX VIVO
This protocol describes Breast Cancer (BrC) PDX tumor collection, preparation, and handling in detail. The method elaborates on the preparation of 300-μm tumor slices suitable for culture. Following the protocol allows the generation of tissue slices and culture conditions that preserve the viability of the original tissue. Next, the microfluidic device is described in more detail. The device is a closed version of the comPLATE (Bi/ond, Delft, The Netherlands). This protocol contains a step-by-step explanation of the procedure, including cleaning, assembly, control during the run, and disassembly of the device. This protocol allows breast cancer tissue slices to be cultured for up to 14 days with or without treatment. A maximum of six slices can be handled within a single setup. A video demonstration of how to assemble the comPLATE can be found at https:// youtu.be/ SKmaQeLZYP8. 2. Place the tissue in a 10-cm dish filled with medium and prepare the tissue for slicing using a scalpel. 4. Pour the 4% agarose/PBS (kept in a 60°C water bath) into one well of a 6-well plate to fill it entirely.

Materials
5. Monitor the temperature until the agarose cools to 37°C.
6. Place the tissue in the 4% agarose and wait until the agarose solidifies (Fig. 1A).
The tissue will sink to the bottom of the well before the agarose solidifies. Ensure that the tissue stays at the top of the plate by holding it in position with metal tweezers. This step will aid the slicing process by positioning the tissue in the preferred location. Alternatively, let the tissue sink to the bottom of the plate and invert the agarose block before cutting. 11. Begin slicing the tissue at a speed of 0.6 to 1 mm/s, amplitude 2.00 mm, and thickness 300 μm.
These settings can be adjusted according to the quality of the tissue. If the tissue is difficult to process, the speed may be decreased and the amplitude increased.
12. Use sterile metal tweezers to transfer the slices to a 10-cm dish filled with breast medium.
Collecting the slices in a dish before placing them into separate wells of a 6-well plate allows randomization. If sequential slices are preferable, they can be collected directly into the 6-well plate.
13. Place each slice in a separate well of a 6-well plate and place it on an orbital shaker at 60 rpm in a 37°C, 5% CO 2 incubator.
Tissue slices can be cultured using this method until the microfluidic setup is complete; the maximum culture time is 1 day.
14. Cut the tubing to appropriate lengths, clean it, and connect it to the perfusion system. Prepare the medium reservoir (15-ml Falcon tube) and collection tubes (15-ml Falcon tubes) and fix them in the appropriate tube holders connected to the shelf inside the CO 2 incubator ( Fig. 2A). Make sure the ends of the tubing are submerged in the medium in both reservoir and collection tubes to avoid air bubbles.
The tubing should be long enough to reach from the perfusion system to the plate and from the plate to the reservoir (the length will depend on the perfusion system selected for this assay). Cut the tubing with sharp scissors or a knife to avoid deformation. 16. Prepare the medium with or without additional (chemotherapeutic) compounds. 19. Place the tissue on the chip using the Mebiol hydrogel ( Fig. 1C and 1D). 20. Close the system with a top plate and tighten with the white ring ( Fig. 1E and 1F). Fill and connect the medium reservoir to the holder attached to the incubator shelf ( Fig.  2A). Connect the collection reservoir to the holder opposite the medium reservoir. Ensure the tubing is fully submerged in both reservoirs.

Komar, van Gent and Chakrabarty
21. Connect the CoC device to the FLPG Plus pumping system with an MFCS-EZ flow control system for continuous perfusion according to manufacturer directions.
22. Connect the microfluidic flow control system (Fluigent MFCS-EZ) and flow rate sensors (Fluigent FLOW UNIT-S) to the comPLATE containing tumor slices in each well using tubing and adaptors. Ensure the MFCS-EZ is connected to the MAESFLO software to check the flow rate during the run.
23. Make sure that the flow rate control system is connected to the Fluigent Microfluidic Automation Tool (MAT) with the predesigned program set for an inlet flow rate of 5 μl/min through the top and bottom channels.
Alternatively, a perfusion system connected to a syringe pump can be used instead of steps 21-23. 24. Run the CoC device for the number of days required for the experiment while checking the system for leaks or disturbances daily. Ensure there is sufficient medium in the reservoir for at least 1 day. 27. While incubating the comPLATE on ice, prepare a 6-well plate with 3 ml breast medium per well.
Prepare enough wells to fit one tissue slice per well.
28. Disassemble the comPLATE and place tissue slices in the medium-filled wells.
Do not forget to properly label the wells before transferring the tissue slices from the OoC device.
29. Add EdU (3 μg/ml) to each well and incubate for 2 hr while placing the tissue on an orbital shaker at 60 rpm in a 37°C and 5% CO 2 incubator.
For fixation instructions, continue to Basic Protocol 2.
30. After the run, thoroughly clean all tubing, adaptors, connectors along, and flow sensors with water, biofilm-removing detergent, and isopropanol. The detergent should be diluted with ultra-pure MQ water (20% IO-Biofilm-Entferner and 80% MQwater). The comPLATE is cleaned with isopropanol followed by ultra-pure MQ water.

HISTOLOGY AND IMMUNOFLUORESCENCE ANALYSIS OF TUMOR TISSUE ARCHITECTURE, CELL PROLIFERATION, AND CELL DEATH
This protocol describes sample fixation and analysis. Tissue slices are fixed using the formalin-fixation paraffin embedding (FFPE) technique. Following this protocol allows the generation of 4-μm tissue sections that can be used for immunostaining or immunohistochemistry. Analyses described in this protocol allow assessment of tissue proliferation (EdU), apoptosis (TUNEL), DNA damage induction (53BP1), and histological tumor architecture (H&E) (Fig. 3).

of 16
Figure 3 Schematic representation of the steps involved in tissue slice processing. Tissue slices are fixed in formalin for 1-3 days. Next, the sample is transferred to cassettes and placed in ethanol (4°C). Samples can be kept in ethanol until the Histokinette processing is done with the tissue processor. The tissue slices are then embedded in paraffin in a block mold. Paraffin blocks can be sliced using a microtome to generate 4-μm slices. Samples prepared by this method are suitable for immunohistochemistry assays. 1. Place the tissue slice in a 15-ml Falcon tube filled with 10 ml formalin and incubate for 1-3 days at room temperature.

Materials
2. Move the tissue slices to 70% ethanol while placing them into cassettes.
Considering the small size of the tissue, it is helpful to use two foam pads and biopsy paper while placing the tissue slices into cassettes.
3. Handle the material with the tissue processor Histokinette and embed samples in paraffin.
5. Proceed to an appropriate staining protocol.

Immunostaining was performed with EdU/DAPI, TUNEL, 53BP1, and H&E. Basic Protocol 2 describes the procedure for EdU/DAPI/53BP1 staining. The manufacturer provides instructions for TUNEL staining.
6. Remove the paraffin from the slides by washing in xylene for 3 min. Repeat 3 times.
7. Rehydrate the sample by washing the slides in an ethanol series, 1 min per wash: twice at 100%, twice at 95%, once at 80%, once at 70%, and once at 50%).
8. Wash the samples twice in demineralized water for 5 min.
9. Boil the samples in pre-heated antigen retrieval buffer for 15 min.
This step can be performed using a microwave. Pre-heat the antigen retrieval buffer during the second deionized water wash. Place samples in the pre-heated antigen retrieval buffer and microwave at 600 watts for 15 min.
10. Let the samples cool to room temperature while keeping them in the antigen retrieval buffer for approximately 30 min.
11. Perform a series of 5-min washes: twice in demineralized water, twice in PBS, and twice in 0.1% Tween-20.
12. Mark tissue areas on the slide using the Immedge Hydrophobic Barrier pen.
This step can be performed during the series of washes described in step 11. Take a single slide, carefully remove the remaining liquid from the slides, and mark a circle around the tissue using the hydrophobic pen. This will allow the liquid to stay within the circle while immunostaining. Place the demineralized water or PBS (depending on the washing steps for the remaining slides) inside the marked circle. Repeat for all slides.
14. Add the Click-iT cocktail and incubate for 30 min.

Tween-20, 0.1%
In a 20-ml volume, prepare 0.1% Tween-20 in PBS. Volume is sufficient for 20 samples. Unused solution can be stored for up to 6 months at room temperature.

Background Information
OoC devices provide a way to maintain cells or tissues under constant flow to mimic the natural environment inside the body as much as possible. Most devices are closed systems in which cells can be inserted via the flow channel(s) (Dsouza, Kuthethur, Kabekkodu, and Chakrabarty, 2022;Ingber, 2022). This protocol describes a method for culturing tissue slices loaded into the OoC device before the system is closed. Several designs have been described elsewhere (Table 1), but most are custom systems that are not commercially available. We described an OoC system that is commercially available . We considered several important features in addition to the option to load tissue specimens before closing the system.
First, experiments must be reproducible. We investigated reproducibility in PDX tumors to demonstrate the ability to perform biological replicates in a controlled setting. Tumor slices from breast and prostate cancers showed good viability, proliferative capacity, and maintenance of tissue integrity for 7 to 14 days. We also confirmed that in vivo responses were mimicked in the OoC device: cisplatin sensitivity in the device was consistent the in vivo characteristics of breast cancer PDX models (Fig. 4), and androgendependent prostate tumors were sensitive to the anti-androgen apalutamide .
Second, the tumor microenvironment should be maintained for the duration of the experiment. We observed no significant differences in tumor architecture, including the condition of stromal fibroblasts. Gene expression analysis showed no significant changes between days 0 and 7, suggesting that the controlled environment was sufficient to maintain the integrity of all major cell types in the tissue. The primary limitation of these experiments is that PDX tumors are much less rich in stromal cells than primary or metastatic tumors directly derived from patients. Therefore, additional characterization of the culture system with these tumor samples provides a more complete understanding of its ability to faithfully maintain all stromal cell types (including immune cells) for the duration of the experiment.
Third, the OoC device should not influence therapeutic efficacy. We showed this is indeed the case for cisplatin and apalutamide, suggesting that absorption did not hamper drug efficacy. We hypothesize that the amount of PDMS in the device is sufficiently limited and absorption is minimal. This represents a substantial advantage of this device, which consists mainly of silicon in the membrane layer, leaving only some absorption by the PDMS. However, we cannot exclude the possibility that other compounds might behave differently. We are currently investigating paclitaxel and PARP inhibitors.
Fourth, multiple parallel experiments should be possible. The 6-well format of the device enables multiplexing. For example, duplicate experiments can be performed for untreated tumor slices and two drug concentrations to discriminate between sensitive and resistant tumors. Each culture well is large (3 × 3 mm), allowing relatively large pieces of tissue to be cultured, and the height of the chamber (several mm) is sufficient to allow growth. When the system is used for much smaller biopsies or parts of biopsies, it might be useful to redesign the chamber to optimize it for such specimens. The tissue slices are immobilized with a thermoreversible hydrogel to prevent movement after loading into the OoC device. This will also help to localize (very) small tissue pieces after the culture period, which is a major problem in standard 6-or 24-well ex vivo cultures.
Finally, the OoC device should be sufficiently flexible to adapt its configuration for membrane porosity, sensor insertion, and optical accessibility. The PDMS layer containing the membrane allows easy redesign of the membrane properties, for which only the chip must be reconfigured. Furthermore, sensors can be integrated into the chip, although this requires more extensive redesign. Optical access is possible from the bottom of the device through the transparent PDMS window. The main disadvantage is the relatively long working distance, as the tissue specimen is on top of the membrane, with the bottom channel between the microscope and the tissue.
We believe the Bi/ond OoC device is well suited for studies of tumor biology, treatment response, and tumor evolution because tumor characteristics can be maintained for a relatively long time. In addition to end-point measurements that can be done when the device is disassembled, it is also possible to measure tumor-derived components in the outflow (such as extracellular vesicles, cell-free DNA, and cytokines) and observe the tumor specimen directly using microscopic techniques Komar, van Gent and Chakrabarty via the transparent window. We have not yet explored these possibilities in detail, so additional analysis is needed to determine the level to which these analyses will be possible. Furthermore, the top and bottom flow design mimics the administration of therapeutic compounds via the bloodstream by coating the bottom channel with endothelial cells.

Critical Parameters
The microfluidic device used in this protocol has several critical parameters. A continuous flow of medium must be provided to the tissue to ensure its survival. Therefore, leaks should be avoided. If the system is set up properly, this issue should not occur, but leaks may occur if, for instance, bubbles form.
Handling of the perfusion system should be done in a way that avoids bubble formation. A regular check of the system is required to ensure the experiment is carried out correctly. This should be done approximately once a day. For other issues with running the microfluidic setup, sample staining, and possible solutions, see Table 2.

Understanding the Results
The Bi/ond microfluidic CoC device was developed to culture relatively large tumor slices inside a microfluidic chip connected to a system that circulates medium inside the chip and removes waste for an extended period. Precision-cut tumor slices are prepared using a Leica VT1200S vibratome and cultured inside the chip, which holds a silicon frame with microfluidic channels embedded in a PDMS film (Fig. 1A-D). The silicon frame has two openings (inlet and outlet) to the channels in the PDMS film (Fig. 1D). The design of comPLATE, which hosts the microfluidic chips, is based on a simple 6-well plate design that provides relative ease of handling for loading and disassembly (Fig. 1E-G). Each well can hold one tumor slice, enabling users to perform six independent experiments in a single run (Fig. 1E).
In each run, six chips can be located on the bottom and covered with a top (Fig. 1F-G). The optical window beneath the well of the microfluidic chip facilitates microscopy during culture (Fig. 1E). The design of the Stain with DAPI before embedding using Vectashield without DAPI microfluidic chip facilitates culture of relatively large tumor slices for drug testing. This is important given that tumors are heterogenous, with varying proportions of tumor and stromal cells significantly impacting drug response. We used a thermoreversible hydrogel to attach the tumor slice inside the microfluidic chip to minimize the effect of shear forces during prolonged culture for up to 14 days. The fluidic system consists of an FLPG pump connected to a microfluidic flow control system (Fluigent MFCS-EZ) and flow rate sensors (Fluigent FLOW UNIT-S) that maintain a flow rate of 5 μl/min. The entire system is placed on a shelf inside the CO 2 incubator ( Fig. 2A and 2B). Fluid pressure is monitored throughout the run in real-time using a Fluigent Microfluidic Automation Tool (MAT) (Fig. 2C). Previously, we showed that tumor slices could be cultured in our CoC device for up to 14 days without any gross change in tumor architecture, cell viability, proliferation, or death .
We cultured PDX-derived tumor slices for 3 days in the presence or absence of cisplatin. We compared cisplatin-sensitive and cisplatin-resistant breast PDX models to validate the system. Histology, cell proliferation (EdU staining), and cell death (TUNEL staining) analysis showed that proliferation decreased considerably in the cisplatin-sensitive model but not in the resistant model. Conversely, cell death increased in the sensitive model but not in the resistant model, demonstrating that the CoC device can accurately predict cisplatin sensitivity in these PDX mod-els ( Fig. 4A and 4B). Consistent with these results, DNA damage analysis showed increased DNA damage in the cisplatin-sensitive PDX (53BP1 foci formation). In contrast, cisplatinresistant PDX showed minimal DNA damage upon cisplatin exposure, confirming that the sensitive model has impaired DNA damage repair ( Fig. 4C and 4D).

Time Considerations
Basic Protocol 1 can be performed in one week. In case of the late arrival of the tumor tissue, overnight incubation at 4°C is possible. Alternatively, slicing can be performed immediately and the slices incubated at 37°C on a rotating platform overnight before assembling the CoC device the following day. Incubation time depends on the treatment chosen. In this setup, tissue slices were incubated with chemotherapy drugs for 3 days. Fixation requires 2 hr for EdU incubation and overnight incubation in formalin. Tissue processing following fixation requires 2-3 days. Immunostaining requires 4-5 hr. The entire procedure can be performed within 1 to 2 weeks from tumor sample retrieval.