An agarose-alginate microfluidic device for the study of spheroid invasion, ATRA inhibits CAFs-mediated matrix remodeling

Growing evidence demonstrates that cancer-associated fibroblasts (CAF) are responsible for tumor genesis, growth, metastasis, and treatment response. Therefore, targeting these cells may contribute to tumor control. It has been proposed that targeting key molecules and pathways of proliferative functions can be more effective than killing CAFs. In this regard, multicellular aggregates, like spheroids, can be used as human tumor models. Spheroids closely resemble human tumors and mimic many of their features. Microfluidic systems are ideal for cultivation and study of spheroids. These systems can be designed with different biological and synthetic matrices in order to have a more realistic simulation of the tumor microenvironment (TME). In this study, we investigated the effect of all-trans retinoic acid (ATRA) on 3D spheroid invasion of MDA-MB cells exposed to hydrogel matrix derived from CAFs. The number of invasive cells significantly decreased in CAF-ECM hydrogel treated with ATRA (p < 0.05), which indicates that ATRA could be effective for CAFs normalization. This experiment was done using an agarose-alginate microfluidic chip. As compared with common methods, such hydrogel casting is an easier method for chip fabrication and can even reduce costs.


Introduction
In 2020, the World Health Organization estimates about 10 million deaths in the world will be caused by cancer, or nearly one death every six (WHO. cancer overview. 2021). Despite many advances in cancer treatment, metastasis is one of the main causes of death in this population. Metastasis is a multi-step process and many cellular and molecular factors are involved in it. Because of the high complexity of this process, metastasis simulation in conventional models like as cell culture flasks, multi-well plates, animal models and etc., has produced unrealistic results. A new generation of 3D models, such as microfluidic systems, has been successful in mimicking cancer-related phenomena in recent years. Several cell types may be cultivated at the same time in microfluidic chips, and different types of extracellular matrix can be implanted in the chip. However, fibroblasts co-culture has gained a lot of interest in microfluidic metastasis researches (Ao et al. 2015;Fang et al. 2021;Zhao et al. 2020;Ro et al. 2022).
Cancer-associated fibroblasts (CAFs), which are present in the stroma around solid tumors, have been shown to play a significant role in tumor initiation, progression, metastasis, and response to anticancer therapies (Kalluri 2016;Öhlund et al. 2014;Santi et al. 2018;Crawford et al. 2009;Sahai et al. 2020). In solid tumor stroma, CAF subpopulations exist in different functional subgroups (Öhlund et al. 2017;Kieffer et al. 2020;Davidson et al. 2020;Wu et al. 2020Wu et al. , 2021 and some of them have the ability to decelerate tumor growth (Özdemir et al. 2014;Rhim et al. 2014). Recent research indicates that they are dynamically interconvertible (Biffi et al. 2019;Dominguez et al. 2020). CAFs similar to myofibroblasts (myoCAF) produce collagen-rich extracellular matrix (ECM) in addition to alpha smooth muscle actin (αSMA) (Öhlund et al. 2017;Kieffer et al. 2020;Davidson et al. 2020). There is evidence of myoCAF presence in the tumor stroma in human solid tumors (Öhlund et al. 2017;Davidson et al. 2020;Dominguez et al. 2020;Costa et al. 2018). The high level of SMAexpressing CAF abundance shown in breast cancer patients was associated with poor prognose as per higher histological grade, lymph node metastases, and high rates of distant recurrences (Kieffer et al. 2020;Yamashita et al. 2012;Yazhou et al. 2004). Furthermore, CAFs derived from patients and grown in culture, retain their characteristics of myoCAFs, which can promote tumor growth when transplanted with cancer cells (Kieffer et al. 2020;Orimo et al. 2005;Hernandez-Fernaud et al. 2017;Hu et al. 2008;Kojima et al. 2010;Tyan et al. 2011). Pancreatic Ductal Adenocarcinoma (PDA) and murine breast cancer models have shown significant increase in cancer growth and metastasis when proliferating SMA-expressing CAFs were reduced or prevented by inhibiting the hedgehog pathway (Özdemir et al. 2014;Becker et al. 2020;Bailey et al. 2008;Jiang et al. 2020).
The extracellular matrix is also altered by CAFs. Due to its composition and mechanical characteristics, ECMs are thought to drive tumor pathology (active drivers) (Pickup et al. 2014;Alexander and Cukierman 2016;Kai et al. 2019;Nissen et al. 2019). Consequently, targeting ECM formation would have a therapeutic benefit (Olive et al. 2009;Diop-Frimpong et al. 2011;Liu et al. 2012;Takai et al. 2016;Polydorou et al. 2017). ECM components, particularly collagen, are known to promote tumor growth. In breast cancer models, overexpression of collagen I (Col1a1) increases tumor formation and progression, and genetic reduction of collagen VI (Col6a1) decreases tumor development and growth (Chen et al. 2021). Desmoplastic stroma is also promoted by collagen, which makes it hard to develop a tumor vasculature that transports medicine and draws immune cells. A better understanding of ECM production might be tailored to combat cancer . As a result of co-migrating with tumor cells and remodeling the surrounding matrix, CAFs appear to be involved in the formation of distant metastatic sites, leaving a trail behind as a guide for invasive carcinoma cells (Gaggioli et al. 2007;Lee et al. 2018). Definitely, this cell invasion is related to matrix rearrangement.
Direct targeting of CAFs has been connected with little effectiveness due to their variety and diverse sources (Özdemir et al. 2014;Rhim et al. 2014). Instead of eliminating CAFs, another strategy to inhibit cancer progression may be to reprogram them to target particular molecules and pathways that govern their metastasis-promoting function. It is also known that CAF activation promotes tumors by reprogramming the metabolism, which provides a range of nutrients to tumor cells, including lactate, pyruvate, glutamine, alanine, proline, ketone bodies and vitamins (Guido et al. 2012;Zhang et al. 2015;Yang et al. 2016;Sousa et al. 2016;Wang et al. 2016;Olivares et al. 2017;Eckert et al. 2019;Bertero et al. 2019;Zhu et al. 2020). The pancreatic stellate cells (PSCs), one of the progenitors of CAFs (Neuzillet et al. 2019), express retinoic acid receptors, which react with vitamin A metabolites like all-trans retinoic acid (ATRA). PSCs can't remodel their ECM when activated by ATRA, and if the ECM's stiff, they can't feel external mechanical signals (Chronopoulos et al. 2016). As a result, ATRA was chosen in this investigation because of its participation in CAFs metabolism and the formation of the extracellular matrix.
For more than forty years, multicellular aggregates, such as spheroids, have been used as human tumor models (Sutherland 1988;Sutherland et al. 1971). Due to their shape, high cellular density, and chemical environment, spheroids resemble in vivo tumors of the human body. Cells in this threedimensional culture environment can develop certain cellular functions they can't in two-dimensional dishes (Imamura et al. 2015;Ravi et al. 2017). It has become possible with microfluidic culture systems to precisely control these microenvironments and monitor 3D dynamic cellular behaviors (Shin et al. 2012;Wang and Jeon 2022). These systems, however, must be easily manufactured and processed as well as simulating precisely. Hydrogel casting is a method with significant potential in this field (Nie et al. 2020). It requires less equipment and produces fewer costs than traditional microfluidic manufacturing methods. In this study, we investigate the effect of ATRA on 3D spheroid invasion of MDA-MB cells exposed to hydrogel matrix derived from CAFs. As mentioned, ATRA can prevent the activation of PSCs to CAFs. There will have to be further testing to determine whether ATRA is also capable of reverting CAFs to an inactive state. A microfluidic chip made of agarose has been used for this testing, which offers the advantages of a simple manufacturing process and a low cost. Considering the microwell in this chip provides the possibility of producing spheroids and facilitates the possibility of further investigations in the same chip. In addition to invasion studies, this chip with the same design can be used in cell migration (Yuan et al. 2019) and angiogenesis (Nashimoto et al. 2020) studies.

MDA-MB 231 cell culture and staining
Cell lines were purchased from the Iranian Biological Resource Center (IBRC, Tehran, Iran) and were labeled with IBRC C10684. These cells were grown in a high glucose DMEM medium containing 10% fetal bovine serum (FBS) and 1% penicillin/ streptomycin. MDA-MB231 Cell lines were incubated under 5% CO2 and 37 °C humidification in incubator.
By labeling tumor cells with fluorescent proteins, we were able to track them over time. GFP transfection is carried out in a 24-well cell culture plate. 70,000 cells were seeded 24 h before transfection in 500 µl of DMEM/F12 with 10% FBS per well. To starve the cells prior to transfection, 200 µl of serum-free culture medium were replaced with previous medium for 1 h. We used PEI 25KD (Gold standard) as a nanocarrier for GFP expressing PFGFP plasmid delivery according to a previous protocol (Shayestehfar, et al. 2022). Briefly, we prepared 1 µl of PEI 25KD stock solution (1 mg/ ml) in 25 µl sterile water and 1 µg plasmid in 25 µl sterile water, then PEI solution was added to plasmid solution, then pipetted for 30 times and incubate for 30 min in room temperature (RT). This complex was added to each well in the next step. cells were incubated for 4 h, and then 500 µl of fresh medium with FBS and Pen/Strep were replaced. After 48 h, we evaluated the results by fluorescent microscopy and flow cytometry.

Isolation, culturing and characterization of fibroblasts
We obtained informed consent from all patients for fibroblast isolation and were approved by the Research Ethics Committee of Iran University of Medical Science (IR.IUMS.REC.1398.663). Cancer specimens were collected from patients at Firoozgar Hospital (Tehran, Iran) who had undergone tumor resection surgery. The tissue was cut into 1 mm pieces using a sterile scalpel. These small pieces of tissue were placed in a cell culture flask containing low glucose DMEM medium with 10% FBS and 1% penicillin/streptomycin. The flask was incubated under 5% CO 2 and 37 °C humidification in incubator for 7 days until fibroblasts crowding. Unattached cells (Immune cells, etc.) were removed the next day with a medium change. After this, we separate CAFs from this mixture using serial (differential) trypsinization. fibroblasts are generally more sensitive to trypsin than tumor cells. Trypsin detaches the fibroblasts within 30-60 s (while tumor cells require 2-3 min). New medium is added to the flask after the detached fibroblasts are collected. To get a homogenous CAF population, we repeat this step 2-3 times. A maximum of six passages was used to isolate CAFs. A total of three CAF samples were collected from three different patients for this study. We describe the pathological characteristics in the Supplementary table.
We used an immunocytochemical test to identify CAFs. First, the cells were fixed with 4% paraformaldehyde and then washed 3 times with PBS every 5 min. Triton 0.3% was added to the samples for 30 min in order to permeabilize the cell membrane, and at the end, they were washed with PBS. We added 10% goat serum (1:100 diluted in PBS) to the samples to block the antibody reaction for 45 min. After removing the goat serum, the primary antibody diluted with PBS was added to the samples and incubated for 24 h in a refrigerator (2-8 °C). After that, it was washed 4 times with PBS at 5-min intervals. Then secondary antibody was added to the samples with a dilution of 1:150 and incubated at 37C for 90 min in the dark. Then the samples were transferred to the dark room and after washing 3 times with PBS, DAPI was added to them. After 20 min, samples washed with PBS and finally, imaging was done with a fluorescent microscope (Xcellence, Olympus). Figure 1 shows a summary of these steps.

Preparation of decellularized fibroblast sheets
We seeded NF (Primary Lung Fibroblast, Normal) and CAF cells with passage numbers between 10 and 12 at 8000 cells/cm2 on six-well plates. The cell culture medium was DMEM with 20% FBS, 20% Ham F12 medium, 500 mM ascorbic acid, and 1% penicillin/streptomycin. Under 5% CO2 and 37 °C, the cells were incubated and during this time the medium was changed every three days. After five weeks of proliferation, the cells were separated as a sheet from the surface of the plate. (Fig. 2A, B). Decellularization was performed after the cell sheets detached from the six well plates. A plate shaker was used at room temperature to shake the plates for 1 h at 75 rpm after withdrawing medium and adding the first decellularization solution. Following this process, three PBS washes were performed for ten minutes at 75 rpm with a plate shaker. In the next step, a second decellularization solution was added after the PBS had been carefully removed. At room temperature, shaking was performed for 30 min at 75 rpm. PBS was again applied three times at 75 rpm for 10 min. The cell sheets were rinsed in DMEM with 20% FBS for 48 h at room temperature at 75 rpm and then immersed in fresh media. A final wash step was carried out in PBS three times for ten minutes. The decellularized extracellular matrix Fig. 1 The steps we took to reach CAFs. A After breast surgery, the samples transferred to lab in cold PBS. B The samples were minced and cultured in the flask for 10 days until fibroblasts were crowded and replicated. C Immunocytochemistry was done for fibroblast characterization using Pan-CK (as Negative) and FAP (as Positive) markers (dECM) can be stored at 4 °C in PBS. The removal of nuclear components was evaluated by staining dECM structures with DAPI (Fig. 2C, D). More information about decellularization solutions is provided in the Supplementary information section.

dECM hydrogel preparation
Choosing the appropriate dissolution protocol is crucial to the hydrogel-forming ability of ECM, so we used the Uriel method (Uriel et al. 2009) for matrix preparation for minimum proteins degradation. In order to homogenize the dECM, a high salt buffer solution containing protease inhibitor cocktail was used (3.4 M sodium chloride, 0.05 M Tris pH 7.4, 2 M N-ethylmaleimide and 4 M ethylenediaminetetraacetic acid). Following homogenization, the mixture was centrifuged three times at 7000 g for 15 min, and the supernatant was removed. 1 ml of 2 M TBS buffer (0.15 M sodium chloride and 0.05 Tris pH 7.4) was added to the pellet and incubated overnight at 4 °C while stirring. Following centrifugation at 14,000 g for 20 min, the supernatant was stored and the pellet was redissolved in urea buffer. Once again, the pellet was centrifuged at 14,000 g for 20 min. We used the supernatants from both centrifugation stages for the next steps.

Fabrication of microfluidic metastatic chip
In the first step, we designed a 3D mold of each layer of the chip using SolidWorks 2018 (Dassault Systems, Velizy-Villacoublay, France), and then it was made using an SLA 3D printer. Basically, this chip has three inputs, three outputs, and five channels. One of the side channels has 400 micron-deep and 400 micron-wide wells intended to form spheroids. Using the hydrogel casting method, we fabricated the microfluidic chip in 2 layers (Fig. 3A-H). In this method, sequential castings allow for the creation of two coaxial layers of different hydrogel compositions and cellular contents (Ling et al. 2007;Heidari and Taylor 2020). Different proportions of agarose and alginate were investigated for making chips. A combination of 4% agarose and 2% alginate as optimal combination was used to make each layer ( Figure 5A). Both casted layers transferred from their well-plates to petri dishes before assembly and exposed by high-power mercury UV light for 10 min. This step prepares both layers for connection. To connect two layers, the twice crosslinking technique was employed (Nie et al. 2018;Yajima et al. 2014;Sun et al. 2016). The chip was then immersed in calcium chloride at a temperature of 60 degrees for ten minutes, and the process was then repeated Decellularization assessments under the microscope before and after the procedure. E and F Schematic figure of hydrogel preparation from extracted ECMs after 30 min of cooling (RT). In this procedure the bonding mechanism is based on the ionic crosslinking of alginate with calcium ion (Ca 2+ ). Different percentages of calcium chloride were exposed for binding strength testing. At the end, the chip was exposed to UV rays for 20 min after for sterilization. Below is a detailed flow chart showing the process step-by-step. Supplementary information provided more details about chip feature sizes.

Device loading and spheroid treatments
In our chip, the first channel to be filled is the barrier channel. Using the phase-guided patterning method (Shin et al. 2012), a hydrogel matrix should cover this channel with surface tension through the micro-posts. After that, adjacent cells are prevented from passing through it, but the micronutrients and molecules in the central channel can easily reach the cells in the culture channel. To plug this channel, we used 10 µl of cold Matrigel diluted in PBS (2:1). Incubation at 37C for 5 min completed gelation. In the next step, A 1.5% agarose solution was first prepared and autoclaved. 10 µl of 37 °C FBS were mixed with 40 µl of agarose after being cooled to 60 degrees. The mixture was introduced to the central stimulation channel subsequently. At the end, approximately 100,000 GFP positive MDA-MB231 cells are injected into the cell culture channel of the microfluidic device, and after 10 min, 10 µl of fresh medium are injected to remove any cells that do not fit into the microwells. The channels and how to load the chip are indicated in Fig. 4. Spheroids are formed by incubating microfluidic devices for 24 to 48 h in an incubator. As soon as the spheroids had formed in the wells, we added extracted hydrogel matrices to the channel and monitored the invasion of the spheroids over time. The cold gel matrix was diluted with cold culture medium (4 °C) (with a ratio of 1:3) and then entered into the cell culture channel. After that, the chip was placed in the incubator to complete the gelation process.

Cellular viability
AO/PI Cell Viability Kit (Labtech) was used to test cell viability according to the manufacturer's instructions. As a brief overview, we used 2 mg/mL Fig. 3 Step-by-step microfluidic chip fabrication. A Main layer design, B Positive mold designed for casting main layer, C Positive mold designed for casting ports layer, D, E, F 3D printed molds and their walls, G UV sterilization of casted layers H Chip under the microscope acridine orange in the culture channels, incubated at 37 °C for 40 min, rinsed with PBS, labeled with 2 mg/ mL Propidium Iodide, then imaged. We observed the images using an inverted fluorescent microscope (Xcellence, Olympus). For each condition, at least four images from different chambers were analyzed using ImageJ software. Based on the number of viable cells divided by the total number of cells, a percentage of viable cells was determined.

Spheroid imaging and image analysis
A fluorescence microscope (Xcellence, Olympus) was used to capture all cell images. Statistical analysis and image analysis were performed with Image-Pro Plus 6.0 (Media Cybernetics, Silver Spring, MD) and GraphPad Prism 9. Experiments were replicated three times. Data are expressed as mean x standard deviation. Comparisons between groups were made using one-way analysis of variance (ANOVA). Statistics were considered significant when P-values were less than 0.05 and are indicated with asterisks (*).

CAF ICC assessment
Numerous studies have demonstrated that several subsets of CAF may exist, each characterized by distinct biomarkers. These biomarkers may contribute to a variety of biological functions. CAF subgroups are currently poorly defined, and they appear to overlap due to a variety of variables such as varied tissue heterogeneity, several categorization criteria, multiple biomarkers, and labeling methods chosen by different laboratories. Primary CAF cells should be elongated spindle-shaped and negative for epithelial (EpCAM, Pan-CK), endothelial (CD31) and leukocyte (CD45) markers. It has been discovered that many of the biomarkers represent tumor-promoting CAF subsets, including FAP, PDGFR, Vimentin, PDPN and CD70, as well as some newly identified markers (CD49e, CD10/GRP77 and MHCII/CD74), whereas CD146 + CAFs, CAV1 high CAFs, and PDGFRα + Saa3 − have been identified as tumor-suppressive CAF subgroups (Sahai et al. 2020). We selected two of the most common markers for identifying CAFs from all these markers. Based on standard immunocytochemical analysis, our extracted CAF cells exhibited negative for pan-CK and positive for FAP (80% expression) markers (Fig. 5).

Chip performance
The flexibility of agarose-based hydrogels enables sub-millimeter geometries to be cast even at low polymeric concentrations (1% w/v and lower) and elastic moduli (1-10 kPa and less). As pure agarose has no cell-adhesive domains, it lacks the ability to adhere to cells for long periods, making it unsuitable for long-term cultures. The activation of hydroxyl groups can be used to immobilize adhesion ligands on crosslinked chains, which would resolve the biocompatibility issue (Luo and Shoichet 2004). A second option would be to mix polymers like agarose with natural biopolymers and proteins from the extracellular matrix (Rosser et al. 2015). We used an ECM extracted from patient tissue instead of commercial materials such as collagen, Matrigel, etc., which sets our work apart. As a family, ECM is composed of many members, including, but not limited to proteins (collagen, elastin, fibronectin, laminin, etc.) glycosaminoglycans, fibroblasts, growth factors, and enzymes (Laremore et al. 2009). Additionally, the ECM maintains organization homeostasis by maintaining its equilibrium. It is achieved by balancing metalloproteinases and tissue inhibitors of metalloproteinases, controlling crosslinking enzyme activity, and binding growth factors to the ECM (Mott and Werb 2004). The ECM regulates cell proliferation and differentiation as well as the hosts' response to the ECM (Bonnans et al. 2014;Crapo et al. 2011;Young et al. 2011). By sensing ECM properties through receptors and focal adhesion complexes, cells that are in contact with the ECM maintain tissue homeostasis. The ECM also regulates the expression of enzymes and components in cells. A balance is then reached between deposition and degradation of ECM components by creating a feedback mechanism between cells and ECM (Starr and Fridolfsson 2010). As a result of all of these reasons, the ECM can be used to better reconstruct the cell microenvironment because it has a higher level of complexity. Also, in personalized research for patients, our method for obtaining extracellular matrix can be used.
For multilayered micro-casting, we first have to figure out how layer bonding depends on gel concentration. In a hydrogel chip, a higher concentration of gel makes the chip less transparent. Conversely, in low concentrations of hydrogel, the structures are not sufficiently strong and may be damaged during chip loading. Therefore, an optimal concentration of hydrogel should be considered based on transparency and strength. Agarose is more transparent and stronger than alginate, so it is better to add more to the hydrogel composition, but on the other hand, the amount of alginate must be enough to make double cross-links under the influence of calcium chloride. These are not parameters that can be easily measured, but are mostly determined when working with the chip (Nie et al. 2018;Yajima et al. 2014;Sun et al. 2016. (Several testing during the chip production process with varied concentrations of both types of hydrogels confirmed that 3% agarose and 2% alginate produced an optimum chip in terms of Staining of fibroblast primary cultures with anti-pancytokeratin and anti-fibroblast activated protein antibodies from 3 breast cancer patients. All fibroblasts are negative for pan-cytokeratin (a biomarker for epithelial cells) but CAFs are positive for FAP (a biomarker for active fibroblasts). Blue color indicates nucleus staining (DAPI), while green color indicates an Ab positive staining flexibility, transparency and strength. Furthermore, we conducted tensile testing to explore bonding mechanical strength. In this experiment, it was shown that increasing calcium chloride concentration strengthens the connection, but negatively affects the transparency of the chip. We considered the optimal concentration for two layers to be 300 mM. During chip loading, barrier channel must be filled in an integrated way so that the blocking barrier can function properly. Leaks in adjacent channels may also cause them to not be properly filled during loading and lead to problems with their laminar flow. The leakage of the channel was checked after phenol red was mixed with Matrigel diluted in PBS and introduced into the blocking channel. Visually, there was no leakage at this stage (Fig. 6).
Other aspects of chip performance are characterized by two distinct but related cell behaviors: viability, and proliferation. Viability is calculated as the percentage of cells that survived three days in vitro. Our results show that spheroids survival in the presence of Matrigel, which is a commercial matrix, and CAF-ECM, which is derived from the CAFs, is different (Fig. 7). An evaluation of proliferation in 3D culture is qualitative, in relative terms, based on how the volumetric density has increased over three days. It is essential to consider all of these characteristics when selecting a gel composition for multilayered chips.
Matrix invasion assay and in-situ spheroid treatment Invasion in spheroids may be measured using two approaches. The diameter of the expanded region in the matrix is quantified in the first approach, and the number of cells that have lost their spherical form and extended their filopodia for invasion/migration is counted in the second way (Vinci 2015;Evensen et al. 2013). In this study, cell invasion was assessed by counting the average number of invaded cells in three repeats of pictures from each group. After the spheroids had formed in the wells within 24-48 h, we added the gel matrix to the culture channel and began to investigate their invasion. After two days in vitro microchamber imaging, the cell-spreading or invaded cells are defined as the number of cells showing one or more filopodia and/or clearly nonspherical shape. We compared the invasion level of cells in Matrigel matrix, NF-ECM, CAF-ECM and ATRA treated 9). As shown in Fig. 8, although there were no significant differences between Matrigel and NF-ECM, spheroids exposed to CAF-ECM exhibited higher invasion rates. This result indicates that the matrix extracted from a cancerous tissue can produce different results from a matrix derived from a normal tissue.
Spheroid formation occurs as a result of homotypic tumor cells adhesion. It has been demonstrated by early pioneering studies that cancer cells having a strong propensity to invade are better able to form homotypic aggregates than their Fig. 6 Biophysical measurement of chip function. A Different ratios of gels were investigated and the best concentration was determined. B Based on the tensile test, it was found that the degree of connection between the two layers is directly related to the concentration of calcium chloride. However, concentrations higher than 300 mM significantly increase chip turbidity. Therefore, the most suitable concentration of 300 mM was considered. C The possibility of leakage was checked by injecting food coloring followed by a microscopic examination counterparts with low metastatic potential (Updyke and Nicolson 1986). A tumor spheroid can infiltrate blood vessels and attach to their endothelium (Nashimoto et al. 2020). It is believed that these aggregates have a higher metastatic potential. The usual method for producing spheroids is to culture cells in non-adherent 96-well plates. After forming spheroids, they are immersed in a hydrogel matrix for invasion assessment or transmit to another plate for drug treatment. In this method, the spheroids may not have the same size and usually need to be sorted before examination and may also be Proliferation and Viability charts. We followed these items to check the chip's performance over 3 days. The results of the proliferation graph generally show that cell proliferation is decreasing in the absence of matrix and increasing in the presence of matrix. The matrix extracted from CAFs is more effective on cell proliferation. Viability graph results show that over time, the cell survival in the chip decreases, but this decrease has a smaller slope with the matrix present than without it Fig. 8 Matrix invasion assay. As the ECM is added onto the spheroids on the chip, the cells begin to take morphology, but retain the spheroid structure destroyed during transportation. In contrast, chips similar to ours don't have this problem. Our chip includes cell culture channels embedded with wells. Filling these wells with the correct number of cells results in spheroids of the same size, with a diameter and height of 400 microns.
The role of ATRA in the normalization of CAFs While CAFs have traditionally been associated with tumor promotion as a consequence of paracrine signaling from growth factors and cytokines (Hwang et al. 2008;Wu et al. 2017), their resistant activation impairs not only the biochemical but also the biomechanical synchronization of the tumor microenvironment (Calvo et al. 2013;Otranto et al. 2012). A correlation exists between fibroblasts' ability to contract collagen fibers and their remodeling capacity of the ECM (Calvo et al. 2013). In fact, this could even explain why the diversity of extracellular proteins in the matrix of cancer tissues is different from that of normal tissues, and also why this difference in the arrangement of extracellular proteins is observed. The analogue of vitamin A, ATRA (alltrans retinoic acid), has been shown to maintain the inactive CAFs by targeting the RAR-β (retinoic acid receptor-β) (Chronopoulos et al. 2016). Clinical trials of ATRA for breast cancer (NCT04113863), prostate cancer (NCT03572357), and PDAC (NCT04241276) are currently in progress.
During a CAF-ECM extraction procedure, we added ATRA dissolved in ethanol at a concentration of 1 µM to the culture medium to test its effects. In comparison with other CAF cell sheets, those treated with ATRA had less thickness and less stiffness after 5 weeks. In the microfluidic chip, we found that the amount of cell invasion was less in spheroids exposed to ATRA-treated CAF-ECM than in spheroids exposed to CAF-ECM alone (Fig. 9). ATRA reduces the traction forces imposed by CAFs on their substrates and inhibits their capacity to respond to external mechanical stimuli, which can disrupt CAF mechanosensory activation.
Stroma-targeting strategies for cancer treatment have encountered limited success in ablating or genetically deleting desmoplastic stroma. Considering the multifaceted role played by the desmoplastic stroma in both promoting and retarding cancer progression, stromal-reprogramming strategies that reprogram and inactive CAFs are highly desired. The vitamin A receptor has recently showed promise in this regard. We propose that ATRA or selective RAR-b agonists with improved toxicity profiles can open new doors in the treatment of breast cancer by biomechanically reprogramming CAFs and restoring the microenvironment's biomechanical homeostasis to inhibit cancer invasion and metastasis. The obtained results show that although ATRA reduces invasion and migration of cancer cells, it has a different effect on cell proliferation, which confirms previous animal studies (Coyle, et al. 2018). As a result, combining this drug with an antiproliferative agent might help control metastasis more effectively, and more studies are needed in this area. Fig. 9 ATRA treatment effect on MDA-MB spheroids. The first column shows the spheroids before adding the matrix. In the next two days, the rate of cell invasion in the spheroids was monitored. It is obvious that the matrix that forms in the presence of ATRA is less likely to promote cell invasion. scalebar is 200 µm

Conclusion
Aims to understand tumor metastatic phenotypes as well as molecular pathways, it is imperative to study tumor cells in conjunction with their stromal counterparts. Our investigation was powered by the development of an agarose-alginate microfluidic chip that simulates the invasion of tumor spheroids. Spherical aggregates are more successful at re-creating the original tissue's morphological and functional features than conventional suspension and monolayer cultures. Heterotypic spheroid models, which combine several types of cells or matrices into one tumor microenvironment (TME), are increasingly being used to study noncancerous cells within tumors. Cancer initiation, progression, invasion, and drug sensitivity are influenced by these TMEs (Junttila and Sauvage 2013;Wang et al. 2017).Therefore, we can study tumor/stroma pairing, merging and complexity of heterotypical cell-matrix interactions. To achieve this goal, we used matrices that were driven by normal and cancer associated fibroblasts. Aside from being easy to manufacture and process, agarose based microfluidic device also have a low cost overall. Our agarose-alginate chip can provide a practical and useful tool to describe metastatic processes, dissect biological mechanisms, and find new therapies for cancer.