Cell-imprinted-based integrated microfluidic device for biomedical applications: Computational and experimental studies


 It has been proved that cell-imprinted substrates molded from template cells can be used for the re-culture of that cell while preserving its normal behavior or to differentiate the cultured stem cells into the template cell. In this study, a microfluidic device was presented to modify the previous irregular cell-imprinted substrate and increase imprinting efficiency by regular and objective cell culture. First, a cell-imprinted substrate from template cells was prepared using a microfluidic chip in a regular pattern. Another microfluidic chip with the same pattern was then aligned on the cell-imprinted substrate to create a chondrocyte-imprinted-based integrated microfluidic device. Computational fluid dynamics (CFD) simulations were used to obtain suitable conditions for injecting cells into the microfluidic chip before performing experimental evaluations. In this simulation, the effect of input flow rate, number per unit volume, and size of injected cells in two different sizes of the chip were examined on exerted shear stress and cell trajectories. This numerical simulation was first validated with experiments with cell lines. Finally, chondrocyte was used as template cell to evaluate the chondrogenic differentiation of adipose-derived mesenchymal stem cells (ADSCs) in the chondrocyte-imprinted-based integrated microfluidic device. ADSCs were positioned precisely on the chondrocyte patterns, and without using any chemical growth factor, their fibroblast-like morphology was modified to the spherical morphology of chondrocytes after 14 days of culture. Both immunostaining and gene expression analysis showed improvement in chondrogenic differentiation compared to traditional imprinting methods. This study demonstrated the effectiveness of the cell-imprinted-based integrated microfluidic devices for biomedical applications.


Introduction
It has been proven that cells' natural environment effectively controls cells' function, and cells lose their normal behavior after isolating themselves from their natural environment [1][2][3] . For example, chondrocytes' spherical morphology will be lost after a mono-layer culture on a polystyrene plate, and they will gain broblast-like morphology 4 . Also, providing healthy chondrocytes from the patient or donor and in vitro culture for knee osteoarthritis treatment is one of the most common non-fatal diseases, which is expected to increase dramatically in the coming decades among the world population, is challenging [2,3]. In order to overcome the limitations of using chondrocytes, many researchers have been gradually attracted to adipose-derived mesenchymal stem cells (ADSCs), which can be differentiated for generating cartilage [1,2,4,5].
Therefore, in regenerative medicine based on stem cell manipulation, researchers try to imitate the cells' natural environment by creating similar conditions for their growth and differentiation. The extracellular matrix (ECM), as a supportive structure for a cell, provides the physical, chemical, and mechanical conditions for cell adhesion, growth, and differentiation 5 . Morphological structure and mechanical loading change cell fate during growth in the embryonic period 6 . The traditional substrate for cultivating cells in a laboratory is made of transparent and hydrophobic polystyrene. However, this substrate's rigidity induces a hard tissue, like the bone, in cells and, therefore, increases the osteogenic differentiation probability 7 . The role of culture substrate in cellular behavior has been evaluated in several studies [8][9][10][11] .
The method of inducing differentiation in stem cells through cell shape engineering (imprinting) was rst implemented by Mahmoudi et al. 2 . A cell-imprinted substrate was fabricated from the chondrocyte shape as a physical stimulus for inducing chondrogenic differentiation in stem cells. Moreover, the imprinting method can enhance cardiomyogenic differentiation e ciency in induced pluripotent stem cells 24 .
Tenogenic, osteogenic, kertainogenic, and Schwann cell differentiation in stem cells were also obtained with the imprinting method [25][26][27][28][29] . Also, the effect of physical topography on the cancer cells' response to the conventional anti-cancer drugs was investigated in 30 . Cancer bioimprinting and cell shape recognition can improve detection limits or eliminate the need for a thorough patient samples analysis. 31 . So, the cell-imprinted substrate can manipulate cell phenotypes and regulate their function 24,32 . However, this process's e cacy is poor because of the lack of control over the cells' location.
The rapid development of micro uidic technology is a way of mimicking an in vivo-like microenvironment [33][34][35][36] . The primary purpose of producing microstructures is to convert today's complicated and costly laboratories into fast, inexpensive, and high-e cient micro-scale laboratories 37,38 . This technology can provide and transport cell culture medium and even air, while the waste products by cellular activities are discharged like the human circulatory system. Micro uidic devices' other bene ts are biocompatibility, high surface-area-to-volume ratio, continuous and homogenous feeding of cells, automated cell culture media perfusion, and ease of handling. Moreover, they provide spatial-temporal control on the microenvironment scale (0.1-100 µm). Micro uidic devices are used with the controlled microenvironment to investigate the effects of external factors on cell fate [39][40][41][42][43][44][45] .
Therefore, many research efforts are now focused on using micro uidic devices for cell culture studies in developing medicines and biological research applications, such as drug toxicity or metabolism studies also for stem cell research 33,37,46−49 . Speci cally, the organ on a chip concept has attracted researchers, which is a micro uidic cell culture platform consisting of a continuous perfusion system with living cells that can mimic the tissue or organ 50 . Stem cell culture and differentiation require careful control of several cell culture microenvironment signals regulating intracellular signaling and, eventually, cell phenotype. The development of such precise monitoring is di cult for traditional cell culture systems [40,54]. In addition, in comparison with conventional methods, micro uidic systems can simultaneously combine physical and biochemical factors to provide precise and repeatable stimulation for controlled stem cell differentiation, which is very important in regenerative medicine 33,51−59 .
A hybrid micro uidic system was developed by putting aligned polydimethylsiloxane (PDMS) microgrooves on the surface of biodegradable polymers as physical cues for regulating hiPSC neural differentiation and creating a dynamic microenvironment. Neuronal-speci c gene expression on the Page 4/24 micro uidic device was shown to be considerably higher than traditional systems: an indicator of improved differentiation of hiPSCs into neuronal cells in the micro uidic device 58 .
In traditional imprinting methods, the template cells have entirely random placement, so the secondary cell's probability of being placed exactly on the rst cell template is low. Therefore, herein to increase the traditional imprinting methods e ciency, a micro uidic-based platform is introduced. The template cells are rst cultured in a micro uidic chip on a cell culture plate in this method. Their topography is transferred to a silicone replica by mold casting in a regular pattern. The regular cell-imprinted pattern is then used as a second culture substrate under the other micro uidic chip aligned to the regular cells pattern. The cell culture environment is both predictable and controllable since the entire process is performed inside the chips.
Furthermore, cells' culture is dynamic, and the cell culture medium passes continuously over cells, which mimics the uidic ow like in the body. Using a micrometer level cell-imprinted-based integrated micro uidic device reduces the number of cells needed in one experiment. It introduces the su ciently accurate, reproducible, and low-cost substitute of traditional cell culture plates to control cells' fate. This procedure can be used in cell therapy or drug analysis while preserving normal cell activity or stem cell differentiation into target cells.
Although these experimental methods are reliable, they are very time-consuming to characterize the uid ow in a micro uidic cell chip. Computational uid dynamics (CFD) is a powerful tool for overcoming these limitations and enables complete characterization of ow elds. The design of micro uidic cell culture systems and, therefore, their associated ow and patterns can be theoretically evaluated before construction. Certain parameters such as uid inlet velocity and its effect on shear stress applied to cells or injection cell concentrations and its effect on lling micro uidic chip microchannels can also be evaluated to predict their effect better and obtain appropriate conditions before entering the laboratory and testing on cells. In fact, we will have a virtual lab that will save time and reduce the cost. Therefore, in this study, numerical evaluations of uid ow and cell tracing in the micro uidic chip were performed and validated with an experimental assessment on cell lines. The appropriate conditions of cell injection that have been found through simulation were used to prepare the chondrocyte-imprinted substrate. After chondrogenic differentiation of stem cells, in vitro assessments such as immunocytotoxicity and real-time PCR were done.

Micro uidic chip design
The location of the template cells should be monitored to improve the e ciency of the imprinting process, so a set of 128 micro-channels (20 mm length and 50 µm depth) microchannels with a width of 25 µm and 40 µm comparable to that of template cells have been considered ( Fig. 1.a and FigS1.a).
Approximately 2×10 6 cells (with an average diameter of 8 µm) can be placed in regular and parallel lines in a micro uidic chip with 40 µm microchannels. At the end of each microchannel, three diamond-shaped microposts were considered to inhibit cells' removal from their ends during cell injection. These microposts limited the available space's width to 2 µm per side of the microchannel for moving cells while providing the cell culture medium exchange ( Fig. 1.d). The input channel's width is 0.6 mm and 0.96 mm in the chip with 25 µm and 40 µm microchannels, respectively.

Computational uid dynamics analysis
Computational uid dynamics (CFD) modeling is a useful technique that has been used in the eld of microscale cell culture. It allows a deeper understanding of the function of the hydrodynamic environment and the factors that regulate it. CFD is generally applied to chemical and mechanical engineering; recently, it is used to consider the effects of uid ow on cell function and offers valuable insights into micro uidic cell culture chip design and optimization. Thus, before fabrication, micro uidic cell culture device designs and their respective ow rates and patterns can be theoretically evaluated and characterized. Further precise parameters such as uid inlet velocities and channels dimension can also be varied to better predict their effect on shear stresses, thus optimizing cell growth conditions 35,37,60,61 . So in this study, to better understand the micro uidic chip's ow characteristics, a numerical simulation of the microchip was computed using the COMSOL Multiphysics software. For the simulation, a 2D creeping ow model based on the steady-state Navier-Stokes' equation and the particle tracing model for uid ow were used. An in-compressible uid with 1000 kg/m 3 density and 0.001 Pa.s dynamic viscosity, were considered.
The boundary conditions of inlet velocity and zero pressure were used at the inlet and outlet, respectively, and no-slip conditions were applied to all walls. Particles were released in random position at the chip's inlet with a coupling velocity of the creeping ow model, and in the particle tracing model, drag force based on Stokes' equation was applied. For the initial assessment, the injection ow rate of 2.12 ml/h for syringe pump (equivalent to 0.012291 m/s and 0.019666 m/s inlet velocities in the chip with 40 µm and 25 µm microchannels respectively) and 2×10 6 cells in 170 µl of culture medium with an average diameter of 12 µm and normal distribution of particles were considered.
A convergence and mesh independence study was performed for various mesh sizes to investigate mesh element number effect on average shear stress and velocity in the whole chip surface ( Fig. 1.b-c and As shown in the gures, the percentage error with the converged value can be ignored for the selected mesh. Therefore, the selected mesh provides the answer with high accuracy and appropriate calculation time. Finer mesh has been used in the areas around the microposts, which have increased velocities due to reduced passage width. After the appropriate mesh selection, the uid ow and cells tracing model inside the micro uidic chip were solved. Also, various parameters such as uid inlet velocity, number of cells, and cells' sizes were changed to obtain suitable laboratory conditions. First, the results of this study were veri ed by experimental evaluations on cell lines. The appropriate values for cell concentration per unit volume of cell culture medium and injection ow rate of syringe pump chosen based on simulation results were then used to prepare chondrocyte-imprinted substrate and future stem cells chondrogenic differentiation in a cell-imprinted-based integrated micro uidic device.

Cell culture
All the experiments were approved by the ethics committee of the Pasteur Institute of Iran, and all methods were performed in accordance with the relevant guidelines and regulations.
The study was carried out in compliance with the ARRIVE guidelines. In this study, HUVEC, L929, and SW1353 cell lines also isolated Adipose-derived stem cells (ADSCs) and chondrocytes from 6-month-old male New Zealand white rabbits were used. Stem cells and chondrocytes were isolated from sacri ced animals in other studies according to the protocols established at the National Cell Bank of Iran 2,26 . In short, anesthesia was induced by injecting ketamine (35 mg/kg) and xylazine (8 mg/kg) intramuscularly.
Then to harvest samples, barbiturate (100 mg/kg) was injected intraperitoneally. Harvested samples of Hyaline cartilage were washed multiple times with cell culture medium, sliced, and added to the trypsin-EDTA solution (0.25 %, Sigma, USA) and placed in the incubator (37°C). After 30 min, the samples were digested overnight in collagenase type II solution (0.08 mg/ml, Sigma, USA) in the incubator (37°C and 5% CO2). The chondrocytes have the spherical morphology of mature cells a short time after isolation, but they de-differentiate and gain a spindle-shaped morphology after cultivation in a cell culture plate and more extended incubation (~ 14 days); so we used freshly isolated chondrocytes in this study 2 .
For ADSCs isolation, the adipose tissue was collected from the rabbit interscapular region. First, it was put in the DMEM cell culture medium containing antibiotic/antimycotic solution (1%, Invitrogen, USA). After separation of connective tissues, blood vessels, and fragmentation, the fragments were washed with PBS solution containing antibiotic/antimycotic (1%, Invitrogen, USA). Afterwards, they were added to collagenase type I (0.02 mg/ml, Sigma, USA) and were kept for 1 hour in the incubator at 37°C. The cells were then centrifuged, washed, and transferred to the culture medium containing Dulbecco's Modi ed Eagle's Medium (DMEM, GIBCO, Scotland)/Ham's F12 supplemented with 100 µg/ml streptomycin, 100 U/ml penicillin (Sigma, USA), and 10% fetal bovine serum (FBS, Seromed, Germany).
According to our previously published report 62 , ADSCs' multi-potency was assessed in vitro for adipogenesis, osteogenesis, and chondrogenesis.

Micro uidic device fabrication
Basic photolithography accompanied by deep reactive-ion etching (DRIE) of silicon using an oxide mask was used for the master fabrication process. Afterwards, using soft lithography, the pattern was transferred to PDMS (Sylgard 184 Silicon Elastomer Kit, Dow Corning) with the 10:1 weight ratio of the base polymer to the curing agent according to the previously published report 63 .

Cell-imprinted-based integrated micro uidic device fabrication
First, a cell-imprinted substrate was prepared using a micro uidic chip. For this purpose, after sterilization by treatment in the autoclave, a micro uidic chip with channels side facing down was placed on a cell culture plate. The solution with the concentration of 6×10 6 of template cells in 150 µl of culture medium was prepared and injected into the chip using a syringe pump with a ow rate of 50 µl/min. The cell injection was continued until all the microchannels were lled with the cells and gained the desired pattern. In order to ensure the lling of microchannels, the micro uidic chip was observed under a microscope during the injection.
The set was put inside the incubator for 7 hours to enable the cells to adhere to the cell culture plate's surface while obtaining the micro uidic chip pattern. After removing the micro uidic chip from the cell culture plate, the plate's surface, which had the pattern of template cells, was washed with PBS. Then the adhered template cells were xed by 4% glutaraldehyde solution for 1 hour. After washing the xed template cells with distilled water and, after drying, PDMS casting was done. In order to transfer the cell pattern to the PDMS curing process was carried out at 37°C for 1 day. After completing the curing process, the silicone layer was peeled off from the cell culture plate. The cell-imprinted substrate was then washed with 1 M NaOH solution to remove the residues.
For the fabrication of a cell-imprinted-based integrated micro uidic device, a new micro uidic chip should be aligned on the cell-imprinted substrate and attached to it using argon plasma. As the micro uidic chip's design is in parallel lines, and the cell-imprinted substrate pattern is the same as the micro uidic chip pattern, they can easily align under the microscope after argon plasma treatment. Then in order to ensure the bonding of the upper micro uidic chip and the bottom cell-imprinted substrate, the cellimprinted-based integrated micro uidic device was placed on the 80 ºC hot plate for 1 hour. This cellimprinted-based integrated micro uidic device can be used for future cell culture for biomedical applications such as drug analysis on template cells or stem cell differentiation to template cells. After removing cells with trypsin injection, washing and sterilization, the cell-imprinted-based integrated micro uidic device can be used again.

Application of the cell-imprinted-based integrated micro uidic device in stem cell differentiation
A cell-imprinted-based integrated micro uidic device was fabricated based on chondrocyte as template cell according to the above procedure. It was then sterilized using autoclave treatment. There is a static and a dynamic stage for differentiation of ADSCs in the cell-imprinted-based integrated micro uidic device. In the static stage, the solution with the concentration of 3×10 6 of ADSCs in 150 µl of culture medium was prepared and injected into the integrated micro uidic device using a syringe pump with a ow rate of 50 µl/min until the cells lled all the microchannels. In order to allow the attachment of cells to the surface of the cell-imprinted substrate, the integrated micro uidic device should be placed inside the incubator for 4 hours.
A parallel network of 4 cell-imprinted-based integrated micro uidic devices was used to have more differentiated ADSCs to chondrocytes simultaneously. This network was connected to a 25 ml syringe full of cell culture medium. In a dynamic stage during 14 days of ADCSs differentiation to supply the cell culture medium dynamically, a syringe pump with a ow rate of 1 ml/day was connected to the network mentioned above, and this set was placed in the incubator for 14 days. There is no need to change the cell culture medium by an operator like traditional cell culture by this method.
Trypsin-EDTA treatment was done using an insulin syringe instead of pippets in the traditional digestion methods in cell culture plates to remove differentiated cells from inside the cell-imprinted-based integrated micro uidic device after 14 days.

Microscopy observations
Scanning electron microscopy (SEM) was used to characterize the cell-imprinted substrate and observe its structure and morphology.
For uorescence microscopy, the upper micro uidic chip was not bonded to the bottom cell-imprinted substrate for better and easier imaging. So, after ADSCs culture on a cell-imprinted substrate in a micro uidic chip, the micro uidic chip can be removed. For this purpose, to prevent leakage, two rigid plexiglass sheets were used to compress the upper micro uidic chip and the bottom cell-imprinted substrate, which were aligned under a microscope. Cultured cells on the cell imprinted substrate were then xed for 20 minutes in paraformaldehyde (4%, Sigma, USA) before staining. Antibodies employed for staining are shown in Table 1.  64 .
In addition, after 14 days of chondrogenic differentiation of cultured ADSCs on a cell-imprinted substrate, immuno uorescence staining of collagen type II was done 26,39 . The samples were washed with ice-cold PBS two times and incubated for 10 minutes in 0.25 % Triton X-100 to permeabilize the cell membrane. Then, they were washed with PBS three times for 5 min each. Afterwards, they were incubated with 1% BSA for 30 minutes to block the secondary antibody reaction as the additional background color and subsequently incubated with the primary antibody (1: 100 dilution with PBS) for 1 hour at room temperature, followed by washing with PBS three times for 5 min each. Cells were then incubated in the dark with secondary antibody (1:150 dilution with PBS) for 1 hour at room temperature, followed by washing three times for 5 min each in the dark. After four washes in the dark, DAPI (Invitrogen, USA) was added and removed immediately. Then PBS was poured onto the samples, and they were evaluated using a uorescent microscope (Labomed tcs400).

Gene expression analysis
In order to determine the expression of Aggrecan, Collagen I, Collagen II, and Sox9 genes, the Real-time PCR assay by StepOne instrument was used (Applied Biosystems, USA), and to design forward and reverse primers, the sequences of target genes were obtained from the NCBI database (Table 2). In this study, undifferentiated (normal) ADSCs and differentiated ADSCs (on the traditional cell-imprinted substrate and in a cell-imprinted-based integrated micro uidic device) were considered as control and test groups, respectively.
According to the manufacturer's instruction, a blood/Cultured cell total RNA mini kit (Yekta Tajhiz Azma, Iran) was used for total RNA isolation from samples after 14 days. Recombinant DNase I (TaKaRa, Japan) was used for removing Genomic DNA. Then the PrimeScript RT Reagent Kit (TaKaRa, Japan) was used for reverse transcription of total RNAs to produce single-strand cDNA. Finally, SYBR Premix Ex Taq II (Takara, Japan) was used for the Real-time PCR analysis with GAPDH as an endogenous control. The comparative Ct method was used for analyzing each gene expression quantitatively. Each target gene Ct value was normalized to their respected GAPDH.

Computational uid dynamics analysis
As it was mentioned in Sect.  shows the velocity pro le around terminal microposts. As can be seen in the micro uidic chip, we have laminar ow. Around the terminal microposts, where the available space for uid passage decreases, the velocity increases, and maximum velocity occurs in these 2 µm free spaces between the chip and microposts walls.
When high shear stress exists, the rupture of a cell membrane occurs, and this phenomenon is called cell disruption. It is the principal physical cause of the death of cells. Born et al. studied the damage to suspended cells due to shear stress and reported the shear stress range of 200-700 Pa for cell disruption in laminar ow. Figure 2.c shows the shear stress pro le in the micro uidic chip with 40 µm microchannels and around terminal microposts. Figure 2.d shows the pressure contour around terminal microposts. As shown in the micro uidic chip, the shear stress applied to the cells is almost the same along the microchannels. Around the terminal microposts, where the available space for uid passage decreases, the shear stress increases, and maximum shear stress occurs in these 2 µm free spaces between the chip and microposts walls. The surface average shear stress in the micro uidic chip is 0.1504 Pa. This number is below the physiological shear stress of 1 Pa experienced by vascular endothelial cells 65   the time when the rst cells reach the end of the microchannels can be obtained. As shown in Fig. 3.a, since the micro uidic chip design is symmetrical about the x-axis, the particle distribution on the chip surface is also symmetrical. The movement of cells in the top, bottom, and two central microchannels is behind the other microchannels, and these microchannels are the last microchannels to be lled. The movement of cells in the micro uidic chip can be obtained as animation using simulation and can be used as a guide before laboratory experiments (Supplementary video 1). This video can also be compared and validated by experimental injection of cells in the laboratory. Figure 3.b shows the histogram of particle distribution throughout the micro uidic chip when the rst cells reach the terminal microposts. As can be seen, almost all microchannels have received a similar distribution of cells, and the number of cells in the inlet sections, which are larger, is higher than the 40 µm microchannels. Also, the number of cells decreases along the micro uidic chip from the inlet towards the outlet.
Next, we investigated the effect of cells' injection ow rate (input velocity in simulation). Figure 3.c shows the velocity changes along the centerline of the input channel for different inlet velocities in the micro uidic chip with 40 µm microchannels. As can be seen, the velocity slightly increases, reaches a constant value, and decreases again after reaching the ow's partition in the next channels. Figure 3.d shows velocity changes along a cut line on the centerline (before reaching the central channel's terminal microposts) for different inlet velocities. Also, Fig. 3.e shows velocity changes in the vertical direction of a micro uidic chip in eight consecutive microchannels for different inlet velocities. As can be seen, all eight microchannels have the same velocity pro le, and this is true for all microchannels; and according to Fig. 3.d, the velocity remains constant along microchannels. So, according to Fig. 3.d and Fig. 3.e, it can be concluded that all microchannels of the chip have almost the same velocity pro le. Figure 3.f shows velocity changes in the direction of a 2 µm vertical free distance between the micropost and the microchannel wall for different inlet velocities. As can be seen in this area, the amount of velocity is higher than other areas, and it is in parabolic shape because, in contact with the walls, the velocity is zero, but in the distance between the micropost and the microchannel wall, the velocity increased by a reduction in available passage space. As can be seen, the shear stress is higher near the walls and tends to almost zero as it approaches the microchannel's midpoint.
Also, the higher the input velocity, the higher the shear stress applied to the cells. Figure 3.h shows shear stress changes along the 2 µm vertical free distance between the micropost and the microchannel wall for different inlet velocities. Shear stress increases near the micropost and microchannel wall. In addition, to study the effect of input velocity changes, similar to the above simulations, which were performed for cells with an average diameter of 12 µm in normal size distribution ( Fig. 4.a), other simulations were performed for cells with different diameters. Thus, the inlet velocity and the number of cells in 170 µl cell culture medium were xed at 0.012291 m/s and 2×10 6 respectively, but the cell diameter and cell size distribution were almost similar to freshly isolated chondrocytes, L9292 and HUVEC cell lines (average diameters of 8, 14 and 19 µm respectively). The particle size distribution pro les intended for these cells are shown in Fig. 4.a. As can be seen, a wider particle size distribution is considered for larger cells according to laboratory observations.   Therefore, considering the application of future cell culture in the cell-imprinted-based integrated micro uidic device, more differentiated stem cells are needed to place on scaffolds and transplant in the animal's body; it is better to use a micro uidic chip with 40 µm microchannels, which has a higher capacity. Another point to note is that it takes less time for the rst cells to reach the terminal microposts in a chip with smaller microchannels, and it should be noted that this short injection time is more di cult to control, and prolonging the injection time causes excessive cell accumulation. It affects their adhesion and may push cells out of the microchannels' outlet by applying more pressure.
In this part, by simulation, the parameters which affect the experiment were investigated. Therefore, for cell-imprinted substrate preparation, in addition to selecting the appropriate inlet velocity for that does not exert too much shear stress on the cells and does not allow the cells to be out of the incubator for a long time during injection so as not to damage the cells and cause the cells to settle inside the insulin syringe, the su cient concentration of injected cells should be selected so that we can have a regular pattern of cells in parallel lines. Also, the injection ow rate of the dynamic stage cell culture medium can be selected so that applied shear stress in the integrated micro uidic device is consistent with the cartilage space's interstitial uid level. Simulation helps us get an overview of experimental conditions before entering the lab without wasting materials and time.

Validation of numerical analysis with experimental analysis
The SW1353 cell line with an average diameter of 12 µm and a concentration of 1380000 cells in 170 µl cell culture medium in a micro uidic chip with 40 µm microchannels and a syringe pump ow rate of 2.12 ml/h were used to validate the simulation results. For this purpose, after preparing the cell injection conditions using a light microscope and a camera, a video of the movement of cells within uid ow at a frame rate of 30 frames per second was recorded. To validate the simulation, we applied exactly the same conditions in the simulation and prepared an animation of the cell trajectories at the same frame rate. The position of the chip under the microscope was tried to be exactly the same as the part of the chip in the simulation from which the animation was prepared. Then both videos were put beside each other for comparison (Supplementary video 2).
Comparing simulation and experimental videos, we concluded that the simulation accurately predicted the movement of cells in the micro uidic chip and that the cell movements differed in velocity in a few hundredths of a second. This slight difference in simulation and experimental results may be due to the following: 1. When making a micro uidic chip, after several times of molding and removing the PDMS layer from the silicon wafer, some of the microposts may not form properly. They may remain inside the silicon wafer, causing some microchannels to miss some of their terminal microposts. So the resistance against the uid ow and passing cells in some microchannels may be decreased. Compared with the ideal simulation of the same microchannels, this can cause some differences in experimental.
2. In experimental, cell mixing (by pipette up/down) may not be performed well, and the cell concentration inside the insulin syringe may not be uniform, and some cells may make aggregates.
3. There might be an error in cell counting, and the cell number might differ from what was counted by the hemocytometer slide and trypan blue staining (human error).
4. The syringe pump's ow rate used to inject the cell into the micro uidic chip in the laboratory may not be exactly the set value.
Despite the above reasons, the simulation results agree with the experimental results, which shows the power of computational methods that in a virtual laboratory, suitable conditions for experimental studies can be obtained without wasting time and money in a real laboratory. Figure 5.a shows the schematic of the cell-imprinted-based integrated micro uidic device fabrication procedure. Before experimenting with chondrocytes as the main cells in this study, which needs to be isolated from the rabbit's cartilage, the micro uidic chip's functionality in trapping cells and creating a regular pattern the same as the micro uidic chip geometry was evaluated for HUVEC and L929 cell lines. Figure 5.b shows 40µm microchannels of the micro uidic chip after injecting the HUVEC cell line with a syringe pump. The microchannels were almost lled with cells in regular controlled places, and this pattern transferred to the cell-culture plate and looks like regular paving (Fig. 5.c). Figure 5.d also shows the same results for the L929 cell line in 25µm microchannels of the micro uidic chip after injection and their regular pattern transferred to the cell-culture plate (Fig. 5.e). As it was mentioned, it was proved that instead of cell culture in conventional rigid polystyrene plates, cell-imprinted substrates based on their topography provide conditions similar to those of cells' natural growth environment, but in traditional imprinting methods, the placement of cells was random, and the probability of the cells to place exactly on the cell-imprinted topography was low.

Microscopy observations
In this study, by applying a micro uidic chip, the topography of template cells regularly transfers to the cell culture plate. So, a regular cell-imprinted substrate can be made after mold casting on these regular cellular topographies by PDMS. These regular cell-imprinted substrates can provide platforms for anticancer drug analysis in future studies by culturing the same cell line as the template cell line that has been used for cell imprinting.
After evaluating our micro uidic chip's functionality with test cells (HUVEC and L929 cell lines), experiments were performed with chondrocytes, and they were injected into the micro uidic chip with 40µm microchannels. Figure 6.a shows the microchannels were almost lled with chondrocytes in regular controlled places and this pattern transferred to the cell-culture plate ( Fig. 6.b).
The SEM image of one line of the regular chondrocytes' pattern on the cell-imprinted substrate is shown in Fig. 6.c. As can be seen, chondrocytes' topography in a regular arrangement similar to the micro uidic chip's parallel microchannels was transferred to the PDMS replica by mold casting.
Also, proteoglycan's presence secreted by differentiated ADSCs on the cell-imprinted substrate surface in the micro uidic chip was con rmed by Alcian blue staining (Fig. 6.d).
Phalloidin staining, WGA staining, and optical microscopy of ADSCs cultured for ve days on a chondrocyte-imprinted substrate in a micro uidic chip are shown in Fig. 6.e-g, respectively. As it can be seen, just after ve days, the spindle morphology of ADSCs was almost converted into chondrocyte's spherical morphology. Also, ADSCs created a regular pattern like that of the micro uidic chip on parallel lines in the predicted locations on the regular cell-imprinted substrate.
As it was mentioned, uncertainty over whether ADSCs have been speci cally positioned on the chondrocyte pattern on the cell-imprinted substrate is the key issue with the traditional imprinting methods. The location of the cells is completely unpredictable. In this study, based on the parallel lines in micro uidic chip design, a similar chondrocyte pattern is formed on the cell-imprinted substrate. In the cell-imprinted-based integrated micro uidic device fabrication procedure, before the bonding stage, the upper micro uidic chip and the chondrocyte-imprinted substrate can be easily aligned on their similar lines. This approach is objective, controlled, and non-random, unlike previous imprinting techniques.
ADSCs' pathway is predicted in the cell-printed-based integrated micro uidic device to be precisely positioned on the chondrocyte-imprinted pattern. After 14 days of differentiation, ADSCs will get the chondrocyte phenotype.
In addition, after 14 days, ADSCs cultured on the cell-imprinted substrate were stained with collagen II antibody, and cell nuclei were stained with DAPI ( Fig. 7.a). Also, evaluation of chondrogenic differentiation in 3 samples according to image processing with IMAGE J software showed collagen II (chondrocyte speci c gene marker) relative expression of about 68%.

Gene expression analysis
The gene expression analysis of the cultured ADSCs in the cell-imprinted-based integrated micro uidic device was compared with cultured ADSCs on the traditional cell-imprinted substrate (Fig. 7.b.). Undifferentiated ADSCs which were grown on standard cell culture asks were considered as control. In ADSCs cultured in the cell-imprinted-based integrated micro uidic device compared with cultured ADSCs on the traditional cell-imprinted substrate, collagen II expression (a chondrocyte speci c gene marker) is up-regulated while the expression of collagen I is signi cantly down-regulated.
As a criterion for comparison between chondrogenic differentiations in different methods, the ratio of collagen type II to collagen type I expressions was evaluated for cultured ADSCs in the cell-imprintedbased integrated micro uidic device and on the traditional cell-imprinted substrate. According to Fig. 7 In this study, in order to eliminate the drawbacks of traditional imprinting methods, the placement of template cells was predicted by using a micro uidic chip, and imprinted cells got the same pattern as the parallel line of the micro uidic chip. In order to increase the e ciency of traditional imprinting methods using another micro uidic chip aligned on the cell-imprinted substrate, the culture of the secondary cells, which can be the same as template cell (while preserving normal cell activity) or stem cell (to be differentiated to the template cell) based on the biomedical application, is non-random and targeted. Also, the probability of secondary cells placement on the cell-imprinted substrate was increased. In addition, cell culture in this cell-imprinted-based integrated micro uidic device was dynamic using a syringe pump, and the culture medium passed continuously over the cells. In contrast with conventional cell culture methods, there was no need to change the cell culture medium and daily care by an operator, which increases the risk of error. Also, in the case of stem cell differentiation, a network of these cellimprinted-based integrated micro uidic devices was connected to a syringe pump. It simultaneously supplied more differentiated ADSCs to chondrocytes without any chemical growth factors and only with a physical signal and improved stem cell differentiation e ciency, increasing the success of future cell transplantation procedures. This study's cell-imprinted-based integrated micro uidic device can be washed after removing cells by trypsin-EDTA treatment and autoclaved again for further use. All of the above advantages led to a signi cant reduction of the nal cost.
Phalloidin, WGA, and Alcian blue staining results showed that the spindle morphology of ADSCs cultured in the micro uidic device on the cell-imprinted substrate was converted into spherical morphology of chondrocyte by placement into the chondrocyte-imprinted topography. Collagen II and Alcian blue staining as a criterion of chondrogenic differentiation showed positive results. The gene expression analysis showed that ADSCs differentiation in the cell-imprinted-based integrated micro uidic device was successful by increasing collagen II expression and decreasing collagen I expression than control, which was the undifferentiated ADSCs cultured on a cell culture ask. Compared to traditional imprinting, ADSCs differentiation was improved because the ratio of collagen II to collagen I expressions was 2.3 times higher for ADSCs differentiation in the cell-imprinted-based integrated micro uidic device than traditional cell-imprinting methods.
Previous experiments have shown that all cells with possible chondrogenic differentiation capability can be differentiated into chondrocytes on the traditional chondrocyte-printed substrate [26]. So in our chondrocyte-imprinted-based integrated micro uidic device, too, all cells with potential chondrogenic differentiation can be differentiated into chondrocytes.
This method, which is safe, cheap, reproducible, and well-controlled, can be used for all applications involving cell culture instead of cell culture plates in the laboratory or clinic. It can also be generalized to other adherent template cells or other cell transplantation methods, such as heart, skin, neuron, bone, etc. According to the nal application, the template cell can be isolated from the tissue or chosen from cell lines.

Conclusion
In this study, a cell-imprinted-based integrated micro uidic device was presented for biomedical applications, which improved the traditional imprinting cell-imprinted substrate e ciency by controlling the cell culture space. In this method, template cells were prepared in a regular pattern employing a micro uidic chip. After mold casting by PDMS, it was used as a cell-imprinted substrate under another micro uidic chip with the same pattern. When secondary cells were injected into the cell-imprinted-based integrated micro uidic device, there were only template cell patterns under them where they could be placed on it. In addition to PDMS, any materials with nanometer and micrometer dimensions can execute the process of making a cell-imprinted-based integrated micro uidic device based on the cell membrane topography.
In our method, cell culture was objective, non-random, and dynamic in contrast with traditional imprinting methods. Also, the cell culture medium continuously passed through cells.
Also, by applying micro uidic devices and reducing the experimental device's size to micrometer levels, the amount of cell culture medium and the number of cells required in an experiment was reduced leads to a more economical process.
In the stem cell differentiation application, it was possible to supply more differentiated stem cells using a network of multiple integrated micro uidic devices connected to a syringe pump, which can be useful for future clinical trials.
In comparison with rigid polystyrene cell culture plates, as the topography of the cell-imprinted substrate is consistent with the target cells' natural phenotype, the cells' function is much closer to their normal behavior in the body. So, this cheap and reproducible procedure can be used in all cell culture applications, such as growth and proliferation (while preserving normal cell activity) for drug analysis applications or differentiation for various tissue engineering and cell therapeutic applications.
Numerical simulation results can also be used as a guide to determine the effective factors in experimental conditions before entering the laboratory. The simulation results in this study showed that parameters such as injection speed, number, and size of cells, as well as channel dimensions, are effective in the experimental results, and suitable conditions were obtained using simulation before experimental analysis. Validation of simulation results with experimental results showed the power of numerical methods, which saves time and money. In fact, without attending the laboratory, we designed a virtual lab that does not require cell culture materials, the preparation of cell lines, the isolation of cells from the animals, and the operator for culturing, passage, and daily care of cells.

Declaration of competing interests
There are no con icts to declare.