PDMS leaching and its implications for on-chip studies focusing on bone regeneration applications

Polydimethylsiloxane (PDMS) is among the most widely used materials for organ-on-chip systems. Despite its multiple bene ﬁ cial characteristics from an engineering point of view, there is a concern about the effect of PDMS on the cells cultured in such devices. The aim of this study was to enhance the understanding of the effect of PDMS on cellular behavior in a context relevant for on-chip studies. The focus was put on an indirect effect of PDMS, namely leaching of uncrosslinked oligomers, particularly for bone regeneration applications. PDMS-based chips were prepared and analyzed for the potential release of PDMS oligomers within the micro ﬂ uidic channel when kept at different ﬂ ow rates. Leaching of uncrosslinked oligomers from PDMS was quanti ﬁ ed as silicon concentration by inductively coupled plasma - optical emission spectrometry and further con ﬁ rmed by mass spectrometry. Subsequently, PDMS-leached media, with a silicon concentration matching the on-chip experiment, were prepared to study cell proliferation and osteogenic differentiation of MC3T3-E1 pre-osteoblasts and human mesenchymal stem cells. The silicon concentration initially detected in the media was inversely proportional to the tested ﬂ ow rates and decreased to control levels within 52h. In addition, by curing the material overnight instead of 2h, regardless of the curing temperature (65 and 80 (cid:1) C), a large reduction in silicon concentration was found, indicating the importance of the PDMS curing parameters. Furthermore, it was shown that PDMS oligomers enhanced the differentiation of MC3T3-E1 pre-osteoblasts, this being a cell type dependent effect as no changes in cell differentiation were observed for human mesenchymal stem cells. Overall, this study illustrates the importance of optimization steps when using PDMS devices for biological studies, in particular PDMS curing conditions and extensive washing steps prior to an experiment.


Introduction
Over the past decade, a variety of novel in vitro platforms that aim at recapitulating physiologically relevant functional units of tissues and organs have been developed (Benam et al., 2015;Bhatia and Ingber, 2014).One of these involves organs-on-chip.Organs-on-chip are predominantly based on microfluidic technology, meaning that cells can be cultured while being perfused through channels of only tens to hundreds of micrometers (Whitesides, 2006).The main advantage of using such systems is the ability to control and adjust environmental parameters to physiologically relevant levels, such as fluid shear stress, mechanical load and biochemical concentration gradients (Bhatia and Ingber, 2014), which would not be possible using traditional static well plate cultures.
One of the most common materials used to fabricate these devices is polydimethylsiloxane (PDMS), a synthetic silicone polymer.Due to its low cost, ease of manipulation and replication procedure that allows rapid prototyping of micron-sized structures, PDMS is among the most widely used materials in microfluidic device fabrication (Duffy et al., 1998;Berthier et al., 2012).Furthermore, PDMS also gains from its optical transparency and gas permeability.A well-known example of a PDMS microfluidic device is the human lung-on-chip that was published a decade ago (Huh, 2010).Whereas PDMS has played a key role in the development of the area, concerns regarding whether PDMS is an appropriate material to be used when working on biomedical applications involving cell culture have been raised (Berthier et al., 2012;Sackmann et al., 2014;Varma and Voldman, 2018).Various studies have reported on the behavior of cells when in direct contact with PDMS.For example, cell attachment and cell growth of four mammalian cell types showed response dependency on PDMS composition (i.e.modifying the ratio of base-to-curing agent) (Lee et al., 2004).Specifically, when PDMS samples were prepared with excess curing agent, primary human umbilical artery endothelial cells and human epithelial cervical cancer cells showed reduced cell growth, while 3T3 fibroblasts and osteoblast-like (MC3T3-E1) cells were not affected.In another study, human intestinal-like cells (Caco-2) that were grown on PDMS showed a lower level of cell adhesion compared to a polystyrene control (Wang et al., 2009).Nevertheless, functionalization of PDMS with extracellular matrix proteins, in particular type I collagen or fibronectin, enhanced Caco-2 cell adhesion.Whereas these results indicate that PDMS could be a suitable platform to sustain cells once optimized, it also becomes apparent that one should carefully consider the effect of PDMS on cellular behavior.
When taking the effect of PDMS on cellular behavior into account, it is essential to look beyond the effects of direct contact (i.e.culturing cells on a PDMS surface).A well-known disadvantage of PDMS is its inherent ability to absorb small hydrophobic molecules.These molecules may deplete from the surrounding medium, diffuse into the bulk of the polymer and may cause a nutritional misbalance in the culture medium or influence the outcome of drug screening (Regehr et al., 2009;Toepke and Beebe, 2006;Van Meer et al., 2017).Another common disadvantage of PDMS is the leaching of uncrosslinked oligomers, even after prolonged curing times (i.e.24 h of curing at 70 C) (Lee et al., 2003).It has previously been shown that PDMS oligomers could be detected in water that had been statically incubated in a PDMS microfluidic channel for 24 h at 37 C (Regehr et al., 2009).In the same study, energy-dispersive X-ray spectroscopy analysis showed that silicon, which was presumably released by PDMS, was present in the cell membranes of mouse mammary epithelial cells that had been cultured in PDMS microchannels for 24 h.Although the implications on cellular behavior were not further evaluated, these findings again suggest the importance of careful interpretation of data obtained from PDMS-containing culture systems.
Since the 1970s, increasing evidence has proven the importance of silicon for bone formation and bone health.Carlisle identified silicon to be an important initiator of bone mineralization and showed that silicon deficiencies in the diet attenuate bone formation in chicks (Carlisle, 1972).Around the same time, Schwarz and Milne reported that rats on a low-silicon intake diet suffered from impaired growth, skull deformations and distortions in the bone surrounding the eye socket (Schwarz and Milne, 1972).Years later, Jugdaohingh et al. examined the effect of dietary silicon intake in humans and found a positive association between silicon intake and skeletal health in men and women, in particular for premenopausal women (Jugdaohsingh et al., 2004).In summary, these results indicate the potential of silicon-containing compounds to affect bone tissue and the behavior of cells that develop this tissue.
Since PDMS microfluidic devices are used as cell culture platforms, it is crucial to raise awareness about how these devices could influence the results of in vitro studies.The overall aim of the current work was to enhance the understanding of cellular behavior within the microenvironmental conditions created by PDMS, particularly caused by the leaching of PDMS oligomers.Firstly, the release of PDMS oligomers into culture medium that had flown through PDMS microfluidic chips was assessed.Subsequently, PDMS-leached media, with a relevant silicon concentration [Si] for organs-on-chip studies, were prepared by using PDMS samples that were cured for different curing times and at different temperatures.Lastly, cell proliferation and osteogenic differentiation were evaluated with calvarial preosteoblasts and human mesenchymal stem cells (hMSCs) grown in PDMS-leached media.

Assessment of PDMS leaching in dynamic conditions
To assess the amount of PDMS oligomers leached out from PDMS when under flow conditions, PDMS-PDMS chips were fabricated and the outcoming medium was characterized for its [Si] by inductively coupled plasma -optical emission spectrometry (ICP-OES).
PDMS (Dow Corning, SYLGARD™ 184) was prepared according to manufacturer's instructions in a 10:1 base:curing agent ratio.Subsequently, the PDMS was cast in stereolithographic 3D printed (Formlabs, Form 2) molds that defined the chip dimensions.After curing for 2 h at 65 C, the PDMS casts were treated with oxygen plasma (Diener electronic, Atto Plasma Cleaner) and afterwards bonded together, creating a tight seal.This resulted in a functional chip with a single microfluidic channel (length ¼ 22 mm, width ¼ 1 mm, height ¼ 0.1 mm).
Minimum Essential Medium (MEM)-α (Gibco™, ref. nr.A1049001) was flown through the chips at flow rates of 0.5, 1.5 or 4.5 μl/min (Harvard Apparatus, PHD, 2000 Infusion) over a time period of 52 h at room temperature.At predetermined time intervals, between 100 and 150 μl of medium was collected from each chip (Fig. 1A).In addition, non-perfused chips were included as a control.These chips were kept under static conditions for 4 h, after which the medium from the microfluidic channel was collected.The collected volume in the flow experiment and the subsequent dilutions were based on the detection limit of silicon for ICP-OES analysis.
The samples were diluted in Milli-Q® water, filtered using a 0.2 μmpore-sized membrane filter (Whatman™, ref. nr. 6781-2504) and finally diluted 1:2 in 2% HNO 3 ref. nr. 438,073).The presented results are adjusted for the appropriate dilution factor.The samples were introduced into the ICP-OES machine (PerkinElmer, Avio 200 ICP Optical Emission Spectrometer) at a flow rate of 1 ml/min, with a delay of 60 s between the initial introduction of the sample and the spectroscopic measurement.As a blank and calibration standard, Milli-Q® water and an aqueous silicon standard solution (PerkinElmer, ref. nr.N9300150) were used respectively.Duplicates of each condition were included in the experiment.

Preparation and [Si] quantification
To evaluate the effect of PDMS on cell proliferation and differentiation, PDMS-leached media were prepared.The large surface-area-tovolume ratio of the PDMS chip limited the preparation of sufficient volumes of PDMS-leached media, which were required for the long-term cell studies.Therefore, these media were instead prepared off-chip.In order to expose the cells to PDMS-leached media relevant for on-chip studies, the focus was put on the preparation of media that fell within the [Si] range found in the on-chip leaching experiment.
PDMS discs (height ¼ 5 mm, diameter ¼ 10 mm) were made by curing 2 g of PDMS in 12 well plates for either 2 h or overnight (O/N, i.e. 19 h) at 65 C or 80 C (Fig. 1B).After curing, the PDMS discs were taken out and transferred into 50 mL Falcon tubes.PDMS-leached media for MC3T3-E1 cells were prepared by incubating each PDMS disc in 10 mL MEM (HyClone™,ref. nr. SH30265.01),supplemented with 10 v/v % fetal bovine serum (FBS) and 1 v/v % penicillin/streptomycin (Gibco™, ref. nr. 15140122), which is later referred to as supplemented MC3T3-E1 medium.After 24 h of leaching at room temperature, the PDMS discs were removed, the media were filter-sterilized and stored at 4 C until further use.MC3T3-E1 supplemented medium that had not been in contact with PDMS was included as a control.The PDMS-leached media samples were named according to the temperature and time the PDMS was previously cured at, i.e.PDMS-65C-2h, PDMS-65C-O/N, PDMS-80C-2h and PDMS-80C-O/N.
For studies with hMSCs, the same procedure was followed, the exception being the medium, which was Dulbecco's Modified Eagle To quantify the [Si] in the culture media, ICP-OES was used as described in section 2.1.Five replicates of each condition were evaluated.

Mass spectrometric detection
For further mass spectrometric evaluation, one sample was selected.PDMS (l ¼ 20 mm, w ¼ 8 mm, h ¼ 6 mm) was cured for 2 h at 65 C.Each sample was incubated in 15 mL supplemented Milli-Q® water in a 50 mL falcon tube.After 24 h incubation at room temperature, the PDMS was removed.The sample was diluted 1:1 with methanol and directly infused via a syringe pump into an ESI LTQ mass spectrometer (Thermo Scientific), using positive ion mode in the mass range of 50-1000.For spectra evaluation, 0.5 min of the resulting chronogram were averaged and compared to a control sample of Milli-Q® water.

Protein absorption
To assess the protein absorption of media components onto PDMS during the preparation of the PDMS-leached media, the total protein concentration in PDMS-leached medium was assessed.For this experiment, PDMS samples were prepared as for mass spectrometric detection.Each sample was incubated in 15 mL supplemented MC3T3-E1 medium in a 50 mL falcon tube for a period of 24 h at room temperature.As a control, supplemented MC3T3-E1 medium that had not been in contact with PDMS was used.The total protein concentration in the media was determined using a micro BCA protein assay kit (Fisher Scientific, ref. nr. 23227), as per manufacturer's instructions.In short, a 30 μl aliquot was taken from each sample and combined with microBCA working solution in a 1:8 sample:working solution ratio.After 30 min incubation at 37 C, the absorbance was read at 562 nm (TECAN, Spark®).Triplicates of each condition were evaluated.

Cell culture conditions
MC3T3-E1 murine calvarial preosteoblasts (subclone 14) were purchased from the American Type Culture Collection (ATCC, CRL-2594).The cells were maintained in MEM-α medium (Gibco™), supplemented with 10 v/v % FBS and 1 v/v % penicillin/streptomycin.This medium was used to culture the cells prior to the experiments since it is free of ascorbic acid, a compound known to stimulate cell differentiation (Franceschi and Iyer, 1992).hMSCs were kindly donated by Prof. Martin Stoddart from the AO Foundation.The cells were maintained in supplemented hMSC medium.Both cell types were kept at 37 C in a humidified atmosphere with 5% CO 2 .
For proliferation and differentiation studies, PDMS-leached medium was prepared as described in section 2.2.1.and additionally supplemented with differentiation factors.Specifically, MC3T3-E1 cells were cultured in supplemented MC3T3-E1 medium with 50 μg/ml ascorbic acid (Sigma-Aldrich, ref. nr.A7631) and 10 mM beta-glycerophosphate (Sigma-Aldrich, ref. nr.G9422), which is later referred to as MC3T3-E1 differentiation medium.Experiments with MC3T3-E1 were performed with cells at passage number 6.In the case of hMSCs, the cells were grown in supplemented hMSC medium with 50 μg/ml ascorbic acid, 10 mM beta-glycerophosphate and 10 nM dexamethasone (Sigma-Aldrich, ref. nr.D4902), which is later referred to as hMSC differentiation medium.hMSCs were used at passage number 2 and 3.
For cell proliferation and differentiation, MC3T3-E1 cells were seeded at 10,000 cells/cm 2 in 48-well plates in supplemented MC3T3-E1 medium.Similarly, hMSCs were seeded at 20,000 cells/cm 2 in supplemented hMSC medium, in 48-well plates for cell proliferation and differentiation, and in 24-well plates for gene expression.After 48 h (MC3T3-E1) and 24 h (hMSCs), the supplemented media were replaced for differentiation media.During the experiment, the media were replaced every two days.After 3, 7 and 14 days in differentiation media, cell proliferation, differentiation and gene expression were assessed.As a control, differentiation medium that had not been in contact with PDMS was included.

Proliferation
Cell proliferation was evaluated using the lactate dehydrogenase (LDH) assay (Sigma-Aldrich, ref. nr.TOX7-1 KT) as an indirect method to quantify the cytosolic enzyme LDH of cells that had previously adhered to the well plate.LDH reduces NAD þ to NADH, which can be measured through a reaction in which a red formazan product is formed.The well plates were rinsed with PBS (Gibco™ ref. nr.14200067) and lysed using 0.1 v/v % Triton-X (Sigma-Aldrich, ref. nr.T8787) dissolved in PBS for 60 min at 37 C.After lysing, a 50 μl aliquot was taken and incubated with 100 μL LDH assay reagents in a 96-well plate.After 25 min of incubation at room temperature protected from light, LDH activity was determined by measuring the absorbance at 490 nm and background absorbance at 690 nm (TECAN, Spark®).Experiments were performed three times, with three samples per condition in each experiment.

Differentiation
Cell differentiation was assessed by measuring alkaline phosphatase (ALP) activity, using a colorimetric method based on the conversion of pnitrophenyl phosphate into p-nitrophenol in the presence of ALP.A 50 μl aliquot of the prepared cell lysate was taken from each sample and combined with 100 μl of alkaline phosphatase substrate (Sigma-Aldrich, ref. nr.P7998) in a 96-well plate.The samples were incubated at room temperature protected from light for 15-40 min.Production of p-nitrophenol was determined by measuring the absorbance at 405 nm (TECAN, Spark®), after which the values were compared to a standard curve with known concentrations of p-nitrophenol (Sigma-Aldrich, ref. nr.N7660).ALP activity was determined by normalizing the calculated p-nitrophenol concentrations to total protein concentration and reaction time in minutes.The total protein concentration was determined using a micro BCA protein assay kit, as described in section 2.2.3.Experiments were performed three times, with three samples per condition in each experiment.

RNA isolation and quantitative real-time PCR (qRT-PCR) analysis on hMSCs
Given the results from the characterization of PDMS-leached medium and the proliferation and differentiation studies on hMSCs, qPCR was performed only on PDMS-65C-2h and PDMS-65C-O/N samples.After 3, 7 and 14 days in hMSC differentiation medium, the well plates were rinsed with PBS and total RNA was isolated using an RNeasy Micro kit (QIAGEN,ref. nr. 74004) according to manufacturer's instructions.After RNA quantification (NanoDrop™ , 2000), cDNA was prepared using a High Capacity RNA-to-cDNA kit (Applied Biosystems™,ref. nr. 4387406).The following genes were selected for investigation: RUNX2 (Runt-related transcription factor 2), COL1A1 (collagen I), SPP1 (osteopontin), BGLAP (osteocalcin) and housekeeping gene ACTB (β-actin) (TaqMan™).To analyze differences in Ct values, the relative quantification method (also known as ΔΔCT) was used.First, the mean cycle threshold for the target gene minus the mean cycle threshold for the endogenous control (ACTB) was calculated, resulting in the ΔCT value.Subsequently, the ΔΔCT values were calculated by deducting the ΔCT from the control from the ΔCT of the target conditions.Lastly, the fold change in gene expression relative to the control was calculated by 2 ÀΔΔCT .Experiments were performed twice, including two samples in each condition and three technical replicates per sample.

Statistical analysis
Statistical analysis was performed using Minitab version 17.The data was evaluated by one-way analysis of variance (ANOVA), two-sided, at a significance level of α ¼ 0.05.Post-hoc Dunnett's test or Tukey test was performed to investigate differences from the control or differences between samples, respectively.The results are presented as mean AE standard deviation from one representative experiment.

On-chip leaching of PDMS oligomers into culture medium
To evaluate the leaching of PDMS oligomers from PDMS-PDMS chips, culture medium that was flown through the chips at different flow rates was assessed for [Si] with ICP-OES.As can be seen from Fig. 2, the [Si] initially detected in the media was inversely proportional to the flow rate.Additionally, for all flow rates, the highest [Si] was detected at the first time point measured, which was within the first 4 h after starting flow.Specifically, at a flow rate of 4.5 μl/min, the peak in [Si] was 0.06 AE 0.02 mM, which decreased within 14 h, reaching steady medium control levels (i.e.0.03 AE 0.02 mM).At a flow rate of 1.5 μl/min, the highest [Si] found was 0.09 AE 0.01 mM, which decreased towards medium control levels over the studied period of ~52 h.The lowest studied flow rate of 0.5 μl/min resulted in a peak in [Si] of 0.19 AE 0.02 mM, which decreased linearly over time but did not reach the control level after 52 h, showing a value of 0.06 AE 0.01 mM at the end of the experimental period.The medium collected from non-perfused chips after 4 h showed a much higher [Si] of 0.68 AE 0.07 mM.

Silicon concentration and mass spectrometric evaluation
PDMS-leached media were prepared by incubating PDMS discs, which had been cured for 2 h or O/N at 65 C or 80 C, in supplemented medium for 24 h.As depicted in Fig. 3, all media samples that had been incubated with PDMS, regardless of curing time and temperature, showed significantly higher [Si] compared to the control medium (p < 0.0005).Medium incubated with PDMS cured for 2 h resulted in an almost 10 times higher [Si] than that of the control medium, whereas if the PDMS had been cured O/N, the resulting [Si] was about 4 times higher than that of the control medium.In other words, PDMS cured for only 2 h caused a significantly higher [Si] than PDMS cured O/N (p < 0.0005).The media samples that were in contact with PDMS cured for the same time showed no differences in [Si], regardless of the curing temperature.Data acquired using direct infusion mass spectrometry showed that the silicon detected in ICP-OES is bound to carbon.In particular, as shown in Fig. 4, the peak at m/z 73 that was detected in the sample is known to be an abundant fragment of PDMS consisting of trimethylsilyl (Si(CH 3 ) 3 ).

Protein concentration
To evaluate the effect of PDMS on the composition of the culture medium, particularly in terms of protein content, total protein was quantified in media samples that were statically incubated with PDMS for 24 h.As can be seen from Fig. 5, no difference in protein content was observed between media samples that had been incubated with PDMS and control media samples.

Cell proliferation and differentiation after exposure to PDMS-leached medium
To evaluate the effect of PDMS oligomers on cell proliferation and differentiation over 14 days, MC3T3-E1 preosteoblast-like cells and hMSCs were cultured in PDMS-leached differentiation medium.Regarding cell proliferation of MC3T3-E1 cells, in all conditions, cell proliferation increased noticeably from day 3 to day 7, with statistical significance obtained for PDMS-65C-2h, PDMS-80C-2h, PDMS-80C-O/N and the control sample (Fig. 6A and Table S.M. 1).As depicted, no significant differences were observed between cells grown in PDMS-leached medium and control medium over the entire time period of 14 days.Regarding cell differentiation, ALP activity increased with statistical significance from day 3 to day 7 for all samples, which continued from day 7 to day 14, with statistical significance for PDMS-65C-2h, PDMS-80C-2h, PDMS-65C-O/N and the control sample (Fig. 6B and Table S.M. 2).On day 3, the ALP levels in all conditions were similar to the control medium.However, on day 7, a general trend was observed, in which all samples that had been in contact with PDMS-leached medium showed an increase in ALP, with statistically significant differences determined for media incubated with PDMS-65C-ON (p ¼ 0.003) and PDMS-80C-ON (p < 0.0005).On day 14, all samples reached similar values in ALP activity again.
To assess potential cell type specific effects of PDMS on cellular proliferation and differentiation, an undifferentiated human cell type, hMSCs, was included in this study.As displayed in Fig. 7A, over the time period of 14 days cell proliferation increased in a time-dependent manner in all conditions, which was statistically significant both from day 3 to day 7 for PDMS-80C-2h, PDMS-65C-O/N, PDMS-80C-O/N and the control sample and from day 7 to day 14 for all conditions (Table S.M. 3).
Similar to MC3T3-E1 cells, hMSCs cultured in PDMS-leached medium did not show differences in cellular proliferation compared to the control over the studied time period.The ALP activity showed a slow increase over the first 7 days, which became significantly elevated from day 7 to day 14 for PDMS-65C-2h, PDMS-80C-2h, PDMS-65C-O/N and the control sample (Fig. 7B, Table S.M. 4).No statistical differences in ALP activity were seen for the hMSCs cultured in PDMS-leached medium compared to control medium.However, on day 14, ALP activity of cells cultured with PDMS, particularly PDMS-65C-2h, showed a higher value than that of cells cultured in the control medium.
To further assess the effect of PDMS on hMSC differentiation towards osteoblasts, gene expression of early osteogenic markers RUNX2, COL1A1 and late osteogenic markers BGLAP, SPP1 was measured.For this study, only samples incubated with PDMS cured at 65 C were selected since this condition released a higher amount of silicon in solution (Fig. 2).As can be seen in Fig. 8, for all time points, neither upregulation or downregulation of any master gene (fold change ~1.0) was observed.For the early osteogenic markers, at the three time points, almost all evaluated samples showed a slight downregulation in comparison to the control medium (fold-change < 1).This trend was more pronounced on days 7 and 14.In contrast, for the late osteogenic markers, a strong downregulation was only observed for BGLAP when cells were cultured with PDMS-leached media on day 14.Regardless, over the time period of 14 days, there were no statistically significant differences in gene expression levels between the PDMS-leached media samples and the control medium for any of the studied genes.

Discussion
Although PDMS is one of the most widely used materials to fabricate microfluidic systems, there is a concern within the scientific community about whether it could alter certain biological functions of cells contained in such systems, either through direct or indirect contact.This may be of special interest for bone regeneration studies evaluated on-chip, since silicon, the inorganic ion constituting PDMS, is known to enhance bone growth.
Previous studies have reported on the tendency of PDMS to release low molecular weight oligomers.For example, Gross examined multiple PDMS-based items of daily use and showed significant release of low molecular weight oligomers from these products (Gross, 2015).Not surprisingly, in our study, both when under flow and under static conditions, silicon atoms, which were assumed to be PDMS oligomers, were detected in the cell culture medium that had been exposed to PDMS (Figs. 2 and 3).Further evaluation by mass spectrometry showed a peak at m/z 73, which is a known abundant fragment of PDMS consisting of trimethylsilyl (Si(CH 3 ) 3 ) (Fig. 4) (Timko et al., 2009).These findings are in line with work by Regehr et al. in which it was shown that PDMS oligomers can be identified inside PDMS microfluidic channels (Regehr   et al., 2009).The authors reported on the detection of a dimethylsiloxane monomer (around 20 to 90 subunits) in water that had been statically incubated in a PDMS microfluidic channel for 24 h at 37 C. Similar findings were described in a study by Sun et al. in which the main aim was to reduce the amount of uncrosslinked oligomers from PDMS microchips, to avoid hindrance for applications requiring mass spectrometry detection (Sun et al., 2010).
Over the years, various approaches have been explored to remove these uncured PDMS oligomers.In the current study, we evaluated the effect of two curing parameters of PDMS, time and temperature, on the release of oligomers in cell culture medium.It must be noted that the focus of this study was not put on the detection and analysis of PDMS oligomers, rather, it was assumed that the [Si] detected in the cell culture medium was related to the amount of PDMS oligomers.The analysis of [Si] in the PDMS-leached media showed that the amount of PDMS oligomers was inversely proportional to the curing time, irrespective of the tested curing temperature (Fig. 3).This result is in agreement with earlier studies.For example, Sun et al. showed that extended curing times (i.e. from 2 to 72 h at 75 C) significantly reduced the amount of noise from PDMS oligomers in the obtained mass spectra (Sun et al., 2010).In another study, a combination of two approaches was used, which consisted of optimizing the base-to-curing agent ratio and afterwards intensively curing the PDMS (i.e.48 h at 70 C), together resulting in a lower amount of PDMS oligomers being detected (Huikko et al., 2003).Other approaches to decrease the amount of oligomers released by PDMS include Soxhlet extraction and dissolution of PDMS oligomers in different organic solvents (Lee et al., 2003;Kim et al., 2000).When using solvents to extract uncrosslinked PDMS oligomers from their PDMS microfluidic device, Sun et al. found a further decrease in PDMS oligomers (Sun et al., 2010).According to the authors, although the extraction was not complete, longer-term extraction (i.e.16-24 h) did not result in different mass spectra, indicating that the remaining contaminants were either too far  away or migrated too slowly to the surface.Comparing the methodology of changing some curing parameters with that of solvent extraction, Sun et al. proved that increasing the curing time was not as effective as using solvent extraction (Sun et al., 2010).
Although it has been proven that serial solvent extractions of PDMS allows the reduction of PDMS oligomers contaminating the cell culture medium within a microfluidic device, drawbacks of this approach are that it is time-consuming and often involves toxic chemicals, which necessitate extensive washing steps to ensure complete removal before continuing with cell experiments.For this reason, in this work the aim was to evaluate the biological influence of oligomers present in PDMS by modifying the curing conditions.In addition, when our PDMS chips were kept under flow, the amount of detected silicon in the medium was reduced within 14-52 h to or towards media control levels (Fig. 2).As expected, the lower the flow rate, and therefore the longer the medium was in contact with the PDMS, the higher the amount of detected silicon and the slower the reduction of silicon levels over time.This was further confirmed by analysis of medium incubated for 4 h in a non-perfused device (i.e.within the range of typical static incubation times for cells to adhere on-chip), which showed an almost 4 times higher [Si] than the lowest flow rate tested (i.e.0.5 μl/min) (Kim et al., 2007).
Taken together, these findings indicate that when desired, washing of the chip prior to starting the experiment in combination with considering PDMS curing conditions could minimize the amount of uncrosslinked oligomers in a simple yet effective manner.It must be noted that although the on-chip leaching experiments were performed at room temperature and humidity without further supplementing the basal medium, no significant differences in [Si] were found when perfusing medium supplemented with FBS and penicillin/streptomycin through chips maintained at 37 C in a humidified atmosphere with 5% CO 2 (Figure S.M. 1).
Millet et al. have previously illustrated the importance of being aware of the biological effects of uncrosslinked oligomers when using PDMS (Millet et al., 2007).PDMS microfluidic devices were fabricated for culturing hippocampal neurons, which could survive and differentiate within PDMS microfluidic devices only if these were previously subjected to serial extractions of PDMS using solvents prior to assembly.
Silicon plays a key role in bone formation during bone development and remodeling.The fact that PDMS is a synthetic silicone polymer suggests its potential to affect cellular response, particularly when it comes to osteogenic behavior.To assess the biological impact of these PDMS oligomers on osteogenic behavior, MC3T3-E1 cells and hMSCs were cultured in PDMS leached-medium and evaluated over a period of 14 days.The PDMS-leached media used for cell culture were designed to have a [Si] within the range of detected silicon in the dynamic onchip study (Fig. 2), indicating that the cells were exposed to a relevant [Si] for on-chip applications.It must be noted that the studied [Si] (0.11-0.24 mM) was much higher than what has been reported to be physiological, which is in the range of 2-30 μM (Reffitt et al., 2003).In addition, it was verified that the method to prepare PDMS-leached medium (i.e.incubating PDMS in supplemented medium for 24 h) did not lead to a decrease in protein concentration (Fig. 5), suggesting that any change of cell behavior would only be due to the release of PDMS oligomers.In agreement with this, it was shown that on-chip, both over the studied time period of 52 h under flow and after 4 h in non-perfused chips, proteins were not significantly depleted from the medium, indicating that protein absorption would most likely not affect cell response in the initial phase of on-chip experiments (Figure S.M. 2).In this study, PDMS oligomers leached out in cell culture media did not affect cellular proliferation of MC3T3-E1 cells and hMSCs (Figs.6A  and 7A).However, it was demonstrated that MC3T3-E1 cell differentiation was enhanced after 7 days in culture.A higher cell differentiation was observed for cells cultured with PDMS-65-ON and PDMS-80-ON, although these samples were not the ones with highest [Si] (Fig. 6B).In contrast, none of the PDMS-leached media significantly enhanced differentiation in hMSCs, indicating that the effects of PDMS on cellular behavior are cell type specific (Fig. 7B).Although not in full agreement, these results match similar studies in which the effect of silicon on cellular behavior was evaluated.Kim et al. (2013) found that ALP activity of MC3T3-E1 cells was significantly increased after culture with 5 μM (i.e.[Si] ~3.6 μM) and 10 μM sodium metasilicate (i.e.[Si] ~10.7 μM) for 7 days, while only cells cultured with 100 μM sodium metasilicate (i.e.[Si] ~96.4 μM) showed significantly higher ALP activity than the control at day 14.When comparing these findings to our study on MC3T3-E1 cells, a similar trend can be seen, in which after 7 days of culture an increase in ALP activity was observed, which became negligible after 14 days of culture.In another study, hMSCs were cultured in osteogenic culture medium containing between 0.2 and 125 μM silicic acid (Costa-Rodrigues et al., 2016).Over the studied time period of 21 days, DNA content kept increasing and became statistically significant for concentrations !1 μM silicic acid (i.e.expected silicon concentration [Si] exp !~0.3 μM).Regarding ALP activity, it was found that cultures treated with !5 μM silicic acid (i.e.[Si] exp !~1.5 μM) exhibited a statistically significant increase in ALP activity compared to the control after 14 days of culture, with 25 μM silicic acid (i.e.[Si] exp ~7.3 μM) eliciting a maximum response.In addition, in the presence of silicon, a significant increase in the expression of both early and late osteogenic markers was found after 21 days.Over the past years, multiple other studies have been performed with the aim of identifying the effects of silicon on osteogenic behavior.Nevertheless, the variability in silicon sources and tested concentrations, make it complex to compare cellular response in detail and might explain the differences in results.However, in general, our observations seem in line with these previous studies, which indicate that silicon could affect cellular behavior in a time-and concentration-dependent manner.Particularly, the enhanced differentiation of MC3T3-E1 cells and the difference in cellular response between cell types emphasize the importance of being cautious when using PDMS in cell culture platforms and the significance of assessing the effects of PDMS oligomers on different cell types.
In general, it should be noted that the obtained silicon concentrations from the on-chip leaching study are all dependent on the chosen experimental setting.Changing the PDMS preparation parameters, chip fabrication, sterilization methods and flow rates are all factors that have influenced the experimental outcome.

Conclusions
PDMS is a globally used material for on-chip applications, the latter aiming to mimic physiological conditions in vitro.To determine the potential influence of PDMS on results obtained from on-chip experiments, uncured oligomers leached from PDMS, both on-chip and in static conditions, were evaluated by assessing the [Si] in the cell culture media.Moreover, the effect of PDMS-leached oligomers on MC3T3-E1 and hMSC cell proliferation and osteogenic differentiation was studied.This work shows that such released oligomers can affect cellular behavior, in particular differentiation of MC3T3-E1 cells, and that this is a cell type dependent effect.It was proven that the amount of released oligomers could be significantly decreased by modifying the curing conditions of PDMS, in particular, by extending the curing period.Another method demonstrated to be efficient to reduce the quantity of leached oligomers consisted of pre-incubating the PDMS, which in this work was done by flowing cell culture medium through PDMS microfluidic chips.Overall, given that PDMS devices are used as cell culture platforms, these findings indicate the great importance of considering the material properties of PDMS in a biological context.

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

Fig. 1 .
Fig. 1.Schematic illustrating (A) the assessment of PDMS leaching on-chip (B) preparation of PDMS-leached medium for cell culture.
qRT-PCR was carried out with 20 μl reaction volume, consisting of 10 μl 2x TaqMan Universal PCR Master Mix (Applied Biosystems™, ref. nr.A15297), 5 μl diluted cDNA, 1 μl primer (Applied Biosystems™) and 4 μl of RNAse-free water.For amplification, the CFX Connect System (Bio-Rad) was used.The CFX Manager software automatically calculates the Ct (cycle threshold) values.Samples with a Ct value less than or equal to 35 were considered for analysis.

Fig. 2 .
Fig. 2. ICP-OES detection of the silicon concentration ([Si]) in culture medium flown through PDMS-PDMS chips at flow rates of 0.5, 1.5 and 4.5 μl/min for 52 h and in medium collected from non-perfused chips after 4 h.The PDMS pieces were previously cured for 2 h at 65 C. The data was plotted at the average time point of each sample collection period.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4 .
Fig. 4. Mass spectrum obtained for PDMS-leached Milli-Q® water.The PDMS pieces were previously cured for 2 h at 65 C.

Fig. 5 .
Fig. 5.Total protein content in culture medium incubated with PDMS (previously cured for 2 h at 65 C) for 24 h and control medium (i.e.not exposed to PDMS).

Fig. 6 .
Fig. 6.Effects of PDMS-leached compounds on MC3T3-E1 (A) cell proliferation and (B) differentiation over a period of 14 days.Culture medium not exposed to PDMS was included as a control.Bars labeled with different letters indicate statistically significant differences between time-points within each treatment (p < 0.05).*corresponds to a significant difference from the control (p < 0.05).

Fig. 7 .
Fig. 7. Effects of PDMS-leached compounds on hMSC (A) cell proliferation and (B) differentiation over a period of 14 days.Culture medium not exposed to PDMS was included as a control.Bars labeled with different letters indicate statistically significant differences between time-points within each treatment (p < 0.05).

Fig. 8 .
Fig. 8. Gene expression of early osteogenic markers (A) RUNX2 and (B) COL1A1 and late osteogenic markers (C) BGLAP and (D) SPP1 in hMSCs after exposure to PDMS-leached media.Culture media not exposed to PDMS was included as a control.