A dialysis medium refreshment cell culture set‐up for an osteoblast‐osteoclast coculture

Culture medium exchange leads to loss of valuable auto‐ and paracrine factors produced by the cells. However, frequent renewal of culture medium is necessary for nutrient supply and to prevent waste product accumulation. Thus it remains the gold standard in cell culture applications. The use of dialysis as a medium refreshment method could provide a solution as low molecular weight molecules such as nutrients and waste products could easily be exchanged, while high molecular weight components such as growth factors, used in cell interactions, could be maintained in the cell culture compartment. This study investigates a dialysis culture approach for an in vitro bone remodeling model. In this model, both the differentiation of human mesenchymal stromal cells (MSCs) into osteoblasts and monocytes (MCs) into osteoclasts is studied. A custom‐made simple dialysis culture system with a commercially available cellulose dialysis insert was developed. The data reported here revealed increased osteoblastic and osteoclastic activity in the dialysis groups compared to the standard nondialysis groups, mainly shown by significantly higher alkaline phosphatase (ALP) and tartrate‐resistant acid phosphatase (TRAP) activity, respectively. This simple culture system has the potential to create a more efficient microenvironment allowing for cell interactions via secreted factors in mono‐ and cocultures and could be applied for many other tissues.

and prevent waste product accumulation, which has been shown to inhibit cell growth (Glacken et al., 1986).
To overcome this waste accumulation problem, cell culture systems have been developed in which culture medium was filtered or dialyzed to remove waste products (Rose et al., 1958) (Rose, 1966).
The principle of dialysis relies on the exclusion of molecules based on their size (Vis et al., 2020). Fresh medium contains low molecular weight (MW) molecules such as nutrients, amino acids, and vitamins.
Depending on the chosen MW cut-off (MWCO), the dialysis membrane allows for exchange of those molecules. In this way waste products can diffuse out of the culture medium while nutrients and vitamins can diffuse back in. High MW components such as growth factors produced by the cells or added to the medium are retained in the medium compartment (Vis et al., 2020).
This dialysis culture principle has already been reported in 1958 by G.G Rose who introduced the concept of using a cellophane membrane to divide the tissue culture part from the culture medium exchange part (Rose et al., 1958) (Rose, 1966). He kept advancing the set-up and eventually engineered a 12-chambered tissue culture system with dialysis membranes (Rose, 1967). Decades later, E.A. Vogler reported a cell culture device based on Rose's work and showed long-term cell culture (30 days) of a variety of mammalian cells (epithelioid [canine], fibroblastic [monkey], hybridoma, primary thymocytes and splenocytes [all murine]) in the stable culture environment created by dialysis (Vogler, 1989). Vogler's group further expanded this idea of "simultaneous growth and dialysis" and over the years reported work in the field of bone tissue engineering and breast cancer metastasis with cell cultures ranging from twodimensional (2D) to three-dimensional (3D) (Dhurjati et al., 2006) (Mastro & Vogler, 2009) (Krishnan et al., 2011) (Krishnan et al., 2015).
Currently, dialysis for re-use of culture medium is not frequently reported. It is remarkable that these techniques are not used more often in cell culture systems while they have the potential to improve cell function tremendously. Dialysis cultures have both economic and biological benefits. The economic benefits come from the fact that less of the costly macromolecules for cell proliferation and differentiation are needed in the culture medium. The biological benefit originates from the maintenance of the cells' secretome over time, which leads to a more stable cell culture environment. This seems particularly imperative for studies investigating cell-cellcommunication via cell-secreted macromolecules or extracellular vesicles. Through the retention of those factors in the medium, dialysis cultures could be beneficial for the proliferation and differentiation process of many cell types, particularly for the culture of in vitro tissue models. These models aim at representing the in vivo situation as closely as possible to for example test the effect of drugs on cells. It is hypothesized that a microenvironment providing more stability through less fluctuations could contribute to the efficacy and reliability of such models.
In the most recent study of Krishnan and colleagues, Vogler's bioreactor was used to create a 3D bone remodeling model with murine osteoblasts and osteoclasts cultured over a period of 2-10 months (Krishnan et al., 2015). With a goal towards in vitro boneremodeling models for drug testing and personalized medicine, a human model would be preferred over a murine model due to interspecies differences. Moreover, a 2-10 month culture period would generate practical issues for high throughput drug testing. Therefore, we propose a fully human 3D bone-remodeling model, including the use of human platelet lysate (hPL) instead of the generally used fetal bovine serum (FBS) (de Wildt et al., 2022a). We present a simple, cost-effective method for incorporating dialysis into the cell culture. A silk fibroin (SF) scaffold will be integrated to reduce the amount of time needed to create the 3D bone tissue (Figure 1a).
For in vitro bone tissue formation, the extracellular matrix (ECM) production by the cells is of uttermost importance. Osteoblasts produce collagen type 1 ECM that is mineralized both intrafibrillarly and extrafibrillarly (de Wildt et al., 2019). The cells' secretome is essential in this process. Continuous removal or periodic exchange of culture medium considerably disturbs these mineralization processes, making it difficult to induce in vitro bone tissue formation (Dhurjati et al., 2006). Therefore, in this study, a dialysis culture approach is investigated for an in vitro bone remodeling model by differentiation of human mesenchymal stromal cells (MSCs) into osteoblasts and monocytes (MCs) into osteoclasts ( Figure 1b). A custom-made simple dialysis culture system with a commercially available cellulose dialysis insert is developed. It is believed that the dialysis of culture medium could contribute to a more efficient and more physiological environment for cell proliferation and differentiation, as the cells' secretome remains in the culture.
Silk-salt blocks were immersed in 90% methanol in UPW for 30 min to induce β-sheet formation (Tsukada et al., 1994). NaCl was extracted in UPW for 2 days. Scaffolds were cut into disks of 1 mm height, punched with a 5-mm diameter biopsy punch, and autoclaved in phosphate buffered saline (PBS) at 121°C for 20 min for sterilization.

| Fabrication of the dialysis culture plates
A holder with 48 holes was custom-made (poly carbonate, designed in Inventor Professional, Autodesk) to fit a 48-wells plate (677180, Greiner) ( Figure 2). The holder was autoclaved at 121°C for 20 min Experimental set-up: (a) A silk fibroin scaffold was seeded with cells and cultured in a custom-made dialysis culture dish. The dialysis membrane ensured retaining of high molecular weight growth-and communication factors in the cell culture insert. Nutrients and waste products of low molecular weight can pass the dialysis membrane and will be supplied and removed by diffusion across the membrane. (b) In the monoculture experiment the silk fibroin scaffold is seeded with MSCs. In the coculture experiment the scaffold is seeded with MSCs and MCs. MSCs, mesenchymal stromal cells. with a MWCO of 20 kDa were sterilized using UV-light on both sides for 15 min each. The mini dialysis devices were carefully placed in the holder. All membranes were prewetted using sterile PBS which was pipetted in both the base and the insert of the wells (Figure 1). Later, the PBS was removed and replaced with the appropriate cell culture medium.
After 9 days, cells were detached using 0.25% trypsin-EDTA (25200, Thermo Fisher Scientific) and directly used for experiments at passage 4. A dynamic seeding process was used as previously described (Melke et al., 2020). Briefly, SF scaffolds were incubated with a cell suspension (3 × 10 5 cells/4 ml expansion medium) in 50-ml tubes placed on an orbital shaker at 150 rpm for 6 h in an incubator at 37°C and 5% CO 2 . Next, the samples were transferred to the dialysis wells plates and incubated at 37°C and 5% CO 2 for a total 4 weeks in osteogenic monoculture medium containing DMEM (low glucose, 22320, Thermo Scientific), 10% human platelet lysate ( | 1123 50 ng/ml macrophage colony stimulating factor (M-CSF, 300-25, PeproTech). The cells were counted and 1.9 × 10 6 monocytes were resuspended in 20 µl coculture medium and per scaffold added on top of the scaffolds that had previously been seeded with MSCs as described above (chapter 2.4). After 1.5 h incubation at 37°C and 5% CO 2 , the rest of the 180 µl medium was added to the insert. 500 µl medium was added to the base. Medium was changed according to Table 1. After 2 days, 50 ng/ml receptor activator of NFκ-B ligand (RANKL, 310-01, PeproTech) was added to the coculture medium and maintained for the rest of the culture. The culture was maintained for 28 days.

| Glucose assay
A glucose assay was used to measure the glucose concentration in the medium at day 7, 14, 21, and 28 in both the insert and the base (n = 3).
The glucose concentration was measured to ensure passage of glucose as a nutrient component over the dialysis membrane during the 28 day culture period. The method was adapted from Hulme and colleagues (Hulme et al., 2012). Briefly, on the day of assay, a buffer/chromophore reagent was prepared by mixing an equal volume (

| Lactate assay
A lactate assay was used to measure the lactate concentration in the medium at day 7, 14, 21, and 28 in both the insert and the base (n = 3).

| Statistical analysis
Statistical analyses were performed, and graphs were prepared in were equal and differences were tested with two-way ANOVA and  F I G U R E 3 Functionality of the dialysis membrane. (a, b) Glucose and (c, d) lactate measurements on culture medium of the monoculture cell experiment over 28 days performed in our dialysis system (n = 3). Measurements were taken in both the insert and the base of the system. At each timepoint the difference between insert and base is nonsignificant (p > 0.05, Welch's t-tests).

F I G U R E 4
Cell metabolic activity (presto blue) of (a) the monoculture and (b) coculture measured during 28-day cell experiments shown as relative fluorescence versus medium control (n = 4). At each timepoint the difference between the control and dialysis samples was determined, with significant differences for p < 0.5 (Mann-Whitney tests) indicated with an asterisk.

F I G U R E 5
Osteoblastic differentiation of MSCs in monoculture. (a) ALP activity divided by DNA (n = 4); significant differences are determined by two-way ANOVA with Tukey's post hoc test. All differences are significant (p < 0.05), unless indicated with "ns." (b) Collagen type 1 production by the cells visualized by immunostaining at Day 28. (c) Mineralization shown by Alizarin Red staining at Day 28. (d) RUNX-2 and osteopontin expression visualized by immunostaining at Day 28. ALP, alkaline phosphatase; MSCs, mesenchymal stromal cells.

| Functionality of the dialysis membrane: Cell metabolic activity
To ensure that no toxic components build up in the cell culture system, cell metabolic activity was followed over the culture time.
Cell metabolic activity was measured in both the mono-and coculture for the standard and dialysis groups. In the monoculture, the cell metabolic activity of the standard group compared to the dialysis group was similar for each timepoint with no significant differences (Figure 4a). In the coculture, dialysis seems to be beneficial compared to the standard as the cell metabolic activity of the dialysis group was significantly higher (co-standard vs. co-dialysis) on Day 7, 21, 28 (Figure 4b).
The results indicate that both in mono-and coculture, the cells were equally or at some timepoints more metabolically active in the dialysis group compared to the standard showing that the dialysis system could beneficially influence the cell metabolic activity.

| Osteoblastic differentiation of MSCs in monoculture
ALP activity is a widely used marker for indirectly quantifying osteoblastic activity (Remmers et al., 2021). ALP activity on the cells' surface was determined weekly in the monoculture over a 28 day time period and normalized against the amount of DNA.
The ALP activity was lowest in the negative control throughout all timepoints and was only slightly elevated at Day 28. As expected, the ALP activity in the standard and dialysis groups was increased and this difference was significant compared to the negative control starting from Day 14. There was a trend that the ALP activity was slightly higher in the dialysis group compared to the standard group with significant differences at Day 14 and 21

| Osteoclastic differentiation of MCs in coculture
TRAP activity is a widely used marker for indirectly quantifying osteoclastic activity (Remmers et al., 2021). TRAP activity was measured weekly in the coculture medium of both the base and the insert over a 28 day time period. Over time, TRAP activity in the base of both groups was stable and low (Figure 6a), most likely originating from the hPL (de Wildt et al., 2022a). In the insert TRAP activity Dialysis membranes are designed to be nonfouling. However, cell culture medium consists of nutritious liquids which are slightly viscous and sticky and could cause the membrane to clog over time (Poon, 2022). Also, cell mediated mineral deposition on the cell side of the dialysis membrane has been reported (Krishnan et al., 2010).
However, it was not reported whether this deposition influenced the function of the dialysis membrane. Our system was functional over the culture period of 28 days. Small molecules such as glucose (180 Da) and lactate (90 Da) were able to pass the membrane continuously over the whole culture time. Large molecules such as TRAP (30-35 kDa) were maintained in the cell culture compartment.
Moreover, the cells remained metabolically active, an indicator that the membrane still provided nutrients and removed waste products. The dialysis culture system was used to study osteogenic differentiation of MSCs and MCs into osteoblasts and osteoclasts.
It was hypothesized that the maintenance of auto-and paracrine factors in the culture medium would contribute to a more physiological microenvironment for the cells. The data reported here revealed higher osteoblastic and osteoclastic activity in the dialysis groups compared to the standard, shown by significantly higher ALP and TRAP activity, respectively. Therefore, the dialysis system Osteoclastic differentiation of MCs in coculture. (a) TRAP activity; significant differences (p < 0.05) are determined by comparing the inserts the of the standard to the dialysis group for each timepoint using Welch's tests. contributes to an excellent cell culture environment. Remarkably, this significant effect seen in the biochemical assays was not visible in the matrix production (collagen type 1 and mineralization). We hypothesize that this might be due to a difference in overall concentration of ascorbic acid and β-glycerophosphate. In the dialysis system, only the base medium is exchanged (500 µl) while in the standard system both the insert and base media are exchanged (200 + 500 µl), resulting in a lower overall concentration of these two factors. Ascorbic acid has been shown to increase the secretion of collagen type 1 and βglycerophosphate is the phosphate source that is needed for mineralization (Langenbach & Handschel, 2013). It is recommended to study this effect in future experiments.
To analyze in vitro models over time, nondestructive methods including medium analysis are desired (Owen & Reilly, 2018). A limitation of the current dialysis system is that the possibility to take medium samples from the insert is very limited. The principle of the system relies on the fact that the insert stays undisturbed. Also, the current design allows for only 200 µl of cell culture medium in the insert. Taking medium samples would lead to a necessity for adding fresh medium. Therefore, at each timepoint samples were killed to enable the performance of assays on the medium. Preferably we would have analyzed more markers, such as cathepsin K and carbonic anhydrase II to confirm OC differentiation (Bernhardt et al., 2017) To overcome this limitation, noninvasive assays or sensors could be used such as for example biosensing by particle mobility (BPM) (Yan et al., 2020) or aptamer based sensors (Xiao et al., 2007). Such sensors would allow for continuous monitoring of specific biomarkers without the need for medium sampling. A simpler solution could be the use of a larger volume culture medium in the insert. The amount of medium sampled should be very small relative to the total amount of medium in the insert, so that it would not lead to disturbance of the microenvironment. However, making the volume too large, could lead to loss of the effect of the cells' secretome on the microenvironment due to dilution (Vis et al., 2020).
A crucial process of in vitro bone tissue formation is ECM production and mineralization. The data presented in this study, mainly in the coculture, indicate limited mineralization (alizarin red) in both the standard as the dialysis system. Furthermore, this mineralization occurs primarily in the scaffold material and not in the matrix produced by the cells. There are two possible explanations for this observation. First, the presented experimental set-up cultures statically. It has been shown that for mineralized matrix production the cells prefer mechanical stimulation (Vetsch et al., 2017). Here, the MSCs were initially stimulated by dynamic seeding (Melke et al., 2020), but this effect was limited compared to what is usually seen in a dynamically loaded environment (Melke et al., 2018). In future experiments, the use of a bioreactor that can apply for example fluid flow induced shear stress would be desired to improve the mineralization (Melke et al., 2018). But implementing the dialysis system with its membrane will concomitantly affect the fluid flow.
Second, it is still a major challenge to use the right type of coculture medium as both the MSCs and the MCs should be stimulated to differentiate into osteoblasts and osteoclasts respectively. Ideally, no exogenous osteogenic and osteoclastic factors need to be added and the cells interact with each other in homeostasis. The current setup is limited since it starts from OB and OC precursors that first need to differentiate into mature cells, but it has recently been shown that this differentiation process might also happen without exogenous supplementation (Schulze et al., 2018). The coculture medium used in our study may not have offered the ideal balance yet, probably resulting in a less mineralized matrix compared to the monoculture.
MCs need a mineralized surface to attach to and to become osteoclasts. Recently, de Wildt and colleagues reported a premineralized SF scaffold that acts as a bone-mimetic template (de Wildt, van der Meijden, et al., 2022b). They used poly aspartic acid and simulated body fluid to premineralize silk fibroin scaffolds before cell seeding. The mineralized scaffolds supported both osteoclastic resorption and osteoblastic mineralization while in our study, the monocytes relied on MSCs first differentiating into osteoblasts and producing a mineralized matrix. Using the premineralized template in combination with a dialysis culture system could lead to an even faster method to generate a physiological bone remodeling model.
The proposed dialysis culture system is not limited to bone and could be beneficial for a wide variety of cell culture applications.
There is a huge variety of existing materials and MWCO options for the membrane, making it customizable for different needs. The MWCO can be chosen based on the cell culture medium ingredients and expected secretome. Caution however has to be taken to possible toxic elements in the manufactured dialysis membranes, such as glycerol, which has been shown to inhibit cell proliferation in several cell lines (Thermo Scientific, 2013) (Wiebe & Dinsdale, 1991).
In conclusion, we have demonstrated the feasibility of a simple to use dialysis cell culture system for bone tissue engineering applications. The dialysis system enables retention of the cells' secretome and thereby omits the extra effort that the cells have to make to restore their communication after culture medium exchange. The system creates a stable microenvironment for the cells to differentiate into the osteogenic and osteoclastic lineage. This simple culture system has the potential to be applied in other TE fields and is recommended to be used for differentiation of various cell types.

ACKNOWLEDGMENTS
We thank Jurgen Bulsink for the design and fabrication of the 48-well plate holder. We thank the ICSM studio for creating the illustrations in Figure 1. This work is part of the research program TTW with project number TTW 016.Vidi.188.021, which is (partly) financed by the Netherlands Organization for Scientific Research (NWO).