Coculture with monocytes/macrophages modulates osteogenic differentiation of adipose‐derived mesenchymal stromal cells on poly(lactic‐co‐glycolic) acid/polycaprolactone scaffolds

Abstract The effects of immune cells, in particular macrophages, on the behaviour of mesenchymal stromal cells (MSCs) have recently gained much attention for MSCs‐based tissue‐engineered constructs. This study aimed to evaluate the effect of monocytes/macrophages on the osteogenic differentiation of adipose‐derived mesenchymal stromal cells (ADMSCs) in three‐dimensional (3D) cocultures. For this, we cocultured THP‐1 monocytes, M1 macrophages, or M2 macrophages with ADMSCs on 3D poly(lactic‐co‐glycolic) acid (PLGA)/polycaprolactone (PCL) scaffolds using osteogenic medium for up to 42 days. We found that osteogenic differentiation of ADMSCs was inhibited by monocytes and both macrophage subtypes in 3D scaffolds. Furthermore, coculture of monocytes/macrophages with ADMSCs resulted in downregulated secretion of oncostatin M (OSM) and bone morphogenetic protein 2 (BMP‐2) and inhibited expression of osteogenic markers alkaline phosphatase (ALP), bone sialoprotein (BSP), and runt‐related transcription factor 2 (RUNX2). Compared with both macrophage subtypes, monocytes inhibited osteogenic differentiation of ADMSCs more significantly. These data suggest that the mutual interactions between monocytes/macrophages and ADMSCs negatively affect MSC osteogenic differentiation and thus possibly bone healing capacity, which highlights the importance of the micro‐environment in influencing cell‐based constructs to treat bone defects and the potential to improve their performance by resolving the inflammation ahead of treatment.


| INTRODUCTION
Inflammation is the first stage of bone healing after bone injury. The state of inflammation has been indicated to affect the delicate balance between bone formation and bone degradation (Loi et al., 2016).
Monocytes and macrophages are vital modulators of inflammation (Nich et al., 2013) and display the transition of different phases in tissue regeneration (Wynn & Vannella, 2016). The crosstalk between monocytes/macrophages and cells involved in tissue regeneration, such as mesenchymal stromal cells (MSCs), is critical for normal tissue formation and healing (Guihard et al., 2012;Guihard et al., 2015;Vi et al., 2015). Upon injury, monocytes are recruited from the peripheral circulation and enter injured sites, where they differentiate into macrophages (Rickard & Young, 2009). The recruited macrophages respond to signals from the micro-environment in which they reside by acquiring different phenotypes (Wynn & Vannella, 2016). These macrophages are generally classified as either classically activated macrophages (M1) or alternatively activated macrophages (M2; Murray et al., 2014;Spiller, Freytes, & Vunjak-Novakovic, 2015). Based on current knowledge, M1 macrophages are responsible for angiogenesis and the removal of necrotic tissue at an early stage, whereas M2 macrophages are responsible for immune regulation, matrix deposition, and tissue remodelling at a later stage (C. Chen, Uludag, Wang, Rezansoff, & Jiang, 2012).
Recent studies reported a switch of macrophage subtypes from proinflammatory M1 macrophages to pro-wound healing M2 macrophages during the bone healing process (Tasso et al., 2013;Wu et al., 2015), demonstrating the crucial role of monocytes and different macrophage subtypes in bone healing.
To further elucidate the interaction of different macrophage subtypes with bone forming cells, such as MSCs, in vitro, our group has previously established a two-dimensional (2D) coculture system where different types of macrophages were cocultured with adipose-derived mesenchymal stromal cells (ADMSCs; Zhang et al., 2017). This study demonstrated that M2 macrophages, rather than M1 macrophages, can promote the osteogenic differentiation of ADMSCs. Although culturing cells on 2D substrates has been considered a standard technique for in vitro cell culture, it is recognized that cells more closely mimic native tissues when cultured in a three-dimensional (3D) environment. In 3D cell cultures, cells adhere to each other via the extracellular matrix (ECM) and form specific cell-cell contacts, which differentially regulate cell growth, migration, and differentiation (Lee, Cuddihy, & Kotov, 2008). This is supported by findings of significant divergence of cell-cell interactions for cells in 2D and 3D culture systems in previous studies (D. Y. Chen et al., 2013;Valles et al., 2015).
Furthermore, 3D scaffolds are widely used for tissue engineering applications. The most widely used materials for tissue engineering are polymeric materials because they are easily processable, biocompatible, and biodegradable and can be modified with desired properties (e.g., dimensions and porosity; Ceccarelli et al., 2013). In recent years, polymers have been processed via electrospinning to fabricate nanofibres for different applications in skin (Duan et al., 2006), blood vessel (Vaz, van Tuijl, Bouten, & Baaijens, 2005), and bone tissue regeneration (Zhang et al., 2008). Electrospun fibres represent morphological similarity to natural ECM, which makes them attractive for cells to proliferate and function effectively (Yang, Yang, Wang, Both, & Jansen, 2013). The interfibre pores that are obtained within electrospun fibre meshes render such scaffolds highly interactive with its surrounding tissue due to the high specific surface area (Holzwarth & Ma, 2011). To make full use of the functionality of multiple polymer types in one electrospun mesh, the blend electrospun method, which allows the simultaneous combination of multiple polymers during the electrospinning process, has gained much attention (Hiep & Lee, 2010). An attractive polymer combination for electrospun meshes includes poly(lactic-co-glycolic) acid (PLGA), which is suitable for cell adhesion and proliferation due to its hydrophilic properties, and polycaprolactone (PCL), which is a flexible biopolymer that can be used to overcome the brittle and low elongation properties of PLGA (Kim & Cho, 2009 (Hofmann et al., 2008;Levenberg et al., 2005) and MSCs for osteogenic differentiation Ma et al., 2014). As a nutritional supplement, fetal bovine serum (FBS) has been commonly used for multiple cell types. However, the major drawback of this supplement is the possibility to trigger an immunological response due to the presence of xenogeneic antigens (Bieback et al., 2009). Consequently, it has been postulated that the use of FBS should be avoided for human cell cultures (Ma et al., 2015). In contrast, platelet lysate (PL) is of human origin, can be applied as an autologous nutritional supplement for primary cells, and contains various growth factors and cytokines, including platelet-derived growth factor (PDGF), basic fibroblast growth factor (bFGF), insulin-like growth factor 1 (IGF-1), and transforming growth factor β (TGF-β; Doucet et al., 2005). A vast amount of scientific literature has reported on the capacity of PL to promote the proliferation and differentiation of MSCs into different lineages (Altaie, Owston, & Jones, 2016;Fekete et al., 2012;Shanskii et al., 2013). In particular, PL has been demonstrated to be an optimal serum supplement to culture ADMSCs for bone regeneration (Hayrapetyan, Bongio, Leeuwenburgh, Jansen, & van den Beucken, 2016;Ma et al., 2015).
Furthermore, PL was also used in tissue-engineered scaffolds to benefit the innate immune response for superior tissue regeneration. It was found that PL can induce an anti-inflammatory response of monocytes/ macrophages (Linke et al., 2017). These findings suggest the potential of using PL to culture human cells for the clinical usage.
The objective of this study was to evaluate the effect of monocytes and macrophage subtypes on osteogenic differentiation of ADMSCs cultured on 3D PLGA/PCL scaffolds using a direct coculture model. We hypothesized that monocytes and macrophage subtypes would differentially affect the osteogenic differentiation of ADMSCs compared with ADMSCs monocultures on PLGA/PCL scaffolds.
All cell culture flasks, dishes, and plates were purchased from Greiner Bio-One (Frickenhausen, Germany).

| Isolation, preculture, and characterization of ADMSCs
ADMSCs isolation was performed as described previously (Varma et al., 2007). Briefly, human subcutaneous adipose tissue was obtained from the Department of Plastic Surgery (Radboudumc, Nijmegen, the Netherlands) after ethical approval (CMO Radboudumc; dossier# 2017-3252) and written informed consent. Resected fat tissue was minced using surgical scalpels and scissors and washed with phosphate-buffered saline (PBS) for three times. The tissue was digested with 0.1% collagenase A in PBS containing 1% BSA at 37°C for 60 min with intermittent shaking. The digested tissue was centrifuged for 10 min at 600 g, and the cell pellet was resuspended in 5 ml of PBS/1% BSA and filtered with a 100-μm nylon mesh (Roche Diagnostics). Cells were then subjected to a Ficoll density centrifugation (Lymphoprep™, 1,000 g, 20 min; Axis-Shield, Oslo, Norway) step to remove erythrocytes and were seeded at a density of 1 × 10 5 cells/cm 2 in αMEM containing 10% FBS, 100-U/ml penicillin, 100-μg/ml streptomycin, and 1-ng/ml bFGF, and cultured in a humidified atmosphere of 5% CO 2 at 37°C. Medium was changed twice a week. When near confluent (90%), cells were detached with 0.5-mM EDTA/0.05% trypsin and passaged or frozen in 1 × 10 6 cells/ml aliquots in liquid nitrogen. ADMSCs from passages 3 to 5 were used in further experiments.
The expression of surface antigens was evaluated by incubating ADMSCs at 4°C for 1 hr with the respective antibodies in 100-μl FACS buffer (1-mM EDTA in PBS with 0.5% BSA; Sigma). The following antibodies were used for evaluation: FITC mouse anti-human CD45, APC mouse anti-human CD73, PerCP-Cy 5.5 mouse anti-human CD90, and PE mouse anti-human CD105 (all from BD Pharmingen, Piscataway, USA). Cells without antibodies were used as negative controls. Labelled cells were washed twice in 1-ml FACS buffer and analysed with the FACSAria II flow cytometer (BD biosciences, San Jose, CA, USA). Data were processed using Flowing software 2.5.1 (University of Turku, Turku, Finland), and the percentage population of each antibody that stained positively for the respective markers was compared with negative controls.
M1 and M2 phenotypes were characterized by measuring the concentration of TNF-α and TGF-β in collected supernatants from polarized macrophages using the respective ELISA kits. ELISA kits were used according to the manufacturer's instructions. M1 and M2 phenotypes were further characterized by immunostaining for the M1 macrophage marker CCR7 and M2 macrophage marker CD36 (Stewart, Yang, Makowski, & Troester, 2012). The cells were fixed with 4% paraformaldehyde for 15 min and then blocked with incubation buffer (1% BSA in PBS) for 1 hr at room temperature. Samples were incubated with the primary antibodies rabbit anti-human CCR7 (1:500) and

| Preparation of PL
PL was prepared as described previously (Prins et al., 2009). Briefly, pooled platelet products containing approximately 1 × 10 9 thrombocytes/ml were purchased from the Sanquin Blood Bank (Nijmegen, the Netherlands). The product was divided into 5-ml aliquots in 15-ml tubes (Greiner Bio-One), subjected to one freeze/thaw (−80°C/37°C) cycle and stored at −80°C until use. Before adding to the medium, PL was thawed and centrifuged at 2,000 g for 10 min to remove remaining platelet fragments.
The 3D scaffolds were fabricated using a so-called wetelectrospinning technique in a commercially available electrospinning set-up (Esprayer ES-2000S, Fuence, Tokyo, Japan). The optimal processing parameters for stable formation of electrospun fibres were selected based on an earlier publication (Yang et al., 2013). Briefly, the prepared polymer solution was fed into a glass syringe and delivered to an 18G nozzle at a feeding rate of 50 μl/min. A high voltage (20-25 kV) was applied at the nozzle to generate a stable polymer jet by overcoming the surface tension of the polymer solution. A grounded bath filled with 100% ethanol located 15 cm under the nozzle was used to collect the fibres. To control the size of resulting fibre meshes, the process was stopped every 15 min for fibre mesh collection. Subsequently, the wet-electrospun scaffolds were washed thoroughly in MilliQ and freeze-dried (VirTis BenchTop Pro with Omnitronics Freeze Dryer, SP Scientific, NY, USA) for 3 days. The obtained scaffold displayed an uncompressed structure with an average fibre diameter of 1.98 ± 0.51 μm and a porosity of 99% (Yang et al., 2013).
Disk-shaped scaffolds with a diameter of 6 mm and a thickness of about 2 mm were punched out using a biopsy punch (Kai medical, Gifu, Japan) from each wet-electrospun mesh and subsequently sterilized in 70% ethanol for 2 hr and soaked in proliferation medium overnight.
2.6 | Analyses 2.6.1 | Cell loading efficiency and DNA content To evaluate the cell loading efficiency for each group, DNA content of the loaded cells was measured after 24 hr of seeding. Scaffolds with cells were also collected on Days 3, 7, 14, and 28 for DNA content measurement (n = 3). Samples were washed twice with PBS, transferred to 1.5-ml Eppendorf tubes, and digested with 0.1% collagenase A in PBS and 1% BSA for 16 hr at 37°C with intermittent shaking. The digested samples were centrifuged for 5 min at 2,000 g. The supernatant was aspirated, and 1 ml of MilliQ was added to each tube after which repetitive freezing (−80°C) and thawing (room temperature) cycles were performed; 5 × 10 5 ADMSCs with or without 5 × 10 5 of another type of cells were suspended in 1 ml of MilliQ, which was regarded as 100% control. DNA content was quantified using the Loading efficiency in each group was calculated through division of the result by the respective 100% control.

| ALP activity
Gross alkaline phosphatase (ALP) activity was measured using the same samples as described for the DNA content measurement. For the assay,

| Characterization of M1 and M2 macrophages
Human THP-1 monocytic cells were induced into M0 macrophages via activation with PMA and further differentiated into M1 or M2 macrophages using LPS/IFN-γ or IL-4/IL-13, respectively. Activation with PMA changed the THP-1 cells from cells growing in suspension to adherent cells. Morphologically, M1 macrophages showed a more spindle-like shape compared with M2 macrophages (data not shown).
Immunostaining for macrophage phenotypes showed a mixture of

| Cell seeding efficiency and DNA content
Twenty-four hours after cell seeding, the DNA content of the adherent cells for each experimental group was measured to assess the cell seeding efficiency (Figure 3a)

| ALP activity
ADMSCs monoculture and M1-ADMSCs and M2-ADMSCs cocultures showed a similar trend for ALP activity over the culture period, that is, a rise during the early stage and a decline in the later stage

| Mineralization
Mineralization levels for all experimental groups were relatively low on Days 14 and 28, without significant differences among groups (p > 0.05; Figure 3d). On Day 42, the calcium content for the ADMSCs monoculture was significantly higher compared with all other groups (p < 0.001).

| Real-time qPCR
Osteogenic differentiation of ADMSCs was inhibited upon coculture with monocytes or macrophage subtypes along with the decrease in the expression of osteogenesis-related genes, which was examined by real-time PCR analysis ( Figure 5). For ALP gene expression, no significant differences among the experimental groups at Day 3 were observed ( Figure 5a). Significantly, higher ALP gene expression was observed for the ADMSCs monoculture compared with the three coculture groups at Days 7 (p < 0.001), 14 (p < 0.05), and 28 (p < 0.01). For BSP gene expression, significantly higher expression was observed for the ADMSCs monoculture compared with the three coculture groups at Days 3 (p < 0.01), 7 (p < 0.05), 14 (p < 0.01), and 28 (p < 0.001; Figure 5b). No significant differences between ADMSC and the coculture groups were observed for  Regarding interaction between MSCs and monocytes/ macrophages, several previous studies cocultured macrophages with MSCs and reported diverse effects (i.e., stimulatory or inhibitory) on the osteogenic differentiation of MSCs (C. Chen et al., 2012;Z. Chen et al., 2014;Fernandes et al., 2013;Guihard et al., 2012). This variation can be attributed to multiple factors, including the source of stem cells, utilized polarization protocols for macrophages, and cell ratios.
Therefore, the exact role of monocytes/macrophages on osteogenic differentiation of MSCs requires a more comprehensive and more accurate research set-up. Our group has developed a delicate, indirect 2D coculture system using transwells and the THP-1 cell line as monocyte source and showed that different types of macrophages differentially affected the behaviour of cocultured ADMSCs (Zhang et al., 2017). To be more specific, M2 macrophages, rather than M1 macrophages, promoted the mineralization of ADMSCs and proved that this is mediated through paracrine signalling pathways. However, given the cell behavioural difference in 3D and 2D culture systems and the fact that 3D scaffolds are a crucial part of cell-based bone constructs, we here established a direct 3D coculture system by using human primary ADMSCs, THP-1 cells and electrospun scaffolds. In contrast to the previously observed stimulatory effects of macrophages on the osteogenic differentiation of ADMSCs (Zhang et al., 2017), monocytes, M1 macrophages, and M2 macrophages significantly inhibited the osteogenic differentiation of cocultured ADMSCs. This is evidenced by decreased ALP activity, mineralization content, and and 3D culture models (D. Y. Chen et al., 2013;Valles et al., 2015).
With regard to the mechanism behind this inhibitory effect of monocytes and macrophages, pro-inflammatory and anti-inflammatory cytokines were measured during the culture time, but no significant differences in inflammatory cytokine concentrations in the coculture medium were detected. However, a significantly decreased OSM and BMP-2 secretion was found in cocultures compared with the monoculture. This indicated that monocytes/macrophages inhibited certain osteogenic signalling pathways in our 3D coculture system. The observed inhibitory effect may come from other cytokines than inflammatory cytokines such as TNF-α and TGF-β secreted by monocytes/macrophages (C. Chen et al., 2012) or the direct interaction between monocytes/macrophages, scaffold, and ADMSCs. Furthermore, PL contains a high concentration of TGF-β as reported previously, with fluctuations from 900 to 15,000 pg/ml (Fekete et al., 2012;Renn et al., 2015;Salvade et al., 2010). In our research, the concentration of TGF-β in blank controls (i.e., culture medium incubated with scaffold only) was 2,426.6 ± 185.9 pg/ml (data not shown). These high values led to negative values after correction for blank controls.
This observation suggests that the cells consume a large amount of TGF-β in the process of proliferation and differentiation. Alternatively, the low concentrations of TNF-α might imply the transition of M1 macrophages into M2 macrophages after 3 days of coculture with ADMSCs (Yin, Pang, Bai, Zhang, & Geng, 2016), which is a contradictory finding to an earlier publication suggesting that PL-treated MSCs support the maintenance of macrophages in a pro-inflammatory condition (Ulivi, Tasso, Cancedda, & Descalzi, 2014 There are also several limitations to our study. First, to address multivariate research questions that require large numbers of cells and for reproducibility of results, the THP-1 cell line rather than primary human monocytes and macrophages was used. Although THP-1 cells have been reported to retain all necessary markers and morphologic features of primary monocytes (Auwerx, 1991;Qin, 2012), further studies using macrophages derived from primary monocytes of different donors are desired. Second, due to the lack of exclusive markers for M1 and M2 macrophages, we cannot dynamically monitor the macrophage behaviour during the culture time. A delicate staining method to follow the fate of macrophages and ADMSCs and to explore the cell-cell interaction would greatly aid in elucidating the mechanism of the observed inhibitory effects of monocytes/ macrophages in vitro.
In conclusion, this study used cocultures of monocytes/ macrophages and ADMSCs on 3D PLGA/PCL scaffolds to evaluate effects of cell-cell interactions on the osteogenic differentiation of ADMSCs. We found that monocytes and macrophage subtypes inhibit the osteogenic differentiation of ADMSCs on 3D PLGA/PCL scaffolds. Cocultured monocytes/macrophages decreased the expression of osteogenic markers ALP, BSP, and RUNX2. These data highlight the ignored fact that inflammation may regulate osteoblast activity of MSC-based bone constructs within the bone microenvironment. It implies that strict control of inflammation may be necessary to create an anabolic environment that improves the performance of cell-based bone constructs. Additionally, compared with macrophage subtypes, monocytes played a stronger inhibiting role on the osteogenic differentiation of ADMSCs. Therefore, it seems that the transient activation of monocytes after fracture injury is important for fracture repair.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.