*2.4. Chondrogenic and Osteogenic Differentiation of hMSC in Biphasic Marine Scaffolds*

A sequential seeding procedure for the biphasic collagen scaffolds was performed (Figure 5). In the first step of sequential cultivation, hMSC were suspended in alginate solution, the jellyfish collagen phase of the biphasic scaffolds was infiltrated with this mixture with a cell density of <sup>6</sup> × <sup>10</sup><sup>6</sup> cells/cm3 and the constructs were cultivated with complete chondrogenic medium for 9 to 12 days. At the same time, hMSC from the same batch were seeded into flasks and cultivated in the presence of osteogenic medium for 9 to 12 days. After 9 to 12 days of pre-stimulation, the osteogenically induced cells were seeded into the mineralized salmon collagen layer of the biphasic scaffolds with an initial cell density of <sup>6</sup> × 105 cells/cm3. The constructs were cultivated until d21 from the initial seeding with osteochondral medium, containing 5 ng/mL TGF-β3, ITS, 10−<sup>7</sup> M Dex and 50 μg/mL AAP.

The biphasic constructs were stable during the whole cultivation period. MTT staining of viable cells after 1 and 9 days of cultivation demonstrated that the chondrogenically induced cells did not migrate out of the jellyfish collagen/alginate phase to the mineralized salmon collagen phase below (Figure 6). This was confirmed by confocal laser scanning microscopic (cSM) investigations at the area between the two phases (Figure 7).

**Figure 5.** Pre-differentiation and sequential seeding of hMSC onto biphasic scaffolds from jellyfish collagen and mineralized salmon collagen.

At the end of the cultivation, a contraction of the biphasic scaffolds was visible. This contraction has been observed before in monophasic collagen scaffolds and it has been shown to be diminished with the application of alginate as cell carrier. Histological sections of biphasic scaffolds after 21 days of cultivation showed the presence of cells in both phases (Figure 8A–C). Furthermore, toluidine blue staining of histological sections revealed the production of cartilage extracellular matrix in the chondrogenic part of the scaffold (Figure 8D–F). The cellularity of the constructs is considerably lower compared to pellet cultures of chondrogenically stimulated cells. However, we have shown in a previous study, that higher cell densities in porous collagen scaffolds did not increase extracellular matrix production per cell [9].

**Figure 6.** MTT Staining of viable cells in biphasic collagen scaffolds, freshly seeded with alginate-embedded hMSC (d1), after 9 days of chondrogenic differentiation (d9) and after 21 days of cultivation, seeded with osteogenically pre-differentiated hMSC at day 9. Scale bars represent 2 mm.

**Figure 7.** (**A**) cLSM images of a cross section of biphasic collagen scaffold seeded with hMSC after 9 days of chondrogenic differentiation; (**B**) transition area between jellyfish collagen (upper) and salmon collagen (lower) phase, after 9 days of chondrogenic differentiation before seeding of osteogenic cells, cytoskeleton stained with Alexa Fluor 488 phalloidin (green), nuclei stained with DAPI (blue). Scale bars represent 200 μm.

Chondrogenic differentiation of the cells in the jellyfish collagen/alginate phase was verified by gene expression analysis of collagen II, which was also detected at protein level (Figure 9).

Gene expression of collagen II increased from d1 to d12, but decreased slightly during the cultivation of the biphasic constructs in the osteochondral medium. Similar results were obtained for the quantification of collagen II by ELISA; however, there were always high variations between the different samples (Figure 9b). Additionally, the production of sulfated glycosaminoglycans increased during chondrogenic differentiation of jellyfish collagen/alginate embedded hMSC, but it did not further increase during cultivation in the osteochondral medium (Figure 9c).

Osteogenic differentiation of hMSC in the mineralized salmon collagen phase was demonstrated by ALP gene expression, which was strongest immediately after seeding of the biphasic scaffolds with osteogenically pre-differentiated hMSC. During further cultivation in osteochondral medium the ALP gene expression decreased (Figure 9a). Furthermore, gene expression of osteocalcin was analyzed in both scaffold parts, which was relatively low in all examined samples, however, cells in the chondrogenic layer showed down-regulation of osteocalcin (Figure 9).

**Figure 8.** Histological sections of biphasic scaffolds after 21 days of osteochondral cultivation: Haematoxylin/Eosin staining of (**A**) jellyfish collagen phase, (**B**) mineralized salmon collagen phase and (**C**) transition zone between jellyfish collagen part (**top**) and mineralized salmon collagen part (**bottom**). Deeply purple stained regions in (**B**,**C**) represent mineralized salmon collagen which is stained by haematoxylin too. Toluidine blue staining of (**D**) jellyfish collagen phase, (**E**) mineralized salmon collagen phase and (**F**) transition zone between jellyfish collagen part (**top**) and mineralized salmon collagen part (**bottom**).

**Figure 9.** (**a**) RT-PCR products of different osteogenic and chondrogenic marker genes jc = jellyfish collagen, sc = mineralized salmon collagen; (**b**) Collagen II content in the jellyfish collagen phase of biphasic scaffolds, detected by ELISA; (**c**) concentration of sulfated glycosaminoglycanes content in the jellyfish collagen phase of biphasic scaffolds. *n* = 3, mean +/− standard deviation.

#### **3. Discussion**

Biphasic, but monolithic scaffolds, exclusively from the marine biopolymers jellyfish collagen, biomimetically mineralized salmon collagen and alginate were fabricated for the first time. Due to the concerted freeze drying of the two phases before crosslinking the layers were tightly connected showing interconnecting pores through the area of the different layers. In the wet state, the two layers showed different swelling behavior. While the biomimetically mineralized salmon collagen phase did not swell, the volume of the fibrillized jellyfish collagen phase increased somewhat in wet state. The reason for the different swelling behavior could be the difference between the collagen types. Collagen II was shown to have higher swelling capacity compared to collagen I [17]. Furthermore, the presence of nanocrystalline hydroxyapatite connected to the collagen fibers of the salmon collagen could be responsible to the reduced swelling in this phase. Nevertheless, the different swelling behavior did not impair the union between the two phases, which have been cross-linked with higher EDC concentrations compared to the monophasic scaffolds. Despite the high concentration of carbodiimide for crosslinking to allow the tight union between the phases, we did not detect any negative effects on cytocompatibility. In a similar approach to our method mineralized and non-mineralized collagen (from equine origin) were combined to a triphasic scaffold for osteochondral regeneration [18]. In contrast to our study, crosslinking of the single layers was performed before freeze-drying, which required an additional knitting procedure to anchor the layers. Biphasic scaffolds from collagen I in the chondrogenic part and collagen I mineralized with Mg2+ substituted hydroxyapatite in the osteogenic part were recently prepared by Sartori and co-workers [19]. Also in this study, the collagen layers were cross-linked before freeze-drying. Nevertheless, the resulting biphasic scaffolds were stable enough to withstand subcutaneous implantation in mice for up to 8 weeks. In vitro investigations for MSC differentiation, however, were only performed with monophasic scaffolds in this study.

For the generation of osteochondral tissue constructs in vitro progenitor cells like human mesenchymal stroma cells need to be differentiated into two different lineages in spatially separated phases of the same scaffold. Optimally, scaffold structure and chemical composition provide the necessary stimuli for differentiation into the osteogenic and chondrogenic cell line. Jellyfish collagen from R. esculentum has a similar structure to human collagen II, since it consists of α-chain homotrimers and shows a degree of glycosylation similar to that of vertebrate collagen type II [20]. Chondrogenic differentiation is triggered by clustering of hMSC, which can be realized by 3D pellet formation or seeding of scaffolds with high cell densities [21]. Additionally, embedding of cells into alginate hydrogel, infiltrated into porous collagen scaffolds, induced a chondrogenic phenotype and increased collagen II expression of the cells [22]. Therefore, the upper (chondral) phase of the marine biphasic scaffolds was seeded with hMSC in high density (6 × 106 cells/mL) which were incorporated into alginate hydrogel completely filling the pores of the jellyfish collagen part of the scaffolds. Osteogenic differentiation is favored by the presence of hydroxyapatite, and it has been shown in porous collagen scaffolds in vitro, that specific ALP activity which is the main osteogenic marker increases with decreasing seeding density [23]. Similar results were also obtained in the present study for monophasic scaffolds from mineralized salmon collagen (Figure 3c). The lower phase of the biphasic marine scaffolds was therefore seeded with hMSC in a tenfold lower density compared to the chondral part, and the cells were allowed to attach directly at the pore walls of the mineralized salmon collagen without addition of alginate. Nevertheless, as shown in our previous studies with monophasic scaffolds from jellyfish collagen, scaffold structure and composition as well as cell density alone are not sufficient to induce chondrogenic differentiation of hMSC. Likewise, biomimetically mineralized collagen scaffolds, despite their bone-like composition of collagen and nanocrystalline hydroxyapatite, are not osteoinductive. The addition of osteogenic stimuli is necessary to induce osteogenic differentiation of MSC seeded in the scaffolds. For simultaneous osteogenic and chondrogenic differentiation of hMSC in the marine biphasic scaffolds in vitro, a suitable osteochondral medium needed to be developed. Cell culture media for the chondrogenic and osteogenic differentiation of MSC have equal and

distinct components [24]. While dexamethasone and ascorbate are included in both chondrogenic and osteogenic differentiation medium, both media have exclusive components (Table 1).

**Table 1.** Main differences in the composition of commonly used osteogenic and chondrogenic differentiation media (FCS = fetal calf serum, ITS = insulin, selen, transferrin mix, AAP = ascorbic acid-2-phosphate, Dex = dexamethasone, β-GP = β-glycerophosphate, TGF-β = transforming growth factor β).


To develop an osteochondral medium supporting both osteogenic and chondrogenic differentiation of MSC (additionally to the stimuli, which are exerted by the scaffold matrix) it is not sufficient just to combine the single media 1:1. In a previous study we have analyzed the impact of FCS on the chondrogenic differentiation of hMSC in monophasic scaffolds of jellyfish collagen [25] and demonstrated, that even small amounts of FCS (2%) in the culture medium significantly decrease the mRNA expression of chondrogenic markers. In addition, reduction of glucose content caused a decreased mRNA expression of chondrogenic markers as well as a decreased extracellular matrix production of the chondrogenically differentiated cells. In contrast, osteogenic differentiation of hMSC in mineralized salmon collagen scaffolds was not affected when the glucose concentration of the medium was increased (data not shown). Furthermore, osteogenic differentiation of hMSC is even favored by low serum conditions (1% and 5% compared to 10%) [26]. B-glycerophosphate which is an integral component of osteogenic differentiation media to provide a phosphate source for the mineralizing osteoblasts, is regarded as hypertrophy promoting reagent for chondrocytes [27]. The main component of chondrogenic differentiation medium, TGF-β, was shown to cause a downregulation of ALP expression [28], which was also found in our study for MSC seeded in scaffolds from mineralized salmon collagen (Figure 4). Based on these results we propose an osteochondral medium containing a reduced amount of TGF-β (5 ng/mL), no FCS, high glucose content, bovine serum albumin, as well as the factors stimulating both osteogenic and chondrogenic differentiation: dexamethasone and ascorbic acid-2-phosphate. However, even the combination of (1) different seeding densities in the osteogenic and chondrogenic part (2) different scaffold layer composition (3) alginate infiltration in the chondrogenic part preventing cell adhesion to the scaffolds pores and therefore supporting chondrogenic cell phenotype and (4) osteochondral medium was not sufficient to stimulate osteogenic and chondrogenic differentiation spatially separated in the respective scaffold parts. Similar results were obtained by Gupta et al. who tried to simultaneously differentiate rat MSC into chondrogenic and osteogenic lineage in gradient PLGA scaffolds with encapsulated chondroitin sulfate for chondrogenic priming as well as tricalcium phosphate for osteogenic priming [29]. The authors admittedly reported a better interaction of cells with the materials and a greater cellularity, but it was not possible to drive differentiation specifically into either of the planned directions. Caliari and Harley investigated the impact of scaffold local biophysical properties like mineral content and density on hMSC differentiation in the presence of mixed soluble signals for osteogenic and chondrogenic differentiation [30]. Unexpectedly, the authors observed an increased osteogenic response just in nonmineralized scaffolds with low density, which were intended to induce chondrogenic differentiation. Therefore, in the present study, a sequential seeding and pre-differentiation approach was developed. It has been shown before, that temporal stimulation of chondrogenic cells with TGF-β is sufficient to induce the chondrogenic phenotype in MSC. Buxton and co-workers stimulated hMSC in hydrogels with TGF-β1 and demonstrated, that total production of collagen II after three weeks of cultivation was not decreased in comparison to the controls, when TGF-β was withdrawn after 7 days of prestimulation [31]. Likewise Fensky and co-workers demonstrated, that a 10 day stimulation of hMSC embedded in a collagen I hydrogel is sufficient to induce upregulation of the chondrogenic marker genes for collagen II and aggrecan after three

weeks of cultivation [32]. Short delivery of high TGF-β doses (100 ng/mL) for 7 days on bovine MSC embedded in hyaluronic acid hydrogels was sufficient to induce and maintain the chondrogenic phenotype over a period of 9 weeks [33]. Chondrogenically pre-differentiated hMSC maintained their chondrogenic potential after embedding in methacrylated hyaluronic acid gels [34]. Osteogenic pre-differentiation of MSC before application in bone regeneration has successfully been applied by several groups. Peters et al. demonstrated that the injection of osteogenically pre-differentiated MSC enhanced healing of a critical bone defect in rats [35]. It is hypothesized that the application of osteogenically pre-differentiated instead of undifferentiated hMSC may prevent the transplanted cells from neoplasia and tumor formation [36]. Osteogenic pre-differentiation of hMSC has been successfully applied in our study. Osteogenically pre-differentiated hMSC seeded into monophasic scaffolds from mineralized salmon collagen showed significantly increased ALP activity both at the start of the 3D cultivation and after 14 days. Furthermore, the negative effect of TGF-β on ALP activity is less pronounced for osteogenically pre-differentiated cells compared to non-differentiated MSC, which is a further point to apply osteogenic pre-differentiation for osteochondral constructs. Osteocalcin gene expression, which is a further marker of osteogenic differentiation, was quite low in the biphasic scaffolds. This is in accordance to a fundamental study of Jaiswal and co-workers, demonstrating that vitamin D3 is necessary to induce adequate osteocalcin expression in osteogenically differentiated hMSC in vitro [37]. We refrained from adding vitamin D3 to the osteochondral medium to preserve the chondrogenic phenotype in the chondral layer. However, it might be beneficial to include vitamin D3 into the osteogenic pre-differentiation medium in future. In the present study, osteogenic and chondrogenic pre-differentiation was performed simultaneously. While chondrogenic pre-differentiation was realized already during cultivation of alginate-embedded hMSC in the jellyfish collagen part of the biphasic scaffold, osteogenic pre-differentiation of hMSC of the same batch was performed in the monolayer. Gene expression analysis revealed collagen II expression exclusively in the chondral part of the scaffolds; however, there was a slight decrease in collagen II expression during the cultivation with osteochondral medium in the presence of osteogenically pre-differentiated cells. Likewise, collagen II production on protein level and production of sGAG were somewhat decreased during co-culture (Figure 9). It has been demonstrated that continuous treatment of hMSC pellet cultures with TGF-β provided significantly higher production of extracellular matrix and expression of chondrogenic genes compared to short-time TGF-β supplementation of 3 and 10 days [38]. Possibly, the reduced TGF-β supplementation during osteochondral cultivation was not sufficient to stabilize the chondrogenic phenotype at the starting level. Similar observations were made for the osteogenic part. Highest ALP expression was detected immediately after seeding of the biphasic scaffolds, followed by a decrease of ALP expression during further osteochondral stimulation. Since ALP is an early osteogenic marker, the decrease could also relate to further differentiation along the osteogenic lineage. However, it has to be noted, that the applied osteochondral medium does not provide optimal conditions for osteogenic differentiation. Nevertheless, the RNA amount isolated from the mineral part of the biphasic scaffolds after 3 weeks of cultivation was not reduced compared to the start of 3D cultivation, suggesting that the osteogenically pre-differentiated cells survived the osteochondral conditions with serum deprivation. After three weeks of cultivation osteochondral constructs were obtained in vitro with spatial separated expression of chondrogenic ECM and osteogenic differentiation.

#### **4. Materials and Methods**
