Transcriptome Analysis of MSC and MSC-Derived Osteoblasts on Resomer® LT706 and PCL: Impact of Biomaterial Substrate on Osteogenic Differentiation

Background Mesenchymal stem cells (MSC) represent a particularly attractive cell type for bone tissue engineering because of their ex vivo expansion potential and multipotent differentiation capacity. MSC are readily differentiated towards mature osteoblasts with well-established protocols. However, tissue engineering frequently involves three-dimensional scaffolds which (i) allow for cell adhesion in a spatial environment and (ii) meet application-specific criteria, such as stiffness, degradability and biocompatibility. Methodology/Principal Findings In the present study, we analysed two synthetic, long-term degradable polymers for their impact on MSC-based bone tissue engineering: PLLA-co-TMC (Resomer® LT706) and poly(ε-caprolactone) (PCL). Both polymers enhance the osteogenic differentiation compared to tissue culture polystyrene (TCPS) as determined by Alizarin red stainings, scanning electron microscopy, PCR and whole genome expression analysis. Resomer® LT706 and PCL differ in their influence on gene expression, with Resomer® LT706 being more potent in supporting osteogenic differentiation of MSC. The major trigger on the osteogenic fate, however, is from osteogenic induction medium. Conclusion This study demonstrates an enhanced osteogenic differentiation of MSC on Resomer® LT706 and PCL compared to TCPS. MSC cultured on Resomer® LT706 showed higher numbers of genes involved in skeletal development and bone formation. This identifies Resomer® LT706 as particularly attractive scaffold material for bone tissue engineering.


Introduction
Human multipotent mesenchymal stem cells (MSC) are multipotent stem cells and represent a particularly attractive source for tissue engineering, because they are readily isolated and expanded and can differentiate into several mature cell types, such as adipocytes, chondrocytes, osteoblasts and myocytes [1,2 and references therein]. Due to their mesenchymal origin and their osteogenic differentiation capacity, MSC are particularly promising cells for bone replacement. Using a specific culture medium, they can differentiate towards osteoblasts in vitro within three weeks [3]. However, a biomaterial scaffold is required for bone tissue engineering (BTE) to allow for immobilisation of cells in a spatial structure. A variety of potential biomaterials exist, which are cytocompatible for MSC, e. g. collagen, fibrin and non-oxide ceramics [4][5][6][7][8]. For scaffolds in general, cyto-and biocompatibility are the criteria of upmost importance and they need to meet further specific application-dependant criteria. For BTE, a scaffold should be osteoinductive or at least osteoconductive and has to be long-term degradable to allow for an autologous replacement of the transplanted area. Besides such specific requirements, a fundamental bottleneck must be overcome to use MSC for BTE -the adequate supply of the cells. This problem will become more critical when aiming at engineering of bulk tissue to fill bone defects after injury or elimination of tumors, particularly when autologous bone production is desired. Such goals necessitate the maintainance of large numbers of undifferentiated cells embedded in biocompatible matrices to provide sufficient starting material.
A number of commercially available cell culture materials is accessible, such as standard tissue culture poly(styrene) (TCPS), Primaria TM , poly(ethylene terephthalate) (PET) and TC plastics biocoated with Matrigel TM or single extracellular matrix proteins, but given properties of the polymers make them ineligible for tissue engineering. Further, frequently the animal origin of biocoating products disqualify them from use in human clinical applications. Currently, the published literature contains a large number of studies in which stem cells are seeded on one or more readily available materials but frequently such studies lack systematic approaches to determine cell responses.
In our recent work we established a biomaterial test platform to assess the compatibility of stem cells and biomaterials for tissue engineering in a highly standardised and systematic manner. The assessment of stem cell/biomaterial interactions is multifactorial and requires a stringent analysis of parameters, such as material surface and bulk properties, cytotoxicity, cell adhesion, cell morphology, viability, proliferation, necrosis and apoptosis [5]. In our present study, we analysed in more detail the interactions of MSC with two commercially available synthetic and resorbable polymers of our panel of biomaterials, Poly(e-caprolactone) (PCL) and poly[(L,L-lactide-co-trimethylene carbonate) 7/3 ] ( = PLLA-co-TMC = ResomerH LT706).
PCL is a linear, semicrystalline, synthetic aliphatic homopolyester, which has a degradation time of two to three years [9][10][11]. PCL is the most studied degradable polymer of the polyester family [12] and is already approved by the Food and Drug Administration for diverse applications in the human body. Furthermore, PCL is currently under consideration for use in bone tissue engineering [13][14][15][16].
In contrast, there is only poor knowledge on ResomerH LT706. This linear and semi-crystalline poly-lactide-based polymer is similar in chemistry to Suprathel, which is now used in clinical trials for skin replacement after burns [17,18]. Further, poly(1,3trimethylene carbonate) copolymerised with D,L-lactide was shown to be long-term biodegradable and biocompatible for soft tissue engineering [19], but so far, polymers related to ResomerH LT706 were not analysed with respect to their usefulness for bone replacement. The random copolymer PLLA-co-TMC (ResomerH LT706) is characterised by an intermediate-term degradation, slower than poly(glycolic acid) (PGA), poly(D,L-lactide acid) (PDLLA), and poly(glycolic acid-co-trimethylene carbonate) (PGA-co-TMC), but faster than the long-term degradable PCL [20][21][22]. Although, ResomerH LT706 has some advantages regarding its mechanical properties and processing behavior, it is used only in a few studies as potential scaffold for hard and soft tissue engineering [23,24].
In the present study, we analysed the two polymers in comparison to TCPS for their osteoinductive capacity and thus for their potential use in BTE. Therefore, we seeded MSC on PCL and ResomerH LT706 samples with defined, ultraflat topography to exclude topography-dependant changes in cell adhesion, morphology and proliferation. MSC adhered on both polymers and showed only minor differences in morphology and viability. When subjected to osteogenic induction medium (OIM), MSC differentiated towards the osteoblastic fate. Alizarin red stainings, realtime PCR, scanning electron microscopy (SEM) and energydispersive X-ray spectroscopy (EDS) analysis revealed a similar frequency of MSC-derived osteoblasts on the two polymers. However, detailed studies on the molecular level including whole genome expression analysis unravelled differences in the biomaterial-based propensity towards the osteogenic fate of MSC supported by these two polymers.
Poly(e-caprolactone) (PCL) PCL with a molecular weight of 80.000 g/mol was purchased from Sigma-Aldrich GmbH (Germany). For PCL foils, 3 g of granules were used. These granules were placed on TeflonHcovered metal plates, the temperature was raised to 85uC and maintained for 5 min. A load of 1000 kg was applied for 1 min at 85uC. After cooling down to room temperature, foils were then washed several times with isopropanol (Fluka, Germany), aqueous 0.02 mM Tween 80 (Roth, Germany) and 8 M urea (Roth, Germany) and then rinsed vigorously with demineralised water. PCL samples were then dried in a vacuum oven for 24 h at 40uC. Samples were placed in TCPS wells, stored at 4uC, and protected from light. All steps were carried out under sterile conditions.

ResomerH LT706
Resomer type poly(L-lactic acid-co-trimethylene carbonate) ( = P(LLA-co-TMC; lactic acid-trimethylene carbonate ratio 70:30, LT706, 1.2-1.6 dL/g) was purchased from Boehringer Ingelheim Pharma GmbH & Co. KG (Germany). Foils were prepared by melt-pressing technique. Therefore, granules were ground to powder in a cryo-mill and 1.2 g of the powder were placed between TeflonH-covered metal plates. This was followed by a 5 min incubation at 180uC. A load of 1000 kg was applied for 9 min. The P(LLA-co-TMC) foil was allowed to cool down to room temperature, removed from the metal plates and further processed as for as for PCL.

Coating of biomaterials with radiolabeled fibronectin and vitronectin
100 ml of either fibronectin or vitronectin (100 mg diluted in 100 ml PBS; BD, Heidelberg, Germany) were mixed in a silanised counter vial (3.5 ml volume) with 1 ml Na 125 I solution (0.1 mCi; 3.7 MBq; 45 pmol; Amersham Europe, Freiburg, Germany) and 5 ml Chloramine-T solution (10 mg Chloramine-T trihydrate, solved in 100 ml PBS) using a magnetic stir bar. After 5 min, the incorporation of 125 I into fibronectin and vitronectin was measured by TCA precipitation. Therefore, a melted glass capillary was immerged first into the iodinising mixture and then into 200 ml of BSA solution (10% (m/m) BSA, 1% (m/m) NaI, 0.01% (m/m) NaN 3 ). After vigorously mixing of the solution, an aliquot of 10 ml was admixed with another 200 ml BSA solution. Afterwards, 2 ml ice-cold 10% trichloroacetic acid was added. The precipitated material consisting of BSA and target protein was spinned down, separated from the supernatant and measured using a gamma counter COBRA II Auto-Gamma (Packard, Dreieich, Germany). To separate labelled protein from nonincorporated iodine, the mixture was purified by gel filtration using a SephadexH G-25 NAP TM 5 Column (Pharmacia Biotech AB, Uppsala, Sweden). The column was equilibrated with PBS before use. The identified radiolabelled proteins were stored at 220uC. Radiolabeled fibronectin and vitronectin were dissolved in PBS and DMEM with 10% FCS in a concentration of 10 mg/ ml.
Radiolabeled protein adsorption study was carried out by incubation of TCPS, ResomerH LT706 and PCL disc surfaces with 2 ml of the different solutions, each with 125 I-labeled fibronectin or vitronectin (10 mg/ml), respectively. The activity was adjusted to 100.000 cpm/500 ml (1.67 kBq/500 ml; 45.0 pCi/ 500 ml) using a tracer. To test the resulting activity, samples of 500 ml volume were measured using the gamma counter COBRA II Auto-Gamma during a period of 3 min. After treatment of biomaterial substrates with radiolabelled protein (incubation time 1 h) and subsequent washing, activity was measured using the gamma counter during a period of 5 min.
Significant differences between samples were analysed using student's t-test for 10 independent measurements of each coating.

Isolation of human mesenchymal stem cells
Human mesenchymal stem cells (MSC) were isolated from femoral head spongiosa of patients with total hip joint endoprosthesis (TEP) after written consent was obtained from the patients. The study was approved by the Ethics Committee of the University Clinics, Aachen, Germany. MSC were characterised as previously described [5][6][7][8]25; Fig. S2]. In brief, after rinsing the spongiosa with stem cell medium, spongiosa was removed and the remaining cell suspension was centrifuged for 10 min at 5006g. Thereafter, the cell pellet was resuspended in stem cell medium and cells were seeded in a T75 culture flask and cultured at 37uC in a 21% O 2 and 5% CO 2 humidified atmosphere. After 24 hours, non adherent (hematopoietic) cells were removed by medium change. Mesenchymal stem cells were expanded in growth medium (PAN Biotech, Aidenbach, Germany) consisting of 60% DMEM low glucose, 40% MCDB-201, 2% FCS, 16ITS-plus (insulin-transferrin-selenic acid+BSA-linoleic acid), 1 nM dexamethasone, 100 mM ascorbic-acid-2-phosphate, and 10 ng/ml EGF. Medium was replenished every 3-4 days. At 80-90% confluency, stem cells were trypsinised with stem cell trypsin (CellSystems, St. Katharinen, Germany) and reseeded in a density of 5,000 cells/cm 2 for optimal proliferation.

Differentiation of MSC towards osteoblasts
For osteogenic differentiation, MSC were seeded in a density of 3.1610 4 cells/cm 2 on TCPS and on foils of the two polymers PCL and ResomerH LT706. Twenty-four hours after seeding, the growth medium was replaced with osteogenic induction medium (OIM) consisting of DMEM low glucose (PAA, Coelbe, Germany), 100 nM dexamethasone, 10 mM sodium b-glycerophosphate, 0.05 mM L-ascorbic-acid-2-phosphate (all Sigma, Steinheim, Germany) and 10% FCS (PAN Biotech, Aidenbach, Germany). Medium was changed every 2-3 days. After 21 days of osteogenic differentiation, cells were fixed with 70% ethanol for 1 hour, washed three times with demineralised water and then stained with an Alizarin red solution (40 mM, pH 4.1, Sigma) for 10 minutes. Finally, cells were washed three times with PBS (phosphate-buffered saline).
For quantification, the Alizarin red precipitates were solubilised. Briefly, stained samples were incubated with 800 ml acetic acid (10%) for 30 min. Then, supernatant was transferred into a 1.5 ml tube and boiled for 10 min at 85uC, followed by a 5 min incubation on ice. After centrifugation (15 min, 15,0006g), 500 ml of the supernatant were transferred into another 1.5 ml tube and mixed with 200 ml of 10% ammonium hydroxide. Samples of 150 ml were transferred into a 96 well microtiter plate and optical density was measured at 405 nm using a standard ELISA reader.
P-value to detect statistically relevant differences for the different biomaterials was calculated with student's t-test (n = 3 with two replicates each).

Scanning electron microscopy (SEM)
MSC/polymer biohybrids were fixed in 3% glutaraldehyde for at least 24 hours, rinsed with sodium phosphate buffer (0.2 M, pH 7.39, MERCK, Darmstadt, Germany) and dehydrated by incubating consecutively in 30%, 50%, 70% and 90% acetone and then three times in 100% acetone for 10 minutes. The biohybrids were critical-point-dried in liquid CO 2 , and then sputter-coated with a 30 nm gold layer. Samples were analysed using an environmental scanning electron microscope (ESEM XL 30 FEG, FEI, PHILIPS, Eindhoven, The Netherlands) in a high vacuum environment.

Energy-Dispersive X-ray Spectroscopy
Energy dispersive X-ray spectroscopy (EDS) is an analytical technique used for element analysis samples. EDS was performed on the XL 30 FEG scanning electron microscope (FEI, Eindhoven, The Netherlands) using an EDAX Falcon Genesis Spectrum 5.21 energy-dispersive X-ray spectroscopy system with an ultrathin window liquid nitrogen cooled Si(Li) X-ray detector (EDAX Inc. Mahwah, NJ, USA). For the EDS an accelerating voltage of 20 kV was used.

Reverse Transcriptase Polymerase Chain Reaction (RT-PCR)
Total RNA was isolated using the RNeasy mini Kit according to the manufacturers' instructions (Qiagen, Hilden, Germany). Reverse transcription was done with 1 mg of total RNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Darmstadt, Germany). PCR was as follows: denaturation at 95uC for 1 min, annealing at 58uC (osteocalcin), 60uC (alkaline phosphatase) for 1 min, extension at 72uC for 1 min (30 cycles), and a final extension at 72uC for 10 min. PCR products were analysed by electrophoresis using a 2% agarose gel and visualised with ethidium bromide. Primer sequences were used as listed in Table 1.

RealTime PCR (RT-qPCR)
RealTime PCR was performed using SybrGreen and Taqman technology. Briefly, 10 ml SybrGreen Master Mix (Applied Biosystems, Darmstadt, Germany) was mixed with 1 ml (10 pg) Primer forward, 1 ml (10 pg) Primer reverse, 6.8 ml water and 1.2 ml (60 ng) template. Then the samples were subjected to the following program: initial denaturation at 95uC for 10 min, followed by 40 cycles of denaturation at 95uC for 15 sec and annealing/extension at 60uC for 1 min. Primer sequences and sizes of amplicons are listed in table 1.

Whole Genome Expression Analysis
For whole genome expression analysis, MSC were seeded in a density of 3.1610 4 cells/cm 2 on TCPS, ResomerH LT706 and PCL and cultured for 21 days in growth medium (GM) or in osteogenic induction medium (OIM). Cells at day 0 served as control. Total RNA was isolated using the RNeasy micro Kit according to the manufacturers' instructions (Qiagen, Hilden, Germany). The RNA quality was assessed using RNA 6000 NanoChips with the Agilent 2100 Bioanalyzer (Agilent; Santa Clara, CA, USA). Probes for the GeneChipH Human Gene 1.0 ST Arrays (Affymetrix, Santa Clara, CA, USA) were prepared and hybridised to the arrays according to the Affymetrix GeneChipH Whole Transcript (WT) Sense Target labeling Assay Manual. Briefly, for each sample, 300 ng of total RNA was reverse trancribed into cDNA using a random hexamer oligonucleotide tagged with a T7 promoter sequence (59-GAATTGTAATAC-GACTCACTATAGGGNNNNNN-39). After second strand synthesis, double-stranded cDNA was used as template for amplification with T7 RNA polymerase to obtain antisense cRNA. Random hexamers were then used to reverse transcribe the cRNA into single stranded sense strand cDNA. The cDNA was then fragmented by UDG (uracil DNA glycosylase) and APE1 (apurinic/apyrimidic endonuclease 1). Fragment size was checked using the Agilent 2100 Bioanalyzer (fragment size between 50-200 bp). Fragmented sense cDNA was biotin-labelled with TdT (terminal deoxynucleotidyl transferase) and probes were hybridised to the GeneChipH Human Gene 1.0 ST Arrays at 45uC for 16 hours. Hybridised arrays were then washed and stained on Fluidics Station 450 and scanned on a GeneChipH Scanner 3000 7G (both Affymetrix).
The image data were analysed with GCOS (Affymetrix). For statistical analysis data were processed by R software (R Development Core Team, 2005). Gene expression levels were normalised with RMA algorithm [26]. Principal component analysis (PCA) and hierarchical clustering were done on whole genome transcripts by the R Stats Package [27]. Hierarchical clustering was performed using Pearson correlation coefficient and the Average linkage method. RankProd, a non-parametric statistic, was employed for identification of differentially expressed genes [28]. Transcripts with a fold change more than 2 and Pvalue less than 0.01 between two conditions were considered as being differentially expressed. Gene ontology (GO) and pathway over-representation analysis was done using DAVID bioinformatics resources [29].

ResomerH LT706 and PCL enhance MSC differentiation towards osteoblasts
In a previous study, we analysed MSC on a grid-based platform in contact to different polymers to determine cytocompatibility of polymers for future MSC-based tissue engineering applications [5]. Now, MSC were cultured for 21 days either in growth medium (GM) or in osteogenic induction medium (OIM) on two of the cytocompatible, synthetic, biodegradable polymers, Reso-merH LT706 and PCL, which are potentially osteoinductive. To analyse the osteogenic differentiation, calcium accumulations were visualised by Alizarin red staining (Fig. 1A), followed by dissolving the dye and subsequent quantification of the staining (Fig. 1B). Alizarin red stain residues were not retained by any of the biomaterials (Fig. S3). As shown in Figure 1, polymers did not induce osteogenic differentiation by themselve in GM and OIM was always required to guide MSC into osteogenic fate. However, although OIM was required to initiate osteogenic differentiation of MSC on the two polymers, biomaterial properties influenced calcium accumulations.
As shown in Figure 1, both polymers resulted in a stronger alizarin red staining than TCPS. Further, the amount of calcium accumulations on ResomerH LT706 was significantly higher, than on TCPS and on PCL. Since the Alizarin red staining is an indirect measure for osteogenic differentiation, we analysed the differentiation in more detail. The ultrastructural analysis depicted in Figure 1C illustrates the morphology of MSC on the two polymers after a three-week culture period in GM or in OIM (Fig. 1C). Our results show that although MSC morphology on the two polymers differed during initial adhesion [5], MSC morphology is identical on both polymers after 21 days of culture and independent of the culture medium (GM vs. OIM). Higher magnifications showed analogous particles densely covering the biomaterial surfaces under differentiation conditions (OIM) on both polymers, which were qualitatively identified as consisting of calcium and phospate by EDS analysis (Fig. 1D). Such calciumphosphate particles are indicative for an advanced osteogenic differentiation of MSC on both polymers, ResomerH LT706 and PCL. These results were confirmed by the expression of osteogenic markers, such as alkaline phosphatase (which is already expressed in unstimulated MSC), osteocalcin, bone sialoprotein and the transcription factor Cbfa-1 after 21 days of cultivation in OIM on the polymers (Fig. 2). By demonstrating (i) positive Alizarin red staining and (ii) expression of standard osteogenic markers on the RNA level, well-accepted standard assays were performed to show osteogenic differentiation of MSC on ResomerH LT706 and PCL. However, we reasoned that conventional standard assays are not sufficient to comprehensively investigate the osteogenic fate of MSC and thus we postulated, that there might be differences in the quality of the MSC-derived osteoblasts on the molecular level. Therefore, we performed whole genome expression analysis to determine biomaterial-related differences in osteogenic differentiation.

Whole genome expression analysis identifies specific responses of ResomerH LT706
To determine the molecular events ongoing in MSC differentiation on ResomerH LT706 and PCL, samples were analysed by whole genome expression profiling. First, data were subjected to principal component analysis (PCA, Fig. 3A) to discriminate related samples (similar gene expression) from distant samples (different gene expression). Figure 3A Fig. 3A). Hierarchical clustering (Fig. 3B) of the same data set also revealed general changes in gene expression over time (samples D1-3, grey dottet line), donor variations as well as the shift between growth and differentiation conditions (blue dottet lines vs. red dottet lines, respectively). Additionally, this analysis also identified clustering of ResomerH LT706 with TCPS, independent of donor and independent of growth (GM) and differentiation (OIM) conditions (Fig. 3B). Thus for each donor, ResomerH LT706 and TCPS consistently clustered together when cultured in GM or in OIM (e.g. L3 OIM/T3 OIM or L2 GM/T2 GM). In most cases, these neighboring clusters were adjacent to the corresponding donor in the respective medium on PCL (e.g. P3 GM is next to the L3 GM/T3 GM pair). Although the Alizarin red staining was much stronger on ResomerH LT706 than on TCPS (Fig. 1A, B), the impact of ResomerH LT706 on MSC gene expression seemed to be comparable to TCPS, both after culture in GM and OIM. In contrast, PCL resulted in a different gene expression profile compared to ResomerH LT706 and TCPS, but in most cases, PCL samples were in direct proximity to TCPS and ResomerH LT706 in the respective culture conditions (blue and red dottet lines; Fig. 3B).
The results shown in Figure 3 indicate donor variations, as expected for primary cells, but a similar influence on gene expression by TCPS and ResomerH LT706, which was independent of culture medium. Hence, we show that culture medium has a stronger influence on MSC differentiation than the biomaterial substrates. Alizarin red staining was stronger for PCL and ResomerH LT706 compared to TCPS (Fig. 1A, B), yet PCL  affected more genes within the same culture conditions than ResomerH LT706 when compared to cells on TCPS. This analysis describes the overall impact on the gene expression profile and the identity of differentially expressed genes is discussed below.

Biomaterial substrate and culture medium impact on gene expression of MSC and MSC-derived osteoblasts
ResomerH LT706, PCL and TCPS are semi-crystalline polymers with a characteristic, textured structure and ultraflat topography. Thus, the biomaterial surface physicochemistry could be responsible for the specific changes in gene expression rather than the biomaterial topography. Depending on the surface physicochemistry, proteins adsorb to the biomaterial, yet the underlaying mechanisms are still not fully understood. In this context we emphasize that cells do not adhere directly to biomaterial surfaces but rather adhere to a protein layer, which promptly adsorbes to biomaterials, when exposed to cell culture medium. Within the two cell culture media (GM and OIM) the composition of proteins is quite different, which presumably results in different protein layers on the biomaterial surfaces when cultured either in GM or in OIM (2% FCS vs. 10% FCS, respectively). Further, the amount of adsorbed proteins and the orientation of active groups of the adsorbed proteins could differ depending on the biomaterial physicochemistry. Accordingly, this leads to different cell behaviours on the very same biomaterial, when cultured in different media. However, within the same culture medium, different biomaterials can influence cells in different ways.
The heatmap of expression data shown in Figure 4A demonstates the impact of biomaterial substrate and culture medium on gene expression of MSC and MSC-derived osteoblasts. We detect two main gene expression pattern: one is representative for cells cultured in GM (Fig. 4A, right) and one for cells cultured in OIM (Fig. 4A, left), regardless of the biomaterial substrate meaning that the impact of the growth medium is higher than the impact of the biomaterials. However, within the two groups, MSC cultured on ResomerH LT706 and TCPS cluster together, while PCL form a separate branch, which is in line with the results depicted in Figure 3. Again, these results demonstrate the influence of culture medium on MSC, which is more prominent than the influence of biomaterial substrates.
As shown by hierarchical clustering (Fig. 3 and Fig. 4A), MSC on ResomerH LT706 and TCPS show a similar gene expression pattern and this translates into a small number of differentially expressed genes. As described above, MSC do not attach directly to the biomaterial surface, but to the protein layer, which covers the biomaterials after exposure to the cell culture medium. As shown in Figure S1, fibronectin (FN) and vitronectin (VN), two main components of serum involved in cell adhesion, are adsorbed to TCPS, ResomerH LT706 and PCL. While the amount of adsorbed FN is very similar for TCPS and PCL (approx. 120 ng/ cm 2 ), ResomerH LT706 is covered with approx. 150 ng FN/cm 2 . However, the relation between absorbed FN and VN is the same for TCPS and ResomerH LT706 (fibronectin to vitronectin ratio = 1.2), while the relation between FN and VN on PCL is 1.04. These results were not significantly different either when we used a buffer system including FN and VN or serum-containing medium including FN and VN (not shown). Thus, the ratio of adsorbed FN to adsorbed VN might be key to MSC gene expression.
Finally, differentially regulated genes of MSC cultured on ResomerH LT706, PCl and TCPS under differentiation conditions (OIM) were subjected to hierarchical cluster analysis and data depicted in heat map format (Fig. 4B). This analysis shows that ResomerH LT706 and TCPS clustered together and PCL formed a distinct cluster. Some donor variations were also seen in this heat map representation, which is in line with the PCA analysis (Fig. 3A). In addition, this analysis identified a large cluster of genes, with a similar expression pattern in ResomerH LT706 and TCPS (Fig. 4B, upper boxed area). Furthermore, we also identified genes that showed a similar expression in MSC on ResomerH LT706 and PCL, but are differentially expressed in MSC on TCPS, e.g. CCL2 (Fig. 4B, bottom boxed area).
Well-known osteogenic marker genes were expressed in MSC cultured on all three biomaterials (TCPS, ResomerH LT706 and PCL; compare Fig. 2). Thus, the expression and interaction of other genes involved in osteogenesis and chondrogenesis (Fig. 4C) might be important to regulate MSC differentiation towards osteoblasts either via osteogenesis or ossification on these biomaterials. To further support the data of whole genome gene expression analysis, we perfomed quantitative PCR. Therefore, we choosed a panel of genes involved in skeletal development and osteogenic differentiation, highlighted in orange in Figure 4B. As shown in Figure 4C, gene expression of MSC-derived osteoblasts was influenced by biomaterial substrates or -by keeping in mind the results on FN and VN described above -by serum proteins which are adsorbed to the biomaterials before cell attachment.
Gene ontology (GO) overrepresentation analysis detects biological categories related to skeletal development and bone formation Gene functions are described in a controlled vocabulary format referred to as gene ontology (GO; www.geneontology.org) and thus differentially regulated genes between the three biomaterials cultured in GM and OIM were subjected to GO overrepresentation analysis, respectively. The main categories of the two culture conditions are shown for MSC cultured on ResomerH LT706, PCL and TCPS in GM (Table 2) and for MSC cultured on ResomerH LT706, PCL and TCPS in OIM (Table S2). Some categories exist in both lists, like developmental process, system development and extracellular region. However, bone specific categories, such as bone remodelling, calcium ion binding, bone mineralisation, ossification and biomineral formation are -as expected -not found in the GO list for the growth conditions (GM), but confined to the GO list for the differentiation conditions (OIM).

Discussion
MSC are the natural precursor cells of mesenchymal tissue, such as fat, bone and cartilage. They reside in a tissue-specific niche, awaiting signals for tissue regeneration, if necessary [30]. In the last decade, MSC were extensively investigated for their usefulness in bone tissue engineering and diverse scaffold materials were suggested to provide an appropriate three-dimensional environment [31][32][33][34]. Promising tissue-engineered constructs are already on the way from bench to bedside [35]. However, the impact of scaffolds on osteogenic differentiation on the molecular level was rarely analysed and restricted to some well-known osteogenic marker genes, such as alkaline phosphatase, runt related transcription factor 2, type 1 collagen, bone sialoprotein and osteocalcin [34,36].
In the present study, we initially investigated a panel of different polymers for their influence on the osteogenic differentiation of MSC. We then focused on the two most prominent synthetic polymers ResomerH LT706 and PCL. Both polymers are linear, semi-crystalline polyesters, which are hydrolytically or enzymatically degradable. ResomerH LT706 is degradable within 8-12 month, while PCL is long-term degradable (.24 month) [20].
PCL is an FDA proved and widely used engineering polymer with excellent physical, chemical and mechanical properties [37,38]. It is already analysed in the context of bone tissue engineering, either as pure PCL [39], as composite scaffold e.g. together with hydroxyapatite or calcium phosphate [40,41] or modified with a peptide layer such as RGD [16]. All of these studies suggest PCL as suitable scaffold material for bone tissue engineering.
In contrast, ResomerH LT706 is a relatively novel and lessinvestigated synthetic polymer, described for the first time as Poly(L-lactide-co-TMC) in 2005 by Pospiech and coworkers [42]. Materials with related chemistry are shown to be biodegradable and biocompatible for soft tissue engineering [19] and used in clinical trials for skin replacements [17,18]. However, our recent work showed that a prediction of cell behaviour on a chemically related material is not possible [5]. ResomerH LT706 was not analysed in the context of bone tissue engineering or osteogenic differentiation of MSC so far.
We compared the osteogenic differentiation of MSC on the two elastomeric and long-term degradable synthetic polymers Reso-merH LT706 and PCL. TCPS served as control. Our XPS data revealed that the two polyesters ResomerH LT706 and PCL do not differ qualitatively in their elemental composition, but quantitatively, indicated by the C/O ratio. The C/O-values for ResomerH LT706 and PCL are 1.7 and 2.6, respectively (Table S1). Further, XPS analysis of our biomaterial surfaces shows that PCL contains less oxygen atoms than ResomerH LT706, indicating PCL as more hydrophobic than ResomerH LT706. However, contact angle measurements detected the opposite, with ResomerH LT706 being more hydrophobic than PCL with the corresponding contact angles of 75u and 69u, respectively. The higher hydrophobic characteristic of ResomerH LT706 is a result of its molecular structure. ResomerH LT706 consists of 68-72% L-lactide units and thus includes sterically demanding and hydrophobic methyl groups in high frequencies at the surface. In contrast, PCL only includes methylene groups in its backbone. TCPS -in contrast to polystyrol -does not consist of phenyl rings, but also includes surface modifications (established by physical plasma) of hydroxyl-, carboxyl-and amino-groups on the surface and a contact angle of only 54u (Table S1).
To allow for an adequate adhesion of MSC on our substrates (TCPS, ResomerH LT706, PCL and TCPS), an initial adsorption of cell adhesion mediators from culture medium, such as fibronectin (FN) and vitronectin (VN), is crucial, since in general, the less adsorbed proteins the less cell adhesion is mediated. As shown in our adsorption studies with radiolabelled FN and VN ( Figure S1), ResomerH LT706 is covered with the highest amount of FN proteins, compared to TCPS and PCL. Further, the ratio of FN to VN on ResomerH LT706 is comparable to that on TCPS, albeit absolut values are higher for ResomerH LT706. It is known that extracellular matrix (ECM) proteins are involved in cell adhesion not only by their own presence, but also by their relation to other ECM proteins [43,44]. This ratio of different ECM proteins then can result in differences in cell adhesion, morphology, proliferation, gene expression as well as ECM secretion. The similar ratio of FN/VN on ResomerH LT706 and TCPS compared to PCL (1.2 and 1.04, respectively) might be the reason for a comparable initial adhesion and morphology of MSC, which differs from that on PCL, where the cells are more roundish after 24 h of incubation [5]. However, MSC proliferation within 7 days after cell seeding was comparable for TCPS, ResomerH LT706 and PCL [5] and cell morphology was identical after 21 days of culture in GM or OIM (Fig. 1C). Although we detected a comparable proliferation of MSC on all three substrates, osteogenic differentiation capacity was different.
In the present study, we found that both ResomerH LT706 and PCL cyused stronger alizarin red stainings compared to TCPS, which however was most significant for ResomerH LT706. The fact that the alizarin red staining is stronger for ResomerH LT706 than for PCL, though MSC proliferation is comparable on both polymers [5], suggests that ResomerH LT706 is more potent in supporting osteogenic differentiation than PCL. A positive Alizarin red staining is indicative for osteogenic MSC fate, but stains only calcium accumulations and thus provides only indirect evidence. We were thus interested in the quality and the molecular mechanisms underlying osteogenic differentiation of MSC on ResomerH LT706 and PCL. As initial step in this direction we performed whole genome expression analysis using Affymetrix gene arrays. Bioinformatic tools then unravelled (i) donor variations as expected for primary cells, (ii) clusters of samples representative for MSC cultured in growth medium (GM) or in osteogenic induction medium (OIM), indicating a stronger influence of culture medium than of biomaterial substrates, and (iii) clusters of TCPS and ResomerH LT706 under growth (GM) and differentiation conditions (OIM), demonstrating a similar influence on MSC gene expression. Pairs of TCPS and ResomerH LT706 were detected in all bioinformatic analysis. Although both materials are quite different e.g. in surface chemistry (polarity, hydrophobicity, surface charge) and bulk properties (e.g. stiffness), the impact on gene expression related to osteogenesis of MSC is comparable. Yet, TCPS is not useful for bone tissue engineering because of its fabrication characteristics based on its chemical and physical properties. Again this supports our recent study, showing that related biomaterials do not inevitably result in the same cellular response, while unrelated materials might do [5]. The heat map in Figure 4B shows that most genes were expressed on a similar level in MSC cultured on TCPS and ResomerH LT706, while only a few genes were differentially expressed on these two materials.
For all of these genes published data suggest a role in skeletal development or osteogenic differentiation. SFRP4 is expressed in periosteum and bone tissue but overexpression of SFRP4 suppresses osteoblast proliferation [45]. PRELP is a connective tissue matrix protein, which is expressed in cartilage and in osteoblasts [46]. COMP is involved in skeletal development and osteoblast differentiation [47] and detectable in MG-63 cells, an osteoblast cell line. Mutations of COMP are related to specific diseases, such as pseudoachondroplasia and multiple epiphyseal dysplasia [48]. COL11A1 is essential for normal skeletal development, but has to be suppressed for terminal osteoblast differentiation [49]. Elastin degradation products promote osteogenic differentiation and elastin-related calcification is suggested to be involved in tissue repair processes [50]. Finally, CCL2 -which is known as chemokine for the recruitment of cells of the immune system, such as monocytes -is secreted by MSC and this secretion increases the differentiation into mature osteoblasts [51].
Compared to MSC cultured on TCPS and ResomerH LT706, all genes showed lower expression in MSC cultured on PCL, except CCL2. Since all these genes are involved in skeletal development and bone formation, our results suggest ResomerH LT706 as more suitable for bone tissue engineering than PCL.  In summary, we analysed osteogenic differentiation of MSC on ResomerH LT706 and PCL. Alizarin red stainings and expression of conventional osteogenic transcripts show that both biomaterials support osteogenic differentiation fate. However, whole genome expression analysis revealed differences in gene expression and genes involved in skeletal development and bone formation are more expressed at higher levels in MSC cultured on ResomerH LT706. Thus, this novel, long-term degradable and osteoconductive synthetic polymer is suggested as particularly attractive scaffold material for bone tissue engineering with superior properties compared to the currently being used material PCL. The in vitro transformation of MSC on ResomerH LT706 to more osteogenic genotypes might also translate to phenotypes. This hypothesis has to be verified in future in vivo models.