**Di**ff**erences in the Emission of Volatile Organic Compounds (VOCs) between Non-Di**ff**erentiating and Adipogenically Di**ff**erentiating Mesenchymal Stromal**/**Stem Cells from Human Adipose Tissue**

**Ann-Christin Klemenz 1, Juliane Meyer 2, Katharina Ekat 2, Julia Bartels 1, Selina Traxler 1, Jochen K. Schubert 1, Günter Kamp 3, Wolfram Miekisch 1,**† **and Kirsten Peters 2,\*,**†


Received: 4 June 2019; Accepted: 9 July 2019; Published: 10 July 2019

**Abstract:** Metabolic characterization of human adipose tissue-derived mesenchymal stromal/stem cells (ASCs) is of importance in stem cell research. The monitoring of the cell status often requires cell destruction. An analysis of volatile organic compounds (VOCs) in the headspace above cell cultures might be a noninvasive and nondestructive alternative to in vitro analysis. Furthermore, VOC analyses permit new insight into cellular metabolism due to their view on volatile compounds. Therefore, the aim of our study was to compare VOC profiles in the headspace above nondifferentiating and adipogenically differentiating ASCs. To this end, ASCs were cultivated under nondifferentiating and adipogenically differentiating conditions for up to 21 days. At different time points the headspace samples were preconcentrated by needle trap micro extraction and analyzed by gas chromatography/mass spectrometry. Adipogenic differentiation was assessed at equivalent time points. Altogether the emissions of 11 VOCs showed relevant changes and were analyzed in more detail. A few of these VOCs, among them acetaldehyde, were significantly different in the headspace of adipogenically differentiating ASCs and appeared to be linked to metabolic processes. Furthermore, our data indicate that VOC headspace analysis might be a suitable, noninvasive tool for the metabolic monitoring of (mesenchymal stem) cells in vitro.

**Keywords:** adipose tissue-derived mesenchymal stromal/stem cells (ASCs); cell differentiation; volatile organic compounds; metabolic monitoring

#### **1. Introduction**

Due to their capacity for self-renewal and multipotent differentiation, mesenchymal stem/stromal cells (MSCs) have been identified as playing an essential role in tissue homeostasis and regeneration [1]. In recent years, increasing attention has been paid to MSCs from human adipose tissue (adipose tissue-derived MSCs, referred to as ASCs), as they show promising potential as a clinical alternative to other MSCs, like bone marrow-derived MSCs (bmMSCs) [2–4]. Zuk et al. were the first to characterize the mesenchymal differentiation potential of ASCs in more detail [5,6]. Furthermore, nonmesenchymal differentiation (e.g., neuron-like morphology) of ASCs has also been demonstrated [7].

Data concerning the energy metabolism of MSCs in vitro are available but have not been worked out in detail. It has been shown that bmMSCs mainly facilitate glycolysis with subsequent lactate production rather than oxidative phosphorylation [8]. Furthermore, it was demonstrated that nondifferentiating and osteogenically differentiating bmMSCs facilitated oxidative phosphorylation as well as glycolysis in order to fulfill their energy metabolic needs [9]. It has been reported that the function of mitochondria also regulates the differentiation of MSC [10,11]. Adipogenic differentiation of ASC is accompanied by increasing mitochondrial enzyme activities, indicating a growing capacity for oxidative phosphorylation and β-oxidation and, thus, a shift towards lipid metabolism [12]. In addition, in human differentiated adipocytes, the enzymes involved in mitochondrial metabolism showed significantly higher activity rates than in the CD34<sup>+</sup> stromal vascular fraction of human adipose tissue containing the nondifferentiated progenitors of adipocytes, the ASCs [13]. Thus, in recent years the idea matured that metabolic pathways regulate cellular differentiation processes that go beyond ATP production [14]. To determine metabolic activities in different phases of differentiation, further information is needed. Ideally, gathering this data should not influence or even disturb the cell culture.

For conventional metabolomics studies in vitro, sufficient cell material is necessary. Especially when working with human stem cells, often low cell numbers are available. In contrast to those conventional methods, an analysis of volatile organic compounds (VOCs) has the great advantage of being nondestructive and noninvasive [15]. Such a destruction-free analysis can be done by means of bidirectional preconcentration combined with a versatile, standardized set-up and has, therefore, gained more attention during the last decades. Besides the occurrence of VOCs in human breath, they are also emitted in trace concentrations by cell and bacteria cultures [16–18].

Headspace analysis by means of needle trap microextraction (NTME) coupled with gas chromatography and mass spectrometry (e.g., GC-MS with electron impact quadrupole detection) allows the identification and quantification of compounds down to the parts per trillion by volume (pptV) range. Conventional headspace techniques require larger volumes and, therefore, may affect the investigated system itself. Hence, microextraction techniques are better suited for this kind of setup [19]. In contrast to solid phase microextraction (SPME), the sensitivity of NTME can be enhanced by increasing the sampling volume, as the sample volume is usually limited. Bidirectional NTME sampling can be applied to reach a high sensitivity in the VOC extraction [20] without significant effects on volume or pressure of the in vitro system. A crucial step for a reliable and reproducible VOC headspace analysis in in vitro cultures is also a standardized set-up [21].

Data from several studies suggest that different cell lines and bacterial strains can be distinguished from each other by means of their volatile emissions [22–24]. Some volatiles have been shown to be associated with different pathways in cell metabolism [25]. For example, acetone is produced by the decarboxylation of acetoacetate, and acetaldehyde is produced by ethanol oxidation [26]. Previous work has established that volatile substances over cell cultures correlate with stem cell growth [27].

Therefore, the aim of our study was to assess VOC emissions in the headspace of cell cultures to gather complementary information on metabolic changes of ASCs during differentiation. This raises the question of whether we can detect the differences in VOC emissions from cell cultures between nondifferentiating and differentiating ASCs. This in turn leads to the next question: can the differences in VOC emission be related to metabolic changes in ASCs during differentiation?

#### **2. Materials and Methods**

#### *2.1. Isolation, Cultivation, and Di*ff*erentiation of Adipose Tissue-Derived Mesenchymal Stromal*/*Stem Cells (ASCs)*

ASCs used in this study were isolated from human lipoaspirate. Tissue donation was approved by the ethics committee of the Rostock University Medical Center (http://www.ethik.med.uni-rostock.de/) under the registration number A2013-0112. It complies with the ethical standards of the World Medical Association Declaration of Helsinki. Informed consent was obtained from all patients. The material for the different analyses was retrieved from four patients. Data, as supplied by the surgeons, showed one male and three female patients. Patients were on average 39.5 years old (ranging from 32 to 47). Liposuction procedures were performed by waterjet-assisted liposuction or a tumescent suction technique.

ASC isolation was performed as previously described [12]. ASCs were cryopreserved in passage 2 until they were used for the experiments. To achieve this, the cell suspension was transferred into a cryovial (Greiner bio-one, Germany) containing 150 μL DMSO (Sigma-Aldrich, Germany) and 350 μL fetal calf serum (FCS, PAN Biotech, Germany). The cryovials were cooled to −80 ◦C overnight at 1 ◦C per minute in a freezing container (Thermo Fisher Scientific, Berlin, Germany). Vials were then stored in liquid nitrogen at −165 ◦C until further use. For use in the planned experiments, the cells were thawed stepwise by gently shaking the vial at 37 ◦C in a water bath for 1 min. The cell suspension was then gradually transferred into cell culture medium (Dulbecco's Modified Eagle Medium/DMEM GlutaMAX-I, Thermo Fisher Scientific, Germany), containing 10% FCS and antibiotics (100 U/mL penicillin, 100 mg/mL streptomycin, Thermo Fisher Scientific, Germany; hereinafter called maintenance medium) at room temperature and centrifuged for 5 min at 400× *g*. The resulting cell pellet was resuspended in maintenance medium and centrifuged again for 5 min at 400× *g*. Cells were resuspended in maintenance medium and seeded onto cell culture flasks for cultivation at 37 ◦C and 5% CO2 in a humidified atmosphere. When confluency was reached after 5 d, cells were detached from the cell culture flasks by incubation with 0.25% Trypsin EDTA for 5 min at 37 ◦C. The cells were seeded into cell culture petri dishes (Greiner bio-one, Frickenhausen, Germany) at a density of 20,000 cells/cm2. After 24 h of incubation the medium was replaced. After further 48 h (Day 0), additional to the nondifferentiating cultures, adipogenic differentiation of ASCs was induced by adding a differentiation-stimulating medium: i.e., a maintenance medium containing 1 μM dexamethasone, 500 μM IBMX, 200 μM indomethacin, and 10 μM insulin (Sigma-Aldrich, Munich, Germany). Adipogenic stimulation (AS) took place with every replacement of medium three times a week (every second or third day). The medium replacement of differentiating ASCs and undifferentiating ASCs without specific differentiation factors was performed (unstimulated/US) simultaneously. To also assess the emissions of the cell culture medium, pure culture medium without ASC (medium control) was treated and analyzed in a manner identical to the ASC cultures.

#### *2.2. Analysis of Cell Numbers*

Cell numbers were determined according to the manufacturer's instructions using the Nucleocounter NC200 and the Via1-Cassettes™ (ChemoMetec, Lillerod, Denmark).

#### *2.3. Analysis of Adipogenic Di*ff*erentiation*

In order to assess the adipogenic differentiation of the ASCs over the course of the experiment, fluorescent staining of the nuclei and lipid-filled vacuoles was done at the corresponding time points of 1, 7, 14, and 21 d. Cells were seeded into 96-well μClear® cell culture plates (Greiner, Frickenhausen, Germany) at a density of 20,000 cells/cm2. Media changes were done according to the same schedule as the VOC quantification. At the corresponding time points the cells were washed twice with PBS and fixed with 4% paraformaldehyde for 30 min at room temperature. Thereafter, the cells were incubated with a Bodipy/Hoechst-staining solution (100 μL of 1 μg/mL Bodipy (Life Technologies, California, Carlsbad, USA) and 5 μg/mL Hoechst 33,342 (AppliChem, Darmstadt, Germany) in 150 mM NaCl) for 10 min at room temperature in the dark. After incubation, the cells were washed twice with PBS and twice with water for injection. Subsequently, pictures were taken with the microscopic plate scanning system Hermes WiScan (IDEA Bio-Medical, Rehovot, Israel) with a 10-fold magnification objective.

#### *2.4. Volatile Organic Compound (VOC) Sampling by Means of Needle Trap Micro Extraction*

For headspace sampling, three cell culture dishes (without lid) containing culture medium and ASCs were introduced into a hermetically closed sampling box under a sterile hood (schematically shown in Figure 1). The sampling box was constructed from emission-free materials (Teflon® and glass) to ensure a reliable measurement of trace VOC profiles [28]. As a negative control, three cell culture dishes containing pure cell culture medium without ASCs were analyzed in parallel in a second

box. After introduction of the cell culture dishes, the boxes were flushed with 3 L of clean synthetic air (containing 75% N2, 20% O2, and 5% CO2; Air Liquide, Düsseldorf, Germany). After an incubation time of 60 min at 37 ◦C in an incubator, headspace sampling by means of NTME was done. A detailed description of the sampling setup was described before [23]. Headspace samples were taken at day 4 after seeding the cells, which means 24 h after changing from maintenance to adipogenic differentiation medium and on day 7, 14, and 21 of differentiating cultures, nondifferentiating cultures, and medium control dishes, respectively (see experimental setup in Figure 1).

**Figure 1.** Schematic depiction of the experimental setup. VOC—volatile organic compound.

NTME was done with needle trap devices (NTDs, needleEX), obtained from Shinwa Ltd. (Kyoto, Japan). The NTDs were equipped with 3 cm of a copolymer of methacrylic acid and ethylene glycol dimethacrylate. Before sampling, NTDs were preconditioned for 30 min at 200 ◦C in a heating device (PAS Technology Deutschland GmbH, Magdala, Germany) under a helium flow. Teflon caps (PAS Technology Deutschland GmbH, Magdala, Germany) were used to seal the NTDs before and immediately after collecting the headspace samples.

For VOC preconcentration, NTDs were connected to a 1 mL sterile singleuse syringe (Omnifix-F, B. Braun Melsungen AG, Melsungen, Germany) and pierced through the septum of a Luer Lock cap (IN-Stopper, B. Braun Melsungen AG, Melsungen, Germany) at one connection port of the sampling box. Bidirectional headspace sampling was done by filling and releasing 1 mL of headspace gas 20 times, as described previously [20].

#### *2.5. Gas Chromatography and Mass Spectrometry (GC-MS) Analysis*

For VOC analysis, an Agilent 7890 A gas chromatograph coupled to an Agilent 5975 C inert XL MSD (Agilent, Santa Clara, CA, USA) with a triple axis detector was used. At an injector temperature of 200 ◦C, VOCs were thermally desorbed from the NTDs. Sample injection was operated in splitless mode (60 s splitless). The GC was equipped with a 60 m RTX-624 column (0.32 mm ID, 1.8 μm column thickness). Helium carrier gas flow was constant at 1.5 mL min<sup>−</sup>1. The temperature program was as follows: 40 ◦C for 5 min, 8 ◦C min−<sup>1</sup> to 120 ◦C for 2 min, 10 ◦C min−<sup>1</sup> to 220 ◦C, and 20 ◦C min−<sup>1</sup> to 240 ◦C for 4.5 min. Analysis of the samples was performed via electron impact ionization (EI—70 eV) in full scan mode, a mass range of 35–250 amu, and a scan rate of 2.73 scan/s. Volatile organic compounds were tentatively identified by mass spectral library search (NIST Version 2.0). Substance attribution was confirmed by comparing GC retention times and mass spectra of all selected substances with those of pure reference substances. Quantifications were made in parts per billion per volume (ppbV) and then calculated to nmol/L. To quantify the detected substances, a six-point calibration curve (from 1 to 500 ppbV) was established. Humidity-adapted standards were prepared by means of a liquid calibration unit (LCU, Ionicon Analytik GmbH, Austria). Limits of detection were calculated as a signal-to-noise ratio of 3:1, whereby limits of quantification corresponded to a signal-to-noise ratio of 10:1. Ten blank NTDs were analyzed for this purpose. Quantitative parameters (limit of detection (LOD), limit of quantification (LOQ), and standard deviation (SD)) for all identified VOCs are shown in Table S1 in the Supplementary Materials.

#### *2.6. Reference Substances*

Acetone, 2-butanone, heptanal, octanal, and acetaldehyde were acquired from Ionimed Analytik GmbH (Insbruck, Austria). Benzaldehyde and 1,3-di-tert-butylbenzene were purchased from Sigma-Aldrich (Germany); tert-butanol, ethylbenzene, as well as 2-ethylhexanol were purchased from TCI (Eschborn, Germany) and pentane from Fluka (Munich, Germany).

#### *2.7. Statistical Analysis*

Figures 2–5 show numerical data represented as medians and percentiles (25–75%). Correlation analysis and statistical testing was performed by using RStudio (version 1.0.136) and R software (version 3.3.2\_ 2016-10-31). By using the Kruskal–Wallis test with the Nemenyi post hoc test, statistically significant differences in median values between all groups were identified. Values of *p* < 0.05 were considered statistically significant.

**Figure 2.** Depiction of adipose tissue-derived mesenchymal stromal/stem cells (ASCs) under nondifferentiating conditions (referred to as non-diff ASC, subfigures (**a**,**c**,**e**,**g**) for cultivation days 1, 7, 14, 21, respectively) and under adipogenic stimulation (adipo ASC, subfigures (**b**,**d**,**f**,**h**) for cultivation days 1, 7, 14, 21, respectively) Ffluorescence staining of nuclei (blue) and lipids (green), scale bar: 100 μm.

**Figure 3.** Emissions of acetaldehyde, pentane, 1,3-di-tert-butylbenzene, ethylbenzene, benzaldehyde, and heptanal from nondifferentiating (non-diff ASC, blue) and adipogenically differentiating ASC (adipo ASC, green). Concentrations in the headspace are shown in nmol/L on the Y-axis. The X-axis shows the time points of measurements. The boxplots represent data from three independent experiments. Significance was tested within all groups. Symbols (\*, #, \$) indicate significant differences to the corresponding highlighted group (*p*-values < 0.05).

**Figure 4.** Example of two VOCs showing lower emissions in adipogenic differentiating ASC (adipo ASC) cultures than in the culture media controls. Concentrations in the headspace in nmol/L are shown on the Y-axis. The X-axis shows the time point of measurements. The diagram shows cell culture with adipogenically differentiating ASC (referred to as Adipo ASC, in blue) and medium controls (samples without cells, in red). Boxplots represent the data from three independent differentiation experiments. Significance was tested within all groups. Symbols (\*, #, §, \$) indicate significant differences to the corresponding highlighted group (*p*-values < 0.05).

**Figure 5.** Acetone and 2-ethylhexanol as exemplary medium-dependent VOCs in adipogenically differentiated ASCs (referred to as Adipo ASC). The concentration in the headspace in nmol/L is shown on the Y-axis. The X-axis shows the measurement time points. The diagram shows cell culture with adipogenically differentiating ASC (referred to as Adipo ASC, in blue) and medium controls (samples without cells, in red). Boxplots represent data of three independent differentiation experiments. Significance was tested within all groups. Symbols (\*, #, \$) indicate significant differences to the corresponding bold highlighted group (*p*-values < 0.05).

#### **3. Results**

#### *3.1. Adipogenic Di*ff*erentiation of ASCs*

The assessment of adipogenic differentiation by fluorescent lipid staining revealed that nondifferentiating conditions did not induce lipid accumulation in ASCs at any time point (Figure 2a,c,e,f), After 24 h, adipogenic stimulation did not lead to lipid accumulation (Figure 2b). After 7 d of adipogenic stimulation, distinct lipid vacuole formation was visible (Figure 2d). With the progression of adipogenic differentiation of ASCs (day 14 and 21), the lipid vacuoles appeared larger and showed a higher fluorescence intensity (Figure 2f,h). Thus, considerable adipogenic differentiation was clearly visible in almost all cells of the adipogenic cell culture model. This was not the case in nondifferentiating ASCs.

#### *3.2. Comparision of VOC Emissions of Non-Di*ff*erentiating and Adipogenically Di*ff*erentiating ASCs*

Thirty-one potential volatile marker substances were identified in the headspace above the cell cultures (concentration ranges and quantitative parameters are depicted in Table S1 of the Supplementary Materials). To focus on the most relevant substances, we excluded substances having concentrations below the limit of detection (LOD) at more than 3 time points as well as substances having concentrations in the same range (± 10%) as in pure media samples. Based on these criteria, 11 VOCs were selected for further analysis. These substances were identified and quantified by using pure reference substances. Six of these VOCs that were differentially emitted and dependent on the differentiation are depicted in Figure 3.

The compounds acetaldehyde, pentane, and 1,3-di-tert-butylbenzene displayed different trends of emission during cultivation with maintenance media compared with cultivation with differentiation medium (Figure 3).

Concentrations of the emitted acetaldehyde changed significantly between days 7, 14, and 21 of adipogenically differentiated ASCs. In nondifferentiating ASCs, no significant concentration differences could be detected between the days of cultivation.

Pentane concentrations decreased in nondifferentiating ASCs, whereas a slight increase in adipogenically differentiating ASCs was detected from day 14 to day 21 of differentiation. The production of pentane in the adipogenically differentiating cell cultures peaked on day 1 and was significantly higher than on day 7 and on day 14.

The concentration of 1,3-di-tert-butylbenzene in the headspace of nondifferentiating ASCs remained lower as well as constant over the time, whereas the emissions in the headspace of adipogenically differentiating cell cultures showed a slight increase.

Concentrations of ethylbenzene and benzaldehyde differed between nondifferentiating and adipogenically differentiating ASCs in terms of concentration and time course. Both substances showed lower emissions in adipogenically differentiating ASCs compared with nondifferentiating ASCs. For benzaldehyde, these differences were significant. Heptanal showed a significant decrease over time in nondifferentiating ACSs and an increase in adipogenically differentiating ASCs. Concentrations of heptanal on day 21 of cultivation were higher in adipogenically differentiating compared with nondifferentiating ASCs. Detailed statistical information can be found in the Supplementary Materials (Table S2).

To emphasize the differences between nondifferentiating and adipogenically differentiating ASCs, Table 1 presents concentrations of emitted VOCs normalized to the cell number at day 21. All substances other than benzaldehyde showed higher concentrations in differentiating cells than in undifferentiating ASCs. The most prominent differences were found for acetaldehyde and heptanal. In both substances, a nearly threefold increase of emitted concentrations could be detected on day 21 of the experiment.

deviation (SD). Statistical significance was tested by a *t*-test (*p* < 0.05). **Substance Normalized VOC Concentration [pmol**/**L per 1** <sup>×</sup> **<sup>106</sup> cells** <sup>±</sup> **SD] Statistically Significant Nondi**ff**erentiating ASCs Adipogenically Di**ff**erentiating ASCs** Acetaldehyde 229.3 (± 34.5) 686.4 (± 14.4) Yes Pentane 64.2 (± 8.7) 86.9 (± 1.1) Yes 1,3-bis-(1,1-dimethylethyl) Benzene 0.368 (<sup>±</sup> 0.04) 1.4 (<sup>±</sup> 0.2) Yes Ethylbenzene 0.94 (± 0.1) 1.9 (± 0.4) Yes Benzaldehyde 6.17 (± 2.1) 5.93 (± 1.2) No Heptanal 9.6 (± 1.9) 25.4 (± 5.1) Yes Octanal 11.5 (± 1.1) 23.8 (± 0.6) Yes

**Table 1.** Ratio of VOC production in the headspace per cell from nondifferentiating and adipogenically differentiating ASCs at day 21 of cultivation. Concentrations were calculated per 1 <sup>×</sup> 10<sup>6</sup> cells from one experiment as an example. VOC concentrations are shown in pmol per L (pmol/L) ± standard

#### *3.3. Comparision of VOC Emissions from Medium Control and Corresponding Cell Cultures*

Apart from emissions of the cells, VOCs were also emitted from the cell culture medium without cells. Therefore, the headspace of the medium without cells was used as a "medium control" for each time point tested. The detailed depictions of emissions of differentiating and nondifferentiating ASCs compared to the medium control are shown in Figures S1 and S2, respectively. In the following, exemplary results from adipogenically differentiating ASCs are presented.

Acetaldehyde, pentane, and 1,3-di-tert-butylbenzene showed medium independent emissions in adipogenically differentiating cultures. Acetaldehyde and 1,3-di-tert-butylbenzene emissions increased at every measurement time point. Most other VOCs (especially aldehydes) showed lower and decreasing concentrations compared to the medium controls.

#### 3.3.1. VOC Consumption during Adipogenic Differentiation

The emissions of three aldehydes (heptanal, octanal, and benzaldehyde) and one aromatic hydrocarbon (ethylbenzene) were higher in the pure cell culture media compared with cell cultures. As examples, ethylbenzene and benzaldehyde of this VOC group are depicted in Figure 4 (heptanal and octanal can be found in Figure S1 of the Supplementary Materials).

Ethylbenzene concentrations were significantly higher in medium controls compared with cell culture samples at three time points analyzed. For benzaldehyde, we could find significant differences in the emissions from all cell culture samples compared with the pure medium control at all time points. In nondifferentiating ASCs, ethylbenzene and benzaldehyde emissions showed the same profile of higher emissions in the medium control compared with the cell culture (see Figure S2). Thus, the specific consumption or binding of these VOCs by the cells is indicated.

#### 3.3.2. Culture Medium-Dependent VOCs

The emissions of 2-ethylhexanol, acetone, tert-butanol, and 2-butanone from adipogenically differentiating ASCs complied with the emissions of the medium controls. Heptanal and octanal emissions also appeared media-dependent in nondifferentiating ASCs (see Figure S2 in the Supplementary Materials).

The emissions of two exemplary VOCs, acetone and 2-ethylhexanol, over 21 days of cultivation in adipogenically differentiating ASCs are presented in Figure 5. The emissions of tert-butanol and 2-butanone for adipogenically differentiating ASCs showed a slight decreasing trend over the first two weeks of differentiation and peaked at day 21. This trend was similar in the medium control (Supplementary Figure S1). The more comprehensive depiction of adipogenically differentiating and nondifferentiating ASC VOCs from Figures 4 and 5 can also be found in the Supplemenary Figure S1.

#### **4. Discussion**

A nondestructive technique that could save time and material and gain more detailed information on cellular metabolism would be desirable. The analysis of VOCs by means of NTME preconcentration of headspace air coupled with GC-MS could be a fast and nondestructive method for the analysis of metabolic processes in cell cultures [29–31].

In this study, our previously developed versatile sampling system [28] was applied to monitor VOC emissions of cellular differentiation under standardized conditions. This method, however, presents some challenges. Physical, chemical, and biological processes (e.g., the degradation of FCS in culture medium) might affect VOC emissions. Therefore, the analysis of an adequate control—in this case pure cell culture medium without cells—is a crucial step in order to obtain reliable results. Large variations in VOC concentrations emitted from pure culture medium have been observed previously [21] and might be due to physical and chemical processes during heating and cooling as well as aging of the media. Furthermore, lipophilic VOCs (such as ethylbenzene, heptanal, or 2-ethylhexanol) might also be bound to lipids of cellular structures (e.g., membranes) in general [32].

In our study we utilized an established cell culture model that showed clear differences between nondifferentiated cells and cells in adipogenic differentiation. Analysis of the formation of lipid-filled vacuoles clearly confirmed the differentiation of the ASCs after adipogenic stimulation, whereas the nondifferentiated ASCs did not show any signs of adipogenesis. This cell culture model has already been examined for energy metabolic aspects, and it has been shown that adipogenic differentiation induces a shift towards lipid metabolism [12].

In the headspace of both ASC cultures (adipogenically differentiating and nondifferentiating), some VOC emissions seem to be related to culture media emissions rather than to emissions from the cells in culture. This applies to four substances (2-ethylhexanol, acetone, tert-butanol, and 2-butanone) in the headspace of adipogenically differentiating ASC cultures and two substances (heptanal and octanal) in the headspace of nondifferentiating ASC cultures. Whether these changes are specific and biochemically driven or merely induced by unspecific chemical conversion remains unclear. However, since the VOCs acetaldehyde, pentane, and 1,3-di-tert-butylbenzene showed clear emission differences dependent on the differentiation conditions, a specific link to metabolic processes might be possible.

Our data indicated that ASCs undergoing adipogenic differentiation increased the emission of acetaldehyde, whereas nondifferentiated ASCs reduced the emissions down to almost zero. Acetaldehyde emissions were nearly three times higher on day 21 of cultivation in adipogenically differentiating ASCs compared with nondifferentiating cultures. When taking into account the cell number, which is lower in adipogenically differentiating ASCs due to a stop in proliferation [12], this effect becomes even more distinctive. Thus, in our study, acetaldehyde is the VOC with the most variable differentiation-dependent behavior.

Acetaldehyde might be connected to mitochondrial pyruvate decarboxylation [33,34]. Whether acetaldehyde is consumed or produced could indicate the demands for intracellular acetyl-CoA. The energy metabolism of nonstimulated cells and adipogenically stimulated ASCs might be based predominantly on carbohydrate oxidation with a flux of pyruvate to acetyl-CoA. Under these conditions, it is possible that acetaldehyde can be used as well for acetyl-CoA production.

However, acetaldehyde can also be the product of ethanol degradation. Since some differentiation-inducing additives are dissolved in ethanol, a general, slightly increased level of acetaldehyde emissions compared with ASCs in maintenance medium might result from higher ethanol levels in the differentiation medium. This might be reflected in the increase of acetaldehyde emissions in the medium controls at an early time point (up to seven days). Ethanol dehydrogenase activity was detected in brown adipose tissue of rats [35]. Whether ethanol dehydrogenase is also present in the ASCs of white adipose tissue needs to be determined.

1,3-di-tert-butylbenzene emissions showed a significant increase in adipogenically differentiating ASCs compared to nondifferentiating ASCs. Tang et al. [36] found 1,3-di-tert-butylbenzene emitted differentially in cancer cells lines compared with nonmalignant cells. No further information is available about this substance in the current literature.

When normalized to cell numbers, pentane concentrations were higher in adipogenically differentiating ASCs than in nondifferentiating ones. Pentane is produced through the peroxidation of lipids and other biomolecules under the action of reactive oxygen species [37]. As shown before, adipogenic differentiation of ASCs correlates with an increase of the antioxidative defense activities of glutathione reductase [12]. This could conform with our current findings, which indicate an increase in the production of reactive oxygen species in adipogenically differentiating cells, and which could be followed by an increase in the antioxidative defense systems of ASCs.

Ethylbenzene showed higher concentrations in the headspace of medium controls than in the headspace of cultivated cells. In general, lower and less-deviating concentrations were observed in the headspaces of adipogenically differentiating ASCs compared with nondifferentiating ASCs. We, thus, conclude that cells in culture with and without differentiation stimulation consume or bind this compound. Previous studies established that ethylbenzene undergoes α- and ω-oxidation [38].

The aldehydes benzaldehyde, heptanal, and octanal were also consumed or bound by the cells from the culture media under both cultivation conditions. A consumption of aldehydes from the culture media might be based on the activity of aldehyde dehydrogenase [39,40] (e.g., benzaldehyde concentrations were reduced in the headspace of all cells but were even lower in adipogenically differentiating than in nondifferentiating cells). Zimmermann et al. [41] described NAD-aldehyde dehydrogenase as a possible enzyme for metabolizing benzaldehyde in fatty acid metabolism and tryptophan metabolism [42,43]. Amino acids are necessary for proliferation and differentiation, and their consumption was also observed in stem cell lines [27].

Heptanal and octanal showed relatively higher concentrations on day 21 in the adipogenically stimulated ASC cultures when related to the cell number. Since heptanal can be a product of lipid peroxidation [44,45], increasing enzyme activity for lipid oxidation during differentiation could lead to a higher turnover of heptanal. Other studies linked octanal to oxidative stress [46,47]. While this process includes lipid oxidation, octanal might increase during adipogenic differentiation due to an increase of the oxidation processes.

#### **5. Conclusions**

In this study we were able to identify differentially emitted VOCs, which might be linked to a changed metabolic activity in ASCs dependent on cellular differentiation. Whether these changes are driven by specific, enzyme-mediated processes or are the result of unspecific reactions has to be investigated. Irrespective of the fact that these are early stage results, the data presented are in agreement with the results of adipogenic-differentiating cells to date. Thus, further experiments might reveal if headspace VOC measurements serve as a marker for the quality and quantity of adipogenic differentiation of ASCs.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/7/697/s1, Table S1: Quantitative parameters of VOCs: Concentration ranges, limits of detection (LOD), limits of quantification (LOQ) and standard deviation (SD) of 12 substances detected in ASC and pure culture medium, Table S2: P-values of statistical comparison between non- differentiating ASC (non-diff ASC) and adipogenically differentiating (adipo ASC), Figure S1: VOCs during the adipogenic differentiation. Figure S2: Vocs of nondifferentiating ASC.

**Author Contributions:** Conceptualization, A.-C.K. and K.P.; methodology, A.-C.K., S.T., J.M., and K.P.; formal analysis, A.-C.K., J.B., K.E., and J.M.; investigation, A.-C.K., J.B., S.T., J.M., and K.E.; resources, W.M., J.K.S., and K.P.; data curation, A.-C.K.; writing—original draft preparation, A.-C.K., J.B., J.M., and K.P.; writing—review and editing, A.-C.K., K.P., W.M., J.K.S., K.E., and G.K.; visualization, A.-C.K., J.M., and S.T.; supervision, K.P., W.M., J.K.S., and G.K.; funding acquisition, A.-C.K. and K.P.

**Funding:** This research was funded by a scholarship of the Landesgraduiertenfoerderung MV and the European Regional Development Fund (ERDF) together with the Federal State of Mecklenburg-Vorpommern ("Entwicklung eines Systems zur automatisierten Zellfraktionierung aus humanem Fettgewebe für neuartige regenerative Anwendungen und Therapien (ARENA)", TBI-V-003-VBU-001).

**Acknowledgments:** The authors thank Jürgen Weber (Ästhetikklinik Rostock, Germany), Klaus Ueberreiter (Park-Klinik Birkenwerder, Germany) and their patients for kindly providing liposuction tissue. Special thanks are also given to Nina Gehm and Anne Wolff for their technical support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Collagen I Promotes Adipocytogenesis in Adipose-Derived Stem Cells In Vitro**

#### **Nadja Zöller, Sarah Schreiner, Laura Petry, Stephanie Ho**ff**mann, Katja Steinhorst, Johannes Kleemann, Manuel Jäger, Roland Kaufmann, Markus Meissner and Stefan Kippenberger \***

Department of Dermatology, Venereology and Allergy, Johann Wolfgang Goethe University, Theodor-Stern-Kai 7, D-60590 Frankfurt/Main, Germany; Nadja.Zoeller@kgu.de (N.Z.); S\_Schreiner@gmx.de (S.S.); Laura.Petry@kgu.de (L.P.); Stephanie.Hoffmann@kgu.de (S.H.); katja.Steinhorst@kgu.de (K.S.); Johannes.Kleemann@kgu.de (J.K.); Manuel.Jaeger@kgu.de (M.J.); kaufmann@em.uni-frankfurt.de (R.K.); Markus.Meissner@kgu.de (M.M.) **\*** Correspondence: kippenberger@em.uni-frankfurt.de; Tel.: +49-69-6301-7734

Received: 7 March 2019; Accepted: 28 March 2019; Published: 1 April 2019

**Abstract:** A hallmark of ageing is the redistribution of body fat. Particularly, subcutaneous fat decreases paralleled by a decrease of skin collagen I are typical for age-related skin atrophy. In this paper, we hypothesize that collagen I may be a relevant molecule stimulating the differentiation of adipose-derived stem cells (ASCs) into adipocytes augmenting subcutaneous fat. In this context lipogenesis, adiponectin, and collagen I receptor expression were determined. Freshly isolated ASCs were characterized by stemness-associated surface markers by FACS analysis and then transdifferentiated into adipocytes by specific medium supplements. Lipogenesis was evaluated using Nile Red staining and documented by fluorescence microscopy or quantitatively measured by using a multiwell spectrofluorometer. Expression of adiponectin was measured by real-time RT-PCR and in cell-free supernatants by ELISA, and expression of collagen I receptors was observed by western blot analysis. It was found that supports coated with collagen I promote cell adhesion and lipogenesis of ASCs. Interestingly, a reverse correlation to adiponectin expression was observed. Moreover, we found upregulation of the collagen receptor, discoidin domain-containing receptor 2; receptors of the integrin family were absent or downregulated. These findings indicate that collagen I is able to modulate lipogenesis and adiponectin expression and therefore may contribute to metabolic dysfunctions associated with ageing.

**Keywords:** adipose-derived stem cells; adipocytes; differentiation; collagen I; adiponectin; integrins; discoidin domain-containing receptor; ageing; subcutaneous fat

#### **1. Introduction**

Adipose-derived stem cells (ASCs), residing within fat tissue, display multilineage plasticity, which makes them interesting for regenerative medicine. Minimal invasive procedures allowing liposuction facilitate access, particularly to superficial adipose tissue. After purification and plating, the fate of native ASCs can be directed by specific culture conditions. As a result, different phenotypes can be induced that are characteristic for adipocytes [1,2], fibroblasts [3], endothelial cells [4], osteoblasts [5], chondrocytes [6], cardio-myocytes [7], neural-like cells [2], hepatocytes [8], pancreatic cells [9], and others.

Skin ageing is triggered by intrinsic and extrinsic factors, leading to a massive remodeling of the extracellular matrix (ECM) [10,11]. In this study, we hypothesize that the age-related decrease of ECM molecules also has an impact on the adipogenic differentiation potential in ASCs. In this context, we focus on the role of collagen I, which is the most prominent constituent of ECM molecules distributed throughout the interstitium, making up to 90% of the total connective tissue. Besides providing mechanical rigidity, collagen fibrils also act as guidance structures for contacting cells. Among cellular collagen receptors, the surface receptors of the integrin family are most prominent. They are composed of noncovalently associated α and β subunits forming a heterodimer that recognizes specific amino acid motifs within collagen molecules, hereby initiating intracellular signaling cascades. Besides integrins, two receptor tyrosine kinases, namely discoidin domain receptors 1 and 2 (DDR1 and DDR2), have also been identified to be activated upon collagen binding [12,13]. Particularly, DDR2 is characteristic for cells of mesenchymal origin, such as fibroblasts and smooth muscle cells [14].

In order to characterize adipogenic differentiation, we put particular emphasis on lipogenesis by measuring the lipid content. Moreover, another characteristic of adipocytes is the signature pattern of secreted factors that have been collectively termed 'adipokines'. Prominent among those are leptin, adiponectin, resistin, and visfatin, as well as cytokines and chemokines, such as tumor necrosis factor-α and interleukin-6 [15]. It is assumed that these factors contribute to a subinflammatory state that triggers the development of many chronic obesity-correlated diseases. In this present paper, we focus on adiponectin, a multifaceted adipokine, known as a modulator of inflammation [16,17].

The data presented in this paper show that the contact of ASCs to collagen I is a relevant factor in adipogenic differentiation as characterized by lipogenesis and adiponectin secretion. It could be speculated that collagen synthesis under physiological conditions might control the fate of ASCs.

#### **2. Materials and Methods**

#### *2.1. Ethics Statement*

This study was conducted according to the Declaration of Helsinki Principles and in agreement with the Local Ethic Commission of the faculty of Medicine of the Johann Wolfgang Goethe University (Frankfurt am Main, Germany). The Local Ethic Commission waived the need for consent.

#### *2.2. Isolation and Characterization of ASCs*

Isolation and initiation of human ASC cultures were performed as described [18]. As source served abdominal subcutaneous fat tissue derived from plastic surgeries, generously provided by Dr. Ulrich Rieger (Klinik für Plastische und Ästhetische Chirurgie, Wiederherstellungs und Handchirurgie, Markus Krankenhaus, Frankfurt/Main, Germany). The fat tissue was placed in PBS with 2% penicillin/streptomycin (Biochrom, Berlin, Germany) incubated overnight (4 ◦C). On the next day, skin and blood vessels were mechanically removed by scissors and forceps. Small pieces, with approximately 5 mm lengths, were given to a collagenase type I solution (Worthington, Lakewood, USA) and incubated for 3 h at 37 ◦C. Cell debris was discarded by filtration through sterile gauze. Then, the cell suspension was centrifuged (400× *g*, 6 min, 4 ◦C), the cell pellet resuspended in medium, and passed through a cell strainer (70 μm, Greiner, Frickenhausen, Germany). Next, cells were separated by density filtration using a Biocoll solution with a specific density of 1.077 g/mL. After another centrifugation (400× *g*, 30 min, 4 ◦C), ASCs were isolated from the opaque interphase and seeded in DMEM supplemented with 1% UltroSerG (Pall, Dreieich, Germany) and 1% penicillin/streptomycin. The medium was renewed every 3 days.

#### *2.3. Flow Cytometry*

Phenotypical characterization of ASCs was performed using the BD FACSCalibur and analyzed with the CellQuest software (v. 1.0.1, Becton–Dickinson, Heidelberg, Germany). The cells were trypsinized, placed on ice for 30 min, and treated with the following labeled stemness-associated antigen markers: CD31-, CD34+, CD45–, CD54–, CD90+, CD105+, CD166+, HLA-ABC+, HLA-DR–. CD34-PE, CD90–FITC, and CD105-PerCP were part of the BD Stemflow Kit (Becton Dickinson, 562245). The other antibodies with the indicated specifications were purchased separately: CD31-FITC (R & D, Systems, Wiesbaden, Germany, FAB3567F), HLA-DR-PE (Becton Dickinson, 347401), CD166-PE

(Becton Dickinson, 559263), CD34-APC (Becton Dickinson, 555824), CD54-FITC (Beckman Coulter, Krefeld, Germany, PN IM0726U), HLA-ABC-FITC (Becton Dickinson, 555552), and CD45-PE (Becton Dickinson, 555483). Table 1 shows the results from three donors. All experiments shown in this paper were performed with cells until reaching passage 5.


**Table 1.** Characterization of ASCs.

#### *2.4. Adhesion Assay*

For the adhesion experiments, ASCs were trypsinized and seeded at a concentration of <sup>1</sup> <sup>×</sup> <sup>10</sup><sup>4</sup> cells/well onto microtiter plates coated with collagen I (Becton-Dickinson, Heidelberg, Germany). The coating procedure was performed as previously described [19]. Briefly, collagen type I was dissolved with sterile acetic acid and given in the specific vessels (5 mg/cm2). After drying, solvent and collagen remnants were removed by washing with PBS. Regular plastic dishes served as controls. After seeding, the cells were allowed to attach for 30, 60, 90, and 120 min at 37 ◦C. Then, non-anchored cells were removed by two washings with PBS, and the nuclei of anchored cells were stained with the DNA-binding fluorochrome bisbenzimide (2 mg/mL, 20 min, RT). After two washing steps, fluorescence was detected using the CytoFluor multi-well plate reader (Applied Biosystems, Langen, Germany) at 360/460 nm [19,20]. Experiments were performed three times with three triplicates.

#### *2.5. Induction of Adipogenesis*

ASCs were seeded in 24-well culture dishes at a concentration 5 <sup>×</sup> 104 cells/well. Adipogenic differentiation was induced by Adi-medium consisting of DMEM supplemented with 10 μM insulin, 0.5 mM 3-isobutyl-1-methylxanthine, 1 μM dexamethasone, 200 μM indomethacine, 1% penicillin/streptomycin solution, and 2% UltroserG [2,18].

#### *2.6. Detection of Lipids*

ASCs were cultured in Adi-medium (or standard medium) on collagen I-coated and non-coated supports for 9, 11, and 15 days. Consecutively, the lipid content was quantitatively detected as described [21]. Briefly, cultures were washed with PBS and then stained with nile red (1 μg/mL, 20 min, 37 ◦C). After two washing steps with PBS, fluorescence documented by fluorescence microscopy or quantitatively measured using a multiwell spectrofluorometer (Cytofluor, Applied Biosystems, Langen, Germany) equipped with 485/560 nm filters for neutral lipids. Moreover, cell nuclei were stained with bisbenzimide (2 μg/mL, 20 min, 37.4 ◦C) and the fluorescence, as a measure for cell count, was detected at 360/460 nm. In order to normalize the lipid values to different cell counts, a ratio was formed between nile red and bisbenzimide measurements. The values derived from cells cultured on plastic were set to 100% and all other values were related to that. Experiments were performed in quadruplicate, with 4 parallel determinations for each condition.

#### *2.7. Western Blot*

Cells cultured as described above were lysed in 100 mL SDS sample buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS, 10% glycerol, 50 mM DTT, 0.1% bromphenol blue), sonicated and boiled for 5 min, and separated on SDS-polyacrylamide gels. Consecutively, proteins were immunoblotted to a PVDF membrane. The membrane was blocked in blocking buffer (TBS [pH 7.6], 0.1% Tween-20, 5% nonfat dry milk) for at least 3 h at 4 ◦C followed by incubation with the primary antibody in TBS (pH 7.6), 0.05% Tween-20, and 5% BSA. Bound primary antibodies were detected using anti-mouse IgG-horseradish peroxidase conjugate and visualized with the LumiGlo detection system (Cell Signaling, Frankfurt, Germany). The following primary antibodies were used: Integrin β1 (Santa Cruz, sc-6622), integrin α1 (R & D, MAB5676-SP), integrin α2 (Becton-Dickinson, 611016), integrin α11 (R & D, AF4235-SP), and discoidin domain-containing receptor 2 (DDR2) (R & D, AF2538).

#### *2.8. Adiponectin ELISA*

Cells were seeded on collagen or plastic supports and exposed to either Adi-medium (or standard medium) for 9, 11, and 15 days. Cell-free supernatants were obtained and assayed for human adiponectin using a commercial ELISA test kit (R & D) according to the manufacturers' instructions. Briefly, supernatants were placed in microwell plates coated with antibodies against adiponectin. After incubation with a biotin-labeled secondary antibody, a streptavidin horseradish-peroxidase conjugate was added. A colorimetric reaction was exerted by addition of a substrate (tetramethylbenzidine/peroxide), giving rise to a colored product measured at 450 nm in a scanning multiwell spectrophotometer (ELISA reader MR 5000, Dynatech, Guernsey, UK).

#### *2.9. Real-Time RT-PCR Analysis*

Total cellular RNA was isolated from cells cultured as described above using the ExtractMe Total RNA Kit (Blirt, Gdansk, Poland). After DNase digestion, a total amount of 25 ng RNA was used for first-strand cDNA synthesis using a QuantiTect SYBR Green RT-PCR Kit (Qiagen, Hilden, Germany). Real-time PCRs were performed on a Light Cycler system 2.0 (Roche Diagnostics, Mannheim, Germany). The following primers for adiponectin (accession no. ENST00000320741.6) and PBDG (accession no. ENST00000278715) were used:

Adiponectin-for: 5 -TGTGGTTCTGATTCCATACCAG-3 Adiponectin-rev: 5 -CGGGCAGAGCTAATAGCAGTA-3 PBDG-for: 5 -CCATGTCTGGTAACGGCAAT-3 PBDG-rev: 5 -GTCTGTATGCGAGCAAGCTG-3 The relative expression of transcripts was determined using the 2-ΔΔCT method [22].

#### *2.10. Statistics*

All data are presented as mean values ± standard deviation. Statistical significance of the data was calculated by Wilcoxon-Mann-Whitney-U-test (BIAS, Frankfurt, Germany). Data sets were statistically compared as indicated in the graphics.

#### **3. Results**

Although most cell cultures were carried out on uncoated plastic dishes, it is known that extracellular matrix molecules have impacts on cell physiology. Therefore, we initially tested the adherence of ASCs held in standard medium to collagen I in comparison to uncoated plastic (Figure 1). Representative photographs of time-dependent adhesion are shown in Figure S1. It was found that ASCs in suspension feature a significant faster adherence to collagen I-coated substrates in the first 30 min. In the further course, the cell count adherent to either plastic or collagen I was similar in the observed time span. These results indicate that naïve ASCs also express adhesion molecules with avidity to collagen I.

**Figure 1.** Quantitative adhesion of ASCs on plastic vs collagen I. ASCs were plated at a density of <sup>1</sup> <sup>×</sup> 104 cells per well into microtiterplates either non-coated or coated with collagen I. After 30, 60, 90, and 120 min, non-adherent cells were discarded and the remaining cells were stained with bisbenzimide. Fluorescence, as a measure for adherent cells, was quantified using a CytoFluor multi-well plate reader at 360/460 nm. Each point represents the mean of five independent experiments. The standard deviations are indicated. \* *p* < 0.05. The whole experiment was repeated three times with similar results.

Furthermore, we investigated the impact of adhesion to collagen I on cell morphology and differentiation (Figure 2). Cells were seeded on supports coated with collagen I or plastic for control and then cultured for 19 days in either standard medium (DMEM) or adipogenic medium (Adi-medium). On plastic (Figure 2A), ASCs feature a domed cell center with filigree dendritic plasma extrusions in the periphery. These dendritic branches are less pronounced in cells on collagen I (Figure 2B). When cells were held in adipogenic medium, the cell morphology changed massively. Cells form a large interconnected network with many extended cell branches reminding on a network of nerve cells (Figure 2C,D). Moreover, cells accumulate tightly packed vesicles in the cell center, which putatively represent lipid droplets. In order to verify the nature of these droplets, a nile red staining was performed (Figure 2E–H). Positive nile staining was found around the cell nucleus, stained in blue with bisbenzimide, indicating the presence of lipids. Of note, on the photographs, the expression of lipids seems to be more distinct in cells cultured on collagen I.

**Figure 2.** *Cont*.

**Figure 2.** ASCs cultured on plastic or collagen I. (**A**,**B**) show the cell morphology after cultivation in ASC standard medium (ctr) for 19 days, (**C**,**D**) after cultivation in adipogenic medium (Adi). (**E**–**H**) display cellular lipids (red) after cultivation in adipogenic medium for 19 days stained with the fluorescence dye nile red; cell nuclei (blue) are stained with bisbenzimide. Photographs show representative sections.

In order to validate this first impression, nile red staining was quantified at different time points (9, 11, 15 days) by fluorometric means as described. Figure 3 shows the content of neutral lipids of ASCs in dependence to culture medium (DMEM vs Adi-medium) and cell support (plastic vs collagen I). The value determined for ASCs cultured in DMEM on plastic was set to 100%. In these cells, which are absent of lipid droplets, nile red stains showed lipid-containing membranes only. All other measured values were related to this. It was found that cultivation on collagen I led to a slight, but significant increase in lipid content at all examined time points in DMEM (black bars). Moreover, the change to Adi-medium induced the expected massive and significant increase in lipids in cells on plastic and collagen I (striped bars). Of note, cultures on collagen I showed an increase in lipids compared to their counterparts held on plastic. In sum, these results indicated a lipogenic effect of collagen I in ASCs.

**Figure 3.** Quantitative lipogenesis of ASCs on plastic or collagen I. ASCs were plated at a density of <sup>1</sup> <sup>×</sup> 104 cells per well into microtiterplates either non-coated (plastic) or coated with collagen I. One part was held in regular DMEM medium (ctr) and the other in adipogenic medium (Adi). After (**A**) 9 days, (**B**) 11 days, and (**C**) 15 days, cells were stained with nile red and bisbenzimide and the fluorescence was measured (see Materials and Methods). Each point represents the mean of 16 independent experiments. The standard deviations are indicated. Data sets were statistically compared as indicated. \* *p* < 0.05.

Next, the release of adiponectin, a prototypical adipokine, was evaluated in cell-free supernatants by ELISA after 9, 11, and 15 days (Figure 4A–C). ASCs cultured in DMEM showed no significant release of adiponectin in the period under observation regardless of whether they were cultured on plastic or collagen I (black bars). Changing the medium to Adi-medium induced a massive release of adiponectin (striped bars). Comparison of both cell substrates shows that the measured levels of adiponectin on collagen I were significantly reduced.

**Figure 4.** Adiponectin release of ASCs on plastic or collagen I. ASCs were plated at a density of <sup>1</sup> <sup>×</sup> 104 cells per well into microtiterplates either non-coated (plastic) or coated with collagen I. One part was held in regular DMEM medium (ctr) and the other in adipogenic medium (Adi). After (**A**) 9 days, (**B**) 11 days, and (**C**) 15 days, supernatants were examined for adiponectin by ELISA. Each point represents the mean of 20 independent experiments. The standard deviations are indicated. Data sets of Adi/plastic and Adi/collagen I were statistically compared as indicated. \* *p* < 0.05.

To learn more about the level of this regulation, mRNA transcripts of adiponectin were quantified by real-time RT-PCR analysis (Figure 5). As found for the adiponectin release, almost no adiponectin mRNA expression was found in ASCs held in DMEM (black bars). In contrast, massive induction was initiated by changing the medium to Adi-medium (striped bars). Similar to the measured protein levels, this induction was gradually diminished by cultivation on supports coated with collagen I. The measured Ct values ranged from ca. (circa) 29 for non-differentiated to ca. 22 for differentiated cells. The Ct values for reference gene (PBGD) expression were ca. 27 independent from culture conditions.

**Figure 5.** Adiponectin mRNA expression of ASCs on plastic or collagen I. ASCs were plated at a density of 1 <sup>×</sup> 104 cells per well into microtiterplates either non-coated (plastic) or coated with collagen I. One part was held in regular DMEM medium (ctr) and the other in adipogenic medium (Adi). After (**A**) 9 days, (**B**) 11 days, and (**C**) 15 days, total RNA was extracted and real-time RT-PCR analysis was performed. The values for ASCs in DMEM are less than 1%. Data sets of Adi/plastic and Adi/collagen I were statistically compared as indicated. \* *p* < 0.05.

Our results demonstrate that collagen I has an impact on the cell physiology in ASCs. Therefore, the expression of collagen receptors was determined by western blot analysis (Figure 6). Surface receptors of the integrin family form heterodimers composed of α and β subunits, which convey substrate specific recognition. For the recognition of collagens, the α1, α2, α11, and β1 subunits are relevant. Positive controls for the detection of α1 and α11 were protein lysates derived from

normal human fibroblasts; for α2 and β1, lysates from HaCaT keratinocytes were used. It was found that ASCs in DMEM produce distinct amounts of α1 on plastic as well as on collagen I (Figure 6A). A change to Adi-medium significantly reduced this expression. Likewise, α11 was massively reduced in ASCs cultured in Adi-medium. Interestingly, the basal level in ASCs held in DMEM was in the range of the positive control. Integrin α2 was only present in the positive control. The β1 subunit was even more pronounced than in the positive control, particularly in ASCs held in DMEM. A shift to Adi-medium caused a decrease in β1 expression. Moreover, DDR2, another collagen receptor present in mesenchymal cells, was investigated (Figure 6B). Here, a moderate expression in ASCs cultured in DMEM was found. Of note, a shift to Adi-medium caused a profound upregulation. Lysates derived from fibroblasts served as a positive control.

**Figure 6.** Expression of collagen I receptor molecules. ASCs, plated on regular plastic supports or supports coated with collagen I, were held for 9 days in standard DMEM medium (ctr) or in adipogenic medium (Adi). Protein extracts were subjected to western blot and tested for (**A**) α1, α2, α11, β1 integrin, and (**B**) DDR2. Equal loading was monitored by using antibodies directed against α-tubulin. The blots show representative results (*n* = 3). + indicates positive controls.

#### **4. Discussion**

Collagen I, as the most abundant ECM molecule, is produced by fibroblasts. It is known to provide a structural scaffold for cell attachment with impacts on tissue organization and tissue homeostasis by affecting cell growth, motility, viability, and differentiation [23]. The omnipresence of collagen in the mesenchym makes this molecule interesting as a potential trigger factor determining the cell's fate. It is known that the amount of collagen decreases during ageing with implications to the above mentioned physiological parameters [24]. An imbalance of the production and degradation leads to the accumulation of fragmented collagen molecules on which fibroblast cannot efficiently attach to. The absence of a mechanical load, which is a trigger factor for proliferation [25], marks the entry to a self-perpetuating detrimental cycle, accelerating the decrease of collagen [24]. Moreover, another hallmark of ageing is the redistribution of fat, with a decrease of peripheral subcutaneous fat [26]. It is an interesting issue if both observations are somehow functionally linked. ASCs residing in subcutaneous fat tissue have the potential to transdifferentiate into adipocytes, serving to compensate the age-dependent volume loss. In this paper, we showed that ASCs produce more lipids, a marker for adipogenic differentiation, when cultured in collagen coated dishes. This effect was measured both in cells cultured in standard medium containing no differentiation factors, and also in cells cultured in Adi-medium, described to promote adipogenic differentiation. Against this background, the stimulation of collagen synthesis by topical retinoic acid [27–29] or carbon dioxide laser resurfacing [30] may also support the differentiation of ASCs into adipocytes. Furthermore, injectable preparations containing animal collagen, which are widely used in cosmetic medicine as injectable fillers to increase lip volume, to minimize wrinkles, or to correct post-acne and traumatic scars [31], impact this differentiation process.

In addition to lipogenesis, the expression of adiponectin was also examined, a fat-derived hormone that is predominantly produced by adipocytes [32]. Corroboratively, ASCs in standard medium produce no adiponectin while cultivation in Adi-medium leads to a massive induction, indicating successful transdifferentiaton to adipocytes. Although adiponectin has pleiotropic metabolic effects, it is mainly known for its action as an insulin sensitizer with anti-apoptotic and anti-inflammatory properties [33–35]. Interestingly, the presence of collagen suppressed the expression of adiponectin. In humans, an inverse correlation between fat mass and adiponectin plasma levels was observed [36,37]. In this context, the inhibitory effect of collagen may represent a countermeasure in response to the massive induction of lipogenesis.

After we learnt that collagen impacts on metabolic parameters in ASCs, we looked at the potent receptors conveying such a signal, namely integrins and DDR2. In higher vertebrates, there are 18 α subunits and 8 β subunits, which form 24 distinct integrins. Specific for collagen binding are the following heterodimers: α1β1, α2β1, α10β1, and α11β1 [38]. Therefore, in western blot analysis, the expression of these integrins were investigated. An inverse regulation between lipogenesis/adiponectin expression and the expression of α1, α11, and β1 was found (α2 was not detectable in ASCs independent from culture conditions). In contrary, DDR2 was massively induced in ASCs when transdifferentiated to adipocytes. These findings suggest a contribution of this receptor in the collagen I-induced regulations of lipogenesis and adiponectin expression.

In sum, the presented results identified collagen I as a new regulator of adipogenic differentiation of ASCs. These findings have implications for the understanding of changes in age-related fat metabolism.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/4/302/s1, Figure S1: Time-dependent cell adhesion of ASCs to plastic vs collagen I.

**Author Contributions:** Conceptualization, S.K., M.M., R.K., J.K., N.Z. and M.J.; Methodology, S.S., L.P., S.H., K.S., Writing—Original Draft Preparation, S.K.

**Acknowledgments:** We are grateful to Ulrich Rieger (Klinik für Plastische und Ästhetische Chirurgie, Wiederherstellungs- und Handchirurgie, Markus Krankenhaus, Frankfurt/Main, Germany) for providing us with fat tissue.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
