Expansion of Semi-Automatic Processed Human Adipose-Derived Stem Cells in Medium Supplemented with Autologous Serum and Antioxidants

1Department of Medical Research, Buddhist Tzu Chi General Hospital, Hualien, Taiwan 2Institute of Medical Sciences, Tzu Chi University, Hualien, Taiwan 3Department of Hematology and Oncology, Taichung Tzu Chi General Hospital, Taichung, Taiwan 4Department of Obstetrics and Gynecology, Min-Sheng General Hospital, Taoyuan, Taiwan 5Institute of Biomedical Sciences, Academia Sinica, Taipei, Taiwan 6Center for Medical Genetics, Buddhist Tzu Chi General Hospital, Hualien, Taiwan 7Department of Surgery, Taipei Tzu Chi General Hospital, Taipei, Taiwan 8Department of Medicine, College of Medicine, Tzu-Chi University, Hualien, Taiwan #These authors contributed equally to this study

ADSCs derived from the direct excision of orbital fat mass [3,16] and buccal fat pad excised by surgery [17] compared to those isolated from lipoaspirated abdominal subcutaneous fat have similar differentiation potential [18]. Although the size of excised fat tissue (EFT) samples are commonly smaller, ADSC yields from EFT are often greater than that from AFT [6], suggesting that EFT could be an alternative source of ADSCs for individuals with lower BMI [3,6,16,18]. Currently, EFT must be minced manually with surgical scissors in a biological safety cabinet prior to further incubation with collagenase for SVF isolation [3,18]. The manual method presents major limitations including a lengthier time frame required to process the tissue for the extraction of SVF and variability in the resulting isolated cells due to intraoperator processing. The repeated use of surgical scissors also presents risks of cross-contamination. Several bench-top closed systems have been developed for SVF isolation from AFT [1], including the Cytori Celution™ system [19] and the Tissue Genesis' TGI 1000™. However, an appropriate automated system for the isolation of SVF or ADSCs from EFT has yet to be developed.
Fetal bovine serum (FBS) are commonly used for the ex vivo expansion of MSCs, which is a potential source of zoonotic or other infectious agents such as prions, viruses and mycoplasma. FBS is a chemically ill-defined supplement with high batch-to-batch variability, which may lead to unpredictable cell growth or variation in differentiation [20][21][22][23]. On the other hand, the expansion of MSCs using human serum-supplemented medium is less efficient [24][25]. Several serum-free media (SFMs) have been recognized to be safe and free from animal products and related issues [26]. SFMs also do not affect the immunophenotype, proliferation potential and multipotency of ADSCs [22,[27][28][29][30][31][32][33]. These SFMs include MesenCult ® -XF Medium (Stemcell Technologies, Canada) [34], StemPro ® MSC SFM (Life Technologies, US), MSC Nutristem ® XF Medium (Biological Industries, Israel) and BD Mosaic™ (Becton Dickinson, US). However, the cost of these SFMs and their lack of clinical validation limit their clinical applications.
Autologous cell transplantation is the best choice for patients who require cell therapy, where EFT harbors an excellent source of autologous ADSCs. We developed a semi-automated process complying with Good Tissue Practice (GTP) and Good Manufacturing Practice (GMP) guidelines to replace the current manual techniques to minimize human-induced variability and comply with clinical trial standards. In addition, we have also formulated a human autologous serum (HAS)-and antioxidant-supplemented medium for the expansion of ADSCs, and compared it with media containing either animal sera or HAS alone.

Clean room
Procedures for the isolation of SVF and expansion of ADSCs from EFT were conducted in Biological Safety Cabinets (Model 1286, Thermo Scientific Forma Class II Type A2, Thermo Scientific, Waltham, MA) in a clean room (ISO14644 Class 7) located in Gene and Stem Cell Production Laboratory of Hualien Buddhist Tzu Chi General Hospital. Each suite is supported by positive-pressure (10-15 Pa), HEPA-filtered air, at a temperature of 22°C and relative humidity of 55-65%. The flow suite is unidirectional, with entry and exit air locks. All personnel were trained in respect to the GMP guidelines.

SVF isolation from EFT
The experimental protocol was approved by the Institutional Review Board (IRB) of Min-Sheng General Hospital, Taoyuan, Taiwan. EFT from abdominal subcutaneous fat was obtained from 11 pregnant women (average age of 33.7 ± 3.3). About 25 grams of adipose tissues were collected from each donor, and the tissues were minced either by disposable disperser tubes (DT-20 gamma, IKA ULTRA TURRAX Tube drive, Staufen, Germany) or manually with surgical scissors. The process of isolation is shown in Figure 1. About 3 grams of EFT was resuspended in 12 mL of Iscove's modified Dulbecco's medium (IMDM; GIBCO-Invitrogen, Grand Island, NY), placed in a 15 mL disposable disperser tube, and homogenized by centrifugation at 1200 rpm for 10 seconds thrice. Collagenase (Animal Origin Free)-A (LS004154, Worthington Biochemical Corporation, Lakewood, NJ; final concentration: 0.4 mg/mL, volume: 15 mL) was added for enzymatic digestion, and incubated in MO-01 hybridization oven (15 rpm, 37°C; Double Eagle Enterprise, New Taipei City, Taiwan) for 8 hours ( Figure 1D). After dispersing the digested adipose tissues several times with a 10 ml sterile pipette (CORNING, Tewksbury, MA), the digested tissues were centrifuged at 2000 rpm for 10 min to generate SVF pellets ( Figure 1E and 1F). The SVF pellet was then resuspended in PBS and filtered through a 70 μm strainer (Becton Dickinson, San Jose, CA) and the SVF cells were ready for ADSCs culture.

Cell number and viability assays
For cell number and viability assays of SVF, cells were counted using an automatic cell counter, ADAM-MC (NanoEnTek, Seoul, Korea), with Propidium Iodide (PI) according to the manufacturer's instructions. For cell proliferation and viability assays, ADSCs were detached by trypsinization and then counted in triplicates using Vi-CELL AS cell counter (Beckman Coulter, Indianapolis, IN) as previously described [35]. For total ADSCs yield, the attached cells from SVF culture under various expansion media were detached by exposure to trypsin-EDTA on the 5 th day.

Flow cytometric analysis
For the characterization of cell surface phenotype, the ADSCs were cultured in 75 cm 2 tissue culture flasks (Becton Dickinson) at an initial density of 1000 cells/cm 2 in a humidified 5% CO 2 and 37°C incubator. After 6 days, single-cell suspensions were obtained by trypsinization, resuspended in culture medium, and further incubated at 37°C in CO 2 incubator for 30 minutes. Cells were stained with anti-human CD13, CD34, CD44, CD73, CD90 and CD105 (Becton Dickinson) conjugated with fluorescein isothiocyanate or R-phycoerythrin for 1 hour at 4°C (for negative controls, primary antibodies were omitted). The expression profiles were examined by flow cytometry (FACSCalibur, Becton Dickinson) and analyzed by CXP Software (Beckman Coulter, CA).

Analysis of gene expression and karyotyping
After incubation under different culture media for 6 days (n=3 for each medium, initial cell density of 1000 cells/cm 2 ), the ADSCs were harvested to examine the expression of stemness and growth factor genes. Isolation of total RNA, cDNA synthesis and quantitative polymerase chain reaction (qPCR) were carried out as previously described [36]. Briefly, total RNA was isolated from cells using the RNeasy kit (Qiagen, Valencia, CA), and cDNA was synthesized from 10 μg of total RNA using the Advantage RT-for-PCR kit (Clontech, Palo Alto, CA). qPCR and product detection were performed using the Fast Start Essential DNA Green Master (Roche, Indianapolis, IN) and the PikoReal Real-Time PCR System (Thermo Scientific) with the gene-specific primers as listed in Table 1. The results of amplification were analyzed using PikoReal 2.0 software (Thermo Scientific, PA). Gene expression levels were normalized to β-actin, which served as the internal control.
For karyotyping analysis, 75 cm 2 tissue culture flasks seeded with ADSCs in 10% HAS+MCA-supplemented medium with an initial density of 3000 cells/cm 2 were placed in a CO 2 incubator. Once the cells reached 50-60% confluence, they were washed twice with phosphate buffered saline (PBS), detached by 0.25% trypsin-EDTA (Gibco BRL) and then replated at a density of 3000 cells/cm 2 under the same culture conditions. Karyotyping of ADSCs was performed following one month culture by established protocols [37].

In vitro differentiation
After incubation under various culture media conditions (10% FBS, 10% HAS, 10% HAS+MCA or 2% HAS+MCA, n=3 for each medium) for 6 days, the ADSCs were seeded into 6-well plates (Becton Dickinson) for differentiation assays. Osteogenesis, adipogenesis and chondrogenesis of ADSCs were induced according to established protocols [38,39]. Osteogenic differentiation was evaluated by alkaline phosphatase (ALP) staining and the expression of cbfa1 (core-binding factor subunit alpha-1), OC (osteocalcin) and COL IA1 (collagen type I, alpha 1) on day 14. Adipogenic differentiation of ADSCs was evaluated using Oil red O staining and the expression of PPARγ (proliferatoractivated receptor γ) and aP2 (adipocyte fatty acid-binding protein) on day 7. Chondrogenic differentiation was evaluated using Alcian Blue (Sigma) staining of sulfated proteoglycan-rich matrix (blue color) and the expression of ACAN (aggrecan) and COL IIA1 (collagen type II, alpha 1) on day 14. Methods for ALP staining, Oil red O staining and Alcian Blue staining were as reported previously [4,33,35].

Cytokine and growth factor array
Human cytokines and growth factors from culture supernatants were detected and quantitated using Human Cytokine Antibody Array 1 (Cat #AAH-GF-1, RayBiotech, Norcross, GA) according to the manufacturer's instructions. Cytokine and growth factor expression from culture supernatants were quantified by ImageJ 1.40 g software (NIH), using the positive control (POS) on the same membrane as the internal control. Data are presented as the relative ratio of cultured to un-cultured groups.

Statistical analysis
Statistical analysis for cell proliferation, gene expression and cytokine contents in supernatants were performed with Microsoft Excel using t-tests. Cell numbers and viabilities following extraction via the two differential methods were analyzed with Wilcoxon signed rank test using Prism 4 (GraphPad Software, CA). p<0.05 was considered statistically significant. Experiments were performed at least twice, and each data point is presented as mean ± SD.

SVF isolation from EFT by disposable disperser tubes
We have developed a novel semi-automated method to isolate SVF from EFT. For comparison, the EFT was minced either with surgical scissors ( Figure 1B) or by the semi-automated disposable disperser tubes ( Figure 1C). The processed fat tissues were then digested with collagenase for 8 hours ( Figure 1D). Finally, SVF cells were isolated from the digested fat tissues by centrifugation into cell pellets ( Figure 1E and 1F). The semi-automated disposable tube tissue dispersion method reduced the operation time for mincing tissue from approximately one hour to a few minutes, and also avoided the use of scissors, a potential source of contaminants.

Cell yield and viability of SVF and ADSCs prepared using the semi-automated tube disperser and manual operation
The yield and viability of SVF and ADSCs prepared using either the semi-automated tube disperser method or the manual operation with surgical scissors was compared in 9 donors (Figure 2). Three grams of fat tissues were used for both methods. To compare total yield of ADSCs, SVF cells were seeded at the same initial cell density (10,000 cells/cm 2 ) and incubated in 10% HAS+MCA-supplemented medium for 5 days, and then detached using trypsin-EDTA. The cell yield of SVF prepared using tube disperser was 1.3-10.2 folds higher than the conventional method. The cell viability of SVF prepared using tube disperser was 0.9-2.4 folds higher than the conventional method ( Figure 2A). Moreover, the total yield of ADSCs was 1.2-7.8 folds higher when compared to the conventional method, and the S: Sense; A: Antisense   Figure 2B). Although the cell viability did not significantly differ between the two methods for both SVF and ADSCs (p=0.25, p=0.3594, respectively), the total yields were significantly higher for both SVF and ADSCs using the tube disperser method (p=0.0039, p=0.0039, respectively; Figure 2).

ADSCs expansion in different culture media
We compared the growth of ADSCs in the medium containing 10% HAS and 10% FBS, and found that the growth is slower in the 10% HAS-supplemented medium ( Figure 3B). To improve the growth of ADSCs without the use of animal serum, we added two antioxidants (NAC and AsA2P) to the HAS-supplemented medium, where FGF-2, NAC and AsA2P supplements were collectively named MCA as previously described [37]. The morphology and growth of ADSCs between the following conditions were compared: 10% FBS, 10% HAS, 10% HAS+MCA and 2% HAS+MCA. ADSCs were found to maintain their spindle-shaped morphology under all conditions on day 3 ( Figure  3A). ADSC expansions in the 10% HAS+MCA condition significantly increased from 155% to 324% between days 1 to 7 as compared to the 10% HAS condition ( Figure 3C). However, ADSC expansions in the 2% HAS+MCA condition markedly decreased when compared to the 10% HAS+MCA condition ( Figure 3D). HAS+MCA-supplemented medium for 1 month were shown in figure 4A and 4B. No abnormality was observed. The expression of 2 stemness genes ( Figure 4C), 2 stem cell-related genes ( Figure 4D), and 3 growth factors ( Figure 4E) of ADSCs maintained in various culture supplements were compared. The expression of Nanog, SOX2 (sexdetermining region Y-box 2), CXCR4 (C-X-C chemokine receptor type 4), TERT (telomerase reverse transcriptase), IGF-1, HGF, and EGF were enhanced in the 10% HAS-, 10% HAS+MCA-and 2% HAS+MCA-supplemented media as compared to those in the 10% FBS-supplemented medium ( Figure 4C and 4D).

Osteogenic, chondrogenic and adipogenic differentiations
Following culturing in differentiation-inducing media, the differentiation status of ADSCs was assessed. Alkaline phosphatase staining ( Figure 5A) and von Kossa staining ( Figure 5B) revealed the osteogenic differentiation of ADSCs. Expression of several osteogenic genes (i.e. cbfa 1, OC and COL IA1) after osteogenic induction was also studied ( Figure 5C). In general, the expression of these genes increased for all HAS-containing conditions as compared to those in the FBSsupplemented medium. Sulfated proteoglycan-rich matrix (alcian blue staining), indicating chondrogenic differentiation, was the highest in ADSCs cultured in the 2% HAS+MCA condition ( Figure 6A). Expressions of ACAN and COL IIA1 increased significantly in ADSCs cultured in the 10% HAS+MCA-and 2% HAS+MCA-supplemented media ( Figure 6B). Oil Red O-positive lipid vacuoles, indicating adipogenic differentiation, were evident in ADSCs for all conditions ( Figure 6C). The expression of PPARγ increased in ADSCs cultured in the 10% HAS+MCA-and 2% HAS+MCA-supplemented media, while the expression of aP2 was increased only in the 10% HAS+MCA condition ( Figure 6D). Thus, following the addition of differentiating media, ADSCs conditioned in both HAS and MCA further promoted all three types of differentiation when compared to ADSCs conditioned in media supplemented with FBS or HAS alone.

Surface markers of ADSCs
We have examined the phenotype of ADSCs grown in different culture conditions at passage 1 ( Figure 7A) and passage 2 ( Figure 7B) for the identification of MSCs. The expression levels of CD44 and CD73 were higher on ADSCs grown in the HAS-containing media (10% HAS, 10% HAS+MCA or 2% HAS+MCA) compared with those grown in the FBS-supplemented medium during passage 1, whereas the expression level of CD90 was lower on ADSCs grown in all HAS-containing media as compared with those grown in the FBS-supplemented medium. In addition, only ADSCs grown in the FBS-supplemented medium expressed CD34 at passage 1. These differences in expression levels between the HAS-and FBS-containing media were small but steady during passage 1. However, no differences in the expression of these markers were found during passage 2.  Karyotyping analyses were carried out using ADSCs from donors 3 and 8 following culturing in 10% HAS+MCA-supplemented medium for 1 month. Quantitative PCR were performed to quantify the relative gene expression of (C) the stemness genes: Nanog, SOX2, (D) the stem cell-related genes: CXCR4, TERT, and (E) the growth factor genes: IGF-1, HGF and EGF in the ADSCs that were cultured in the different culture media for 7 days (n=3). The gene expression was normalized to β-actin, serving as the internal control, and made relative to the expression levels of each gene measured for cells cultured under the 10% FBS condition. Each bar represents means ± SD. *p<0.05. **p<0.01. ***p<0.005.

Cytokine profiles of ADSCs
increased after 4 days of culture in the 10% HAS condition ( Figure  8A and 8B). Comparing between the 10% HAS and 10% HAS+MCA conditions, PDGF-AA, PDGF-AB and PDGF-BB content found in the supernatant decreased in the 10% HAS+MCA condition, while the VEGF R3 increased. At the same time, the secretion of HGF and VEGF increased in the 10% HAS+MCA condition and the secretion of IGF-1 and IGFBP-4 decreased ( Figure 8C). These results indicate that MCA can affect the autocrine and/or paracrine activities of the ADSCs in the microenvironment.

Discussion
The development of GTP/GMP processes for the isolation of SVF or ADSCs using small amount of EFT represents a major advance in the fields of cell therapy and tissue engineering. However, the current manual operation with surgical scissors in EFT processing has serious limitations. To minimize these limitations, we developed a semiautomated tissue processing method for the isolation of SVF ( Figure  1). This semi-automated process has three important advantages. First, the new process can markedly reduce the operation time for mincing tissues from one hour down to a few minutes. Second, the yields of SVF and ADSCs can be increased by the semi-automatic protocol. Third, the use of a disposable disperser tube instead of surgical scissors (requiring repeated disinfection and aseptic validation) eliminates a potential source of contamination. To the best of our knowledge, this    (C) Relative ratio of cytokine expression levels from cytokine array membranes. Cytokine and growth factor expression from culture supernatants were normalized against an internal positive control (POS) on the same membrane and presented as the relative intensity ratio of cultured, experimental (Ex) groups at day 4 to uncultured, control (Crl) group at day 0 (Ex/Crl). Data are presented as mean ± SD (n=3). The relative ratios of less than 1 are reduction of content in the supernatant, and the relative ratios of greater than 1 are classified as secretion. *p<0.05. **p<0.01. ***p<0.005. excision of fat tissues has been found previously to increase cell yield [6]. Thus, processing of EFT by our semi-automatic protocol is inevitably necessary. Our study demonstrated that the total SVF cell yield of EFT (n=9) was 2.65 ± 2.2 × 10 5 /ml by surgical scissors and 3.53 ± 2.2 × 10 5 / ml by tube dispersers. The cell viability was 66.22 ± 14.12% by surgical scissors and 74.22 ± 11.36% by tube dispersers, indicating that the semiautomatic method enhances the viable cell yields. Studies regarding the total SVF yield from both EFT and AFT are rarely reported in the literature. Although one study has reported total SVF cell yields of 1.44 ± 1.62 × 10 6 /ml from males (n=36) and 2.05 ± 1.46 × 10 6 /ml from females (n=78) [2], SVF yields obtained from AFT in our laboratory was 3.28 ± 5.59 × 10 5 /ml (n=11) (data not published). The differences in cell yield may be due to differential cell counting methodology, where the large amount of red blood cells present in the SVF samples are more easily excluded using Propidium Iodide (PI) instead of the Trypan Blue dye. In comparison to the cell yields obtained from our AFT samples, cell extractions from EFT samples by our semi-automated method, on average, lead to higher cell yields. At the same time, fat tissue extraction by EFT is more suitable for individuals with low BMI.
Serum is the most variable factor in the expansion of ADSCs at clinical grade standard. The use of animal sera, such as FBS, in culture medium increases the risk of transmitting zoonotic or other infectious agents [20][21][22]. On the other hand, the cost of proprietary chemicallydefined, xeno-free SFM presents another barrier to the production of ADSCs for wide clinical applications. In fact, the clinical effects of SFM remain unclear. Some replacements of FBS such as pooled human serum, HAS or platelet lysate (PL) have been considered as alternatives for producing MSCs in clinical applications [24,[40][41][42]. Although HAS exhibits batch-to-batch variability, it is still the safest option due to the elimination of the risk of contamination from xenogeneic or allogeneic products. In substitution for FBS, HAS and/or PL has been tested for ADSC cultures [24,41]. However, HAS and/or PL supplement may cause cell growth arrest and even cell death [20,41]. We had similar observations where HAS supplementation alone resulted in ADSC growth arrest on day 7 ( Figure 3B), but the addition of MCA supported, and increased, the continual growth of ADSCs to the 7 th day ( Figure  3C).
Cells must be retrieved in sufficient numbers for therapeutic efficacy [2,43,44]. In our previous study, we demonstrated that MCA promoted ADSCs to the S phase in the cell cycle by suppressing CDK inhibitors, p21 and p27, and resulted in rapid cell proliferation similar to that observed under hypoxic conditions [35]. In the present study, we found that the 10% HAS+MCA-supplemented medium enhanced ADSCs proliferation in vitro as compared to the 10% HAS condition ( Figure 3C). The addition of MCA also reduced the amount of HAS necessary for ADSC expansion ( Figure 3D). Thus, our formulation facilitates rapid ADSC expansion for individuals who require urgent autologous cell transplantation.
In our previous study, medium containing FBS and MCA caused significant reduction in the expression of CD29, CD90 and CD105 in ADSCs, while at the same time enhanced their osteogenic and adipogenic differentiating potentials [35]. In the present study, we did not find significant differences in the expression level of surface markers on ADSCs grown in media supplemented with either HAS alone or HAS+MCA at passage 1 or 2. However, the expression levels of CD34, CD44 and CD90 on ADSCs grown in the HAS-supplemented medium were different to those cultured in the FBS-supplemented medium at passage 1. The expression levels of surface markers on ADSCs have been demonstrated to be associated with the differentiation potential of ADSCs [45,46]. The higher expression of CD44 (TGF-β receptor) on ADSCs grown in the HAS-supplemented medium may explain the enhanced osteogenic ( Figure 5) and chondrogenic potentials ( Figure  6) [47].
ADSCs and their secretory factors such as VEGF, HGF and FGF-2 are known to be beneficial in skin repair and regeneration [48]. While hypoxia has been found to regulate the production of a variety of growth factors and enhance the paracrine effect of ADSCs to promote hair growth [49], we have found that MCA may also affect the paracrine and/or autocrine activities of ADSCs via changes of cytokine and growth factor content in the microenvironment (Figure 8). Our results showed that the 10% HAS+MCA supplement induced the reduction of PDGF-AA, PDGF-AB and PDGF-BB found in the supernatant during ADSCs proliferation that may be due to an enhanced uptake, binding to cell surface receptors, or degradation of cytokines, whereas the secretion of HGF and VEGF were enhanced ( Figure 8C). To the best of our knowledge, this is the first report demonstrating that MCA can regulate the autocrine and/or paracrine activities in ADSCs.
We have developed a semi-automated process that is particularly suitable for isolating human ADSCs from small amounts of EFT and formulated a xeno-free medium using HAS (2 or 10%) and MCA for their efficient expansion. The use of HAS and chemically-defined MCA comply with the GTP and GMP requirements for human tissues and cell products, respectively. These rapidly proliferating ADSCs have normal karyotypes and typical surface marker profiles, while maintaining stemness gene expression and differentiation potential. In summary, the combination of this novel semi-automated system with the HAS+MCA-supplemented medium may provide a better solution for the clinical applications of small amounts of EFT for regenerative medicine and tissue engineering.