Single 50–60 mL BM aspirates from the iliac crest were collected from three different volunteer male donors after obtaining informed consent (Lonza Walkersville, Inc., Walkersville, MD, United States) (Figure 1). BM donors #1, #2, and #3 were 21, 24, and 20 years old, respectively. The bone marrow aspirates from each of the first 2 donors, BM #1 and BM #2 were divided into 5 aliquots and BM aspirate from the 3rd donor, BM #3, was divided into six aliquots of approximately 10 mL each. Four sites (#1, #2, #3, and #4) received 1 aliquot of marrow aspirate from all 3 donors; while site #5 received 1 aliquot from BM #1 and BM #2 and 2 aliquots from BM #3. All BM aliquots were shipped to participating facilities, four of which were located within the United States and one site in Japan. FedEx conducted overnight shipments to the United States participating sites using containers with insulated packaging to maintain ambient temperature. Shipment to Japan was facilitated by MNX Global Logistics (Irvine, CA, United States) using the Evo Smart shipper (BioLife Solutions, Bothell, WA, United States) to maintain ambient temperature. Upon arrival of the BM aspirate, each site plated the BM using their own methodology to produce MSC preparations in culture.

Cell viability and cell count assessments, culture expansions and cryopreservation were performed according to each individual sites preferred method (Tables 1–3). For all three BM aspirates, two of the sites (site #2 and #3) plated cells from whole BM and two sites (site #1 and #4) plated cells from a mononuclear cell (MNC) fraction (Table 2). Site #5 plated whole BM for all three BM aspirates and received an additional BM aliquot from BM donor #3, which they plated cells from a MNC fraction (Table 2). Initial and subsequent passage seeding densities are indicated in Table 3. The adherent cells were cultured according to each sites’ in-house protocol and upon reaching confluence; the cells were harvested and passaged as indicated in Tables 2, 3. Four sites used fetal bovine serum (FBS) as media supplement, and one site used hPL (Cell Xpand^TM, University of Utah) (Table 2). At the end of each passage, cell counts and viabilities were performed and the number of population doublings (PDs) were calculated according to the following equation: PD(hrs) = t log 2/(log Nf/Ni), where PD = population doubling, t = time in culture, Nf = number of cells harvested, Ni = number of cells seeded. At the end of the culture period, each site prepared multiple aliquots of MSCs and used their own cryopreservation method to freeze the cultured-derived MSCs at a concentration that ranged from 1.0 × 10⁶ to 10 × 10⁶ cells/mL (Table 4). Laboratories routinely tested for bacterial and mycoplasma contamination. Cryopreserved cells were maintained in LN2 storage until they were shipped in LN2 to centralized testing laboratories where they were analyzed for immunophenotype, tri-lineage differentiation potential, CFU-F activity, gene expression, and immunosuppressive activity.

Evaluation of post-thaw MSCs generated in culture at each of the 5 sites was conducted by a centralized facility using flow cytometry to determine the expression or lack of expression of surface markers for CD105, CD73, CD90, CD45, CD14, CD34, HLA-DR, and Stro-1. Approximately 250,000 cells were transferred into each of five separate tubes. Cells from one tube were stained with primary antibodies, anti-CD73 conjugated to allophycocyanin (APC) (BD Biosciences), anti-CD90 fluorescein isothiocyanate (FITC) (BD Biosciences) and anti-CD105 phycoerythrin (PE) (Miltenyi Biotec). In a second tube, cells were triple stained with anti-CD45, anti-CD34, and anti-CD14 that were all conjugated with PE-Cy^TM5 (BD Biosciences). In a third tube cells were stained with anti-Stro1 APC (BioLegend) and anti-HLA-DR FITC (BD Biosciences). Additional tubes were stained with isotype controls and one tube was left unstained as a control. All tubes included the use of 7-aminoactinomycin D (7-AAD) to evaluate cells for viability. Following staining, the cells were washed with PBS-BSA, resuspended in PBS-BSA, and were analyzed on a BD FACSCanto (BD Biosciences) flow cytometer.

Adipogenesis, osteogenesis, and chondrogenesis differentiation in vitro assays were performed by a centralized laboratory using commercially available StemPro^TM kit adipogenesis, StemPro^TM kit chondrogenesis and Gibco^TM kit osteogenesis as per the manufacturer’s differentiation assay protocols. Briefly, cells from each MSC preparation provided by each site were washed and resuspended using Gibco^TM MesenPro RS medium (ThermoFisher Scientific). Cells from each MSC preparation were then seeded into four wells (one control well to remain undifferentiated, plus three conditional replicates) of a 12-well tissue-culture plate. Adipogenic (and control) wells were seeded at 1 × 10⁴ cells/cm². Osteogenic (and control) wells were seeded at 5 × 10³ cells/cm². MSC cultures were incubated at 37°C in a humidified atmosphere of 5% CO₂ for 48–50 h, then MesenPro RS medium in the triplicate conditional wells was removed and replaced with either adipogenesis or osteogenesis differentiation medium (StemPro^TM kit). The cultures were continued for 12 (adipogenic) or 18–20 (osteogenic) more days. Differentiation medium was exchanged every 2–3 days. Control wells received MesenPro RS media exchanges concurrently as the conditional wells received osteogenesis or adipogenesis media exchanges.

Chondrogenic (and control) wells were each seeded with five “micromass” 5-μL droplets of cells concentrated at 1.6 × 10⁷ cells/mL. Cells were maintained at 37°C in a humidified atmosphere of 5% CO₂ for two hours, after which the StemPro^TM Differentiation kit chondrogenesis medium (ThermoFisher Scientific) was added to the triplicate conditional wells and MesenPro RS medium to the controls. Micromass/medium cultures were maintained at 37°C/5% CO₂ for 21 days. Differentiation medium was exchanged every 2–3 days. Control wells received MesenPro RS media exchanges concurrently with the conditional wells receiving osteogenic or adipogenic media exchange. Control wells received MesenPro RS media exchanges concurrently as the conditional wells received chondrogenesis media exchanges.

Control and differentiated cell cultures were fixed with 10% formalin, and then stained with Oil Red O (adipogenesis), Alizarin Red S (osteogenesis) or Alcian Blue (chondrogenesis) dyes (all from Sigma-Aldrich). Differentiation of cells (or lack thereof) was then scored by three different staff members via visualization under an inverted light microscope from zero to four plus. A score of zero was equal to no differentiation while a score of 4+ was equal to maximum differentiation. A final score was determined as an average of the three scores as assigned by the observers.

To determine relative Colony-Forming Unit (CFU-F) activity of for each MSC preparation, a centralized facility received frozen cultured-derived MSCs from each of the five sites derived from BM#1, BM#2, and BM#3. The centralized laboratory thawed and plated the MSCs in triplicate into 6 well plates. Each well was plated with 200 MSCs in a total of 2 mL of MesenCult MSC Basal Medium (StemCell Technologies, Vancouver, Canada). The cells were cultured for 14 days at 37°C with 5% CO₂. After 14 days, the media form each well was aspirated and adherent cells were washed with 2 mL PBS, then 2 mL ice-cold methanol was added to each well for 5 min to fix the cells. The methanol was aspirated and 2 mL Wright-Giemsa Stain (Sigma Aldrich, St. Louis, MO, United States) was added to each well and stained for 10 min. The stain was aspirated and the wells washed twice with 2 mL PBS. The plates were air-dried and colonies were counted using a light microscope.

The immunosuppressive properties of MSCs were compared using MLR assays (SAIC-Frederic, Frederic, MD, United States). Ficoll-separated peripheral blood mononuclear cells were plated in 96-well plates at 1 × 10⁵ responders per well. Responders were co-cultured with 2500 cGy irradiated stimulator peripheral blood mononuclear cells at a concentration of 1 × 10⁵ cells per well. MSCs from different centers were added at concentrations of 1 × 10⁴, 4 × 10⁴, and 10 × 10⁴ cells/well. Culture plates were incubated for 6 days in a humidified 5% CO₂ incubator at 37°C. On the day of harvest, 0.5 μ Ci of ³H-thymidine was added to each well for 4 h with lymphocyte proliferation measured using a liquid scintillation counter. The effect of MSCs on MLR was calculated as the percentage of the suppression compared with the proliferative response of the control without MSCs, where the control was set to 0% suppression. The experiments were performed three times for each variable described.

Total RNA extractions were performed on the samples from each of the different centers using RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocol. RNA was quantified using Nanodrop 8000 (Thermo Fisher Scientific, Wilmington, DE, United States). Total RNA integrity was evaluated following isolation using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, United States). Samples with an RNA Integrity Number value ≥8 were used for gene expression analysis.

Microarray gene expression analysis was performed on 4 × 44 K Whole Human Genome Microarrays (Agilent Technologies, Santa Clara, CA, United States) according to the manufacturer’s protocol. In general, 200ng of total RNA from each sample was amplified, labeled, and hybridized on the array chip using a Quick Amp Labeling kit (Agilent). Array images were obtained by Agilent Scanner G2600D. Then images were extracted using Feature Extraction 12.0 software (Agilent). Partek Genomic Suite 6.4 (Partek Inc., St. Louis, MO, United States) was used for data visualization and hierarchical cluster analysis.

Principal Component Analysis (PCA) was used to visualize the similarities and differences among the samples from different centers. Unsupervised Hierarchical clustering was performed by Partek Genomic Suite using whole gene set to group similar samples into clusters.

The Source of Variation Analysis was performed to present the relative contribution of each factor included in the ANOVA by Partek Genomic Suite and the F-value provided by this analysis is the ratio of between group variance/within group variance. A large F-value for a factor indicates that between group variation is greater than within group variation.

BM aspirates (i.e., 50–60 mL each) were collected from three different volunteer donors and the BM from each donor was divided into aliquots of approximately 10 mL each (Figure 1). The BM aliquots were shipped to 4 sites located in the United States and to 1 site located in Japan. The average time from the collection to the receipt of the BM at each site is shown in Table 5. For those samples transported within the United States the average transit time was approximately 24 h while the average shipment time to the site located in Japan was approximately twice as long. Upon arrival at each site, the initiation of processing was executed within 5 h of receipt of the BM sample (Table 5).

Each site evaluated the BM aliquots for volume, cell viability, and total number of nucleated cells (Figure 2). The requested minimum volume from each of three donors to be supplied by the procurement agency to each site was 10 mL. As shown in Figure 2A, all sites received a minimum of 10 mL from each donor with some sites receiving volumes of greater than 10 mL. Cell viabilities as performed by each site (Table 1) were typically 98–100%, with one site, site #3, reporting viabilities from 80–90%, and one site, site #5, reporting that one of its aliquots had a viability of approximately 71% (Figure 2B). Average viable TNC concentrations for BM aspirates #1, #2, and #3 from the 5 laboratories were 25.5 ± 6.6 × 10⁶/mL, 21.8 ± 4.1 × 10⁶/mL, and 28.7 ± 7.1 × 10⁶/mL with coefficients of variation of 25.7, 21.5, and 24.6%, respectively (Figure 2C).

The culture schemes used by each of the five centers are summarized in Tables 2, 3. Two sites used unfractionated whole BM to plate cells, two sites subjected the BM aspirate to a density gradient separation to plate cells from a mononuclear cell fraction (MNC), and one site used either whole marrow or a MNC cell fraction to initiate MSC cultures. Laboratories cultured cells in alpha minimal essential media (MEM), MEM-Eagle or Dulbecco’s Modified Eagle Medium (DMEM). Fetal bovine serum (FBS) was used by 4 facilities as a media supplement, and 1 laboratory used a commercially available human platelet lysate-serum (PL-S) (Cell Xpand, University of Utah). The final concentration of FBS ranged from 10% to 20% and the final concentration for PL-S was 10% (Table 2). Antibiotics were added to the medium by four of the five centers (Table 2). Three sites cultured the cells in T-flasks, one site in a multiple layer flask and one site in dishes of 15 cm diameter (Table 2). Primary cultures or the initial seeding cell densities for BM#1, BM#2, and BM#3 ranged from 0.2 × 10⁵ cells/cm² to 2.8 × 10⁵ cells/cm² (Table 3). After the initial seeding, the five sites inoculated MSCs from the primary cultures (i.e., P0) at seeding densities that ranged from 50 cells/cm² to 4.0 × 10³ cells/cm² to obtain P1 MSCs. After harvesting MSCs at P1, sites 1 and 4 cryopreserved their cells without any further culturing. Three sites sub-cultured P1 MSCs at seeding densities that ranged from 2.3 × 10³ cells/cm² to 4.0 × 10³ cells/cm² to obtain MSCs from a P2 harvest (P2) (Table 3). After a P2 harvest, site 3 cryopreserved their cells without any further culturing. Two sites inoculated cultures with P2 MSCs to obtain MSCs that were cryopreserved after passage 3 (P3). No site passaged MSCs beyond P3 (Table 3). This translated into overall culture periods that ranged from 21 to 28 days for BM #1, 20 to 34 days for BM #2, and 21 to 35 days for BM #3 (Figures 3A–C). We did not only observe inter-laboratory variances for the time cells spent in culture at a particular passage among the five laboratories, but there were also intra-laboratory differences in the amount of time that the MSC cultures from different BM donors spent at a specific passage (Figures 3A–C). Each site cryopreserved cells according to their in-house protocol (Table 4). Four of the sites used 10% DMSO and one site used 5% DMSO as the final concentration of cryoprotectant. Only one site used a control rate freezer to drop the cyrovial temperature to −100°C, and three sites used a Thermo Scientific Nagene Mr. Frosty Freezing container to drop the temperature to −80°C before transferring to a liquid nitrogen freezer. One site wrapped the cryovials and placed them at −80°C and maintained them at this temperature for shipment to the centralized testing facilities.

In the end, different numbers of passages were executed by the five sites for each of the BM donors. For BM #1, the final harvest of cells for two sites occurred at P1, 1 site at P3 and one site failed to obtain cells. For BM#2, the final harvesting of cells occurred for three of the sites at P1 and for two sites at P3. Finally, for BM#3, three sites harvested at P1, one site at P2 and one site at P3. Overall, this resulted in 14 different final preparations of MSCs that were frozen according to each sites’ own specific freezing protocol for maintenance in LN2 prior to their analysis for phenotype and function. In summary, among the five laboratories the variations in methodologies to produce MSCs in culture included: (1) how cells were plated (i.e., cell density, whole BM versus MNC preparations); (2) the culture medium, (3) the time in culture; (8) the number of passages performed before the final cell harvest was cryopreserved and; (9) how each laboratory cryopreserved their cells.

Cellular proliferation responses as measured by theoretical yield and population doubling time from 14 different seeding events from three BM donors from five laboratories are shown in Figures 3D–F4A–C. Overall, inter-laboratory differences in cell proliferation responses and cell yields were observed from each BM donor. The most notable observations include the following: (1) Cell yields for all three BM donors showed that site #1 consistently produced the most cells and this was accomplished with only a P1 harvest. (2) Site #2 reported cell yields from BM #1 and BM #2 that were comparable to those obtained by site #1, but this was only achieved after site #2 completed a P3 harvest (Figures 3D,E4A,B). (3) Cell cultures for BM #2 performed by site #5 showed that the cellular proliferation responses lagged relative to the other four sites (Figures 3E 4B). (4) In general proliferation responses from the MSCs isolated from BM #3 were muted relative to the proliferation responses for cells from BM #1 and BM #2 (compare Figure 3F to Figures 3D,E and Figure 4C to Figures 4A,B).

To determine whether inter-laboratory manufacturing differences affected the phenotype and function of the cultured MSCs after cryopreservation, the frozen MSCs from 14 final preparations or lots were thawed and examined for cell surface markers, cell viability, CFU-F activity, tri-lineage differentiation potential, immunosuppressive activity, and their gene expression profile. This meant that assays were performed on 14 lots of MSCs that were cryopreserved at different passages during the manufacturing process. Sites #1 and #4 provided 6 lots of P1 MSCs and site #2 provided 3 lots of P3 MSCs from each of the 3 BM donors. Site #3 provided 2 lots of P2 MSCs from BM#1 and BM#2 donors. In addition, site #5 provided 1 lot of P3 MSCs from BM donor #2, and 1 lot of P1 and 1 lot of P2 MSCs from BM donor #3.

To assess whether each of the 14 final preparations of MSCs met the criteria established for a MSC phenotype, each lot of cells was examined by flow cytometry for their expression of cell surface antigens CD105, CD73, CD90 and lack of express for CD45/14/34 and HLA-DR) (Dominici et al., 2006). All final harvests of cells expressed the antigens CD105, CD73, and CD90 and lacked the expression of CD45/14/34 and HLA-DR as well as the Stro-1 antigen without apparent inter-laboratory differences (Figure 5).

Next, we examined each of the 14 lots of MSCs to determine whether inter-laboratory differences in culture strategies affected their function. Post-thaw MSC viabilities ranged from 74 to 92% for BM#1, from 61 to 96% for BM#2 and from 23–90% for BM#3 (Figure 6A). Site #4 had the lowest viabilities and the poorest CFU-activity for MSCs produced from BM donors #1 and #2. While site #1 had the lowest MSC viabilities and CFU-F activity for BM donor #3. Overall, MSC viabilities and CFU-F activity from BM#3 tended to be lower than for MSCs from BM donors #1 and #2 (Figures 6B–D). Inconsistencies in tri-lineage development among the laboratories were also observed and again site#4 showed some of the poorest developmental activity (Table 6). Of note is the fact that site#4, which was a laboratory located in Japan, experienced the longest transport times from BM collection to culture initiation and also shipped their cells at −80°C. Also, greater inconsistencies in tri-lineage development were observed from MSC lots produced from BM#3 than from MSCs isolated from the other two BM donors (Table 6). Interestingly, despite poorer viabilities, CFU-F activity and tri-lineage differentiation potential for MSCs from BM#3, the immunosuppressive activity of MSCs from BM#3 were comparable to that of MSCs from BM#1 and BM#2 (Figure 7).

All 14 lots of MSCs were analyzed by global gene expression analysis including PCA and unsupervised hierarchical analysis by using the entire set of expressed genes to identify relationships among the samples. Gene expression of MSC lots clustered by manufacturing site and donors (Figures 8A,B). Notably, the MSC lots clustered stronger according to site compared to donors (Figures 8A,B). This was confirmed by unsupervised hierarchical clustering analysis showing that the grouping of samples was related to site and donor with sites having a greater effect (Figure 8C). To further evaluate the relationship between the effects of manufacturing site and donor variability on MSC characteristics, the entire set of expressed genes from the 14 MSC lots was subjected to source of variation analysis. This analysis confirmed that the variability among MSCs samples was due to both the site and the BM donor, but the manufacturing site had a greater influence on MSC characteristics (Figure 8C).