A Chemically Defined Medium-Based Strategy to Efficiently Generate Clinically Relevant Cord Blood Mesenchymal Stromal Colonies

Abstract

During the last decade it has been demonstrated that mesenchymal progenitors are present and can be isolated also from cord blood (CB). Recently, we managed to set up a standard protocol allowing the isolation of mesenchymal stromal cells (MSCs) with high proliferative potential and multiple differentiation capabilities, whereas the generation rate of MSC-initiating colonies could still be further improved. Herein, we strikingly succeeded in defining some simple and basic culture conditions based on the use of a chemically defined medium that increased the colony isolation efficiency up to almost 80% of processed CB units. Importantly, this result was achieved irrespective of CB unit white blood cell content and time elapsed from delivery, two limiting parameters involved with processing CB units. Thus, this high efficiency is guaranteed without strict selection of the starting material. In addition, since we are profoundly concerned about how different culture conditions can influence cell behavior, we devoted part of this study to in-depth characterization of the established CB-MSC populations to confirm their stemness features in this novel isolation and culture system. Therefore, an extended study of their immunophenotype, including classical pericytic markers, and a detailed molecular analysis addressing telomere length and also stemness-related microRNA contribution were performed. In summary, we propose a straightforward, extremely efficient, and reliable approach to isolate and expand thoroughly characterized CB-MSCs, even when poor-quality CB units are the only available source, or there is no space for an isolation to fail.

Introduction

Mesenchymal stromal cells (MSCs) are multipotent and represent one of the major players in cell-based clinical trials, and various studies have found that they possess unique therapeutic properties^1,2. Until now no single mechanism of action has been defined, although the prevailing theory describes MSCs as living drugstores capable of secretion and delivery of numerous cytokines and growth factors to damaged tissues ³ , along with vesicular products implicated in many physiological processes ⁴ . Nonetheless, MSCs also retain the potential to differentiate into mesodermal lineages and specifically into adipocytes, chondrocytes, and osteocytes ⁵ , and this was the characteristic that first drew attention to MSCs as tools in regenerative medicine for tissue repair purposes.

Currently, cord blood (CB) has the undeniable advantage of having a completely safe and noninvasive harvesting procedure. Another advantage that will facilitate extensive use of CB-MSCs in the clinic is the existence of international networks of public and private banks storing thousands of frozen cord blood units (CBUs) that, in the future, could be used for MSC isolation for therapeutic purposes. Like MSCs from other sources, CB-MSCs can also undergo adipogenesis, chondrogenesis, and osteogenesis upon induction ⁶ and secrete biologically active molecules ⁷ . Currently, even if several attempts have been made and different methodological approaches have been implemented in order to efficiently isolate stromal cell colonies from CB, high-quality CBUs are often required, specifically with regard to time from collection to processing, white blood cell (WBC) content, and volume^8,9. For this reason many stem cell researchers have hesitated until now to consider CB-MSCs as a real alternative to bone marrow-derived mesenchymal stromal cells (BM-MSCs). Indeed, this main drawback could represent a huge hurdle, especially when feasibility of translation to the clinic has to be evaluated.

Recently, we described the existence of at least two stromal populations in CB, both capable of differentiation into mesodermal derivatives and both expressing typical MSC surface markers. Nonetheless, marked differences were found with regard to other biological features, such as growth properties including proliferation rate, life span in vitro, and ability to generate colonies under low-density seeding conditions. In addition, a robust statistically significant difference was detected at the molecular level following the analysis of telomere length, which was positively correlated with enhanced growth properties. Thus, we decided to name CB-MSCs endowed with higher proliferation properties and longer telomeres long-living (LL)-CB-MSCs, while we named CB-MSCs showing shorter life span and shorter telomeres short-living (SL)-CB-MSCs. We also proposed LL-CB-MSCs as the best choice in the context of stem cell therapy settings, on the basis of their biological features. For instance, the higher proliferation properties of LL-CB-MSCs allow for obtaining clinically relevant ex vivo cell expansions.

Notwithstanding the crucial advance in the knowledge of these important players in the regenerative medicine field, the establishment of such a population occurred in 12% of total isolated colonies ¹⁰ . In order to improve the efficiency of our standard isolation protocol and at the same time to generate CB-MSCs showing the same or improved qualities of LL-CB-MSCs, the main aims of the present study were threefold. First of all, we investigated whether we could provide novel cell culture conditions to improve MSC colony appearance and avoid some of the strict CBU quality requirements. The idea was to offer the best environment to facilitate CB mesenchymal progenitor attachment to a plastic surface, which is considered essential for their survival ¹¹ , and also to better support the generation and growth of stromal colonies. Second, considering the great influence of the medium composition on MSC stemness and other biological features ¹² , we sought to further evaluate the peculiar characteristics of the CB-MSCs cultured under these new conditions to provide more complete information about their effect and on their properties. Finally, we wanted to investigate if such an optimal culture environment could be applied to rescue the SL-CB-MSC phenotype and induce them to adopt the same clinically relevant proliferative properties demonstrated by LL-CB-MSCs in standard culture conditions.

Materials and Methods

Isolation of Human CB-MSCs

Human umbilical cord blood (hUCB) was collected in a multiple system bag (Machopharma, Mouvaux, France) containing 29 ml of citrate phosphate dextrose as an anticoagulant. The samples were not included in the study if the presence of clots or hemolysis was detected before or after the implementation of the isolation procedure. The authors state that this study was performed according to the amended Declaration of Helsinki. In addition, written informed consent was obtained from all of the cord blood donors involved in the study, and use of human tissue and cells was approved by the Ethical Committee of our institute, Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico. Two standard isolation procedures routinely performed in our laboratory were used for CB-MSC isolation under new culture conditions: a classic BM-MSC isolation protocol and a CB-MSC-tailored one ¹⁰ . In the first case, samples were seeded as whole blood at 0.5 × 10⁴ total nucleated cells (TNCs)/cm². This protocol is fast and simple, but given the high number of TNCs in the sample, only a small CB fraction (≤4 ml in this study) can be processed, since seeding all CB would imply the use of an excessive and unmanageable number of culture flasks (e.g., for a 20-ml CB sample with 10 × 10⁶ TNCs/ml, a 4,000-cm² cell growth area corresponding to more than 50 75-cm² flasks would be needed). On the other hand, higher seeding density cannot be applied because red blood cells and adherent white blood cells would heavily interfere with MSC attachment or colony growth. Alternatively, the samples were centrifuged at 680 × g for 15 min, and the plasma was discarded, while the buffy coat (BC) was collected. MSC precursors were obtained by negative immunodepletion of CD3⁺, CD14⁺, CD38⁺, CD19⁺, glycophorin A, and CD66b⁺ cells using a commercial kit (RosetteSep Mesenchymal Stem Cell, StemCell Technologies, Vancouver, BC, Canada) followed by density gradient separation. Briefly, for immunodepletion the recovered TNCs were incubated with 50 μl/ml RosetteSep MSC enrichment cocktail for 20 min at room temperature (RT). After incubation, the sample was diluted 1:3 with phosphate-buffered saline (PBS; Gibco, Grand Island, NY, USA), ethylenediaminetetraacetic acid (EDTA; Sigma-Aldrich, St. Louis, MO, USA), and human serum albumin (Kedrion, Barga, Italy) and then separated under standard density gradient conditions (Ficoll Paque Plus 1.077 ± 0.001 g/L; GE Healthcare, Little Chalfont, UK). The ring of immunodepleted mononuclear cells (MNCs) was transferred to a new tube with a Pasteur pipette, washed, and then plated at a concentration of 1 × 10⁶ cells/cm². This protocol allows the concentration of CB and its enrichment in MSC progenitors. Thus, all the sample is processed, the undesired and committed cell types are discarded, and a higher seeding density compared to the whole blood protocol can be applied.

In both cases, the sample fractions were suspended in culture medium composed of SPE-IV alone (ABCell-Bio, Paris, France) or supplemented with 20% fetal bovine serum (FBS; Gibco) and, when specified, with L-glutamine 1× (Gibco), hereafter referred to as complete medium (CM). SPE-IV (ABCell-Bio) is a defined serum-free medium developed for optimal growth support of human MSCs containing 25 ng/ml recombinant human insulin-like growth factor-1 (rh-IGF-1) and 0.33 ng/ml recombinant human-basic fibroblast growth factor (rh-bFGF). Other components, whose exact concentration is not disclosed in detail by the manufacturer, are clinical-grade human albumin, synthetic iron carrier, rh-insulin, nucleosides, L-glutamine, α-monothioglycerol, synthetic lipids, and αMEM. When specified, culture surfaces were coated with 10 μg/cm² collagen I–III (ABCell-Bio) for 2 h at RT.

Cell cultures were maintained at 37°C in a humidified atmosphere containing 5% CO₂. After 48 h, nonadherent cells were removed, and fresh medium was added. From that point, culture medium was changed every 3 days and cultures examined for colony appearance everyday. At 80% confluence, the cells were harvested with 25% TrypLE Select 1× (Gibco), washed with PBS (Gibco), and subcultured at a concentration of 1.5 × 10³/cm². LL-CB-MSCs and SL-CB-MSCs were isolated and cultured as previously reported ¹⁰ in the standard medium composed of αMEM-GlutaMAX (Gibco) supplemented with 20% FBS (Gibco) (standard CB-MSC culture control). To ensure proper culture conditions and as a routine quality control in our laboratory, the FBS batches used in this study were previously tested for optimal CB-MSC growth in the aforementioned standard culture conditions.

Cumulative Population Doublings

Population doublings were calculated for each established CB-MSC population using the equation: population doubling (PD) = log₁₀(N)/log₁₀(2), where N is the ratio between the number of cells harvested at the end of the culture and the number of seeded cells.

Cumulative PD (CPD) was calculated for each passage as the sum of the current and all the previous PD values.

Colony-Forming Unit-Fibroblast (CFU-F) Assay

Two hundred CB-MSCs were plated per 100-mm Petri dish (BD Biosciences, Franklin Lakes, NJ, USA) in replicate (n = 2). The medium was replaced with fresh medium after 1 week; at day 14 the cells were washed with PBS (Gibco), fixed with methanol (154903; Sigma-Aldrich), and stained with carbolic Gentian violet solution (Ral Diagnostics, Cedex, France). After two washing steps with milliQ-grade water, colonies with diameters greater than 1–2 mm were counted by sight.

Morphological Analysis

The morphological parameter calculated as the ratio between cell major axis length and nucleus diameter was considered indicative of cell shape and used to distinguish spindle-like cells from those showing a more compact shape. The two parameters were measured (n = 10 for each cell type) with Media Cybernetics Image-Pro Plus 2 (Media Cybernetics, Rockville, MD, USA). Images of cells in culture were taken with a Nikon Eclipse TS100 microscope (Nikon, Tokyo, Japan).

Telomere Length

Telomere length was assessed by real-time qPCR as previously described ¹³ . Briefly, CB-MSC DNA was extracted from cell pellets stored at −80°C with QIAamp DNA Blood Mini Kit (51104; Qiagen, Hilden, Germany). The telomere sequence and single-copy gene (36B4)-specific amplification reactions were performed in triplicate in 96-well plates on a CFX96 machine (Bio-Rad, Hercules, CA, USA). The data analysis was carried out using CFX Manager software (Bio-Rad). In order to present mean telomere length, the values of each CB-MSC population for each passage were normalized on its passage 0 value, and then the mean was calculated.

Gene Expression

Isolation of total RNA and real-time qRT-PCR assays were conducted in duplicate as previously reported ¹⁴ . The following primers were used.

C/EBP-β (FW 5′-GACAAGCACAGCGACGAGTA-3′, RV 5′-AGCTGCTCCACCTTCTTCTG-3′; COL10 A1 (FW 5′-ACTCCCAGCACGCAGAATCCA-3′, RV 5′-TGGGCCTTTTATGCCTGTGGGC-3′); MYOD (FW 5′-TGCTCCGACGGCATGATGGAC-3′, RV 5′-GACACCGCCGCACTCTTCCC-3′); PAX3 (FW 5′-GCCCAACCACATCCGCCACA-3′, RV 5′-CTTGGAGACGCAGCCGTGGG-3′); ALP (FW 5′-TACAAGGTGGTGGGCGGTGAACGA-3′, RV 5′-TGGCGCAGGGGCACAGCAGAC-3′); RUNX2 (FW 5′-AACCCACGGCCCTCCCTGAA-3′, RV 5′-CTGTGCCTGCCTGGGGTCTGTA3-′); BDNF (FW 5′-AGGTGAGAAGAGTGATGA-3′, RV 5′-CGGATGTTTGCTTCTTTC-3′); MUSASHI (FW 5′-GCGGCTGTTCGTGTTTTGGG-3′, RV 5′-AATGAGGGAGAGGGAAGCTAAGTG-3′); SOX2 (FW 5′-GGGGGAAAGTAGTTTGCTGCCTCT-3′, RV 5′-TGCCGCCGCCGATGATTGTT-3′); ANGPT1 (FW 5′-AGCAGCCTGATCTTACACGGTGC-3′, RV 5′-GCATCAAACCACCATCCTCCTGT-3′); ICAM-1 (FW 5′-AAGGTGACCGTGAATGTG-3′, RV 5′-GCCTGTTGTAGTCTGTATT-3′); VCAM-1 (FW 5′-TGTCATTGAGGATATTGGAA-3′, RV 5′-TAACTGTATTCTTGGGTGAT-3′); RPLP0 (FW 5′-TGTGGGCTCCAAGCAGATGCA-3′, RV 5′-GCAGCAGTTTCTCCAGAGCTGGG-3′).

Gene expression levels were normalized on RPLP0 “housekeeping” gene expression, following the 2^ΔΔCt method. For each CB-MSC population and each analyzed gene, passage 3 expression level was used as baseline.

Flow Cytometry

CB-MSCs were extensively characterized by flow cytometry at passages 3 and 5. Cells were washed with PBS (Gibco), centrifuged at 270 × g for 8 min and resuspended in PBS (Gibco) plus 2% FBS (Gibco), and then incubated for 20 min in the dark with the following combinations of directly coupled mouse anti-human fluorochrome-conjugated antibodies: CD45-FITC (21270453X2; ImmunoTools, Friesoythe, Germany), CD73-PE (550257; BD), CD90-PE-Cy5 (PNIM3703; Beckman Coulter, Brea, CA, USA), CD117-PE-Cy7 (PNIM3698; Beckman Coulter), CD105-APC (130-094-926; Miltenyi Biotec, Bergisch Gladbach, Germany), CD146-biotin (5050-B100T; BioCytex, Marseille, France) revealed with streptavidin APC-Cy7 (BD) for the first tube; CD31-FITC (555445; BD), CD144-PE (PNA07481; Beckman Coulter), CD34-PE-Cy5 (A07777; Beckman Coulter), CD133-APC (130-092-880; Miltenyi Biotec), CD146-biotin (BioCytex) revealed with streptavidin APC-Cy7 (BD) for the second tube; alphaSMA-FITC (F3777; Sigma-Aldrich), PDGFR-beta-PE (FAB1263P; R&D Systems, Minneapolis, MN, USA), CD146-biotin (BioCytex) revealed with streptavidin APC-Cy7 (BD) for the third tube; NG2-PE (PNIM3454U; Beckman Coulter), CD56-PE-Cy5 (555517; BD) for the fourth tube; SSEA4-FITC (560126; BD), CD271-PE (557196; BD), CD105-APC (Miltenyi Biotec) for the fifth tube. After staining, the cells were washed once with PBS (Gibco) containing 0.1% BSA (Kedrion). Fifty thousand events per sample were acquired with a FACSCanto II (BD) running FACSDiva 6.1.3 software, and isotype-matched mouse immunoglobulins were used as controls under the same conditions. Histograms and plots were generated using FlowJo analysis software v8.8.7 (Tree Star, Ashland, OR, USA). Mean fluorescence intensity (MFI) ratio was calculated as the ratio between sample and control MFI for each marker under analysis. Only MFI ratios greater than 2 were taken into consideration.

Cell Lineage Differentiation

CB-MSCs were induced to differentiate into adipogenic, osteogenic, and chondrogenic cell types. Commercial media (Lonza, Basel, Switzerland) were used following manufacturer's protocols. Adipogenesis was promoted with human MSC adipogenesis induction medium (PT-3102B; Lonza) and human MSC adipogenesis maintenance medium (PT-3102A; Lonza). Lipid vacuoles were stained with Oil red O solution (O0625; Sigma-Aldrich), following manufacturer's instructions. Osteogenic differentiation was induced with human MSC Osteogenic Medium (PT-3002; Lonza). Calcium deposits were stained with Alizarin red S (A5533; Sigma-Aldrich), following manufacturer's instructions. Chondrogenesis was induced with human MSC chondrocyte differentiation medium (PT-3003; Lonza), and the presence of extracellular matrix-producing cells was assessed with Alcian blue staining (B8438; Sigma-Aldrich). Due to high proliferation of CB-MSCs cultured in SPE-IV, initial seeding densities were modified to 0.5 × 10³ cells/cm², and the induction lasted only 2 weeks. Images of stained cells were taken with a Nikon Eclipse TS100 microscope (Nikon).

MicroRNA Expression Profiling

For miRNA studies, total RNA was isolated using the miRNeasy Mini kit (Qiagen) following manufacturer's instructions. RNA purity was determined by measuring the absorbance A260/A280 in a Nanodrop spectrophotometer. RNA integrity was assessed using electrophoretic techniques. MicroRNA expression profiling was performed in triplicate with an RT² miRNA PCR Array System (Qiagen) following the manufacturer's instructions. Briefly, 4 μg of total RNA, including small RNAs like miRNAs, was reverse transcribed into first-strand cDNA using the RT² miRNA First Strand Kit. Then the templates were mixed with RT² SYBR Green qPCR Master Mix (Qiagen), and 25 μl was aliquoted into each well of the eight 96-well plates containing the 704 predispensed miRNA-specific assays based on the Sanger mirBASE Release 14 (University of Manchester, Manchester, UK). Amplifications were carried out in a Bio-Rad CFX96 Real-Time PCR Detection System instrument (Bio-Rad Laboratories). After amplification, relative expression of each miRNA was determined with the ΔΔCt method using the online software RT² Profiler PCR Array Data Analysis version 3.5 (http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php) and the global mean normalization procedure^5,15,16. To detect only the reliable amplifications, all the wells giving rise to ambiguous double or flattened amplified product melt curve peaks were removed. In addition, following manufacturer's instructions, an absolute Ct value of 35 was set as the maximum threshold, since low-expressed miRNAs giving rise to high Ct values might be PCR amplification artifacts or miRNA machinery by-products ¹⁷ . Then array data of two small RNAs (miR-16 and miR-29b) were confirmed by qPCR using the miScript II RT Kit for cDNA synthesis (Qiagen) and miScript SYBR Green PCR Kit (Qiagen) as per the manufacturer's instructions. Briefly, 1 μg of total RNA was first polyadenylated and then reverse transcribed to generate first-strand cDNA. cDNA was diluted in RNAse-free water and prepared for qPCR with miRNA-specific forward primers (miScript Primer Assay, Qiagen), a universal qPCR primer included in the miScript SYBR Green PCR mix, and SYBR Green reference dye. All reactions were run in a Bio-Rad CFX96 Real-Time PCR Detection System instrument (Bio-Rad Laboratories). Relative miRNA levels were normalized to RSU snRNA, a verified endogenous control. Verified mRNA targets of selected miRNAs were identified using the miRWalk database (http://www.umm.uni-heidelberg.de) ¹⁸ . Final target lists were then uploaded to GeneCodis free software (Centro Nacionál de Biotecnología, Madrid, Spain) for functional interpretation using default filters^19–21.

Statistical Analysis

Nonparametric Kruskal–Wallis tests followed by Dunn's multiple comparisons tests were performed to evaluate CBU characteristics. One-way analysis of variance (ANOVA) followed by Newman–Keuls (mean normalized telomere length) (Fig. 3A) or Tukey's (relative telomere length) (Fig. 3B) multiple comparisons tests were performed to evaluate CB-MSC telomere length after normal distribution of the datasets was verified by a D'Agostino–Pearson omnibus normality test. GraphPad Prism version 6.00 for Windows (www.graphpad.com; GraphPad Software, San Diego, CA, USA) was used to perform all of the previously described statistical analyses. Concerning the miRNA target study, the GeneCodis free software (Centro Nacionál de Biotecnología, Madrid, Spain) was run to find statistically miRNA-dependent enriched biological processes with respect to the entries of the whole human genome^19–21. A value of p < 0.05 was used to determine statistical significance.

Results

Chemically Defined Medium-Based Culture Conditions for the Isolation of Cord Blood Stromal Colonies

A total number of 100 CBUs were processed for this study, and different standard experimental approaches were investigated. The aim was to determine which isolation procedure best supported the generation of stromal colonies from CB under new culture conditions, taking into account efficiency but also feasibility of the procedure (for the complete schematic, see Fig. 1).

Figure 1.

Complete experimental schematic. The two graphs represent isolations performed without (A) or with (B) serum supplementation, and they specify further experimental conditions (diagram branches) and isolation success as percentage (dotted line rectangles). The following CB unit parameters were evaluated for the different whole blood (WB) and immunodepleted fraction (ID) isolation experiments: time from collection to processing (C), WBC content (D), and processed volume (E, F). Box and whisker graphs show 25th and 75th percentiles (plot hinges), median, minimum, and maximum values (C, D) or 10th and 90th percentiles with outside values (E, F). **p < 0.01; ***p < 0.005; ****p < 0.001; ns = statistically nonsignificant difference.

In a first series of isolations the importance of the collagen I–III coating was investigated for the optimal use of a chemically defined medium containing, among other bioactive molecules, two important growth factors: IGF-I and bFGF. Whole blood samples versus immunodepleted fractions were compared, and the appearance of the first fibroblastic colony was considered as a positive event. Positive events on coated surfaces were obtained from five whole blood samples (14 total units) and from two immunodepleted fractions (5 total units). On the contrary, without the coating no positive event was observed (5 whole blood samples and 11 immunodepleted fractions) (Fig. 1A).

In a second set of experiments, the addition of a standard percentage of serum was investigated as a possible replacement of the collagen I–III coating, based on the common practice of coating plastic surfaces with serum. In this way, positive events were obtained from three whole blood samples (6 total units) and from 17 immunodepleted fractions (30 total units), representing the most efficient isolation strategy for the experimental conditions reported herein.

The analysis of CBU characteristics demonstrated that the observed results were not biased by major differences between the experimental groups in terms of time elapsed from CB collection to processing (Fig. 1C) or WBC content (Fig. 1D), two crucial parameters for CB-MSC isolation^8,9. The evaluation of processed sample volume revealed that no statistically significant differences were present within whole blood (WB, WB coating, and WB serum conditions; Fig. 1E) and immunodepleted (ID, ID coating, and ID serum conditions; Fig. 1F) experimental groups.

Following this experimental design, the use of serum as a substitute for collagen I–III coating to culture the CB immunodepleted fraction appeared as the most promising approach to further ameliorate the generation of stromal cell colonies, due to its higher isolation efficiency. Thus, in order to further improve this already relatively high rate of positive events, L-glutamine supplementation as a major energy source for cultured cells was tested in immunodepleted fraction isolation experiments (complete medium) to offer a more favorable environment for colony generation, along with a preference for significantly richer-in-volume (≥20 ml threshold) CBUs compared to the other immunodepleted samples (ID serum and L-glutamine condition; Fig. 1F). Under these conditions, positive events were observed in 79% of the processed samples (23 out of 29 units) (Fig. 1B). Due to this high efficiency, we focused on this method for the isolation and expansion of CB-MSC populations for all of the subsequent studies.

CB-MSC Proliferative Properties

Following the most efficient isolation conditions (n = 23 units with positive events), the first colonies of plastic adherent cells appeared reasonably soon [a median of 13 days after seeding (min 5, max 22 days)], with a colony number per processed unit ranging from 1 to 5. Interestingly, there were almost no other “contaminant” adherent cell types, like osteoclast-like or nonproliferative fibroblast-like cells. The first trypsinization occurred with a median of 24 days after seeding (min 7, max 41) and the median number of cells harvested from passage 0 was 0.37 × 10⁶ cells (min 0.05, max 2.59 × 10⁶ cells).

Three CB-MSC populations were established, which were named CB-MSC 1, 2, and 3, each starting from a single colony derived from a distinct CBU (unrelated donors). All three CB-MSC populations were extensively characterized along long-term culture for growth kinetics, immunophenotype, and molecular profile. The CB-MSC growth curves are shown in Figure 2A. For all of the passages, 1,500 cells/cm² was the optimal seeding density, which allowed us to trypsinize the cells once a week with medium changes every 3–4 days. In two cases (CB-MSC 2 and 3) growth arrest occurred only after reaching a CPD of 75, and only after passage 13, while CB-MSC 1 reached a CPD of 45 and arrested its growth at passage 9. We also noted that the cells proliferated with a constant rate without high fluctuation in terms of population doublings (CB-MSC 1, P1–P9: mean (m)PD = 5.1, standard deviation (SD) = 1.6; CB-MSC 2, P1–P15: mPD = 5.1, SD = 0.7; CB-MSC 3, P1–P13: mPD = 5.8, SD = 1.3), or of days between two subsequent trypsinizations during almost all the culture (a mean of 7 days from P1 to P7 for the three populations; SD = 1.3). If we consider the three CB-MSC populations together from passage 1 to passage 9, the result is a mean growth curve that approximates almost perfectly a linear function, constantly increasing the CPD of 5.6 after every passage (Fig. 2B).

Figure 2.

Biological characterization of established CB-MSC populations. Proliferation rate (A) and mean proliferation rate (B) of the three CB-MSC populations. The error bars in (B) represent standard deviation. A representative image of CB-MSC CFU-F is shown (C). (D–F) Morphology study: images of CB-MSCs cultured under standard conditions (D) or relying on the complete medium (E); (F) summarizes the results.

At passage 1, the CFU-F assay was performed, and all the three CB-MSC populations could generate colonies under low-density seeding conditions (a mean of 76 colonies; SD = 35) (Fig. 2C).

CB-MSC Morphology

In order to give quantitative information about cell morphology, two parameters were measured by image analyzer software: diameter of the nucleus and cell major axis. The morphology of CB-MSC 1 (Fig. 2D) was compared with that of LL-CB-MSCs cultured in the standard medium (Fig. 2E). The results are summarized in Figure 2F. Although no statistical significance was observed, the ratio seemed to be lower for CB-MSC 1, as shown by both the mean and median, but concomitantly also presented the greatest degree of variation, as shown by SD and minimum and maximum values. This was consistent with previously reported data ¹² , comparing CB-MSC morphology of the same population in two different media. No significant differences in cell morphology were observed among CB-MSC 1, 2, and 3 or between LL-CB-MSCs and CB-MSC 2 or 3 (data not shown).

Telomere Length Assessment

To further characterize the CB-MSC populations, telomere length was assessed (Fig. 3A and B). Evaluating the established CB-MSC populations together, a dramatic and significant decrease in mean telomere length was observed from passage 0 to passage 1 (Fig. 3A), after which telomere length was maintained at a constant (statistically nonsignificant differences between P1 and P9 samples). Intriguingly, the CB-MSC populations evaluated separately at passage 0 showed a similar relative telomere length for CB-MSC 2 and 3, while CB-MSC 1 had a telomere length significantly shorter (three times).

Figure 3.

Telomere length. For telomere length analysis, passages 0, 1, 3, 5, 7, and 9 were studied. The histogram in (A) shows the mean value for each passage of the three analyzed CB-MSC populations with the standard deviation. The data are presented as percentage relative to P0 telomere length value. The histograms in (B) show mean value from triplicates of real-time qPCR analysis, again associated with standard deviation. ***p ≤ 0.005; *p ≤ 0.05. (C) Steady-state transcription profile. The mRNA levels (P3-5-7) of genes involved in adipogenesis, chondrogenesis, myogenesis, and osteogenesis were taken into consideration. Also the expression of neurotrophic and angiogenetic growth factors as well as neural and endothelial markers was analyzed. The histograms represent mean gene expression of the three CB-MSC populations analyzed, while the error bars show the corresponding standard error of the mean (SEM) values.

Steady-State Gene Expression Profile Along Prolonged Culture

The genes analyzed included typical MSC differentiation markers: C/EBP-β for adipogenesis, COL10A1 for chondrogenesis, and ALP and OSTERIX for osteogenesis. Additionally, genes typically expressed by cells from other lineages, such as neural cells [brain-derived neurotrophic factor (BDNF), MUSASHI, sex-determining region y-box 2 (SOX2)], or implicated in the angiogenic [angiopoietin 1 (ANGPT1), intercellular adhesion molecule 1 (ICAM-1), vascular cell adhesion molecule 1 (VCAM-1)] and myogenic (MYOD, PAX3) processes were considered. The mean gene expression levels of the CB-MSC populations evaluated together revealed stable and mostly unchanged expression along prolonged culture of the majority of the genes taken into consideration (Fig. 3C). The only genes that showed a steady increasing trend in their transcription levels, albeit not significant, were chondrocytic COL10A1 and ANGPT-1, whose expression lead to a threefold difference between P3 and P7 samples.

Immunophenotype of CB-MSC Populations

All CB-MSCs stained positively for canonical MSC surface markers CD90, CD105, and CD73, whereas they were negative for CD45 and CD34 (Fig. 4A) (data not shown for CB-MSC 3). A deeper analysis with a more extensive surface marker panel was applied to investigate if the observed differences among CB-MSCs in growth properties and passage 0 relative telomere length could correlate with a specific immunophenotype. With this in mind, CB-MSC 1 was compared to CB-MSC 2 at passages 3 and 5. CB-MSC 2 was considered representative of both CB-MSC 2 and 3 behavior, since they shared the same growth characteristics and telomere length. Both populations were strongly positive for CD146. In addition, CB-MSCs were completely negative for CD31, CD133, CD117, and CD144, and almost completely negative for α-smooth muscle actin (α-SMA) and stage-specific embryonic antigen-4 (SSEA4) (Fig. 4A). Concerning platelet-derived growth factor receptor (PDGFR)-β, nerve/glial antigen 2 (NG2), CD56, and CD271 some marked differences were observed (Fig. 4B). PDGFR-β, NG2, and CD56 were more expressed in CB-MSC 2. Mean fluorescence intensity (MFI) ratios summarized in Figure 4C showed a marked difference along passages for PDGFR-β in both populations. The case of CD271 was peculiar since CB-MSC 1 did not express CD271 at passages 3 nor 5, whereas CB-MSC 2 at passage 3 showed a portion of CD271⁺ cells (MFI ratio = 5.0) corresponding clearly to a distinct subpopulation, as shown in the density plot in Figure 4D. However, this subpopulation was absent at passage 5 (MFI ratio = 1.8).

Figure 4.

Immunoprofile of CB-MSCs. Classical MSC surface marker panel with percentages of expression (A) for passages 3 and 5 CB-MSCs. Representative histograms (P3) of some novel MSC surface markers show marked differences between the two populations (B). (C) summarizes MFI data for both analyzed passages. The dot plot (D) shows passage 3 CD271⁺ CB-MSC 2 subpopulation and its percentage within the whole population. In the histograms and in the dot plot, for every plotted event the y-axis represents the percentage of maximum expression, while the x-axis shows fluorescence intensity on a logarithmic scale. (E–G) Differentiation properties of CB-MSCs. Representative images of CB-MSC-derived fat droplet-producing adipocyte-like cells, calcium deposit-producing osteocyte-like cells, and Alcian blue-positive chondrocyte-like cells are shown, respectively, in (E–G). No significant differences were observed with regard to differentiation capabilities between the different CB-MSC populations.

Maintenance of Differentiation Potential

To evaluate the differentiation potential some changes in the standard protocols were developed. Mostly because the proliferation potential of these CB-MSCs was too high, it was difficult to keep the standard induction protocols without an earlier cell detachment. CB-MSCs were differentiated toward the adipogenic, osteogenic, and chondrogenic lineages. Adipogenesis was confirmed by Oil red O staining of lipid vacuoles (Fig. 4E). Osteogenesis was assessed by Alizarin red S staining, which evidenced the conspicuous production of calcium deposits (Fig. 4F). Chondrogenesis was detected by presence of Alcian blue-positive cells (Fig. 4G). No macroscopic differences in the differentiation capacity toward the adipogenic, osteogenic, and chondrogenic lineages were observed for the three CB-MSC populations at the analyzed passage.

Identification of miRNA Expression Profiles

To unravel whether molecular differences could also be found between CB-MSC 2 and 3, which demonstrated homogenous cell growth, clonogenic properties, telomere length, and molecular signature, a whole-genome comparative gene expression profiling analysis was performed using arrays able to monitor the expression of 704 miRNA sequences. After a quality control procedure, 243 and 244 miRNAs out of the total 704 available sequences were detected in CB-MSC 2 and CB-MSC 3, respectively. In order to exclude potential erroneous miRNA expression changes introduced by handling or technical pitfalls, random individual real-time qRT-PCR reactions on cDNA obtained from two biological replicates of CB-MSC 2 and 3 were performed. Thus, the expression of two miRNAs belonging to unrelated families (miR-16 and −29b), both normalized with the small nuclear RNA RNU6-2 used as a housekeeper, was analyzed, and then the values were compared with those obtained in the arrays (also normalized with RNU6-2). As shown in Figure 5A and B, the independent samples showed consistent values (within twofold modulation) with the array data. We also checked whether the ratios between CB-MSC 2 and CB-MSC 3 in the single PCR amplifications for these two miRNAs were consistent in comparison with the array data and again such values did not show modulations higher than twofold units (Fig. 5C). All of these tests on independently processed and assayed samples confirmed the validity of the proposed array results, although the design of the arrays with single probes leaves open the possibility of few potential technical-dependent miRNA expression variations.

Figure 5.

Validation of array data through real-time qRT-PCR on independent samples (A–C). Total RNA isolated from biological replicates of CB-MSC 2 and 3 samples was reverse transcribed and used in real-time qRT-PCR reactions to compare normalized expression levels (y-axis) of miR-16 and miR-29b with the normalized values obtained with microRNA arrays set as 1. None of the miRNAs differ by a factor higher than 2 (represented in the graphs by the black lines). Mean values and standard deviations of two independent biological replicates are shown. Comparison of genome-wide miRNA expression profiles (D). Correlation of miRNA expression levels (normalized Ct, x- and y-axis) between CB-MSC 2 and CB-MSC 3. Dotted lines represent a boundary corresponding to normalized Ct value difference of 2 U (fourfold differential expression). Rescue of short-living phenotype (E, F). Representative morphology of the same SL-CB-MSC population cultured either in standard or complete medium (E). Increased number of passages and CPD shown by SL-CB-MSCs cultured in the complete medium (Complete) and compared to those obtained by the same SL-CB-MSCs grown in the standard medium (Standard) are represented in the two graphs (F). For SL-CB-MSCs cultured in complete medium only the passages after thawing and switch to complete medium are shown.

To compare miRNA expression patterns between CB-MSC 2 and 3, a pairwise comparison of the two expression profiles was performed generating a distribution of normalized intensity with a correlation coefficient of 0.96 (Fig. 5D). This finding demonstrates that the two CB-MSCs share a strikingly similar miRNA expression pattern, thus accounting for possible phenotypic and functional overlap. Nevertheless, a deeper characterization was undertaken selecting miRNAs with fourfold or higher differential expression, corresponding to a difference of at least two normalized Ct value units. Distinct molecular signatures were extracted consisting of increased expression of four miRNAs (miR-450a +10.4, miR-369-3p +4.5, miR-10a +4.1, miR-10b +4.1) and downregulation of one miRNA (miR-27a* −4.3) in CB-MSC 2 compared to CB-MSC 3. Notably, none of the differentially expressed miRNAs is among the most expressed ones. In fact, the five most expressed miRNAs are conserved and constant between the two CB-MSCs (miR-21, miR-125b, miR-16, miR-1260, and miR-199-a-3p) with the two most expressed (miR-21 and miR-125b) recently proposed as stem cell markers^22–24. Next, a gene target identification for the five more differentially modulated miRNAs was performed using miRWalk, a database that presents literature information on experimentally validated miRNA interactions with genes ¹⁸ . Mining the miRWalk database, 239 validated targets were shown to be regulated by these miRNAs. Then, on the identified genes, the GeneCodis software showed that the two most statistically enriched biological processes were connected with regulation of transcription [from RNA polymerase II promoter—Gene Ontology (GO): 0045944, p < 0.001; DNA-dependent—GO: 0045893, p < 0.001], a result in agreement with the most enriched cellular component of the submitted gene products that resulted to be the nucleus (GO: 0005634, p < 0.001). Finally, such a relationship with gene regulation was supported by the molecular functions uncovered by these regulated genes that appeared to be directly linked with an enrichment in the protein-binding category (GO: 0005515, p < 0.001), especially associated with DNA-binding transcription factor activity (GO: 0003700, p < 0.001) and DNA-binding properties (GO: 0043565, p < 0.001).

Rescue of SL-CB-MSC Short-Living Phenotype

Two SL-CB-MSC populations previously obtained in the standard medium were thawed at passage 2 and cultured in the complete medium to observe whether their short-living phenotype could be rescued. From a morphological point of view the cells grown in the standard medium seemed to possess a more spindle-like versus a more compact morphology, compared to the same cells grown in the complete medium (Fig. 5E). Strikingly, both SL-CB-MSC populations showed an increased life span and higher CPD peaks (from a value of 5.4-10.8 to 20.2-21.3), doubling the maximum number of passages (from P2–3 to P7) in the complete medium compared to the standard one (Fig. 5F). Notwithstanding this ameliorated growth kinetic, they did not reach the same high performances of CB-MSC 2 and 3 of the present study.

Discussion

Over the last years, the clinical use of allogenic CB for hematopoietic stem cell (HSC) transplantation has increased dramatically, and more recently CB was also used as a source of MSCs ²⁵ . Unfortunately, the low frequency of these MSCs in CB and the consequent low isolation efficiency still is a major problem, which keeps this source far from being widely recognized for its possible role in regenerative medicine. In this regard, several attempts and different methodological approaches have been implemented in order to more efficiently isolate MSCs from CB. Notwithstanding, this source still offers some advantages in comparison to the more classical ones, including the easy procurement, the absence of risks to donors, and the reduced risk of transmitting infections ²⁶ . Up until now, one of the crucial CB characteristics that makes CB-MSC isolation feasible is the “freshness” of the CBU (≤2 h from CB collection to processing) ²⁷ .

In our laboratory we set up a standard strategy ¹⁰ allowing the isolation of long-living (P > 7), high proliferative CB-MSCs with multipotent differentiation capabilities (LL-CB-MSCs). Yet, the efficiency of this process should be further improved, especially if clinical application is the final goal. In the last years several commercially available media have been proposed on the market for MSC culture; however, they are mainly developed and tested for MSCs from bone marrow and adipose tissue, whose isolation efficiency is close to 100%. In this context, with respect to the first goal of our work, we succeeded in defining some culture conditions for obtaining CB-MSC colonies at high efficiency, overcoming a major drawback, which hampers a widespread availability of this kind of therapeutic cell. The adopted strategy combines an immunodepletion treatment of CBUs and a chemically defined medium specifically developed for optimal MSC isolation and culture. In this way we also skipped the use of a collagen I–III coating that could possibly make the translation of this system in a clinical-grade setting in some countries more difficult. With this approach and a preference for rich-in-volume samples, the successful rate of CB-MSC colony generation was dramatically improved up to 79%, avoiding strict quality selection of the starting material in terms of cellularity and “freshness.” In addition, no osteoclast-like cell contamination, which badly affects the newly born stromal colonies ²⁸ , was observed. This could be due to the presence of IGF-1 in the defined medium, which has been demonstrated to reduce monocyte adhesion properties ²⁹ . The same effect was reported also for bFGF, the other growth factor present in the chemically defined medium ³⁰ .

The second goal of this work was to investigate the biological features of CB-MSCs obtained in this new isolation and culture system, focusing on growth properties, on the molecular pattern, and also on the protein level. The proliferation curves of the expanded CB-MSCs reached higher CPD peaks in comparison with CB-MSCs isolated and cultured under standard conditions ¹⁰ , and they showed a stable, linear, and homogeneous growth. Based on the International Society for Cellular Therapy (ISCT) classification ³¹ , the CB-MSCs obtained expressed the classical MSC markers and maintained them during passages and possess multipotent differentiation abilities. The cells were also able to generate colonies under low-density seeding conditions, even if their morphology differed from that of colonies obtained from CB-MSC cultured in standard conditions ¹⁰ , being more compact and presenting a higher cell density. This again could be ascribed to the synergic action of IGF-1 and bFGF, which add their effect to that of serum growth factors.

Telomere length was dramatically decreased from passage 0 to passage 1, but then was maintained at the same level along culture, suggesting that the medium tested is not sufficient to prevent repetitive telomere sequence loss at the initial phases of the culture establishment, but nonetheless it could play some role in stabilizing its decrease. Unfortunately, the extreme difficulty to have more than one colony able to give rise to a CB-MSC population from the same CBU has hampered, until now, the ability to perform proper experiments to address this issue, in which standard and novel media effects on telomere length could be compared. Some minor differences were found between the CB-MSC populations. In particular, telomere length dissimilarities correlated with different replicative potentials: CB-MSC 1 presented the shortest telomere and arrested its growth at passage 9, while CB-MSC 2 and 3 showed higher telomere length and could be cultured at least until passage 13.

Moving to an extensive immunophenotype profiling, some noncanonical surface markers were found to be expressed. These surface markers were all classic pericytic markers such as CD146, NG2, CD56, and even the more relevant PDGFR-β ³² . Intriguingly, pericytes have been proposed as the precursors and sort of in vivo counterpart of MSCs ³³ , so that such an immunophenotype could mirror a CB-MSC state reminiscent of its pericytic precursor.

In addition, CB-MSCs showed very homogenous gene expression profiles. The study revealed stable expression of key genes involved in distinct differentiation pathways, suggesting that the tested medium succeeded in maintaining the CB-MSC cultures in an undifferentiated state. To confirm the homogeneity found at the molecular level between the CB-MSCs endowed with the longest telomeres, we addressed their miRNA signatures using a miRNA array. Interestingly, we observed that in these CB-MSCs the most expressed miRNAs were stemness-related miR-21 and miR-125b^22–24.

Based on our recent report ¹⁰ describing two CB-MSC populations distinguished by telomere length and life span, we wondered if the analyzed populations were bona fide SL- and LL-CB-MSCs, even if characterized by growth properties enhanced by a culture medium much richer than the standard one. This hypothesis was substantiated by the “rescue” experiments, as SL-CB-MSCs previously isolated following the standard methodology increased the number of total passages and of population doublings in response to this novel culture system. Nevertheless, the “rescued” SL-CB-MSCs still possessed a shorter life span compared to LL-CB-MSCs, confirming that two distinct stromal populations exist in cord blood irrespective of the culture conditions adopted to isolate or maintain them. It is also true that even the observed partial rescue of the SL-CB-MSC short-living phenotype could represent a relevant effect to take into account when such a population is the only one isolated from a specific CB unit. This is of paramount importance, most of all in the context of private CB banking, where clinically relevant CB-MSC numbers must be obtained necessarily from related or autologous CBUs.

In conclusion, we assessed how medium composition can exert a heavy influence on cell behavior, so that our last message in the bottle is that new media with more complex composition than the standard ones can have advantages on cell culture, although a deep characterization is strongly encouraged to really define how the new culture system may modulate the cells you are dealing with.