Intercellular cytosolic transfer correlates with mesenchymal stromal cell rescue of umbilical cord blood cell viability during ex vivo expansion

Background aims. Mesenchymal stromal cells (MSC) have been observed to participate in tissue repair and to have growth-promoting effects on ex vivo co-culture with other stem cells. Methods. In order to evaluate the mechanism of MSC support on ex vivo cultures, we performed co-culture of MSC with umbilical cord blood (UCB) mononuclear cells (MNC) (UCB-MNC). Results. Significant enhancement in cell growth correlating with cell viability was noted with MSC co-culture (defined by double-negative staining for Annexin-V and 7-AAD; P<0.01). This was associated with significant enhancement of mitochondrial membrane potential (P<0.01). We postulated that intercellular transfer of cytosolic substances between MSC and UCB-MNC could be one mechanism mediating the support. Using MSC endogenously expressing green fluorescent protein (GFP) or labeled with quantum dots (QD), we performed co-culture of UCB-MNC with these MSC. Transfer of these GFP and QD was observed from MSC to UCB-MNC as early as 24 h post co-culture. Transwell experiments revealed that direct contact between MSC and UCB-MNC was necessary for both transfer and viability support. UCB-MNC tightly adherent to the MSC layer exhibited the most optimal transfer and rescue of cell viability. DNA analysis of the viable, GFP transfer-positive UCB-MNC ruled out MSC transdifferentiation or MSC-UCB fusion. In addition, there was statistical correlation between higher levels of cytosolic transfer and enhanced UCB-MNC viability (P< 0.0001). Conclusions. Collectively, the data suggest that intercellular transfer of cytosolic materials could be one novel mechanism for preventing UCB cell death in MSC co-culture.


Introduction
Mesenchymal stromal cells (MSC), also known as multipotent mesenchymal stem cells, have been used in various clinical applications such as tissue engineering (1), regenerative medicine (2) and immunoregulatory therapy (3). MSC can be isolated easily from various tissue sources (4), readily expanded in culture (5) and differentiated into various cell types, such as osteoblasts, chondrocytes and adipocytes, with suitable stimulation (6,7). Therefore, MSC have been proposed as a promising source of cells for site-specifi c repair of bone, cartilage, muscle, tendon, marrow stroma and other connective tissues (6). In addition, it has been shown that MSC have immunosuppressive and anti-infl ammatory effects that are useful for the treatment of transplant-related complications such as graft-versus-host disease (GvHD) (8 -10). Currently, many clinical trials are underway to explore the above-mentioned properties of MSC (11).
Despite the use of MSC in many clinical trials, the underlying mechanism by which it affects tissue repair and immunosuppression remains largely unknown. MSC have been postulated to differentiate into cells of the target tissue and functionally regenerate damaged, dysregulated or aged cells. In addition, direct and indirect effects of either cellcell interaction or secreted, soluble factors may play Cytotherapy, 2012;14: 1064-1079 a role in creating a local immunosuppressive environment (12). However, the above biologic effects of MSC still remain controversial (13) and may not represent the complete picture.
MSC have been found to enhance the viability of some cell types in co-culture systems (14 -16). For the ex vivo expansion of hematopoietic stem (HSC) and progenitor cells (HPC), it has been suggested that the MSC layer used in the co-culture system could provide soluble factors such as cytokines/chemokines, extracellular matrix (ECM) proteins and adhesion molecules that serve to regulate survival and maintenance of ' stemness ' . One key consequence of using MSC co-culture for HSC expansion is the ability of HSC to bypass the need for CD34 selection prior to expansion (17,18). While several groups have demonstrated the utility of MSC in umbilical cord blood (UCB)-mononucleated cell (UCB-MNC) expansion, it remains to be determined whether this expansion is the result of increased cell proliferation or reduced apoptosis. As such, we have used a co-culture system of MSC with umbilical cord blood (UCB) cells as a model to investigate the mechanism by which MSC could support the ex vivo expansion of UCB- MNC. Our data show that the viability supporting effect of MSC is most prominent when UCB-MNC and MSC are in physical contact, and results in reversal of early apoptosis and enhancement of the mitochondrial membrane potential of the UCB-MNC. Furthermore, we show that there is transfer of cytosolic content from MSC to UCB-MNC, and this could contribute to the reversal of early apoptosis in the UCB-MNC cells. We propose that this mechanism may, at least partly, contribute to the effect of MSC observed in clinical studies where tissue viability is restored after injection or implantation.

Cell cultures
The human embryonic stem cell (ESC)-derived MSC (ES-MSC) HuES9.E1 cell line was generated according to the protocol published by Lian et al. (19) and was maintained on a gelatin-coated (0.1% w/v gelatin in Ultrapure H 2 O; Millipore, Billerica, MA, USA) tissue culture plate/fl ask surface (Becton Dickenson Falcon, San Jose, CA, USA) in Dulbecco ' s modifi ed Eagle medium (DMEM; Invitrogen, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Hyclone, Thermo Scientifi c, Waltham, MA, USA), non-essential amino acids (MEM NEAA; Invitrogen) and penicillin-streptomycin-glutamine (PSG; Invitrogen) at 37 ° C in a humidifi ed, 5% CO 2 atmosphere. In some experiments, the HuES9.E1 cell line was transduced with green fl uorescent protein (GFP) using lentivirus as the vector to obtain the GFP ES-MSC.
The NIH-3T3 and human bone marrow (BM) MSC (BM-MSC) cell lines were maintained on a tissue culture plate/fl ask surface (Becton Dickenson Falcon) in DMEM (Invitrogen) supplemented with 10% and 20% FBS (Hyclone, Thermo Scientifi c), respectively. The BM-MSC was isolated from consented donor BM from Singapore General Hospital (SGH, Singapore) as approved by the hospital ethics committee. The BM samples were plated at a density of 3 ϫ 10 5 cells/cm 2 on tissue culture fl ask surfaces (Becton Dickenson Falcon) and grown till a confl uent feeder layer was achieved. The BM-MSC cells were checked for expression of MSC markers such as CD44, CD73, CD90, CD105, CD166 and HLA-ABC (20,21).
Fresh UCB was obtained from Singapore Cord Blood Bank (SCBB), and the use of the UCB samples was reviewed and approved by the institutional review boards (IRB) of each cord blood collection hospital as well as those of the Singapore General Hospital (SGH, Singapore) and National University of Singapore (NUS, Singapore). The UCB-MNC were isolated using Ficoll Histopaque-1077 (Sigma Aldrich, St Louis, MO, USA) density-gradient centrifugation and counted before cyropreservation in 90% v/v donor autoplasma with 10% v/v dimethyl sulfoxide (DMSO; Sigma Aldrich) for subsequent use. CD34 ϩ selected cells were obtained using Magnetic Activated Cell Sorting (MACS) cell-separation columns (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). The cryopreserved UCB-MNC were thawed using a thawing solution containing human albumin (HAS; 20% w/v; Health Sciences Authority, Singapore) and Onkovertin 40 (10% w/v; B. Braun, Melsungen, Germany). The cells were centrifuged at 400 g for 15 min at 10 ° C. The cells were then washed with Dulbecco ' s phosphate-buffered saline (DPBS; Hyclone, Thermo Scientifi c), followed by centrifugation at 300 g for another 15 min. The cells were fi nally resuspended in StemSpan-SFEM (STEMCELL Technologies, Vancouver, Canada) supplemented with 100 ng/mL human stem cell factor (SCF; PeproTech, Rocky Hill, NJ, USA), 100 ng/mL human thrombopoietin (TPO; PeproTech, Rocky Hill, NJ, USA), 50 ng/mL human Flt3-Ligand (Flt3; PeproTech) and 20 ng/mL human insulin-like growth factor binding protein 2 (IGFBP2; R&D Systems, Minneapolis, MN, USA). The initial seeding density for human UCB-MNC was set at 2.5 ϫ 10 5 cells/mL.
Viable cell counts of the ES-MSC, BM-MSC and NIH-3T3 were performed using trypan blue (Invitrogen), while crystal violet (Invitrogen) was used to count the nucleated UCB-MNC. A standard hemocytometer and upright microscope (bright fi eld with 10 ϫ magnifi cation) were used.

Stromal layer (ES-MSC, BM-MSC and NIH-3T3) and UCB-MNC co-cultures
For co-cultures that involved direct contact between the stromal layer and UCB-MNC, the ES-MSC, BM-MSC and NIH-3T3 were seeded at a density of 3.5 ϫ 10 4 cells/cm 2 in 1 mL media on 24-well plates (BD Falcon, Franklin Lakes, NJ, USA) and grown for 1 day (confl uency Ͼ 80%). Prior gelatin coating of the culture plates was required for the ES-MSC. The media were aspirated, followed by rinses with DPBS (Hyclone, Thermo Scientifi c), and 2.5 ϫ 10 5 cell/ mL UCB-MNC in StemSpan-SFEM (STEMCELL Technologies) containing the basal cytokines (SCF, TPO, Flt3 and IGFBP2) was inoculated to start the co-culture. UCB-MNC cultures without the stromal layer served as a control. The cultures were maintained in a humidifi ed, 5% CO 2 incubator at 37 ° C for 1, 2, 3, 7 and 11 days. Cytokine replenishment using the StemSpan-SFEM media was usually done on day 7 of the co-culture.
For co-cultures where ES-MSC were not in contact with UCB-MNC, a transwell insert (polyethylene terephthalate track-etched membrane, pore size 0.4 microns; BD Falcon) was used to separate these two cell populations in companion six-well plates (BD Falcon). ES-MSC-seeding parameters were the same as those described above except for group V (Figure 4), where the ES-MSC were grown on the lower side of the transwell membrane. In this case the seeding density of ES-MSC was increased three times. The transwells for group V ( Figure 4) were incubated in the inverted position in a humidifi ed, 5% CO 2 incubator at 37 ° C for 2 h before placing them in the tissue culture plate for further incubation. ES-MSC culture media were aspirated, and the ES-MSC were rinsed with DPBS (Hyclone, Thermo Scientifi c) before adding 3.0 mL/well UCB-MNC containing StemSpan-SFEM media (STEMCELL Technologies). Transwell inserts were subsequently placed into each well at a height of 0.9 mm above the bottom of the well. UCB-MNC in transwell inserts without ES-MSC in the wells served as controls. Cultures were maintained in a humidifi ed, 5% CO 2 incubator at 37 ° C. On specifi ed days, the cultures were harvested by washing with DPBS (Hyclone, Thermo Scientifi c).
For experiments requiring concentrated ES-MSC-conditioned media (CM) ( Figure 5A, B), the culture media from the ES-MSC fl ask were collected after 3 days and concentrated using dialysis. For the fresh fractionated ES-MSC CM ( Figure 5C), a Vivaspin 6 polyethersulfone column (Sartorius Stedim Biotech, France) with a molecular weight cutoff (MWCO) of 30 kDa was used. The two portions ( Ͼ 30 kDa and Ͻ 30 kDa) of the fractionated CM were obtained after centrifugation of the VivaSpin 6 column at 4000 g in a swing bucket centrifuge.

Separation of the combined, non-adherent and adherent fraction of the UCB-MNC after co-culture with the stromal layer
Upon co-culturing, the UCB-MNC that remained tightly attached to the ES-MSC (or BM-MSC and NIH-3T3) layer were defi ned as the adherent fraction of the UCB-MNC, while the rest, which were fl oating freely in the growth media, were defi ned as the non-adherent fraction of the UCB-MNC (22). In order to separate the two UCB-MNC fractions, the media were fi rst gently removed from the wells, followed by a wash using DPBS (Hyclone, Thermo Scientifi c) to allow the collection of the non-adherent UCB-MNC from the co-culture system. For the adherent fraction, 0.5 mL Accutase (PAA Laboratories GmbH, Pasching, Austria) was added to the wells and incubated for 5 min at 37 ° C in a humidifi ed 5% CO 2 atmosphere. The wells were then washed with DPBS (Hyclone, Thermo Scientifi c) and the adherent fraction of the UCB-MNC, along with the ES-MSC, was collected. A combination of the non-adherent and adherent fraction was defi ned as the combined fraction. Harvesting of the combined fraction was similar to the non-adherent fraction except that more rigorous washing was carried out with the DPBS.

Determination of intercellular transfer of cytosolic materials
The lentivirus GFP-expressing or quantum dot (QD) (Qtracker 525; Molecular Probes, Invitrogen, Grand Island, NY, USA)-labeled ES-MSC were seeded at a density of 7.5 ϫ 10 4 cells/well in 1 mL media in gelatin-coated 24-well plates (BD Falcon) and grown for 1 day. The media were aspirated, followed by rinsing with DPBS (Hyclone, Thermo Scientifi c, UT, USA), and 2.5 ϫ 10 5 cell/mL UCB-MNC were seeded in 1 mL UCB expansion media. Cell cultures without the GFP ES-MSC layer and the GFP ES-MSC layer without the UCB-MNC served as controls. Cultures were maintained for 1, 2 and 3 days in the incubator (37 ° C in a humidifi ed, 5% CO 2 atmosphere) and the adherent and non-adherent UCB-MNC were harvested as described above. The labeling of the ES-MSC with QD (Molecular Probes, Invitrogen) was performed according to the manufacturer ' s protocol.

Sorting of CD45
The cryopreserved UCB-MNC were thawed by the process described above. The cells were then labeled with Annexin-V -FITC (Beckman Coulter Inc.), 7-AAD (Beckman Coulter Inc.) and CD45 -PE -Cy7 (BD Pharmingen). To sort for the CD45 ϩ GFP ϩ cocultured UCB-MNC, the adherent and non-adherent UCB-MNC were obtained as described above. The cells were then labeled with CD45 -PE -Cy7 (BD Pharmingen). The labeled cells were sorted using a BD FACsAriaII (Customized Unit at National Cancer Center, Singapore). Single-stained cells and fl uorescent minus one (FMO) controls were used to determine the appropriate gating of the sort. The height and width of the forward and side scatters were used to eliminate tetraploid cells (see the supplementary Figure 1 to be found online at http:// www.informahealthcare.com/doi/abs/10.3109/14653 249.2012.697146). A 100-micron nozzle was used for the sort and the sorting effi ciency was Ͼ 90%.
CD34-selected UCB-MNC served as controls. Cultures were maintained in a humidifi ed 5% CO 2 incubator at 37 ° C for 4 days. A similar methodology was used to obtain the adherent and non-adherent fractions of the CD34-selected UCB-MNC. The collected CD34 cells were labeled with primary antibody CD45 (mouse-anti-human; BD Pharmingen, Franklin Lakes, CA, USA) (because an initial fl ow cytometer-based analysis of the CD34 cells showed 100% phenotypic expression of CD45) followed by a secondary antibody Alexa Fluor 568 (goat-antimouse; Molecular Probes, Invitrogen). The labeled CD34 cells were transferred to a microscopic slide by cytospin centrifugation and mounted using cover slips. Imaging (Z series) of the slides was done using a Carl Zeiss LSM710 system (40 ϫ oil immersion lens). The captured images were analyzed with ZEN 2009 Analysis software.

DNA extraction and analysis of CD45 ϩ GFP ϩ UCB-MNC
DNA of the CD45 ϩ GFP ϩ UCB-MNC was determined by variable number tandem repeat (VNTR). Genomic DNA from the sorted CD45 ϩ GFP ϩ UCB cells was extracted using a DNeasy blood and tissue kit (Qiagen, Valencia, CA, USA). Polymerase chain reaction (PCR) amplifi cation of human-specifi c locus D1S80 (23) was performed using the following primers: 5 ¢ -GAAA CTGGCCTCCAAACACTGCCCGCCG-3 ¢ and 5¢ -CTTGTTGGAGATGCACGTGCCCCTTGC-3 ¢ .  PCR was carried out using the DyNAzyme ™ EXT DNA polymerase (Finnzymes, Thermo Fisher Scientifi c, Vantaa, Finland) with the following conditions. DNA was fi rst denatured at 95 ° C for 2 min, then it was amplifi ed for fi ve cycles at 95 ° C for 30 s, 62 ° C annealing for 30 s, and this was followed by one cycle of elongation at 72 ° C for 45 s and a fi nal extension at 72 ° C for 10 min. Genomic DNA samples from non-co-cultured UCB-MNC and MSC were processed in parallel. Amplifi ed PCR products were electrophoresed in 2.6% agarose (Invitrogen) gels at 80 V, stained with ethidium bromide (Sigma Aldrich), and visualized under ultraviolet (UV) light (Bio-Rad, Hercules, CA, USA).

Data analyses
All the experiments were repeated at least three times, and data are presented as mean Ϯ SEM. Statistical analyses included t -tests, with a P -value of Ͻ 0.05 considered to be statistically signifi cant.

ES-MSC support ex vivo expansion and viability of UCB-MNC in co-culture
Previous studies have demonstrated that BM-MSC co-culture enhances ex vivo expansion of UCB-MNC (17,18,24,25). To confi rm this observation under our experimental conditions, we fi rst elucidated the effect of ES-MSC co-culture on ex vivo expansion of UCB-MNC. As shown in Figure 1A, the presence of the ES-MSC layer signifi cantly increased the UCB-MNC expansion in terms of total viable cells (12.5 Ϯ 0.5 versus 8.2 Ϯ 0.3), CD34 ϩ CD38 Ϫ cells (107 Ϯ 9 versus 61 Ϯ 4) and CFU-GM populations (201 Ϯ 14 versus 150 Ϯ 20) over the time -course of 11 days.
UCB hematopoietic progenitor cells, which have been shown to have a higher self-renewal capacity as assessed by cell-surface markers and gene expression profi le coupled with in vivo functional assays, have a greater tendency to adhere tightly to MSC in in vitro co-culture systems (22,26 -28). As such, in our MSC-UCB co-culture system, we studied the behavior of the adherent, non-adherent and combined UCB fractions separately.
We investigated the viability of the CD45 ϩ cells using the Annexin-V/7-AAD double-staining method by fl ow cytometer, where viable cells are defi ned as double negative for Annexin-V and 7-AAD. We demonstrated that immediately after thawing the UCB-MNC, the viability was 57.3 Ϯ 1.5%. After a 3-day culture period without co-culture, the UCB-MNC viability decreased to 24.6 Ϯ 0.7% (Figure 1B, D). In contrast, when UCB-MNC were co-cultured with ES-MSC, BM-MSC or NIH-3T3 (i.e. with stromal support), the percentage of viable CD45 ϩ cells was signifi cantly higher in the combined, nonadherent and adherent fractions of the UCB-MNC ( Figure 1B, C). In the presence of ES-MSC, the viability of the combined, non-adherent and adherent UCB-MNC was 53.7 Ϯ 0.6%, 49.5 Ϯ 0.8% and 71.5 Ϯ 0.9%, respectively ( Figure 1B). The NIH-3T3 and BM-MSC co-culture system also demonstrated a similar viability, enhancing the effect on the co-cultured UCB-MNC ( Figure 1B, C). In all our experiments, the adherent CD45 ϩ cells exhibited the highest viability compared with the other fractions of co-cultured UCB-MNC. Interestingly, the adherent cells from the ES-MSC, BM-MSC and NIH-3T3 co-culture had a reduction in early apoptotic cells (CD45 ϩ Annexin-V ϩ ) to less than 10% ( Figure 1C).
To confi rm further the viability of the supporting effect of the ES-MSC, we sorted for the early apoptotic CD45 ϩ Annexin-V ϩ UBC-MNC from a freshly UCB thawed sample ( Figure 1E) and cocultured the sorted cells with ES-MSC. As shown in Figure 1E, the co-cultured group (62.4 Ϯ 0.5%) had a signifi cantly higher population of viable CD45 ϩ Annexin V Ϫ cells compared with the non-co-cultured group (46.6 Ϯ 0.9%) on day 2.

The presence of ES-MSC prevented the loss of mitochondrial membrane potential in UCB-MNC
To determine whether the observed viability support effect of ES-MSC on UCB-MNC was correlated with the extent of loss of mitochondrial membrane potential, the mitochondrial membrane potential was studied using a JC-1 assay (Figure 2A). In healthy cells, JC-1, being a positively charged lipophilic dye, enters the intact and negatively charged mitochondrial membrane matrix, where it accumulates to form the red fl uorescent J-aggregate. However, in apoptotic cells the mitochondrial membrane potential collapses, thus preventing the accumulation of JC-1 in the mitochondrion and formation of the J-aggregate. Thus the green fl uorescent monomeric form remains in the cytoplasm of the apoptotic cells. The ratio of the red to green fl uorescence of a JC-1-stained cell, as measured with a fl ow cytometer, can be used to assess the mitochondrial-dependent apoptotic state of cells. In contrast with UCB-MNC cultured in the absence of ES-MSC (3.0 ϫ 10 5 Ϯ 0.2), a 3-day coculture with ES-MSC (6.8 ϫ 10 5 Ϯ 0.2) signifi cantly increased the number of UCB-MNC that exhibited intact mitochondrial membrane potential. In addition, the adherent UCB-MNC (4.6 Ϯ 0.2) had a signifi cantly higher JC-1 fl uorescence ratio (red/green) compared with the non-adherent cells (2.5 Ϯ 0.2) and the UCB-MNC without co-culture ( Figure 2B, C). We also observed a signifi cant reduction in caspase 3/7 and 9 activity when the UCB-MNC were co-cultured with ES-MSC (see the supplementary Figure  Taken together, these results supported the ability of ES-MSC to rescue the thawed UCB-MNC from early apoptosis by preventing the loss of mitochondrial membrane potential.

Transfer of cytosolic components from the ES-MSC to the UCB-MNC
Lentivirus-transduced ES-MSC expressing intracellular GFP were co-cultured in direct contact with UCB-MNC to elucidate the possibility of transfer of cytosolic components from the ES-MSC to the UCB-MNC. As depicted in Figure 3A, C, the adherent UCB-MNC (4.2 Ϯ 0.2%) had a signifi cant transfer of GFP from the ES-MSC, as evidenced by the presence of the CD45 ϩ GFP ϩ population by fl ow cytometry analysis. There was no background GFP fl uorescence observed in the control groups, which included non-co-cultured UCB-MNC and ES-MSC alone; in addition, signifi cantly lower percentages of transfer-positive cells were observed in the combined (0.8 Ϯ 0.04%) and non-adherent (1.0 Ϯ 0.1%) fractions over a 3-day culture period (Figure 3A, C).
A more detailed study of the adherent population revealed that the percentage of CD45 ϩ GFP ϩ cells increased from 1.0 Ϯ 0.03% to 3.3 Ϯ 0.6% from day 1 to day 3 ( Figure 3B). This provided strong evidence of the transfer of cytosolic material from ES-MSC to UCB-MNC over the 3-day co-culture period, in which the UCB-MNC CD45 ϩ cells became positive for green fl uorescence, which was not initially present in them. For further confi rmation of cytosolic transfer, ES-MSC were labeled with QD and cocultured with CD34-selected UCB-MNC. Within 24 h of co-culture, the UCB-MNC (CD45 ϩ cells) became positive for QD ( Figure 3D).
Moreover, to ensure that the GFP marker was actually enclosed by the CD45 ϩ cells, we carried out Z-series imaging of the CD45 ϩ GFP ϩ cells using fl uorescent confocal microscopy ( Figure 3E). The outer surface of the cell membrane was predominantly red because of the labeling of the CD45 antigen by CD45 (primary) and Alexa Fluor 568 (secondary) antibody. As we probed inside the cell (Z-plane 4; Figure 3E) we were able to detect the presence of the intracellular GFP (green arrow) bound by the red cell membrane (red arrow) of the CD45 cells.

The viability supporting effect of ES-MSC and cytosolic transfer from the ES-MSC to early HPC in UCB-MNC was most prominent when ES-MSC and UCB-MNC were in direct contact
To investigate whether direct contact between UCB-MNC and ES-MSC was necessary to prevent the UCB-MNC from undergoing early apoptosis and maintaining their viability, we cultured the UCB-MNC under three different conditions: in the presence of ES-MSC that were in direct contact with UCB-MNC (Figure 4; group VI), in the presence of ES-MSC but with the two cell populations separated by a transwell insert (Figure 4; groups III and V), and in the absence of ES-MSC co-culture (Figure 4; groups I and II). As demonstrated in Figure 4A, the optimal viability support for both the non-adherent (52.2 Ϯ 0.3%) and adherent (71.5 Ϯ 1.1%) fractions was achieved when the UCB-MNC were in direct contact with the ES-MSC layer on day 3. Large (0.9 mm) (groups III and IV) or small (membrane thickness of the transwell) (group V) physical separation of the two populations of cells by the transwell also signifi cantly reduced the viability. Furthermore, the use  of a non-viable MSC layer, whereby the MSC cells were fi xed with paraformaldehyde or by freezing, did not enhance UCB viability and its subsequent expansion (data not shown). Similarly, cytosolic transfer of GFP from the ES-MSC to the adherent UCB-MNC occurred only when the two cell populations were in direct contact with each other ( Figure 4B), with negligible transfer taking place in the other groups. A statistical correlation between higher levels of cytosolic transfer and enhanced UCB cell viability was observed ( Figure 4C).

Enhanced UCB-MNC viability is not the result of a paracrine effect or transdifferentiation of ES-MSC
Several research groups have reported that MSC helps in tissue regeneration and repair via the release of small biomolecules (29 -34) (paracrine effect) or by transdifferentiating (35 -38) into the required cell types. As such, we investigated whether these mechanisms could have played a role in the viability enhancement of the UCB-MNC. Concentrated CM from ES-MSC was used to culture the UCB-MNC cells ( Figure 5A, B). As seen in Figure 5A, the UCB-MNC viability cultured in the ES-MSC-concentrated (1 ϫ ) CM (17.6 Ϯ 0.3%) showed a signifi cant reduction compared with the co-cultured group (70.4 Ϯ 1.0%) after a 3-day culture period. Furthermore, in order to study the effect of the CM concentration, a higher (10 ϫ ) concentrated CM was used, and a similar decrease in viability (10.7 Ϯ 0.1%) was observed. The low viability of the UCB-MNC in the concentrated CM groups (1 ϫ and 10 ϫ ) translated to a signifi cantly lower expansion of the viable CD45 ϩ cells over an 11-day culture ( Figure 5B).
A similar experiment was repeated using fresh ES-MSC CM that was fractionated using a 30-kDa MWCO membrane. The viability of the UCB-MNC cultured in the Ͼ 30-kDa fraction (24.7 Ϯ 0.8%) and Ͻ 30-kDa fraction (32.0 Ϯ 1.4%) was significantly lower than the direct ES-MSC co-cultured control (64.2 Ϯ 0.8%) in a 3-day culture system. Finally, to rule out the possibility of cell fusion between the GFP-expressing ES-MSC and CD45 ϩ UCB-MNC or transdifferentiation of the ES-MSC to give rise to a CD45 ϩ GFP ϩ population, we performed fl uorescence-activated cell sorting to purify the specifi c diploid CD45 ϩ GFP ϩ population from the MSC-UCB co-culture system. Analysis of the extracted DNA was performed using VNTR at the polymorphic human locus D1S80. VNTR, which are repeated DNA sequences, are arranged in tandem with 7 -100 base pairs. Thus the number of repeats at the VNTR locus is unique for the MSC and UCB cells used in the coculture system, thereby enabling the recognition of the genetic identity of the respective cell population. PCR amplifi cation of D1S80 and agarose gel electrophoresis revealed that the amplifi ed DNA of the CD45 ϩ GFP ϩ cells corresponded with the DNA pattern of the UCB-MNC (Figure 6A), thus suggesting that the transfer GFP ϩ cells was of UCB-MNC origin but not an outcome of MSC transdifferentiation or MSC-UCB cell fusion. Also, a detailed assessment of the viability of CD45 ϩ GFP ϩ cells revealed that approximately 70 -80% of the transfer of GFP ϩ cells were viable over a time -course of 3 days ( Figure 6B). The CD45 ϩ GFP ϩ were also able to form colony forming units (supplementary Figure 3 to be found of cytosolic components from MSC to UCB-MNC is possible, and that the extent of transfer of GFP from MSC to UCB-MNC appears to be dependent on the contact of these two cell populations, which also correlates with the viability supporting effect. We were also able to identify the intercellular transfer taking place in primitive, viable HPC populations, as defi ned by phenotypic markers such as CD34 and CD38. In another study, Spees et al. (39) reported that cells could rescue surrounding cells from apoptosis by direct intercellular transfer of mitochondria. The phenomenon of direct mitochondria transfer between cells could play a role in the stromal cell viability supporting effect that has been observed in MSC -cardiomyocyte co-culture systems, as ischemic cardiomyocytes were rescued from cell death (40 -43), which is an observation similar to our current UCB ex vivo expansion system. Similar observations of cytosolic transfer have been reported by Gillette et al. (44), who identifi ed that the transfer of cytosolic materials from HPC to osteoblasts plays an important role in the production of stromal-derived factor-1 (SDF-1) by the osteoblasts, which in turn control HPC homing in the BM. Therefore it may be reasonable to suggest that the process of cytosolic transfer is bi-directional, with multiple biologic functions that help in regulating the ex vivo osteoblastic and vascular niche (45).

Discussion
To our knowledge, this is the fi rst study that addresses intercellular transfer as a mechanism of MSC enhancing the viability of UCB cells during ex vivo expansion. We have demonstrated that MSC co-culture supports the viability of frozen-thawed UCB cells, and this effect was best seen in the presence of cellcell contact with MSC. We have also shown that the transfer of cytosolic materials occurs between MSC and co-cultured UCB cells, and that a statistically signifi cant positive correlation exists between cytosolic transfer and enhanced UCB cell viability.
Our data indicate that the viability supporting effect of stromal cells (ES-MSC, BM-MSC as well as NIH-3T3) can be mediated through the reversal of early apoptotic events in UCB-MNC, as demonstrated by the reduction in phosphatidylserine externalization coupled with the prevention of the loss of mitochondrial membrane potential and reduction of caspase activation. The viability supporting effect of the ES-MSC was optimal only during direct contact, with adherent UCB-MNC exhibiting the highest viability, illustrating that secreted factors are not likely to be the most important determinant, but rather a cellto-cell contact-dependent process might be involved. The use of concentrated or fractionated ES-MSC CM did not exhibit a similar viability supporting effect on the UCB-MNC. The use of a GFP-labeled ES-MSC layer indicated that intercellular transfer organelles (51) and mRNAs (52). Valadi et al. (52) have demonstrated the transfer of functional genetic materials in the form of mRNAs and microRNAs through exosomes.
According to our data, direct contact accompanied by the adhesion of UCB-MNC to the MSC layer is the single most important factor in regulating the intercellular transfer of GFP from the ES-MSC to the UCB-MNC. This observation is in accordance with those made by Gillette et al. (44), who reported that cytosolic transfer from HPC to the osteoblasts would only occur when the two cell layers were in direct contact with each other. Moreover, Wagner et al. (22,27) have demonstrated that HPC that have a signifi cantly higher self-renewing ability (CD34 ϩ CD38 Ϫ as opposed to CD34 ϩ CD38 ϩ ) dominate the adherent fraction. In these studies, uropod formation by the HPC (which was also observed in our co-culture system) to anchor itself to the MSC was described as the main mechanism of cellular contact and adhesion between the two cell layers (22,27,44,53,54). Adhesion proteins such as Ncadherin, cadherin-11 and very late antigen-4 (VLA-4) are also known to be highly expressed in the adherent portion compared with the non-adherent portion (44,45). Furthermore, in vivo studies involving connexin 43 (Cx43)-defi cient mice have shown that gap junctions are critical in regulating hematopoiesis in BM and thymus, via gap junction-mediated intercellular communications (53). In addition, ECM (consisting mainly of collagen and fi brinogen) created by the viable MSC layer may help in maintaining the viability of the UCB-MNC by mimicking an in vitro hematopoietic stem cell niche (55).
Based on the current data, it would be important to identify the type of cellular elements, signaling molecules, soluble factors, nutrients and/or organelles, that are transferred between the ES-MSC layer and UCB-MNC, as the identifi cation of the cellular elements would shed new light on how these two cell populations communicate and the mechanism possibly responsible for the viability supporting effect of ES-MSC. Based on the size of the surrogate markers GFP and QD, which were shown to transfer from the ES-MSC to the UCB-MNC, it may be reasonable to suggest that molecules such as nutrients, antioxidants or even mRNAs, which are much smaller than GFP or QD, could also transfer in great quantity and mediate the biologic response of higher viability in the UCB-MNC.
In conclusion, our current study has demonstrated the transfer of cytosolic content from ES-MSC to UCB-MNC, and this phenomenon might mediate the reversal of early apoptosis of UCB. Further investigation of the intercellular communication between ES-MSC and UCB cells would shed new light on