Efficiency of Exosome Production Correlates Inversely with the Developmental Maturity of MSC Donor

1Institute of Medical Biology, A*STAR, 8A Biomedical Grove, 138648 Singapore 2National University of Singapore, Graduate School for Integrative Sciences and Engineering, 28 Medical Drive, 117456 Singapore 3Laboratory of Experimental Cardiology, University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, the Netherlands 4Bioprocessing Technology Institute, A*STAR, 20 Biopolis Way, 138671 Singapore 5Department of Surgery, YLL School of Medicine, NUS, 5 Lower Kent Ridge Road, 119074 Singapore 6Interuniversity Cardiology Institute of the Netherlands, Catharijnesingel 52, 3511 GC Utrecht, the Netherlands 7Department of Maternal Fetal Medicine, KK Women’s and Children’s Hospital, 100 Bukit Timah Road, 229899 Singapore 8Key Laboratory of Agricultural Animal Genetics, Breeding and Reproduction of Ministry of Education, Key Laboratory of Freshwater Animal Breeding, Ministry of Agriculture, College of Fisheries, Huazhong Agricultural University, Wuhan, Hubei, 430070, People’s Republic of China

The therapeutic efficacy of MSC transplantations in the treatment of different diseases had been rationalized on the differentiation potential of MSCs to replace lost or injured cells. However, this differentiation-based mechanism has become increasing untenable. In animal models and even patients where MSC transplantation elicited a therapeutic response, <1% of transplanted MSCs reach their target MSC secretion as the therapeutic agent has been proposed [65][66][67][68][69][70][71][72][73]. Our group demonstrated that MSC secretion alone could improve cardiac function in pig and mouse models of acute and chronic myocardial ischemia [74][75][76] and identified exosome, a membrane vesicle as the therapeutic factor in the secretion [77,78]. Others subsequently confirmed exosomes as the factor mediating the therapeutic trimester fetal tissues [77,79] for the production of secretion and exosomes. Since developmental stage of the donor is a dominant determinant of the therapeutic potency of MSCs and exosome is the factor mediating this potency, we tested here if exosome production is correlated to the developmental stage of the donor tissue by comparing exosome production from hESC-derived MSC and the developmentally more mature tissues such as the umbilical cord from a full term delivery.
In our hands, MSCs derived from umbilical cord could not be expanded beyond six passages while MSCs derived from human ESCs and fetal tissues could be passaged for more than 20 passages to generate up to 10^19 cells [79,80]. In addition, cord MSCs have a longer doubling time. Therefore, expanding cord MSCs to produce sufficient exosomes for comparative studies with exosomes from either ESC-or fetal MSCs was not feasible. To circumvent this issue, we immortalized the cells by over-expressing MYC gene. We had previously use this method to immortalize human ESC-derived MSCs and observed that despite some changes, the immortalized cells retained many fundamental MSC characteristics including the production of cardioprotective exosomes [81].
In this study, we characterized MYC-immortalized 109 cord MSCs, their production of exosomes and the efficacy of their exosomes in reducing reperfusion injury in a mouse model of ischemia/reperfusion injury, and compared these against our previous analysis of MYCimmortalized ESC-MSCs.

Derivation of cord MSCs
The collection of umbilical cords of term babies after obstetric delivery was carried out in KK Women's and Children's Hospital under an IRB approved protocol (CIRB 2009/289/D). Umbilical cords were stored in DPBS with 10 μg/ml gentamycin at 4°C during the transfer to the lab. To isolate MSCs, the cord was cut into 3 cm-long pieces, rinsed with DPBS + gentamycin to remove as much blood as possible and then cut lengthwise to remove blood vessels. They were digested with 300 U/ml collagenase, 1 mg/ml hyaluronidase and 3 mM CaCl 2 in PBS for 1 h at 37°C with occasional agitation. The cord pieces were then crushed with forceps to release cells from the Wharton's jelly and then digested with 0.05% trypsin-EDTA for 30 min at 37°C before being crushed again with forceps. The cell suspensions were combined, washed and cultured as previously described [75].

Oncogenic transformation of cord MSCs
MYC transformation of cord MSCs was performed using a lentivirus carrying the CMYC gene as previously described [81]. Briefly, cord MSCs were plated at 10^6 cells per 10 cm dish and infected with viruses at a MOI of 5 in the presence of 4 μg/ml polybrene overnight. The following day, culture medium was replaced with fresh medium and then 48 h later with medium containing puromycin (2 μg/ml). After 72 h of puromycin treatment, the surviving cells were allowed to expand. Clonal lines from each of three independently infected cell cultures were derived by limiting dilution. When individual clones were expanded to 10^7 cells per clone (or a confluent 15 cm culture dish), the cells were designated p1. Three clonal lines were generated and named CMSC3A1, CMSC3A3 and CMSC3A4 lines, respectively. Integration of the CMYC or GFP transgene was confirmed by amplifying genomic DNA using specific primers for exon2 and exon3 of CMYC respectively: 5'-GCCCCTGGTGCTCCATGAGGAGACACC'-3' and 5'-ACATTCTCCTCGGTGTCCGAGG-3' using the following PCR conditions: one cycle of 94, 2 min; 32 cycles of 94°C, 15 s; 60°C, 30 s; 72°C, 90 s and one cycle of 72°C 5min. The PCR products were resolved on a 1% agarose gel. Differentiation of the MYC-MSCs to adipocytes, chondrocytes and osteocytes was performed using adipogenic, chondrogenic and osteogenic hMSC Differentiation Bullet Kits, respectively (Lonza, Walkersville, MD) according to manufacturer's instructions. Karyotyping by G-banding was performed by the Cytogenetics Laboratory, KKH. The MYC-specific primer set is 5' ACT TAG TTG CGT TAC ACC C 3' and 5' AAA TAA AGC CAT GCC AAT CTC 3' .

Telomerase activity
Relative telomerase activity was measured by SYBR ® Green real time quantitative telomeric repeat amplification protocol assay using a modified method as described by Wege et al. [82]. Briefly, 3×10^6 cells were harvested and cell lysate was prepared using a commercially available mammalian cell extraction kit (Cat K269-500-1, BioVision, Milpitas, CA). The reagents for the PCR amplification was 1 μg of protein cell lysate, 10 μL of 2 X SYBR Green Super Mix (Cat 170-8880, BioRad, Hercules, CA) with 0.1 μg of TS primer (5'-AATCCGTCGAGCAGACTT-3'), 0.1 μg of ACX primer (5'-GCGCGG[CTTACC]3CTAACC-3') and 10 mM EGTA in a total volume of 25 μL. The reaction was first incubated at 25°C for 20 min to allow the telomerase in the cell lysate to elongate the TS primers followed by 2 min incubation at 95°C to inactivate telomerase activity and denature the primers. The telomerase product was amplified by PCR for 40 cycles of 95°C, 30s; 60°C, 90s. The relative telomerase activity was assessed against that of HEK293 cells using the threshold cycle number (or Ct value) for 1 μg protein cell lysate.

Rate of cell cycling
To assess cell cycle rate, 2×10^7 cells were pre-labelled in 2 ml of 10 μM CFDA (Molecular Probe, Eugene, OR) in PBS at 37°C for 15 min, cultured for 24 h and then replated at 5×10^4 cells per well in 6-well coated with gelatin. At 0, 24, 48, and 72 h, cells from duplicate wells were harvested, and fixed in 2% paraformaldehyde, and analyzed on FACSplus (Becton Dickinson; San Jose, CA). The number of cell cycles per 24 h was calculated assuming that each halving of cellular fluorescence represented one cell division. Therefore, the number of cell cycles per 24 h (n) was calculated as n=lg(F-Fn)/lg2 where F is initial average cellular fluorescence and Fn is the average cellular fluorescence after 24 h. The number of cell cycles was then plotted against time to derive the average time per cell cycle.

Illumina gene chip analysis
Total RNA was prepared in technical triplicates from different passages of MSCs using Illumina ® TotalPrep RNA Amplification Kit (Ambion, Inc., Austin, TX). The MSCs were HuES9-E1 MSCs at p15 and p16; E1-MYC 21.1 at p3, p4, and p5; E1-MYC 16.3 line at p4, p7, and p8; CMSC3A1 at p4, p5, and p6; CMSC3A3 at p4, p5, and p6; and cord MSCs at p1 and p2. 500 ηg RNA was converted to biotinylated cRNA using the Illumina RNA Amplification Kit (Ambion, Inc., Austin, TX) according to the manufacturer's directions. 750 ηg of the biotinylated cRNA were hybridized to the Sentrix HumanRef-8 Expression BeadChip Version 3 (Illumina, Inc., San Diego, CA). Washing and scanning were performed according to the Illumina BeadStation 500x manual. The data were analyzed using Genespring GX 10. Quantile normalization was performed by a shift to 75th percentile, and the normalized data were baseline transformed to the median of all samples.

Exosome preparation
Exosomes were purified from MSC conditioned medium (CM) by size exclusion using HPLC as previously described [77,81].

Mouse model of myocardial ischemia/reperfusion injury
Cardioprotective effect of the exosomes was tested in a mouse model of ischemia and reperfusion injury. MI was induced by 30 min left coronary artery (LCA) occlusion and subsequent reperfusion. 5 min before reperfusion, mice were intravenously infused with 200 μl saline solution of 0.3 μg exosome protein purified from culture medium conditioned by MYC-MSCs. Control animals were infused with 200 μl saline. After 24 h reperfusion, infarct size (IS) as a percentage of the area at risk (AAR) was assessed using Evans' blue dye injection and TTC staining as described previously [75].

Statistical analysis
Two-way ANOVA with post-hoc Dunnett was used to test the difference in infarct size between groups. Correlation coefficient of each pairs of array was assessed using Pearson correlation test.

Immortalization of cord MSCs
Primary umbilical cord derived MSCs have a limited expansion capacity of about 6 passages in our hands. As such, a single cord preparation would potentially generate only 10 8 MSCs against >10 19 MSCs from a single preparation using human ESCs or fetal tissue. To generate sufficient quantities of exosomes for comparative analyses against those produced by hESC-derived MSCs (HuES9.E1) [80] and the MYC-234 immortalized HuES9.E1 or E1MYC [81], primary cord MSCs, p3 were infected with lentivirus carrying the MYC oncogene. After puromycin selection, surviving cells were re-plated at low density ranging from 20,000-50,000 cells per 10 cm plate to produce physically well separated colonies and selected colonies were expanded to establish clonal lines. Three colonies from three independent infections were eventually selected to establish CMSC3A1, CMSC3A2 and CMSC3A3 clonal lines. The transformed cells were smaller and rounder with prominent nuclei. They had reduced adherence to plastic culture and reduced contact inhibition at confluence so that the cells formed clusters instead of adhering to the plastic dish as a monolayer ( Figure 1A). PCR amplification of genomic DNA revealed that the MYC transgene was integrated into the genome ( Figure 1B). The level of MYC transcripts in CMSC3A1 was higher than that in HuES9.E1 MSCs, but lower than E1MYC16.3 MSCs ( Figure 1C). Since MYC was reported to promote cell immortalization by activating telomerase to maintain telomeric repeats [83], we determined telomerase activity and observed a concordance between telomerase activity and MYC expression level. Telomerase activity in CMSC3A1 cells was higher than that in HuES9.E1 but lower than that in E1MYC16.3 which was established by MYC transformation of HuES9.E1 ( Figure 1D). CMSC3A1 also has a normal 46 XY karyotype ( Figure 1E). Consistent with the lower telomerase activity, MYC-transformed cord MSC lines have a cell cycle of ~13 hours which is longer than the 11 hours for E1MYC16.3 but shorter than the 19 hours for HuES9.E1 ( Figure 1F).

Characterization of MYC-immortalized cord MSCs
The surface antigen profile of the MYC-transformed cord MSCs, CMSC3A1, was qualitatively similar to that of MYC-transformed E1MYC16.3. The cells were CD29 + ,CD44 + , CD49a + CD49e + , CD73 + CD105 + , CD166 + , MHC I¯, HLA-DR¯, CD34¯ and CD45¯ (Figure 2A). The in vitro differentiation potential of CMSC3A1 was also similar to that of immortalized human ESC-derived MSC line, E1MYC16.3 [81]. Like E1MYC16.3, CMSC3A1 cells differentiated readily into chondrocytes and osteocytes but not adipocytes ( Figure 2B) [84]. During induction of adipogenesis which consisted of 4 cycles of a 6-day treatment of 3 days' exposure to induction medium and 3 days' exposure to maintenance medium, most CMSC3A1 cells like E1MYC16.3 died during exposure to the induction medium. These observations suggested that MYCtransformed cord MSCs cannot undergo adipogenic differentiation which is a defining property of MSCs.

Gene expression profile
The genome-wide gene expression profile of HuES9.E1, primary cord-derived MSCs, E1MYC16.3, E1MYC21.1, CMSC3A1 and CMSC3A3 was determined by microarray hybridisation on the Illumina Sentrix HumanRef-8 Expression BeadChip containing more than 24,000 unique features, and assessed for the relatedness between cell types. The expression profile of CMSC3A1 resembled that of MYCtransformed human ESC-derived cell lines, E1MYC16.3 (correlation coefficient, r 2 = 0.95) more than that of its parental cord-derived MSCs (r 2 = 0.92) or HuES9.E 1 (r 2 = 0.92) (Fig. 3A) as illustrated by hierarchical clustering ( Figure 3B). Upon MYC immortalization, 295 and 377 genes were respectively up-and down-regulated by 2.0 fold in cord-derived CMSC3A lines while only 86 and 120 genes were similarly up-or down-regulated in the human ESC-derived lines. Of these genes, 25 genes were up-regulated and 39 down-regulated genes in both MYCtransformed MSCs ( Figure 3C).

Isolation of exosomes from culture medium conditioned by CMSC3A1
We had previously demonstrated that exosomes secreted by ESCderived MSCs and their MYC-transformed progeny were protective in a mouse model of myocardial ischemia and reperfusion injury [75,77,81]. To test if transformed cord MSCs also produced similar exosomes, CMSC3A1 were grown in a chemically defined medium, the conditioned culture medium (CM) was harvested and exosomes purified as previously described [77,85]. The HPLC protein profile of the CM was similar to that of CM from ESC-derived MSCs and their MYC-transformed progeny [77] (Figure 4A) with the fastest eluting fraction having a retention time of about 12 minutes. Dynamic light scattering analysis of this peak revealed the presence of particles with a hydrodynamic radius range of 50-65 ηm. Western blot analysis of this peak also revealed the presence of exosome-associated proteins such as CD9 and CD81 ( Figure 4B), suggesting that this peak contains the exosome fraction of the CM. Notably, MYC protein was not detected in any of the exosome fractions.     Figure 5A). The relative infarct size (IS/AAR) in mice treated with E1MYC16.3 exosomes or CMSC3A1 exosomes was 22.6 ± 4.5%, and 19.8 ± 2.9%, respectively and their relative infarct sizes were significantly lower than the relative infarct size of 38.5 ± 5.6% in salinetreated mice (P<0.002 and P<0.001, respectively) ( Figure 5B).

Discussion
Human cord-MSCs, like human ESC derived-MSCs could be immortalized by over expression of MYC gene to increase telomerase activity, enhance rate of proliferation and bypass senescence. The immortalized MSCs retained many of the MSC characteristics. Their genome-wide gene expression profile was highly similar to that of their parental cells with a correlation coefficient of 0.92. However, we observed this gene expression profile was more similar to that of MYC-immortalized ESC-MSCs with correlation coefficient of 0.95. The immortalized cells also have the characteristic MSC surface antigen profile: CD29 + , CD44 + , CD49a + CD49e + , CD105 + , CD166+, MHC I¯, HLA-DR¯, CD34¯ and CD45¯. However, in contrast to a previous report that observed no fundamental changes in MSC properties after MYC immortalization [86], several MSC features were altered.
These alteration included a reduced adherence to plastic and a failure to undergo adipogenesis, and were similarly observed in ESC-MSCs after MYC immortalization [81]. Notwithstanding these changes, the MYC-immortalized cord MSCs like the MYC-immortalized ESC-MSCs retained a normal karyotype and the potential albeit limited to differentiate, suggesting a non-tumorigenic phenotype. Indeed, MYC-immortalized ESC-MSCs failed to engraft when transplanted in immune compromised mice (unpublished data). In addition, MYCimmortalized cord MSCs also secreted exosomes that were equally efficacious as those from MYC-immortalized ESC MSCs and could reduce infarct size in a mouse model of ischemia/reperfusion injury [81]. As ESC-MSC exosome-mediated reduction in infarct size was previously shown to correlate with an improvement in cardiac function, ATP/ADP and NADH/NAD ratios, immune cell infiltration and survival signalling [78], the similar reduction mediated by exosomes from MYC-immortalized cord MSCs implied that they have similar efficacy in restoring cardiac function.
The most notable difference between the MYC-immortalized cord-and ESC-MSCs was the exosome yield. A liter of culture medium conditioned by MYC-immortalized cord MSCs yielded 177 μg exosomes while that conditioned by MYC-immortalized ESC MSCs yielded 1282 μg exosomes. This exosome yield for MYC-immortalized cord-MSCs (177 μg per liter) was even lower than that of nonimmortalized ESC-MSCs and fetal-MSCs which produced ~500 and ~800 μg per liter of conditioned medium respectively (unpublished data) but was comparable to MYC-immortalized adult bone marrow-MSCs (unpublished data). We generally observed that after MYC immortalization, MSCs produced more exosomes and we attributed this to the smaller cell size of the immortalized MSCs which resulted in  a higher cell density for each confluent culture. The inverse correlation between the exosome production with developmental maturity of the donor tissues supported the correlation between therapeutic efficacy of MSC secretion and therefore MSCs with the "youthness" of the donor. It is possible that the composition of the exosomes from MSCs of donors at different developmental stages is different. However, this difference in composition is not likely to contribute to the efficacy of the exosomes from MSCs of different donors as we observed that similar efficacy was achieved with the same exosome dosage. We therefore propose that the correlation between MSC therapeutic efficacy and developmental stage of the donor is underpinned by exosome production.