A highly efficient non-viral process for programming mesenchymal stem cells for gene directed enzyme prodrug cancer therapy

Mesenchymal stem cells (MSCs) driven gene-directed enzyme prodrug therapy has emerged as a potential strategy for cancer treatment. The tumour-nesting properties of MSCs enable these vehicles to target tumours and metastases with effective therapies. A crucial step in engineering MSCs is the delivery of genetic material with low toxicity and high efficiency. Due to the low efficiency of current transfection methods, viral vectors are used widely to modify MSCs in preclinical and clinical studies. We show, for the first time, the high transfection efficiency (> 80%) of human adipose tissue derived-MSCs (AT-MSCs) using a cost-effective and off-the-shelf Polyethylenimine, in the presence of histone deacetylase 6 inhibitor and fusogenic lipids. Notably, the phenotypes of MSCs remained unchanged post-modification. AT-MSCs engineered with a fused transgene, yeast cytosine deaminase::uracil phosphoribosyltransferase (CDy::UPRT) displayed potent cytotoxic effects against breast, glioma, gastric cancer cells in vitro. The efficiency of eliminating gastric cell lines were effective even when using 7-day post-transfected AT-MSCs, indicative of the sustained expression and function of the therapeutic gene. In addition, significant inhibition of temozolomide resistant glioma tumour growth in vivo was observed with a single dose of therapeutic MSC. This study demonstrated an efficient non-viral modification process for MSC-based prodrug therapy.


Results
An efficient non-viral linear polyethylenimine (LPEI) based transfection method for AT-MSCs modification. AT-MSCs (Age group [18][19][20][21][22][23][24][25][26][27][28][29][30] were transfected with a plasmid encoding GFP reporter gene in 24-well tissue culture vessels to evaluate the transfection efficiencies of LPEI and Lipofectamine 3,000 (L3K). Although, there were more cells transfected using LPEI, the number of adherent cells were less than when using L3K (Fig. S1a). While the cell viability post-transfection remained high, there was a significant reduction in adherent cell number after LPEI-mediated transfection when compare to un-transfected control. The number of adherent cells further reduced with the use of increasing amounts of plasmid DNA (pDNA) (Fig. S1b), consistent with previous observations [39][40][41] . Attempts to attain high adherent cell number by transfecting AT-MSCs at 200 ng of pDNA with lower amounts of polymers only resulted in significantly reduced transfection efficiency (Fig. 1).
Next, we explored the use of the Enhancer 36 with low amount of pDNA (200 ng) and various ratios of DNA:polymer for the enhancement of transfection (Fig. 1). More than 80% of AT-MSC cells were transfected (Fig. 1a), with comparable number of adherent cells and viability to non-transfected control (Fig. 1b, c). With no apparent reduction in cell number and viability, this data suggests transfection of AT-MSCs in the presence of Enhancer does not affect cell proliferation or incur cytotoxicity. We next extended the study to include other AT-MSC isolated from another donor (Age group [31][32][33][34][35][36][37][38][39][40][41][42][43][44][45]. Using the same protocol (at the ratio of 1 µg pDNA to 5 µL LPEI), the efficiency was as high as 80% of cells transfected with minimally 200 ng of pDNA (Fig. S2). This condition was used in further studies.
Determination of the functionality of CDy::UPRT_AT-MSCs. In order to generate AT-MSCs expressing fused cytosine deaminase and uracil phosphoribosyltransferase (CDy::UPRT_AT-MSCs), cells were transfected with the CpG free expression plasmid encoding the transgene using LPEI. Based on immunocytochemistry analysis, transfection was significantly improved in the presence of Enhancer even at low amount of pDNA (200 ng). In the absence of the Enhancer, increasing pDNA amount modestly increased transfection efficiency of LPEI and Lipofectamine 3,000 (Fig. 2a), suggesting modulation of intracellular trafficking to Scientific RepoRtS | (2020) 10:14257 | https://doi.org/10.1038/s41598-020-71224-2 www.nature.com/scientificreports/ contribute to the high transgene expression. Extending this observation, we constructed a fusion gene encoding cytosine deaminase, uracil phosphoribosyltransferase and green fluorescent protein (CDy::UPRT:GFP) for direct visualization and quantification. In the presence of Enhancer, transfection efficiency was significantly increased (~ 80%) as compared to the use of LPEI alone (~ 40%; Fig. 2b), with no significant change in viability (Fig. 2c). Notably, there was no significant difference in the anticancer efficiency of AT-MSC modified with CDy::UPRT:GFP or CDy::UPRT (Fig. S3), suggesting that addition of GFP tag did not affect CDy::UPRT function. Collectively, the results demonstrated a significant improvement in the transfection of AT-MSCs by the use of the Enhancer, which likely shares a similar mechanism in facilitating intracellular trafficking of pDNA in BM-MSC 36 . Next, we examined the sensitivity of modified MSCs to 5FC using MTS assay. Exposure of CDy::UPRT modified AT-MSCs to 5FC reduced cell viability over time (Fig. S4a). This was to verify that the modified MSCs could express CDy::UPRT at levels sufficiently high that could induce significant cell death in the presence of prodrug. CDy::UPRT_AT-MSCs demonstrated increased sensitivity to the active cytotoxic drug 5FU (Fig. S4b). This effect was likely to be due to the activity of UPRT transgene, which catalyses the conversion of 5FU to 5-fluorouridine monophosphate 20 , consistent with previous observations 18,42 . Phenotypic characteristics of AT-MSC are not affected by the LPEI based transfection method. To explore the possibility that high transfection may modify the phenotype of AT-MSC, immunophenotyping of CDy::UPRT_AT-MSCs was carried out by standard FACS analysis using markers as defined by the International Society for Cellular Therapy (ISCT) 43 . The CDy::UPRT_AT-MSCs displayed identical profiles to un-transfected AT-MSCs where both cell types were found to be positive for CD90, CD73 and CD105 while negative for CD14, CD20, CD34, CD45 and HLA-DR (Fig. 3a). Expression of CDy::UPRT did not affect the differentiation capability of AT-MSCs into osteogenic (Fig. 3b) and adipogenic (Fig. 3c) lineages. Evidently, the presence of oil droplets in transfected cells indicated that the potential to differentiate into adipogenic lineage was unaffected by transfection and transgene expression (Fig. S5). In a separate study, chondrogenic differentiation was also unaffected after transfection using this method (unpublished data). Twentyfour hours later, bright field and fluorescent images were captured. Representative images are presented. Then, cells were trypsinised, pelleted and resuspended in 1XPBS for flow cytometry analysis. Transfection efficiency was calculated as the percentage of GFP positive cells normalized to the total number of cells as quantified by FACS. Bar graph represents mean ± SD, n = 3. In the same experiment, (b) total number of adherent cells and (c) cell viability of each condition was determined with NC-3000 cell counter. Un-transfected population serves as control. Results are presented as mean ± SD, n = 3.  cultured in 24-well vessels were transfected at various amount of CDy::UPRT expression plasmid with LPEI or Lipofectamine 3,000, using centrifugation or manufacturer's protocol respectively. After 24 h of incubation, cells were fixed with 4% paraformaldehyde and stained for CDy (green) and nucleus (Hoechst stain, blue). Representative images are shown. Scale bar represents 400 µm. (b) AT-MSCs in 6 well culture vessels were transfected with 1 µg of CDy::UPRT::GFP pDNA complexed with LPEI (1 µg of DNA to 5 µL of LPEI). After centrifugation, transfection mixture was replaced with fresh media (with or without Enhancer). One day later, representative images were acquired, and cells were trypsinised for FACS analysis. Results are presented as mean ± SD (n = 4). Un-transfected AT-MSCs served as negative control for FACS analysis. Significant differences between the transfection conditions were calculated using two tailed Student's t-test. **P < 0.01. (c) In the same experiment, total number of the cells and cell viability of each condition was determined with NC-3000 cell counter. The percentage of total adherent cells in transfected population at control (Un-transfected) was calculated. Data represented mean ± SD, n = 3. Here, the invasion assay was used to examine the vectorial migration of AT-MSCs through extracellular matrix in the presence of cancer cells. Invasion of AT-MSCs through extracellular matrix was significantly induced by MDA-MB-231, U-251MG and MKN1 but not HEK293T (Fig. 4a). This observation was consistent with previous reports of the chemo attraction of MSCs to cancer cells but not HEK293T 45,46 . Comparable numbers of migrated AT-MSCs and CDy::UPRT_AT-MSCs to cancer cells were observed, suggesting that the tumour homing capability was unaffected by transfection and the over-expression of CDy::UPRT. The number of CDy::UPRT_AT-MSCs that invaded through the extracellular matrix was correlated to the number of cancer cells and higher numbers of cells migrated towards U-251MG and MKN1, and lesser towards MDA-MB-231 cell lines (Fig. 4).

Scientific RepoRtS
CDy::UPRT_AT-MSC/5FC mediated cytotoxicity in vitro. Demonstrating a cytotoxic effect of the CDy::UPRT_AT-MSCs on target cancer cells is a prerequisite for the adoption of this non-viral method of therapeutic gene modification in generating theranostic MSCs for prodrug cancer therapy. The effect of cytosine deaminase/5FC in proliferation inhibition is commonly assessed by MTS assay 47,48 . With the same approach, we first compared the anticancer effects of CDy::UPRT_AT-MSC/5FC and 5FU in glioma, breast cancer and gastric cancer cell lines (Fig. S6). At 1:1 ratio of CDy::UPRT_AT-MSC to cancer cells, the anticancer effects were comparable to the direct pharmacological effects of 5FU. To further examine the therapeutic potential of CDy::UPRT_AT-MSC/5FC, cells were directly co-cultured with target cancer cells at various MSC to cancer cell number ratios (Fig. 5a). Proliferation inhibition by almost 57%, 69% and 89% were observed even at co-culture ratios of 1:50 of CDy::UPRT_AT-MSC/5FC to U-251MG, MDA-MB-231, and MKN1, respectively. This ratio of mixed culture represents 2% of therapeutic cells within the cancer populations. With 10% of therapeutic cells to cancer cells, the extent of proliferation inhibition was greater than 86%. Notably, 85% proliferation inhibition was seen with only 1% of therapeutic cells in the MKN1 population. Proliferation inhibition was not observed in co-cultures without the addition of the prodrug 5FC (Fig. 5b). www.nature.com/scientificreports/ Next, we explored the anticancer effects in scenarios where the therapeutic cells might not be in direct contact with the cancer cells by seeding modified MSCs in the upper chambers of transwells. Four day after exposure of MDA-MB-231 to CDy::UPRT_AT-MSC/5FC, close to 90% proliferation inhibition was observed (Fig. 5c). The anticancer efficiency of CDy::UPRT_AT-MSC/5FC in the absence of cell-cell contact was highly comparable to the direct co-culture model. Taken together, these data suggested that a potent anticancer effect can be exerted when therapeutic cells are in contact or in close proximity to the target cells. We next extended the study to compare the sensitivity of Hs738 (a non-transformed human fetal gastric/intestinal cells) and 5 gastric cancer cell lines. CDy::UPRT_AT-MSC/5FC exerted anticancer effect selectively to the gastric cancer cell lines (Fig. S7), suggesting preferential targeting of the therapeutic cells/5FC to cancerous but not non-transformed cells.
LPEI/enhancer generates highly potent CDy::UPRT_AT-MSCs. We hypothesized that high expression of suicide gene is necessary for generating high efficacy therapeutic AT-MSCs. We compared the potencies of AT-MSCs produced by transfection with L3K and LPEI with or without the use of Enhancer (Fig. 2, Fig. S1). As expected, the anticancer efficacies of the therapeutic cells prepared with the different protocols were highly dependent on transfection efficiencies of each protocol (Fig. 6). The anticancer efficacy of CDy::UPRT_AT-MSCs generated in the presence of Enhancer significantly surpassed effects observed with cells modified with L3K. At the ratio of 1 MSC to 10 cancer cells, complete inhibition of proliferation was observed in all cancer cell lines co-cultured with CDy::UPRT_AT-MSCs generated in the presence of Enhancer. It is worthy to note that the transfection protocol using the Enhancer generated modified MSCs with similar potencies regardless of cell sources (adipocyte, bone marrow or umbilical cord derived MSCs (Fig. S8). Furthermore, we have successfully transfected MSCs with another suicide gene, Herpes Simplex Virus-1 Thymidine Kinase (HSV-TK) (Fig. S9). These data suggest the transfection workflow described herein is agnostic to MSC types and transgene. Potential therapeutic application in targeting chemo-resistant Glioblastoma. Evidently, the cytotoxic effect of CDy::UPRT_AT-MSCs is agnostic against cancer types (Fig. 5, 6). Next, we hypothesized that this approach could potentially target chemo-resistant cancer too. A stable TMZ-resistant variant of U251-MG cell line (termed U251R) was established as described previously 49 . Proliferation inhibition of U251R in the presence of CDy::UPRT_AT-MSCs and 5FC (Fig. S11) suggested the potential use of this approach for targeting TMZ resistant Glioblastoma.
To provide further evidence of a therapeutic utility, CDy::UPRT::GFP_AT-MSCs were injected directly into the subcutaneous (s.c) tumour of U251R. Previous studies have shown contradictory outcomes when native MSCs were used for experimental cancer treatment 50 . To evaluate the anti-tumour effect was due to the expression of CDy::UPRT::GFP, tumour growths were monitored in 3 groups of mice. One group of animals received native MSCs and treated with 5 FC (MSC plus 5FC), another with the prodrug treatment alone (5FC), and the third group where the mice were given modified MSCs and treated with 5FC (CDy::UPRT::GFP/5FC). The tumour growth in the MSC plus 5FC group of animals did not show significant difference from the animals treated with 5FC alone. Expectedly, significant inhibition of tumour growth was observed in the CDy::UPRT::GFP/5FC group of animals (Fig. 8), where a single cycle of treatment resulted in an average of 45% reduction in tumour size (Fig. 8A). However, the tumour size in the CDy::UPRT::GFP/5FC group appeared to increase in size but www.nature.com/scientificreports/ yet significantly smaller than the 5FC or MSC plus 5FC groups of animals on day 7 and day 10 post treatment (Fig. 8B). This is consistent with the hypothesis that anti-tumour effect requires the MSCs to express the therapeutic suicide gene.

Discussion
Conflicting reports have recently emerged regarding the roles of MSCs in tumour inhibition 51,52 and growth [53][54][55][56] . These contradictions were thought to be largely due to technical differences and inherent biological heterogeneity 57 . Regardless, genetically modified MSCs is thought to offer a more suitable strategy for cancer therapy, as they are safer and more efficient than the unstable and heterogeneous naive MSCs 58,59 . This study demonstrated the successful modification of AT-MSCs at high efficiency for the generation of theranostic AT-MSCs for prodrug cancer therapy, without the use of viruses. About half of the cell population was transfected with the commercially available polymer (PEI MAX) and the efficiency was significantly improved with low toxicity in the presence of the Enhancer (Fig. 1). This modification process did not require purification nor antibiotic selection for high expression MSCs of > 70% CD expressing cells, in line with release testing for human clinical trial 60 . Attempts to develop novel cationic polymers and lipids to modify MSCs have met with limited success due to low efficiency of transfection or high cytotoxicity 61 . Recently, a poly(β-amino-esters) (PBAE) polymer structure was reported to transfect MSCs with high efficiency and low toxicity 62 . Although the cells were well modified, the migration ability was notably affected.
For AT-MSCs to be used as targeted drug delivery vehicles for therapy, the processes used to modify them must not change their phenotypic characteristics and behaviour, including their multipotency and their capacity for migration and invasion. No significant difference in the expression of phenotypic markers and differentiation potential of modified and native AT-MSCs was observed (Fig. 3), an essential criterion for theranostic application of the modified AT-MSCs 43 . The inherent tumour tropism is an essential feature of the homing/migration property of MSCs as a cellular vehicle for delivery of therapeutic agents 8,63 . Despite the high over expression of the transgene, the migration ability of the modified cells was comparable to the native MSCs in the presence of cancer cells in vitro (Fig. 4).
A number of GDEPT systems is being explored for cancer treatment to improve the efficacy and safety of conventional cancer chemotherapies 19 . Among the enzyme/prodrug systems tested in a recent study, the CDy::UPRT is the most effective 64 and this has been used in stem cells based clinical trials 60,65 . In our study, CDy::UPRT modified cells inhibited the growth of MDA-MB-231, U-251MG, MKN45 and MKN1 cell lines efficiently, with as little as 10% of therapeutic AT-MSCs. Notable was that MDA-MB-231 proliferation was inhibited by ~ 90% at the ratio of 1 therapeutic MSC to 10 cancer cells (Fig. 5a). At the similar ratio, Kucerova et al. demonstrated only 40% proliferation inhibition of the same cell type when using AT-MSCs modified by retroviral transduction 66 . Yet another study reported ~ 60% reduction in cell number in the co-culture of MDA-MB-231 with virally transduced CDy::UPRT_MSCs at a ratio of 1 MSC to 4 cancer cells 67 . Furthermore, Kwon et al. 68 and Nouri et al. 64 using www.nature.com/scientificreports/ selected modified MSC, reported the inhibition of tumour growth but not regression. It is likely that the modification using the process developed herein resulted in increased payload resulting in more efficacious killing of cancer cells which is consistent with the high anticancer effects on chemoresistant glioma cells in vivo (Fig. 8).
Although significantly less than the control groups (5FC alone or native MSC with 5FC), the apparent increase in tumour size over a period of time with a single treatment may be further improved with multiple treatment cycles and this is the subject of an ongoing investigation. MSCs mediated CD/5FC treatment has been suggested as a strategy to overcome the systemic toxicity of 5FU 14,68 . In the in vivo study, we did not observe significant change in the weight of subjects or other direct side-effects (data not shown), consistent with other studies 14,68 . Because of the alleviation of systemic toxicity, repeated injection of CD-MSC is possible to enhance the antitumour activity. Additionally, it is noteworthy that the therapeutic MSCs itself is sensitive to the CDy::UPRT/5FC system (Fig. S4), thus limiting the survival of the therapeutic cells; fulfilling a crucial requirement of a 'hit and run' strategy, leaving no trace of the cellular vehicle 8 .
Depending on the route of administration and location of the tumour, it is anticipated that 1 to 4 days are required for MSCs to biodistribute to residual tumour and home to distant foci of tumour 69 . In the recent clinical trial on advanced gastrointestinal cancer, patients were given three treatment cycles with modified MSCs followed by the prodrug administered 48-72 h later 65 . In parallel, 4 days after the administration of modified neural stem cells, prodrug (5FC) was administered and the modified cells were functional during the entire 7-day course of 5FC 60 . Hence, it is conceivable that the transiently transfected AT-MSCs with prolonged expression of CDy::UPRT (Fig. 7) should be effective over the duration of a treatment regime.
In order to reduce toxicity due to prolonged exposure to the polyplex, a low speed spinning step was used 36,70 , which may limit the production of theranostic AT-MSC in large scale. Process design and optimization is underway for scalability in production by precluding the low speed centrifugation step.
In conclusion, we herein described an in vitro, non-viral process for engineering theranostic AT-MSCs for GDEPT with high efficiency and high cell viability using an off-the shelf cationic polymer. We showed that, despite the high over expression of the transgene, the phenotypic characteristics and migration ability of the modified cells were comparable to the native MSCs. These cells were highly efficient in inhibiting proliferation of www.nature.com/scientificreports/ cancer cells in vitro. Hence, this process to modify AT-MSCs constitutes an effective and safe alternative to viral transduction for stem cell-based cancer therapy and may be useful for applications beyond the scope of this study.
In vitro drug susceptibility. Quadruplicates of AT-MSC, MKN1, MKN45, MDA-MB-231 (10,000 cells per well) and U-251MG (5,000 cells per well) for each treatment were plated into 96-well plates. Twenty-four hours later, culture medium was replaced for medium containing various concentration of 5-Fluorocytosine (5FC, InvivoGen) or 5-Fluorouracil (5FU, InvivoGen). One to five days later, plates were subjected to the CellTiter 96 Aqueous One Solution Cell Proliferation Assay (Promega). The colorimetric read out was measured spectrophotometrically at 490 nm. Results were expressed as the percentage of cell viability, in relative to cells in condition without 5FC or 5FU (set to 100%).
Anticancer efficacy of CDy::UPRT producing AT-MSCs in vitro. Direct co-culture. Quadruplicates of gastric cancer cell lines and breast cancer cell line (5,000 cells) and U-251MG (2,000 cells) were plated in 96well plates. Five hours later, various numbers of either un-transfected or CDy::UPRT-producing AT-MSCs at the ratios of 1 AT-MSC to 1, 5, 10, 50 and 100 cancer cells were added to the cancer cells. One day later, the culture media was replaced with DMEM supplemented with 2% FBS, with or without 5FC (0-150 µg/mL). Five days later, cell viability was measured by proliferation assay. Conditions without 5FC was set to 100%.
Indirect coculture. MDA-MB-231cells were plated on 24-well plate (5 × 10 4 cells per well). AT-MSCs or CDy::UPRT_AT-MSCs (5 × 10 4 cells per well) were plated on transwell (Corning, C05/3,422). After 6 h of cultivation, inserts with therapeutic cells were transferred into the wells with MDA-MB-231cell line, with or without 5FC. Cytotoxic effect was evaluated after 4 days of incubation. Transwells were removed and culture media was replaced with 1XPBS containing 1 µg/mL of Hoechst 3,222. Stained cells were analysed using Synergy H1 microplate reader at excitation and emission wavelength of 358 nm and 461 nm, respectively. With gain setting at 80, Relative Fluorescence Unit (RFU) at 9 areas of the cell culture were recorded. Proliferation inhibition after treatment will be calculated relative to the control (coculture of untransfected AT-MSC and MDA-MB-231 cells).
Anticancer efficacy of CDy::UPRT producing AT-MSCs in vivo. Five to six-week old female nude mice were purchased from InVivos and used for the in vivo studies under National University of Singapore Institutional Animal Care and Use Committee (IACUC) approved protocol (R18-1,383). All methods were carried out in accordance with relevant guidelines and regulations. Mice were anesthetized by isoflurane inhalation and 5 × 10 6 Temozolomide resistant U-251MG cells suspended in 100 μL DMEM (50% Matrigel) were injected s.c. in dorsal flank regions. The growth of tumour was monitored by digital caliper. When tumours measured an average volume of 80-200 mm 3 , treatment was started. All mice were randomly distributed into 3 groups each containing 5 mice. Prodrug control group received daily injections of prodrug. Cell control group received intratumoural injection of 1 × 10 6 MSCs plus daily injections of prodrug. Treatment group received intratumoural injection of 1 × 10 6 CDy::UPRT_AT-MSCs plus daily injections of prodrug. Modified or non-modified MSC were administrated intratumourally on day 0 (single dose). One day later, mice received i.p. administration of 500 mg/kg of 5FC for 4 consecutive days. Before cell injection (Day 0) and Day 7, 11 and 15 after MSC administration, tumour sizes and body weights were measured.
Cell invasion assay. The tumour tropism of AT-MSCs was determined using BD Biocoat matrigel invasion chambers (BD Biosciences). Cancer cell lines or HEK293T cells were loaded in the lower well of the 24-well plates. Twenty four hours later, un-transfected and CDy::UPRT-producing AT-MSCs in serum-free DMEM were added onto the invasion chambers. Lower wells were washed with 1XPBS, filled with serum free DMEM, for the invasion assay. After 24 h incubation, non-invading cells and matrigel were removed from the top chamber of the insert. Invaded cells were stained with Hoechst 33,342 (ThermoFisher Scientific) and photographed through the imaging system. Number of cells in 3 frames were counted.

Statistical analysis.
Where Student's t-test, was used, an unpaired two-tailed test was used, with the assumption that changes in the readout are normally distributed. www.nature.com/scientificreports/