E1A-engineered human umbilical cord mesenchymal stem cells as carriers and amplifiers for adenovirus suppress hepatocarcinoma in mice

Gene therapy is an attractive approach for hepatocellular carcinoma (HCC) patients. Nevertheless, efficient transgene delivery remains a challenge. In this study, we explored a new targeted system based on human umbilical cord-derived mesenchymal stem cells (HUMSCs), which were engineered to deliver adenovirus to tumor sites, and to replicate and assemble into new adenovirus against HCC. Our results showed that HUMSCs infected by Ad-hTERTp-IL24 followed by LentiR.E1A infection could specifically migrate to HepG2 tumor cells and support adenoviral replication in vitro and in vivo 36 h after LentiR.E1A infection. Ad-hTERTp-IL24 specifically inhibited HepG2 cells growth, and this inhibitory effect was enhanced by low doses of 5-fluorouracil (5-Fu), because the expression levels of coxsackie adenovirus receptor (CAR) and integrin ανβ3 on tumor cells were significantly increased, causing higher viral uptake. Compared with the no treatment groups, Ad-hTERTp-IL24 and LentiR.E1A co-loaded HUMSCs exhibited significant anti-tumor activity in vivo, particularly in combination with low doses of 5-Fu. In summary, this study provides a promising targeted gene therapeutic strategy dependent on the tumor tropism of HUMSCs, to improve the outcome of virotherapy for tumor patients especially those with metastatic diseases.


INTRODUCTION
Gene therapy is an attractive and promising approach to cancer treatment. Currently, adenoviral vectors have been employed for gene therapy due to their low pathogenicity, high titer and lack of integration adenovirus serotypes 5 (Ad5), in which the E1 and E3 regions of the genome are deleted, are the most widely used [1]. Clinical studies have shown that administration intraperitoneally and intravesically is a safe, feasible and effective antitumor strategy against many types of cancers [2]. Nevertheless, the major concerns over the use of such tumor sites due to the fact that the metastatic tumors are often smaller and directly inaccessible. Moreover, systemic administration of high doses of adenovirus is associated with systemic toxicity and rapid elimination of the virus by the immune system before reaching the tumor [3]. Thus, it is absolutely critical to develop an Research Paper www.impactjournals.com/oncotarget administration of adenoviral vectors to improve the clinical outcome of patients with recurrent and metastatic lesions.
Mesenchymal stem cells (MSCs) have been shown to migrate toward malignant tumors and track microscopic metastasis when administered by intravenous injection in vivo [4,5]. Further, engineered MSCs have been indicated as a potential vehicle to deliver anticancer agents to primary and metastatic tumors [6][7][8]. At present, scientists have successfully taken advantage of MSCs to deliver antitumor agents, including cytokines [5], interferons [9], pro-drugs [10] and conditionally replicating virus [11]. Human umbilical cord's Wharton's jelly (WJ)-derived by McElreavey et al [12]. HUMSCs share common characteristics of MSCs, such as immunosuppression, markers (CD90, CD105 and CD73), multi-differentiation potential to the osteogenic, adipogenic and chondrogenic lineages [13], and ability to accumulate at sites of addition, HUMSCs are advantageous in term of rapid cell expansion, yield, ease of procedure, lack of ethical problems and are suitability for genetic engineering with viral vectors [14]. These characteristics make HUMSCs to be a promising platform for targeted delivery of anticancer agents for a variety of cancers.
To enhance the transfer of adenovirus to tumor cells, we engineered the HUMSCs to produce an adenovirus adenoviral vectors based on Ad-5 can be propagated in complementing human cell lines that provide the E1 proteins [15]. Therefore, if E1A proteins, which are essential for the replication of the adenovirus, were supplied in HUMSCs, the HUMSCs would permit proteins. The engineered HUMSCs not only delivered adenoviral vehicles to tumor or metastatic tumor sites but also supported the adenoviral replication. Ultimately, the released, allowing the infection of the surrounding tumor cells to express the target therapeutic proteins. target therapeutic protein carried by the adenoviral vectors. apoptosis in a variety of cancer cells without affecting the normal cells in vitro [17,18], in vivo [17,19,20], and in neighboring cancer cells can be killed by the bystander by the human telomerase reverse transcriptase (hTERT) promoter, which is highly active in over 85% of human cancer cells but inactive in most somatic cells. The hTERT promoter has shown great potential in regulating the celltherapeutic system in vitro and in vivo in hepatocarcinoma models. Our results showed that adenovirus-loaded HUMSC.lentiR.E1A could support adenoviral replication and viral particle release to infect the tumor cells. Moreover, virus-loaded HUMSCs were still capable of migrating to hepatocellular carcinoma. The tumor suppressive effect of this dual targeted therapeutic system was also observed in vitro and in vivo. Furthermore, we investigated the synergistic antitumor effect of this dual targeted therapeutic system in combination with

RESULTS
The HUMSCs, obtained from the WJ of human umbilical cord with informed consents, had a typical the morphology reported by others [12,27]. Although mesenchymal markers CD73, CD90 and CD105 ( Figure  1A) and negative for typical hematopoietic antigens CD34, CD45 and CD19 ( Figure 1B) as previously described [13]. Moreover, the HUMSCs were able to undergo adipogenesis ( Figure 1C) and osteogenesis ( Figure 1D showed evident transcriptional activity, with values of 1.73±0.71% and 2.99±1.07% of the positive control activity in cancer cells but not in normal cells. E1A and the adenoviral expression vectors pAd-hTERTp-HUMSCs using AdTrack at 48 h after infection. Flow respectively ( Figure 1G). Based on these results, for their growth ( Figure 1H).
To directly verify whether the replication-hexon gene, a late adenoviral gene, intracellularly and in the supernatant of virus-loaded HUMSCs at the co-infection. Hexon gene expression was measured by quantitative PCR to determine the copy number of set as 0 h. As the expression of E1A increased gradually ( The total concentration (intracellularly and in the supernatant) of viral DNA was measured at different time points in in three independent samples of HUMSC sequentially infected The intracellular and supernatant concentration of viral DNA were detected at different time points respectively. Electron micrographs show viral particles in the HUMSC sequentially www.impactjournals.com/oncotarget 55.5±3.5% and 72.4±3.9% at HUMSCs: HepG2 ratio of 1:20, 1:10, 1:4 and 1:1, respectively ( Figure 2E). These observations indicated that the released viral particles have the ability to infect cells.
shown to have a homing predisposition to tumor cells in vitro and to the tumor site in hepatocarcinoma models, and their migration capacity presents a concentrationdependent pattern [28]. However, in this study, the virus-loaded HUMSCs would be lysed after 36 h of and packaging was activated by the expression of E1A, as described in Figure 2B. To test the migration capacity of virus-loaded HUMSCs within 36 h after infection, in vitro migration assays using Transwell loaded HUMSCs migrated towards HepG2 cultures in after infection ( Figure 3A, 3B). These results indicated that human hepatocellular carcinoma HepG2 cells were capable of stimulating the migration of HUMSCs and that the migration ability of HUMSCs was not affected by adenoviral and lentiviral co-infection.
Next, to investigate the homing capability of virusloaded HUMSCs in vivo, we designed an adenoviral Representative photographs showed the migrated HUMSCs stained with crystal violet in vitro migration assays using Transwell plates. The numbers of migrated HUMSCs in three independent samples were expressed as mean ± SD. tumor sites was monitored by bioluminescence imaging using Xenogen imaging system at indicated times after tail vein injection. The signal was detected at the tumors site. 2 days after the injection, the signal further increased, and it lasted until injection group, the intense imaging signal could also be detected at the tumor site 1 day after the injection, and disappeared after 4 days. The signal was discovered once again 7 days after the injection, indicating that the surrounding tumor cells ( Figure 3D). The inhibitory rate curve of HepG2 cells treated with 5-Fu gradient concentrations was analyzed. the presence or absence of low doses of 5-Fu were tested by CCK8 assay and expressed as mean ± SD. *P<0.05, ** P<0.01 compared with P P Apoptosis ratios were detected when 5-Fu P<0.05, ** P P P www.impactjournals.com/oncotarget with AdTrack (Table 1). A similar synergistic effect was discovered in the apoptosis assays when 5-Fu was used in group, and promoted p38MAPK phosphorylation ( Figure  4F), correlating cell killing with the activation of the p38MAPK pathway. Furthermore, Bax protein expression levels increased, while Bcl-2 protein expression decreased, and other apoptosis-related proteins such as PARP, caspase-3 and 9, were cleaved and activated. All of these 3 was improved when combined with low doses of 5-Fu ( Figure 5A). Flow cytometry analysis also showed that group was higher than the adenoviral treatment alone ( Figure 5B). To test whether the synergistic effect was due to increased susceptibility to adenoviral infection in the presence of 5-Fu, HepG2 cells were infected with AdTrack  Figure 5C). Next, we investigated whether the expression levels of the viral attachment receptor CAR and the major internalization receptors 3 5 integrins were increased in response to 5-Fu HepG2 cells, the baseline CAR membrane levels reduced gradually with the increase of time in culture, and they were 17.60±4.38%, 13.95±4.45%, 6.53±3.31% at 24, 48 and 72 hours, respectively. While in the presence of 5-Fu, 3 , baseline expression was low (<10%), and it slightly increased after 5-Fu treatment ( Figure 5E). For integrin 5 , baseline expression was as high as 100%, and 5-Fu treatment did not affect its expression levels ( Figure  5F). All these observations might explain the excessive adenoviral uptake and the enhanced cytotoxicity when the adenovirus was combined with low doses of 5-Fu.
We further investigated the antitumor potential of

DISCUSSION
We successfully established a new targeted treatment system based on HUMSCs, which were engineered to enable its replication and assembling into new viruses to tumor sites in a mouse model of hepatocellular carcinoma. Our results showed that engineered HUMSCs have the ability to incorporate into tumors and release cancer-killing the transplanted hepatocarcinoma, which were mediated effect was greatly strengthened when combined with low doses of 5-Fu, due to the increased expression of CAR in an orthotopic mouse xenograft model of lung cancer. Compared with this study, our results not only showed the effectively anti-tumor activity of this strategy but also provided more evidence of adenoviral replication in vitro and in vivo (Figures 2 and 3).
The tumor-migrating tropism of MSCs has been acknowledged in recent years. Due to its applications to tumor treatment, increasingly frequent investigations have been performed on the kinetic distribution of systemically administered MSCs in vivo and on the time required for MSCs to the tumor site [11,33,34]. Our laboratory previously demonstrated by luciferase bioluminescence in vivo that MSCs migrate and selectively accumulated at the tumor site at 24 h after intravenous injection [34]. Recently, MSCs have been demonstrated to deliver conditionally replicative adenovirus (CRAd) to various malignant tumors. These viruses are able to destroy tumor cells by replication and consequent oncolysis, which enables the newly produced virus to be released to the surrounding tumor tissues to prevent tumor growth [6,11,33,35]]. However, a considerable number of CRAdloaded MSCs were also found in the lung, liver, and spleen, with the exception of the tumor site after systemic administration [33]. These ectopic CRAd-loaded MSCs would produce virus to injury normal tissues as well, study, we used E1A-engineered HUMSCs to deliver a assemble into new viruses to tumor sites. The replicationnot tumor cells or other normal cells because of the complementary expression of E1A in HUMSCs.
Combining adenoviral constructs with chemotherapeutics has represented an appealing strategy to increase their potency [36]. Several studies have presented combinatory cytotoxic effects in esophageal carcinoma by Ad-delE1B55 in combination with 5-Fu [37], in pancreatic adenocarcinoma model by Ad-dl922-947 in combination with 5-Fu or gemcitabine [29], and in patients with recurrent head and neck cancer by intratumoral ONYX-015 in combination with cisplatin or 5-Fu [30]. The reasonable explanation for combinatory cytotoxic effects on tumors was that chemotherapeutic agents increased adenoviral targeted therapeutic strategy using E1A-engineered adenovirus against hepatocarcinoma to tumor sites. The therapeutic strategy provides a new, effective and safe inaccessible and/or metastatic tumor sites. Meanwhile, it also solves the potential safety hazard of HUMSCs. However, this therapeutic system retains some problems that will need to be improved in the future. For example, the expression of E1A can be controlled by an inducible promoter, which would ensure more HUMSCs migrate to tumor sites before lysis, enhancing the tumor-suppressing effect. Additional studies are warranted to demonstrate the superiority of this therapeutic strategy in the tumor metastasis models, which would make it highly appealing for the treatment of tumor patients with metastatic diseases. HUMSCs were isolated from human umbilical cord Wharton's jelly (WJ) as previously described [27]. HUMSCs were seeded at a density of 8×10 3 cells/cm 2 in detached using a 0.125% trypsin/1 mM EDTA solution, and re-seeded using the same growth media for subsequent passages. Cells at passage number 3-5 were used for the following experiments. www.impactjournals.com/oncotarget HUMSCs were seeded in 6-well plates at a density of 1×10 5 cells/well and incubated overnight at 37°C. On the next day, the HUMSCs were infected with Ad-6 hours. Then, the culture medium was replaced with Santa Clara, CA). Twelve hours later, the medium was replaced. HUMSCs and supernatants were harvested after the indicated periods of lentiviral infection, and used for Nucleic Acid Extraction Kit (Roche, Basel, Switzerland).
technology was used to detect a 286-bp-long amplicon (nucleotides 21049-21334) within the conserved region of the Ad5 hexon gene. The primers were designed as manufacturer's instructions for SYBR Premix Ex Taq reagent (Takara, Dalian, China). The standard curve for of pAdTrack plasmid. The experiments were repeated for three times.
The migration of virus-loaded HUMSCs was diameter (BD Falcon, New York, USA). 12 hours after coinfection, 1×10 5 HUMSCs were plated in the top chamber cells were seeded at a density of 5×10 4 cells/well in the lower chamber in fresh medium containing 2% FBS. After 20 h incubation at 37°C, cells that had not migrated from the upper side of the membrane were scraped off with a cotton swab, and membranes were stained with 0.1% crystal violet at 37°C for 45 min. Cells that had migrated performed in triplicate.
For the in vivo HUMSC migration assays, we HepG2 cells inoculation into the mouse right armpits, when the solid tumors reached 100-200 mm 3 in size, 1×10 6 injected into mice via the tail vein to detect cell migration.
All animal procedures were approved by the Committee on the use and care of animals, Chinese Academy of Medical Science. 5-6-weeks-old female with 5×10 6 HepG2 cells into right armpits. 7 days after tumor inoculation, when solid tumors reached 100-200 mm 3 in size, mice were randomized into 9 groups (7 The HUMSCs were co-infected as described above and injected intravenously with 1×10 6 cells/mouse. 5-Fu was i.p. injected at a dose of 10 mg/kg for 5 continuous days starting 3 days after HUMSCs injection. Growing tumors were measured every three days using a vernier caliper in two perpendicular dimensions. The tumor volumes were anesthesia when the tumors reached 2000 mm 3 in size, and tumor tissues were harvested and weighed. a series of graded-ethanol and PBS solutions. Antigen retrieval was performed by heated in a hot bath. Sections were then treated with goat serum for 30 min at room temperature followed by incubation with rabbit antiand mouse anti-GFP monoclonal antibody (Abbkine, CA) at 4°C overnight. On the following day, sections were incubated with secondary polyclonal donkey antibody for 30 min. 4, 6-diamidino-2-phenylindole staining. The stained sections were imaged by confocal www.impactjournals.com/oncotarget Apoptotic cells in the tumors were detected by terminal deoxynucleotidyl transferase dUTP nucleotide Biotechnology, Shanghai, China) was used according positive cells on the sections were detected by confocal microscopy.
Data were analyzed using an independent sample t-tests and were represented as the mean ± SD. P<0.05 P<0.01 www.impactjournals.com/oncotarget