Abstract We established a semi-high-throughput in vivo screening platform using hyperactive piggyBac (hyPB) transposons (designated as PB-miR) to identify microRNAs (miRs) that inhibit hepatocellular carcinoma (HCC) development in vivo, following miR overexpression in hepatocytes. PB-miRs encoding six different miRs from the miR-17-92 cluster and nine miRs from outside this cluster were transfected into mouse livers that were chemically induced to develop HCC. In this slow-onset HCC model, miR-20a significantly inhibited HCC. Next, we developed a more aggressive HCC model by overexpression of oncogenic Harvey rat sarcoma viral oncogene homolog (HRAS^G12V) and c-MYC oncogenes that accelerated HCC development after only 6 weeks. The tumor suppressor effect of miR-20a could be demonstrated even in this rapid-onset HRAS^G12V/c-MYC HCC model, consistent with significantly prolonged survival and decreased HCC tumor burden. Comprehensive RNA expression profiling of 95 selected genes typically associated with HCC development revealed differentially expressed genes and functional pathways that were associated with miR-20a-mediated HCC suppression. To our knowledge, this is the first study establishing a direct causal relationship between miR-20a overexpression and liver cancer inhibition in vivo. Moreover, these results demonstrate that hepatocyte-specific hyPB transposons are an efficient platform to screen and identify miRs that affect overall survival and HCC tumor regression. Keywords: miR-20a, transposon, gene therapy, piggyBac, oncogene, hepatocellular carcinoma, HRAS, c-MYC, non-viral vector, miR-17-92 Introduction MicroRNAs (miRs) are small, single-stranded, non-protein-coding RNA molecules that are 22 nt in length, and they have been shown to control cell growth, differentiation, and apoptosis.[37]^1 Consequently, impaired miR expression has been implicated in tumorigenesis. Abnormal miR expression has been found in both solid and hematopoietic tumors and is associated with altered expression of “classical” oncogenes.[38]^2 Typically, miRs located in genomic regions that are amplified in cancer function as oncogenes, whereas miRs located in chromosomal regions that are deleted in cancer function as tumor suppressors.[39]3, [40]4, [41]5, [42]6 Notably, expression profiling of both miRs and protein-coding genes can be used to improve the accuracy of cancer subtype classification, diagnosis, and prognosis.[43]7, [44]8, [45]9 Interestingly, about one-half of the annotated human miRs map within fragile regions of chromosomes, which are areas of the genome that are associated with various human cancers.[46]^10 Changes in miR expression between normal and tumor cells may not necessarily alter the cancerous phenotype because the main interactions of a given miR with its various targets could have antagonistic rather than synergistic or additive biological consequences. In particular, some miRs are organized in clusters, and the role of each miR within a given cluster would need to be assessed for each miR individually. To validate a miR for diagnostic and therapeutic purposes, it is therefore essential to first establish a causal relationship between its differential expression and the cancerous phenotype in in vivo models. The present study aims at addressing some of these outstanding questions in miR biology and cancer using hepatocellular carcinoma (HCC) as a model, given its poor prognosis and high prevalence. HCC is one of the most common causes of cancer-related mortality worldwide.[47]^11^,[48]^12 The altered expression of several miRs (i.e., miR-18, miR-20a, miR-21, miR-34, miR-17-92, let-7a, let-7c, miR-92, miR-122, miR-195, miR-199a, miR-200a, miR-341, and miR-370) has been associated with HCC in mice and/or humans, but experimental evidence establishing a causal relationship between abnormal expression of these miRs and HCC is generally lacking.[49]13, [50]14, [51]15, [52]16, [53]17, [54]18, [55]19, [56]20, [57]21, [58]22, [59]23, [60]24, [61]25, [62]26 Nevertheless, using miR microarray analysis in paired patient-derived samples of tumoral tissue and non-tumoral adjacent tissues, several miRs (i.e., miR-92, miR-20, miR-18, and miR-18 precursor) were found to be inversely correlated with the degree of HCC differentiation.[63]^27 Of particular interest is miR-20a, a member of the miR-17-92 cluster, because it increases cell apoptosis and inhibits cell proliferation and migration in transformed HCC cell lines in vitro.[64]^14^,[65]^15 In addition, downregulation of miR-20a was observed in primary HCC of patients following liver transplantation.[66]^14 Hence, this suggests a possible role between miR-20a and HCC initiation and/or progression. However, formal proof that overexpression of miR-20a results in suppression of HCC in cancer models in vivo is lacking. Consequently, establishing a causal relationship between a given miR and malignancy following hepatic overexpression of the miR should allow us to strengthen the intrinsic diagnostic and therapeutic relevance of miRs in HCC and exclude unrelated associations between miR expression levels and the cancerous phenotype. In the current study, we designed a hepatocyte-directed gene expression platform for in vivo miR screening and validation based on hyperactive piggyBac (hyPB) transposons that stably express miR from a robust hepatocyte-specific promoter, thereby avoiding ectopic expression in non-liver tissues.[67]28, [68]29, [69]30 Co-delivery of PB transposons and constructs expressing PB transposases results in DNA transposition into the target genome through a cut-and-paste mechanism.[70]^31^,[71]^32 Consequently, PB transposon-based gene transfer offers the advantage of sustained gene expression following genomic integration.[72]30, [73]31, [74]32, [75]33, [76]34, [77]35 We also used this PB platform to develop a rapid-onset model for HCC based on the selective hepatocyte-specific overexpression of oncogenic Harvey rat sarcoma viral oncogene homolog (HRAS^G12V) and c-MYC oncogenes that overcomes some of the limitations of current HCC models such as slow onset of tumor development, non-specific expression or effects unrelated to HCC, and time-consuming generation of transgenic models.[78]^36 Using this hepatocyte-specific PB system to functionally screen miR-17-92 cluster members in vivo, we identified miR-20a as a tumor suppressor gene in two complementary HCC models, both the conventional slow-onset chemically induced HCC model and the rapid-onset HRAS^G12V/c-MYC HCC model. Results Hepatocyte-Specific hyPB Transposon System for Liver-Targeted Overexpression of Exogenous miRs In Vivo We first established an in vivo screening platform based on PB transposons that were designed to specifically express any given miR in hepatocytes. miRs associated with HCC were specifically selected, based primarily on previous studies in cell lines.[79]13, [80]14, [81]15 We focused primarily on the miRs from the miR-17-92 cluster, namely miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92a-1. In addition, other miRs that are differentially expressed in HCC (i.e., miR-370, miR-1188, miR-341, miR-221, miR-222, miR-106b-25, miR-93, miR-106b, and miR-25) were also tested since it is unclear whether they have a direct impact on tumor growth in liver cancer. The respective miR genes and approximately 50–200 bp of the surrounding genomic sequence were PCR amplified from mouse genomic DNA. Subsequently, because in general miRs are frequently located within introns, these miR-containing fragments were cloned into an intron present in a human FIX mini-gene (designated as hFIXIA). One of the main advantages of this platform is that the miR gene is then embedded in the FIX mini-gene, which encodes a secretable FIX protein that serves as reporter. This facilitates the continuous monitoring of in vivo expression kinetics and determination of transfection efficiency because FIX was proven to be non-immunogenic after liver-directed gene therapy.[82]28, [83]29, [84]30 To ensure hepatocyte-specific miR expression, a robust de novo designed hepatocyte-specific promoter was used that was composed of the transthyretin (TTR[min]) minimal promoter linked to a hepatocyte-specific cis-regulatory module (designated as HS-CRM8), which was identified by genome-wide data mining.[85]28, [86]29, [87]30 The expression cassette containing the HS-CRM8-TTR[min] hepatocyte-specific promoter driving the expression of the hFIXIA minigene harboring the miR gene was inserted into the PB transposon, which was flanked by its inverted repeats (IRs) ([88]Figure 1). All of the distinct miRs (i.e., miR-17, miR-19a, miR-18a, miR-20a, miR-19b-1, or miR-92a-1) are encoded by the same factor IX (FIX) transgene and embedded in exactly the same configuration within the first intron (intron IA) of this FIX transgene. Since the design of all the PB-hFIXIA-miR transposons is identical for each miR, FIX serves as an appropriate surrogate reporter gene whose expression can be readily monitored over time in the plasma of the recipient animals without sacrificing the mice.[89]37, [90]38, [91]39, [92]40 The miR gene is ideally positioned within intron A of the hFIXIA minigene to directly link its expression driven from the HS-CRM8-TTR[min] hepatocyte-specific promoter to that of the FIX protein as reporter. This intron-based miR design was successfully implemented and validated in other studies, albeit with distinct reporter genes.[93]37, [94]38, [95]39, [96]40 Figure 1. [97]Figure 1 [98]Open in a new tab Schematic Representation of Hepatocyte-Specific PB Transposons (A) Organization of miR-17-92 cluster members within PB-hFIXIA-miR-17-92. (B) PB-hFIXIA contains the human factor IX (hFIX) gene harboring 1.4 kb of truncated intron A, designated hFIXIA, and PB-hFIXIA-miR was derived from PB-hFIXIA by inserting the corresponding miR gene (i.e., miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92 a-1) into intron A. Both transposons carry a wild-type inverted repeat (IR[wt]). (C) PB-HRAS^G12V, PB-c-MYC, PB-MDM2, and PB-MDM4 encode the constitutively active form of Harvey rat sarcoma virus oncogene (HRAS) with a G12V amino acid mutation, the myelocytomatosis oncogene (c-MYC), the transformed mouse 3T3 cell double minute 2 (MDM2) oncogenic protein, and the transformed mouse 3T3 cell double minute 4 (MDM4) oncogenic protein, respectively. All PB transposons expressing oncogenic proteins contain truncated inverted repeats (IR[micros]). A liver-specific chimeric promoter composed of the minimal transthyretin (TTR[min]) promoter and hepatocyte-specific cis-regulatory module (HS-CRM8) was used to drive gene expression in all PB constructs. The bovine growth hormone polyadenylation (bGH-polyA) functioned as a terminating signal. We first performed a proof-of-concept study to validate the hepatocyte-specific hyPB transposon system for miR overexpression in normal C57BL/6JRj mice ([99]Figure 2). Normal C57BL/6JRj male mice were subjected to hydrodynamic co-transfection of the PB-hFIXIA-miR transposon individually expressing each member of the miR-17-92 cluster (i.e., miR-17, miR-19a, miR-18a, miR-20a, miR-19b-1, or miR-92a-1) in conjunction with a plasmid encoding a hyperactive PB transposase (hyPBase). Control C57BL/6JRj mice were co-transfected with PB-hFIXIA devoid of any miR and the hyPBase-encoding plasmid. The hFIX reporter gene was stably expressed at high levels that were not significantly different among each of the different miR-containing transposon constructs ([100]Figure 2A). This indicates consistent stable transfection efficiencies and transposition across all cohorts. Stable FIX expression was depending on the expression of hyPBase since expression declined to about ∼90% of the initial levels at 10 days post-transfection in the absence of any hyPBase ([101]Figure 2A). This indicates that prolonged expression of the FIX/miR constructs was mainly determined by hyPBase-mediated stable transposition. Figure 2. [102]Figure 2 [103]Open in a new tab Validation of Hepatocyte-Specific hyPB Transposon Platform for Overexpression of the miR-17-92 Cluster Members in Normal Mice Six-week-old C57BL/6JRj male mice were hydrodynamically transfected with 10 μg of either the PB-hFIXIA-miR or PB-hFIXIA construct and 2 μg of PB transposase-encoding plasmid. Mice transfected with PB-hFIXIA without PB transposase-encoding plasmid and PBS-injected mice were used as a controls (PB-hFIXIA, n = 3 mice; PB-hFIXIA+hyPBase, n = 3 mice; PB-hFIXIA-miR-17+hyPBase, n = 3 mice; PB-hFIXIA-miR-18a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19b-1+hyPBase, n = 3 mice; PB-hFIXIA-miR-20a+hyPBase, n = 3 mice; PB-hFIXIA-miR-92a-1+hyPBase, n = 3 mice). (A) hFIX protein expression analysis on days 1, 7, and 10 post-transfection measured by enzyme-linked immunosorbent assay (ELISA) to monitor the transfection efficiency of the PB transposons. Graph represents the mean ± SEM of data obtained from individual mice; unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01, ***p < 0.001). PB-hFIXIA-miR/hyPBase-treated mice were compared to the PB-hFIXIA control group devoid of hyPBase (PB-hFIXIA, n = 3 mice; PB-hFIXIA+hyPBase, n = 3 mice; PB-hFIXIA-miR-17+hyPBase, n = 3 mice; PB-hFIXIA-miR-18a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19b-1+hyPBase, n = 3 mice; PB-hFIXIA-miR-20a+hyPBase, n = 3 mice; PB-hFIXIA-miR-92a-1+hyPBase, n = 3 mice). (B) Quantitative miR expression levels in the mouse liver on day 10 post-transfection measured by quantitative PCR (qPCR) to assess the elevated selected miR expression in PB-hFIXIA-miRs/hyPBase-treated mice compared to control devoid of miR (i.e., PB-hFIXIA/hyPBase). Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 3 mice; PB-hFIXIA+hyPBase, n = 3 mice; PB-hFIXIA-miR-17+hyPBase, n = 3 mice; PB-hFIXIA-miR-18a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19b-1+hyPBase, n = 3 mice; PB-hFIXIA-miR-20a+hyPBase, n = 3 mice; PB-hFIXIA-miR-92a-1+hyPBase, n = 3 mice). (C) Relative hFIX mRNA expression levels in mouse livers on day 10 post-transfection measured by qPCR to assess the relative hFIX mRNA expression in the PB-hFIXIA-miR/hyPBase-treated mice compared to the PB-hFIXIA/hyPBase control group. Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01, ***p < 0.001) (PB-hFIXIA, n = 3 mice; PB-hFIXIA+hyPBase, n = 3 mice; PB-hFIXIA-miR-17+hyPBase, n = 3 mice; PB-hFIXIA-miR-18a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19b-1+hyPBase, n = 3 mice; PB-hFIXIA-miR-20a+hyPBase, n = 3 mice; PB-hFIXIA-miR-92a-1+hyPBase, n = 3 mice). (D) Quantitative PB transposon copy number per diploid genome in the mouse liver on day 10 post-transfection measured by qPCR. All PB-hFIXIA-miR/hyPBase-treated groups were statistically compared to PB-hFIXIA/hyPBase control group. Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01, ***p < 0.001) (PB-hFIXIA, n = 3 mice; PB-hFIXIA+hyPBase, n = 3 mice; PB-hFIXIA-miR-17+hyPBase, n = 3 mice; PB-hFIXIA-miR-18a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19a+hyPBase, n = 3 mice; PB-hFIXIA-miR-19b-1+hyPBase, n = 3 mice; PB-hFIXIA-miR-20a+hyPBase, n = 3 mice; PB-hFIXIA-miR-92a-1+hyPBase, n = 3 mice). Mice were subsequently euthanized to assess miR expression levels after PB-mediated transposition in the mouse livers. Typically, a significant increase in miR expression levels could be attained following transfection of mouse livers with PB-hFIXIA-miRs and hyPB compared to the controls devoid of miR (i.e., PB-hFIXIA and hyPB) ([104]Figure 2B). An average 15-fold increase in miR expression was observed after in vivo transfection with the PB-hFIXIA-miR transposons in normal C57BL/6jRj mice. Comparable levels of hFIX reporter gene expression, relative expression of hFIX mRNA, and PB transposon copy number confirmed that the transfection efficiency was similar among PB-hFIXIA-miRs/hyPBase versus PB-hFIXIA/hyPBase cohorts ([105]Figures 2A, 2C, and 2D. Collectively, these data validate the hepatocyte-specific hyPB system as a versatile platform to efficiently overexpress selected miRs in hepatocytes, making it a suitable system to assess the impact of miR overexpression on HCC initiation and/or progression in mice. Hepatocyte-Specific hyPB Transposon System for the Functional Screening of miR in a Slow-Onset HCC Model Next, the impact of miR overexpression on hepatocarcinogenesis was first assessed following PB transposon-mediated hepatic gene delivery in a slow-onset, chemically induced mouse model of HCC. HCC formation was pre-induced by diethylnitrosamine (DEN) in C57BL/6JRj male mice, followed by hydrodynamic transfection of the PB-hFIXIA-miRs or PB-hFIXIA control using the hyPB system, as previously described.[106]^30 The mice were then euthanized at 36 weeks post-DEN injection to assess tumor burden. Notably, the results showed that miR-20a led to a significant reduction in tumor burden based on the reduction in HCC tumor size and number of HCC tumor nodules between mice transfected with the PB-hFIXIA-miR-20a transposon and the control PB-hFIXIA transposon, which lacked any miR (p < 0.05) ([107]Figures 3A–3C). Consistently, miR-20a expression was selectively and significantly increased up to 28-fold and sustained in the livers of recipient mice that were transfected with the PB-hFIXIA-miR-20a transposon and the hyPB transposase ([108]Figure 3D). Significant miR overexpression was also observed in the mice transfected with PB-hFIXIA-miR-17 (∼7-fold), PB-hFIXIA-miR-18a (∼22-fold), PB-hFIXIA-miR-19a (∼27- fold), PB-hFIXIA-miR-19b-1 (∼14-fold), or PB-hFIXIA-miR-92a-1 (∼19-fold), respectively ([109]Figure 3D). An average 20-fold increase in miR expression was observed after in vivo transfection with the PB-hFIXIA-miR transposons in the slow-onset DEN-treated C57BL/6jRj mouse model. However, it did not contribute to a significant difference of HCC tumor sizes and nodule numbers compared to the control PB-hFIXIA/hyPBase mice (p > 0.05) ([110]Figures 3B and 3C). Similar results were obtained based on miRs that did not belong to the miR-17-92 cluster but that were differentially expressed in HCC, such as miR-370, miR-1188, miR-341, miR-221, miR-222, miR-106b-25, miR-93, and miR-25 ([111]Figure S1). Although a slight reduction in tumor size was apparent in the case of miR-106b, the number of tumor nodules was not statistically significantly different from that of controls ([112]Figure S1). The PB-hFIXIA-miR transposons were efficiently transfected into the mouse liver of the DEN-treated mice and stably expressed the hFIX protein ([113]Figure 3E) that is encoded by the miR-containing FIX transcripts ([114]Figure 1). There are no significant differences in expression among the different miR-containing constructs, in accordance with the comparable FIX protein levels obtained with each PB-hFIXIA-miR transposon encoding miR-17, miR-19a, miR-18a, miR-20a, miR-19b-1, or miR-92a-1 ([115]Figure 3E). This is consistent with the comparable PB transposon copy numbers per diploid genome among the different cohorts ([116]Figure 3F) and with the results obtained in normal C57BL/6JRj mice ([117]Figure 2). Figure 3. [118]Figure 3 [119]Open in a new tab Evaluation of the miR-17-92 Cluster on Hepatocarcinogenesis in a DEN-Induced Mouse Model C57BL/6JRj male mice were subjected to a pre-weaning protocol of diethylnitrosamine (DEN) initiation. DEN (12.5 mg/kg) was injected intraperitoneally in 14-day-old mice. Then, at 4 weeks post-injection, mice were hydrodynamically transfected with 10 μg of either the PB-hFIXIA-miR or PB-hFIXIA construct and 2 μg of PB transposase-encoding plasmid (PB-hFIXIA, n = 17 mice; PB-hFIXIA-miR-17, n = 8 mice; PB-hFIXIA-miR-19a, n = 9 mice; PB-hFIXIA-miR-18a, n = 9 mice; PB-hFIXIA-miR-20a, n = 8 mice; PB-hFIXIA-miR-19b-1, n = 6 mice; PB-hFIXIA-miR-92a-1, n = 8 mice). After 36 weeks of DEN initiation, mouse livers were assessed to compare the tumor burden among the different treatments. (A) Macroscopic tumor burden. Scale bars represent 5 mm. (B) Number of tumor nodules. Graph presents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 17 mice; PB-hFIXIA-miR-17, n = 8 mice; PB-hFIXIA-miR-19a, n = 9 mice; PB-hFIXIA-miR-18a, n = 9 mice; PB-hFIXIA-miR-20a, n = 8 mice; PB-hFIXIA-miR-19b-1, n = 6 mice; PB-hFIXIA-miR-92a-1, n = 8 mice). (C) Maximum size of tumor nodules. Graph presents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 17 mice; PB-hFIXIA-miR-17, n = 8 mice; PB-hFIXIA-miR-19a, n = 9 mice; PB-hFIXIA-miR-18a, n = 9 mice; PB-hFIXIA-miR-20a, n = 8 mice; PB-hFIXIA-miR-19b-1, n = 6 mice; PB-hFIXIA-miR-92a-1, n = 8 mice). (D) Quantitative miR expression levels in the mouse liver at 36 weeks post-treatment measured by qPCR to assess the elevated selected miR expression in hFIXIA-miRs-treated mice. Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 5 mice; PB-hFIXIA-miR-17, n = 3 mice; PB-hFIXIA-miR-19a, n = 4 mice; PB-hFIXIA-miR-18a, n = 5 mice; PB-hFIXIA-miR-20a, n = 4 mice; PB-hFIXIA-miR-19b-1, n = 4 mice; PB-hFIXIA-miR-92a-1, n = 3 mice). (E) Quantitative hFIX protein expression levels at 36 weeks post-treatment measured by ELISA to monitor the transfection efficiency of PB transposons. Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 17 mice; PB-hFIXIA-miR-17, n = 8 mice; PB-hFIXIA-miR-19a, n = 9 mice; PB-hFIXIA-miR-18a, n = 9 mice; PB-hFIXIA-miR-20a, n = 8 mice; PB-hFIXIA-miR-19b-1, n = 6 mice; PB-hFIXIA-miR-92a-1, n = 8 mice). (F) Quantitative PB transposon copy number per diploid genome in the mouse liver at 36 weeks post-treatment measured by qPCR to confirm the presence of the PB transposon in mouse hepatocytes after transfection. Graph represents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 5 mice; PB-hFIXIA-miR-17, n = 4 mice; PB-hFIXIA-miR-19a, n = 4 mice; PB-hFIXIA-miR-18a, n = 5 mice; PB-hFIXIA-miR-20a, n = 4 mice; PB-hFIXIA-miR-19b-1, n = 4 mice; PB-hFIXIA-miR-92a-1, n = 3 mice). Collectively, these results suggest that miR-20a is a tumor suppressor gene that inhibits HCC development and/or progression in the slow-onset DEN-induced murine model. This justifies the focus of the subsequent studies on miR-20a since it is only miR-20a that has tumor suppressor effects in the slow-onset DEN-induced HCC model. Establishment and Characterization of the Rapid-Onset HCC Model One of the limitations of the DEN chemically induced HCC model is the slow onset of HCC development. This reflects, at least in part, the random nature of the DEN-induced mutations. Although transgenic cancer models with a more rapid tumor progression have been developed, the effects are not always cell type-specific since ubiquitous promoters are often employed over cell type-specific ones. Moreover, such transgenic models are based on germline modifications and often lead to developmental perturbations. Typically, germline modifications in transgenic models do not adequately model the effect of somatic oncogene activation in adults. To overcome the limitations of existing HCC models, we first developed an HCC model using the PB transposon platform to simultaneously overexpress somatically multiple oncogenes in hepatocytes of adult mice. Previous studies have shown that dysregulation of the ras viral oncogene homolog (Ras) and c-Myc oncogene-dependent pathways promotes malignant transformation and tumor progression in the liver.[120]^41^,[121]^42 In addition, transformed mouse 3T3 cell double minute 2 (MDM2) and MDM4 act as oncogenes and induce HCC by targeting the tumor suppressor p53.[122]^43^,[123]^44 Hence, HRAS^G12V, c-MYC, MDM2, and MDM4 are attractive candidate oncogenes to establish a rapid-onset HCC model by somatic gene transfer in adult hepatocytes. PB transposons were therefore designed that expressed HRAS^G12V, c-MYC, MDM2, or MDM4 under the control of the hepatocyte-specific TTR[min]/HS-CRM8 promoter ([124]Figure 1).[125]^29 Subsequently, PB-HRAS^G12V was co-transfected in combination with PB-c-MYC, PB-MDM2, or PB-MDM4 transposons using the hyPB platform. HCC tumor nodules were present in mouse livers stably transfected with the PB-HRAS^G12V/PB-c-MYC combination, whereas the other combinations (i.e., PB-HRAS^G12V/PB-MDM2 or PB-HRAS^G12V/PB-MDM4) and single oncogene expression (i.e., PB-HRAS^G12V or PB-c-MYC) did not induce any liver tumor development ([126]Figures 4A and 4B). This is consistent with a possible cooperativity between the HRAS^G12V and c-MYC oncogenes allowing rapid-onset HCC development. We observed the pathological abnormalities only in liver tissues of PB-HRAS^G12V/PB-c-MYC-transfected mice, consistent with the presence of hepatic tumor lesion and large vacuoles in the tumor area ([127]Figure 4C, left panel, black solid arrow). Consistently, a marked accumulation of lipid content was present in the vacuolated region within the liver tumor of PB-HRAS^G12V/PB-c-MYC-transfected mice ([128]Figure 4C, right panel, black dotted arrow; also see [129]Figure 3E). PB-HRAS^G12V/PB-c-MYC-transfected mice showed an increase of hepatic fibrosis in liver tissues compared to other groups ([130]Figure 4C, middle panel; also see [131]Figure 3F). In particular, the histological features of hepatocytes with ballooning degeneration ([132]Figure 4D, upper panel), Mallory-Denk bodies ([133]Figure 4D, upper panel), and infiltrating inflammatory cells ([134]Figure 4D, middle panel) were uniquely identified in PB-HRAS^G12V/PB-c-MYC-transfected group. Collectively, this indicates that HRAS^G12V/c-MYC-induced liver tumors displayed the typical characteristics of steatohepatitic HCC (SH-HCC).[135]^45^,[136]^46 Figure 4. [137]Figure 4 [138]Open in a new tab Establishment and Characterization of Rapid-Onset HCC Model Six-week-old C57BL/6JRj male mice were hydrodynamically co-transfected with 1 μg of a single transposon encoding an oncogene (PB-c-MYC, PB-MDM2, or PB-MDM4) and PB-HRAS^G12V and 1 μg of hyperactive PB-encoding plasmid. As a control to study the effect of HRAS^G12V and c-MYC cooperation, mice were hydrodynamically co-transfected with 1 μg of single transposon plasmid expressing either HRAS^G12V or c-MYC (PB-HRAS^G12V or PB-c-MYC, respectively) and 1 μg of hyperactive PB-encoding plasmid. PBS-injected and non-injected mice were used as a negative control in this study. After 8 weeks of treatment, mouse livers were observed to compare the tumor burden among the different treatments. (A) Macroscopic tumor burden and number of tumor nodules from mice transfected with transposons expressing HRAS^G12V and c-MYC at 8 weeks post-injection. Graph presents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (n.s., not significant; *p < 0.05, **p < 0.01). Scale bars represent 5 mm (PB-HRAS^G12V+PB-c-MYC, n = 4 mice; PB-HRAS^G12V+PB-MDM2, n = 4 mice; PB-HRAS^G12V+PB-MDM4, n = 4 mice). (B) Macroscopic tumor burden from mice transfected with PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, PB-c-MYC, PBS, and without injection at 8 weeks post-injection. Scale bars represent 5 mm. (C) Microscopic representatives of mouse liver tissue sections stained with hematoxylin and eosin (left panel, original magnification, ×200), Sirius red (middle panel, original magnification, ×200), and oil red O (right panel, original magnification, ×200) to examine the presence of histopathology features of HCC, formation of liver fibrosis, and lipid droplet accumulation in mice transfected with PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, PB-c-MYC, PBS, and without injection. Scale bars represent 20 μm. Solid arrows indicated large empty vacuolated regions. Dotted arrow indicates lipid droplets (stained in red). Tumor (T) and non-tumor (NT) areas in liver tissue are divided using dotted lines. (D) Microscopic representatives of tumor (T, upper and middle panels) and non-tumor (NT, lower panel) areas of mouse liver tissue sections from PB-HRAS^G12V+PB-c-MYC-injected group stained with hematoxylin and eosin (original magnification, ×400). Black arrowhead indicates the presence of Mallory-Denk body. “B” indicates ballooned hepatocyte. Black arrow indicates an infiltration of inflammatory cells in liver tissue. (E) Percentages of sirius red-positive area in liver tissues from mice transfected with PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, PB-c-MYC, PBS, and without injection at 8 weeks post-injection. Graph presents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (***p < 0.001) (PB-HRAS^G12V+PB-c-MYC, n = 4 mice; PB-HRAS^G12V, n = 4 mice; PB-c-MYC, n = 4 mice; PBS-injected, n = 4 mice; non-injected, n = 4 mice). (F) Percentages of oil red O-positive area in liver tissues from mice transfected with PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, PB-c-MYC, PBS, and without injection at 8 weeks post-injection. Graph presents the mean ± SEM of data obtained from individual mice; two-tailed unpaired Student’s t test (***p < 0.001) (PB-HRAS^G12V+PB-c-MYC, n = 4 mice; PB-HRAS^G12V, n = 4; PB-c-MYC, n = 4 mice; PBS-injected, n = 4 mice; non-injected, n = 4 mice). Molecular Analysis of RNA Expression Profiles and Pathway Analysis during HRAS^G12V/c-MYC-Mediated Hepatocarcinogenesis To better characterize the rapid-onset HCC model and gain better insights into the molecular interplays between HRAS^G12V and c-MYC to induce HCC development in mice, we conducted a more comprehensive RNA expression profile study of 95 selected genes that are known to associate with liver cancer using qRT-PCR ([139]Table S4). Subsequently, the gene expression levels from the PB-HRAS^G12V/PB-c-MYC-transfected cohort were compared with those from the non-injected control to identify differentially expressed genes (DEGs). Of those 95 genes associated with liver cancer, the expression pattern of 14 DEGs (13 upregulated genes and 1 downregulated gene) was distinctly observed only in the PB-HRAS^G12V/PB-c-MYC-transfected cohort versus the non-injected controls ([140]Figures 5A and 5B). Among the 14 DEGs, Aurka, Cdc20, and Igf2 exhibited a marked upregulation (i.e., 137-fold, 262-fold, and 1,034-fold change, respectively). In contrast, in the PB-HRAS^G12V-transfected cohort, there were three DEGs compared to the non-injected control (i.e., Egf, Stm1, and Igf2) ([141]Figures 5A and 5B). In the PB-c-MYC-transfected cohorts, there were seven DEGs compared to the non-injected control (i.e., Adam17, Birc2, Epo4, Pdgfra, Akt1, Cdkn1a, and Rb1) ([142]Figures 5A and 5B). Figure 5. [143]Figure 5 [144]Open in a new tab Molecular Analysis of RNA Expression Profiles and Pathway Analysis during HRAS^G12V/c-MYC-Mediated Hepatocarcinogenesis Six-week-old C57BL/6JRj male mice were hydrodynamically co-transfected with 1 μg of PB-c-MYC, 1 μg of PB-HRAS^G12V, and 1 μg of hyperactive PB-encoding plasmid. As a control to study the cooperative effect of HRAS^G12V and c-MYC, mice were hydrodynamically co-transfected with 1 μg of single transposon plasmid expressing either HRAS^G12V or c-MYC (PB-HRAS^G12V or PB-c-MYC, respectively) and 1 μg of hyperactive PB-encoding plasmid. Non-injected mice were used as a negative control in this study. After 8 weeks of treatment, total extracted RNA from mouse liver tissues was used as a template for qRT-PCR using primers targeting 95 genes that are known to be associated with hepatocarcinogenesis. (A) Heatmap representing log[2] fold change of all 95 gene expression profiles from PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, PB-c-MYC, and non-injected groups. Genes that showed p < 0.05 for differential expression were defined as differentially expressed genes (DEGs); otherwise, they were defined as non-differentially expressed genes (non-DEGs). Red and blue indicate upregulation and downregulation of the genes, respectively. DEGs were ranked based on fold difference (from high to low), and the number on the left of the heatmap indicates an average fold difference (PB-HRAS^G12V+PB-c-MYC, n = 4 mice; PB-HRAS^G12V, n = 4; PB-c-MYC, n = 4 mice; non-injected, n = 4 mice). (B) Venn diagram of DEGs in PB-HRAS^G12V+PB-c-MYC, PB-HRAS^G12V, and PB-c-MYC groups compared to non-injected groups. The genes in intersected areas were identified as common DEGs between groups whereas the genes in non-intersected areas indicated group-specific DEGs. (C) Protein-protein interaction (PPI) networks of all DEGs of PB-HRAS^G12V+PB-c-MYC-transfected group versus non-injected group generated by the STRING database ([145]https://string-db.org). Next, a protein-protein interaction (PPI) network of the 14 DEGs, HRAS^G12V, and c-MYC was generated using STRING software to better understand how these proteins systemically interacted during hepatocarcinogenesis. The constructed PPI network was highly connected and non-random (clustering coefficient = 0.771, PPI enrichment p = 9.55e−13) with an average node degree of 4.25 ([146]Figure 5C). Notably, Aurka, Akt, and Trp53 exhibited a relatively greater number of interactions (node degree = 6, 13, and 14, respectively) than the average. Taken together, this suggests that Aurka, Akt, and Trp53 may play a crucial role in HRAS^G12V- and c-MYC-dependent hepatocarcinogenesis.[147]47, [148]48, [149]49 We further explored the biological significances of 14 DEGs using Profiler software for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses. The results obtained from GO analysis showed that the DEGs were significantly involved with regulation of cell cycle (9–11/14 DEGs), apoptotic (11/14 DEGs), and metabolic processes (11–13/14 DEGs) ([150]Table S1). Similarly, significant biological pathways of the DEGs based on KEGG pathway enrichment consist of pathways involved in apoptosis (6/14 DEGs), cellular senescence (5/14 DEGs), p53 signaling (4/14 DEGs), and cell cycle (4/14 DEGs) ([151]Table S1). Long-Term Follow-Up of Tumor Progression in Rapid-Onset HCC Model To better characterize the disease onset and tissue-specific tumorigenesis of the HCC model using the hepatocyte-specific hyPB platform, a long-term follow-up study was carried out in mice stably transfected with PB-HRAS^G12V in combination with PB-c-MYC or with either the PB-HRAS^G12V or PB-c-MYC transposon plasmid alone. These PB transposons were co-transfected with an expression construct encoding the hyperactive PB transposase to enable the efficient stable integration and expression of HRAS^G12V and/or c-MYC in the transfected hepatocytes. Kaplan-Meier survival analysis revealed that the median survival of PB-HRAS^G12V/PB-c-MYC-transfected mice was 93 days, which was 2.6-fold faster than that of mice transfected with only PB-HRAS^G12V ([152]Figure 6A). In contrast, mice stably transfected with PB-c-MYC, PB-hFIXIA, or PBS survived for 365 days ([153]Figure 6A), at which time the experiment was terminated. The decrease in median survival observed in PB-HRAS^G12V/PB-c-MYC-transfected mice was consistent with rapid HCC development and/or progression, including early-onset tumorigenesis and massive HCC tumor growth, compared to those in the other cohorts ([154]Figure 6B). In particular, hepatocarcinogenesis progressed more rapidly in PB-HRAS^G12V/PB-c-MYC-overexpressing mice than in DEN-treated mice ([155]Figure 6A). Tumorigenesis was restricted to the liver and absent in other organs and tissues, consistent with the use of the hepatocyte-specific HS-CRM8-TTR[min] promoter to drive the expression of HRAS^G12V and c-MYC ([156]Figure 6B). The elevated expression of the proliferation marker protein Ki-67 (Mki67) gene, one of the most common proliferative markers in HCC, was clearly observed in HRAS^G12V and c-MYC liver tissues ([157]Figure 6C).[158]50, [159]51, [160]52 In addition, we identified a dysregulation of other biomarkers (i.e., GluI, Hadha [hydroxyacyl-coenzyme A dehydrogenase trifunctional multienzyme complex subunit alpha], Acly [ATP-citrate synthase], Sc1, and Fasn [fatty acid synthase]) in HRAS^G12V/c-MYC liver tissues, which were known to be associated with HRAS^G12V and c-MYC-induced hepatic tumors ([161]Figure 6C).[162]^50 RNA expression analysis confirmed that only the mice transfected with PB-c-MYC/PB-HRAS^G12V expressed high levels of both HRAS and c-MYC mRNA ([163]Figure 6D). In contrast, control mice transfected with either the PB-HRAS^G12V or PB-c-MYC transposon only expressed increased levels of HRAS^G12V or c-MYC mRNA, respectively ([164]Figure 6D). The long-term presence of the HRAS^G12V or c-MYC transgene in transfected livers is consistent with the stable genomic integration of the PB transposon responsible for the long-term HRAS^G12V or c-MYC mRNA expression ([165]Figure 6E). Notably, an increase of copy number of PB transposon was observed in liver tumor compared to non-tumor liver tissue in the HRAS^G12V/c-MYC group ([166]Figure 6F). In conclusion, we established a rapid-onset HCC model following the hepatocyte-specific delivery and expression of the HRAS^G12V and c-MYC oncogenes in adult mouse livers using the hyPB platform. These results indicate that the HRAS^G12V and c-MYC oncogenes cooperate in the development and/or progression of HCC. Figure 6. [167]Figure 6 [168]Open in a new tab Long-Term Follow-Up of Tumor Progression in Rapid-Onset HCC Model Six-week-old C57BL/6JRj male mice were hydrodynamically co-transfected with 1 μg of PB-c-MYC in combination with PB-HRAS^G12V, PB-HRAS^G12V, PB-c-MYC, or PB-hFIXIA and 1 μg of hyperactive PB-encoding plasmid. As controls, mice were injected with PBS. (A) One-year Kaplan-Meier survival curve analysis of treated mice from all groups. Graph presents survival days of individual mice; log-rank test (PB-HRAS^G12V+PB-c-MYC, n = 10 mice; PB-HRAS^G12V, n = 7 mice; PB-c-MYC, n = 7 mice; PB-hFIXIA, n = 7 mice; PBS-injected, n = 7 mice). (B) Representative macroscopic morphology of mouse organs from all groups. Tumor formation was only observed in mouse livers of the combined PB-c-MYC and PB-HRAS^G12V and individual PB-HRAS^G12V treatment groups at different onset periods. Scale bars represent 5 mm. (C) Relative mRNA expression of a proliferative gene marker (i.e., Mki67) and other genes that are responsible for glutamine metabolism (i.e., Glul) and fatty acid metabolism (i.e., Hadha, Acly, Scd1 [stearoyl-coenzyme A desaturase 1], and Fasn) from the combined PB-c-MYC and PB-HRAS^G12V and healthy (non-injected) groups. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-c-MYC+PB-HRAS^G12V, n = 5 mice; healthy (non-injected), n = 5 mice). (D) RNA expression analysis of HRAS and c-MYC in mouse livers of the combined PB-c-MYC and PB-HRAS^G12V, individual PB-HRAS^G12V, and PB-c-MYC treatment groups to determine the elevated oncogene expression according to the types of transfected oncogenes. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant; *p < 0.05, **p < 0.01) (Healthy (non-injected), n = 5 mice; PB-c-MYC+PB-HRAS^G12V, n = 6 mice; PB-HRAS^G12V, n = 6 mice; PB-c-MYC, n = 6 mice). (E) Quantitative PB transposon copy number per diploid genome in the mouse livers of the combined PB-c-MYC and PB-HRAS^G12V, individual PB-HRAS^G12V, and PB-c-MYC groups measured by qPCR to ensure the presence of the PB transposon in mouse hepatocytes after transfection. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant) (PB-c-MYC+PB-HRAS^G12V, n = 6 mice; PB-HRAS^G12V, n = 6 mice; PB-c-MYC, n = 6 mice).(F) Quantitative PB transposon copy number per diploid genome in the HCC tissue and normal liver tissue of combined PB-c-MYC and PB-HRAS^G12V mice measured by qPCR. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (***P < 0.001). (n = 4 tissue samples of HCC regions and corresponding adjacent normal regions from 4 individual PB-c-MYC+PB-HRAS^G12Vmice). miR-20a Suppresses Tumorigenesis in the Rapid-Onset HCC Model To assess the robustness of the tumor suppressor effects of miR-20a, which were initially validated in the slow-onset chemically induced HCC model ([169]Figure 3), we subsequently explored whether miR-20a could also function to suppress HCC initiation in the more aggressive, rapid-onset HRAS^G12V/c-MYC HCC model. Ultimately, this would provide more comprehensive insight into the inhibitory roles of miR-20a not only on tumor progression but also tumor initiation. HCC formation was induced by the liver-directed transfection of PB-HRAS^G12V and PB-c-MYC in C57BL/6JRj male mice in conjunction with a plasmid encoding a hyperactive PB transposase. The recipient mice were also transfected in parallel either with the PB-hFIXIA-miR-20a transposon or its corresponding PB-hFIXIA control. Consistent with the results obtained in the chemically induced HCC model, a significant reduction in HCC tumor formation was apparent in the HCC mice transfected with the PB-hFIXIA-miR-20a transposon compared to that in the PB-hFIXIA control group ([170]Figure 7A). Macroscopic liver examination did not reveal any HCC formation in any cohort at 3 weeks post-transfection. Although all mice developed HCC at 6 weeks post-transfection ([171]Figure 7A), a relatively robust and significant 4-fold reduction in HCC tumor nodules was apparent in the PB-hFIXIA-miR-20a-transfected livers compared to that in the controls ([172]Figure 7B). Moreover, miR-20a overexpression in HCC mice following PB-hFIXIA-miR-20a transfection significantly prolonged their overall survival for at least 140 days, whereas the median survival of the control group was only 85 days ([173]Figure 7C). This is consistent with a significantly 44-fold higher miR-20a expression in PB-hFIXIA-miR-20a-transfected HCC mice compared to the control PB-hFIXIA group ([174]Figure 7D). In the liver tissues of the PB-hFIXIA-miR-20a-transfected HCC group, a marked 4-fold increase of miR-20a expression level in non-tumor area compared to tumor area was observed ([175]Figure 7E). In addition, the copy numbers of all three PB transposons (i.e., PB-HRAS^G12V, PB-c-MYC, and PB-hFIXIA-miR-20a) present in non-tumor region PB-hFIXIA-miR-20a-transfected livers were comparable, suggesting at least 40% of overall transfection efficiency of each PB transposon ([176]Figure 7F). FIX levels ([177]Figure 7G) and transposon copy number ([178]Figure 7H) were not significantly different between the PB-hFIXIA-miR-20a-transfected recipient mice and the PB-hFIXIA-transfected controls. These results confirm that miR-20a is a robust tumor suppressor gene capable of inhibiting HCC development even in a rapid-onset HRAS^G12V/c-MYC HCC model. Figure 7. [179]Figure 7 [180]Open in a new tab Analysis of the Tumor-Suppressive Effects of miR-20a in the Rapid-Onset HCC Model Six-week-old C57BL/6JRj male mice were hydrodynamically co-transfected with (1) 0.25 μg of PB-c-MYC, (2) 0.25 μg of PB-HRAS^G12V, (3) 10 μg of either PB-hFXIA or PB-hFIXIA-miR-20a, and (4) 2.25 μg of hyperactive PB-encoding plasmid (PB-hFIXIA, n = 8 mice; PB-hFIXIA-miR-20a, n = 7 mice). (A) Representative macroscopic tumor burden was assessed at 3 and 6 weeks post-injection. Scale bars represent 5 mm. (B) Number of tumor nodules was counted at 6 weeks after injection to compare the hepatic tumor burden. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant; *p < 0.05, **p < 0.01) (PB-hFIXIA, n = 8 mice; PB-hFIXIA-miR-20a, n = 7 mice). (C) Kaplan-Meier survival curve analysis of PB-hFIXIA and PB-hFIXIA-miR-20a-treated HCC mice. Graph presents survival days of individual mice; log-rank test (PB-hFIXIA, n = 8 mice; PB-hFIXIA-miR-20a, n = 7 mice) (PB-hFIXIA, n = 6 mice; PB-hFIXIA-miR-20a, n = 6 mice). (D) Quantitative miR-20a expression in the liver tissues of PB-hFIXIA-miR-20a- and PB-hFIXIA-treated HCC groups at 6 weeks post-treatment measured by qRT-PCR. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (*p < 0.05) (PB-hFIXIA, n = 5 mice; PB-hFIXIA-miR-20a, n = 5 mice). (E) Quantitative miR-20a expression in the tumor and non-tumor liver tissues of PB-hFIXIA-miR-20a-treated HCC group at 6 weeks post-treatment measured by qRT-PCR. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (**p < 0.01) (n = 5 tissue samples of tumor and corresponding adjacent non-tumor regions from five individual PB-hFIXIA-miR-20a-treated mice). (F) Quantitative copy number of all PB transposons (i.e., PB-HRAS^G12V, PB-c-MYC, and PB-hFIXIA-miR-20a) per diploid genome in the mouse liver of the PB-hFIXIA-miR-20a-treated group at 6 weeks post-treatment measured by qPCR. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant) (n = 5 mice). (G) Quantitative hFIX protein expression levels at 1 week post-treatment measured by ELISA to monitor the transfection efficiency of PB transposons. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant) (PB-hFIXIA, n = 6 mice; PB-hFIXIA-miR-20a, n = 6 mice). (H) Quantitative overall PB transposon copy number per diploid genome in the mouse liver at 6 weeks post-treatment measured by qPCR to confirm the presence of the PB transposon in mouse hepatocytes after transfection. Graph represents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant) (PB-hFIXIA, n = 6 mice; PB-hFIXIA-miR-20a, n = 6 mice). Effects of miR-20a on Gene Expression Signatures in the Rapid-Onset HCC Model We subsequently determined whether miR-20a altered the expression of any of the putative miR-20a target genes such as phosphatase and tensin homolog (Pten), Unc-51-like autophagy activating kinase 1 (Ulk1), and induced myeloid leukemia cell differentiation protein (Mcl1) ([181]Table S5).[182]^14^,[183]53, [184]54, [185]55, [186]56 The effects of miR-20a overexpression on the expression at the post-transcriptional and translational levels of the putative target genes were examined. Interestingly, Pten expression, a known tumor suppressor gene in HCC, was significantly upregulated in HRAS^G12V/c-MYC HCC mice that were stably transfected with the PB-hFIXIA-miR-20a transposon to levels that were comparable to those in the control mice injected with PBS ([187]Figures 7B, 7C, and [188]8A).[189]^57^,[190]^58 In contrast, no differential expression of Pten at the mRNA level was apparent ([191]Figure S2D). This suggests that miR-20a potentially modulates Pten expression at the translational level during HCC suppression in vivo possibly through an as yet unknown indirect mechanism, potentially involving inhibition of a negative regulator of Pten. In contrast to Pten, overexpression of miR-20a following liver-directed transfection with PB-hFIXIA-miR-20a transposons affected neither RNA nor protein expression levels of the other putative miR-20a candidate targets (i.e., Mcl1 and Ulk1; [192]Figure S2). Figure 8. [193]Figure 8 [194]Open in a new tab Effects of miR-20a on Gene Expression Signatures in the Rapid-Onset HCC Model Six-week-old C57BL/6JRj male mice were hydrodynamically co-transfected with (1) 0.25 μg of PB-c-MYC, (2) 0.25 μg of PB-HRAS^G12V, (3) 10 μg of either PB-hFIXIA or PB-hFIXIA-miR-20a, and (4) 2.25 μg of hyperactive PB-encoding plasmid. After 6 weeks of treatment, total extracted RNA from mouse liver tissues was used as a template for qRT-PCR using primers targeting 95 genes that are known to be associated with hepatocarcinogenesis. Total proteins were also extracted to compare the target protein expression level between PB-hFIXIA an PB-hFIXIA-miR-20a-transfected HCC groups. (A) Representative image and (B) quantitative Pten protein expression normalized to Gapdh (glyceraldehyde 3-phosphate dehydrogenase) expression from western blotting analysis. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant; *p < 0.05) (PB-hFIXIA, n = 6 mice; PB-hFIXIA-miR-20a, n = 6 mice). (C) ELISA analysis of Pten was performed to compare the protein expression level among all treatments. Graph presents the mean ± SEM of data obtained from individual mice; Mann-Whitney U test (n.s., not significant; **p < 0.01) (PB-hFIXIA, n = 6 mice; PB-hFIXIA-miR-20a, n = 6 mice). (D) Heatmap representing log[2] fold change of all 95 gene expression profiles from PB-hFIXIA- and PB-hFIXIA-miR-20a-treated HCC groups. Genes that showed p < 0.05 for differential expression were defined as differentially expressed genes (DEGs); otherwise, genes were defined as non-differentially expressed genes (non-DEGs). Red and blue indicate upregulation and downregulation of the genes, respectively. DEGs were ranked based on fold difference (from high to low), and the number on the left of heatmap indicates an average fold difference (PB-hFIXIA-miR-20a, n = 4 mice; PB-hFIXIA, n = 4 mice). (E) Protein-protein interaction (PPI) networks of all DEGs of PB-hFIXIA-miR-20a-transfected HCC group versus PB-hFIXIA HCC group generated by the STRING database. To gain a better understanding of the possible tumor suppressor mechanisms of miR-20a in the rapid-onset HRAS^G12V/c-MYC HCC model, RNA expression profiling of the selected 95 liver cancer-related genes in mouse liver tissues of PB-hFIXIA-miR-20a and PB-hFIXIA-transfected cohorts was carried out using qRT-PCR ([195]Table S4). We found that 14 genes were significantly dysregulated in the hFIXIA-miR-20a cohort versus hFIXIA controls (13 downregulated genes and 1 upregulated gene) ([196]Figure 8D). Notably, the oncogene Tgfa and the tumor suppressor gene Cdh13 showed a more pronounced differential expression pattern compared to other DEGs (i.e., 20-fold downregulation and 86-fold upregulation, respectively). The PPI network of 14 DEGs from the hFIXIA-miR-20a cohort versus hFIXIA control (clustering coefficient = 0.51, PPI enrichment p = 1.67e−09, average node degree = 2.43) showed that BCL2-like 1 (Bcl2l1) exhibited the highest number of interactions among other genes (six interactions), suggesting its central role during miR-20a-mediated HCC suppression in mice ([197]Figure 8E). These DEGs were functionally enriched in genes involved in the regulation of apoptotic processes (12/14 DEGs), programmed cell death (12/14 DEGs), and cell cycle (10/14 DEGs) based on GO analysis ([198]Table S2). Correspondingly, functional pathway enrichment of all DEGs using the KEGG database includes pathways involved in apoptosis (5/14 DEGs), cellular senescence (4/14 DEGs), and other signaling pathways (i.e., nuclear factor κB [NF-κB] signaling pathway, 4/14 DEGs and TNF signaling pathway, 4/14 DEGs) that are known to be associated with cell death and apoptosis.[199]^59^,[200]^60 Discussion In the current study, we validated a semi-high-throughput in vivo screen based on the hyPB transposon platform specifically designed to selectively express miRs in hepatocytes and identify those that affect HCC development and progression. We systematically analyzed the impact of each miR in the miR-17-92 cluster (i.e., miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92a-1) and other miRs that are thought to play a role in HCC (i.e., miR-370, miR-1188, miR-341, miR-221, miR-222, miR-106b, miR-25, miR-93, and miR-106b). Notably, the screening revealed that miR-20a functions as a tumor suppressor gene inhibiting HCC development and progression in both the slow-onset chemically induced and rapid-onset mouse models, which had not been shown previously. Our results indicated that miR-20a overexpression in the liver significantly decreased the number and expansion of HCC nodules, consistent with the significant prolongation of overall survival in the rapid-onset HCC model. For the chemically induced DEN HCC model, the PB-hFIXIA-miR transposons were transfected into the mice, 4 weeks after tumor initiation by chemical induction. Hence, the miR-20a likely suppressed tumor progression rather than tumor initiation per se since the tumors had already been initiated by DEN, 4 weeks prior to miR-20a overexpression. The DEN HCC model depends on the stochastic and random mutations that contribute to the subsequent initiation (and progression) of HCC. Typically, Hras, Braf, Egfr, and/or Apc genes are thought to play a role in DEN-induced hepatocellular carcinogenesis.[201]^61 Consequently, this process is far less controllable and typically takes much longer for HCC tumors to develop (>9 months). In the rapid-onset HCC model, the PB transposons encoding HRAS^G12V/c-MYC and miR-20a were all injected at the same, implying that the miR-20a may also have an impact on tumor initiation per se in this model, although a possible effect on tumor progression cannot be excluded. The rationale for testing the effect of miR-20a in a rapid-onset model was to evaluate the impact of miR-20a in hepatocytes that from the onset already expressed high levels of HRAS and c-MYC oncogenes that synergistically contribute to HCC initiation and/or progression and that are therefore poised to rapidly progress toward full-blown HCC tumors. The demonstration that miR-20a can even suppress HCC tumors in such an aggressive, rapid-onset model as compared to a slow-onset model strongly suggests that miR-20a is a relatively potent tumor suppressor gene. Moreover, it shows that miR-20a can suppress HCC regardless of the method or mechanism of HCC initiation. This indicates that the tumor suppressor effects of miR-20a are not restricted to a single HCC model and suggests its therapeutic potential and broad implications for HCC in general. To our knowledge, this is the first in vivo demonstration that miR-20a is a possible tumor suppressor gene in HCC. This is consistent with previous in vitro studies in tumor cell lines indicating that miR-20a inhibits cell proliferation and invasion and promotes cancer cell sensitization to apoptosis.[202]^14^,[203]^15^,[204]^62 Moreover, miR-20a expression is downregulated in HCC samples from patients compared to that in normal tissue.[205]^14^,[206]^15^,[207]^62 However, the biological significance of these previous observations was unknown. Our current study now establishes a direct causal link between miR-20a overexpression and suppression of HCC in in vivo models. The previous publications were based on in vitro studies in hepatic cell lines only or patient biopsies were circumstantial at best and never formally demonstrated based on in vivo HCC models that miR-20a is a tumor suppressor gene.[208]^14^,[209]^15 These previous studies merely showed differential miR-20a expression levels between normal and tumor cells, which does not formally establish a causal link between elevated miR-20a expression and tumor suppression in HCC in vivo. In fact, with the exception of miR-20a, we demonstrated that other candidate miR genes that were previously found to be differentially expressed in HCC compared to normal hepatocytes had no impact whatsoever on the HCC tumor initiation and/or progression in vivo ([210]Figure 3; [211]Figure S1).[212]^13^,[213]^16^,[214]^17^,[215]19, [216]20, [217]21, [218]22 Hence, our current findings caution against merely relying on in vitro data, which are prone to false positives and negatives, or patient biopsies, which do not formally establish a causal relationship between mi-R20a expression and HCC suppression. This emphasizes the need for in vivo confirmation to assess the real impact of a given miR on tumor initiation/progression in HCC models. We demonstrated that miR-20a altered the mRNA expression levels of several HCC-related genes in the HRAS^G12V/c-MYC-induced HCC mode, suggesting that each of them could play a role in HCC suppression by miR-20a. Notably, most of these differentially expressed genes were shown to functionally interact with Bcl2l1, which is itself also differentially expressed upon miR-20a overexpression. The Bcl2l1 oncogene belongs to the Bcl-2 family, which exhibits anti-apoptotic activity through sequestration of BH3 domain-only molecules from Bak activation during apoptosis.[219]^63 In addition, overexpression of Bcl2l1 was observed in liver tissues of HCC patients and resulted in tumor progression.[220]^64^,[221]^65 This is consistent with downregulation of the Bcl2l1 oncogene and reduction of HCC tumor burden following PB-hFIXIA-miR-20a transfection in the rapid-onset HRAS^G12V/c-MYC-induced HCC model. Alternatively, a significant upregulation of the Cdh13 tumor suppressor gene and downregulation of the Tgfa oncogene may also have contributed to HCC suppression by miR-20a overexpression through inhibition of cell proliferation in liver.[222]^66^,[223]^67 Cdh13, a glycosylphosphatidylinositol (GPI)-anchored cadherin devoid of transmembrane and cytoplasmic domain, regulates hepatocyte function such as albumin secretion and urea synthesis. Interestingly, overexpression of Cdh13 suppresses proliferation in gastric cancer cells.[224]^68^,[225]^69 Moreover, significant downregulation of CDH13 expression was observed in liver tissues of patients with metastatic HCC.[226]^66 Tgfa encodes transforming growth factor α (TGF-α) and binds to epidermal growth factor receptor and subsequently activates a signaling pathway that promotes cell proliferation.[227]^70 Particularly in HCC, TGF-α enhances DNA synthesis and extracellular signal-regulated kinase (ERK)-mitogen-activated protein kinase (MAPK)-dependent cell proliferation. Moreover, TGF-α expression was upregulated in the liver of HCC patients.[228]71, [229]72, [230]73, [231]74 Functional pathway enrichment analysis suggests that a miR-20a-mediated inhibitory mechanism may be associated with apoptotic processes, which are known to regulate HCC development.[232]^75^,[233]^76 Moreover, miR-20a may potentially suppress HRAS^G12V/c-MYC-mediated hepatocarcinogenesis by upregulating Pten protein expression levels, a known tumor suppressor gene in HCC.[234]^77^,[235]^78 Consistent with an increase of Pten protein expression following tumor regression in our study, previous findings showed that Pten protein deficiency promotes HCC development.[236]^78^,[237]^79 It is possible that indirect mechanisms account for the increase in Pten protein expression, for example through miR-20a-mediated inhibition of negative regulators of Pten, particularly since miR-20a overexpression did not affect the steady-state Pten mRNA levels. The prioritization of the 95 genes associated with HCC permitted identification of possible lead candidates that could play a key role in miR-20a-mediated hepatocarcinogenesis. However, it cannot be excluded that other genes may also play a role warranting future comprehensive transcriptomics and mechanistic studies, which are beyond the scope of the current study. Previous studies have shown that the overexpression of short hairpin RNA (shRNA) can cause low-level hepatotoxicity that can facilitate the ability of the c-MYC oncogene to induce liver tumorigenesis.[238]^80^,[239]^81 Moreover, oversaturation of cellular microRNA/shRNA pathways may result in death, possibly by interfering with karyopherin exportin-5. However, the results of the present study suggest that enhanced tumorigenicity and/or hepatotoxicity may not be a concern for miR-20a overexpression because it diminishes tumorigenicity, even in the current rapid-onset HCC model that relies on the forced overexpression of the MYC and Ras oncogenes. The tumor suppressor effect of miR-20a may not be restricted to HCC and may have broader implications for other types of malignancies. In particular, it was recently shown that miR-20a expression in liver sinusoidal endothelial cells (LSECs) inhibits colorectal metastasis to the liver.[240]^82 Hence, to further assess the broader impact of miR-20a in oncology and confirm our current findings beyond HCC, future studies are needed to further investigate the role of miR-20a in cancers of different origins and different (sub)types. One of the main advantages of the current in vivo approach based on the selective hepatic overexpression of a given miR is that it allows the establishment of a causal relationship between miR expression and HCC, thus excluding possible unrelated associations. Interestingly, miR-20a was the only miR in the miR-17-92 cluster that had an impact on HCC, challenging the prevailing assumption that miRs located in the miR-17-92 cluster typically function as oncogenes, as suggested by transgenic studies.[241]^1^,[242]^2^,[243]^13^,[244]^83 In the current study, a rapid-onset HCC model was also developed to further validate and confirm the tumor suppressor properties of miR-20a. This rapid-onset HCC model was developed by the somatic and sustained overexpression of HRAS^G12V and c-MYC in hepatocytes following stable PB transposon-mediated genomic integration of the HRAS^G12V and c-MYC oncogenes. This rapid-onset model offers several advantages over conventional HCC models. The current rapid-onset HCC model is generated by co-delivery of PB-c-MYC and PB-HRAS^G12V and is based on the de novo expression in adult, post-mitotic hepatocytes, complementing the shRNA/c-MYC model described by Beer et al.[245]^81 Whereas both models result in rapid-onset HCC, the current PB-c-MYC and PB-HRAS^G12V HCC model did not require a transgenic model as in the case of the Tet-regulatable c-MYC transgenic mouse models used in the Beer et al. study. Hence, liver-specific somatic overexpression of oncogenes using the PB transposon platform provides a rapid, cost-effective, and relatively straightforward alternative to viral vector-mediated and transgenic HCC mouse models and obviates time-consuming and elaborate breeding schemes for different transgenic mouse strains. Moreover, these different HCC models likely contribute to rapid HCC development through different molecular pathways, which may provide new complementary insights into HCC development. A low level of shRNA expression causes gross perturbation of the miR signatures in the liver of the c-MYC transgenic mice, regardless of the shRNA type or specificity, whereas the current HRAS^G12V/c-MYC model results in different molecular signatures. The initial objective of the Beer et al. study was to selectively downregulate the expression of the tumor-suppressor gene p53 using p53-specific shRNA while conditionally expressing the transgenic c-MYC oncogene. As expected, the shRNAs silenced hepatic p53 and accelerated liver tumorigenesis when c-MYC was concurrently expressed. One of the unexpected and intriguing findings of the Beer et al. study was that various irrelevant control shRNAs (i.e., directed against hepatitis B virus surface antigen against or α1-antitrypsin) similarly induced a rapid onset of tumorigenesis, comparable to carbon tetrachloride (CCl[4]), a potent carcinogen. Even marginal shRNA doses can already trigger histologically detectable hepatotoxicity and increased hepatocyte apoptosis. This suggests that the accelerated effect of shRNA on tumorigenesis is not shRNA-specific and may be due to a general cytotoxic effect that nonspecifically increased hepatocyte proliferation. In contrast, in our current study we demonstrated that the tumor suppressor effect of miR-20a is miR-20a-specific and has not been observed with any other miR that was tested, at least in the DEN HCC model. Hence, modulating tumorigenicity following miR overexpression may be dependent on specific miRs, whereas shRNA overexpression may influence tumorigenicity in a more general, non-specific fashion, perhaps by interfering with the cellular RNA interference (RNAi) machinery.[246]^80 In addition, it offers a relatively short latency period compared to that of DEN-induced HCC, which usually requires 9–12 months.[247]^36 Alternatively, other transposon systems, such as Sleeping Beauty, can be used to selectively express oncogenes in the liver.[248]^50^,[249]84, [250]85, [251]86, [252]87, [253]88 However, a comparative analysis of different transposon systems revealed that hyPB led to an approximately 10-fold improvement in transgene expression compared to that with SB100X, thus suggesting a potential advantage of the hyPB transposon system for in vivo sustained gene expression.[254]^89 This increased expression likely reflects an increase in transposon copy number per se rather than an increased mRNA expression level per transposon copy. Nevertheless, the SB system may yield a more random integration pattern, whereas PB transposons favor integration into genes.[255]^90^,[256]^91 Another advantage of the current system is that the synthetic hepatocyte-specific promoter allows for selective expression in only hepatocytes, avoiding undesirable effects or interfering tumor lesions in non-hepatocytes or ectopic tissue.[257]28, [258]29, [259]30^,[260]^92 The rapid-onset HRAS^G12V/c-MYC HCC model provides proof that the HRAS^G12V and c-MYC oncogenes cooperate in the development of HCC. This is consistent with previous studies showing that dysregulation of the Ras and c-Myc oncogene-dependent pathways promotes malignant transformation and tumor progression in the liver.[261]^41^,[262]^42 Our data suggests that the cooperative effects on cancer development mediated by the Ras/MAPK pathway and c-MYC may require c-MYC stabilization and Ras/MAPK/ERK signaling, which ultimately contributes to hepatocarcinogenesis.[263]93, [264]94, [265]95 This is supported by a substantial upregulation and multiple functional interactions of Aurka in mice overexpressing HRAS^G12V and c-MYC as observed in the current study. Phosphorylated c-MYC forms a protein complex with Aurka that stabilizes c-MYC.[266]^96 Moreover, c-MYC induces Aurka mRNA expression and Aurka induces c-MYC mRNA expression, suggesting a positive feedback loop.[267]^97 In addition, Aurka binds to HRas and further induces ERK phosphorylation through Ras/MAPK signaling.[268]^98^,[269]^99 Akt1, one of the highly interactive differentially expressed genes during HRAS^G12V/c-MYC HCC development, may prevent c-MYC degradation through Ras-driven activation of Akt.[270]^100 Upregulation of Cdc20 and Igf2 following HRAS^G12V and c-MYC overexpression could perturb cell cycle regulation at the G[2]/M phase and drive hepatocyte proliferation in mice, ultimately contributing to HCC.[271]^101^,[272]^102 In contrast, rapid-onset HCC development was found to not depend on the cooperation between HRAS^G12V and the MDM2 or MDM4 oncogenes.[273]^43^,[274]^44 This is in contrast to what has been observed in other tumors, such as colon carcinoma and melanomas. Indeed, in these cases, RAS is frequently activated with the concomitant overexpression of MDM2 and/or MDM4, suggesting an important role of the MDM2/MDM4 axis in inhibiting p53-dependent arrest in RAS-expressing cells.[275]^103^,[276]^104 Consistent with previous studies, overexpression of Ras and c-Myc further suppresses fatty acid metabolism by decreasing the expression of enzymes that are responsible for fatty acid accumulation and degradation.[277]^50 In addition, c-Myc-mediated liver tumorigenesis is known to downregulate the expression of glutamine synthetase (Glul) and subsequently reduce glutamine synthesis.[278]^105 One of the merits of the slow- and rapid-onset HCC models described in the present study is that they obviate the need for orthotopic transplantation of transformed HCC cell lines, which does not mimic natural HCC initiation and progression.[279]106, [280]107, [281]108 Orthotopic transplantation typically involves subcutaneous implantation of already established HCC cell lines in a non-natural context and microenvironment and does not allow investigation of the early steps of HCC tumor initiation.[282]^109 The models used in the current study overcome these limitations, and the results demonstrate unequivocally that miR-20a inhibits HCC initiation and progression in both the slow- and rapid-onset HCC models, which more closely mimic the natural context and microenvironment that foster HCC development. To our knowledge, no prior study has assessed the impact of overexpressing miRs using transposon systems on tumor initiation and progression in similar HCC mouse models that are based on de novo carcinogenesis. Moreover, the current study complements previous studies showing that Sleeping Beauty transposons can be used to express miRs in pulmonary fibrosis mouse models.[283]^110 The current study strengthens the diagnostic, prognostic, and therapeutic relevance of miR-20a in HCC. In particular, it supports the notion that miR-20a overexpression is associated with a more favorable prognosis in patients with HCC compared to that in patients who have only low or no miR-20a expression in their HCC biopsies. Moreover, the delivery of miR-20a may offer new therapeutic perspectives for the treatment of HCC. Recently, the first phase I clinical trial of miR therapy using liposome-based delivery was launched to evaluate the safety of treatment in patients with primary liver cancer (ClinicalTrials.gov: [284]NCT01829971), and it may be applicable to miR-20a. Alternatively, the use of folic acid-modified nanoparticle-like lipid-based protocells was shown to be an effective method for the targeted delivery of transposons into cancer cells in a xenograft model, resulting in an in vivo transfection efficiency of up to 35%.[285]^111 Ideally, the sustained and robust expression of miR-20a may be preferred to increase the therapeutic efficacy and prevent HCC recurrence. The presence of a gap junction allows the intracellular miR transfer between transfected cells and non-transfected adjacent cells, thereby potentially increasing its therapeutic impact.[286]^112 Ultimately, this study may aid in the development of other clinically relevant gene delivery platforms, such as adeno-associated viral vectors or lentiviral vectors, to achieve sustained miR-20a expression in patients with HCC.[287]^80^,[288]^113^,[289]^114 Materials and Methods PB Transposon and Transposase Constructs The PB transposon backbone used in this study was previously developed by our group.[290]^30 To generate the PB transposons harboring miR-17-92 cluster-derived miRs, specifically including miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92a-1, the respective miRs flanked with approximately 50–200 bp of genomic sequences from mouse genomic DNA were first PCR amplified and subsequently cloned into the 1.4 kb-truncated human factor IX (hFIX) intron 1 (intron IA), namely hFIXIA ([291]Table S3). The transgenes are flanked by truncated optimized IRs (IR[micros]) ([292]Figures 1B and 1C) composed of the 40 bp of 5ʹ and 67 bp of 3ʹ IRs.[293]^115 The expression of the oncogenes was driven by the liver-specific chimeric promoter containing the TTR[min] promoter as well as HS-CRM8.[294]28, [295]29, [296]30 The bovine growth hormone polyadenylation site (bGH-polyA) functioned as a terminating signal. The hyperactive PB transposase was used with the PB transposons for efficient transgene integration.[297]^30 The coding sequences of four oncogenes, including c-MYC, HRAS^G12V, MDM2, and MDM4, were PCR amplified using specific primers harboring the NheI/BglII restriction sites and Phusion hot start high-fidelity DNA polymerase (Thermo Fisher Scientific, Merelbeke, Belgium) activity ([298]Figure 1C) and cloned into the corresponding sites of the PB transposon plasmid.[299]^30 For efficient in vivo transposition, we employed the hyperactive version of PB transposase containing 7 aa substitutions, which can improve transgene integration up to 8-fold compared to that with wild-type transposase.[300]^116 Animal Experiments C57BL/6JRj male mice were purchased from Janvier (France) and Taconic (Denmark). All animal procedures were approved by the Institutional Animal Ethics Committee of the Free University of Brussels (VUB) (Brussels, Belgium) and the University of Leuven (Leuven, Belgium). Husbandry was carried out in individually ventilated Thoren cages that contained hygienic animal bedding from Lignocel. Temperature was maintained at approximately 21°C with 50%–60% humidity. Animals were fed SsniFF laboratory animal food (ABEDD Vertriebs, Vienna, Austria) ad libitum. For ectopic gene expression in mice, the PB transposons carrying transgenes were co-delivered with hyperactive-PB transposase-encoding plasmids (2:1 molar ratio of transposon/transposase) by hydrodynamic injection through the tail veins of 6-week-old mice. For DEN-induced hepatocarcinogenesis, 12.5 mg/kg body weight of DEN (Sigma-Aldrich, Diegem, Belgium) was injected intraperitoneally (i.p.) into 14-day-old mice.[301]^117 Four weeks later, mice were hydrodynamically injected with the transposon plasmids if required, whole blood was collected into buffered citrate by phlebotomy of the retro-orbital plexus, and the citrated plasma was stored at −80°C. According to the experimental setting, the gross morphology and tumor formation were assessed after euthanasia to determine the hepatic tumor incidence by enumerating the total numbers of tumor nodules, macroscopically measuring total sizes of hepatic tumor nodules, and weighing the total mass of gross liver tissues. The data were collected for further analysis and statistics. Statistical Analysis GraphPad software (GraphPad, La Jolla, CA, USA) was used for data analysis. A two-tailed independent Student’s t test was performed when the assumptions were valid, and the non-parametric Mann-Whitney U test was applied when the distribution of data was not normal and two variances were not equivalent. Differences in Kaplan-Meier survival curves from two populations were assessed to determine statistical significance by the log-rank test. Results are shown as the mean ± standard error of mean (SEM). n.s., not significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Author Contributions J.T., M.D.M., and W.T. performed and designed the experiments, collected and processed the data, and wrote part of the paper. E.S.-K. and J.K. performed the experiments and/or collected and processed the data. J.A.V.G. collected the data. T.V. and M.K.C. designed the experiments, analyzed the data, wrote and approved the paper, and led the study. T.V. and M.K.C. applied for and obtained financial support for this study. Conflicts of Interest The authors declare no competing interests. Acknowledgments