SYNCRIP Modulates the Epithelial-Mesenchymal Transition in Hepatocytes and HCC Cells

Heterogeneous nuclear ribonucleoproteins (hnRNPs) control gene expression by acting at multiple levels and are often deregulated in epithelial tumors; however, their roles in the fine regulation of cellular reprogramming, specifically in epithelial–mesenchymal transition (EMT), remain largely unknown. Here, we focused on the hnRNP-Q (also known as SYNCRIP), showing by molecular analysis that in hepatocytes it acts as a “mesenchymal” gene, being induced by TGFβ and modulating the EMT. SYNCRIP silencing limits the induction of the mesenchymal program and maintains the epithelial phenotype. Notably, in HCC invasive cells, SYNCRIP knockdown induces a mesenchymal–epithelial transition (MET), negatively regulating their mesenchymal phenotype and significantly impairing their migratory capacity. In exploring possible molecular mechanisms underlying these observations, we identified a set of miRNAs (i.e., miR-181-a1-3p, miR-181-b1-3p, miR-122-5p, miR-200a-5p, and miR-let7g-5p), previously shown to exert pro- or anti-EMT activities, significantly impacted by SYNCRIP interference during EMT/MET dynamics and gathered insights, suggesting the possible involvement of this RNA binding protein in their transcriptional regulation.


Introduction
Epithelial-mesenchymal transition (EMT) is a cellular reprogramming mechanism that allows epithelial cells to acquire mesenchymal properties. This transdifferentiation process has a key role in physiology and pathology, being required in the embryo for gastrulation and morphogenesis, in the adult for wound healing, and in epithelial tumors for several functions, such as stemness, resistance to therapy and, mainly, malignant progression. Transitional cells, indeed, can migrate and disseminate, allowing carcinoma cells to metastasize. In secondary sites, mesenchymal cells can reacquire an epithelial phenotype by undergoing mesenchymal-epithelial transition (MET), which is regulated by tumor niche (reviewed in [1]).
Heterogeneous nuclear ribonucleoproteins (hnRNPs) represent a class of RNA Binding Proteins (RBPs), with conserved RNA-binding domains (RBDs), that control different classes of cellular RNAs. HnRNPs, indeed, are involved in translational regulation, alternative splicing, mRNA stabilization, pri-miRNAs processing as well as miRNAs compartmentalization (reviewed in [11]). HnRNPs can also bind pyrimidine-rich DNA sequences, including those at promoters, and are involved in chromatin remodeling and transcription, telomere elongation, and monitoring the genome integrity [12][13][14][15]. Because of their pleiotropic functions, hnRNPs are often deregulated in pathological conditions, particularly in tumors. However, while it is conceivable that hnRNPs can be involved in the fine regulation of cellular reprogramming, their functions in the regulation of EMT is still largely uncharacterized. Current evidence is limited to the regulation of (i) Snail by hnRNP-A2/B1 and hnRNP-F, respectively, in lung [16] and bladder [17] cancer cells, (ii) disabled-2 (Dab2) and interleukin-like EMT inducer (ILEI) by hnRNP-E1 in mammary gland cells [18,19], and, finally, (iii) invasion by PCBP-1 in hepatoma cells [20].
Despite this body of evidence, a link between SYNCRIP overexpression and the ability of cancer cells to metastasize has not yet been clarified.
Here, SYNCRIP upregulation was found to occur during transforming growth factor (TGF)β-induced EMT. The impairment of this hnRNP in hepatocytes interfered with the responsiveness to TGFβ in terms of morphological changes as well as the modulation of epithelial and mesenchymal gene expression. Moreover, in murine invasive HCC cells, SYNCRIP knockdown was demonstrated to impair migration as well as mesenchymal phenotype.

SYNCRIP Is Involved in TGFβ-Induced EMT of Hepatocytes
To explore the possible SYNCRIP function in the EMTs, we treated non-tumorigenic hepatocytes with a TGFβ cytokine previously found to be sufficient to induce Snail expression in these cells and the loss of the epithelial differentiated phenotype [8,52].
As shown in Figure 1A, the analysis of SYNCRIP levels highlighted that this RBP was upregulated in the transdifferentiation process. In order to evaluate whether SYNCRIP modulation was only correlative or rather causal to the EMT, we analyzed the effects of SYNCRIP knockdown. To achieve this aim, hepatocytes were stably infected with retroviral vectors expressing different shRNAs against SYNCRIP (shSYN) and a scrambled sequence as a control (shCTR). RT-qPCR and Western blot analysis validated that SYNCRIP levels in shSYN cells were significantly reduced by the viral vectors targeting its transcript (Figure 1B,C; in line with [33]).   The image is representative of four independent experiments. GAPDH was used as loading control. The densitometry analysis (right panel) was conducted by using the Image J software. Data are means ± SD of four independent experiments. Statistically significant differences are reported (* p < 0.05; ** p < 0.01; ns = no significance).
As shown in Figure 1D,E, morphological and immunofluorescence analysis in TGFβtreated shCTR hepatocytes confirmed the occurrence of EMT by a morphological change from an epithelial, cobblestone-like phenotype to a more spindle-shaped mesenchymal phenotype with (i) Snail and fibronectin positive regulation and (ii) delocalization of the epithelial markers ZO-1 and E-Cadherin. Conversely, SYNCRIP-interfered cells retained a more epithelial morphology after EMT induction, with the partial retention of ZO-1 and E-Cadherin in the membrane and undetectable Snail and fibronectin expression. Coherently, qRT-PCR ( Figure 1F) and WB ( Figure 1G) analysis showed that while the SYNCRIPknockdown did not interfere with the TGFβ-mediated upregulation of the EMT-TF Zeb2, it significantly prevented the induction of Snail protein and the downregulation of its main epithelial target genes (E-cadherin and HNF1α).
Overall, these findings indicate that SYNCRIP can act as a mesenchymal gene, positively regulated in TGFβ-induced EMT, and provide evidence of a functional role for this hnRNP in the transdifferentiation of the hepatocytes by contributing to the modulation of Snail and its mediated gene regulation.

SYNCRIP Impairment Affects Mesenchymal Phenotype and Migratory Properties of HCC Cells
To shed light on the functional role of SYNCRIP in invasive HCC cells, the effects of its knockdown were further assessed in murine hepatoma mesenchymal-like BW1J cells. As shown in Figure 2, SYNCRIP impairment in BW1J cells (Figure 2A,B) allowed a MET with the rescue of a more differentiated phenotype. Morphological ( Figure 2C) and molecular analysis ( Figure 2D,E,F), indeed, highlighted the strong negative regulation of the mesenchymal markers Vimentin and Fibronectin, the downregulation of the EMT-TF Zeb2 and N-cadherin with the positive modulation of E-cadherin levels.  Notably, SYNCRIP was found to be involved in the acquisition of the migratory ability of HCC cells, as demonstrated by scratch assays of BW1J cells in which SYNCRIP knockdown significantly reduced cell motility ( Figure 2G). Overall, these data demonstrate a positive role for SYNCRIP in allowing migration as well as in the maintenance of the mesenchymal phenotype of HCC cells.
Firstly, we monitored mature miRNAs levels in hepatocytes undergoing EMT. As shown in Figure 3A, RT-qPCR analysis demonstrated that the expected TGFβ repression of miR-200a-5p and miR-122-5p, and induction of miR-181a1-3p and miR-181b1-3p were significantly interfered with by SYNCRIP knockdown. Then, to evaluate at which level of control of miRNAs regulation SYNCRIP could participate, the expression levels of primary miRNAs (pri-miRs) transcripts, whose processing generates mature miRs, were further investigated. As shown in Figure 3B, in both silenced and control hepatocytes undergoing EMT, the induction of pri-miR-181a1, pri-miR-181b1, and the repression of pri-miR-122 matched that of the respective miRs. The same modulation of the corresponding mature form, even if not significant, was observed for pri-miR-200a. These data indicate that SYNCRIP impacts the TGFβ-mediated regulation of specific miRs and, as its effect was detectable at the pri-miR level, suggest that this regulation may occur at the transcriptional level.   Of note, as shown in Figure 4A,B, the SYNCRIP influence on miR-181a1-3p, miR-181b1-3p, and miR-122-5p levels was further confirmed in BW1J cells. Interestingly, miR-let7g-5p, another anti-EMT regulator [49][50][51], as found modulated limitedly to hepatoma cells ( Figure 4A and data not shown). Coherently, in BW1J cells modulations of pri-miRs-181, pri-miR-122, and pri-miR-let7g were strictly correlated to that of the corresponding mature forms. Overall, these data indicate that SYNCRIP exhibits a function in miRNA regulation in both the induction (i.e., during EMT) and the maintenance (i.e., in transformed invasive cells) of the mesenchymal state of transdifferentiated hepatocytes.  pro-EMT microRNAs (miR-181a1-3p and miR-181b1-3p) in BW1J shSYN compared to control cells (empty). The values are calculated via the 2 (−∆Ct) method, normalized to the cel-miR-39, expressed as fold enrichment, and shown as mean ± SD. Statistically significant differences are reported for four independent experiments (* p < 0.05; ns = no significance). (B) qRT-PCR analysis of intracellular levels of the indicated pri-miRNAs in the same cells as in (A). The values are calculated via the 2 (−∆Ct) method, normalized to the 18S ribosomal RNA levels, and shown as mean ± SD. Statistically significant differences are reported for seven independent experiments (* p < 0.05; ** p < 0.01; ns = no significance).

Discussion
The main finding of this work was to ascribe a role to the hnRNP-Q, also known as SYN-CRIP, in the modulation of EMT/MET dynamics. This RNA-binding protein was found, indeed, as positively regulated in non-tumorigenic hepatocytes by TGFβ and, notably, its impairment prevented the full transdifferentiation. Moreover, SYNCRIP knockdown in HCC invasive cells allowed the rescue of a more differentiated phenotype by MET.
As pleiotropic regulators of gene expression, and often deregulated in epithelial cancers (reviewed in [11]), hnRNPs are conceivably implicated in mediating EMT reprogramming, but their role in this context remains largely unexplored. Here, we demonstrated for the first time, to our knowledge, a direct correlation between the function of SYNCRIP and the EMT outcome. Specifically, SYNCRIP-interfered cells undergoing TGFβ-induced EMT showed a limited Snail induction, in turn correlated to a minor downregulation of its main epithelial targets, E-cadherin and HNF1α, controlling hepatocyte differentiation and the maintenance of the epithelial phenotype [53,54]. These observations were corroborated by functional data in HCC invasive cells, where SYNCRIP knockdown induced MET and significantly impaired their migratory capacity. This evidence provides a possible molecular link between the well-known SYNCRIP overexpression in HCC [40] and the ability of cells to metastasize.
Interestingly, we observed that SYNCRIP knockdown, in TGFβ-treated hepatocytes as well as hepatoma cells, leads to a significant modulation of both the pri and mature forms of the specific miRNAs investigated, thus indicating the conceivable involvement of this hnRNP in their transcription. Even if, to our knowledge, there is no previous evidence of a possible function of SYNCRIP as a transcriptional regulator, the here provided observations suggest, for SYNCRIP, a role in the chromatin context that might extend the function of SYNCRIP in the regulation of miRNAs. It is previously known, indeed, that SYNCRIP mediates (i) their partition between intracellular and extracellular compartments by binding the hEXO motif (GGCU/A) [33,34]; (ii) the processing of let-7a by recognizing the UAGAAU sequence on the apical loop of the correspondent pri-miRNA [22].
Further studies are required to clarify whether the here suggested transcriptional trans activation of specific miRNAs by SYNCRIP can require the direct binding of this hnRNP to specific DNA sequences, or if it can be indirect, implying its association with multiprotein complexes of transcriptional regulators and/or epigenetic modifiers or a possible SYNCRIP-mediated control of their synthesis. Note that other hnRNPs, such as hnRNP-U and hnRNP-K, were shown to bind to chromosomal DNA [57,58]. Furthermore, hnRNP-U interacts with p300, thus controlling the hyperacetylation of histones [59], while hnRNP-K can recruit the chromatin remodeling enzyme histone methyltransferase [60] and acts as a transcriptional factor [12,13]. Moreover, given that the role of miRNAs in the fine-tuning of EMT/MET dynamics is not limited to the hepatocytes (e.g., miR-200 family regulation is involved in the progression of different epithelial tumors, including breast and colorectal cancer [46,61,62]), the impact of SYNCRIP in their regulation could also have relevance in other cell types.
Interestingly, SYNCRIP transcription, promoted by the lncNT5E, has a role in pancreatic cancer progression [39].
Further studies are required to dissect whether the function of SYNCRIP in modulating the EMT correlates to the acquisition of drug resistance by HCC cells. This in line with the role of other hnRNPs in chemoresistance [63][64][65] and the observation that the SYNCRIP target miR-200a-3p affects the drug resistance of Hep3B cells [66]. On the other hand, SYNCRIP could potentially be involved in molecular pathways known to be the target of pharmacological approaches [67].
In conclusion, we believe that the major conceptual advance implied by our results is that SYNCRIP represents a new modulator of EMT. Moreover, our data point to the involvement of this regulator in the TGFβ-mediated transcription of specific miRNAs. The further understanding of the mechanism by which SYNCRIP acts will hopefully be instrumental for research on hepatocyte reprogramming and tumorigenesis.

SYNCRIP Knockdown
Stable SYNCRIP knockdown was achieved through the infection of 3A cells with pSUPER retroviral constructs (Oligoengine, Seattle, WA, USA) expressing different shRNAs against SYNCRIP and a scrambled sequence as a control (as reported in [33]). Viral supernatants were collected 48 h after the transfection of 293 gp packaging cells, filtered, then added to cells. Selection was performed starting from 48 h post infection with 2 µg/mL puromycin for at least 1 week before successive analysis. To achieve the SYN-CRIP knock-down, BW1J cells were transfected with Lipofectamine LTX and Plus reagent (Invitrogen, San Diego, CA, USA) by using equal amounts of the pSUPER shSYNCRIP constructs. Analyses of RNAs and proteins were performed 48 and 72 h after transient transfection, respectively. Diluted cDNA samples were used for qPCR in a total volume of 10 µL using GoTaq qPCR Master Mix (Promega, Madison, WI, USA) and the reactions were carried out in a Bio-Rad-iQ-iCycler. Relative amounts of mRNAs and pri-miRNAs, obtained using the 2 (−∆Ct) method, were normalized with respect to the housekeeping gene 18S rRNA. Relative amounts of miRNAs, obtained using the 2 (−∆Ct) method, were normalized with respect to the synthetic cel-miR-39 (NORGEN, Thorold, ON, Canada). The same amount of cel-miR-39 was added to 150 ng of miRNAs of each sample before reverse transcription and was used as an internal control in Real-Time qPCR. Primer sequences are reported in Table 1.
Digital images were processed with Adobe Photoshop 7 software (Adobe Systems). The same enhanced color levels were applied for all channels.

Scratch Assay
BW1J cells, transfected with pSUPER shSYNCRIP constructs (Oligoengine, Seattle, WA, USA) or with empty vector (as above), were maintained in culture medium to reach 100% confluence, then shifted to serum-depleted culture medium to inhibit cell proliferation (as reported in [69]). A scratch wound was created on the cell layer using a micropipette tip, and micrographs were taken at 0.24 and 48 h after the scratch. Cell-devoid areas at times of 0.24 and 48 h after the scratch were quantified through the Fiji-ImageJ image processing package.

Statistical Analysis
For qRT-PCR, Western blotting, and scratch assay, statistical differences were assessed with the one-tailed paired Student's t-test using GraphPad Prism Version 6 (GraphPad Software). A p-value p < 0.05 was considered statistically significant (* p < 0.05; ** p < 0.01 and *** p < 0.001). Data were obtained from independent experiments and expressed as means ± s.e.m.