The pituitary tumour‐transforming gene 1/delta‐like homologue 1 pathway plays a key role in liver fibrogenesis

Abstract Background and Aims PTTG1 is almost undetectable in adult livers but is highly expressed in hepatocarcinoma. While little is known about its involvement in liver fibrosis, PTTG1 expression is associated with DLK1. We assessed the role of the PTTG1/DLK1 pathway in fibrosis progression and the potential therapeutic effect of PTTG1 silencing in fibrosis. Methods Pttg1 and Dlk1 were studied in liver and isolated cell populations of control and fibrotic rats and in human liver biopsies. The fibrotic molecular signature was analysed in Pttg1 −/− and Pttg1 +/+ fibrotic mice. Finally, Pttg1 silencing was evaluated in rats as a novel antifibrotic therapy. Results Pttg1 and Dlk1 mRNA selectively increased in fibrotic rats paralleling fibrosis progression. Serum DLK1 concentrations correlated with hepatic collagen content and systemic and portal haemodynamics. Human cirrhotic livers showed greater PTTG1 and DLK1 transcript abundance than non‐cirrhotic, and reduced collagen was observed in Pttg1 Pttg1 −/− mice. The liver fibrotic molecular signature revealed lower expression of genes related to extracellular matrix remodelling including Mmp8 and 9 and Timp4 and greater eotaxin and Mmp13 than fibrotic Pttg1 +/+ mice. Finally, interfering Pttg1 resulted in reduced liver fibrotic area, lower α‐Sma and decreased portal pressure than fibrotic animals. Furthermore, Pttg1 silencing decreased the transcription of Dlk1, collagens I and III, Pdgfrβ, Tgfrβ, Timp1, Timp2 and Mmp2. Conclusions Pttg1/Dlk1 are selectively overexpressed in the cirrhotic liver and participate in ECM turnover regulation. Pttg1 disruption decreases Dlk1 transcription and attenuates collagen deposition. PTTG1/DLK1 signalling is a novel pathway for targeting the progression of liver fibrosis.

L ay Summar y PTTG1 and DLK1 transcription are increased in rats and patients with hepatic cirrhosis. PTTG1 is involved in fibrotic extracellular matrix remodeling and its silencing decreases portal hypertension and alliviates fibrosis progresion.

| INTRODUC TI ON
Cirrhosis is a major determinant of morbidity and mortality and predisposes to hepatic failure and liver cancer. Halting the progression of fibrosis to cirrhosis is considered as a foremost goal in patients with liver disease. Anti-inflammatory agents, arresting hepatic stellate cells (HSC) activation substances, renin-angiotensin system inhibitors, cannabinoid receptor antagonists, hepatoprotective peptides, transforming growth factorβ (TGFβ) or platelet-derived growth factor (PDGF) antagonists and chemokine receptor antagonists are among the numerous candidates assessed to limit or reverse liver fibrogenesis. [1][2][3][4] However, most of these compounds have shown limited efficacy and/or adverse side effects, and consequently, an antifibrogenic pharmacological treatment for liver fibrosis is currently lacking.
The pituitary tumour-transforming gene (PTTG1) is the index mammalian securin. 5 PTTG1 is overexpressed in a variety of cell lines including hepatocellular carcinoma (HCC). 6 It encodes a multifunctional protein involved in the regulation of faithful chromatid segregation during mitosis, DNA repair, apoptosis, metabolism and gene transcription. 7 Interestingly, PTTG1 modulates extracellular matrix (ECM) turnover regulating several matrix metalloproteinases (MMPs). 8,9 Despite overexpression of PTTG1 in liver biopsies from patients with HCC, very little data are available on its expression in preneoplastic conditions such as advanced liver fibrosis and cirrhosis. This is particularly striking as several factors induce PTTG1 expression, including estrogens, fibroblast growth factor, insulin, insulin growth factor-1 and hepatocyte growth factor 7,10 all increased under conditions of chronic liver injury. [11][12][13][14] Moreover, microenvironmental hypoxia occurring in damaged hepatic tissue could also regulate PTTG1 expression through the hypoxia-inducible factor 1. 15 PTTG1 also acts to regulate growth factors, angiogenesis and exhibits transforming activity in vitro and in vivo. 16,17 Furthermore, hepatic PTTG1 expression is upregulated after partial hepatectomy and has been proposed as a new marker of proliferation in liver regeneration. 18 These findings support the exploration of whether PTTG1 could contribute to the activation of fibroproliferative processes in liver disease. In addition, delta-like homologue 1 (DLK1) was identified as one of the most abundantly expressed PTTG1 targets. 19 The DLK1 gene encodes a single-pass transmembrane protein that belongs to a family of epidermal growth factor (EGF) repeat-containing proteins. 20 DLK1 is a non-canonical ligand of Notch receptors that mediate a metabolic shift from lipid storage to peripheral lipid oxidation in adipocytes, participate in differentiation processes and behave as a growth factor. 21 It consists of six EGFlike tandem repeats, a juxtamembrane region with a tumour necrosis factor-alpha converting enzyme (TACE)-mediated cleavage site, a transmembrane domain and a short intracellular tail. 22 19 Given this background, we aimed to explore the hypothesis that PTTG1/DLK1 signalling should play a central role in the activation of the fibrogenic process in liver disease.

| Induction of hepatic cirrhosis in rats
This study was performed in control (n = 32) and male Wistar rats with different degrees of fibrosis (n = 77) (Charles-River, Saint Aubin Les Elseuf, France). Fibrosis was induced by repetitive carbon tetrachloride (CCl 4 ) inhalation. 23 The rats were fed ad libitum with standard chow and water containing phenobarbital (0.3 g/L), as drinking fluid. Animals were exposed to a CCl 4 atmosphere twice a week, starting with 0.5 minutes for three sessions. Afterwards, the duration was increased to 1, 2, 3, 4 and 5 minutes until the end of the investigation. To induce variable degrees of hepatic fibrosis CCl 4treated rats were studied at the 8th, 13th, 16th and 19th week after starting the fibrosis induction protocol. Control rats were studied following similar periods of phenobarbital administration. When scheduled, animals were anaesthetised and a haemodynamic study was performed. Afterwards, a blood sample was obtained and animals were sacrificed by isoflurane overdose (Forane, Abbott Laboratories S.A., Madrid, Spain). Organ samples were snap-frozen or fixed in 10% buffered formalin.

| Induction of fibrosis in mice
This study was performed in fibrotic and control male Pttg1 wild-type (Pttg1 +/+ ) and knock out (Pttg1 −/− ) mice. Pttg1 −/− mice with C57BL/6 genetic background were provided by Dr Shlomo Melmed and the origin of these mice has been described previously. 24

| In vivo Pttg1 interference
A group of fibrotic rats randomly received i.v. Pttg1 small-interfering RNA (siRNA, assay ID s133880, 0.25 mg/kg/dose bwt, n = 6) or scrambled siRNA (Ambion in vivo negative control No. 1, n = 6) as the negative control (C − siRNA) every 10 days from the 9th to the 13th week after starting the fibrosis induction protocol. In vivo transfection was performed using Invivofectamine 3.0 kit (Invitrogen, Life Technologies Corporation, Carlsbad, CA, USA) following the manufacturer's instructions. Six control rats were also included. Rats were studied in the 14th week.

| Fibrosis quantification and staging
The liver of rats treated with CCl 4 showed macroscopic finely granulated surfaces. According to the time of CCl 4 exposure, we observed progressive ECM accumulation, evolving from a light deposition, mainly in the portal area, to numerous and thicker septa in those animals submitted to longer CCl 4 exposure periods. Most animals exposed to the toxin for the longest periods of time developed cirrhosis. Consequently, rats were staged according to the percentage of fibrotic area with respect to the total area of the liver biopsy: mild and moderate fibrosis was defined when the percentage of fibrotic area <6% (n = 6), severe fibrosis 6%-11%, (n = 8) and cirrhosis >11% (n = 11). Control rats (n = 13) displayed no appreciable alterations in liver histology. Figure 1A shows representative Sirius red staining from a control liver, a liver with mild/moderate fibrosis, a liver with severe fibrosis and a cirrhotic liver. Fibrotic/cirrhotic rats had important alterations in liver function tests, which were more pronounced in cirrhosis (Table S1).

| Hepatic Pttg1 and Dlk1 mRNAs parallel the intensity of liver fibrosis and selectively occurs in this organ
Progression of liver fibrosis was associated with a concomitant increase in Pttg1 mRNA expression ( Figure 1B). Pttg1 expression significantly increased in rats with severe fibrosis and reached maximum levels in rats with cirrhosis. Interestingly, Pttg1 transcript was selectively detected in the liver of cirrhotic animals, but not in the spleen, lungs, kidneys, heart, aorta or brain ( Figure 1C). Pttg1 mRNA abundance was also assessed by droplet digital PCR (ddPCR). Results were in line with those obtained in real-time PCR (RT-PCR) experiments. The liver of cirrhotic rats showed a much higher abundance of Pttg1 transcripts (411 ± 67 copies ̸ μl) than that found in control livers (20 ± 2 copies ̸ μl, P ˂ 0.001). In contrast, with the exception of heart (7 ± 1 vs 28 ± 2 copies/μl, P ˂ 0.05) no differences were found between spleen (614 ± 94 vs 947 ± 172 copies/μl), kidney (27 ± 4 vs 38 ± 5 copies ̸ μL), lung (74 ± 8 vs 87 ± 5 copies/μl) aorta (4 ± 1 vs 27 ± 12 copies ̸ μL) and brain (17 ± 3 vs 21 ± 1 copies ̸ μl) of cirrhotic and control rats. The pattern expression of Pttg1 was paralleled by a similar profile for Dlk1 mRNA ( Figure 1B). Dlk1 mRNA abundance progressively increased, the lowest levels observed in rats with mild/ moderate fibrosis, the highest in cirrhotic rats. Indeed, Dlk1 activation was selectively detected in the cirrhotic liver but not in other assessed organs ( Figure 1C).

| Pttg1 and Dlk1 are mainly expressed in hepatic parenchymal tissue
To identify the cellular source of altered expression of both Pttg1 and Dlk1 in hepatic tissue, we isolated primary cells from the liver of cirrhotic and control rats. Both, Pttg1 and Dlk1 exhibited low or almost negligible mRNA expression in different control cell types ( Figure 2A). By contrast, marked Pttg1 mRNA abundance was F I G U R E 1 Expression of Pttg1 and Dlk1 in experimental liver fibrosis. A, Staging of CCl 4 -treated rats based on liver-collagen content. Sirius red staining of representative liver sections (×100). B, Hepatic Pttg1 and Dlk1 mRNA of control (n = 13) and CCl 4 -treated rats with mild/moderate fibrosis (n = 6), severe fibrosis (n = 8) and cirrhosis (n = 11). C, Pttg1 and Dlk1 mRNA in organs from control (n = 5) and cirrhotic rats (n = 5). Results are expressed as mean ± SE. *P < 0.05, **P < 0.01, ***P < 0.001 vs control; ## P < 0.01 vs cirrhosis. One-way ANOVA with Newman-Keuls post hoc test or Kruskal-Wallis test with Dunn post hoc test observed in three types of liver cells isolated in cirrhotic rats, the highest abundance being found in HSC ( Figure 2A). This was paralleled by striking activation of Dlk1 mRNA but was largely observed in hepatocytes (HEP) (Figure 2A). In an attempt to further delineate the relative contribution of HEP and HSC to the acute increase in Pttg1 and Dlk1 in cirrhotic liver, we next measured the absolute concentration of these transcripts in the isolated cells. In line with the RT-PCR results, the absolute Pttg1 mRNA values were similar in both types of cells (HEP: 125 ± 7 copies ̸ μl, HSC: 131 ± 3 copies ̸ μl), whereas Dlk1 mRNA values were lower in HSC (80 ± 11 copies ̸ μl) than in HEP (154 ± 9 copies ̸ μl). Overactivation of the PTTG1/DLK1 axis in human cirrhosis was further confirmed. Paralleling the increased abundance of collagen I alpha 1 (COL1α1) messenger, higher expression of both PTTG1 and DLK1 mRNA was observed in samples derived from cirrhotic patients in comparison to non-cirrhotic biopsies. Next, we performed histological immunolocalization of PTTG1 and DLK1 in the liver of cirrhotic and control rats. Both proteins were almost undetectable in control samples. However, in cirrhotic livers, they were clearly identified either in the parenchymal area or close to the portal tracts and fibrous septa ( Figure 2C). Additionally, DLK1 expression clearly differs from that of other well established profibrogenic substances, since DLK1 hepatic protein content only exhibited a clear relationship with fibrosis intensity ( Figure 2D).  . Results are expressed as mean ± S.E. *P < 0.05, **P < 0.01 vs control, † †P < 0.01 vs cirrhotic HEP; # P < 0.05, ## P 0.01 vs cirrhotic EC. !P < 0.05 vs control HEP. B, COL1α1, PTTG1 and DLK1 mRNA in liver from cirrhotic (n = 12) and non-cirrhotic patients (n = 7). *P < 0.05, ***P < 0.001 vs control. C, Immunnolocalization of PTTG1 and DLK1 in rat control and cirrhotic liver (200×). D, Western blots for rat hepatic DLK1, TGFβ, TNFα and angiotensin II (AII) MAP inversely correlated with serum DLK1 in CCl 4 -treated rats (r = −0.69, P < 0.001) ( Figure 3C). Furthermore, serum concentration of DLK1 also depicted a direct relationship with the degree of portal hypertension in fibrotic/cirrhotic animals (r = 0.41, P < 0.05) ( Figure 3D).  Figure 4C). We also observed a significant increase in Ccl11 and Mmp13. The former encodes eotaxin a chemokine that has been described to be upregulated in senescent HSC 26 whereas the latter encodes for a metalloprotease involved in the degradation of a fibrotic liver matrix. 27

| Assessment of Pttg1 siRNA in cultured rat hepatocytes and fibrotic rats
To investigate the efficacy and duration of gene silencing in cultured rat hepatocytes, we transfected CC-1 cells. Following treatment, F I G U R E 3 DLK1 serum levels in rats with experimental fibrosis. A, Serum concentrations of DLK1 in control (n = 13), mild/moderate fibrosis (n = 6), severe fibrosis (n = 8) and cirrhotic (n = 11) rats. Each point represents single DLK1 value in serum. Horizontal lines indicate the mean value for each group. ***P < 0.001 vs control; ### P < 0.001 vs cirrhotic. Oneway ANOVA with Newman-Keuls post hoc test. Correlation of DLK1 serum levels with (B) histological quantification of liver fibrosis (r = 0.74; P < 0.001); C mean arterial pressure (r = −0.69; P < 0.001; and D, portal pressure (r = 0.43; P < 0.01) in control (n = 13) and CCl 4 -treated (n = 25) rats. Pearson two-tailed test Pttg1 mRNA was significantly lower than that in the siRNA C − group at 24, 48 and 72 hours ( Figure 5A). These results indicate that Pttg1 siRNA effectively suppresses Pttg1 expression in rat-cultured hepatocytes. Pttg1 siRNA was also effective at silencing the enhanced expression of Pttg1 mRNA in fibrotic rats ( Figure 5B). In fact, whereas fibrotic rats treated with scrambled siRNA displayed approximately 15 times higher levels of Pttg1 mRNA than control animals, an abundance of this transcript in the liver of fibrotic rats receiving Pttg1 siRNA was not different from that found in healthy animals. In addition to silencing hepatic Pttg1 mRNA, administration of Pttg1 siRNA also inhibited hepatic Dlk1 mRNA expression in fibrotic rats ( Figure 5B).

| Effect of Pttg1 siRNA on liver histology, portal pressure and profibrogenic genes in fibrotic rats
Fibrotic rats treated with C − siRNA showed initial stages of the characteristic pattern of perivenular and periportal deposition of connecting tissue with development of portal-to-portal septa and evidence of architectural distortion resulting in micronodular fibrosis ( Figure 6A). However, biopsies obtained from fibrotic rats treated with Pttg1 siRNA displayed less remarkable architectural alterations, with thinner septa, and more preserved hepatic parenchyma. This was confirmed by morphometric analysis of Sirius red-stained sections ( Figure 6B). This abrogation of fibrosis was consistently observed in all animals exposed to Pttg1 silencing. Similar results were found when staining alpha 2 smooth muscle actin (α-SMA). We detected α-SMA as linear staining in the portal tracts and fibrous septa of both groups of fibrotic rats ( Figure 6A). Staining was more diffuse in rats receiving Pttg1 siRNA than in C − siRNA. In line, Pttg1 siRNA also showed significantly reduced portal hypertension than fibrotic animals receiving C − siRNA ( Figure 6B). Furthermore, Pttg1-silenced animals significantly decreased hepatic mRNA expression of Tnfα compared to fibrotic rats.
As anticipated, Col1α2 and Col3α1 mRNA was significantly increased in fibrotic rats treated with C − siRNA. Consistently, we also observed activation of key genes involved in profibrogenic mechanisms, such as Tgfßr1 and Pdgfrβ ( Figure 6C F I G U R E 4 Hepatic fibrosis in Pttg1 −/− mice. A, Sirius red staining from healthy and fibrotic Pttg1 +/+ and Pttg1 −/− mice (100×). B, Fiber content in Pttg1 +/+ (n = 4) and Pttg1 -/-(n = 7) mice. ***P ≤ 0.001 vs Pttg1 +/+ , unpaired t-test. C, Volcano plot of the differentially expressed genes in a pair-wise comparison of Pttg1 +/+ (n = 4) and Pttg1 -/-(n = 4) mice. Significance was set to a P value based on a Student's t-test of 0.05 (−log10 [P-value] ≥1.30), the biological cut-off was set to a fold regulation of ±1.5 fold (−1 ≥log1.5 [FC of fibrotic Pttg1 KO/fibrotic WT] ≥1). According with these two criteria, the top 12 differentially expressed genes are labeled with their corresponding gene ID. Insignificant (black), statistically but not biologically significant downregulated (grey), biologically but not statistically downregulated (blue) and upregulated (pink), and both biologically and statistically significant downregulated (green) up regulated (yellow) genes in fibrotic Pttg1 +/+ mice 3.9 | Effect of Pttg1 siRNA on serum markers of liver function Pttg1 siRNA treatment was associated with a tendency towards normalization of most systemic indicators of liver function (Table 2).
Actually, aspartate aminotransferase, lactate dehydrogenase, gamma-glutamyl transferase, total bilirubin and triglycerides were found to be near normal. Overall, these results support the protective effects on hepatic function resulting from Pttg1 mRNA silencing in rats with experimental fibrosis.

| DISCUSS ION
This investigation aimed to explore whether PTTG1/DLK1 signalling contributes to the activation of the fibroproliferative process in liver disease. In agreement with previous studies, Pttg1 mRNA was almost undetectable in healthy animals. In contrast, Pttg1 mRNA levels were markedly overexpressed in the liver of rats with hepatic fibrosis, reaching maximal abundance in cirrhotic rats. Moreover, only hepatic tissue of cirrhotic rats showed a significantly increased abundance of Pttg1 mRNA with respect to control animals.
Paralleling Pttg1 results, Dlk1 mRNA expression markedly increased in fibrotic rats, with a close correlation with collagen deposition.   Administration of Pttg1 siRNA to fibrotic rats also affects the regulation of ECM remodelling. In experimental and human cirrhosis, fibrosis appears to be the result of not only excessive ECM synthesis but also reduced degradation. 37 38 The present study shows that Mmp2 F I G U R E 6 Effect of Pttg1 blockade on fibrosis progression. Control (n = 6) and fibrotic rats receiving C -siRNA (n =6) or Pttg1 siRNA (n = 6). A, Sirius red and α-SMA stainings (100×). B, Quantitative measurement of relative fibrosis and α-SMA positive area, portal pressure and Tnfα mRNA. C, Hepatic messenger expression of Col1α2, Col3α1, Tgfβr1 and Pdgfrβ. D, Hepatic Mmp2, Mmp9, Timp1 and Timp2 mRNA expression. Results are given as mean ± SE. *P < 0.05, **P < 0.01, ***P < 0.001 vs control; # P < 0.05, ## P < 0.01 vs C -siRNA treated rats. One-way ANOVA with Newman-Keuls post hoc test or the Kruskal-Wallis test with Dunn post hoc test expression increases in liver fibrosis, however, Pttg1 interference may have an antifibrogenic effect also by reducing Mmp2 expression and, consequently, by blocking degradation of normal perisinusoidal matrix and promoting activation of quiescent HSC. 42 In this study, we also observed that Pttg1 blocking reduced TIMPs expression. Timp1 and Timp2 are mainly expressed in activated HSC, thus, Timp1 and Timp2 expression could be reduced as a result of diminished HSC activation in these animals. TIMPs also stimulate fibroblast proliferation. 43 Thus, Timp1 and Timp2 downregulation could also contribute to decreased proliferation of activated HSC. A graphical model summarizing the proposed mechanism underlying PTTG1-induced promotion of liver fibrosis is provided in the supplementary information section.
In conclusion, this investigation shows that serial administration of Pttg1 siRNA exerts antifibrotic effects when administered during induction of hepatic damage. Pttg1 gene silencing normalizes expression of Dlk1, arrests activation of HSC, diminishes expression of ECMrelated genes and finally decreases hepatic collagen deposition and reduces portal hypertension. Thus, the PTTG1/DLK1 axis may represent a valuable target for the prevention and treatment of liver fibrosis.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests. Dr Bruix consults for, advises, and is on the speakers' bureau and received grants from

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.