Impact of osteopontin on the development of non‐alcoholic liver disease and related hepatocellular carcinoma

Abstract Background & aims Osteopontin, a multifunctional protein and inflammatory cytokine, is overexpressed in adipose tissue and liver in obesity and contributes to the induction of adipose tissue inflammation and non‐alcoholic fatty liver (NAFL). Studies performed in both mice and humans also point to a potential role for OPN in malignant transformation and tumour growth. To fully understand the role of OPN on the development of NAFL‐derived hepatocellular carcinoma (HCC), we applied a non‐alcoholic steatohepatitis (NASH)‐HCC mouse model on osteopontin‐deficient (Spp1−/−) mice analysing time points of NASH, fibrosis and HCC compared to wild‐type mice. Methods Two‐day‐old wild‐type and Spp1−/− mice received a low‐dose streptozotocin injection in order to induce diabetes, and were fed a high‐fat diet starting from week 4. Different cohorts of mice of both genotypes were sacrificed at 8, 12 and 19 weeks of age to evaluate the NASH, fibrosis and HCC phenotypes respectively. Results Spp1−/− animals showed enhanced hepatic lipid accumulation and aggravated NASH, as also increased hepatocellular apoptosis and accelerated fibrosis. The worse steatotic and fibrotic phenotypes observed in Spp1−/− mice might be driven by enhanced hepatic fatty acid influx through CD36 overexpression and by a pathological accumulation of specific diacylglycerol species during NAFL. Lack of osteopontin lowered systemic inflammation, prevented HCC progression to less differentiated tumours and improved overall survival. Conclusions Lack of osteopontin dissociates NASH‐fibrosis severity from overall survival and HCC malignant transformation in NAFLD, and is therefore a putative therapeutic target only for advanced chronic liver disease.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is the most common form of liver cancer and the cause of approximately one million deaths yearly, with an alarming mortality rate of 94%. 1 Even though hepatitis B-(HBV) and C (HCV) virus infections are major risk factors for HCC, nutrition-related diseases strongly promote the increase in HCC prevalence worldwide. This may be due to a chain of events starting with obesity, metabolic syndrome, type-2 diabetes mellitus (T2DM) and fatty liver and potentially leading to non-alcoholic liver steatohepatitis (NASH), liver fibrosis and finally HCC.
Non-alcoholic fatty liver (NAFL) will likely drive the increase in the incidence of HCC in the next decades. 2 Approximately 20% of patients manifesting simple steatosis further develop non-alcoholic steatohepatitis. 3 NASH mainly occurs when the rate of liver non-esterified fatty acids (NEFA) uptake surpass its capacity for esterification into triglycerides (TG). 4,5 NASH, consequently, is a driver for the development of fibrosis/cirrhosis and finally HCC. 6 Hence, the study of hepatocarcinogenesis on a metabolic syndrome background is gaining significant interest on the individual as well as public health level. However, underlying molecular mechanisms are far from being understood.
Osteopontin (OPN; gene Spp1) is a multifunctional protein highly expressed in activated macrophages and T-cells, but also in hepatic stellate cells and hepatocytes. In obesity, OPN is vastly overexpressed in adipose tissue and induces infiltration and activation of macrophages generating a pro-inflammatory environment, which crucially contributes to the onset of insulin resistance. 7 Hepatic OPN expression is up-regulated in obesity 8 and various models of liver injury. 9,10 Furthermore, OPN is involved in the pathogenesis of NAFL associated with visceral obesity 11 and is a reliable biomarker for NASH/fibrosis in human non-alcoholic fatty liver disease (NAFLD). 12,13 Studies from others and our lab point to a pivotal role of OPN in obesity-driven nutrition-dependent diseases including high-fat diet-induced fatty liver [14][15][16][17] and thus suggest OPN as a treatment target.
Also downstream in the proposed order of events, OPN is highly upregulated in HCC and may even be evaluated as a potential therapeutic target in HCC. 18,19 However, a causal role of OPN in the pathogenesis of NASH and NASH-derived HCC is still not defined.To elucidate the impact of OPN on the sequential development of NASH and derived HCC as occurring in metabolic syndrome we took advantage of a novel mouse model, 20 which starting from hyperglycaemia recapitulates the development of NAFL, NASH, fibrosis up to HCC. Applying this model to wild-type (WT) and OPN-deficient (Spp1 −/− ) animals, we provide evidence for a Janus-type role of OPN in various states of NASH-HCC progression eventually resulting in enhanced NAFL, NASH and fibrosis but also more highly differentiated HCC and improved overall survival rate in Spp1 −/− mice.
Results: Spp1 −/− animals showed enhanced hepatic lipid accumulation and aggravated NASH, as also increased hepatocellular apoptosis and accelerated fibrosis. The worse steatotic and fibrotic phenotypes observed in Spp1 −/− mice might be driven by enhanced hepatic fatty acid influx through CD36 overexpression and by a pathologi-

| Animals
All experimental procedures were approved by the institutional animal care and use committees. Wild-type and Osteopontin knock out (Spp1 −/− , B6.129S6(Cg)-Spp1tm1Blh/J) mice on a C57BL/6J background were purchased from Charles River Laboratories Inc (Wilmington, Massachusetts, USA) and cohoused to minimize potential microbiome effects. Animals were treated as originally described by Fujii and colleagues 20 : two-day-old male newborns of both genotypes received a single subcutaneous STZ (Sigma, Missouri, USA) injection and were fed ad libitum a high-fat diet (HFD, 60 kcal%, D12492; Research Diets, New Jersey, USA) starting at 4 weeks of age. 14-hours fasting blood sugar (FBS) and body weight (BDW) were assessed in 4-week-old mice, and four FBS-matched cohorts per genotype were generated (n = 15). Of each group, a representative subgroup of eight animals (FBS-matched between genotypes) with lower alpha-diversity was used for molecular analyses.
Three-to-four littermate mice were housed together in wooden bedding-containing cages in the presence of cage enrichment in a light-controlled and temperature-controlled facility. At each time point, animals were overnight fasted (dark cycle) and later sacrificed by neck dislocation (light cycle). To assess in vivo proliferative potential of hepatocytes, all mice received an intraperitoneal injection of 5-Bromo-2′-deoxyuridine (BrdU) (Sigma) 2 hours prior to sacrifice.
Blood samples were drawn from the tail vein. Liver, subcutaneous and visceral white adipose tissue (SWAT, GWAT), kidney and small intestine samples were weighted, formalin-fixed and/or snap-frozen for further analyses.

| Computed tomography
Liver X-ray computed tomography (Siemens Inveon µCT, by Siemens Medical Solutions, Knoxville, USA) was used to non-invasively measure number and volume of liver tumours in isoflurane-anaesthetized mice. Before the µCT X-ray examination each mouse was administered 100µL of CT contrast medium intravenously (ExiTron™ nano 6000, Miltenyi Biotec GmbH, Germany) to improve the soft tissue contrast. During the examination, the mice were placed on a heated bed and the vital parameters (body temperature and respiratory rate) were constantly monitored and a protective eye ointment was applied. After completion of the study, mice were sacrificed by neck dislocation and tissue samples were harvested as described above.

| Histological analyses
Liver sections from the left lobe were formalin-fixed, paraffin-embedded and were stained with haematoxylin/eosin, or Sirius red to assess liver fibrosis. Stained sections were evaluated by an expert pathologist, blind to genotype and study period, following the WHO Classification of Tumours of the Digestive System. 21 Pancreas section was formalin-fixed, paraffin-embedded and stained with haematoxylin/eosin to quantitatively evaluate Langerhans islets. Apoptosis

| Plasma biochemistry
Blood glucose was measured using a One Touch Ultra glucose meter

| Liver biochemistry
TG concentration was enzymatically assessed using a commercially available kit (Sigma, Missouri, USA), after chloroform-methanol lipid extraction.

| Targeted lipidomics of diacylglycerols and ceramides
Lipids were extracted, purified and analysed from frozen samples, using lipid chromatography mass spectrometry as adapted from Kumashiro et al. 23

| Statistics
Data are presented as mean ± SEM, and two groups compared by unpaired Student t-test, with a significance level of <0.05. When more than two groups were compared, ANOVA with Tukey´s post hoc analyses were performed. For the survival observations, Kaplan-Meyer analysis was applied.

| Increased hepatic lipid uptake in NASH-HCC OPN-deficient mice promotes liver steatosis
Spp1 −/− and WT mice both developed comparable degrees of hyperglycaemia early after streptozotocin (STZ) treatment (plasma glucose 378 ± 16 mg/dL and 348 ± 20 mg/dL, respectively), indicating comparable efficiency of STZ administration. In line with that, the magnitude of STZ-induced Langerhans islets damage was also comparable between genotypes ( Figure S2B). Mice were sacrificed after 4 weeks of high fat diet (HFD) treatment to evaluate NAFL/NASH phenotype. Body weight as well of weights of GWAT and SWAT and plasma NEFA were similar ( Figure S1). As shown by haematoxylin and eosin-staining ( Figure 1A) and the NAS score ( Figure 1B) as well as hepatic TG quantification ( Figure 1C), there was a significantly increased lipid accumulation in livers of Spp1 −/− mice. Accordingly, plasma alanine aminotransferase (ALT) was markedly elevated in the Spp1 −/− genotype ( Figure 1D).
To evaluate the potential mechanisms of hepatic TG accumulation in Spp1 −/− animals, we measured the expression of genes playing a pivotal role in hepatic lipid homeostasis. Surprisingly, expression of Acaca was comparable between WT and Spp1 −/− mice, while

| Increased liver fibrosis in OPN-deficient mice
Hepatic fibrosis evolves from NAFL/NASH on a metabolic syndrome background. In line with the worse steatotic phenotype of Spp1 −/− livers, markedly increased liver collagen deposition and myofibroblast activation were observed in Spp1 −/− mice as shown by Sirius Red staining and α-SMA immunostaining respectively ( Figure 3A,B). While WT mice just showed mild-to-moderate perisinusoidal and zone-3 fibrosis, fibrosis was also extended to the portal and periportal area resulting in a higher fibrosis score in Figure 3C). Only these livers also showed well-defined bridging fibrosis and fibrotic septa ( Figure 3A). In addition, a significant increase in gene expression of pro-fibrogenic markers (Col1a1, Col 4a1, Timp1) was found in Spp1 −/− mice ( Figure 3D).
Of note, no changes in Tgfb expression were observed between

| Osteopontin deficiency prevents HCC dedifferentiation
Already at 12 weeks of age, which corresponds to the liver fibrotic stage, 87.5% of OPN-deficient animals developed liver tumours, which were histologically defined as HCCs (data not shown). In the WT group, just 1/8 of the mice showed HCCs at this time point ( Figure 3E).
Mice were finally analysed at 19 weeks of age in order to evaluate hepatocellular carcinomas. CT scans showed widespread liver tumours in both WT and Spp1 −/− mice, but no significant differences in tumour size and number were observed ( Figure 6A-C). However, a higher degree of tumour dedifferentiation in WT compared to Spp1 −/− animals was shown in histologic analyses, as indicated by a significantly higher tumour grade in livers of wild-type mice ( Figure 6D).  Figure S2 the HCC stage in WT livers ( Figure 7C). These data strongly suggest that the lack of OPN hampers HCC progression and dedifferentiation by modulating the hepatic inflammatory kinetics.

| Osteopontin deficiency reduces liverrelated mortality
As described earlier, NASH-HCC mice started to die after week 11, which corresponds to the liver pre-fibrotic/fibrotic stage. 30 In the present study, the mortality after 19 weeks in the WT group was 30% and significantly higher than in Spp1 −/− mice (14%; P = .0085) ( Figure 8A). All dropouts showed symptoms of hepatic toxicity, such as microvesicular steatosis ( Figure S3) indicating acute-on-chronic liver failure (ACLF) as described in patients with chronic liver disease. 31 Mortality in clinical ACLF correlates with the magnitude of systemic inflammation, 31 which was significantly enhanced also in NASH-HCC-WT mice as shown by increased serum amyloid P (SAP) levels ( Figure 8B). Hence, OPN promotes a systemic pro-inflammatory milieu, which significantly reduces survival in non-alcoholic fatty liver disease.  shown that lack of OPN protects from HFD-induced hepatic steatosis, due to the preservation of adipose tissue function in obesity, hence preventing ectopic lipid accumulation in the liver. 16,17 In contrast, the model used here is primarily hyperglycaemic but not obese hence revealing the impact of hyperglycaemia in this process while excluding a potential involvement of a dysfunctioning adipose tissue.

| D ISCUSS I ON
Several experimental and clinical studies also indicate that OPN is involved in liver fibrogenesis. 40,41 However, previous research did not use models that replicate clinicopathological features of NAFL-related fibrosis. In our model, hepatic fibrogenesis evolves from NAFL/NASH on a diabetes background. The lack of OPN not only worsened the steatotic and inflammatory phenotypes, but also enhanced hepatic collagen deposition in NASH-HCC-Spp1 −/− mice. We also provide ev- Since hepatocellular glycogen deposition is the principal clinical manifestation of glycogenic hepatopathy in humans, 49 we further investigated glycaemic control in vivo, which was improved in OPN-deficient animals. Moreover the fact that fasting blood sugar showed significant improvement only after four weeks of HFD and that the amount of intact Langerhans islets was comparable between genotypes before the HFD challenge suggests that OPN deficiency did not interfere with initial STZ treatment but improved beta-cell survival or recovery thereafter, maybe by reduced lipotoxicity. 50

D ECL A R ATI O N O F I NTER E S T
The authors declare no competing interests.