Aberrant hepatic lipid storage and metabolism in canine portosystemic shunts

Non-alcoholic fatty liver disease (NAFLD) is a poorly understood multifactorial pandemic disorder. One of the hallmarks of NAFLD, hepatic steatosis, is a common feature in canine congenital portosystemic shunts. The aim of this study was to gain detailed insight into the pathogenesis of steatosis in this large animal model. Hepatic lipid accumulation, gene-expression analysis and HPLC-MS of neutral lipids and phospholipids in extrahepatic (EHPSS) and intrahepatic portosystemic shunts (IHPSS) was compared to healthy control dogs. Liver organoids of diseased dogs and healthy control dogs were incubated with palmitic- and oleic-acid, and lipid accumulation was quantified using LD540. In histological slides of shunt livers, a 12-fold increase of lipid content was detected compared to the control dogs (EHPSS P<0.01; IHPSS P = 0.042). Involvement of lipid-related genes to steatosis in portosystemic shunting was corroborated using gene-expression profiling. Lipid analysis demonstrated different triglyceride composition and a shift towards short chain and omega-3 fatty acids in shunt versus healthy dogs, with no difference in lipid species composition between shunt types. All organoids showed a similar increase in triacylglycerols after free fatty acids enrichment. This study demonstrates that steatosis is probably secondary to canine portosystemic shunts. Unravelling the pathogenesis of this hepatic steatosis might contribute to a better understanding of steatosis in NAFLD.


Introduction
Non-alcoholic fatty liver disease (NAFLD) is the most common liver disorder in men with an estimated prevalence ranging 25% up to 45% worldwide [1]. NAFLD includes related disorders from the earliest stage hepatic steatosis, to the more progressive stage non-alcoholic steatohepatitis, of which the latter can progress to cirrhosis and hepatic cellular carcinoma [1,2]. The pathophysiology, however, is still poorly understood and NAFLD is associated with cardiovascular disease, diabetes mellitus type 2, and chronic kidney disease [2][3][4]. Although murine models resemble monogenic forms of NAFLD [5], these diseases in mice are often incapable of fully mimicking the multifactorial nature of human NAFLD.
Histological evaluation of hepatic biopsies after surgical attenuation of the shunt revealed a decrease in steatosis, suggesting steatosis in CPSS could be induced by hepatic hypoxia or a disturbed fatty acid metabolism [14]. Steatosis in CPSS dogs could be explained by a genetically determined factor [7] or by altered metabolism secondary to disease processes and the resulting hepatic injury [12,14].
This study was performed to evaluate steatosis in canine congenital portosystemic shunting. As steatosis is observed histologically in both shunt types [11][12][13][14][15], we expect that hepatic steatosis occurs secondary to portosystemic shunting. In-depth analysis of the lipid metabolism of dogs with CPSS with gene-and lipid-profiling combined with organoid disease modelling will give insight in the pathogenesis of primary or secondary hepatic steatosis. This in-depth analysis might serve as a model for human steatosis as observed in NAFLD and lead to novel treatment methods for steatosis in human and veterinary medicine.

Animals and samples
Liver material was obtained from privately owned dogs with portosystemic shunts, referred to the University Clinic for Companion Animals (Department of Clinical Sciences of Companion Animals, Utrecht University). Permission was obtained from the dog owners using informed consent. CPSS was diagnosed based on increased fasting plasma ammonia levels (reference values 15-45 μmol/L) [16,17], ultrasound visualization and classification of the shunt, and finally confirmed during surgery. Fresh wedge liver biopsies were taken during the surgical attenuation of the shunt [18]. Liver tissue from healthy dogs was used as a control in this study. These dogs were euthanized for other unrelated research, data were collected according to the Act on Veterinary Practice and the procedure was approved by the local ethics committee (DEC Utrecht), as required under Dutch legislation (ID 2007.III.08.110). Liver tissue was obtained as surplus material (University 3R policy). The absence of an underlying liver disease was confirmed histologically by a board certified veterinary pathologist. The analysis of lipid accumulation by Oil-red-O staining, mRNA expression using quantitative reversed transcriptase PCR (RT-qPCR) and the profiling of neutral lipids and phospholipids was performed on overlapping hepatic tissue of dogs with a shunt (EHPSS n = 7 and IHPSS n = 5) and compared to healthy control dogs (n = 4). For the microarray analysis a cohort of 49 samples (EHPSS n = 32, IHPSS n = 15, and control n = 2) was used. Nine cases for both IHPSS and EHPSS from this analysis were replicated in qPCR and the sample set was supplemented with 46 additional samples (EHPSS n = 19, IHPSS n = 14, and control n = 13). Hepatic tissue of 12 dogs (EHPSS, IHPSS, and healthy controls; n = 4 per group) was used for organoid culture. An overview of the sample use and overlapping samples is given in supplementary data (S1 Fig). For Oil-red-O staining, hepatic biopsies were placed in a Tissue-Tek1 cryo-molds filled with O.C.T. Compound (Sakura Finetek Europe B.V., Alphen aan den Rijn, The Netherlands) and frozen in liquid nitrogen until use. For RNA isolation liver samples were snap frozen in liquid nitrogen.

Oil-red-O staining
Oil-red-O staining was performed as previously described [15]. The frozen samples were cut into 8-μm sections and stained for lipids using a standard Oil-red-O (Klinipath, Duiven, The Netherlands) protocol with haematoxylin counterstaining. All Oil-red-O stained sections were evaluated blind and at random by a board-certified pathologist (GCMG), using a semi-quantitative scoring system of lipid accumulation based on lipid intensity of the stainings. Intensity grading ranged from low to high using a scale from 0 to 4. For lipid intensity analysis, slides were scanned at 20× magnification as described previously [19]. Images were extracted using Aperio ImageScope v12.0.0.5039 (Aperio, Vista, CA, USA) as a TIFF file with jpeg compression. The images were resized to 10% of their original size for digital analysis. Data of ten random snapshots were collected. The RGB data of the images was converted to a 2-bit black/ white image based on thresholding the color of the dye using ImageJ (NIH; http://rsb.info.nih. gov/ij/) software. An average of the black:white ratio was calculated to determine lipid intensity scores.

Expression profiling
Previously published microarray expression data on IHPSS and EHPSS liver tissue [20], available through GEO Series accession number GSE39005 (http://www.ncbi.nlm.nih.gov/geo/ query/acc.cgi?-acc=GSE39005), was used to determine the 20 most up and 20 most down regulated genes in both EHPSS and IHPSS versus control dogs. A corresponding list of lipid related genes for IHPSS and EHPSS was selected for further confirmation. Genes with log2fold changes of more than 1.1 or less than -1.5 were selected to ensure that only robust differences were considered. Involvement of these genes in lipid metabolism, transport, or storage was determined based on Gene Ontology biological processes and literature.
Gene expression differences of 11 selected genes was confirmed using RT-qPCR on available cDNA obtained using the iScript™ cDNA synthesis kit as described by the manufacturers protocol (Bio-Rad, Veenendaal, The Netherlands). Primer design, validation, RT-qPCR conditions, and data analysis were performed as described previously [20]. Normalization was performed using four reference-genes; glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ribosomal protein S5 (RPS5), heterogeneous nuclear ribonucleoprotein H (HNRPH), and ribosomal protein S19 (RPS19) [20] as required under MIQE-precise [21]. Details of the primers are listed in S1 Table. Analysis of neutral-and phospholipids by high-performance liquid chromatography-mass spectrometry (HPLC-MS) Lipids were isolated from frozen tissue by the method of Bligh and Dyer [22] and separated in a neutral lipids and phospholipid fraction on a freshly prepared silica-G column (approximately 10 mg of 0.063-0.200 mm silica (Sigma-Aldrich, St Louis, MO, USA) [23]. Neutral lipids were eluted with two volumes acetone, dried under nitrogen gas and stored at −20˚C. Just before HPLC-MS analysis, the neutral lipid fraction was reconstituted in methanol:chloroform (1:1 v/v) and separated on a Kinetex/HALO C8-e column (2.6 μm, 150 × 3.00 mm; Phenomenex, Torrance, CA, USA). A gradient was generated from methanol:H 2 O (5:5 v/v) and methanol:isopropanol (8:2 v/v) at a constant flow rate of 0.3 ml/min. Mass spectrometry of neutral lipids (triacylglycerols (TAGs) and cholesterol) was performed using positive mode Atmospheric Pressure Chemical Ionization (APCI) on a LTQ-XL mass spectrometer (Thermo, Waltham, MA, USA). Separation of phospholipid classes was performed as described elsewhere [24].

Flow cytometry analysis
After a 24 h incubation with FFA, organoids were collected from the Matrigel with cold advanced DMEM/F12 (Gibco), and subsequently trypsinised with 10x Trypsin (Gibco) containing 0.5 mg/ml DNAse (Sigma). Advanced DMEM/F12 with 10% FCS was added and the cell suspension was spun at 250 g for 5 min at 4˚C. Pellets were resuspended in advanced DMEM/F12. Organoids were incubated with 5 μg/ml LD540 (lipophilic dye, kindly provided by prof. Christoph Thiele, Bonn, Germany) for microscopic imaging of lipid droplets [27] in DMEM medium containing 20 μg/ml fatty acid free BSA, 10μg/ml HEPES (Gibco), and 10 μg/ml Glutamax (Gibco) for 30 min in a water bath at 37˚C. Incubations without LD540 in FFA medium served as a control. Cells were washed twice with HBSS (Gibco) and cells were resuspended in HBSS containing 20 μg/ml fatty acid free BSA, 1 μg/ml HEPES and 1 μg/ml Sytox Red (Molecular probes, Thermo Fisher, Bleiswijk, The Netherlands). Cell analysis was performed on a 488-laser LSRFortessa flow cytometer (Becton Dickinson, Erembodegem, Belgium). A 540/30 nm bandpass filter was installed to measure the optimum of the LD540 emission peak. Fluorescently labelled beads (CS&T beads, Becton Dickinson) were used to check the performance and verify optical path and stream flow of the flow cytometer. Dead cells were excluded with Sytox red using a 635 nm laser with an emission spectrum of 670/30 nm.

Whole mount imaging
For whole mount fluorescent staining canine liver organoids were carefully harvested from Matrigel and fixed in 10% v/v neutral buffered formalin (Klinipath) for 45 min on ice. Fixed organoids were incubated in 0.025 μg/μL LD540 in PBS for 1 h at room temperature. After washing, nuclei were stained with DAPI and organoids were mounted with ProLong Diamond Antifade mounting medium (Life Technologies, Thermo Fisher) and imaged using a confocal microscope (Leica SPE-II).

Statistical analysis
Oil-red-O differences in scoring were evaluated using a Student T-test. P values < 0.05 were considered significant. The results of the microarray analysis were reanalyzed with updated annotations using ANOVA (R version 2.2.1/MAANOVA version 0.98) [28]. Correction for multiple testing (Permutation F2-test using 5,000 permutations) was performed and P < 0.05 was considered statistically significant. In RT-qPCR log-values of normalized relative expression were used to obtain a normal distribution. A Levene's test was used to determine if the data was normally distributed. A Kruskal-Wallis test was performed to observe differences between the EHPSS, IHPSS, and control group and was performed in case of multiple group testing. Any observed differences were confirmed by a Mann-Whitney U test on independent samples. Statistical significance was obtained if P < 0.01. Processing of the LC-MS data of neutral lipids and phospholipids was performed with XCMS under R version 3.0.2 [29,30]. Principal component analysis (PCA) was performed with the R package 'PCAMethods' using the nonlinear iterative partial least squares (nipals) algorithm with pareto scaling [31]. Differences in lipid accumulation after free fatty acid incubation between the EHPSS, IHPSS, and wildtype organoids measured using flow cytometry analysis, were calculated with a Kruskall-Wallis test. Observed differences were confirmed by a Mann-Whitney U test on independent samples.

Difference in lipid accumulation between healthy and shunts liver biopsies
Oil-red-O staining for neutral lipid accumulation was increased in EHPSS and IHPSS slides compared to livers of healthy control dogs (Fig 1A). Image J quantification revealed a 12-fold increased staining intensity in both shunt types (EHPSS P < 0.01; IHPSS P < 0.05) (Fig 1B) compared to the samples of healthy control dogs in our Dutch cohort. Semi-quantitative analysis of these samples confirmed the higher hepatic neutral lipid levels in EHPSS (P < 0.01) and IHPSS (P < 0.05) compared to healthy dogs (S2 Fig). Similar gene-expression patterns of lipid related genes in both shunt types Data retrieved from previously published data sets revealed that similar genes were differentially expressed in both shunts compared to samples of healthy dogs in the microarray analysis (S2 Table). Interestingly 11 out of the selected top 24 differentially expressed genes are related to lipid-metabolism, -transport, or -storage emphasizing the importance of the altered lipid metabolism here. Nine gene products (CBR2, CRP, ELOVL2, FABP1, IGFBP1, ITIH3, ITIH4, PLIN2, and SAA1) were significantly upregulated in both shunt groups whereas HSD3B and SEC14L3 were significantly downregulated compared to healthy control dogs in the microarray analysis. The expression of these 11 genes was validated by RT-qPCR in an independent cohort. For technical reasons, no RT-qPCR data could be obtained for CBR2. Due to difficulties in primer design for ELOVL2 but the interest in this gene, primers were ordered for ELOVL5 and ELOVL6 to gain information about the ELOVL2 pathway. The RT-qPCR analysis confirmed the microarray results for seven genes of interest (P < 0.01), namely CRP, FABP1, HSD3B, IGFBP1, ITIH4, PLIN2, and SAA1 (Fig 2). ELOVL2-pathway, ITIH3, and SEC14L3 expression levels were not significantly different in the validation cohort. and S3 Table). The phospholipidomes of IHPSS and IHPSS were not distinguisable, as can be concluded from the overlap of these samples in the PCA score plot (Fig 3A). When evaluating the phospholipid classes and phospholipid chain length between shunt types or healthy control dogs, no differences are observed (S5 Fig). Interestingly, dogs with a shunt had higher levels of hexaenoic (6 unsaturations per molecule) phospholipid species (Fig 3C) at the expense of tetraenoic (4 double bonds) and, to a lesser extent, pentaenoic (5 double bonds) species compared to the healthy control dogs. From the PCA loading plot (Fig 3B), it can be deduced that this effect is shared among all major phospholipid classes.  (Fig 3A). For neutral lipids, a shift towards shorter chain fatty acids (C16:n) is observed in livers of shunt dogs at the expense of more extended fatty acids (C18:n) measured in tissue of healthy dogs (Fig 4 and S4 Table). No difference is observed in TAGs with a chain length of C56 or above between the groups (Fig 4C).

Similar TAG accumulation in organoids from healthy and shunt livers
Triacylglycerol accumulation in hepatic organoids was evaluated by whole mount LD540 staining. Microscopically, the LD540 accumulation in lipid droplets, displayed in a single section, was more pronounced in the organoids cultured with FFA supplementation and revealed little differences between healthy and both CPSS canine liver organoids regarding basal fatty acid uptake (control medium) or in medium supplemented with FFA ( Fig 5A). In order to quantify the total TAG accumulation flow cytometry analysis of whole organoids was performed. The relative increase in median LD540 fluorescence after FFA enrichment was similar in the healthy and shunt groups (Fig 5B) indicating that there were no differences in total TAG accumulation.

Discussion
This study provides detailed insight into the pathogenesis of hepatic steatosis in CPSS dogs by examination of the quantity of accumulating lipids, mRNA expression of genes involved in lipid metabolism, and lipid profiles in both EHPSS and IHPSS compared to control dogs. To further support our hypothesis, canine organoids of EHPSS, IHPSS and healthy controls demonstrated a similar increase in TAGs after free fatty acids enrichment. Importantly, the results obtained in this study elucidate aspects of steatosis in portosystemic shunting, demonstrate that hepatic steatosis observed in CPSS occurs presumably secondary to portosystemic shunting and finally provide a model to study steatosis in human medicine.
The complex pathogenic mechanisms underlying NAFLD are influenced by multiple factors including genetic, hormonal and nutritional factors [2]. For instance, a missense mutation in human PNPLA3 (patatin-like phospholipase domain-containing 3 protein) is associated with increased hepatic fat deposition and hepatic inflammation, which makes the liver more susceptible for NAFLD [32]. In dogs, hepatic steatosis featured by vacuolar changes within hepatocytes and the presence of lipogranulomas, is a frequently observed finding in liver biopsies of dogs with CPSS [11][12][13]15]. Once a shunt is attenuated causing restauration of the normal liver blood flow, the lipid accumulation seems to be reversible over time [14]. Only a few descriptive studies have investigated this phenomenon in dogs [15,33,34], but none studied the pathophysiology of steatosis in CPSS.
As extra-and intrahepatic portosystemic shunts have a different genetic background [7], the similarity in the quantity of lipids, gene-expression, and lipid profiles suggests lipid accumulation to be a secondary effect of portosystemic shunting. In addition, the FFA-supplemented organoid cultures did not reveal differences between shunt derived organoids (EHPSS or IHPSS) strengthening the idea of a secondary effect. Age could be an influencing factor on Lipid metabolism in canine portosystemic shunt the degree of steatosis in CPSS as the incidence of lipogranulomas (LG) seems to be greater in age-matched dogs with CPSS compared to healthy dogs and LG are generally less observed in dogs under one year of age [33,34]. Another explanation for steatosis occurring secondary to CPSS could be plasma ammonia levels. CPSS are the most frequent cause of hyperammonemia in dogs which, when left untreated, results in hepatic encephalopathy [35]. In case of hyperammonemia, the ammonia is postulated to accumulate in lysosomes. Consequently intralysosomal pH will raise, thereby inhibiting lysosomal enzymes involved in proteolysis and lipid degradation [36,37]. The reduced breakdown of proteins and lipids can contribute to the progressive abnormalities in the brain in case of hepatic encephalopathy [38]. Therefore, the presence of the biomarker ammonia in CPSS [39] might be causative for the secondary lipid accumulations observed in portosystemic shunting. For two reasons it is not likely that insulin resistance, an important factor in the pathogenesis of NAFLD [2], plays a role in steatosis in CPSS dogs. Insulin resistance is associated with hyperinsulinemia [2] and low circulating IGFBP1 levels. Insulin levels vary greatly among diseased dogs [40] and IGFBP1 levels in this study are induced (Fig 2) rather than reduced. Taken together, the presented data together suggests that steatosis in CPSS dogs is a secondary phenomenon in portosystemic shunting possibly influenced by hyperammonemia.
Gene expression profiling and RT-qPCR validation corroborated the importance of seven lipid related genes to both subtypes of portosystemic shunting. The FABPI and PLIN2 upregulation in extra-and intrahepatic shunts (Fig 2) is probably associated with their lipid related functions and the lipid accumulation in CPSS. FABP1 serves as a key regulator of hepatic lipid metabolism by enhancing the cellular uptake, transport, and metabolism of fatty-acids [41]. Notably, this protein can bind bile acids which are often increased in CPSS [41]. Increased PLIN2 expression improves cellular lipid accumulation and regulates (phospho)lipid exchange from lipid droplets [42]. A down-regulation of HSDB3 was observed, in line with our data, in Lipid metabolism in canine portosystemic shunt granulosa cells when FFA concentrations increased [43]. Whether this hepatic down-regulation is directly correlated with FFA content needs to be investigated.
Interestingly, three of the six upregulated lipid related genes (i.e. CRP, ITIH4, and SAA1), serve as acute phase reactants (APR) which are secreted in response to a variety of acute and chronic inflammatory conditions, in particular regulated by IL-6. CRP is an important nonspecific biochemical marker of inflammation which is synthesized in both liver as well in adipose tissue in the presence of obesity [44]. Besides its role as APR, ITIH4 is associated with hypercholesterolemia [45] and may play an important role in liver regeneration [46]. SAA1 is associated with high-density lipoprotein metabolism, and cholesterol metabolism and transport [47]. This protein is also involved in the formation of amyloid deposits in feline and canine hepatic amyloidosis, triggered by inflammatory conditions [48]. Based on these results, the question raises whether the upregulation of these genes in portosystemic shunting is solely caused by an altered lipid metabolism per se or as a consequence of the inflammatory state [49]. The lack of an inflammatory component in hepatic biopsies of CPSS dogs argues in favour of a direct effect of lipid accumulation.
Since the liver plays a vital role in lipid metabolism, any disturbance in the fatty acids and triglycerides pathways leads to an imbalance in the lipid metabolism resulting in hepatic steatosis, and eventually steatohepatitis [50][51][52]. A plethora of biological effects of omega-3 and -6 fatty acids have been described [53]. Remarkable is the shift towards omega-3 fatty acids at the expense of the omega-6 fatty acids in particular the shift of AA (20:4, O-6) to DHA (22:6, O-3) in shunt dogs in comparison to healthy control dogs. AA is the main precursor of eicosanoids, which modulate the immune response via a diversity of pathways [54]. Dietary DHA has the capacity to suppress markers of hepatic damage, hepatic inflammation, oxidative stress and fibrosis in LDLR -/mouse with induced non-alcoholic steatohepatitis [55] and is reported to be beneficial in hepatic encephalopathy [56]. The altered lipid metabolism in elevated levels of DHA in shunt livers compared to healthy control dogs, might therefore be a protective response.
Due to population bottlenecks and inbreeding during the formation of the contemporary dog breeds, canines have a limited phenotypic and genetic diversity which makes the dog population ideal for exploring the genetic basis of a variety of naturally occurring diseases [57][58][59]. The dog has been proposed as a useful model to study inherited diseases in both canine and human research [60], since they are remarkably similar between canine and human diseases based on phenotypic presentation. This also holds true for CPSS, the same subtypes for intrahepatic and extrahepatic shunts in dogs have been recognized in man, although at a much lower frequency in humans [8,61]. Histological features observed in dogs with CPSS are comparable to human [62] and rats with an induced portacaval shunt [63]. The high prevalence but poorly understood pathogenesis of NAFLD urges the search for reproducible and predictive disease model systems. Therefore, studying lipid loading in CPSS dogs, could in a further stadium serve as a model to study the pathophysiology of steatosis and/or novel treatment modalities preventing lipid accumulation.
In conclusion, this study describes excessive hepatic lipid accumulation in portosystemic shunting. Gene expression profiling indicated that the majority of genes changed in CPSS were involved in lipid metabolism. Different TAG composition and a shift in short chain and omega-3 fatty acids were observed in shunt dogs compared to healthy animals. Despite a different genetic background of extra-and intrahepatic shunts, lipid species observed in both shunt types were almost identical. As cultured organoids derived from healthy and diseased animals accumulate TAGs equally, we suggest that lipid accumulation as observed in shunt livers appears not to be related to primary gene defects in liver shunts, but rather be caused by a secondary effect, possibly ammonia related. Histological features observed in dogs with CPSS are comparable to human [62] and rats with an induced portacaval shunt [63]. Since lipid accumulation is a natural phenomenon in CPSS dogs, these animals might represent a simplified NALFD model.