ATGL and DGAT1 are involved in the turnover of newly synthesized triacylglycerols in hepatic stellate cells

Abstract
Hepatic stellate cell (HSC) activation is a critical step in the development of chronic liver disease. During activation, HSCs lose their lipid droplets (LDs) containing triacylglycerol (TAG), cholesteryl esters (CEs) and retinyl esters (REs). Here we aimed to investigate which enzymes are involved in LD turnover in HSC during activation in vitro. Targeted deletion of the Atgl gene in mice HSCs had little effect on the decrease of the overall TAG, CE and RE levels during activation. However, ATGL- deficient HSCs specifically accumulated TAG species enriched in PUFAs and degraded new TAG species more slowly. TAG synthesis and levels of PUFA-TAGs were lowered by the DGAT1 inhibitor T863. The lipase inhibitor Atglistatin increased the levels of TAG in both wt and ATGL-deficient mouse HSCs. Both Atglistatin and T863 inhibited the induction of activation marker α-SMA in rat HSCs, but not in mouse HSCs. Compared to mouse, rat HSCs have a higher turnover of new TAGs and Atglistatin and the DGAT1 inhibitor T863 were more effective. Our data suggest that ATGL preferentially degrades newly synthesized TAGs, synthesized by DGAT1, and is less involved in the breakdown of pre-existing TAGs and retinyl esters in HSCs. Furthermore a large change in TAG levels has modest effect on rat HSC activation.

Introduction
Hepatic stellate cells (HSCs) are the main vitamin A (retinol)-storing cells of the body (1,2). In a healthy liver, HSCs store vitamin A in the form of retinyl esters (REs) in large lipid droplets (LDs), together with triacylglycerols (TAGs) and cholesteryl esters (CEs). HSCs are located in the space of Disse, between the sinusoidal endothelial cells and the hepatocytes. Upon liver injury, quiescent HSCs can transdifferentiate into an activated myofibroblastic phenotype (1). Activated macrophages in concert with the HSCs may initiate this transition by secreting cytokines, such as transforming growth factor beta (TGF-β), which stimulate the synthesis of matrix proteins and the release of retinoids by HSCs (1,3). The loss of retinoids is associated with a gradual disappearance of the LDs inside the HSCs. We previously reported that LD degradation in activated rat hepatic stellate cell occurs in two phases (4). Upon activation of the hepatic stellate cells, the LDs reduce in size, but increase in number during the first 7 days in culture before they disappear in a later phase. Raman and lipidomic studies showed that in the initial phase of HSC activation, the REs are disappearing rapidly, whereas the TAG content is transiently increased (4). Interestingly, this increase in TAGs in rat HSCs is predominantly caused by a large and specific increase in poly-unsaturated fatty acid (PUFA)-containing triacylglycerol species during the first 7 days in culture, mediated by an increase in expression of the PUFA-specific fatty acid CoA synthase (ACSL) 4 and a decrease in expression of the other ASCLs, especially ASCL1 (5). So far, the molecular mechanisms and identity of the enzymes involved in the observed increase in lipid droplet number, and their subsequent breakdown during HSC activation are not well understood. An increase in number can be accomplished by the novo synthesis of new LDs (6) or fission of existing large LDs (7). Breakdown of LDs is best characterized in adipose cells, in which key roles were assigned to adipose triglyceride lipase (ATGL, also known as PNPLA2), its co-activator CGI-58 and hormone sensitive lipase (HSL) (8).

The first two proteins are known to have a more general function as deficiencies in either one leads to neutral lipid storage diseases (9). Both ATGL and CGI-58 were present on LDs in the hepatic stellate cell line HSC-T6 (10), and rat HSCs were shown to express ATGL, although it was down regulated upon activation (11). In mouse HSCs, lipid breakdown was shown to be partially mediated by a lipophagic pathway, as inhibition of autophagy increased the amount of LDs (12,13). Because inhibition of autophagy was shown to impair HSC activation in mice and this effect could be partially reversed by addition of exogenous FAs, it was suggested that LD breakdown is required to fulfill the energy demands of HSCs during activation (13). On the other hand, HSC activation was shown to be relatively undisturbed in the absence of REs and LDs in Lecithin:retinol acyl transferase (LRAT) knockout mice (14).In this study we addressed the question whether a change in lipid metabolism is causally related to the activation process in rat and mouse HSCs. We identified enzymes involved in LD formation and breakdown in HSCs in vitro and studied the effect of inhibition of these enzymes on HSC activation.
D4-palmitate, D8-arachidonate and Atglistatin were purchased from Cayman Chemical (Ann Arbor, MI, USA). Dulbecco’s modified Eagle medium (DMEM), fetal bovine serum (FBS), and penicillin/streptomycin were obtained from Gibco (Paisley, UK). Bovine Serum Albumin (BSA) fraction V was obtained from PAA (Pasching, Austria). T863 and Collagenase (Clostridium histolyticum Type I) was obtained from Sigma-Aldrich (St. Louis, MO, USA), and saponin from Riedel-de Haën (Seelze, Germany). The mouse monoclonal antibody against alpha-smooth muscle actin (α-SMA) was from Thermo Scientific (Waltham, MA, USA). Lipid droplet staining dye LD540 was kindly donated by Dr.C. Thiele, Bonn, Germany. Hoechst 33342 was obtained from Molecular Probes (Paisley, UK), paraformaldehyde (PF) (8%) was obtained from Electron Microscopy Sciences (Hatfield, PA, USA). FluorSave was obtained from Calbiochem (Billerica, MA, USA), all HPLC-MS solvents were from Biosolve (Valkenswaard, the Netherlands) with exception of chloroform (Carl Roth, Karlsruhe, Germany) and were of HPLC grade. Silica-G (0.063–0.200 mm) was purchased from Merck (Darmstadt, Germany).

We used 10-12 week-old male and female Atgl+/+ (wt) and Atgl-/- mice, generated by crosses of Atgl+/- C57BL/6J mice (15) and paired on sex and age from the same nest and adult male Wistar rats (300–400 g).Procedures of mouse and rat care and handling were in accordance with governmental and international guidelines on animal experimentation, and were approved by the Animal Experimentation Committee (Dierexperimentencommissie; DEC) of Utrecht University (DEC-numbers: 2010.III.09.110, 2012. III.10.100, and 2013.III.09.065).Stellate cells were isolated from livers of mice and rats by collagenase digestion followed by differential centrifugation (16) and cultured as described previously (5) in DMEM supplemented with 10% FBS, 100 units/ml penicillin, 100 µg/ml streptomycin and 4 µl/ml Fungizone and cells were maintained in a humidified 5% CO2 incubator at 37°C. Medium was changed every 3 days.Gene silencingGene-silencing experiments were performed using small interfering RNA (siRNA) for target genes or non-targeting siRNA as a control (ON-TARGETplus SMARTpool of 4 siRNAs, Thermo-Scientific, Rochester, NY, USA) according to the manufacturer’s instructions. Briefly, two days after plating, the cells were treated with 40 nM siRNA using 5 µl/ml RNAiMAX (Invitrogen, Breda, the Netherlands) in antibiotic free complete media (with FBS). After 6 hr of transfection, media was changed to standard culturing conditions up to day 7.Total RNA was isolated from HSCs grown in a 24-well plate using RNeasy Mini Kit (Qiagen, Venlo, the Netherlands) including the optional on-column DNase digestion (Qiagen RNase-free DNase kit). RNA was dissolved in 30 μl of RNase free water and was quantified spectrophotometerically using a Nanodrop ND-1000 (Isogen Life Science, IJsselstein, the Netherlands). An iScript cDNA Synthesis Kit (Bio-Rad, Veenendaal, the Netherlands) was used to synthesize cDNA. Primer design and qPCR conditions were as described previously (17). Briefly, qPCR reactions were performed in duplicate using the Bio-Rad detection system.

Amplifications were carried out in a volume of 25 µl containing 12.5 µl of 2xSYBR green supermix (BioRad), 1ul of forward and reverse primer and 1 µl cDNA. Cycling conditions were as follows: initial denaturation at 95°C for 3 minutes, followed by 45 cycles of denaturation (95°C for 10 seconds), annealing temperature (see Supplementary data, Table S1 and S2)for 30 seconds, and elongation (72°C for 30 seconds). To determine the relative expression of a gene, a 4-fold dilution series from a pool of all samples of all genes tested were used. The amplification efficiency was between 95-105%, amplicon sequencing conformed the specificity and for each sample a melt-curve analysis was performed. IQ5 Real-Time PCR detection system software (BioRad) was used for data analysis. Expression levels were normalized by using the average relative amount of the reference genes. Reference genes used for normalization are based on their stable expression in stellate cells, namely, tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein zeta (Ywhaz), hypoxanthine phosphoribosyl transferase (Hprt), and hydroxylmethylbilane synthase (Hmbs). Primers of reference and target genes are listed in Supplementary Data, Table S1 and S2.Freshly isolated HSCs were grown on glass coverslips in 24 well plates at 37°C for 7 days. Staining of cells was performed as follows: cells were fixed in 4% (v/v) PF at room temperature for 30min and stored in 1% (v/v) PF at 4°C for a maximum of 1 week. HSCs were washed twice in PBS, permeabilized (0.1% (w/v) saponin) and blocked with 2% BSA for 1hr at room temperature. After blocking, slides were incubated 1hr with the primary antibody against α-SMA (50-75 ug/mL), washed again, and incubated for 1hr with anti-mouse antibody (15ug/mL) supplemented with Hoechst (4 ug/mL) for nuclear counterstaining and lipid droplet dye LD540 (0.05 ug/ml). Thereafter, coverslips were mounted with FluorSave on microscopic slides.

Image acquisition was performed on a Leica TCS SPE-II confocal microscope at the Center of Cellular Imaging, Faculty of Veterinary Medicine, Utrecht University.To quantify lipid droplet size and numbers per cell, confocal images of LD540 (lipid droplets) and Hoechst33342 (nuclei) were analyzed with CellProfiler v2.1.1. Recognized lipid droplets and nuclei were overlayed on the original image to confirm the identity. The error rate was <5% for lipid droplets and <2% for nuclei.To investigate the dynamics of TAG formation and degradation, we labeled isolated rat HSCs for 1 or 2 days with 25 µM D4-palmitate and 25 µM D8-arachidonate in medium containing 10% fetal calf serum. The labeling medium contained approximately 25 % of unlabeled palmitate and 5% of unlabeled arachidonate from the fetal bovine serum (5-10 µM, and 1-2 µM, respectively; (18)). After the labeling the cells were either harvested or subsequently chased for 1 or 2 days in medium containing 10% fetal calf serum the absence of the stable isotope-labeled FAs to determine the breakdown of the labeled TAGs.Lipids were extracted from a total cell homogenate of HSCs grown in a 12-well plate by the method of Bligh and Dyer (19) after addition of 200 pmol tripentadecanoylglycerol as internal standard. Lipid extracts were dissolved in methanol/chloroform (1/9, v/v), loaded on top of the silica-G column (approximately 10 mg of 0.063–0.200 mm silica) and separated in a neutral and phospholipid fraction (20). Neutral lipids were eluted with two volumes acetone, dried under nitrogen gas and stored at −20°C. Prior to HPLC-MS analysis, the neutral lipid fraction was dissolved in methanol/chloroform (1/1, v/v) and separated on a Kinetex /HALO C8-e column (2.6 micron, 150×3.00 mm; Phenomenex, the Netherlands). A gradient was generated from methanol/H2O (5/5, v/v) to methanol/isopropanol (8/2, v/v) at a constant flow rate of 0.3 ml/min. Mass spectrometry of neutral lipids (TAGs, CEs, RE, and cholesterol) was performed using Atmospheric Pressure Chemical Ionization (APCI) interface (AB Sciex Instruments, Toronto, ON, Canada) on a Biosystems API-2000 Q-trap. The system was controlled by Analyst version 1.4.2 software (MDS Sciex, Concord, ON, Canada) and operated in positive ion mode using settings described previously (4,5).

Typical TAG, CE, and RE ions detected are shown in Supplementary Data, Fig. S1. Cholesterol and CEs give typical ions with a m/z of 369, and REs with am/z of 269, due to the loss of the FA chain. It is important to note that TAG species have retention times on HPLC that are characteristic for TAGs. Subsequently, ionization of the TAG species by APCI generates some DAG fragments. As these DAG fragments are generated after HPLC, they have HPLC retention times that are the same as the TAG species that they are derived from. In contrast, DAG species present in HSCs have DAG-specific retention times. TAG fragments with two non-, single-, and double-labeled palmitoyl chains (16:0,16:0,x) were quantitated by counting ions with m/z of 551, 555 and 559 respectively. TAG fragments with a non- or single-labeled palmitoyl and an oleoyl chain (16:0,18:1,x) were quantitated by counting ions with m/z of 577 and 581, respectively, and non-, single-, double- and triple-labeled di-arachidonoyl-palmitoyl-glycerol (20:4, 20:4,16:0) was quantitated by counting ions with m/z of 903, 907+911, 915+919, and 923, respectively. Typical non-labeled, high abundant TAG species quantitated were: non PUFA, 52:3 (m/z 859), and 54:3 (m/z 885); PUFA, 56:5(m/z 909), 58:6 (m/z 935), 60:9 (m/z 957) and 62:11 (m/z 981). Total TAG was determined by quantitating all ions between m/z 530 and 1050 with a retention time of TAG species and corrected for the presence of second and third isotope peaks. In this mass and retention time range no CEs and REs were detected (See Supplementary Data, Fig. S1). The quantitated lipids were normalized to the amount of cholesterol in the same sample. Cholesterol was found to be a good marker for recovery of cellular material, as the cholesterol/protein ratio was found to be constant during HSC culture.Each experiment was performed in duplicate and repeated at least three times. Comparisons of variables between multiple factors were made by a two-way ANOVA. Comparisons of a variable between two groups were made with unpaired or paired Student’s t test depending on whether the data were normalized to a control before analysis. Differences were considered statistically significant if the P value was less than 0.05.

Results
We investigated the dynamics of neutral lipid breakdown in HSCs isolated from wt and Atgl-/- mice. HSCs from Atgl -/- mice have almost 2-fold higher TAG levels compared to wt mice during the cell culture period, but similar levels of CEs and REs (Fig. 1, panels A-C). These results clearly implicate ATGL in TAG metabolism of HSCs. Surprisingly, however, upon culturing and hence activation of HSCs, the TAG levels in ATGL-deficient HSCs dropped with an apparently similar rate as compared to wt (Fig. 1A), indicating that ATGL is not the main lipase in mouse HSCs involved in TAG hydrolysis during activation. Likewise, there was no effect of ATGL knockout on the decrease of CEs and REs levels upon culturing (Fig. 1B and 1C). To look more closely at a possible role of ATGL in lipid breakdown during mouse HSC activation, we quantitated different TAG species individually and expressed the TAG levels as percent of the levels at day 1 (Fig. 1D-F). In particular TAG species with one or more polyunsaturated fatty acid (PUFA) moieties were increased in ATGL deficient HSCs at day 7 and 14, but decreased in wt HSCs. Non-PUFA TAG species decreased at an almost similar rate in Atgl-/- HSCs. An exact determination of the contribution of the PUFA-TAG species to the total amount of TAG by the APCI-MS method is hampered by the numerous species (>1000) and the observation that the various TAG species fragment to a different degree depending on the saturation of their acyl chains (5). Nevertheless, quantitation of a number of representative intact and fragmented TAG ions allowed us to estimate that wt HSCs at day 1 have 10% TAGs with one or more PUFAs, whereas this was approximately 5% in Atgl -/- HSCs. At day 14, the fraction of PUFA-TAGs in wt HSC was similar as on day 1, but in ATGL-/- cells it was increased to more than 30% (Supplementary Data, Fig. S1D and S2).This allows us to estimate that the PUFA-TAG levels in Atgl -/- HSCs at day 1, 7 and 14 are 0.13, 0.33 and 0.29 nmol/mmol chol, respectively.

We previously reported that the increase in PUFA-TAGs in wt HSCs upon culturing in vitro was dependent on a preferential incorporation of exogenous FAs into TAG (4,5). We therefore determined the incorporation and subsequent chase of deuterium-labeled fatty acids (D4-palmitate and D8- arachidonate) in mouse HSCs. As shown in Fig. 2A, approximately 5% of all the TAG species were labelled with D4-palmitate in either TAG (16:0, 16:0,x) and TAG (16:0, 18:1,x) (sum of labeled species in first and fifth bar). The predominant D4-palmitate labeled species were the TAGs with two palmitoyl groups, from which more than 50% was labeled (Fig.2A, first bar). ATGL is involved in the breakdown of this newly formed TAG pool as the labeled TAG pool was lost at a significantly lower relative rate in Atgl -/- HSCs (half time > 2 days) compared to wt cells (half time < 1 day; Fig. 2C grey and dark grey bars). The breakdown of non-labeled, mainly pre-existing TAGs were not different in the ATGL- deficient cells (Fig. 2C, white bars), assuming an equal re-synthesizis of unlabeled TAG species during the chase in both mice types from FAs from the medium. Similar results were observed for the predominant D8-arachidonate-labeled TAG species (Fig. 2B and 2C), indicating that ATGL is specific for newly formed TAGs, irrespective of the degree of unsaturation of the fatty acid moieties.We previously reported that the enrichment of PUFAs into newly synthesized TAGs is caused by a down regulation of the general ASCL1 isoform concomitant with an upregulation of the PUFA-specific ACSL4 isoform upon rat HSC activation (5). To see whether a similar shift occurs in mouse HSCs to explain the increase in PUFA-TAGs observed in the ATGL-deficient HSCs, we determined the relative mRNA expression of ASCL1 and ASCL4 in mouse HSCs at day 1, and 7. We found that the ratio of ACSL4 mRNA to ASCL1 mRNA levels increased from 0.8 ± 0.2 at day 1 to 4.6 ± 0.7 at day 7 (n=6). No difference in increase in ASCL4/ASCL1 ratio was observed between wt and ATGL-/- HSCs. This demonstrates a clear enrichment of the PUFA-specific ASCL 4 isoform upon activation of mouse HSCs.

Furthermore, we observed that D8-arachidonate was incorporated at a higher rate into TAG in activated HSCs at day 6 than in relative quiescent cells at day 1 (Fig. 3B).To investigate which of the two Acyl-CoA: Diacylglycerol Acyltransferase (DGAT) enzymes is involved in the formation of the new TAG species in mouse HSCs during activation we determined the DGAT1 and DGAT2 mRNA expression in mouse HSCs at day 1 and day 7. The mRNA levels of DGAT1 were higher compared to those of DGAT2 at day 1 (relative mRNA levels of 118 ± 26 and 35 ± 36, respectively; n=6), and the difference between DGAT1 and DGAT2 mRNA expression increased during activation at day 7 (relative mRNA levels of 136 ± 20 and 14 ± 16, respectively; n=6). To test the role of DGAT1 on the synthesis of new TAGs we determined the incorporation of deuterium-labeled FAs into TAG in the presence of the DGAT1-specific inhibitor T863 (21). As shown in Fig. 3A and 3B, the DGAT1 inhibitor inhibited the incorporation of both D4-palmitate and D8-arachidonate into TAG. The DGAT1-specific inhibitor T863 also caused a clear decrease in the levels of PUFA-TAGs in both wt and in Atgl -/- HSCs during activation, but had almost no effect on the non-PUFA TAG levels (Fig 3C). This indicates that PUFAs are preferential incorporated in the newly formed TAG pool in mouse, similar to observed in rat HSCs. As shown in Fig 3C, the levels of CEs were lower in T863-treated mouse HSCs, suggesting that this drug may also affect cholesterol acylation.The Atgl -/- mice also provided a model to study the specificity of the ATGL-specific inhibitor Atglistatin in HSCs (22). Surprisingly, Atglistatin-treatment for 6 days resulted in an almost similar increase in TAG levels in Atgl -/- as compared to wt HSCs (Fig. 3D). This amounted to a 2-fold net increase in the total TAG levels in both wt and Atgl -/- HSCs, when compared to the original TAG levels at day 1 (Fig. 3E).

This clearly demonstrates that Atglistatin not only inhibits ATGL but also targets another, as yet unknown, TAG lipase, which plays a major role in the degradation of the TAG pool in mouse HSCs.To investigate whether a change in LD size and number is causally related to the activation process, we determined the expression levels of α-smooth muscle actin (α-SMA) after the various treatments to alter TAG content of the HSCs and determined the cellular morphology of the cells. α-SMA is considered a marker for HSC activation and was clearly upregulated in mouse HSCs (Fig. 4A). However, neither deletion of ATGL nor addition of the lipase inhibitor Atglistatin or the DGAT1-inhibitor T863 affected the expression of the activation marker α-SMA in mouse HSCs (figs 4A and 4B). In contrast, we observed a 30 % inhibition of α-SMA expression by both Atglistatin and T863 in rat HSCs (Fig. 4B). Imaging studies show that Atglistatin increased both the number and size of LDs in rat and mouse HSCs, whereas the DGAT1-inhibitor T863 decreased these parameters in rat HSCs (Fig. 5). In HSCs from wt mice, T863 had no effect on LD size or number, but T863 decreased the number of LDs in Atgl-/- HSCs (Fig. 5).As we observed effects of Atglistatin and the DGAT-inhibitor T863 on activation and LD number and size in rat HSCs, we investigated neutral lipid metabolism in isolated HSCs from this rodent in more detail. The dynamics of TAG formation and degradation in rat HSC was assessed by labeling freshly isolated rat HSCs for 2 days with deuterated FAs followed by a 1 day chase. As shown in Fig. 6A (sum of labeled species in first and third bar), 20 % of all the TAG species were labelled with D4-palmitate after 2 days. From the TAG species containing two palmitoyl groups even 80% was labeled (first bar). This suggests that a large part of the total TAG pool is renewed if we assume that palmitoyl-containing TAG species behave similar to the other TAG species. Most of the label was replaced after the 24 h chase, especially in the doubly labeled TAGs, which lost more than 90% of their label (see Fig. 6A,chase and 6B).

Similar results were observed for the D8-arachidonate-labeled TAG species upon addition of D8-arachidonate to HSCs for two days (results not shown; cf. (4)). This indicates that rat HSCs contain a large and highly dynamic TAG pool with turn-over rates of less than 8 hrs. In comparison to mouse HSCs, rat cells incorporated more labeled FAs, but also lost their newly made TAGs at a higher relative rate (cf. Figs 2A and 2B to 6A and 6B).Next we tested the effects of the DGAT1-specific inhibitor T863 and Atglistatin on TAG formation and breakdown in rat HSCs. As shown in Fig. 7A, T863 inhibited the incorporation of D4-labeled palmitate in TAG by 80% and caused a similar decrease in the unlabeled TAGs, presumably due to inhibition of the incorporation of unlabeled palmitate from the fetal calf serum. Atglistatin prevented the breakdown of labeled TAG almost completely and also increased the level of non-labeled TAG almost 4-fold during the 2 day chase. This latter observation is caused most likely by inhibition of degradation of the newly formed unlabeled TAGs (Fig. 7A, and 7B). When rat HSCs were incubated for 6 days with Atglistatin the TAG levels were even 10-fold higher compared to control incubated cells (Fig. 7C). This experiment demonstrates the large capacity of rat HSCs to synthesize TAG, since net TAG formation is now unmasked by the virtual absence of the normally large concurrent degradation pathway. The DGAT1- specific inhibitor T863 caused 80% lower TAG levels after a 6 day incubation (Fig. 7D), suggesting that DGAT1 is the principal enzyme in the high TAG formation in activated rat HSCs. The levels of cholesterol esters and retinyl esters, were also lowered by T863, indicating that either these neutral lipids require the presence of the major LD core lipid TAG, or that T863 also affects the enzymes involved in CE and RE synthesis.T863 and Atglistatin might be not completely specific for DGAT1 and ATGL, respectively, as was clearly found in case of Atglistatin in mouse HSCs.

Therefore, we wanted additional evidence for a role of ATGL and DGAT1 in the rapid turn-over in rat HSCs. To this end we measured the expression of various TAG-metabolizing genes in quiescent (day 1) and activated (day 7) rat HSCs. Of the tested genes, Atgl, in combination with its regulator CGI-58, and Dgat1 are likely to play a role in the rapid rat HSC lipid turnover, as HSL mRNA was not detectable and Dgat2 expression was relatively low (Fig. 8A). We found no indication that the expression level of these genes was up-regulated during activation. In contrast, ATGL expression was down-regulated during the 7 day culture of the HSCs. We next explored the function of the candidate genes Atgl, Dgat1 and Dgat2 by siRNA-mediated silencing. The efficiency of silencing was between 80% and 50% as shown in Fig. 8B. In line with the mRNA expression data, knock-down of Atgl and Dgat1, but not Dgat2, affected the levels of TAG in rat HSCs (Fig. 8C). Atgl knock-down clearly inhibited the relative loss of labeled palmitate during the chase and caused a higher level of labeled palmitate containing TAGs at the beginning of the chase, probably by an inhibition of breakdown during the labeling period (see Fig. 8D and 8E). A more quantitative estimation of the roles of ATGL and DGAT1 on TAG levels in rat HSCs was hampered by the relatively low knock-down efficiency in these primary cells (50 % and 60 %, respectively, on mRNA levels; Fig. 8B). The relatively small effects on TAG levels by the knock-down of ATGL and DGAT1 did not cause an effect on HSC activation, since we could find no effect on α-SMA expression by siRNA-mediated knock down of Atgl, Dgat1 and Dgat2 in rat HSCs (Fig. 8F).

Discussion
Our study with ATGL-deficient HSCs showed that ATGL has a limited, but specific role in lipid degradation in this cell type. We could not confirm a significant role of this lipase in the degradation of retinyl esters in culture-activated mouse HSCs as suggested by Taschler et al. (23). Knockout of Atgl did not prevent a decrease in most TAG species without PUFA chains, but caused a preferential accumulation of PUFA-TAG species. This differential behavior was not expected, as no clear preference of ATGL for specific acyl chains has been described (24). We here propose that, rather than an acyl- chain preference, the observed difference in TAG species distribution reflects a preference of ATGL for newly synthesized TAGs, as i) the degradation of newly formed TAGs was specifically decreased in the ATGL-deficient mouse HSCs, ii) PUFAs are preferentially incorporated in newly synthesized TAG in HSCs upon activation by an enrichment of ACSL4 (5), that we now observe also in mouse HSCs, and iii) the DGAT1-specific inhibitor T863, which inhibited the formation of new TAGs in mouse HSCs, also specifically affected the levels of PUFA-TAGs in activated mouse HSCs.The suggestion that ATGL specifically targets newly synthesized TAGs, would imply that in activated HSCs two pools of TAG exist. A original pool of TAG present in the relatively large pre-existing LDs, and a novel pool, made preferential by DGAT1 in combination with ACSL4, and thus enriched in PUFAs. This latter pool may well be located in newly formed, and therefore relatively small LDs. In activated rat HSCs we previously observed that LDs reduce in size, but increase in number and are localized in cellular extensions (4). We speculated that this was caused by fission of the LDs and extensive migration to the periphery (4). Our new data favors an alternative explanation, i.e. that the apparent redistribution of LDs is due to a degradation of the lipid content (both TAGs and RE) of original LDs concurrent with an rapid resynthesis of TAGs in novel lipid droplets by DGAT1 in the cell periphery. Interestingly, DGAT1 has been associated with the formation of small new LDs, whereas DGAT2 was involved in the enlargement of existing LDs in model cells (25).

In wild type mice, the accumulation of these new PUFA-TAGs is less apparent as this pool is probably degraded by ATGL at an equal rate as it is formed. In rat HSCs, however, the formation of new TAGs, preferential containing PUFAs, was shown to be much higher and is apparently not balanced by ATGL. We have strong evidence that DGAT1 is a key enzyme in this rapid TAG synthesis, as TAG levels were affected by siRNA-mediated knock down of DGAT1, but not DGAT2 in rat HSCs, and the DGAT1- specific inhibitor T863 affected the synthesis of new TAGs and resulted in a large drop in TAG species during activation. In rat HSCs, the relatively high rate of TAG synthesis must be compensated by an almost equal rate of degradation to obtain the small net gain of TAG observed (4). The breakdown of the novel TAGs was indeed found to be very high in rat cells, as more than 80-90% of the newly formed TAGs is degraded within 1 day. Furthermore, we found that inhibition of TAG breakdown by Atglistatin caused a massive increase in total levels of TAG, which was estimated to be more than a doubling of the original amount of TAG in one day (as 2 days of Atglistatin resulted in a 4-fold higher level of TAG, and 6 days of inhibition led to a 10-fold higher level; Fig. 7). The higher turnover of new TAGs in rat HSCs would suggest a larger role of ATGL in TAG homeostasis in rats compared to mice. This was corroborated by the clear effect of a limited siRNA-mediated knockdown of ATGL on total TAG levels (Fig. 8).The observation that ATGL-deficient mouse HSCs still lose their TAG content, indicates that another as yet unknown lipase is the main TAG-degrading enzyme in mouse cells. This other lipase, like ATGL, is sensitive to Atglistatin as TAG levels were 3-4 fold increased by this inhibitor in both wt and Atgl -/- HSCs (Fig. 3). This would eliminate monoglyceride lipase, hormone sensitive lipase, lipoprotein lipase and pancreatic lipase, as well as patatin-like phospholipase domain containing (PNPLA) 6 and PNPLA 7, which were shown to be unaffected by this drug (22). PNPLA3, also called adiponutrin (ADPN), is also an unlikely candidate since this enzyme is rather involved in TAG remodeling, than in net lipolysis (26). In human HSCs, PNPLA3 was shown to play a role in retinyl ester degradation (27), rather than TAG breakdown. But, interestingly, a variant of this gene was associated with steatohepatitis by an as yet unknown mechanism (28). Another candidate would be lysosomal acidic lipase (LAL/Lipa), the putative TAG degrading lipase in the lipophagy/autophagy pathway. Inhibition of autophagy was shown to result in TAG accumulation in mouse HSCs (12,13). We have no direct information whether Atglistatin inhibits LAL/Lipa. However, we did not observe an increase in CEs after treatment with Atglistatin, as would be expected in case of LAL/Lipa inhibition (29).

In general, this study also points out that HSCs from different species are not using metabolic pathways in a quantitatively similar way, not even in related animals as rat and mice. This may be relevant to explain the differences observed between mice and rats in the effect of strategies to alter lipid metabolism on HSC activation. The detection of two potent inhibitors of TAG synthesis and degradation in rat HSCs, T863 and Atglistatin, respectively, allowed us to address the question whether a change in TAG metabolism is causally related to the activation process in rat and mouse HSCs. Intriguingly, both inhibitors significantly attenuated activation in rat HSCs to a similar, albeit relatively low degree (30%, Fig. 4). Although we cannot exclude the possibility that these drugs affect HSCs activation by other, non-lipid related pathways, their effects on HSC activation correlated with their effect on TAG levels, as Atglistatin and T863 had less effect on TAG levels in mouse HSCs and did not affect the expression of the activation marker α-SMA in these cells. As inhibition of TAG formation had the same effect as inhibiting its breakdown, rat HSCs seem to require an available TAG pool for optimal functioning. This pool may act as a buffer for FAs required for energy (13), or synthesis of membranes or bioactive lipids, like prostanoids (5). Our study with the DGAT-specific inhibitor T863 is in line with previous studies showing that knockdown of Dgat1 in rat liver inhibited activation of HSCs in vitro (30), whereas knockout of Dgat1 in mice had no effect on subsequent HSC activation in culture (31). Interestingly in the latter study also indicated that a limited effect in vitro, might be meaningful in vivo, as the DGAT1- deficient mice were less sensitive to induction of liver Atglistatin fibrosis (31).