Targets for TNFα-induced lipolysis in gilthead sea bream(Sparus aurata L.) adipocytes isolated from lean and fat juvenile fish

Fish growth is a very complex process that responds to a variety of nutritional, environmental and genetic factors. Life-history decisions are,however, not fixed and depend on critical size and energy sufficiency at a specific stage `opportunity window' several months prior to transformation itself. Thus, the decision in salmonids to become smolt or sexually mature(Shearer and Swanson, 2000; Silverstein et al., 1997; Silverstein et al., 1998) is linked to growth and fat deposition in mid-summer and spring. In Mediterranean fish, the replenishment of body fat stores is also dictated by the seasonal calendar, but current aquaculture practices in European sea bass and gilthead sea bream largely increase fat deposition in fat storage tissues, leading to production and indirect selection of specimens with fatty characteristics. At the same time, the individual variability in fish adiposity is relatively high, and a major goal for the Mediterranean aquaculture is to explore the different lipostat mechanisms operating in farmed fish. For instance, liver steatosis can be induced in gilthead sea bream by high feeding ratios(Sitjà-Bobadilla et al.,2003) and partial or total replacement of fish meal and fish oil with alternative vegetable sources(Benedito-Palos et al., 2008; Caballero et al., 2004; Sitjà-Bobadilla et al.,2005). The underlying mechanisms are not well understood, but the enhanced removal of plasma triglyceride-rich lipoproteins by hepatic lipoprotein lipase (LPL) explains, at least in part, the increased liver fat deposition and hypotriglyceridaemic effect of plant protein diets(Saera-Vila et al., 2005).

The hormone-sensitive lipase (HSL) is the principal mediator of regulated lipolysis in body fat stores(González-Yanes and Sánchez-Margalet, 2006), and its regulation and substrate specificity have been addressed in salmonids(Harmon et al., 1993; Michelsen et al., 1994) and Antarctic fish (Hazel and Sidell,2004), respectively. Thus far, HSL sequences are not available in fish, but recently the gilthead sea bream HSL has been characterized using RT-PCR approaches, and its nucleotide sequence has been introduced into GenBank with the accession number EU254478. There is now also increasing interest in defining the involvement of tumour necrosis factor-α(TNFα) in the regulation of fish lipid metabolism. This pro-inflammatory cytokine affects many aspects of adipocyte function, and its lipolytic action has been demonstrated in rainbow trout and gilthead sea bream adipocytes(Albalat et al., 2005b; Saera-Vila et al., 2007). Recent studies have also shown that TNFα inhibits the differentiation of rainbow trout preadipocytes (Bouraoui et al., 2008). This, together with the high expression level of TNFα in the fat storage organs of gilthead sea bream(Saera-Vila et al., 2007),makes this cytokine a good candidate for playing a key role in reducing the adipose tissue mass. Nevertheless, the regulation and mode of action of TNFα remains mostly unexplored in cultured fish(Albalat et al., 2005b).

In mammals, TNFα regulates the expression of peroxisome proliferator-activated receptors (PPARs) and liver X receptors (LXRs) that belong to the nuclear hormone receptor superfamily with a crucial role in lipid and lipoprotein metabolism(Desvergne et al., 2006; Kim et al., 2007). Three PPAR isotypes (α, β and γ) exist in all the vertebrate species studied to data, including Atlantic salmon(Ruyter et al., 1997),European sea bass (Boukouvala et al.,2004) and gilthead sea bream(Diez et al., 2007; Leaver et al., 2005). Each PPAR isotope is the product of a separate gene and has a distinct tissue distribution and specific function as a sensor of dietary fatty acids(Chawla et al., 2001). Likewise, two isoforms of LXR exist in mammals (LXRα and LXRβ) and they are involved in the regulation of cholesterol homeostasis and fatty acid synthesis (Ou et al., 2001; Peet et al., 1998; Repa et al., 2000). Expression of LXRα is highest in liver and intestine but is also detected in macrophages, adipose tissue, kidney, lung and spleen whereas LXRβ is ubiquitously expressed (Zhang and Mangelsdorf, 2002). Recently, LXR cDNA sequences have also been reported in zebrafish and salmonids (Archer et al., 2008; Cruz-Garcia et al., 2009). As far as we know, LXRs have not been characterized in typically marine fish, but searches in the AQUAFIRST gilthead sea bream database(www.sigenae.org/aquafirst)identified as LXRα (E-value 8×10^–60) a contig of 792 bp in length. This sequence has been introduced into GenBank (FJ502320),and the deduced amino acid sequence for the C-terminal region (142 amino acids) shares 95–97%, 85% and 80% identity with fish, chicken and mammalian counterparts, respectively.

Taking all the above findings into account, the first goal of the present study was to analyze how LPL and HSL are regulated in concert in the adipose tissue of juvenile fish with lean and fat phenotypes in gilthead sea bream. Secondly, the role of PPARs and LXRα as targets for the TNFα-induced lipolysis was monitored in freshly isolated adipocytes to evidence the existence of different mechanisms regulating the size of the adipose tissue.

Two-year-old gilthead sea bream (Sparus aurata L.) were reared from fingerlings in the indoor experimental facilities of the Institute of Aquaculture Torre de la Sal (IATS, Spain). The experimental set-up was conducted in May–June under natural conditions of light (16 h:8 h light:dark) and temperature (19–23°C) at the IATS latitude (40 deg.5′N; 0 deg.10′E). At sampling time, overnight fasted fish were randomly selected and killed by decapitation under anaesthesia (3-aminobenzoic acid ethyl ester, MSS-222; 100 μg ml^–1). Liver and whole right fillets (denuded of skin and bone) were excised, frozen in liquid nitrogen and stored at –80°C until analyses. Mesenteric adipose tissue was also excised, a small piece was frozen, and the remaining tissue was processed for adipocyte isolation. All procedures were carried out according to national (Consejo Superior de Investigaciones Científicas,Institute of Aquaculture Torre de la Sal Review Board) and current EU legislation on the handling of experimental animals.

Adipocytes were individually isolated from 20 fish as described elsewhere(Albalat et al., 2005a). Briefly, mesenteric adipose tissue was cut into thin pieces and incubated in a shaking water bath at 18°C for 60 min with Krebs–Hepes buffer (pH 7.4) pre-gassed with 5% CO₂ in O₂, containing collagenase type II (130 U ml^–1) and 1% bovine albumin serum(BSA). The cell suspension was filtered through a double layer of nylon cloth and then washed three times by flotation. Cells were carefully resuspended in Krebs–Hepes buffer containing 2% BSA at a density of 7×10⁵ cells ml^–1. Cells were counted using a Fuchs-Rosenthal chamber. Aliquots of 1 ml of this final adipocyte suspension were incubated in polypropylene tubes for up to 3 h at 22°C in the absence or presence (100 ng ml^–1) of recombinant human TNFα(Sigma-Aldrich, Madrid, Spain).

At the end of the incubation time, the cells were centrifuged at 18,000 g for 2 min at 4°C, and 300 μl of medium were placed into perchloric acid to give a final concentration of 7% (v/v). Perchloric acid was neutralised for the measurement of glycerol concentration as an index of lipolysis using a spectrophotometric method(Tebar et al., 1996). The remaining medium was removed, and lysis reagent was added for RNA extraction(see below). All products were obtained from Sigma-Aldrich. Control and experimental conditions were conducted in triplicate for each animal and cell preparation.

Total RNA from adipose tissue and adipocytes was isolated using the ABI PRISM™ 6100 Nucleic Acid PrepStation (Applied Biosystems, Foster City,CA, USA). Briefly, tissue and cell samples were homogenized with a guanidine-detergent lysis reagent at given ratios for adipose tissue (25 mg tissue ml^–1) and isolated adipocytes(2.8×10⁶ cells ml^–1). The reaction mixture was treated with proteinase K, and RNA purification was achieved by passing the lysates through a purification tray containing an application-specific membrane. Wash solutions containing DNAse were applied, and total RNA was eluted into a 96-well PCR plate. The RNA purity was checked by absorbance measurements (A_260/280) and was always within the ideal range (1.9–2.1). Reverse transcription (RT) with random decamers was performed with the High-Capacity cDNA Archive Kit (Applied Biosystems). For this purpose, 500 ng of total RNA was reverse transcribed into a final volume of 100 μl. RT reactions were incubated for 10 min at 25°C and for 2 h at 37°C. Negative control reactions were run without reverse transcriptase.

Transcript measurements of lipid enzymes (LPL, HSL), lipid transcription factors (PPARα, PPARβ, PPARγ, LXRα) and lipolytic cytokines (TNFα) were made using an iCycler IQ Real-time Detection System (Bio-Rad, Hercules, CA, USA) as described elsewhere(Calduch-Giner et al., 2003). Briefly, diluted RT reactions were used for PCR reactions in 25 μl volume. Each PCR well contained a SYBR Green Master Mix (Bio-Rad), and specific primers at a final concentration of 0.9 μmol l^–1 were used to obtain amplicons of 77–192 bp in length(Table 1). β-actin was used as housekeeping gene, and the efficiency of PCR reactions for the target and the reference gene varied between 92% and 96%, respectively. The dynamic range of standard curves (serial dilutions of RT-PCR reactions) spanned five orders of magnitude, and the amount of product in a particular sample was determined by interpolation of the cycle threshold value. The specificity of reaction was verified by analysis of melting curves and by electrophoresis and sequencing of PCR-amplified products. Reactions were performed in triplicate and the fluorescence data acquired during the extension phase were normalized to β-actin by the delta-delta method(Livak and Schmittgen, 2001). No significant changes in β-actin expression were found within individuals and cell preparations.

Freeze-dried samples of liver and flesh were used for lipid content determinations. As established in routine procedures, sample aliquots of 0.5 g were desiccated (105°C for 3 h) in porous receptacles before Soxhlet extraction with 50 ml diethyl ether at 120°C (Soxhlet 4001046 Auto Extraction Apparatus; Selecta, Barcelona, Spain).

Data values were checked for normal distribution and homogeneity of variances, and when necessary arcsin transformation was performed before Student's t-test analysis comparing data on growth, adiposity and gene expression in fish with lean and fat characteristics. In isolated adipocyte incubations, lipolytic rates and gene expression after TNFαtreatment were normalized to control values (isolated adipocyte incubations without TNFα) and analyzed for statistical significance. All procedures were performed using SPSS v. 14.0 (SPSS, Chicago, IL, USA).

Adiposity and biometric parameters of fish used in the study are shown in Table 2. Two major groups of fish with lean and fat characteristics were considered on the basis of the mesenteric fat index (MFI): (1) the 10 animals with the highest MFI(1.69–2.86%) were clustered in the `fat group' and (2) the other 10 fish with a reduced MFI (1.01–1.46%) and 10% reduction in the average body mass (M<sub>b</sub>) were put in the `lean group'. Condition factor(fish mass and length³ quotient) and fillet lipid levels (% wet matter) did not differ significantly between groups, although the observed values were slightly higher in the fat than in the lean group. In the same way, a twofold increase in the absolute amount of liver lipids (mg liver lipids per 100 g body mass) was found in fat vs lean fish.

The expression pattern of LPL and HSL in adipose tissue is shown in Fig. 1A. The fat group showed a higher level of LPL transcripts than the lean group (P<0.05),which resulted in an increased LPL/HSL ratio (P<0.05) in the absence of major changes in HSL expression. The expression level of PPARγ and PPARα was also significantly upregulated in the fat group (P<0.05). Conversely, the expression pattern of PPARβ,LXRα and TNFα did not vary significantly with the change of growth parameters and body adiposity (Fig. 1B).

The TNFα-induced effects were tested in isolated adipocytes from the two fish groups. The analyzed response included measurements of glycerol release and mRNA levels of PPARs and LXRα. Both in fat and lean fish,two patterns of adipocyte response were identified on the basis of TNFα-induced lipolysis (glycerol release). In fish with fat characteristics, the group of TNFα-responders (F-R) was composed of three fish (380.6±19.5 g) with an enhanced lipolysis (more than 40%over controls, adipocytes without TNFα)(Fig. 2A) in which expression of PPARβ was significantly lower than that of controls(Fig. 2B); the group of non-responders (F-NR) was made up of seven fish with a lower body mass(324.3±6.7 g), and no apparent effects of TNFα on lipolysis(Fig. 2C) and transcriptional activity were observed (Fig. 2D). Adipocytes from lean fish also showed two different patterns of response corresponding to L-R and L-NR fish: (1) the L-R group comprised six fish (316±14.1 g) with a significant increase in lipolysis (>50%above controls) (Fig. 3A) and no detectable effects on the expression of LXRα and PPARs(Fig. 3B); (2) the L-NR group comprised four fish (285±20 g) with a low lipolytic response (less than 20% above controls) (Fig. 3C)accompanied by a significant downregulated expression of PPARγ and LXRα (Fig. 3D).

Gilthead sea bream is a highly valuable fish for the Mediterranean aquaculture, with an aquaculture production of 125,355 tonnes in 2007(www.fao.org). However, a particular problem concerning both health and quality of farmed fish is related to energy regulation and, particularly, lipid homeostasis. In the current study, a relatively high variability in adiposity was present in fish from the same batch and similar mass range, which happens even under improved management and culture conditions(Benedito-Palos et al., 2007; Benedito-Palos et al., 2008). Thus, two groups of fish were clearly separated according to the MFI, which increased in parallel with liver fat deposition. Our current understanding of the mechanisms by which an excess of lipid deposition progresses to hepatic steatosis is limited in gilthead sea bream and fish in general(Sitjà-Bobadilla et al.,2003; Sitjà-Bobadilla et al., 2005). However, this liver injury may reflect a wide range of lipid and lipoprotein metabolic disorders, including impaired insulin sensitivity and defects of lipid trafficking and lipoprotein processing. Hence, the current consensus is that lipid metabolic disorders are part of a common pathology that has not been properly defined in cultured fish, although it is becoming more and more evident with the currently intensive production systems (Farrel, 2002).

TNFα is synthesized and secreted from adipocytes and, hence, is in a key position to play a paracrine/autocrine role in the control of fat adipose mass. Several clinical studies have reported increased levels of TNFα in the blood of obese patients with signs of insulin resistance or dyslipidemia(Kern et al., 2001; Skurk et al., 2007). However,conflicting results using human and mouse obese models indicate that TNFα expression is increased only in the more extreme forms of obesity(Warne, 2003). In these cases,the stimulated TNFα production acts on the adipocyte to shift lipid metabolism from lipid accumulation towards lipid mobilisation(Fonseca-Alaniz et al., 2007; Guilherme et al., 2008; Skurk et al., 2007). The anti-adipogenic and lipolytic effects of TNFα have also been demonstrated in fish (Albalat et al.,2005b; Bouraoui et al.,2008), and interestingly the expression of TNFα is seasonally upregulated in gilthead sea bream with the replenishment of liver and mesenteric fat depots (Saera-Vila et al., 2007). Thus, in the present study, the lack of changes in TNFα expression with the increase in MFI suggests that fat fish have not reached their peak of fat storage capacity and continue increasing the size of their body fat depots. Consistent with this, the expression of LPL, a key limiting enzyme of tissue fatty acid uptake, was twofold higher in fat fish than in lean fish. This enzyme modulation is not surprising given the conservation of TNFα regulatory elements in the proximal 5′-flanking region of gilthead sea bream LPL(Saera-Vila et al., 2007). This agrees with the observation that LPL activity and expression are upregulated by insulin treatment in the adipose tissue of gilthead sea bream(Albalat et al., 2007). Moreover, experimental evidence indicates that the age-related changes in the tissue-specific profile of LPL may drive the redistribution of fat depots from mesenteric adipose tissue to skeletal muscle(Saera-Vila et al., 2007).

HSL is the rate-limiting step for the breakdown of stored triglycerides to glycerol and fatty acids, which are released into the plasma to be used as metabolic fuels in other tissues. HSL activity has been well characterized in the Antarctic fish Trematomus newnesi(Hazel and Sidell, 2004) but,as far as we know, there have been no reported expression studies in fish tissues until now. In mammals, the short-term regulation of HSL is carried out by reversible phosphorylation and translocation to the surface lipid droplets in response to catecholamines and other lipolytic hormones. However, the long-term regulation takes place at the transcriptional level, and HSL mRNA levels are affected by hibernation, fasting and even severe obesity(Holm et al., 2000). This may also be the case in the present study, and the lack of changes in HSL expression may indicate that the higher MFI of fat fish was primarily due to increased lipid deposition rather than to inhibition of lipolysis. This kind of regulation, in which LPL expression is reduced whereas HSL expression remains at control levels, has also been found in other physiological situations such as lactation in rats (Holm et al., 2000).

As expected, both in this and previous gilthead sea bream studies(Diez et al., 2007; Leaver et al., 2005),PPARγ is the most highly expressed PPAR isotope in the adipose tissue. Moreover, the expression of PPARγ was enhanced in the group of fat fish,which agrees with the idea that this nuclear factor is the master regulator of adipocyte differentiation that stimulates the expression of adipogenic enzymes such as LPL and adipocyte fatty acid binding protein(Rosen et al., 1999; Tontonoz and Spiegelman,2008). When comparing fat and lean phenotypes, we also found a similar expression pattern for the other two PPAR isotopes, although a significant increase in transcript levels was only reported for PPARα. Earlier studies in fish indicate that the expression of PPARα is increased by fasting and the tissue oxidative capacity(Leaver et al., 2005; Leaver et al., 2008). Therefore, the current increase in the expression of PPARγ and PPARα can be viewed as a part of a counter-regulatory system, which may be tissue- and species-specific.

The isolated adipocyte system also evidenced a complex lipid metabolic network, and most fish with fat characteristics (F-NR group) were refractory to TNFα-induced lipolysis. In this group of fish, we also failed to detect any transcriptional effect on LXR and PPARs. This lack of response can be viewed as a steady state with an enhanced refractoriness to lipolytic TNFα action. This is not surprising given the aforementioned increase in the overall LPL/HSL ratio in fish with fat characteristics. At the same time,however, some fish with fat characteristics (F-R group) were sensitive to TNFα-induced lipolysis, which can be interpreted as a selective advantage or adaptive response to limit the size increase of the adipose tissue mass. This agrees with the biggest size of this group of fish, which was probably associated with enhanced feed intake. The precise mechanisms underlying the TNFα-induced lipolysis remain to be fully elucidated,although it might be mediated by the downregulated expression of PPARβ. In mammals, this isotope differs from the other two PPAR isotopes by its almost ubiquitous tissue expression, which suggests a general housekeeping role (Kliewer et al., 1994; Schmidt et al., 1992). However, it has subsequently become clear that this is not true, and analyses in a PPARβ null mouse model (db/db) demonstrate that PPARβdeficiency is associated with multiple developmental and metabolic abnormalities, including demyelization and diminished adipose tissue mass(Barak et al., 2002; Peters et al., 2000). Moreover, PPARβ-specific agonists improve insulin sensitivity, suppress hepatic glucose output and inhibit free fatty acid release from adipocytes in the db/db mouse (Lee et al.,2006), which indicates that high-affinity PPARβ ligands would be useful drugs to effectively target insulin resistance, hyperglycemia and dyslipidemia (Seedorf and Aberle,2007).

In lean fish, dual effects of TNFα were also evidenced in isolated adipocyte cultures. In this case, most fish (L-R group) were highly sensitive to TNFα action, but the induced lipolysis was not transcriptionally mediated by PPARs and LXRα. This observation suggests an intriguing participation of catecholamine receptors and protein kinases, as recently reviewed in mammals (González-Yanes and Sánchez-Margalet, 2006). Alternatively, in the absence of TNFα-induced lipolysis (L-NR group), the downregulated expression of PPARγ and LXRα suggests that the primary action of TNFα on adipocytes from lean fish would be the inhibition of lipid deposition rather than the enhancement of lipid mobilization. Indeed, PPARγ is clearly involved in the activation of lipogenic enzymes and adipocyte differentiation,and its expression is reduced by TNFα in human hepatoma Hep3B cell line(Kim et al., 2007) and mice adipose tissue (Ye, 2008). Experimental evidence also indicates that LXRα upregulates the expression of lipogenic genes such as acetyl-CoA carboxylase, fatty acid synthase and LPL (Al-Hasani and Joost,2005; Zhang et al.,2001). Moreover, PPARγ agonists increase LXRαexpression in different experimental models such as primary cultures of human and murine macrophages, 3T3-L1 adipocyte cells and rat epidymal adipose tissue(Chinetti et al., 2001; Wójcicka et al., 2007). Conversely, PPARγ and LXRα mRNA levels are significantly reduced by TNFα treatment in rabbit adipocytes, and their decrease is accompanied by a reduced cholesterol efflux as the result of a transcriptional cascade mediated by LXRα (Chawla et al., 2001; Zhao and Dong,2008).

In summary, we present new insights on fish lipid metabolism, addressing the gene expression of some relevant genes for the regulation of adipose tissue mass. The results highlight the increased LPL/HSL expression ratio in fish with fat characteristics and increased MSI. Also, the target genes for TNF-induced lipolysis were different in fish with lean and fat characteristics, which reflects the different metabolic capabilities and/or mechanisms operating in lean and fat fish to limit the size increase of the adipose tissue mass (see Fig. 4for schematic drawing). Thus, the TNFα-induced lipolysis in adipocytes from fat fish was transcriptionally mediated by the reduced expression of PPARβ whereas other signalling pathways enhancing lipolysis(post-transcriptional mediated effects) or inhibiting adipogenesis(PPARγ- and LXRα-mediated effects) orchestrated the TNFα-mediated effects in lean fish. The practical consequences of these findings remain to be explored but they open new research opportunities for genetics and comparative physiologists (e.g. linkage studies of fish adiposity and allele polymorphism in TNFα and associated target genes).

LIST OF ABBREVIATIONS

 
• HSI

  hepatosomatic index

 
• HSL

  hormone-sensitive lipase

 
• K

  condition factor index

 
• LPL

  lipoprotein lipase

 
• LXR

  liver X receptor

 
• MFI

  mesenteric fat index

 
• PPAR

  peroxisome proliferator-activated receptor

 
• RT

  reverse transcription

 
• TNFα

  tumour necrosis factor-α