Despite the predominant role of adipose tissue in ruminant’s de novo fatty acid synthesis, the liver also plays an important role in ruminant lipid metabolism [1]. This organ carries out central metabolic functions in various aspects of lipid and lipoprotein metabolism, such as uptake, oxidation and metabolic conversion of non-esterified fatty acids (NEFA), synthesis of cholesterol and phospholipids, and formation and secretion of specific classes of lipoproteins [1]. The ruminants’ liver removes little or no triacylglycerols from blood lipoproteins [2]. Uptake of NEFA is the predominant route by which fatty acids are supplied to the liver [3] and, thus, plasma lipid fatty acid composition should influence the liver fatty acid metabolism and composition [2]. Consequently, the regulation of these liver metabolic pathways may affect fatty acid deposition into lipids of ruminant products [4].

Interest in n-3 long-chain polyunsaturated fatty acids (n-3 LC-PUFA) has increased since it was found that their consumption in most Western populations, particularly those of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), is sub-optimal for protection against the most prevalent chronic diseases [5]. In grazing ruminants, α-linolenic acid content of muscles increases with the concomitant increase in n-3 LC-PUFA contents [6]. In contrast, although the addition of linseed to ruminant diets [7, 8] increases the α-linolenic acid content of muscles, the n-3 LC-PUFA levels stay unchanged or increase only slightly. In fact, Bessa et al. [9] reported that lucerne supplementation with linseed oil promoted an increase in ALA coupled with a decrease in n-3 LC-PUFA in lambs, when compared to the control diet (lucerne only). According to the authors, these results suggest the inhibition of α-linolenic acid metabolism by vegetable oils rich in n-3 PUFA. Therefore, the abundance of n-3 LC-PUFA in ruminants’ tissues appears to depend not only on dietary n-3 PUFA but also on their endogenous synthesis via elongation and desaturation of dietary n-3 PUFA.

The biosynthesis of DHA from α-linolenic acid is performed through alternating steps of desaturation and elongation, followed by a final step of peroxisomal β-oxidation. This metabolic pathway involves two desaturases (Δ5 and Δ6 desaturases), two elongases (elongases 2 and 5) and enzymes of the peroxisomal β-oxidation [10]. The activity of these enzymes is currently regarded as potential limiting steps in this biosynthesis, possibly in a tissue dependent manner. However, recent studies addressed the role of the liver in ruminants’ lipid metabolism, either using in vitro experiments [4] or in vivo assays [11]. These experiments raised some interesting clues on hepatic lipid metabolism, namely the extensive catabolism of α-linolenic acid [4] and the low or negligible expression level of genes encoding for enzymes of fatty acid desaturation and elongation [11]. Therefore, the role of bovine liver, as a central metabolic organ, on lipid metabolism remains to be elucidated.

An experiment with 40 young bulls from two genetically diverse beef cattle breeds, Alentejana and Barrosã, fed either high (70% silage/30% concentrate) or low (30% silage/70% concentrate) silage diets was carried out by our group to study the breed and diet effects on lipid metabolism. Previous reports from this experiment [12, 13] showed that these breeds have a distinct response to the variation in dietary silage level, as assessed by the fatty acid composition and the mRNA levels of key lipogenic factors of the main fat depots and muscle. Bearing this in mind, as well as the studies by Gruffat et al. [4] and Cherfaoui et al. [11], we aimed to investigate whether the same breed-specific response to dietary silage level would be observed in the liver. For this purpose, the detailed fatty acid composition of liver from Alentejana and Barrosã bulls, in parallel to their mRNA levels of key genes associated with lipid metabolism, were determined.

The present study was based on an experiment with two genetically different bovine breeds with distinct maturity rates, Alentejana and Barrosã, fed diets with different silage to concentrate ratio (30/70% vs. 70/30%). Alentejana and Barrosã breeds, despite being phylogenetically distant [14] share more genetic similarities than the breeds used in previous studies addressing the differences between breeds in fatty acid metabolism, mainly based on the Japanese Black and Holstein breeds [15–17]. We observed in these animals that different fat depots, subcutaneous and mesenteric adipose tissues, had distinct features regarding cellularity and fatty acid composition [12]. Results indicated that genetic background and, to a lesser extent diet composition, determine fat content and composition, pointing out to a differential fat partitioning between subcutaneous and intramuscular fat in Alentejana and Barrosã breeds. In addition, the comparison of fatty acid composition and gene expression levels between the muscle and subcutaneous adipose tissue indicated that these tissues play distinct roles in lipid metabolism, reinforcing the prevailing role of the subcutaneous adipose tissue over intramuscular fat in the de novo fatty acid synthesis (data not shown). In fact, the PCA showed a clear separation of the liver from the remaining tissues (muscle, and subcutaneous and mesenteric adipose tissues), to which contributed the fact that this tissue is particularly rich in α-linolenic acid and the products of its desaturation and elongation, the n-3 LC-PUFA. This result suggests that the liver may have an important role in the n-3 LC-PUFA biosynthesis.

The major metabolic fates of long chain fatty acid CoA in the liver are: i) esterification into triacylglycerols and, to a lesser extent, into phospholipids and cholesterol esters, ii) complete oxidation to CO₂ or incomplete oxidation which generates acetate and ketone bodies [1]. De La Torre et al. [18] reported that, once incorporated in vitro into bovine hepatic cells, rumenic and oleic acids are highly catabolised through the β-oxidation pathway. One possible explanation might be the low efficiency of bovines in secreting fatty acids from the liver [19], thus directing fatty acids preferentially towards the oxidative pathway.

The liver is responsible for the uptake of NEFA and subsequent storage as triacylglycerols or release as VLDL. Hepatic triacylglycerols synthesis is the result of various pathways of lipid metabolism, including fatty acid uptake from plasma, oxidation of fatty acids, de novo synthesis of fatty acids and secretion of triacylglycerols via VLDL [20, 21]. Apart from hormones, the metabolic regulatory events in the liver are directed by metabolites levels, either in excess (e.g., NEFA) or in shortage (e.g., glucose) [22]. Once in the hepatocytes, catabolism of fatty acids is mostly directed towards the synthesis of ketone bodies for energy utilisation by tissues [4, 23]. In spite of the higher insulin levels observed in the bulls fed the low silage diet (data not shown), when compared to those fed the high silage diet, it had no impact on INSR mRNA levels in the liver. Zhang et al. [24] reported low INSR gene expression levels in calf cultured hepatocytes in response to high insulin concentrations. However, it should be noted that in the present study even the highest insulin levels were within physiological levels.

Due to gluconeogenesis in liver, both propionate and lactate can indirectly modulate adipose tissue lipogenesis through increased glucose availability. The glucogenic effect of high-starch diets is often accompanied by an enhanced insulin response [25]. The diacylglycerol acyltransferase (DGAT) catalyses the final step in triacylglycerol biosynthesis by converting diacylgycerols and fatty acyl-coenzyme A (CoA) into triacylglycero1s [26]. The DGAT1 gene has also been related to increased hepatic triacylglycerol synthesis in dairy cows [27]. Moreover, mRNA levels of this gene appeared to be susceptible to diet-induced hyperinsulinaemia pre-partum [27]. In the present study, the relative expression levels of DGAT1 was higher in the liver from Barrosã bulls, when compared to Alentejana bulls, although plasma insulin was affected by diet and not by breed as reported in a companion paper [12].

It was suggested that bovine liver converts more efficiently linoleic acid into arachidonic acid than α-linolenic acid into n-3 LC-PUFA [22]. In line with this, in the present study the ARA acid percentages were the highest amongst LC-PUFA. The PCA from the fatty acid pooled data showed a clear separation of the tissues analysed, with particular emphasis on the liver. The LA and α-linolenic acid, along with ARA acid, were identified as the most contributing for the distancing of the liver cluster from those of the remaining tissues. This observation could be explained by the higher phospholipid/triacylglycerol ratio in the liver when compared to the other tissues. Given that LA and α-linolenic acid originate from diet, this suggests that the liver plays a role in their metabolism, possibly through their desaturation and elongation.

Attending to the putative role of liver in fatty acid elongation and desaturation, as described by others [4, 11] and as suggested by the results herein presented for hepatic fatty acid composition, this work was focused on the analysis of gene expression levels of enzymes and transcription factors associated with the n-3 and n-6 LC-PUFA pathway synthesis. The PPARA, a transcription factor that acts as an important regulator of lipid metabolism and energy homeostasis, plays a key role in the control of the pathways involved in fatty acid uptake, fatty acid binding, fatty acid oxidation, ketogenesis, as well as carnitine synthesis [28]. The fact that the mRNA level of CPT1A, which is controlled by PPARA, was unchanged across experimental groups could indicate that there was no activation of PPARA in the liver. Moreover, activation of PPARA is known to be caused by increased plasma concentrations of NEFA. However, in the present study, there were no significant changes in NEFA plasma levels, which is consistent with the similar PPARA levels across experimental groups.

Polyunsaturated fatty acids that escape β-oxidation could be converted into longer and/or more unsaturated fatty acids by the elongation-desaturation pathway, LA and α-linolenic acid being metabolised, theoretically, into 20:4n-6 and 22:6n-3, respectively [29, 30]. In a recent study by Gruffat et al. [4], the conversion of α-linolenic acid into longer and/or more unsaturated fatty acids was not detected, yet 13.5% of LA was converted into ARA. It is widely accepted that members of the n-6 and n-3 families compete for the elongation-desaturation pathway [31]. Furthermore, the conversion of α-linolenic acid and LA into their longer chain homologues is greatly determined by the composition of dietary fats [32]. Harnack et al. [33] suggested that a ratio of 1/1 would lead to the highest formation of n-3 LC-PUFA, given that the conversion of n-3 fatty acids into higher homologues may depend on the ratio of ingested n-6/n-3 fatty acids.

The carnitine palmitoyltransferase (CPT) system is an essential step in the β-oxidation of fatty acids, including LC-PUFA. Gruffat et al. [4] showed that the rate of α-linolenic acid oxidation was higher than that of LA, possibly due to higher mitochondrial activity of CPT1A, as suggested by Ide et al. [34]. The positive correlations between CPT1A mRNA levels and the main SFA and MUFA, as well as the negative relationship with some PUFA (20:2n-6 and 22:4n-6) may be attributed to the role of the enzyme in fatty acid β-oxidation and ketogenesis. The higher SFA and MUFA percentages are a consequence of increased triacylglycerol and NEFA in the hepatic tissue, which would in turn promote fatty acid β-oxidation in order to prevent their excessive accumulation.

SREBF1 is a fundamental regulator of fatty acid biosynthesis, suggesting that it could be a key point of control of membrane lipid homeostasis capable of strongly influencing the lipid composition of membranes [35]. The SREBF1 gene regulates a wide array of genes involved in lipid biosynthesis. In the present work, the SREBF1 expression was shown to be correlated with the mRNA levels of FASN, SCD, PPARA and INSR (data not shown). The fatty acid synthase, encoded by FASN, plays a central role in de novo lipogenesis in mammals. However, the FASN mRNA levels were similar across experimental groups and showed no correlation with the main liver fatty acids, thus suggesting no modulation by diet or silage level under these experimental conditions.

The SCD gene encodes for an important enzyme in unsaturated fatty acid synthesis [36]. SCD activity in bovine liver and duodenal mucosal cells has been reported [37–39]. Furthermore, Chung et al. [39] suggested that some portion of the MUFA in adipose tissues may arise from hepatic and mucosal desaturation of dietary SFA. However, in the present work, there was no apparent association between SCD gene expression levels and the desaturation indices. This finding reinforces the concept of a low SCD activity in the liver, which is not to say that there is no desaturation at all. The matter of fact is that hepatic desaturation activity seems to be carried out mainly by the enzymes encoded by FADS1 and FADS2.

The higher expression levels of FADS1 and FADS2 in the low silage than in the high silage fed bulls, particularly Alentejana bulls, is concomitant with the diet effect observed in the percentages of most LC-PUFA and, therefore, with the correlations found. That is, feeding the low silage diet, poorer in 18:2n-6 and α-linolenic acid than the high silage diet, promoted the expression of the genes encoding Δ5 and Δ6 desaturases in order to incorporate an adequate level of LC-PUFA in membrane phospholipids. The increase in the FADS1 gene expression could have contributed to promote lipid biosynthesis and fatty acid deposition. The FADS1 gene has been considered as one of the rate-limiting enzymes to the endogenous formation of LC-PUFA in humans [40]. In mammals, FADS1 converts dihomo-γ-linolenic acid (20:3n-6) to ARA and eicosatetraenoic acid (20:4n-3) to EPA with, respectively, LA and α-linolenic acid as the initial substrates [40, 41].

Liver seems to be highly active in α-linolenic acid catabolism [18], thus limiting its subsequent availability for deposition in muscles. There are two basic metabolic fates for α-linolenic acid. First, it is subjected to β-oxidation and extensive carbon recycling. Second, it is converted into longer fatty acids via the elongation and desaturation pathway. Elongation of C18 in the LC-PUFA pathway occurs through elongase enzymes, which have been suggested to have a regulatory role on LC-PUFA synthesis and may be transcribed from one or more genes (ELOVL2 and ELOVL5) [42]. Hepatic LC-PUFA metabolism has been linked to energy balance and physiological state in dairy cattle. Long-chain fatty acyl elongases (ELOVL) are endoplasmatic reticulum membrane-bound proteins responsible for the first regulatory step in the fatty acid elongation pathway (condensation of activated fatty acids with malonyl-CoA), elongating fatty acid that are biosynthesized de novo or supplied by the diet [43]. In mammals, ELOVL2 has greatest activity in the elongation of C20 and C22 but low or, in the case of humans, no activity towards C18 PUFA [44]. In contrast, mammalian ELOVL5 is very active towards C18 PUFA but does not appear to have the capacity to elongate beyond C22 [44, 45]. Cherfaoui et al. [11] proposed that the limiting step for elongation of α-linolenic acid in the muscle tissues is the absence of ELOVL5 protein. Moreover, these authors also suggested that the low levels of DHA in the muscle tissues could be a consequence of its preferential peroxidation or of its preferential uptake by other tissues.

Hepatic ELOVL2 and ELOVL5 are both regulated by SREBP transcription factors in mouse [46]. Nonetheless, in the present study there seemed to be no direct correlation between SREBF1 mRNA levels and the expression of ELOVL2 and ELOVL5 genes, in spite of SREBF1 being correlated with total n-6 PUFA. The higher ELOVL2 gene expression, and thus the higher fatty acid elongase with high activity in the final steps of LC-PUFA biosynthesis, in the Barrosã than in the Alentejana bulls could provide a route to promote EPA and DHA deposition. However, LC-PUFA contents in the liver from the present study were influenced, not by breed, but by diet.

In summary, the present study suggests that liver has, among the bovine lipogenic tissues, a specific role in lipid metabolism. In addition, the results indicate a breed modulation of hepatic desaturation/elongation of fatty acids, possibly through the differential expression of genes encoding for enzymes involved in the desaturation and elongation pathway. In some cases, the response to varying silage levels was modulated by the genetic background (FADS2 and PPARA), whereas in others (DGAT1 and ELOVL2) there was a clear breed effect regardless of diet composition. The small differences observed in gene expression levels could have additive effects, which may explain the differences found in the hepatic fatty acid profile, particularly in the LC-PUFA. In summary, the results herein discussed are in line with the previous reports from this experiment, thus stressing the importance of devising custom-made feeding strategies which take into account the genetic background.