Distinct correlations between lipogenic gene expression and fatty acid composition of subcutaneous fat among cattle breeds

Background The fatty acid (FA) composition of adipose tissue influences the nutritional quality of meat products. The unsaturation level of FAs is determined by fatty acid desaturases such as stearoyl-CoA desaturases (SCDs), which are under control of the transcription factor sterol regulatory element-binding protein (SREBP). Differences in SCD genotype may thus confer variations in lipid metabolism and FA content among cattle breeds. This study investigated correlations between FA composition and lipogenic gene expression levels in the subcutaneous adipose tissue of beef cattle breeds of different gender from the Basque region of northern Spain. Pirenaica is the most important beef cattle breed in northern Spain, while Salers cattle and Holstein-Friesian cull cows are also an integral part of the regional beef supply. Results Pirenaica heifers showed higher monounsaturated FA (MUFA) and conjugated linoleic acid (CLA) contents in subcutaneous adipose tissue than other breeds (P < 0.001). Alternatively, Salers bulls produced the highest oleic acid content, followed by Pirenaica heifers (P < 0.001). There was substantial variability in SCD gene expression among breeds, consistent with these differences in MUFA and CLA content. Correlations between SCD1 expression and most FA desaturation indexes (DIs) were positive in Salers (P < 0.05) and Pirenaica bulls, while, in general, SCD5 expression showed few significant correlations with DIs. There was a significant linear correlation between SCD1 and SRBEP1 in all breeds, suggesting strong regulation of SCD1 expression by SRBEP1. Pirenaica heifers showed a stronger correlation between SCD1 and SREBP1 than Pirenaica bulls. We also observed a opposite relationship between SCD1 and SCD5 expression levels and opposite associations of isoform expression levels with the ∆9 desaturation indexes. Conclusions These results suggest that the relationships between FA composition and lipogenic gene expression are influenced by breed and sex. The opposite relationship between SCD isoforms suggests a compensatory regulation of total SCD activity, while opposite relationships between SCD isoforms and desaturation indexes, specially 9c-14:1 DI, previously reported as an indicator of SCD activity, may reflect distinct activities of SCD1 and SCD5 in regulation of FA content. These findings may be useful for beef/dairy breeding and feeding programs to supply nutritionally favorable products.


Background
In recent years, consumers have expressed growing concern regarding the amount and types of dietary fat due to reported deleterious health effects of saturated and trans fatty acids (FAs) [1]. On the other hand, monounsaturated FAs (MUFAs) and polyunsaturated FAs (PUFAs) are recognized as beneficial for human health [2]. The FA composition of meat influences the lipid melting point [3], and an increase in the ratio of MUFA to saturated FA (SFA) increases fat softness, thereby improving palatability. Thus, enhancing MUFA content improves both the quality and nutritional value of animal products [4].
Pirenaica is the most important beef cattle breed raised in the Basque region of northern Spain and is highly appreciated both for its value as a genetic resource as well as for the production system that has developed around it. In addition to Pirenaica, Salers, a rustic cattle breed used for beef production, has grown in importance due to its ready adaptability to local management and environmental conditions [5]. Finally, Holstein-Friesian, primarily as cull dairy cows, are also an integral part of the regional beef supply chain.
There have been a number of studies investigating associations between lipogenic enzyme genotype and FA composition in cattle. In the subcutaneous and intramuscular fat depots of beef cattle, the majority of SFA conversion to MUFA is catalyzed by stearoyl-CoA desaturase (SCD, EC 1.14.19.1 or Δ9-desaturase) [6]. In addition, SCD enzyme can also catalyze the conversion of substrates like vaccenic acid (11 t-18:1) to its corresponding conjugated linoleic acid (CLA) isomer (9c,11 t-18:2 or rumenic acid). The association of SCD1 genotype with FA composition has been previously investigated in Japanese Black [7], Canadian Holstein [8], Fleckvieh [9] and crossbred cattle [10]. In addition to regulation of FA profile by SCD1, the novel Δ9-desaturase isoform SCD5, previously found in humans, has also been identified in cattle which shares 65% identity at the amino acid level [11]. Further, a relationship between genetic polymorphisms in SCD5 and the ratio of SFA to unsaturated FA (UFA) has been reported in Holstein milk fat [12]. It thus appears that both bovine isoforms SCD1 and SCD5 contribute to FA composition. Therefore, the mechanism by which SCD isoforms are activated is a major determinant of FA composition and of great interest to breeders.
Da Costa et al. [13] reported a correlation between SCD1 expression and FA composition of subcutaneous fat in Portuguese cattle, whereas the expression of SCD5 and its relation to the subcutaneous FA profile was not investigated. The SCD1 gene is controlled by the key transcription factor SREBP1 [14]. In Japanese Black beef cattle, SREBP1 polymorphisms have been associated with FA composition [15]. Alternatively, transcriptional regulation of bovine SCD5 remains unclear, although a recent study using human choriocarcinoma trophoblastic cells (JEG3) reported that SREBP1 can bind to the SCD5 promoter [16].
The complex associations between the biochemical pathways regulating fat content and genetic variability of lipogenic genes are not yet fully understood in European cattle breeds, although recent studies have begun to elucidate these relationships in a specific genetic background of Japanese Black cattle [17]. The objectives of the present study were to investigate the expression levels of three key genes controlling Δ9-desaturated FA content, SCD1, SCD5, and SREBP1, and their associations with the FA composition of subcutaneous adipose tissue from the major commercial cattle breeds produced in northern Spain, Pirenaica, Salers, and Holstein-Friesian. Based on the findings of this study, we discuss how these associations may give information on the mechanisms of the differences in meat quality among these cattle commercial types.

Sample collection
In the present study, cattle commercial types typically destined for meat production in northern Spain (Basque region) were examined. Sample collection was designed according to data from the Bovine Identification Document and inferred relationships (parentage and sibships) computed from 29 microsatellites (Software Colony 2.0. 6.2) [18]. Neither parentage nor maternal half-sibs were observed, and paternal half-sibs were maintained at low frequencies (Pirenaica, 0.009; Salers 0.013; Holstein-Friesian, 0.019). A total of 100 subcutaneous adipose tissue samples were collected from pure breed cattle (13 Salers bulls, 37 Pirenaica bulls, 29 Pirenaica heifers, and 21 Holstein-Friesian) slaughtered in a local commercial abattoir (Urkaiko S. Coop., Zestoa, Gipuzkoa, Spain) during 12 days over 5 weeks in June and July 2014. Animals came from different farms [19].
Backfat samples were obtained from the left half carcass between the 5-6th ribs and stored in plastic bags with the air removed for FA analysis or preserved in RNAlater™ (Ambion, Austin, TX) for RNA analysis. All samples were transported to the laboratory in insulated coolers and stored at − 80°C until analysis.
Salers and Pirenaica were yearling calves with similar age (average of 12.9 ± 1.4 months), while Holstein-Friesian were cull cows (70.0 ± 19.43 months) at slaughter, which are regular ages of commercial types used for beef production in the region. Hot carcasses of Salers and Pirenaica commercial bulls were of similar weight (average of 325 ± 38.4 kg) while carcasses of Pirenaica heifers and Holstein-Friesian cows were markedly lighter (291 ± 59.6 and 253 ± 33.1 kg, respectively).
In the abattoir, conformation and degree of fat cover of each carcass were recorded. European regulations were followed for carcass classification at 24 h post-mortem [20] including the EUROP scale for conformation and a 1to-5 scale for fat cover scoring. Each level of both scales was divided in 3 sub-levels and transformed to a numerical scale ranging from 1 to 15, with 15 being the best conformation and the thickest fat cover.

RNA extraction and quantitative real-time PCR
A 100 mg sample of frozen subcutaneous fat tissue was disrupted and simultaneously homogenized to fine powder with a mortar and pestle under liquid N 2 . Total RNA was extracted using the RNeasy Lipid Tissue kit (Qiagen Inc., Valencia, CA, USA) following the manufacturer's instructions. An additional DNase digestion step was performed to remove any contaminating genomic DNA. Concentration and quality of the extracted RNA were assayed by measuring the 260 nm and 280 nm absorbance using a NanoDrop ND-1000 Spectrophotometer (Peqlab, Erlangen, Germany). Absorbance ratios (260/280) of all preparations were at least 1.8. Integrity of RNA was checked by denaturing agarose gel electrophoresis. Aliquots of RNA were stored at − 80°C and dehydrated in RNAstable 96-Well Plates (Biomatrica, San Diego, CA, USA) for long-term storage. Reverse transcription was performed in a 30 μL final reaction volume containing 250 ng total RNA, 3.3 μL RNase/ DNase-free water, 5 μL of 5 × RT buffer, 1.5 μL dNTPs, 0.8 μL RNAase inhibitor, 0.8 μL random primer, and 0. 8 μL high efficient ReverTra Ace reverse transcriptase (TOYOBO, Osaka, Japan). Cycle parameters were 30°C for 10 min, 42°C for 20 min, 99°C for 5 min, and 4°C for 5 min. Custom TaqMan Assays (Applied Biosystems, Foster City, CA, USA) were conducted to measure the relative expression levels of bovine SCD1, SCD5 and SREBP1 using the primers and FAM/TAMRA probes reported in Table 1. Each candidate gene was amplified in multiplex with an internal control (18S rRNA Endogenous Control VIC/TAMRA Probe, Primer Limited) by the co-application reverse transcription method (Co-RT) [27]. This multiplexing approach guarantees the same conditions (thus equal amplification efficiency) and same reverse transcriptase activity for both genes, thereby yielding better normalization and reproducibility. The reaction mixture included primers (10 μM each), FAM-labeled probe (10 μM), 0.6 μL of 18S RNA Endogenous Standard containing VIC-labeled probe and limited primers, and 2 × TaqMan Gene Expression Master Mix (7.5 μL) (Applied Biosystems). Real-time PCR was performed in triplicate using the ABI Prism 7500 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) with a standard twostep cycling program of 40 cycles at 95°C for 15 s and 60°C for 1 min. The average of the gene expression levels was used for further analyses. PCR efficiency was monitored by the increase in absolute fluorescence [28], mainly because this allows PCR efficiency calculation for individual samples/reactions and prevents problems arising from the use of standard curves. Raw data were obtained from the ABI Prism 7500 SDS software v1.4, exported in Rn format, and imported to LinRegPCR (Heart Failure Research Center, Amsterdam, the Netherlands). LinRegPCR determines baseline fluorescence, sets a window of linearity for each amplicon, and calculates the PCR efficiency (E) per sample and amplicons using an iterative algorithm. In this study, efficiencies were over 90% for all samples and correlation coefficients were higher than 0.99.
The comparative threshold cycle method (ΔCt) was employed to calculate relative gene expression based on the following formula:

Statistical analysis
Statistical analysis was conducted using IBM SPSS Statistics 22 for Windows (SPSS Inc., IBM Corporation, NY, USA). First of all, data was checked for normality and homoscedasticity. Then, the following general linear model Y ijk = μ + CT i + A j + HCW k + e ijk was used for analysis of variance (ANOVA), including commercial type (CT; Salers bulls, Pirenaica bulls, Pirenaica heifers, Holstein-Friesian cows) as fixed effect and age at slaughter (A) and hot carcass weight (HCW) as covariates. The effect of sire was also checked but not included in the model as it was statistically not significant. LSD post hoc test was applied for multiple comparison of means among commercial breeds studied. Simple linear regression analyses were also performed to investigate the relationship between genes (gene-gene) for each commercial type studied.
Finally, partial Pearson correlations coefficients adjusted for A and HCW were computed to determine the associations among gene expression (ΔCt) and FA (Δ9 DIs) data.
Three significant figures were used to express the data, and significance was declared at P < 0.05.

Discussion
Fat deposition and the FA composition of fat depots are controlled by a complex regulatory system including lipogenesis and lipolysis pathways. Adipose tissue is the main site for the storage of excess energy in the form of triacylglycerols, with the Δ9-desaturase product oleic acid (9c-18:1) being the predominant FA [29]. Therefore, Δ9-desaturase activity is critical for triglyceride storage in adipose tissue. While several pathways are involved in regulating FA composition, FAs produced from the precursors acetate and NADH, from the hydrolysis of triacylglycerols, and produced and deposited as rumen biohydrogenation metabolites can act as substrates for Δ9-desaturase. Adipose tissue develops in inter-and intra-muscular depots and both have a major impact on the quality and palatability of commercial beef. There is evidence for differential gene expression profiles in these two fat depots [30]. In this regard, the present study aimed to evaluate the regulation of SCD and SREBP1, genes strongly affecting the FA composition of subcutaneous adipose tissue, in three genetically diverse bovine breeds commercialized in the Basque region of northern Spain; Pirenaica, Salers and Holstein-Friesian [5,31,32]. Expression of SCD1 did not differ significantly between Pirenaica bulls and heifers or among young cattle of Salers and Pirenaica. This may be partially explained by a similar feeding regimen, typically including concentrates, when meat production is the final purpose (Fig. 1). However, the content of Δ9 products, such as cis-MUFA, was higher in Salers bulls and Pirenaica heifers than corresponding bulls ( Table 2). The Salers bulls and Pirenaica heifers, together with Holstein-Friesian cows, showed stronger correlations between SCD1 and SREBP1 compared to Pirenaica bulls (Fig. 2). This suggests that, in Pirenaica breed, the FA composition is affected by the lipogenic gene regulation in a sex-dependent manner. Similarly, in a crossbred study, heifers exhibited higher SCD1 mRNA levels and higher MUFA content than bulls in subcutaneous adipose tissue [33], and a possible effect of sex hormones on enzymatic systems affecting lipid metabolism has been suggested [34]. Indeed, the growth hormone, sexually differentiated in mammals, seems to increase SREBP1 and SCD1 gene expression in females [35]. Alternatively, age and diet have been demonstrated to influence adipocyte development in Pirenaica bulls [36]. Hence, the activation of SCD1 due to a potentially higher concentrate consumption [19,37] agrees with the greater total MUFA content of Salers and Pirenaica, while higher MUFA content in Pirenaica heifers than bulls seems to be more sex-dependent ( Table 2). The greater variability in SCD1 expression within Holstein-Friesian cows compared to young Pirenaica and Salers (Fig. 1) could be related to the less homogeneous diet and older age of these animals. Nevertheless, the generally lower SCD1 expression observed in Holstein-Friesian cows was also reported in other mature culled cows [4], in which linoleic acid (18: 2n-6) was suggested as the primary agent depressing SCD gene expression in adipose tissue [38].
We detected variability in SCD5 mRNA expression levels among breeds (P < 0.01) and generally greater expression of SCD1 relative to SCD5 in all breeds. Lengi and Corl (2007) [11] also reported over 40-fold greater expression of SCD1 compared to SCD5 in adipose tissue of bulls (albeit with unspecified feeding). Variation among breeds was observed in the expression of both SCD isoforms, especially between beef and dairy cattle breeds (Salers and Pirenaica vs. Holstein-Friesian), suggesting that even if FA differences are generally small, there may still be differences in the underlying lipogenic gene expression or enzyme profile [39].
These differences in SCD1 and SCD5 expression levels ( Fig. 1) also suggest that SCD5 expression is more breed dependent than SCD1 expression. However, it is also possible that SCD5 expression is more sensitive than SCD1 expression to other environmental factors (i.e., feeding) that differ among commercial types. Our results also revealed a potential opposite association between SCD isoforms within each breed. In general, this opposite correlation of DIs with SCD5 and SCD1 expression levels suggests that regulatory factors that upregulate SCD1 also downregulate SCD5 (and vice versa). However, since both SCD isoforms are expressed in adipose tissue, both may contribute to the maintenance of desaturation. In contrast to Pirenaica and Salers, this opposite association between SCD1 and SCD5 was not observed in Holstein-Friesian cows, a breed selected intensively for dairy production and most often utilized for beef production as cull cows at no specific age. This opposite pattern was more evident when the Holstein-Friesian sample was stratified by age (data not shown). Furthermore, Holstein-Friesian cows may exhibit side effects of dairy selection that differentially affect the genetic sequences containing SCD genes, thereby influencing transcriptional regulation. Moreover, variability in the regulatory DNA sequences of SCD genes may confer differences in gene expression and physiological changes that could also explain the different correlation patterns with DIs observed among commercial types. 9c-14:1 DI has been reported as the best indicator of overall SCD enzyme activity by Corl et al. (2002) [40]. Feedstuffs are normally devoid of 9c-14:1 and, therefore, this FA is produced by de novo FA synthesis. In mammary gland, significant correlation between 9c-14:1 DI and SCD1 expression was observed, whereas 9c-14:1 DI and SCD5 correlation was not [41]. We did not observe significant correlation between 9c-14:1 DI and SCD1, similarly to a previous study in intestinal adipose tissue, skeletal muscle or mammary gland [42]. However, we observed that 9c-14:1 DI and SCD5 were negatively correlated in subcutaneous adipose tissue of Pirenaica heifers (P < 0.05). Thus, these results suggest that SCD gene expression may directly affect 9c-14:1 content, but 9c-14:1 DI correlation with SCD5 and SCD1 might be breed and tissue specific as well.
As previously reported by Horton et al. [14], SCD1 and SREBP1 appear to be directly related as there was a significant linear association between these two genes in all commercial types studied (Fig. 2a). Differences in slope and coefficient of determination (R 2 ) values, however, revealed variability in this relationship among commercial types. A previous study suggested that the FA synthesis pathway is regulated in a coordinated manner by the SREBP family of membrane-bound transcription factors, and regulation of SCD1 by SREBP1 via the SRE binding site of SCD1 has been demonstrated [43].
Significant correlation between SCD5 and SREBP1 specifically observed in Pirenaica bulls and Holstein-Friesian cows suggest that SCD5 expression may be more variable among commercial types than SCD1 expression, possibly due to differences in regulation by SREBP1. For example, according to Lengi and Corl (2012) [16], the early growth response protein 2 (EGR2) and SREBP1 may bind to the same DNA site of the bovine SCD5 promoter. They observed that expression of EGR2 or SREBP1 did not increase endogenous SCD5 mRNA expression but did activate a truncated bovine SCD5 promoter luciferase reporter constructs in human JEG3 cells. Therefore, they attributed the lack of increase in SCD5 expression to the presence of additional negative-regulation sites in this gene. In our case, the absence of significant differences in other commercial types could be due to breed or other environmental factors that could, in part, modulate these putative negative-regulation sites.
Among these commercial types of the Basque region, correlations between lipogenic genes (SCD1 and SREBP1) and calculated DIs were stronger in Salers bulls than Pirenaica bulls and heifers (Fig. 3). In Holstein-Friesian cows, correlations were not significant (Fig. 3). However, SREBP1 and SCD1 correlations with DIs became significant (P < 0.05) when computed without age and HCW as covariates (data not shown). Moreover, correlations between SCD1 expression and DIs were slightly higher in younger Holstein-Friesians, while SCD5 correlations with DIs were higher in older Holstein-Friesians. This suggests an effect of age on gene expression−DI correlations in Holstein-Friesian cows. In addition to 16:1 and 18:1 [42], positive correlations between SCD1 and calculated DIs were observed for other FA species, suggesting desaturase activity also targets minor FAs of subcutaneous fat. In this regard, DIs and MUFA content were more susceptible to the expression of lipogenic genes in Pirenaica heifers than bulls. Furthermore, the effect of lipogenic gene expression on DIs was stronger in Salers than Pirenaica heifers. Our findings are supported by a previous study [13] suggesting that the FA composition of subcutaneous adipose tissue is mainly dependent on genetic background, which may in turn indicate inter-breed differences in lipid metabolism. The effect of breed appears to be more strongly associated with SREBP1 expression level than SCD1 or SCD5 expression level (Fig. 1), whereas the underlying regulation of SCD1 and SCD5 could be responsible for inter-breed differences in DIs and FA profiles.
We also report an opposite effect of SCD isoforms on certain DI values, especially in Salers bulls. This stronger pattern may stem for a more homogeneous production system (Salers breeder, personal communication) that may reduce the influences of extraneous factors. Positive correlations between DIs and SCD1 (Fig. 2b) and contrasting negative correlations between DIs and SCD5 (Fig. 2c) are likely due to genetic compensation. Lower expression of one SCD isoform could well be compensated for by upregulation of the other isoform (Fig. 1). This compensation theory was previously suggested in Caenorhabditis elegans [44]. The reciprocal expression observed between different isoforms and the underlying epigenetic processes require further investigation.
The CLA isomer 10 t,12c-18:2 was examined because it was previously described as an important inhibitor of SCD1 in dairy cattle [25]. In our study, although Pirenaica heifers accumulated the highest amounts of 10 t,12c-18:2 in subcutaneous adipose tissue (Table 2), no significant correlation was observed between 10 t,12c-18:2 and lipogenic gene expression (data not shown). Nevertheless, both isoforms may be differently regulated. In contrast to SCD1, which tends to be reduced by 10 t,12c-18:2 [45], SCD5 appears to be more stable due to lack of an N-terminal PEST sequence for degradation [11]. However, further research is needed to establish relationships among DIs and SCD isoform mRNA expression levels, and to clarify the effects of 10 t,12c-18:2 on bovine adipose and muscle tissues. Analysis of lipogenic gene expression changes with dietary treatment in ruminant species as well as promoter sequencing would provide valuable insight into the regulation of these genes and their impact on the synthesis of MUFAs and PUFAs.

Conclusion
The present study suggests that the differences in subcutaneous fat FA composition among bovine commercial types of the Basque region are related to genetic variability in lipogenic gene expression. The expression of lipogenic genes in Salers bulls showed clear effects on desaturation indexes and FA composition. All breeds show a strong correlation between SREBP1 and SCD1 expression. In addition, distinct correlations between SCD isoforms and DIs suggest a novel genetic compensation mechanism between SCD1 and SCD5 that warrants further investigation.