De Novo Lipogenesis and Stearoyl-CoA Desaturase Are Coordinately Regulated in the Human Adipocyte and Protect against Palmitate-induced Cell Injury*

De novo lipogenesis (DNL) is paradoxically up-regulated by its end product, saturated fatty acids (SAFAs). We tested the hypothesis that SAFA-induced up-regulation of DNL reflects coordinate up-regulation of elongation and desaturation pathways for disposal of SAFAs and production of monounsaturated fatty acids to protect cells from SAFA toxicity. Human preadipocytes were differentiated in vitro for 14 days with [U-13C]palmitate (0–200 μm) to distinguish exogenous fatty acids from those synthesized by DNL. Exogenous palmitate up-regulated DNL (p < 0.001) concomitantly with SCD and elongation (each p < 0.001). Adipocytes from some donors were intolerant to high palmitate concentrations (400 μm). Palmitate-intolerant cells showed lower TG accumulation. They had lower expression of SCD mRNA and less monounsaturated fatty acids in TG, emphasizing the importance of desaturation for dealing with exogenous SAFAs. There was greater [U-13C]palmitate incorporation in phospholipids. SCD knockdown with small interfering RNA caused down-regulation of DNL and of expression of DNL-related genes, with reduced membrane fluidity (p < 0.02) and insulin sensitivity (p < 0.01), compared with scrambled small interfering RNA controls. There was preferential channeling of DNL-derived versus exogenous palmitate into elongation and of DNL-derived versus exogenous stearate into desaturation. DNL may not act primarily to increase fat stores but may serve as a key regulator, in tandem with elongation and desaturation, to maintain cell membrane fluidity and insulin sensitivity within the human adipocyte.

De novo lipogenesis (DNL) is the formation of lipids from nonfat precursors such as glucose and produces the SAFAs myristate (14:0, a minor end product) and palmitate (16:0, the main end product). Palmitate and stearate (18:0) are substrates for stearoyl-CoA desaturase (SCD, or ⌬-9 desaturase), which acts to convert these SAFAs to MUFAs palmitoleate and oleate, respectively. In the liver, the pathways of DNL and fatty acid desaturation by SCD appear to be coordinately regulated (9). Therefore, it would seem that SCD plays a crucial role in maintaining the intracellular equilibrium of SAFAs and MUFAs. However, the literature surrounding the role of SCD in cell function and disease is conflicting.
Paradoxically, SAFAs have been shown to up-regulate lipogenesis. Early studies in rats showed palmitate to stimulate glucose incorporation into TG fatty acids (10,11). Several studies of high fat feeding in rats and mice, whether or not specifically high in saturated fat, show induction of DNL at a transcriptional or flux level (12)(13)(14)(15). In humans, Warensjö et al. (16) recently reported an increased SCD index (16:1 n-7/16:0) in serum phospholipids (PLs) and cholesteryl esters in subjects fed a high saturated fat diet compared with those fed a rapeseed oil-rich diet, implying an up-regulation of SCD by SAFAs.
We hypothesized that the paradoxical up-regulation of DNL seen in response to high fat feeding or SAFA exposure might represent an integral activation of fatty acid modification (desaturation and elongation) with DNL. Thus, as cells attempt to reduce SAFA accumulation by metabolic transformation of the fatty acids, DNL may necessarily be activated in parallel. DNL together with elongation and desaturation would also provide a source of oleate to ameliorate the adverse effects of palmitate excess.
We have investigated this in human adipocytes as a model primary human cell; there is considerable evidence now for the importance of DNL in adipocytes (17)(18)(19). We have investigated the regulation of these pathways and how they may be affected by an exogenous source of palmitate, the end product of DNL. A functional link between DNL and SCD elongation might also be evident in channeling of fatty acids through these pathways, which we have investigated with stable isotope tracer methodology. Our hypothesis makes some critical predictions. One is that cells that fail to up-regulate DNL and desaturation pathways appropriately will be adversely affected by provision of exogenous palmitate. Another is that the loss of SCD would be accompanied by a down-regulation of DNL, which we have tested with siRNA silencing of SCD. We also investigated the effects of SCD loss-of-function on DNL and adipocyte function, particularly cell membrane fluidity and insulin sensitivity, in the presence and absence of exogenous palmitate.

EXPERIMENTAL PROCEDURES
Adipose Tissue Sample Collection-The taking of human adipose tissue samples was approved by the Oxfordshire Clinical Research Ethics Committee, and all subjects gave written, informed consent. Subcutaneous adipose tissue biopsies were obtained by needle aspiration using a 12-gauge needle. Tissue donors consisted of 11 males and 11 females, with a median age of 38 years (ranging from 21 to 53 years) and with a median body mass index of 25.8 kg/m 2 (ranging from 21.0 to 40.1 kg/m 2 ).
Preadipocyte Isolation and Differentiation-Adipose tissue biopsies were taken from abdominal subcutaneous adipose tissue of healthy volunteers using a 12-gauge needle and syringe. Preadipocytes were isolated from subcutaneous adipose tissue as described by Hauner et al. (20) Adipose tissue was mechanically minced using scissors, washed twice with Hanks' buffered salt solution to remove contaminating blood, and then enzymatically digested in 1 mg/ml collagenase (Roche Applied Science) in Hanks' buffered salt solution in a shaking water bath (90 rpm) at 37°C for 45 min. The digested tissue was centrifuged at 1000 rpm for 5 min at 4°C. The buoyant mature adipocytes and supernatant were removed, and the pellet containing the stromal-vascular cells was resuspended in growth medium (Dulbecco's modified Eagle's medium/nutrient mixture F-12 Ham's (v/v, 1:1), 10% fetal calf serum (Invitrogen), 1 l/100 ml fibroblast growth factor, 100 units/ml penicillin, and 0.1 mg/ml streptomycin). The resuspended cells were filtered twice through nylon mesh (SEFAR), first with pore size of 250 m and then through 100 m to remove any large cells. The filtered cells were centrifuged again at 1000 rpm for 5 min at 4°C, and the supernatant was removed and resuspended in growth medium. The cells were plated in tissue culture flasks. After 1 day, contaminating red blood cells were washed off with phosphate-buffered saline (PBS), and the medium was replaced. Preadipocytes were passaged up to five times before being used for experiments.
If no exogenous fatty acids are provided during differentiation, the TG present after 14 days has necessarily all arisen from DNL. "Flux" through DNL is therefore the sum of all TG fatty acids measured by gas chromatography (GC) (as described below), in nmol over 14 days. Flux through elongation was cal-culated as the sum of all TG fatty acids of chain length greater than 16 carbons, and flux through desaturation (SCD) as the sum of all TG fatty acids produced by ⌬-9 desaturation.
Gas Chromatography Analysis of Triglyceride and Phospholipid Fatty Acids-Known amounts of internal standards of phosphatidylcholine dipentadecanoyl (15:0) and glyceryl triheptadecanoate (17:0) were added to samples prior to extraction. The total lipids were extracted from cells using the method described by Folch et al. (21). TG and PL fractions were separated using solid phase extraction columns as described previously, and fatty acid methyl esters were prepared using methanolic sulfuric acid at 80°C for 2 h (22).
GC was performed using an Agilent 6890N GC (Agilent Technologies, Stockport, UK) fitted with a 25-m ϫ 0.32-mm capillary column (FFAP-CB), with stationary phase composed of polyethylene glycol nitroterephthalic acid ester chemically bonded to a fused silica column and a flame-ionization detector. The carrier and make-up gasses were helium and nitrogen, respectively, with a column flow of 1 ml/min, and the samples were run using a splitless injection technique. The GC operated at an initial temperature of 50°C with a series of programmed temperature increments to 240°C, enabling the resolution of fatty acids from 8:0 to 24:0 within 46 min. Individual fatty acid peaks were identified by reference to known FAMEs. The fatty acid concentrations were calculated relative to the internal standard, and the results were expressed as g of fatty acid/10 6 cells.
Use of Stable Isotopes to Trace the Fate of Exogenous Palmitate-[U-13 C]palmitate (CK Gas, Cambridgeshire, UK) was also added during differentiation of preadipocytes to trace the fate of exogenous palmitate with SCD knockdown and to assess possible SCD substrate channeling. Concentrations of 0, 50, 100, and 200 M [U-13 C]palmitate were added to the differentiation medium, and GC-mass spectrometry (19) was used to distinguish exogenous from DNL-derived fatty acids in TG and PL fractions. Fatty acids derived from DNL under these conditions will be unlabeled, whereas those arising from exogenous sources will be 13 C-labeled.
The GC was equipped with a DB-Wax 30m capillary column (from Agilent; inner diameter, 0.25 mm; film thickness, 0.25 m). The sample (1.0 l) was introduced using the splitless injection mode, an injector temperature of 250°C, and an initial oven temperature of 80°C for 5 min. The oven temperature was increased by 25°C/min to 200°C and held for 10 min and then further increased by 25°C/min to 230°C, where it was held for 2 min. Mass-to-charge ratios (m/z) were determined by selected ion monitoring. Dwell time was 50 ms.
SCD Knockdown-SCD knockdown was achieved using Lipofectamine 2000 (Invitrogen) as a carrier. Because siRNA transfection of cells induces a transient knockdown of mRNA, siRNA was added to differentiating preadipocytes at day 6 of differentiation, when SCD mRNA would be expressed (data not shown), and again at days 9 and 12 (a total of three times) to maintain knockdown. Two different SCD-specific siRNAs were used to achieve SCD mRNA knockdown, termed SCD siRNA1 and siRNA2 (Stealth siRNA, HSS109499 and HSS109500, respectively; Invitrogen), with controls of Lipofectamine only, scrambled siRNA negative control (Stealth Negative siRNA 12935-300 (GC content matched to SCD-specific siRNAs; Invitrogen), and one control with no transfection components. The extent of SCD mRNA knockdown was measured as a percentage compared with the scrambled siRNA.
Gene Expression Measurements-The cells were harvested in 0.5 ml of Tri Reagent (Ambion) and transferred to a 2-ml PhaseLock gel Eppendorf tube (catalog number. 955154045; Eppendorf). 100 l of chloroform was added, shaken by hand for 15 s, and incubated at room temperature for 2-3 min before centrifugation at 12,000 ϫ g for 15 min at 4°C. The upper RNAcontaining aqueous phase was transferred to a clean tube, and an equal volume of isopropanol was added. The samples were precipitated at Ϫ80°C overnight. RNA was then collected by centrifugation at 12,000 ϫ g for 20 min at 4°C, isopropanol aspirated off, and then the pellets were washed with 0.5 ml of 70% ethanol. The pellets were once again collected by centrifugation at 7,500 ϫ g for 5 min at 4°C, ethanol was aspirated off, and the pellets were then air dried at room temperature. Once dry, RNA was resuspended in 10 l of RNA storage solution (Ambion). Any contaminating genomic DNA was digested using 1 l of DNase buffer and 0.5 l of DNase I (DNA-free TM ; Ambion).
cDNA was synthesized by reverse transcription from 500 ng of RNA using a SuperScript III first strand synthesis system (Invitrogen). A cDNA standard was created from a pool of each sample and diluted 1/50, 1/100, 1/500, and 1/1000 in 10 mM Tris, pH 8.0. All of the samples were diluted 1/100, and mRNA expression was analyzed in triplicate by real time PCR using the Applied Biosystems (Warrington, UK) Assays-on Demand TaqMan gene expression assays with TaqMan Universal PCR Master Mix (Applied Biosystems) and run on either a Rotorgene real time PCR machine (Corbett Research, Sydney, Australia) or an Applied Biosystems 7900HT real time PCR machine (Applied Biosystems) in a final volume of 8 l (3.6 l of 1/100 dilution of template, 4 l of Master Mix, and 0.4 l of specific Assay-on-Demand). The target genes were as follows: ACACA (acetyl-CoA carboxylase), FASN (fatty acid synthase), ACLY (ATP-citrate lyase), G6PD (glucose 6-phosphate dehydrogenase), SCD (stearoyl-CoA desaturase), ELOVL6 (ELOVL family member 6, elongation of long chain fatty acids), and SREBF1 (SREBP1c), assay identifications Hs00167385_m1, Hs00188012_m1, Hs00153764_m1, Hs00166169_m1, Hs00748952_s1, Hs00225412_m1, and Hs00231674_m1, respectively. mRNA expression was calculated using the ⌬ Ϫ C t transformation method (Q ϭ E[min C t Ϫ sample C t ]) as previously described (23,24). All of the target genes were normalized to the geometric mean of three housekeeping genes: PPIA (cyclophilin), PGK1 (phosphoglycerate kinase 1), and S18 (eukaryotic 18 S rRNA), assay identifications Hs99999906_m1, Hs99999904_m1, and Hs99999901_s1, respectively.
The cells were washed with PBS, and the coverslip was placed on the confocal microscope for analysis.
The images were acquired using an LSM510 confocal microscope and software (Zeiss) using a 488-nm argon ion laser for excitation and ϫ63/1.4 NA (numerical aperture) objective with light collected through a 515-nm long pass filter. The imaging frames were acquired at 4 Hz (4 frames/second) with an image depth of 8 bits. Background fluorescence was assessed by measurements taken 15 s prior to bleaching. For each condition, the measurements were made on 15 different cells.
BODIPY molecules are not entirely nonpolar. Because there was some labeling of intracellular structures, the contribution of internally labeled organelles to plasma membrane fluidity measurements was minimized by using a confocal section of only 1 m for scanning and focusing on the bottom of the coverslip to minimize stray fluorescence. Any small contribution from labeled organelles was then held constant, i.e. in every group measured this was the case. We believe this contributed some noise to the small, but significant differences that we report. The data were analyzed using the Metamorph (version 7.1) software package. Two methods of calculation were used. First, all of the intensities were normalized to the intensity at frame 1, and recovery was calculated as the time taken to halfmaximum recovery. Second, intensities were normalized to a value of 1.0 at maximum recovery, and curve fitting was used to assess recovery rates. Exploration of the data in this way showed three distinct exponential periods (0 -20, 20 -60, and 60 -180 s). Values of r 2 using this approach were from 0.975-1.000. The data are given for exponentials fitted over these periods.
Insulin Sensitivity Measurements-Uptake of 2-deoxy-D- All of the measurements of 2-DOG uptake were corrected for FABP4 mRNA expression, a marker of adipocyte differentiation. This was chosen as a denominator because previous studies showed exposure of Ob1771 preadipocytes to palmitate increased differentiation (28,29). Furthermore, the extent of differentiation of 3T3-L1 preadipocytes has been associated with changes in insulin-stimulated 2-DOG uptake (30).
Statistical Analyses-The differences between control and palmitate-treated cells were statistically analyzed using repeated measures analysis of variance (ANOVA). This was also used to assess the effects of exogenous palmitate and insulin concentrations and SCD knockdown in the insulin sensitivity assays. Differences between scrambled siRNA controls and SCD-specific siRNAs were assessed using a Student's t test (paired). All of the statistical analyses were carried out using SPSS (version 15).

Exogenous Palmitate Up-regulates DNL and SCD as Well as
Fatty Acid Elongation-The effects of exogenous palmitate on DNL and fatty acid modification were assessed by differentiating preadipocytes for 14 days with 0, 50, 100, or 200 M [U-13 C]palmitate to allow exogenous palmitate ( 13 C-labeled) to be distinguished from DNL-derived palmitate (unlabeled). Total DNL flux (Fig. 1A) was increased in response to exogenous palmitate (p ϭ 0.001, ANOVA), especially at lower concentrations (110% increase at 50 M exogenous palmitate). Flux through SCD, calculated as the sum of all TG fatty acids made via SCD action, was also up-regulated by exogenous palmitate, as was flux through fatty acid elongation ( Fig. 1A; p Ͻ 0.001 for each). There was an increase in cellular TG content in response to exogenous palmitate that reflected increases in both DNL and incorporation of exogenous ( 13 C-labeled) palmitate ( Fig.  1B; p Ͻ 0.001 for each). Expression of mRNA for DNL-associated genes and SCD were also up-regulated (p Ͻ 0.05) (Fig. 1C), with a tendency to an increase in ELOVL6 mRNA (p ϭ 0.13).
High Palmitate Has Adverse Effects on the Adipocytes from a Subpopulation of Donors-Differentiation of preadipocytes from 15 donors with 400 M palmitate revealed that the adipocytes from some donors showed reduced cell viability, whereas others thrived. Adipocytes that survived the high, 400 M palmitate were classed as palmitate-tolerant (n ϭ 10), and those that did not were classed as palmitate-intolerant (n ϭ 5). Classification was determined solely by observation of significant cell death, as seen by large areas of cell debris interspersed with occasional, very large lipid droplets, in the presence of 400 M palmitate ( Fig. 2A).
Palmitate tolerance or intolerance appeared to be a characteristic of individual donors. All of the cultures derived from one donor behaved in the same way. Repeated biopsies were obtained from two donors, one of whose adipocytes were tolerant and the other intolerant on the first occasion, and the characteristics were maintained with a second biopsy. Various anthropometric parameters were measured from the adipose tissue donors, such as body mass index, waist-to-hip ratio, age, gender, or percentage body fat. We found no significant differences between the donors of palmitate-tolerant and palmitatetolerant adipocytes (p Ͼ 0.15). The full data are in supplemental Table S1.

Palmitate-intolerant Adipocytes Have Reduced DNL and SCD Compared with Palmitate-tolerant Adipocytes and Higher Saturated Fatty Acid Content in Phospholipids-We investi-
gated what characteristics of the adipocytes would render them tolerant or intolerant to high palmitate concentrations. Palmitate-intolerant cells showed less total TG accumulation (toler-ant, 380 Ϯ 78; intolerant, 87 Ϯ 30 nmol/10 6 cells; p ϭ 0.040), supporting the morphological observations. Expression of DGAT2 mRNA (coding for a key enzyme in TG synthesis) was lower by 91.8% (p Ͻ 0.001) in the palmitate-intolerant cells.
The DNL-derived fatty acid content of adipocyte TG was less in the palmitate-intolerant compared with the palmitate tolerant cells, and mRNA expression for DNL-associated genes was also significantly lower (Fig. 2, B and C). Concomitant with lower activity of DNL in the palmitate-intolerant adipocytes, flux through SCD (Fig. 2D) and SCD mRNA expression (Fig.  2C) were also lower than in the palmitate-tolerant. Furthermore, the proportion of SAFAs in TG (mol%) was higher in the palmitate-intolerant adipocytes than in the palmitate-tolerant (83.6 Ϯ 5.9 and 64.1 Ϯ 0.9, respectively, p Ͻ 0.001). Palmitate-intolerant adipocytes also had less capacity for fatty acid esterification. mRNA expression of PCK1, the gene encoding cytosolic PEPCK, was significantly reduced in palmitate-intolerant compared with palmitate-tolerant adipocytes (0.01 Ϯ 0.06 versus 0.48 Ϯ 0.15, respectively, p ϭ 0.011).
The PL SAFA:MUFA ratio was more than 2.5-fold higher in the palmitate-intolerant adipocytes than in the tolerant group (p ϭ 0.001). The SAFA:MUFA ratio was 1.43 in the palmitatetolerant group, whereas it was Ͼ3 in the intolerant adipocytes. Furthermore, there was greater incorporation of exogenous palmitate into the PL fatty acids than in the tolerant group (p ϭ 0.051).
Effects of SCD Knockdown on DNL-Because desaturation appeared to be important for cell tolerance of exogenous palmitate, we investigated the effects of SCD knockdown in the absence or the presence of 200 M [U- 13 C]palmitate. This was achieved using two different siRNA oligonucleotides, which were compared with a scrambled siRNA control. Both siRNAs produced a consistent, stable knockdown over the 72-h period following transfection. The degrees of knockdown were 66.9 Ϯ 4.5% for siRNA1 and 87.7 Ϯ 1.8% for siRNA2 (p Ͻ 0.001). There was an increase in proportion (mol%) of the SAFA palmitate (44.7 Ϯ 1.2 versus 48.1 Ϯ 1.3 and 48.0 Ϯ 1.6 for control, siRNA1, and siRNA2, respectively) and stearate (4.3 Ϯ 1.1 versus 11.8 Ϯ 2.1 and 11.5 Ϯ 1.9 for control, siRNA1, and siRNA2, respectively) in TG fatty acids in response to SCD knockdown (p Ͻ 0.001).
DNL gene expression was down-regulated in response to SCD knockdown (Fig. 3, top panel). TG fatty acids were also analyzed in preadipocytes differentiated with either 0 or 200 M [U-13 C]palmitate and are shown in Fig. 3, bottom panel. There was a tendency for a reduction in DNL-derived palmitate with SCD knockdown.
SCD Knockdown Reduces Cell Membrane Fluidity and Impairs Insulin Sensitivity of the Adipocyte-Product:precursor ratios of PL fatty acids reflecting SCD action are shown in Fig. 4. Product:precursor ratios were significantly reduced with SCD knockdown. This meant that SCD knockdown caused a significant (p Յ 0.042) increase in PL SAFA:MUFA ratios in cells with no added palmitate (0.89 Ϯ 0.05 versus 1.19 Ϯ 0.10 and 1.14 Ϯ 0.07 for control, siRNA1, and siRNA2, respectively) and in 200 M palmitate-treated cells (1.31 Ϯ 0.16 versus 1.73 Ϯ 0.24 and 1.64 Ϯ 0.22 for control, siRNA1, and siRNA2, respectively). There was a borderline significant difference in SAFA:MUFA between the 0 and 200 M palmitate-treated cells (p ϭ 0.051). More detail on phospholipid fatty acid composition is given in supplemental Table S2.
Analysis of cell membrane fluidity, using FRAP, also revealed that the loss of SCD led to a decreased fluidity (Fig. 5, A and B). Half-times to recovery are shown on Fig. 5B. As described under "Experimental Procedures," we also fitted exponentials to the recovery curves. The effects were mainly seen in the 0 -20-and 20 -60-s periods. The exponential coefficients were significantly lower (recovery slower) for siRNA treatment versus scrambled control (repeated measures ANOVA, p ϭ 0.020 and 0.044 for 0 -20 and 20 -60 s, respectively). For illustration, the size of the effect was quantified in terms of diffusion coefficient (m 2 /cm). For one individual the values were 1.78 Ϯ 0.17 versus 0.89 Ϯ 0.11 and 1.24 Ϯ 0.15 for siRNA1 and siRNA2, respectively (14 -15 cells for each treatment, p Ͻ 0.001 and p ϭ 0.026 by unpaired t test).
Insulin sensitivity was determined from uptake of radiolabeled 2-DOG in response to insulin. The adipocytes differenti- ated with 50 and 200 M palmitate showed reduced insulin sensitivity compared with those differentiated without any exogenous palmitate (p Յ 0.021) (data not shown). Therefore, the effects of preventing SAFA conversion to MUFA via SCD knockdown were investigated. SCD siRNA-treated cells showed lower insulin-stimulated uptake of 2-DOG (Fig. 5C).
Channeling of Fatty Acids toward Elongation and Desaturation-Given the coordinate regulation of DNL, SCD, and elongation observed above, we next investigated whether fatty acids derived from DNL were preferentially channeled into further modification. DNL-derived palmitate was preferentially channeled toward elongation over exogenous palmitate (p ϭ 0.022) (Fig. 6A). Furthermore, stearate made from elongation of DNL-derived palmitate was preferentially channeled toward SCD-mediated desaturation (p Ͻ 0.001) (Fig. 6B). There was no channeling of either DNL-derived or exogenous palmitate toward SCD (p ϭ 0.361).
These distinctions were maintained in the presence of SCD knockdown with siRNA. There was again preferential channeling of DNL-derived stearic acid toward SCD (p ϭ 0.048, repeated measures ANOVA). Furthermore, SCD knockdown also had no effect on the preferential elongation of DNL-derived palmitate. As before, the action of SCD on palmitic acid was not dependent upon the source of this fatty acid, whether from DNL or exogenously derived (p ϭ 0.86).

DISCUSSION
In agreement with previous findings of DNL up-regulation by SAFAs (10 -12), we found that exogenous palmitate also up-regulated DNL in human adipocytes. We showed that upregulation of DNL was accompanied by increased elongation and desaturation and indeed that these pathways are function-   FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 ally coupled, with channeling of palmitic acid derived from DNL through elongation and into oleic acid production. We also showed that down-regulation of SCD with siRNA reduced DNL and had adverse effects on the cell. We identified populations of adipocytes that were unable to tolerate high concentrations of exogenous palmitate; these populations were cells that failed to up-regulate DNL and desaturation appropriately.

DNL and SCD in Human Adipocytes
It would seem counter-intuitive that provision of the end product of DNL, palmitate, should up-regulate this process. However, the finding that SCD and ELOVL6 activities were also increased by palmitate provides a plausible explanation. Because palmitate is a substrate for these enzymes, it appears that the cell up-regulates enzymes capable of modifying this SAFA and converting it to oleic acid. These findings mirror the in vivo finding that DNL and SCD are coordinately up-regulated in response to short term high carbohydrate feeding in humans (9).
Furthermore, there was specific channeling of DNL-derived fatty acids over exogenous fatty acids toward elongation and desaturation. Substrate specificity of SCD is a phenomenon that has previously been described in hepatocytes (31). Specifically, DNL-derived palmitate was channeled toward elonga- FIGURE 5. Changes in cell membrane fluidity and in insulin sensitivity in response to SCD. A shows FRAP raw data curves of fluorescence intensity (normalized to first frame), and B shows the calculated times to half-maximal recovery of adipocytes with SCD silencing. The data are represented as the means Ϯ S.E. from 10 -15 cells/subject (n ϭ 4 subjects). *, p ϭ 0.005, repeated measures ANOVA (mean effects of the two siRNAs). C shows insulin sensitivity measured by 2-DOG uptake. There were significant main effects for SCD knockdown and insulin concentration (p ϭ 0.009 for both). A Tukey's least significant difference post hoc test for pairwise comparisons showed significant differences between scrambled siRNA and SCD siRNA1 and siRNA2 (p ϭ 0.005 and 0.009, respectively). The data are represented as the means Ϯ S.E. (n ϭ 5 subjects). tion to stearate, which in turn was channeled toward SCD to produce oleate. The potential importance of oleate within the cell is 2-fold. First, endogenously derived oleate may be crucial for TG synthesis because DGAT2 (the enzyme that catalyzes the terminal step in TG formation) and SCD are co-localized on the endoplasmic reticulum (32,33). Palmitate up-regulation of DGAT2 mRNA may occur as a knock-on effect of SCD up-regulation, enhancing TG formation. Second, oleate is one of the most abundant fatty acids in the cell membrane, and DNL may provide a source of this fatty acid for PL synthesis. Therefore, it could be that the shunting of endogenously produced palmitate toward oleate synthesis is a physiological mechanism for the adipocyte to optimize TG storage and maintain cell membrane fluidity.
The importance of SCD to cell function in the presence of SAFAs was also highlighted when preadipocytes were differentiated with high concentrations of palmitate. Unexpectedly, the adipocytes from some donors were unable to withstand high palmitate concentrations. The most obvious difference between palmitate-tolerant and palmitate-intolerant adipocytes was in their capacity for DNL and SCD-mediated desaturation. The reduction in DGAT2 mRNA expression and the lower capacity for TG storage seen in the palmitate-intolerant adipocytes may have arisen as a consequence of reduced DNL and SCD activity. Another difference noted between the two groups of adipocytes was their expression of PCK1 mRNA. A reduction in glycerol 3-phosphate in the palmitate-intolerant adipocytes may have limited the capacity for fatty acid esterification and, hence, TG formation. An inability to store fatty acids as TG in the adipocyte would have knock-on effects. As with lipid storage disorders, such a lipodystrophy, an inability to store circulating fatty acids safely in adipose tissue, could result in ectopic fat deposition in nonadipose tissues (34), a situation shown to have lipotoxic effects on many tissues and linked with metabolic disturbances (35,36). It appeared that the inability to store fatty acids as TG led to them being redirected toward incorporation into PL fatty acids. There was a higher incorporation of [U-13 C]palmitate in the PL fatty acids of the palmitate-intolerant cells, coupled with an increased PL SAFA:MUFA ratio, which may have led to rupturing of the cell membrane.
The SCD loss-of-function studies revealed a down-regulation of the DNL pathway within the adipocyte, confirming their mutual regulation. However, Sampath et al. (37) reported that mice fed a high stearate diet up-regulated Scd1 prior to other DNL genes such as Acc, Fas, and Gpat. That study measured lipogenic gene expression in liver and was conducted in mice and thus may not mirror the situation in human adipose tissue. In our study, SCD knockdown was also associated with an increase in PL SAFA:MUFA, reduced cell membrane fluidity, and reduced insulin sensitivity of the adipocyte. Improved cell membrane fluidity as a result of increased SCD is thought to be a mechanism employed by cancer cells to avoid apoptosis and increase their survival (38). Furthermore, SAFAs have also been shown to reduce insulin sensitivity (39), stressing the importance of desaturation of SAFAs in cell function. Although there are claims that SCD inhibition may provide a pharmaceutical means to combat obesity (40,41), this study and others provide evidence that SCD inhibition in the adipocyte may be detri-mental to cell function and lead to unwanted side effects (42)(43)(44).
The cytotoxic effects of palmitate in other cell types involve induction of inflammatory pathways (45). Activation of AMPactivated protein kinase (AMPK) can inhibit this effect (3,4,45), either by increasing fatty acid oxidation or more directly (45). Because AMPK activation would down-regulate DNL, this would seem to be an independent pathway from the one we have described. It could be interesting in future experiments to investigate AMPK activation in adipocytes; if DNL itself were inhibited, but elongation and SCD pathways unaffected, this could be additionally beneficial and might lead to a further rationale for the effectiveness of AMPK-activating drugs such as metformin on insulin sensitivity.
In conclusion, the importance of DNL in the human adipocyte may be as a key regulator of elongation and desaturation, which in turn may assist in the maintenance of cell function. Channeling of DNL-derived palmitate toward elongation and desaturation to produce an endogenous supply of oleate may be necessary for the safe storage of the influx of exogenous palmitate. Consequently, fatty acids may be diverted away from TG storage and instead shunted toward incorporation into plasma membrane phospholipids, which may contribute to a loss of insulin sensitivity. In extreme cases, this may lead to rupturing of the cell membrane. The loss of capacity for TG storage may result in ectopic fat deposition with attendant adverse metabolic consequences.