Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Global transcriptome profiling identifies KLF15 and SLC25A10 as modifiers of adipocytes insulin sensitivity in obese women

Abstract

Although the mechanisms linking obesity to insulin resistance (IR) and type 2 diabetes (T2D) are not entirely understood, it is likely that alterations of adipose tissue function are involved. The aim of this study was to identify new genes controlling insulin sensitivity in adipocytes from obese women with either insulin resistant (OIR) or sensitive (OIS) adipocytes. Insulin sensitivity was first determined by measuring lipogenesis in isolated adipocytes from abdominal subcutaneous white adipose tissue (WAT) in a large observational study. Lipogenesis was measured under conditions where glucose transport was the rate limiting step and reflects in vivo insulin sensitivity. We then performed microarray-based transcriptome profiling on subcutaneous WAT specimen from a subgroup of 9 lean, 21 OIS and 18 obese OIR women. We could identify 432 genes that were differentially expressed between the OIR and OIS group (FDR ≤5%). These genes are enriched in pathways related to glucose and amino acid metabolism, cellular respiration, and insulin signaling, and include genes such as SLC2A4, AKT2, as well as genes coding for enzymes in the mitochondria respiratory chain. Two IR-associated genes, KLF15 encoding a transcription factor and SLC25A10 encoding a dicarboxylate carrier, were selected for functional evaluation in adipocytes differentiated in vitro. Knockdown of KLF15 and SLC25A10 using siRNA inhibited insulin-stimulated lipogenesis in adipocytes. Transcriptome profiling of siRNA-treated cells suggested that KLF15 might control insulin sensitivity by influencing expression of PPARG, PXMP2, AQP7, LPL and genes in the mitochondrial respiratory chain. Knockdown of SLC25A10 had only modest impact on the transcriptome, suggesting that it might directly influence insulin sensitivity in adipocytes independently of transcription due to its important role in fatty acid synthesis. In summary, this study identifies novel genes associated with insulin sensitivity in adipocytes in women independently of obesity. KFL15 and SLC25A10 are inhibitors of insulin-stimulated lipogenesis under conditions when glucose transport is the rate limiting step.

Citation: Kulyté A, Ehrlund A, Arner P, Dahlman I (2017) Global transcriptome profiling identifies KLF15 and SLC25A10 as modifiers of adipocytes insulin sensitivity in obese women. PLoS ONE 12(6): e0178485. https://doi.org/10.1371/journal.pone.0178485

Editor: Xiaoli Chen, University of Minnesota Twin Cities, UNITED STATES

Received: February 9, 2017; Accepted: May 12, 2017; Published: June 1, 2017

Copyright: © 2017 Kulyté et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: Al files are available from the GEO database (accession number GSE94753).

Funding: The funding sources for this project are: Swedish Research Council (www.vr.se, 2016-01376), Strategic research program in diabetes at Karolinska Institutet (http://ki.se/en/srp-diabetes/strategic-research-programme-in-diabetes), EU/EFPIA Innovative Medicines Initiative Joint Undertaking (grant number 115372), The Swedish Diabetes fund (https://www.diabetes.se/diabetesfonden/forskning/beviljade-projekt/), Stockholm county council (https://forskningsstod.vmi.se/Ansokan/start.asp, 20150011). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Insulin resistance (IR) is a key pathogenic factor behind type 2 diabetes (T2D) that often accompanies obesity [1]. Although the mechanisms linking obesity to T2D are not well understood, it is likely that alterations of adipose tissue function are involved [2]. Adipocytes have multiple functions including energy storage/supply, and generating endocrine signals. Expansion of white adipose tissue (WAT) mass leads to increased adipocyte size, accelerated mobilization of fatty acids, increased local inflammation and altered release of adipocyte hormones (adipokines), which may contribute to overall IR [1–3].

However, not all obese are IR as reviewed [4–6]. In a large pan European study insulin sensitivity was, on average, 15–25% lower in obese than in lean people and only 25% of the obese had a marked IR when it was defined as the 10th lowest percentile of insulin sensitivity in lean subjects [7]. There is a genetic predisposition to systemic IR, but otherwise the causes are poorly understood [8]. A previous study, applying global transcriptome profiling on subcutaneous WAT from nondiabetic individuals in the fasted state, reported association between systemic insulin sensitivity and expression of genes related to lipid metabolism and inflammation [9]. Furthermore, we recently investigated gene expression in abdominal subcutaneous WAT of non-obese and obese women in relation to systemic in vivo insulin sensitivity assessed after two hours hyperinsulinemic-euglycemic clamp [10].Surprisingly, systemic insulin resistance had a limited impact on the transcriptional response to insulin infusion and the difference between the insulin resistant and sensitive groups was much smaller than the overall effect of obesity per se on gene expression in WAT. Only about 100 transcribed genes responded differently to hyperinsulinemia in women with systemic insulin resistance compared to insulin sensitive women.

The aforementioned studies addressed systemic insulin resistance but did not directly investigate whether the adipocytes were sensitive or resistant to insulin. To address this we determined insulin sensitivity of isolated adipocytes from abdominal subcutaneous WAT in a large cohort [11]. There are many ways of measuring insulin sensitivity in adipocytes, such as inhibition of lipolysis and effects on glucose metabolism (oxidation, glucose transport, lipogenesis). Insulin-stimulated glucose transport correlates with insulin sensitivity in vivo as measured with euglycemic, hyperglycemic clamp [12]. However, direct assays of human fat cell glucose transport require large amounts of tissue and are therefore not suitable for clinical studies. For that reason we use a simple and less tissue consuming method: lipogenesis in isolated fat cells [11]. Lipogenesis is measured as incorporation of 3H-labeled glucose into fat cell lipids. At low glucose concentrations in the incubation, which is used herein, glucose transport is the rate limiting step. This means that the lipogenesis measurements actually reflect the effects of insulin on glucose transport, in a manner similar to the ““gold standard” 3-0-methyl glucose transport assay[11].

In order to identify adipose expressed genes controlling adipocyte insulin sensitivity, we performed global transcriptome profiling on collected WAT specimen in a subset of 9 lean, 21 obese with insulin resistant adipocytes (OIR) and 18 obese with insulin sensitive adipocytes (OIS); all were women. These investigations identified SLC25A10 and KLF15 as modulators of adipocyte insulin sensitivity in humans, the latter possibly by influencing the expression of several known regulators of insulin action on glucose and lipid metabolism.

Materials and methods

Patients

The subjects were selected from an ongoing study on the genetics of insulin-stimulated lipogenesis in isolated human subcutaneous adipocytes [13]. The subjects were recruited by local advertising and all lived in Stockholm, Sweden. Patients were examined at the Karolinska University Hospital in Huddinge. For the present investigation 419 subjects were assessed [body mass index (BMI) range 18–62 kg/m2 and age 20–64 years] between years 1995 to 2008. To define insulin sensitivity we used insulin-stimulated minus basal lipogenesis in isolated adipocytes (see below) for all healthy lean subjects (n = 71; BMI <25 kg/m2). We adopted the method of Ferrannini et al. using lean healthy subjects to define insulin resistance in order to avoid the influence of confounding factors such as obesity and T2D that could by themselves influence insulin action [7]. Values above the threshold for upper 10th percentile were defined as high insulin sensitivity and values below the threshold for lower 10th percentile were defined as low sensitivity. For this study, we finally selected women from which we had stored frozen abdominal subcutaneous WAT samples and who were either lean (n = 9) or obese (BMI ≥30 kg/m2). The obese were categorized into either high (OIS, n = 21) or low (OIR, n = 18) insulin sensitivity based on insulin-stimulated minus basal lipogenesis in isolated adipocytes according to the definition above. Clinical characteristic are shown in Table 1, including information about medications.

thumbnail
Download:
Table 1. Clinical characteristics.

https://doi.org/10.1371/journal.pone.0178485.t001

For functional studies, subcutaneous WAT was obtained from healthy subjects undergoing cosmetic liposuction. In this group, there was no selection for age or sex. In a subgroup of these tissue donors (n = 12), global transcriptome profiles were determined during in vitro differentiation of primary human precursor cells into adipocytes as has been reported previously [14]. The Regional board of ethics in Stockholm approved the study, and written informed consent was obtained from all participants. All clinical investigation have been conducted according to the principles expressed in the Declaration of Helsinki.

Clinical examination

Patients came to the laboratory for clinical examination following an overnight fast. A nurse determined height, weight, and waist circumference. A venous blood sample was obtained for measuring lipids and glucose at the routine clinical chemistry laboratory. Serum insulin was measured by ELISA (Mercodia, Uppsala, Sweden). Systemic insulin resistance was estimated according to the Homeostasis model adjustment (HOMAIR) formula [15]. Following the clinical examination an abdominal subcutaneous WAT biopsy was obtained by needle aspiration, as described [16]. Tissue pieces were rapidly rinsed in saline, and a small part was immediately used for quantification of adipocyte volume and lipogenesis. The rest was frozen in liquid nitrogen and kept at -70°C.

Cell isolation from subcutaneous WAT biopsies

Mature adipocytes from biopsies were prepared using the collagenase procedure as described [17]. Mean adipocyte weight and volume were determined as described [18, 19].

Primary adipocyte cultures

Primary adipocyte culture for in vitro studies was obtained from subcutaneous WAT from healthy non-obese men and women (BMI ≤30 kg/m2) undergoing cosmetic liposuction. The stroma vascular fraction (SVF) cells were isolated as described [20]. The precursor cells obtained from separate individuals were not mixed. Part of plastic-adherent SVF cells were directly plated (30,000–50,000 cells/cm2) and differentiated to adipocytes as described [20]. The degree of differentiation was controlled under the microscope as accumulation of lipids, and response to insulin in lipogenesis assay was evaluated (see below). Cultures with a differentiation degree below ~80% were discarded. Remaining SVF cells were suspended in fetal calf serum (FCS) supplemented with 10% DMSO and stored in liquid nitrogen until further usage. Upon the experiments, the SVF was thawed, washed with inoculation medium, plated and differentiated as described above.

Insulin-stimulated lipogenesis in ex vivo isolated mature adipocytes

Lipogenesis in mature adipocytes isolated from WAT biopsies by collagenase procedure was conducted as described [19]. In brief, isolated adipocytes were incubated in an albumin containing buffer with (3H) glucose at a final glucose concentration of 1 μmol/l at which glucose transport is the rate limiting step. After 2 h of incubation at 37°C adipocyte lipids were extracted and counted for radioactivity. Data were expressed as amount of glucose incorporated into lipids (nmol glucose x 2 h−1 x [107 adipocytes]−1). The basal lipogenesis (no insulin present) and lipogenesis at the maximum effective concentration for insulin were measured. The sensitivity of the adipocytes to insulin was expressed using equation pD2 = -log(EC50), where EC50 is the concentration (mol/l) of the hormone that produces a half-maximum effect, and was calculated from logistic conversion of the concentration-response curve as described [19]. In methodological experiments we compared insulin stimulated lipogenesis with in vivo insulin-induced glucose disposal rate during euglycemic hyperinsulinemic clamp, the latter was determined as described [10]. In 129 subjects, who in underwent both examinations, the results were positively correlated (r = 0.4; P <0.0001).

Insulin-stimulated lipogenesis in in vitro differentiated adipocytes

To measure lipogenesis in SVF-derived adipocytes differentiated in vitro, the cells were first washed once with DMEM without glucose (Biochrom AG, Berlin, Germany) and incubated in insulin-free DMEM (Biochrom AG) supplemented with 1 μM glucose for three hours. Following the starvation, the cells were incubated for two hours in the presence or absence of 10−7 mol/l insulin and D-[3-3H] glucose (37 MBq/ml; Perkin Elmer-Cetus, Norwalk, CT, USA) diluted 1:1000. Subsequently, the cells were washed three times with cold PBS and lysed in 0.1% SDS/H2O. Ten μl of the lysate was saved for determination of protein concentration using Pierce BCA Protein determination kit (Thermo Fisher Scientific, Lafayette, CO). The rest of the lysate was transferred to cuvettes containing scintillation fluid (toluene with 5 g/l 2.5-Diphenyloxazol and 0.3 g/l 1.4-Bis (4-methyl-5-phenyl-2-oxazolyl)-benzene; Sigma-Aldrich, St. Louis, MO) and CPM was recorded after overnight phase separation.

Transfection of siRNA

In vitro differentiated adipocytes (day 10–12 post induction) were transfected with ON-TARGETplus SMARTpool small interfering RNAs (siRNAs) targeting KLF15, SLC25A10 and SLC2A4 or non-targeting siRNA pool at 40 nM final concentration (Dharmacon, Lafayette, CO) in 24- or 48-well plates and HiPerfect Transfection Reagent (Qiagen), respectively 4.5 or 2 μl according to the manufacturer’s protocol. The cells were incubated for 48 or 72 hours at which time in vitro lipogenesis was assessed in cells plated on 48-well plates and RNA were collected.

RNA isolation, cDNA synthesis and real-time PCR

Total RNA was extracted from WAT specimens (300 mg) using RNeasy Lipid tissue (Qiagen, Hilden, Germany) according to the manufacturer’s recommendations. Total RNA from samples obtained in in vitro experiments was extracted using NucleoSpin RNA II kit (Macherey-Nagel, Düren, Germany) according to the manufacturer’s instructions. RNA concentration as well as purity was measured using a Nanodrop ND-1000 Spectrophotometer (Thermo Fisher Scientific) and high quality RNA was confirmed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). Reverse transcription was performed using the iScript cDNA synthesis kit (Qiagen) and random hexamer primers (Invitrogen, Carlsbad, CA). Quantitative real-time PCR was performed using commercial TaqMan probes (Applied Biosystems). Gene expression was normalized to the internal reference gene LRP10. Relative expression was calculated using the 2[-Delta Delta C(T)] method [21].

Microarray assay

Gene 1.1 ST Arrays were used to profile gene expression in WAT from clinical samples. In a number of previous studies we have shown that by global transcriptome profiling on WAT biopsies it is possible to identify adipocyte expressed genes of clinical relevance. Furthermore, transcriptome profiling on WAT pieces avoids the impact on gene expression caused by the protocol for isolating adipocytes. From high-quality total RNA we prepared and hybridized biotinylated complementary RNA to the arrays, and then washed, stained and scanned the slides using standardized protocols (Affymetrix Inc., Santa Clara, CA, USA). The arrays were subsequently pre-processed using Affymetrix Expression Console with the following settings; summarization: PLIER, and normalization: global-median. To allow comparisons of transcript levels between samples, all samples were subjected to an all-probeset scaling-to-target signal of 250. The Gene 1.1 ST Arrays measure the expression of 28,869 transcripts. We filtered for probesets linked to a gene symbol, and, second, probesets with average signal among all samples >50, leaving 13,798 probesets for phenotypic analysis. The threshold >50 was chosen since it excluded transcripts that should not be present in WAT. Expression levels between the obese OIR and OIS groups were compared with Significance analysis of arrays (SAM) applying 1,000 permutations applying a false discovery rate of (FDR) <5% [22]. For genes represented with more than one probeset on the array, we show results for the probeset with highest average signal.

Gene 2.1 ST arrays were used for global transcriptome profiling of primary adipocyte cultures treated with siRNA against KLF15 (n = 8), SLC25A10 (n = 8), or non-targeting siRNA control (n = 8). These arrays were pre-processed with RMA, which includes normalization, summarization of probes to probesets, and background correction. We used RMA since phenotypic analysis was performed in Limma, which requires Log-based data [23]. One RNA sample was labelled and hybridized twice to the microarrays; both experiments produced similar expression profiles. For this sample we used the average of the expression values according to RMA in subsequent analysis. We filtered for probesets linked to a gene symbol but excluded probesets linked to non-coding RNAs. Next we excluded ~9,000 probesets (9,012 in KLF15 and 9,019 in SLC25A10 analysis) with average expression signal ≤8, and subsequently ~2,340 probesets (2,345 in KLF15 and 2,338 in SLC25A10 analysis) with lowest SD/average expression based on analysis of all samples leaving 11,358 probesets for subsequent phenotypic analysis. Limma software was used to assess the impact of gene knockdown on the transcriptome using a model that takes into account number of subjects, duplicate cell culture samples, and treatment [23]. False discovery rat (FDR) <10% was used as cutoff to identify KLF15 and SLC25A10 target genes; the less stringent threshold was chosen due to small sample size.

Webgestalt was used to identify gene sets that were over-represented among differentially expressed as compared to all analyzed genes [24]. Microarray data have been submitted to GEO with accession number GSE94753.

Statistical analyses

Statistical analyses were performed in JMP v. 12. Shapiro-Wilks test was used to test whether the distribution of clinical variables deviated from normality. Two-group comparisons of normally distributed variables were performed by two-sided Student’s t-test and otherwise by Kruskal-Wallis test. Quantitative relationships were evaluated by least square regression.

Results

Clinical characteristics

Clinical characteristics of the studied cohort are shown in Table 1. The OIR group was defined by insulin-stimulated minus basal lipogenesis in isolated adipocytes below the threshold for lower 10th percentile for all healthy lean subjects, and the OIS group by insulin-stimulated minus basal lipogenesis above the threshold for upper 10th percentile. The two obese groups were matched for BMI, waist circumference and adipocyte size; all of which may influence insulin sensitivity independently of insulin-stimulated lipogenesis. All groups were matched for age. The average age was 38 years (range 22–60 years). The OIR women displayed markedly lower systemic insulin sensitivity as estimated by HOMAIR than the OIS women.

Genes associated with adipose insulin resistance

462 probe sets representing 432 genes were differentially expressed between the OIR and OIS group (FDR ≤5%) (S1 Table). According to the Webgestalt tool, the 432 insulin resistance-associated genes were enriched in pathways related to glucose and amino acid metabolism, and cellular respiration, as compared to all analyzed genes (selected pathways are listed in Table 2; all significant pathways a listed in S2 Table). The Insulin signaling pathway had adjusted P = 0.06. Genes involved in glucose uptake and insulin signaling were with one exception (NUP93) expressed at lower levels in the OIR group, and included SLC2A4, AKT2, as well as genes coding for several enzymes in the mitochondria respiratory chain (Table 3). The expression pattern of these genes with known function was consistent with the difference in sensitivity to insulin between groups.

thumbnail
Download:
Table 2. Selected pathways over-represented among differentially expressed genes between OIR and OIS womena.

https://doi.org/10.1371/journal.pone.0178485.t002

thumbnail
Download:
Table 3. Glucose uptake and insulin signaling genes differentially expressed between OIR and OIS women.

https://doi.org/10.1371/journal.pone.0178485.t003

In general fold change in expression between groups was small. Only nine genes displayed fold change in expression >30% whereas no gene displayed fold change >50% between OIR and OIS women. Among the 45 genes with fold change in expression ≥20% between the OIR and OIS groups, 40 genes displayed directionally consistent, and in most cases nominally significant, difference in expression between all obese versus lean women (Table 4). Among genes expressed at lower levels in the OIR group, some have been reported to be involved in insulin signalling and/or adipogenesis-regulating pathways, i.e. SLC2A4 encoding GLUT4, AXIN2 [25, 26], TF encoding Transferrin [27], KLF15 [28], and CDKN2C [29]. Five genes did not display directionally consistent expression pattern in the OIR versus OIS, and obese versus lean comparisons (AXIN2, DLEC1, ANKRD36B, LOC441666, CAB39L); these genes are besides AXIN2 poorly defined as regards function. All five are expressed at low levels in adipocytes, and none display increased expression during adipogenesis (FDR ≤1%) (Table 4). We did therefore not study these genes in more detail. A flowchart describing the project is shown in Fig 1.

thumbnail
Download:
Fig 1. Flowchart describing the project.

https://doi.org/10.1371/journal.pone.0178485.g001

thumbnail
Download:
Table 4. Genes differentially expressed between OIR and OIS women with FDR ≤5% and fold change in expression 20%.

https://doi.org/10.1371/journal.pone.0178485.t004

As we could not study all differentially expressed genes functionally due to limited resources we narrowed down the list by filtering on the following criteria: genes with fold change ≥20% between the OIR and OIS group, average expression signal in mature in vitro differentiated adipocytes ≥250 arbitrary units, and with increased expression during adipogenesis (FDR ≤1%) [14]. Seven genes fulfilled these criteria, TF, REEP6, KLF15, CDKN2C, GYS1, SLC2A4, and SLC25A10 (Table 2). Based on initial screening experiments and reported gene functions, we considered KLF15 and SLC25A10 as the most likely involved in development of adipocyte IR. KLF15, which encodes a transcription factor, was selected because it has been reported to stimulate adipogenesis and target SLC2A4 in rodent models [28]. SLC25A10, a dicarboxylate transporter, has been reported to play an important role in fatty acid synthesis and hereby to control lipid accumulation in adipocytes [30].

siRNA treatment resulted in efficient knockdown of KLF15 and SLC25A10 mRNA (Fig 2A), as well as inhibited insulin stimulated lipogenesis (Fig 2B). These results are consistent with the lower expression of these genes in OIR compared to OIS women. Knockdown of SLC2A4 was used as a positive control for the lipogenesis assay.

thumbnail
Download:
Fig 2. Effects of KLF15 and SLC25A10 knockdownon lipogenesis in vitro.

A. KLF15 and SLC25A10 were knocked down using 40nM of siRNA in SVF-derived human adipocytes differentiated in vitro and expression of the genes evaluated using real-time PCR. Results were analyzed using Students t-test and are presented as relative fold change ± SD vs. negative control. B. SVF-derived adipocytes differentiated in vitro were transfected with 40 nM of siRNA against KLF15 and SLC25A10 for 48 hours followed by evaluation of basal and insulin-stimulated lipogenesis. Relative insulin-stimulated lipogenesis was calculated against non-targeting siRNA NegC at insulin-stimulated state. Induction of lipogenesis by insulin for NegC was minimum 3-fold in all experiments. Results are based on three to five biological/independent experiments.*p<0.05, **p<0.01 and ***p<0.001.

https://doi.org/10.1371/journal.pone.0178485.g002

KLF15 and insulin resistance in human adipocytes

To delineate pathways controlled by KLF15 signaling in human adipocytes, we characterized the effects of KLF15 knockdown on the global transcriptome. 366 genes were differentially expressed between KLF15 knockdown and control cells with FDR 10% (S3 Table). Overall, 18 genes targeted by KLF15 overlapped with genes differentially expressed between OIR and OIS women. 15 of 18 genes displayed a consistent pattern of expression in the KLF15 siRNA treated cells and in the clinical cohort, e.g. downregulated by KLF15 knockdown and lower expression in OIR women. These 15 genes include one gene in the insulin signaling (GYS1) and one in the glucose transport (GCDH) pathways. Besides GYS1 and GCDH, only PCYT2 of the 15 candidate genes, has previously been linked to relevant metabolic disease phenotypes [31]. However, contrary to expected, PCYT2 knockdown in human adipocytes increased insulin-stimulated lipogenesis (results not shown). KLF15 has been reported to control expression of glucose transporter SLC2A4 in rodents [32]; however KLF15 knockdown did not affect SLC2A4 in our human cells (results not shown).

In order to further elucidate molecular pathways linking KLF15 to insulin sensitivity we regressed insulin stimulated lipogenesis in the obese subset of the clinical cohort on expression of the 366 KLF15 target genes. Expression of 77 genes were nominally associated with insulin stimulated minus basal lipogenesis of which 66 genes displayed a pattern of expression consistent with regulation by KLF15 and influence on insulin sensitivity in the clinical cohort (S4 Table). Interestingly, in analysis of gene sets over-represented among these 66 KLF15-target genes compared to all analyzed genes expressed in the cells, insulin resistance scored highest (adjusted P 9.7x10-5) (Table 5). Adipogenesis was the most strongly enriched Wikipathway (adjusted P 0.005). Ingenuity analysis of the 66 genes identified PPARG and PXMP2 as targets of KLF15, and AQP7 as well as LPL as targets of PPARG (Fig 3). Other KLF15 target genes of potential importance for insulin sensitivity include the NDUFB10 and COQ7 genes encoding mitochondrial enzymes (S4 Table).

thumbnail
Download:
Fig 3. Ingenuity analysis of KLF15 target genes associated with insulin stimulated lipogenesis in vivo.

https://doi.org/10.1371/journal.pone.0178485.g003

thumbnail
Download:
Table 5. Gene sets enriched among target genes in KLF15 knockdown.

https://doi.org/10.1371/journal.pone.0178485.t005

SLC25A10 and insulin resistance

Furthermore, we also examined the effects of SLC25A10 knockdown on the transcriptome in human adipocytes. 165 genes were differentially expressed between SLC25A10 knockdown and control cells with FDR ≤10% (S5 Table). Three genes targeted by SLC25A10 overlapped with genes differentially expressed between OIR and OIS women. Two of three genes displayed a pattern of expression consistent with regulation by SLC25A10 and influence on insulin sensitivity in the clinical cohort, ACVR1C and SLC41A1. SLC25A10 and KLF15 did not target one another. One gene in the glucose transport and insulin signaling pathways was regulated by SLC25A10 knockdown, MAPK6, but this gene was not differentially expressed in WAT of OIR and OIS women. In order to further elucidate molecular pathways linking SLC25A10 to insulin sensitivity we regressed insulin stimulated lipogenesis in the obese subset of the clinical cohort on expression of the SLC25A10 target genes. 27 genes were nominally associated with insulin stimulated minus basal lipogenesis of which 16 displayed a pattern of expression consistent with regulation by SLC25A10 and influence on insulin sensitivity in the clinical cohort (S6 Table). The Webgestalt tool did not identify any over-represented pathways among these 16 genes associated with insulin sensitivity as compared to all analyzed genes. However, specific genes reported to be phenotypically linked to adipose functions and potentially involved in the regulation of insulin sensitivity include ACVR1C, which inhibits lipolysis [33], PRKAR2A which comprises a subunit of protein kinase A [34], SEMA3E which regulates adipose inflammation [35], and SKP2 [36] which controls adipocyte number.

Discussion

The aim of this study was to define novel genes that may cause insulin resistance in human adipocytes independently of obesity. To assess insulin sensitivity we applied lipogenesis assay where glucose transport is the rate limiting step. By transcriptome profiling in a large human cohort including non-obese and obese with or without insulin resistance, combined with functional evaluation in adipocyte cultures of specific genes, this study identifies KLF15 and SLC25A10 as potential modulators of adipocyte insulin sensitivity. In addition, this study highlight potential roles for enzymes involved in glucose uptake and metabolism, and for genes controlling adipogenesis in development of adipocyte insulin resistance in obese women [37]. These findings shed new light on the complex signals linking obesity to development of insulin resistance.

In the study, siRNA knockdown of KLF15 and SLC25A10 resulted in a ~20% inhibition of insulin stimulated lipogenesis. The effect was modest in comparison with the more dramatic differences in insulin stimulated lipogenesis between OIR and OIS women in the clinical cohort. This effect is compatible with a model where insulin sensitivity displays a complex etiology; it is likely that several genes are involved and that different genes might be more or less important in different individuals. Thus, we consider KLF15 and SLC25A10 to be modulators on insulin sensitivity.

KLF15 has previously been reported to be down-regulated in subcutaneous WAT and skeletal muscle from subjects with systemic insulin resistance [9], thus supporting an important role for KLF15 not only in adipose tissue but in systemic insulin resistance. A number of target genes for KLF15 including SLC2A4 have been identified in rodent adipocyte [32, 38–41]. These genes, however, were not detected as targeted by KLF15 in our experiments in human cells. The discrepancy could potentially be due to species specific effects of KLF15, due to the presence of different transcriptional cofactors in the different cell culture systems or due to limited sensitivity in our experiment leading to failure to detect all regulated genes. Importantly, the KLF15 target gene PPARG reported in rodent models was confirmed in our human adipocytes [42]. In rodents, the KLF15-PPARG interaction signal is important for adipogenesis. However, PPARG is also a key factor controlling lipogenesis and insulin sensitivity in mature adipocytes [43]. Our siRNA knockdown of KLF15 was performed late during in vitro differentiation of adipocytes, demonstrating that the KLF15-PPARG signal most likely controls adipocyte insulin sensitivity independently of adipogenesis. Whereas at least three other KLFs have been shown to bind to the PPARG2 isoform promoter, binding sites of KLF15 in the PPARG promoters have yet not been defined [44]. Thus, it is currently unknown whether the control of PPARG gene transcription by KLF15 is direct, or indirect, by controlling the activity of other transcription factors which in turn control PPARG expression. Other transcription factors known to control PPARG gene expression [44, 45] were not among KLF15 target genes.

In the siRNA experiments we identified additional downstream targets of KLF15 implicated in lipid storage including LPL involved in cellular uptake of free fatty acids [46], the glycerol channel AQP7 involved in triglyceride synthesis [47], and PXMP2 which is essential for mammary lipid homeostasis [48].AQP7 disruption elevates adipose glycerol kinase activity, accelerates triglyceride synthesis in adipocytesAQP7 disruption elevates adipose glycerol kinase activity, accelerates triglyceride synthesis in adipocytesAQP7 disruption elevates adipose glycerol kinase activity, accelerates triglyceride synthesis in adipocytes.

Knockdown of SLC25A10 inhibited insulin stimulated lipogenesis with the same magnitude as KLF15. However, the transcriptional effects of SLC25A10 knockdown were modest. A possible complementary explanation is that SLC25A10 has direct effects on insulin stimulated lipogenesis. SLC25A10 encodes a dicarboxylate carrier that transports e.g. malate and succinate across the mitochondrial inner membran dicarboxylate carrier dicarboxylate carrier edicarboxylate carrier dicarboxylate carrier. It has been reported to play an important role in fatty acid synthesis and hereby to control lipid accumulation in adipocytes [30] and knockdown could thus potentially directly impact lipogenesis without need for transcriptional alterations.

WAT from the OIR women was further characterized by lower expression of specific genes implicated in the regulation of adipogenesis, i.e. TF [27], AXIN2 [25], and CDKN2C [29]. This finding is in agreement with the established association between impaired adipogenesis and insulin resistance as reviewed [49]. Also, lower expression of genes in the mitochondria respiratory chain is consistent with reported role of mitochondrial dysfunction in insulin resistance and T2D [50, 51].

Finally, to assess the importance for whole body insulin resistance of the 432 genes whose expression associated adipocyte insulin resistance, we examined their expression in relevant published global transcriptome studies on subcutaneous WAT. Elbein et al. reported that the expression of 172 genes in subcutaneous WAT is associated with systemic insulin resistance [9]. Ten of these genes overlapped with genes associated with adipocyte insulin resistance in the present study; all displayed a directionally consistent change in expression (Table 6). Although the absolute number of genes overlapping between studies was small, possible due to limited sample size and power of both studies, these results highlight a potential role for dysregulated WAT gene expression in development of whole body insulin resistance. Nilsson et al. reported that expression of 197 genes in subcutaneous WAT is differentially expressed between twins discordant for T2D [52]. Six genes overlapped with the present study, of which four displayed consistent expression pattern (Table 6). The poor overlap with T2D may be due to limited importance of adipocyte insulin resistance for development of diabetes in this cohort, or the adipose transcriptome being dominated by changes that are secondary to hyperglycemia.

thumbnail
Download:
Table 6. Genes associated with adipocyte IR also associated with systemic insulin resistance or T2D.

https://doi.org/10.1371/journal.pone.0178485.t006

Limitations of this study include sample size, and the fact that only women were studied. Thus, we cannot from these results draw any conclusions about the control of adipocyte insulin sensitivity in men. Another limitation is that transcriptome profiling on the clinical cohort was performed in WAT specimen whereas the functional studies were limited to adipocytes. Thus, potential indirect effects of stroma-vascular cells on adipocyte insulin sensitivity were not considered in the functional assays.

In conclusion, this study has identified two genes, KLF15 and SLC25A10, potentially important for insulin sensitivity in adipocytes in obese women. KFL15 and SLC25A10 inhibit insulin stimulated lipogenesis under conditions when glucose transport is the rate limiting step. In particular KLF15 could play and important role in development of systemic insulin resistance, as it has also been shown to have effects in other tissues [9, 53].

Supporting information

S1 Table. Genes differentially expressed in subcutaneous adipose tissue between OIR and OIS women with FDR <5%.

https://doi.org/10.1371/journal.pone.0178485.s001

(XLSX)

S2 Table. Gene sets over-represented among differentially expressed genes between OIR and OIS women.

https://doi.org/10.1371/journal.pone.0178485.s002

(XLSX)

S3 Table. Genes differentially expressed between KLF15 and nontargeting (NEGC) siRNA treated cells with FDR <10%.

https://doi.org/10.1371/journal.pone.0178485.s003

(XLSX)

S4 Table. Correlation with insulin stimulated lipogenes of genes differentially expressed between KLF15 and nontargeting siRNA treated cells.

https://doi.org/10.1371/journal.pone.0178485.s004

(XLSX)

S5 Table. Genes differentially expressed between SLC25A10 and nontargeting siRNA (NEGC) treated cells with FDR <10%.

https://doi.org/10.1371/journal.pone.0178485.s005

(XLSX)

S6 Table. Correlation with insulin stimulated lipogenes of genes differentially expressed between SLC25A10 and nontargeting siRNA treated cells.

https://doi.org/10.1371/journal.pone.0178485.s006

(XLSX)

Acknowledgments

The excellent technical assistance of Gaby Åström, Elisabeth Dungner, Kerstin Wåhlen, Eva Sjölin, Katarina Hertel and Yvonne Widlund is highly appreciated. We thank Marika Rönnholm for excellent technical assistance with the microarray, and David Brodin for array processing.

Author Contributions

  1. Conceptualization: PA ID.
  2. Data curation: ID.
  3. Formal analysis: AK AE ID.
  4. Funding acquisition: PA ID.
  5. Investigation: AK AE ID.
  6. Methodology: AK ID.
  7. Project administration: AK ID.
  8. Resources: PA ID.
  9. Software: AE ID.
  10. Supervision: ID.
  11. Validation: AK.
  12. Visualization: AK.
  13. Writing – original draft: AK.
  14. Writing – review & editing: AK AE PA ID.

References

  1. 1. DeFronzo RA, Banerji MA, Bray GA, Buchanan TA, Clement S, Henry RR, et al. Determinants of glucose tolerance in impaired glucose tolerance at baseline in the Actos Now for Prevention of Diabetes (ACT NOW) study. Diabetologia. 2010;53(3):435–45. Epub 2009/12/17.: 10.1007/s00125-009-1614-2. pmid:20012012
  2. 2. Guilherme A, Virbasius JV, Puri V, Czech MP. Adipocyte dysfunctions linking obesity to insulin resistance and type 2 diabetes. Nature reviews Molecular cell biology. 2008;9(5):367–77. Epub 2008/04/11. PubMed Central PMCID: PMC2886982. pmid:18401346
  3. 3. Xu XJ, Pories WJ, Dohm LG, Ruderman NB. What distinguishes adipose tissue of severely obese humans who are insulin sensitive and resistant? Current opinion in lipidology. 2013;24(1):49–56. Epub 2013/01/10. PubMed Central PMCID: PMC3575680. pmid:23298959
  4. 4. Sims EA. Are there persons who are obese, but metabolically healthy? Metabolism: clinical and experimental. 2001;50(12):1499–504. Epub 2001/12/06.
  5. 5. Karelis AD. Metabolically healthy but obese individuals. Lancet. 2008;372(9646):1281–3. Epub 2008/10/22. pmid:18929889
  6. 6. Bluher M. The distinction of metabolically 'healthy' from 'unhealthy' obese individuals. Current opinion in lipidology. 2010;21(1):38–43. Epub 2009/11/17. pmid:19915462
  7. 7. Ferrannini E, Natali A, Bell P, Cavallo-Perin P, Lalic N, Mingrone G. Insulin resistance and hypersecretion in obesity. European Group for the Study of Insulin Resistance (EGIR). The Journal of clinical investigation. 1997;100(5):1166–73. Epub 1997/09/26. PubMed Central PMCID: PMC508292. pmid:9303923
  8. 8. Scott RA, Lagou V, Welch RP, Wheeler E, Montasser ME, Luan J, et al. Large-scale association analyses identify new loci influencing glycemic traits and provide insight into the underlying biological pathways. Nat Genet. 2012;44(9):991–1005. Epub 2012/08/14. PubMed Central PMCID: PMC3433394. pmid:22885924
  9. 9. Elbein SC, Kern PA, Rasouli N, Yao-Borengasser A, Sharma NK, Das SK. Global gene expression profiles of subcutaneous adipose and muscle from glucose-tolerant, insulin-sensitive, and insulin-resistant individuals matched for BMI. Diabetes. 2011;60(3):1019–29. Epub 2011/01/27. PubMed Central PMCID: PMC3046820. pmid:21266331
  10. 10. Ryden M, Hrydziuszko O, Mileti E, Raman A, Bornholdt J, Boyd M, et al. The Adipose Transcriptional Response to Insulin Is Determined by Obesity, Not Insulin Sensitivity. Cell reports. 2016;16(9):2317–26. Epub 2016/08/23. pmid:27545890
  11. 11. Arner P, Engfeldt P. Fasting-mediated alteration studies in insulin action on lipolysis and lipogenesis in obese women. The American journal of physiology. 1987;253(2 Pt 1):E193–201. Epub 1987/08/01.
  12. 12. Bogardus C, Lillioja S, Mott D, Reaven GR, Kashiwagi A, Foley JE. Relationship between obesity and maximal insulin-stimulated glucose uptake in vivo and in vitro in Pima Indians. J Clin Invest. 1984;73(3):800–5. PubMed Central PMCID: PMCPMC425083. pmid:6368588
  13. 13. Dahlman I, Belarbi Y, Laurencikiene J, Pettersson AM, Arner P, Kulyte A. Comprehensive functional screening of miRNAs involved in fat cell insulin sensitivity among women. Am J Physiol Endocrinol Metab. 2017:ajpendo 00251 2016.
  14. 14. Arner E, Mejhert N, Kulyte A, Balwierz PJ, Pachkov M, Cormont M, et al. Adipose tissue microRNAs as regulators of CCL2 production in human obesity. Diabetes. 2012;61(8):1986–93. Epub 2012/06/13. PubMed Central PMCID: PMC3402332. pmid:22688341
  15. 15. Bonora E, Targher G, Alberiche M, Bonadonna RC, Saggiani F, Zenere MB, et al. Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin sensitivity. Diabetes Care. 2000;23(1):57–63. Epub 2000/06/17. pmid:10857969
  16. 16. Kolaczynski JW, Morales LM, Moore JH Jr., Considine RV, Pietrzkowski Z, Noto PF, et al. A new technique for biopsy of human abdominal fat under local anaesthesia with Lidocaine. Int J Obes Relat Metab Disord. 1994;18(3):161–6. Epub 1994/03/01. pmid:8186813
  17. 17. Rodbell M, Krishna G. Preparation of isolated fat cells and fat cell "ghosts"; methods for assaying adenylate cyclase activity and levels of cyclic AMP. Methods Enzymol. 1974;31:103–14. Epub 1974/01/01. pmid:4370663
  18. 18. Hirsch J, Gallian E. Methods for the determination of adipose cell size in man and animals. J Lipid Res. 1968;9(1):110–9. Epub 1968/01/01. pmid:4295346
  19. 19. Lofgren P, Hoffstedt J, Naslund E, Wiren M, Arner P. Prospective and controlled studies of the actions of insulin and catecholamine in fat cells of obese women following weight reduction. Diabetologia. 2005;48(11):2334–42. Epub 2005/10/14. pmid:16222518
  20. 20. van Harmelen V, Skurk T, Hauner H. Primary culture and differentiation of human adipocyte precursor cells. Methods Mol Med. 2005;107:125–35. pmid:15492368
  21. 21. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001;25(4):402–8. Epub 2002/02/16. pmid:11846609
  22. 22. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001;98(9):5116–21. Epub 2001/04/20. PubMed Central PMCID: PMC33173. pmid:11309499
  23. 23. Smyth GK. Limma: linear models for microarray data. In: Gentleman R CV, Dudoit S, Irizarry R and Huber W, editor. Bioinformatics and Computational Biology Solutions Using R and Bioconductor. New York: Springer; 2005. p. 397–420.
    • 24. Zhang B, Kirov S, Snoddy J. WebGestalt: an integrated system for exploring gene sets in various biological contexts. Nucleic Acids Res. 2005;33(Web Server issue):W741–8. Epub 2005/06/28. PubMed Central PMCID: PMC1160236. pmid:15980575
    • 25. Ross SE, Hemati N, Longo KA, Bennett CN, Lucas PC, Erickson RL, et al. Inhibition of adipogenesis by Wnt signaling. Science. 2000;289(5481):950–3. Epub 2000/08/11. pmid:10937998
    • 26. Guo HL, Zhang C, Liu Q, Li Q, Lian G, Wu D, et al. The Axin/TNKS complex interacts with KIF3A and is required for insulin-stimulated GLUT4 translocation. Cell Res. 2012;22(8):1246–57. PubMed Central PMCID: PMCPMC3411167. pmid:22473005
    • 27. Moreno-Navarrete JM, Ortega F, Moreno M, Ricart W, Fernandez-Real JM. Fine-tuned iron availability is essential to achieve optimal adipocyte differentiation and mitochondrial biogenesis. Diabetologia. 2014;57(9):1957–67. pmid:24973963
    • 28. Mori T, Sakaue H, Iguchi H, Gomi H, Okada Y, Takashima Y, et al. Role of Kruppel-like factor 15 (KLF15) in transcriptional regulation of adipogenesis. The Journal of biological chemistry. 2005;280(13):12867–75. Epub 2005/01/25. pmid:15664998
    • 29. Kang JW, Choi Y, Park JH, Kim JS, Park KD, Baek DH, et al. The effects of cyclin-dependent kinase inhibitors on adipogenic differentiation of human mesenchymal stem cells. Biochem Biophys Res Commun. 2008;366(3):624–30. pmid:18047835
    • 30. Mizuarai S, Miki S, Araki H, Takahashi K, Kotani H. Identification of dicarboxylate carrier Slc25a10 as malate transporter in de novo fatty acid synthesis. The Journal of biological chemistry. 2005;280(37):32434–41. Epub 2005/07/20. pmid:16027120
    • 31. Fullerton MD, Hakimuddin F, Bonen A, Bakovic M. The development of a metabolic disease phenotype in CTP:phosphoethanolamine cytidylyltransferase-deficient mice. J Biol Chem. 2009;284(38):25704–13. PubMed Central PMCID: PMCPMC2757972. pmid:19625253
    • 32. Gray S, Feinberg MW, Hull S, Kuo CT, Watanabe M, Sen-Banerjee S, et al. The Kruppel-like factor KLF15 regulates the insulin-sensitive glucose transporter GLUT4. J Biol Chem. 2002;277(37):34322–8. Epub 2002/07/05. pmid:12097321
    • 33. Yogosawa S, Mizutani S, Ogawa Y, Izumi T. Activin receptor-like kinase 7 suppresses lipolysis to accumulate fat in obesity through downregulation of peroxisome proliferator-activated receptor gamma and C/EBPalpha. Diabetes. 2013;62(1):115–23. PubMed Central PMCID: PMCPMC3526038. pmid:22933117
    • 34. Yang H, Yang L. Targeting cAMP/PKA pathway for glycemic control and type 2 diabetes therapy. J Mol Endocrinol. 2016;57(2):R93–R108. pmid:27194812
    • 35. Shimizu I, Yoshida Y, Moriya J, Nojima A, Uemura A, Kobayashi Y, et al. Semaphorin3E-induced inflammation contributes to insulin resistance in dietary obesity. Cell Metab. 2013;18(4):491–504. pmid:24093674
    • 36. Sakai T, Sakaue H, Nakamura T, Okada M, Matsuki Y, Watanabe E, et al. Skp2 controls adipocyte proliferation during the development of obesity. J Biol Chem. 2007;282(3):2038–46. pmid:17082193
    • 37. Carvalho E, Jansson PA, Nagaev I, Wenthzel AM, Smith U. Insulin resistance with low cellular IRS-1 expression is also associated with low GLUT4 expression and impaired insulin-stimulated glucose transport. FASEB J. 2001;15(6):1101–3. pmid:11292681
    • 38. Nagare T, Sakaue H, Takashima M, Takahashi K, Gomi H, Matsuki Y, et al. The Kruppel-like factor KLF15 inhibits transcription of the adrenomedullin gene in adipocytes. Biochem Biophys Res Commun. 2009;379(1):98–103. pmid:19094967
    • 39. Nagare T, Sakaue H, Matsumoto M, Cao Y, Inagaki K, Sakai M, et al. Overexpression of KLF15 transcription factor in adipocytes of mice results in down-regulation of SCD1 protein expression in adipocytes and consequent enhancement of glucose-induced insulin secretion. J Biol Chem. 2011;286(43):37458–69. PubMed Central PMCID: PMCPMC3199492. pmid:21862590
    • 40. Enomoto T, Ohashi K, Shibata R, Kambara T, Uemura Y, Yuasa D, et al. Transcriptional regulation of an insulin-sensitizing adipokine adipolin/CTRP12 in adipocytes by Kruppel-like factor 15. PLoS One. 2013;8(12):e83183. PubMed Central PMCID: PMCPMC3865152. pmid:24358263
    • 41. Zhao ZD, Zan LS, Li AN, Cheng G, Li SJ, Zhang YR, et al. Characterization of the promoter region of the bovine long-chain acyl-CoA synthetase 1 gene: Roles of E2F1, Sp1, KLF15, and E2F4. Scientific reports. 2016;6:19661. PubMed Central PMCID: PMCPMC4726046. pmid:26782942
    • 42. Mori T, Sakaue H, Iguchi H, Gomi H, Okada Y, Takashima Y, et al. Role of Kruppel-like factor 15 (KLF15) in transcriptional regulation of adipogenesis. J Biol Chem. 2005;280(13):12867–75. pmid:15664998
    • 43. Medina-Gomez G, Gray S, Vidal-Puig A. Adipogenesis and lipotoxicity: role of peroxisome proliferator-activated receptor gamma (PPARgamma) and PPARgammacoactivator-1 (PGC1). Public Health Nutr. 2007;10(10A):1132–7. pmid:17903321
    • 44. Wu Z, Wang S. Role of kruppel-like transcription factors in adipogenesis. Dev Biol. 2013;373(2):235–43. pmid:23142072
    • 45. Gao H, Mejhert N, Fretz JA, Arner E, Lorente-Cebrian S, Ehrlund A, et al. Early B cell factor 1 regulates adipocyte morphology and lipolysis in white adipose tissue. Cell Metab. 2014;19(6):981–92. Epub 2014/05/27. PubMed Central PMCID: PMC4109056. pmid:24856929
    • 46. Mead JR, Irvine SA, Ramji DP. Lipoprotein lipase: structure, function, regulation, and role in disease. J Mol Med (Berl). 2002;80(12):753–69.
    • 47. Hibuse T, Maeda N, Funahashi T, Yamamoto K, Nagasawa A, Mizunoya W, et al. Aquaporin 7 deficiency is associated with development of obesity through activation of adipose glycerol kinase. Proc Natl Acad Sci U S A. 2005;102(31):10993–8. PubMed Central PMCID: PMCPMC1182435. pmid:16009937
    • 48. Vapola MH, Rokka A, Sormunen RT, Alhonen L, Schmitz W, Conzelmann E, et al. Peroxisomal membrane channel Pxmp2 in the mammary fat pad is essential for stromal lipid homeostasis and for development of mammary gland epithelium in mice. Dev Biol. 2014;391(1):66–80. pmid:24726525
    • 49. Gustafson B, Hedjazifar S, Gogg S, Hammarstedt A, Smith U. Insulin resistance and impaired adipogenesis. Trends Endocrinol Metab. 2015;26(4):193–200. pmid:25703677
    • 50. Dahlman I, Forsgren M, Sjogren A, Nordstrom EA, Kaaman M, Naslund E, et al. Downregulation of electron transport chain genes in visceral adipose tissue in type 2 diabetes independent of obesity and possibly involving tumor necrosis factor-alpha. Diabetes. 2006;55(6):1792–9. Epub 2006/05/30. pmid:16731844
    • 51. Kim JA, Wei Y, Sowers JR. Role of mitochondrial dysfunction in insulin resistance. Circulation research. 2008;102(4):401–14. PubMed Central PMCID: PMCPMC2963150. pmid:18309108
    • 52. Nilsson E, Jansson PA, Perfilyev A, Volkov P, Pedersen M, Svensson MK, et al. Altered DNA methylation and differential expression of genes influencing metabolism and inflammation in adipose tissue from subjects with type 2 diabetes. Diabetes. 2014.
    • 53. Yamamoto J, Ikeda Y, Iguchi H, Fujino T, Tanaka T, Asaba H, et al. A Kruppel-like factor KLF15 contributes fasting-induced transcriptional activation of mitochondrial acetyl-CoA synthetase gene AceCS2. J Biol Chem. 2004;279(17):16954–62. pmid:14960588