Aldose Reductases Influence Prostaglandin F_2α Levels and Adipocyte Differentiation in Male Mouse and Human Species

Aldose reductases (AKR1B) are widely expressed oxidoreductases whose physiological function remains elusive. Some isoforms are genuine prostaglandin F_2α (PGF_2α) synthases, suggesting they might influence adipose homeostasis because PGF_2α inhibits adipogenesis. This was shown by Akr1b7 gene ablation in the mouse, which resulted in increased adiposity related to a lower PGF_2α content in fat. Yet humans have no ortholog gene for Akr1b7, so the role of aldose reductases in human adipose homeostasis remains to be explored. We analyzed expression of genes encoding human and mouse aldose reductase isoforms in adipose tissues and differentiating adipocytes to assess conserved mechanisms regulating PGF_2α synthesis and adipogenesis. The Akr1b3 gene encoded the most abundant isoform in mouse adipose tissue, whereas Akr1b7 encoded the only isoform enriched in the stromal vascular fraction. Most mouse aldose reductase gene expression peaked in early adipogenesis of 3T3-L1 cells and diminished with differentiation. In contrast with its mouse ortholog Akr1b3, AKR1B1 expression increased throughout differentiation of human multipotent adipose-derived stem cells, paralleling PGF_2α release, whereas PGF_2α receptor (FP) levels collapsed in early differentiation. Pharmacological inhibition of aldose reductase using Statil altered PGF_2α production and enhanced human multipotent adipose-derived stem adipocyte differentiation. As expected, the adipogenic effects of Statil were counteracted by an FP agonist (cloprostenol). Thus, in both species aldose reductase-dependent PGF_2α production could be important in early differentiation to restrict adipogenesis. PGF_2α antiadipogenic signaling could then be toned down through the FP receptor or aldose reductases down-regulation in human and mouse cells, respectively. Our data suggest that aldose reductase inhibitors could have obesogenic potential.

Aldose reductases (AKR1B; Enzyme Classification 1.1.1.21) are cytosolic monomeric enzymes belonging to the aldo-keto reductase (AKR) superfamily that reduce aldehyde or ketone function into corresponding alcohol from various aliphatic or aromatic substrates. This superfamily encompasses more than 150 nicotinamide adenine dinucleotide (phosphate)(reduced)-dependent oxidoreductases distributed in all prokaryotic and eukaryotic kingdoms. In humans, three distinct aldose reductase isoforms encoded by three different genes have been characterized so far: AKR1B1 [human aldose reductase (AR); (1)], AKR1B10 [also designated as human small intestine reductase or aldose reductase-like-1) (2, 3)]; and AKR1B15 (4). Four mouse aldose reductase isoforms have been described: Akr1b3, also referred to as murine aldose reductase is encoded by the ortholog gene of AKR1B1 (5); Akr1b7 [previously named mouse vas deferens protein (6)]; Akr1b8, encoded by the ortholog gene of AKR1B10 [previously named fibroblast growth factor (FGF)-related protein 1 (7)]; and Akr1b16 (4).

Aldose reductases belong to one of the most characterized AKR subgroups. Indeed, human aldose reductase-B1 is notoriously associated with diabetic complications, resulting from its ability to reduce glucose into sorbitol in a nicotinamide adenine dinucleotide phosphate reduced+H⁺-dependent manner, which promotes osmotic and oxidative stresses. In addition to glucose conversion, human and murine aldose reductases display reductase activities for various substrates including aldehydes, retinoids, xenobiotics, and prostaglandins (PG). This wide range of substrates allows them to participate in many pathological processes related to diabetes, tumorigenesis (8–12), or inflammation (13–18). This has prompted the development of specific aldose reductase inhibitors for targeting these pathological manifestations, but their physiological function still remains elusive (19).

Activation of biological pathways that favor adipocyte differentiation from precursor cells results in an increase in the number of adipocytes that may be critical for overall metabolic balance. Therefore, adipocyte expansion is tightly regulated by factors promoting or inhibiting adipocyte development. Among paracrine/autocrine factors produced by white adipose tissue (WAT), PGF_2α acting through the prostaglandin F receptor (FP) coupled to MAPK and/or Ca²⁺ signaling was previously shown to suppress adipocyte differentiation by inhibiting the function/expression of the critical proadipogenic transcription factor peroxisome proliferator-activated receptor-γ (PPARγ) and CCAAT/enhancer-binding protein-α (C/EBPα) (20, 21).

We and others previously showed that mammalian aldose reductase isoforms can be distinguished according to their ability to produce PGF_2α through the reduction of the 9-,11-endoperoxide moiety of the common prostaglandin precursor PGH_2, supplied by cyclooxygenases COX-1 or COX-2 (16, 22). Human aldose reductase-B1 and mouse-b3 and -b7 were shown to possess this PGF_2α synthase activity, whereas aldose reductase-B10 and mouse-b8 isoforms were devoid of it (22). We recently reported that mice deficient for Akr1b7 displayed excessive basal adiposity, resulting from both adipocyte hyperplasia and hypertrophy under a normal diet and in the absence of increased food intake. Aldose reductase-b7 loss was associated with decreased PGF_2α levels in WAT. Moreover, cloprostenol (PGF_2α analog) administration in Akr1b7^−/− mice normalized WAT expansion by altering both de novo adipocyte differentiation and size (23), suggesting that this isoform was an important regulator of WAT homeostasis through PGF_2α-dependent mechanisms (23). Another isoform was proposed to fulfill antiadipogenic function in the mouse. Indeed, PGF_2α produced by aldose reductase-b3 was shown to suppress adipocyte differentiation in the mouse 3T3-L1 cell line (24). Yet the absence of an adipose phenotype in Akr1b3-deficient mice suggested that redundant functions from other isoforms might compensate for the lack of aldose reductase-b3 in vivo (25). Conversely, some specific expression properties of the b7 isoform could explain its nonredundant function as evidenced by the increased adiposity of Akr1b7-deficient mice. Humans have no direct ortholog genes for Akr1b7. Although aldose reductase-B1 was recently shown to be expressed in human subcutaneous adipose tissue (26), its contribution to WAT homeostasis was not established.

The aim of the present study was to provide quantification of mouse and human aldose reductase isoforms in adipose depots and during adipocyte differentiation in culture to explore conserved mechanisms regulating PGF_2α synthesis and to evaluate involvement of human aldose reductases in PGF_2α-mediated regulation of WAT homeostasis. We show herein that murine isoforms present specific expression patterns in adipose tissue/cells and that human aldose reductase could influence adipogenesis in a PGF_2α-dependent manner that can be pharmacologically modulated.

Materials and Methods

Biological materials

129/sv mice were housed in a room-controlled temperature with 12-hour light, 12-hour dark cycles in agreement with international standards for animal welfare. They were fed ad libitum with water and a Global diet (Harlan). Four- to 6-month-old male mice were killed by cervical dislocation, and different tissues, including fat pads from various locations were removed and immediately frozen in liquid nitrogen and stored at −80°C until use.

Human adipose tissues specimens (sc fat) were obtained from seven subjects who were undergoing abdominal surgery at Estaing Hospital, Clermont-Ferrand (Auvergne, France). They were aged between 28 and 61 years with a body mass index (BMI) higher than 25 kg/m². These samples were immediately frozen in liquid nitrogen. The hospital’s Ethics Committee approved this study.

Adipose tissue fractionation

Murine fat pads freshly excised were rinsed in Krebs-Ringer bicarbonate buffer, minced, and digested for 30 minutes at 37°C in DMEM (Invitrogen) containing 1 g/L of type II collagenase (Sigma-Aldrich). Undigested tissues were removed by filtration through a 250-μm nylon sieve. After centrifugation of the filtrate at 750 × g for 10 minutes, the floating adipocyte fraction was separated from the stromal vascular pellet, and each fraction was stored at −80°C until use.

Cell culture and treatment

Mouse 3T3-L1 preadipocytes were cultured, and differentiation was induced as previously described (27). Briefly, for the amplification step, cells were cultured in DMEM (Invitrogen) supplemented with 10% of bovine serum (Biowest), 2 mM L-glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin (Invitrogen) at 37°C in a humidified atmosphere with 5% CO₂. Then 80 000 plated cells/cm² were cultured in DMEM supplemented with fetal calf serum (FCS). Adipocyte differentiation of confluent 3T3-L1 cells was initiated with DMEM supplemented with 10% FCS, 500 nM dexamethasone, 0.1 μM insulin, and 500 μM isobutylmethylxanthine (Sigma-Aldrich).

After 48 hours, the culture medium was replaced with DMEM containing 10% FCS and 0.1 μM insulin. From the fifth day, cells were cultured with DMEM supplemented with 10% FCS until complete adipocyte differentiation. Human multipotent adipose-derived stem (hMADS)-3 cells were cultured and differentiated as previously described (28, 29). hMADS cells were routinely maintained in proliferation in DMEM (Lonza) containing 10% FCS, 2 mM L-glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin at 37°C in a humidified atmosphere with 5% CO₂. Cells were plated at 40 000 cells/cm² in growth medium containing 2.5 ng/mL FGF2 (Peprotech). Two days after seeding, FGF2 was removed from proliferation medium. On the next day, differentiation of confluent cells was induced with DMEM/Ham’s F12 (Lonza) supplemented with 0.86 μM insulin, 10 μg/mL transferrin, 1 μM dexamethasone, 100 μM isobutylmethylxanthine, 1 μM rosiglitazone, and 0.2 nM triiodothyronine (Sigma-Aldrich). At day 3, dexamethasone and isobutylmethylxanthine were removed. To evaluate the involvement of aldose reductases during adipogenesis, some hMADS cell batches were differentiated in the presence of either the aldose reductase inhibitor Statil, 1–10 μM (Santa Cruz Biotechnology), or dimethylsulfoxide (DMSO; vehicle) or in combination with 0.1 μM cloprostenol for various amounts of time as indicated in the figure legends. Cells were then harvested for further analysis.

Cell line authentication

3T3-L1 cell authenticity has been attested herein (see Figure 2) by analyzing their adipogenesis in real-time quantitative PCR (RT-qPCR) analyses using mouse specific primers for adipogenic genes and Oil-Red-O staining. hMADS-3 cell authenticity has been performed recently via analyzing their immunophenotype and their adipogenesis [see Supplemental Table 2 and Figure 2, respectively, of the report of Mohsen et al (30)].

Figure 2

Akr1b3, Akr1b7 and Akr1b8 expression levels are transiently increased during the earlier step of 3T3-L1 adipogenesis. At day 0 (d0), 3T3-L1 preadipocytes were induced to differentiate into adipocytes (inset, Oil-Red-O staining showing progression of adipocyte differentiation of 3T3-L1 cells at days 3 and 9). A, From day 2 to day 15 of culture, mRNA levels of Fabp4, Pparγ, Ptgfr, Akr1b3, Akr1b7, Akr1b8, and Akr1b16 were measured by RT-qPCR and normalized to Ppib. The qPCR Ct values from day 3 are indicated to compare relative expression level of the different Akr1b genes. mRNA quantifications were expressed as percentage of day 2 values. Statistical analyses were performed by one-way ANOVA followed by a Dunnett’s post hoc test. *, P < .05; **, P < .01; ***, P < .001. B, Akr1b3, Akr1b8, Cox-1, Cox-2, and FP protein accumulations were monitored by Western blot throughout 3T3-L1 differentiation. α-Tubulin was used as a loading control.

Oil-Red-O staining

hMADS cells were fixed in 4% paraformaldehyde and stained with Oil-Red-O (Sigma-Aldrich) as previously described (31). Measurement by spectrophotometry of Oil-Red-O staining was performed by dissolving the intracellular lipid droplets with isopropanol. Absorbance was measured at 490 nm.

Gene expression

Total RNA were extracted from human and mouse tissues or hMADS or 3T3-L1 cells using TRI reagent (Molecular Research Center, Inc). RNA concentration and purity were assessed spectrophotometrically using a NanoDrop 1000 (Thermo Scientific). One microgram of mRNA was reverse transcribed for 1 hour at 37°C with 5 pmol of random hexamer primers, 200 U of Moloney murine leukemia virus-reverse transcriptase (Promega), 2 mM deoxynucleotide triphosphates, and 20 U of RNAsin (Promega). Two microliters of a fifth dilution of cDNA were used in each PCR.

RT-qPCR was performed on a Mastercycler ep Realplex (Eppendorf) using MESA GREEN quantitative PCR master mix Plus for SYBR (Eurogentec). Amplification was performed as follows: initial denaturation at 95°C for 2 minutes, followed by 40 cycles of 94°C for 15 seconds, 60°C for 15 seconds, 72°C for 20 seconds. The specificity of each reaction was determined after completion of PCR cycling by analysis of the melting point dissociation curve generated for temperatures from 60°C to 95°C at 0.2°C/s. Each reaction was performed in duplicate for each sample and relative expression was calculated based on the standard curve method normalized to TBP for hMADS cells and Ppib for 3T3-L1 cells (see Supplemental Table 1 for primer sequence). Amplification efficiency was determined for each pair of primers and a 100% ± 5% efficiency conditioned their use in quantitative PCR analyses.

Murine aldose reductases mRNA absolute quantification was performed in various tissues following the method of Pfaffl (32). Briefly, cDNA of each aldose reductase and 18S RNA was cloned in a pGEM-T Easy Vector (Promega) and then was sequenced (GATC Biotech) to confirm the correct insertion of cDNA in plasmid. After mRNA reverse transcription, aldose reductase cDNAs resulting products were quantified by quantitative PCR using calibration curves with known concentrations of plasmid DNA between 10⁻² and 10⁻⁷ ng. Potential RNA quantity variations between tissue extracts were corrected with 18S RNA quantification (see Supplemental Table 2 for primer sequences). Results were expressed as femtomole of aldose reductase mRNA per microgram of total RNA.

For semiquantitative RT-PCR experiments, PCRs were performed by using GoTaq polymerase (Promega). PCR conditions were as follows: 5 minutes at 94°C followed by 22–33 cycles of 45 seconds at 94°C, 45 seconds at the optimal annealing temperature determined by gradient PCR (Actin, 22 cycles, 60°C; Acc, 25 cycles, 65°C; Zfp521, 27 cycles, 65°C; Akr1b3 27 cycles, 65°C; Akr1b7, 33 cycles, 69°C; Akr1b8, 30 cycles, 69°C; Akr1b16, 30 cycles, 65°C), and 30 seconds at 72°C with a final elongation step of 5 minutes at 72°C (Supplemental Table 1). Aliquots of PCR products were analyzed on a 2% ethidium bromide-stained agarose gel and signals were quantified with Multi Gauge Software suite (Fujifilm).

Western blot

Cellular and tissue samples were homogenized in cold extraction buffer containing 20 mM HEPES, 0.42 M NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 2 mM NaF, 2 mM Na₃VO₄, and protease inhibitors cocktail (Complete protease inhibitor cocktail tablets; Roche Diagnostics). After centrifugation at 4°C for 15 minutes at 13 000 rpm, concentration of soluble proteins was determined by the Bradford method (Bio-Rad Laboratories). Fifty micrograms of total proteins were subjected to a denaturing SDS-PAGE and electrotransferred onto Hybond-ECL membrane (GE Healthcare). Detections were performed using primary antibodies (Table 1) raised against FP receptor (1:500; Cayman Chemical) human COX-1 (1:500; Santa Cruz Biotechnology), murine COX-1 (1:500; Cayman Chemical), COX-2 (1:1000; Cayman Chemical), AKR1B1 and AKR1B10 (1:1000, kind gifts of Dr D. Cao, Department of Medical Microbiology, Immunology, and Cell Biology, Simmons Cancer Institute, SIU School of Medicine, Springfield, Illinois), FABP4 (1:500; R&D Systems), Akr1b3 (1:2000, L5), Akr1b7 (1:3000, L4), Akr1b8 (1:1000, L8) (33), α-tubulin (1:20 000; Sigma-Aldrich) and revealed with a peroxidase-conjugate antirabbit, antimouse, antigoat, or antirat secondary antibody (Production d’Anticorps, Réactifs Immunologiques and Services). Detection was performed using Immobilon Western chemiluminescent horseradish peroxidase substrate (Millipore), and signals were quantified with a DNR MF ChemiBis 3.2 camera and Multi Gauge Software suite (Fujifilm).

Table 1

Antibody Table

Peptide/Protein Target	Antigen Sequence (if Known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised (Monoclonal or Polyclonal)	Dilution Used
FP receptor	SMNSSKQPVSPAAGL	FP receptor polyclonal antibody	Cayman Chemical; number 101802	Rabbit; polyclonal	1:500
Human COX-1		Cox-1 (C-20): sc-1752	Santa Cruz Biotechnology; number sc-1752	Goat; polyclonal	1:500
Mouse COX-1	LMRYPPGVPPERQMA	COX-1 (murine) polyclonal antibody	Cayman Chemical; number 160109	Rabbit; polyclonal	1:500
COX-2	DPQPTKTATINASASHSRLDDINPTVLIK	COX-2 (murine) polyclonal antibody	Cayman Chemical; number 160106	Rabbit; polyclonal	1:1000
AKR1B1			Dr Deliang Cao	Rabbit; polyclonal	1:1000
AKR1B10			Dr Deliang Cao	Rabbit; polyclonal	1:1000
FABP4			R&D Systems; number MAB1443	Rat; monoclonal	1:500
Akr1b3	ALMSCAKHKDYPFHAEV	L5	Professor Anne-Marie Lefrançois-Martinez	Rabbit; polyclonal	1:2000
Akr1b7	DLLDARTEEDYPFHEEY	L4	Professor Anne-Marie Lefrançois-Martinez	Rabbit; polyclonal	1:1000
Akr1b8	LLPETVNMEEYPYDAEY	L8	Professor Anne-Marie Lefrançois-Martinez	Rabbit; polyclonal	1:3000
α-Tubulin		Monoclonal anti-α-tubulin antibody	Sigma-Aldrich; number T6074	Mouse; monoclonal	1:20 000

PGF_2α dosage

Culture media samples were collected every 72 hours, immediately frozen in liquid nitrogen, and stored at −80°C. PGF_2α released in culture media was measured by an enzyme immunoassay (EIA) and acetylcholinesterase-linked PGF_2α tracer (Cayman Chemical) according to the manufacturer’s instructions.

Statistical analyses

Results were expressed as means ± SD (or SEM when indicated). Statistical significance of differences between experimental groups was assessed using tests specified in the figure legends. All tests were performed using GraphPad Prism 5 (GraphPad Software).

Results

Comparative tissue expression of mouse aldose reductase genes

mRNA expression levels of the four mouse genes (Akr1b3, b7, b8, b16) were measured by RT-qPCR absolute quantification in a set of 15 tissues. PCR was performed using specific primers that target the most divergent part of the transcripts encompassing the 17 C-terminal amino acids encoding sequences and the 3′-untranslated region (Supplemental Figure 1). Overall, Akr1b3 had the widest expression, with levels ranging from 1.96 to 31.86 × 10⁻⁹ fmol/μg⁻¹ of total RNA in retroperitoneal WAT (rWAT) and testis, respectively (Table 2 and Supplemental Figure 2). It was also the major isoform in all the 15 examined organs except for vas deferens and adrenal. On the contrary, Akr1b7 showed very high expression levels but only in a restricted number of tissues. Its expression culminated in vas deferens and adrenal (25 000 and 2000 times more than in other tissues, respectively). Secondary expression sites for the Akr1b7 were gonadal WAT (gWAT), intestine, and testis (1.78, 0.52, and 0.35 × 10⁻⁹fmol/μg⁻¹ of total RNA, respectively). Akr1b8 had rather low expression levels except in the adrenal and testis (3.95 and 4.68 × 10⁻⁹fmol/μg⁻¹ of total RNA, respectively). Akr1b16 mRNA highest levels were observed in the adrenal (6.94 × 10⁻⁹fmol/μg⁻¹ of total RNA) but were low in vas deferens (2.51 × 10⁻⁹fmol/μg⁻¹ of total RNA) and in kidney (1.23 × 10⁻⁹ fmol/μg⁻¹ of total RNA). Altogether these data indicated that the adrenal gland was a major common expression site for mouse aldose reductase genes.

Table 2

Absolute Quantification of mRNA for Akr1b Isoforms in Mouse Tissues

	×10−9 fmol/μg−1 of Total RNA
	Akr1b3	Akr1b7	Akr1b8	Akr1b16
gWAT	7.04 ± 1.38	1.78 ± 1.553	1.32 ± 0.443	0.80 ± 0.325
iWAT	2.40 ± 0.55	0.08 ± 0.05	0.55 ± 0.29	0.28 ± 0.14
rWAT	1.96 ± 0.51a	0.08 ± 0.02	0.86 ± 0.385	0.66 ± 0.33
BAT	8.34 ± 2.17	0.01 ± 0.002	0.73 ± 0.15	0.88 ± 0.054
Testis	31.86 ± 4.191a	0.35 ± 0.055	4.68 ± 0.491a	0.33 ± 0.19
Intestine	7.47 ± 4.20	0.52 ± 0.084	0.26 ± 0.16	0.08 ± 0.07
Kidney	27.65 ± 3.912	0.16 ± 0.01	0.22 ± 0.08	1.23 ± 0.553
Adrenal	17.61 ± 2.223	311.93 ± 46.932	3.95 ± 0.712	6.94 ± 1.131a
Spleen	3.60 ± 0.26	0.16 ± 0.06	0.12 ± 0.07	0.03 ± 0.004
Liver	14.80 ± 3.274	0.15 ± 0.07	0.05 ± 0.02	0.02 ± 0.004a
Vas deferens	4.47 ± 0.51	3994.30 ± 149.661a	0.27 ± 0.06	2.51 ± 0.532
Lung	2.70 ± 0.15	0.01 ± 0.001a	0.31 ± 0.03	0.20 ± 0.02
Brain	10.58 ± 3.005	0.03 ± 0.01	0.09 ± 0.04	0.72 ± 0.13
Muscle	2.53 ± 1.18	0.05 ± 0.01	0.03 ± 0.01a	0.35 ± 0.02
Heart	6.65 ± 1.17	0.03 ± 0.02	1.00 ± 0.084	0.75 ± 0.21

RT-qPCR analyses were performed using total RNA from a set of 15 tissues (n = 4–9) from 4-month-old male mice. Each isoform level was expressed as femtomoles of the targeted mRNA in 1 μg of total RNA ± SEM. The top 5 expression sites for each Akr1b are in bold, and ranking is numbered on the right.

^a

Highest and lowest expression sites for each Akr1b gene.

In the present quantitative analysis, although brown adipose tissue (BAT) and WAT expressed low mRNA levels of aldose reductase genes, Akr1b3 was still the predominantly expressed member (seven times more than other members), whereas Akr1b8 and Akr1b16 genes were expressed at lower levels (0.86 and 0.65 × 10⁻⁹ fmol/μg⁻¹ of total RNA, respectively) (Table 2). Akr1b7 showed a somewhat unique expression pattern characterized by detectable expression in gWAT and almost undetectable expression in BAT, inguinal WAT (iWAT), and rWAT. This suggested that fat pad-specific mechanisms could modulate its expression in adipose tissues.

Differential expression of aldose reductase genes in adipose fractions

To further characterize aldose reductase genes expression in adipose tissues, we performed cell tissue fractionations. In these experiments, WAT is fractionated into two different cell populations: adipocytes and a heterogeneous population collected in the stromal vascular fraction (SVF) encompassing undifferentiated precursors, macrophages, fibroblasts, leukocytes, epithelial, endothelial, and vascular cells (34). Detection of adipocyte acetyl-coenzyme A carboxylase (Acc) and progenitors-enriched zinc finger protein 521 (Zfp521) allowed confirmation of the quality of tissue fractionation (Figure 1, A–C). In all WAT depots, Akr1b3 expression was detected in both adipocyte and SVF fractions with at least a 2-fold enrichment in adipocytes. In all depots, Akr1b8 and Akr1b16 were expressed similarly in both fractions except for rWAT in which Akr1b16 mRNA was enriched 1.8-fold in mature adipocytes. Thus, expression of Akr1b3, b8 and b16 was detected in both mature adipocytes and in the adipocyte progenitor/precursor-containing SVF. In contrast, Akr1b7 transcripts were absent from adipocytes fraction but were enriched in the SVF from gWAT. Differential expression of aldose reductase-b3 and -b7 (members endowed with PGF_2α synthase activity) was confirmed by Western blot analysis of adipose tissue fractions (Figure 1D). Altogether these data showed coexpression of most genes encoding aldose reductase isoforms in both WAT cellular fractions, whereas Akr1b7 was the only member of the subfamily to be exclusively expressed in SVF.

Figure 1

Differential expression of aldose reductase genes in stromal vascular and adipocyte fractions. Aldose reductase relative mRNA expression was evaluated by RT-PCR after tissue fractionation of gWAT, iWAT, and rWAT depots. A–C, Quantifications of RT-PCR signals. RT-PCR analyses were conducted using four independent RNA pools, each resulting from the fractionation of two to six fat pads. Signal intensity was quantified for each PCR product and normalized according to an Actin signal. For each gene, the mRNA level was expressed relative to the signal obtained for SVF, which was given the value 1. n.d., not detected. Statistical analyses were performed to compare expression of each gene in SVF and adipocyte fraction by unpaired t test. *, P < .05; **, P < .01; ***, P < .001. D, Akr1b3 and Akr1b7 proteins were monitored by Western blot on extracts from SVF and adipocyte fraction. α-Tubulin was used as a loading control.

Differential expression of murine aldose reductase genes during adipogenesis in 3T3-L1 cells

Progenitor cells from SVF constitute a reservoir for adipose tissue maintenance when committed into an adipocyte fate. The mouse preadipocyte 3T3-L1 cell line is a well-characterized and widely used in vitro model to study adipocyte differentiation. To compare mouse aldose reductases expression during adipogenesis, 3T3-L1 differentiation was induced over a 2-week period. RT-qPCR analyses showed increased expression of Fabp4 and Pparγ genes in 3T3-L1 cells at differentiation days 3 and 6, confirming adipocyte differentiation also attested by triglyceride accumulation using Oil-Red-O staining (Figure 2). In comparison with Akr1b3, b8, or b16, the Akr1b7 gene was expressed at very low levels in 3T3-L1 cells (detection threshold quantitative PCR cycle threshold (Ct) values at approximately 33 for Akr1b7 mRNA vs 22–28 for the other isoforms) (Figure 2A). Whereas Akr1b16 expression was unaltered throughout the culture period, Akr1b3, b7, and b8 mRNA levels increased from the onset of the adipogenic program (2.7-, 6-, and 2-fold, respectively) at differentiation day 3 and then progressively returned to basal expression levels found in undifferentiated cells. PGF receptor gene (Ptgfr) mRNA levels (encoding the FP receptor) underwent a transient increase (4-fold) at differentiation day 3, paralleling Akr1b3, b7, and b8 expression patterns (Figure 2A). Moreover, aldose reductase-b3, -b7, -b8, and FP protein levels evaluated by Western blot paralleled their mRNA (Figure 2B). Finally, as previously shown (35, 36), we confirmed that COX-1 protein was constitutively expressed throughout the differentiation program, whereas COX-2 was acutely down-regulated from day 3 onward. We concluded that with the exception of Akr1b16, Akr1b3, b7, and b8, expression levels were all transiently increased during the early steps of adipogenesis with maximal expression at day 3 (at the onset of differentiation) and thereafter decreased during adipogenesis.

Differential expression of human aldose reductases during adipogenesis in hMADS cells

Involvement of human aldose reductases in adipose tissue homeostasis is poorly documented. Recent data reported a positive correlation between PGF_2α release in human preadipocytes isolated from obese women, BMI, and cytokine-induced AKR1B1 expression (26). In contrast, the expression of the AKR1B10 gene as well as that of AKR1C3 and AKR1A1, two genes encoding enzymes (3α-hydroxysteroid dehydrogenase type II and aldehyde reductase, respectively) with known PGF synthase activity, has never been studied in fat cells (37, 38). We also monitored the mRNA levels for COX-2 (PTGS2 gene) and FP_A/FP_B receptors (PTGFR1/PTGFR2 genes). RT-qPCR analyses showed no correlation between BMI and expression of these genes (Supplemental Figure 3). Western blot using specific antibodies (39) and RT-qPCR analyses confirmed AKR1B1 expression in intact sc WAT from obese patients, but no correlation with BMI was shown in our study. The discrepancy with data from Michaud et al (26) could result from the small size of our cohort. Although AKR1B10 transcripts were detected in rather low amounts (average Ct 29.4), the protein remained undetectable in these WAT samples. This suggests that AKR1B10 expression could either be absent or restricted to a minor population within sc WAT.

To analyze the expression of human aldose reductase isoforms during adipogenesis, we used a hMADS cell line that can be differentiated into functional adipocytes under adipogenic culture conditions (40). Adipogenesis was monitored by measuring changes at mRNA or protein levels of markers of the predifferentiation state (ZNF521, a regulator of adipose commitment ortholog to murine Zfp521) and early (PPARγ) and late (LEPTIN, FABP4/aP2) differentiation (Figure 3B and Supplemental Figure 4A). In agreement with progressive adipogenic differentiation, PPARγ mRNA expression increased from day 3 to reach a plateau between day 9 and day 12. As previously described in other preadipocyte lines, a 1.5-fold increase in the ZNF521 mRNA level was transiently observed at day 0 in hMADS cells (41). Consistent with previous reports on leptin production, the highest LEPTIN mRNA level observed at day 6 was decreased 2-fold at day 9 (40). In agreement with full maturation of hMADS adipocytes, fatty acid binding protein 4 (FABP4)/adipose protein 2 (aP2) protein began to accumulate on day 9 (Supplemental Figure 4A). The lowest expression of AKR1B1 mRNA was observed at day 3 and progressively increased to reach 210% of basal level at day 15 (Figure 3A). This expression time course was tightly correlated with changes in aldose reductase-B1 protein accumulation (Supplemental Figure 4A). In contrast, AKR1B10 mRNA levels had their highest expression at day 0 and gradually decreased below 11% of the basal level throughout differentiation (Figure 3A). Unlike AKR1B1 and AKR1B10, AKR1A1 expression was only slightly and transiently increased (1.6-fold) between differentiation day 3 and day 6. mRNA level for AKR1C3 was strongly induced (58-fold) from day 3 onward. Finally, COX-2 expression (PTGS 2 gene) was high in preadipocytes and rapidly turned off after the onset of adipogenesis between day 0 and day 3 (Figure 3C). Taken together, these data suggested that AKR1B10 and AKR1B1 expression could be differentially involved in adipogenesis.

Figure 3

AKR1B1 and AKR1B10 display opposite expression profiles during hMADS cell adipogenesis. At day 0 (d0), hMADS preadipocytes were induced to differentiate into adipocytes. A, AKR1B1, AKR1B10, AKR1A1, and AKR1C3 mRNA levels were measured by RT-qPCR throughout adipogenesis and values were normalized to TBP. B, To validate and follow the progress of adipogenic program, PPARγ, ZNF521, and LEPTIN gene expressions were analyzed by RT-qPCR and values were normalized to TBP. C, Level of PTGS2 transcripts were measured by RT-qPCR and normalized to TBP. mRNA quantifications were expressed as a percentage of day 3 values. Statistical analyses were performed by one-way ANOVA followed by a Dunnett’s post hoc test, each value was compared with day 3 value. *, P < .05; **, P < .01; ***, P < .001.

Aldose reductase-mediated PGF_2α release during hMADS cell adipogenesis

Previous in vitro enzymatic assays using human recombinant proteins showed that aldose reductase-B1 had PGF synthase activity whereas B10 isoform was completely devoid of such activity (22). Therefore, we evaluated the functional link between AKR1B1 expression and PGF_2α production during adipogenic differentiation of hMADS cells. Under differentiation conditions, PGF_2α production rate of flow from hMADS cells increased progressively throughout adipogenesis to reach a 3-fold induction in mature hMADS adipocytes at day 15 (Figure 4A). This increase in PGF_2α output was tightly correlated with changes in AKR1B1 expression at both mRNA and protein levels (Figures 3A and 4A and Supplemental Figure 4B) and independent from changes in the gene expression of other PGF synthases, ie, AKR1A1 and AKR1C3 (Figure 3A). Consistent with the observations from mouse 3T3-L1 cells, expression of the rate-limiting enzymes COX was either constitutive (COX-1) or inhibited (COX-2) upon induction of differentiation and remained very low during hMADS cell adipogenesis (Figure 4B).

Figure 4

PGF_2α flow is correlated to AKR1B1 expression during hMADS cell adipogenesis and is sensitive to Statil inhibitor. A, Culture media were collected every 3 days from day 0 to day 15 during hMADS cell differentiation and levels of released PGF_2α were determined by EIA. AKR1B1 protein was detected by Western blot in hMADS cells extracts. α-Tubulin was used as a loading control. B, COX-1, COX-2, FP_A, and FP_B accumulation were analyzed by Western blot during hMADS differentiation. α-Tubulin was used as a loading control. C, Effects of aldose reductase Statil inhibitor on cumulative PGF_2α production of hMADS cells. hMADS cells were treated with Statil or DMSO (vehicle) during the entire 15-day differentiation period. Culture media were collected every 3 days and PGF_2α levels measured by EIA. D, hMADS cells were differentiated for 9 days in the absence (vehicle) or presence of 10 μM Statil, fixed, and stained for triglycerides with Oil-Red-O. To quantify staining, Oil-Red-O was extracted from differentiating adipocytes and absorbance was then measured at 490 nm. E, hMADS cells were differentiated during 15 days in the presence of Statil or DMSO (vehicle) and expression of AKR1B1, AKR1B10, AKR1A1, AKR1C3, and PTGS 2 genes were analyzed by RT-qPCR, normalized to TBP, and expressed as a percentage of vehicle treatment values. Statistical analyses in panel A were performed by one-way ANOVA followed by a Dunnett’s post hoc test. Statistical analyses in panel C were performed by a two-way ANOVA followed by a Bonferroni post hoc test. In panels D and E, statistical analyses were performed by an unpaired t test. *, P < .05; **, P < .01; ***, P < .001.

As shown in Figure 4B, expression of both FP_A and FP_B isoforms of PGF_2α receptor was maximal in proliferative preadipocytes (day −3). It then decreased dramatically after day 3 (FP_A, 17% of day −3 value) or day 6 (FP_B, 6% of day −3 value) and remained low in maturing adipocytes. This suggested that PGF_2α released by adipocytes may essentially exert its antiadipogenic action in undifferentiated hMADS cells. Soon after differentiation commitment, FP receptors were turned off to trigger the adipogenic program and maintained at low levels throughout terminal differentiation. These low but detectable amounts of FP proteins coexisting with high PGF_2α levels would imply that PGF_2α could also have some function in mature adipocytes.

Exposure of hMADS cells to the specific aldose reductase inhibitor Statil (Ponalrestat) over the 2-week differentiation period resulted in an 18% reduction in cumulative PGF_2α production (Figure 4C). Remaining PGF_2α accumulation could be attributed to constant expression of AKR1A1 or AKR1C3 (Figure 3A) whose PGF synthase activities are insensitive to Statil (38). The ability of hMADS cells to accumulate intracellular lipids when exposed to adipogenic medium in the absence or presence of Statil was then examined by Oil-Red-O staining. As shown in Figure 4D, lipid accumulation was significantly increased by Statil treatment. As expected, Statil had no effect per se on AKR1B1, A1, B10, or C3 gene expression (Figure 4E). We thus concluded that increased PGF_2α release during hMADS adipogenesis was partially associated with aldose reductase-B1 activity that might contribute to delay adipocyte differentiation.

Interestingly, the reduction in PGF_2α production resulting from hMADS Statil treatment led to a concomitant decrease in COX-2/PTGS 2 expression (Figure 4E), whereas COX-1 was unaltered (not shown). This observation was in agreement with previous demonstration of a PGF_2α-dependent up-regulation of COX-2 gene in 3T3-L1 cells (42). Our data thus confirm such a positive feedback loop for PGF_2α in human adipocytes.

Role of aldose reductase-dependent PGF_2α production on the adipogenic program

Our kinetics expression data underscored the main changes occurring during the first 3 days of differentiation for most of the actors involved in PGF_2α production/response. Further experiments were conducted to evaluate the effect of Statil over the first 3 days of differentiation and the impact of aldose reductase-dependent PGF_2α production during this critical period (Figure 5). Compared with the vehicle condition, 10 μM Statil treatment resulted in a 27% decrease in PGF_2α release associated with a concomitant increase in the expression of the transcripts for proadipogenic regulators C/EBPβ and C/EBPα and for adipocyte marker ATGL, the lipolytic adipose triglyceride lipase. As expected, this treatment had no effect on AKR genes or on COX-1 expression and was too brief to alter COX-2 mRNA levels (not shown). To evaluate the ability of PGF_2α to reverse the Statil-dependent increase in adipose conversion, hMADS cells cultured in adipogenic medium for 6 days were concomitantly exposed to the FP agonist cloprostenol in the presence of Statil. As shown in Figure 5, C and D, cloprostenol exposure counteracted Statil effects on both intracellular lipid accumulation and expression of transcripts for proadipogenic factors C/EBPα and PPARγ. Altogether our data indicate that, in hMADS cells, Statil exerts a primary proadipogenic effect at least by inhibiting aldose reductase-dependent PGF_2α synthesis.

Figure 5

Statil enhances hMADS adipocyte differentiation by inhibiting aldose reductase-dependent PGF_2α synthesis. A, Effects of aldose reductase Statil inhibitor on PGF_2α release in culture media were determined by EIA in hMADS cells treated with Statil or its vehicle during the first 3 days of differentiation. B, mRNA levels of C/EBPβ, C/EBPα, and ATGL were monitored by RT-qPCR and normalized to TBP. Results were expressed as percentage of vehicle treatment values. C and D, Effects of Statil and cloprostenol (PGF_2α analog) treatments were examined after 6 days of differentiation on C/EBPα and PPARγ expression by RT-qPCR (C) and on intracellular lipid accumulation by Oil-Red-O staining (D; bright field and contrast phase views). Statistical analyses in panels A and B were performed by one-way ANOVA followed by a Dunnett’s post hoc test, each value was compared with vehicle value. Statistical analyses in panel C were performed by a two-way ANOVA followed by a Bonferroni post hoc test. *, P < .05; **, P < .01, ***, P < .001.

Discussion

Previous data showing that certain aldose reductases are endowed with genuine PGF_2α synthase activity has provided the first evidence for their possible influence on endocrine or metabolic functions [(16, 22, 33); for review see reference 43]. More recently constitutive Akr1b7 gene knockout in mice allowed the in vivo demonstration that decreased PGF_2α production resulting from aldose reductase-b7 loss could affect adipose tissue homeostasis without compensation by other related isoforms (23). Here we performed comparative and quantitative analysis of the expression of aldose reductase (Akr1b subfamily) genes in a set of mouse tissues with a focus on adipose depots and adipose tissue fractions. Our data highlighted the unique adipose-specific features of mouse aldose reductase-b7, in line with its antiadipogenic action in vivo and revealed that in humans aldose reductase-B1 isoform might also regulate adipogenic differentiation at least by a PGF_2α-dependent mechanism.

Aldose reductase isoforms in adipose tissues: minor expression sites for a probable major physiological function

Although aldose reductase-b3 remains the most abundant isoform in mouse adipose tissues, our quantitative studies reveal that fat represents a common expression site for all aldose reductase-like genes (at least one of their top 5 tissues is adipose tissue, Table 2). Previous in vitro experiments have anticipated the possible impact of these proteins on adipose tissue homeostasis. We reported that aldose reductase-b7 had antiadipogenic action when overexpressed in 3T3-L1 preadipocytes (31). Although the underlying mechanisms were not elucidated at that time, later evidence that it was endowed with PGF synthase activity suggested that aldose reductase-b7-mediated PGF_2α production was likely to support inhibition of adipocyte differentiation (22, 33). This was definitively proven in vivo by using Akr1b7 knockout mice that developed excessive adiposity in correlation with decreased accumulation of PGF_2α (23). In the meantime, it was shown that endogenous aldose reductase-b3 could also fulfill the PGF_2α-dependent suppression of adipocyte differentiation in the T3-L1 cell line (24). Interestingly, these authors reported that Akr1b3 expression was regulated throughout the 3T3-L1 differentiation program, mRNA levels showing a transient increase during the first day after the initiation of differentiation and then returning to low levels in maturing cells.

Although we broadly confirm these previous observations, we reveal herein that all aldose reductase genes (except Akr1b16 that was constitutive) and to a lesser extent Ptgfr gene (encoding FP receptor) are differentially expressed during 3T3-L1 adipogenesis with transient and maximal mRNA levels coinciding with the onset of differentiation and decreasing thereafter. Moreover, this observation was confirmed at the protein levels, suggesting that their coordinated down-regulation could be required for completion of the adipogenic program. Although both aldose reductases-b3 and -b7 were reported to negatively regulate 3T3-L1 differentiation through PGF synthase activity, we show that endogenous aldose reductase-b7 amounts are very low compared with that of -b3 (Figure 2). Conversely, Akr1b3 ablation in mice had no impact on adipose tissues, whereas loss of Akr1b7 induced, as expected, an expansion of fat (23). Therefore, aldose reductases function in adipose tissue should be investigated in vivo as much as possible. Here we report in vivo characterization of Akr1b genes expression within the two functional fractions of fat depots, ie, adipocytes and SVF. These data underscore the unique signature of Akr1b7 expression, which is the only gene of the four Akr1b genes in mouse to show a depot-specific pattern and absence of expression in mature adipocytes (Table 2 and Figure 1). Akr1b3 shows higher expression in the adipocyte fraction, thus contrasting with 3T3-L1 data (Figure 2 and reference 24). Altogether these observations suggest that aldose reductase-b7 functions in WAT are not overlapped by other isoforms, which is in good agreement with the adipose phenotype of Akr1b7^−/− mice. The identification of the cell population within SVF, which supports Akr1b7 expression and relays its antiadipogenic action, is currently in progress.

Antiadipogenic action of aldose reductase-dependent PGF_2α production is conserved in humans

Aldose reductase-B1 isoform was shown to be expressed in primary preadipocytes isolated from human fat tissue, in close correlation with cytokine-induced PGF_2α release. However, its impact on adipogenesis remains unknown (26). Our study suggests that the mode of action of these enzymes on adipose tissue homeostasis could differ in human and mouse. Using multipotent cells isolated from human adipose tissue (hMADS line), we show that expression of AKR1B1, the human homolog of mouse Akr1b3, parallels hMADS adipogenic differentiation in close correlation with PGF_2α release. Importantly, we demonstrate that inhibition of PGF_2α production using a specific aldose reductase inhibitor (Statil) during hMADS adipogenesis enhances adipocyte differentiation, whereas exogenous PGF_2α counteracts Statil effects (Figure 5). Hence, in human as in mouse, the aldose reductase-dependent production of PGF_2α might participate in restricting differentiation of preadipocytes in a cell autonomous manner but the means of this restriction may differ. Unlike Akr1b3 in 3T3-L1, the expression of AKR1B1 is not down-regulated but steadily increases throughout adipogenesis. However, PGF_2α receptors expression is turned down after day 3 of hMADS differentiation (compare Figures 2 and 4). This suggests that the need to overcome the repressive effect of PGF_2α on adipogenesis might be mainly achieved by limiting its signaling potential (by down-regulating FP receptors) in human cells and by limiting its synthesis (by down-regulating aldose reductase-b3 and -b7 PGF synthases) in mouse preadipocytes. Nevertheless, we show that in both species, genes encoding aldose reductase-like devoid of PGF synthase activity, ie, Akr1b8 and AKR1B10 show maximum mRNA accumulation in undifferentiated adipocytes and are down-regulated in maturing adipocytes (Figures 2 and 3A).

In terms of structural conservations, substrate spectra and tissue distribution, murine aldose reductase-b8 is considered as the ortholog of human aldose reductase-B10 (44). In addition to PGF_2α, other signaling molecules such as those produced by the vitamin A (retinol) metabolism, retinoic acid and retinaldehyde (Rald) are known to negatively regulate adipogenesis (45, 46). Interestingly both aldose reductase-B10 and aldehyde reductase-C3 display the best Rald reductase activity among aldose and aldehyde reductases (for review see reference 47). Once synthesized, Rald concentration depends both on its rate of irreversible oxidation to retinoic acid and on its rate of reduction back to retinol. Although we showed that Akr1b8/AKR1B10 and AKR1C3 are differentially expressed after adipogenic stimulation, further experiments will be needed to unravel the possible impact of these genes as regulators of adipogenesis by influencing retinoid metabolism.

Taken together, our findings establish aldose reductases as previously unrecognized contributors of adipose tissue homeostasis. By comparing human and mouse isoforms in preadipocyte culture models, we propose that the two species have developed different but convergent strategies to control PGF_2α production and its biological impact on fine-tuning early adipocyte differentiation. Finally, our findings establish the background for further evaluation of pharmacological aldose reductase inhibitors, initially designed to treat diabetic complications, as potential obesogenic compounds.

Acknowledgments

We thank all the members of the team for helpful discussions. We also thank Sandrine Plantade, Khirredine Ouchen, and Philippe Mazuel for excellent animal care.

This work was supported by the Université Blaise Pascal, the Université d’Auvergne, the Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale and by a grant from the Conseil Régional d’Auvergne Contrat de Plan Etat Région/Fonds Européens de Développement Régional 2010 Laboratoire d’Excellence and a PhD grant from the Société Française d’Endocrinologie.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• AKR1B1

  aldo-keto reductase family 1 member B1

 
• AKR

  aldo-keto reductase

 
• aP2

  adipose protein 2

 
• BAT

  brown adipose tissue

 
• BMI

  body mass index

 
• C/EBPα

  CCAAT/enhancer-binding protein α

 
• COX

  cyclooxygenase

 
• CREB

  cAMP-responsive element binding protein

 
• Ct

  cycle threshold

 
• DMSO

  dimethylsulfoxide

 
• EIA

  enzyme immunoassay

 
• FABP4

  fatty acid binding protein 4

 
• FCS

  fetal calf serum

 
• FGF

  fibroblast growth factor

 
• FP

  prostaglandin F receptor

 
• gWAT

  gonadal WAT

 
• hMADS

  human multipotent adipose-derived stem cell

 
• iWAT

  inguinal WAT

 
• PGF_2α

  prostaglandin F_2α

 
• PPARγ

  peroxisome proliferator-activated receptor-γ

 
• PTGFR

  prostaglandin F receptor gene

 
• Rald

  retinaldehyde

 
• RT-qPCR

  real-time quantitative PCR

 
• rWAT

  retroperitoneal WAT

 
• SVF

  stromal vascular fraction

 
• WAT

  white adipose tissue

 
• Zfp521/ZNF521

  zinc finger protein 521.