Previous analyses performed by RNAse protection assays reported very weak levels of Akr1b8 transcript (Lau et al., 1995) in the mouse intestine and absence of expression in the adult liver. These observations were partially revised by Joshi et al. (2010) who showed high Akr1b8 expression all along the adult intestinal tract and moderate expression in the liver. Akr1b8 expression was initially described to be up-regulated during the early phase of the cell cycle and induced by growth promoting agents [fibroblast growth factor (FGF) and epidermal growth factor (EGF)], as well as by hypertonic stress (Donohue et al., 1994; Hsu et al., 1997). Although Akr1b8 displayed a strong expression in mouse colon cancer cells (Joshi et al., 2010) and was suggested to be a target gene for NF-E2 related factor 2 (Nrf2) transcription factor that controls intestinal detoxification response (Varady et al., 2011), its expression and regulation in healthy liver and intestine remained elusive. A recent study investigating bile acids impacts on Akr1b7 expression showed that enterohepatic expression of Akr1b8 was insensitive to bile acids (Schmidt et al., 2011). Akr1b8 enzymatic activities were extensively studied regarding its ability to reduce the highly reactive 4-HNE produced during peroxidation of polyunsaturated fatty acids (Srivastava et al., 1998). In vitro enzymatic studies reported that multiple members of the AKR1B group, including Akr1b8, were endowed with the ability to reduce phospholipid aldehydes derived from lipid peroxidation. Compared with AKR1B1 or Akr1b7, Akr1b8 activity was more efficient with the reduction of short chain aldehydes such as 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine and 1-pamitoyl-2-arachidonoyl-sn-glycero-3-phosphoglycerol (Spite et al., 2007). In line with the high lipid metabolic activities in Akr1b8 producing sites, it was proposed that Akr1b8 was acting physiologically as a scavenger of toxic aldehydes derived from peroxidation of endogenous or dietary lipids.

Beside its aldehyde reductase activity, Akr1b8 was shown to associate with the lipogenic acetyl-CoA carboxylase α (ACCA) in murine colon cancer cells (Joshi et al., 2010). ACCA is a rate-limiting enzyme of de novo synthesis of fatty acids, catalyzing the formation of malonyl-coA by ATP-dependent carboxylation of acetyl-coA. This interaction protects ACCA from proteasomal degradation and consequently allows increased fatty acid synthesis that leads to production of lipid second messengers that promote cell proliferation (Chajès et al., 2006). Whether Akr1b8 is also involved in the stability of ACCA in healthy enterocytes has not yet been determined. In order to associate gene signature with metabolic network, a genetic study integrating quantitative trait locus (QTL) mapping and network modeling led to knock-out (KO) the three corresponding identified genes, among which Akr1b8 (Derry et al., 2010). Akr1b8^−/− mice phenotype was very briefly described in this study suggesting a more specific perturbation in fat tissue homeostasis rather than in the liver or intestine (see below).

The AKR1B10 gene has recently been determined as the human ortholog of Akr1b8 (Joshi et al., 2010). Its expression was primarily detected in healthy colon, small intestine, and liver (Cao et al., 1998) and was found up-regulated with their corresponding tumorigenic transformation (Cao et al., 1998; Yan et al., 2007; Heringlake et al., 2010; Liu et al., 2012). This was also observed in lung and breast cancers suggesting that AKR1B10 overexpression could be associated with a broader tumor phenotype. In human hepatocarcinoma, insulin or EGF enhanced AKR1B10 expression through the activator protein-1 (AP1) mitogenic signaling. This enhanced tumor development and progression through elimination of cytotoxic carbonyls and promotion of lipogenesis (Liu et al., 2012). The treatment of human colon cancer SW-480 and HT-29 cell lines with proteasome inhibitors known to increase the expression of Nrf2-regulated genes induced AKR1B10 expression suggesting that AKR1B10 could be a target of the Nrf2 transcription factor (Ebert et al., 2011). Nrf2 responsiveness of AKR1B10 gene was further demonstrated by co-transfection experiments of AKR1B10 promoter luciferase reporter constructs in human lung adenocarcinoma cell lines (Nishinaka et al., 2011; Figure 3).

AKR1B10 was the first AKR1B protein which was demonstrated to enhance cell proliferation and promote cell survival through its ability to interact with and stabilize ACCA in breast cancer cells (Ma et al., 2008) and in colon cancer cells (Wang et al., 2009). AKR1B10 mediates ACCA stability through physical association. This was shown to affect fatty acid/lipid synthesis, mitochondrial function, and oxidative status. Identification of Akr1b8/AKR1B10 protein domain interacting with ACCA would be helpful to predict this characteristic for other AKR1B proteins and to develop targeted therapeutic strategies for the modulation of de novo fatty acid synthesis in cancer cells.

We are exposed daily to α,β-unsaturated aldehyde and trans-2-hexenal through food and drink consumption (Stout et al., 2008). This persistent exposure to electrophilic carbonyls requires an effective defense system to protect intestinal cells from irreversible damage. Similarly to Akr1b8, in vitro enzymatic studies on AKR1B10 demonstrated a highly specific ability to reduce physiological levels of 4-HNE and other alpha and beta-unsaturated carbonyls in both their free or glutathione-conjugated form (Martin and Maser, 2009; Zhong et al., 2009). Expression of AKR1B10 may reflect the requirement for intestinal cells to protect against the dietary, lumen microbial and lipid-derived carbonyls.

Although physiological integration of all these data remains difficult, this suggests that the beneficial role of Akr1b8/AKR1B10 in the protection of healthy cells against lipid peroxidation could switch, in the case of cancer, into a deleterious role promoting cell proliferation through stabilization of ACCA (Figure 4).

Both Akr1b3 and AKR1B1 are ubiquitously expressed. Akr1b3 transcripts were detected in the liver and small intestine at similar levels (Joshi et al., 2010). AKR1B1 messenger RNAs are found in the liver and small intestine (Cao et al., 1998). Western blot analyses confirmed AKR1B1 expression in the small intestine although it was undetectable in the normal adult liver. However AKR1B1 expression could be induced by alcoholic cirrhosis (O’Connor et al., 1999).

In diabetic mice, Akr1b3 is involved in glucose metabolism in some tissues through its participation in the polyol pathway. In response to chronic hyperglycemia, Akr1b3 catalyzes the rate-limiting reduction of glucose into sorbitol which is in turn converted into fructose by the sorbitol dehydrogenase (SDH). Activation of this pathway is involved in the development and progression of diabetic complications (Brownlee, 2001). Some evidences suggest that it could also be implicated in lipid metabolism. First, diabetic patients exhibit elevated blood triglycerides and non-esterified fatty acids and second, fructose-fed rats display an elevation of blood triglycerides and a reduction of peroxisome proliferator-activated receptor alpha (PPARα) activation (Roglans et al., 2007). These data led Qiu et al. to investigate the effects of hepatic activation of the Akr1b3/polyol pathway on lipid metabolism (Figure 6). In a murine hepatocyte cell line, overexpression of Akr1b3 was able to suppress PPARα transcriptional activity through extracellular signal-regulated kinase (ERK1/2)-mediated inactivating phosphorylations on Ser-12 and Ser-21. This resulted in decreased expression of genes involved in peroxisomal and mitochondrial β-oxidation pathway. In addition, Akr1b3 is up-regulated under high glucose concentrations in mouse hepatocyte AML12 cells and in vivo in diabetic mice. The genetic ablation or pharmacological inhibition of Akr1b3 in diabetic mice decreased ERK1/2-dependent phosphorylation of PPARα and blood triglycerides levels. In diabetic animals, SDH KO was accompanied by a reduction of blood triglycerides. This study clearly indicated that Akr1b3 and the polyol pathway sensed intracellular levels of glucose and adjusted PPARα activity through its phosphorylation/dephosphorylation, which in turn affected cellular lipid homeostasis (Qiu et al., 2008). In db/db mice, inhibition and knock-down of Akr1b3 induced the decrease of both plasma and liver triglycerides levels. Inhibition of Akr1b3 in db/db mice led to an improvement of hepatosteatosis due to the up-regulation of acetyl-coA oxidase and apolipoprotein A-V, two PPARα target genes involved in lipid catabolism (Qiu et al., 2012). Mechanisms by which Akr1b3 influences ERK-dependent inactivation of PPARα remain to be established.

AKR1B1 is also involved in the polyol pathway. Therefore, the relevance of these results in human remains to be determined. However, studies using Akr1b3^−/− mice or transgenic mice overexpressing AKR1B1 revealed no abnormalities in either the liver or small intestine (Yamaoka et al., 1995; Aida et al., 2000).

In addition to this possible implication in metabolism, AKR1B1 could be an actor of small intestine protection against electrophilic carbonyls. Its expression is also induced in human colon cancer cell lines SW-480 and HT-29, by proteasome inhibitors known to induce Nrf2 expression (Ebert et al., 2011). Characterization of Akr1b3 promoter has shown that Nrf2 could regulate its activity through an antioxidant response element 1 (−953 to −945) and an AP1 site (Nishinaka and Yabe-Nishimura, 2005). The arrangement of these elements in the stress response region was conserved between Akr1b3 and AKR1B1 promoters (Figure 3).

Moreover, in a similar way as AKR1B10, AKR1B1 displayed enzymatic properties allowing protection of intestinal cells against dietary electrophilic carbonyls. Indeed, AKR1B1 kinetic parameters are compatible with reduction of glutathione-conjugated carbonyl compounds (Shen et al., 2011). Studies in healthy tissues could allow a better understanding of AKR1B1 implication in cytoprotection against oxidative damage.

Until 2003, there was no mention of AR expression in white adipose tissue. In a transcriptomic study, Moraes et al. reported for the first time the detection of Akr1b7 mRNA in mouse white adipose tissue. In this study, the authors highlighted a decrease in Akr1b7 expression in abdominal white fat tissue from diet-induced obese mice when compared with mice fed a standard diet (Moraes et al., 2003). This decrease reflected either the down-regulation of Akr1b7 gene expression level per cell or the lower proportion of cells expressing the gene. Indeed, Akr1b7 expression appeared to vary depending on the location of white adipose depots and was found enriched in the stromal vascular fraction (containing adipocyte progenitors) whereas it was virtually absent from mature adipocytes (Tirard et al., 2007). Accordingly, Akr1b7 expression decreased during adipogenic differentiation of primary preadipocytes from the stromal vascular fraction and was detectable in the 3T3-L1 preadipocyte cell line.

The transcription factor SF-1, an essential regulator of Akr1b7 expression in the adrenal cortex is absent from adipose tissue. Instead, adipocyte precursors express liver receptor homolog-1 (LRH-1; Clyne et al., 2002), a nuclear receptor that also binds to SF-1 regulatory elements and can substitute for SF-1 in tissues where it is not expressed (Sirianni et al., 2002). Consistent with decreased expression of Akr1b7 during adipogenesis, LRH-1 expression is rapidly down-regulated during adipocyte differentiation (Clyne et al., 2002). We thus evaluated the possibility that LRH-1 stimulated Akr1b7 promoter activity in preadipocytes. Indeed, in co-transfection experiments in 3T3-L1 preadipocytes, LRH-1 induced Akr1b7 promoter through the SF-1 binding site at −458 (unpublished observations). Therefore, the dynamic expression profile of LRH-1 may account for expression of Akr1b7 in the preadipocytes-enriched stromal fraction (stromal vascular fraction) of various fat depots and its transitory expression in the early differentiation steps of 3T3-L1 cells.

Overexpression of Akr1b7 in 3T3-L1 preadipocytes prevented lipid droplets accumulation during adipogenic differentiation. In contrast, knock-down of Akr1b7 accelerated differentiation and lipid accumulation in 3T3-L1 cells. These results allowed us to demonstrate that Akr1b7 is a negative regulator of adipogenesis in vitro, which prevents differentiation by reducing lipid storage (Tirard et al., 2007). The mechanisms through which Akr1b7 inhibits adipocyte differentiation in vivo were recently unraveled by our group using Akr1b7^−/− mice. These will be discussed below.

Akr1b7 has two different enzymatic activities that could potentially be involved in the regulation of white adipose tissue homeostasis. First, just like Akr1b3, Akr1b7 is endowed with PGFS activity allowing synthesis of PGF_2α, a potent inhibitor of adipogenesis (Serrero et al., 1992; Kabututu et al., 2009). Indeed, PGF_2α is able to inhibit differentiation of 3T3-L1 into adipocytes, through binding to the FP receptor. Activation of FP receptor in turn blocks PPARγ and C/EBPα expression through a Gα_q-calcium-calcineurin-dependent signaling pathway (Liu and Clipstone, 2007). PGF_2α was also shown to block adipogenesis through activation of mitogen-activated protein kinase, resulting in inhibitory phosphorylation of PPARγ (Reginato et al., 1998). Second, Akr1b7 reduces 4-HNE, the accumulation of which is known to induce an excessive development of adipose tissue through adipocytes hypertrophy. Detailed analysis of Akr1b7^−/− mice allowed us to show that Akr1b7 behaved as an anti-adipogenic factor limiting white adipose tissue expansion, essentially through the regulation of PGF_2α levels in vivo (Volat et al., 2012; Figure 7).

Akr1b7^−/− mice displayed excessive basal adiposity resulting from both adipocyte hyperplasia and hypertrophy. They further exhibited increased sensitivity to diet-induced obesity. Following adipose enlargement and irrespective of the diet, they developed liver steatosis and progressive insulin-resistance. Akr1b7 loss was associated with decreased PGF_2α white adipose tissue contents. Cloprostenol (a PGF_2α agonist) administration in Akr1b7^−/− mice normalized white adipose tissue expansion by altering both de novo adipocyte differentiation and size. Treatment of 3T3-L1 adipocytes and Akr1b7^−/− mice with cloprostenol suggested that decreased adipocyte size resulted from inhibition of lipogenic gene expression. Hence, Akr1b7 is a major regulator of white adipose tissue development through at least two PGF_2α-dependent mechanisms: inhibition of adipogenesis and lipogenesis (Volat et al., 2012).

Although we showed that Akr1b8 is devoid of PGH₂ 9-,11-endoperoxide reductase activity (Kabututu et al., 2009), QTL mapping allowed identification of Akr1b8 as a possible regulator of adiposity in mouse (Derry et al., 2010). Akr1b8^−/− male mice had a tendency to be fatter than their wild-type littermates, under both standard and high fat diets. This was particularly obvious for the gonadal fat pad. Under high fat diet, Akr1b8^−/− mice also displayed higher serum cholesterol than their wild-type littermates. The mechanisms involving Akr1b8 in the regulation of white adipose tissue homeostasis remain completely unknown and are not necessarily resulting from expression of the gene in the adipose tissue. Indeed, we were unable to detect significant Akr1b8 protein expression in various adipose depots in mouse (Volat et al., 2012). Although we cannot exclude that Akr1b8 could be expressed in variable amounts depending on the mouse genetic background and the location of fat pads, it should be considered that the effect of Akr1b8 ablation could be indirect. Thus the involvement of Akr1b8 in adipose tissue homeostasis should be carefully examined with respect to its expression sites and enzymatic activities.

AKR1B10 expression in the adipose tissue has not yet been determined. Because AKR1B10 and Akr1b8 are not only detoxifying enzymes but also affect de novo fatty acids synthesis in cancer cells, it would be interesting to determine if and which of the enzymatic activities of AKR1B10 could be involved in adipose tissue homeostasis (Ma et al., 2008; Wang et al., 2009; Joshi et al., 2010).

Akr1b3 is expressed in undifferentiated 3T3-L1 preadipocytes and during the early phase of their differentiation. Akr1b3 can synthesize PGF_2α in 3T3-L1 cells, which inhibits their differentiation through stimulation of the FP receptor (Fujimori et al., 2010).

In addition to their PGF_2α synthase activity, Akr1b3 and AKR1B1 also catalyze the isomerization of PGH₂ to PGD₂ (Nagata et al., 2011). The impact of PGD₂ on white adipose tissue homeostasis is still very disputed. Indeed, lipocalin-type prostaglandin D synthase (L-PGDS^−/−) mice show hypertrophy of adipocytes (Ragolia et al., 2005). In contrast, the knock-down of L-PGDS decreases lipid accumulation in 3T3-L1 cells (Fujimori et al., 2007). In agreement with these observations, transgenic mice overexpressing (human hematopoietic-type prostaglandin D synthase (H-PGDS) overproduce PGD₂ and show signs of obesity and pronounced adipogenesis on a high fat diet (Fujitani et al., 2010). These investigations suggest that PGD₂ promotes adipogenesis in vivo.

Because of its involvement in the synthesis of PGF_2α and PGD₂, the involvement of Akr1b3 in white adipose tissue physiology may result from a balance between both enzymatic activities. However the relevance of these data has not yet been studied in vivo. Indeed, no defect in adipose tissue homeostasis have been reported in Akr1b3^−/− mice and the prostaglandin contents of their fat depots have not been analyzed (Aida et al., 2000; Ho et al., 2000).

Although the murine AR Akr1b3 could potentially participate in the homeostatic maintenance of adipose tissue, there is no available information about its human ortholog AKR1B1. Previous studies using transgenic mice with constitutive overexpression of AKR1B1 did not report any effect on adipose tissue (Yamaoka et al., 1995). On the basis of its enzymatic properties (PGF_2α and PGD₂ synthesis and reduction of 4-HNE), AKR1B1 could be considered as a potential actor of adipose tissue physiology (Kabututu et al., 2009; Nagata et al., 2011; Shen et al., 2011). However its expression pattern in human fat depots remains to be determined.