Introduction

Adipose tissue is crucial for lipid storage, thermogenesis and endocrine regulation of whole-body metabolism. Adipose tissue mass is flexibly regulated, and its timely expansion in response to positive energy balance is now considered a beneficial response to prevent ectopic fat accumulation and lipotoxicity.1 Adipocyte hypertrophy and hyperplasia are the two main mechanisms that contribute to adipose tissue expansion. Adipocyte hyperplasia from precursor cells (adipogenesis) is an essential process for adipose mass homeostasis that replaces lost adipocytes throughout life. Therefore, it is critical for the maintenance of metabolic homeostasis, that is, alleviation of systemic insulin resistance by replacing hypertrophic adipocytes with new and small cells. In this regard, recent findings have suggested that impaired adipogenesis is causally linked to hypertrophic obesity phenotypes and metabolic complications.1 However, the molecular mechanisms responsible for adipogenesis are complicated because numerous hormones, lipid mediators and transcription factors are involved.

The wingless-type MMTV integration site family (Wnt)/β-catenin signaling pathway is one of the most important regulators of adipogenic differentiation. Among Wnt family members, Wnt10b is the major molecular switch that strongly represses adipogenesis,2 and its involvement in the pathogenesis of obesity has also been suggested.3 However, the regulatory mechanisms of Wnt10b expression during adipogenesis remain unclear.

MicroRNAs are small noncoding RNAs that function as guide molecules in RNA silencing. Myriad evidence has shown that they are involved in a variety of physiological processes. MicroRNAs also have important roles in adipogenesis, and their possible applications as biomarkers and therapeutic targets for obesity have been suggested.4 Currently, multiple microRNAs have been shown to regulate adipogenesis. Many key molecules involved in adipogenesis have been indicated as potential target molecules for microRNAs.25 It is strongly induced during adipogenic differentiation in mouse 3T3-L1 cells6, 26 and adipose-derived mesenchymal stem cells.7, 27 Moreover, miR-148a levels are increased in obese mice and humans.7 These findings strongly suggest an important role for miR-148a in the regulation of adipogenesis and in the development of obesity or other metabolic disorders. However, the precise regulatory mechanisms involved in its transcription remain unclear. From that point of view, the most notable finding in our study is that XBP1s may be a novel transcriptional activator for miR-148a in differentiating 3T3-L1 cells. When we altered the expression of XBP1 in 3T3-L1 cells, miR-148a levels changed accordingly. The levels of miR-148a were significantly downregulated by XBP1 knockdown during adipogenic differentiation (Supplementary Figure S1). In addition, we revealed an XBP1-specific RE in the promoter region of miR-148a. These findings clearly suggest that XBP1 is an upstream regulator of miR-148a transcription. However, when we compared expression patterns in the early differentiation period, no correlation was found between XBP1s and miR-148a (Supplementary Figure S1). As we previously reported, the level of XBP1 mRNA was increased in two phases: a strong peak at ~6 h and a less potent peak on day 6 after the induction.13 miR-148a expression, consistent with previous reports, was not increased within 24 h, but it began to increase at 6 days after adipogenic induction (Supplementary Figure S1).7, 26 In the same context, the cAMP-response element-binding protein (CREB), which is known to regulate adipogenesis during the very early period, has been recently suggested as a putative upstream regulator of miR-148a.7, 28 Therefore, it is reasonable to assume that additional transcriptional factors or cofactors induced by adipogenic differentiation stimuli may be needed for XBP1- or CREB mediated miR-148a production.

The Wnt/β-catenin pathway has been implicated in many pathophysiological aspects of adipose tissue metabolism and disorders. Therefore, manipulating the Wnt/β-catenin pathway in adipose tissue is an attractive drug-development strategy to combat obesity-associated metabolic complications. Of a total of 19 Wnts, Wnt1 and Wnt10b are regarded as typical anti-adipogenic isotypes,29 and the latter is considered to be the most potent anti-adipogenic Wnt ligand. Previously, we suggested that Wnt10b transcription is specifically repressed by XBP1s during adipogenic differentiation of 3T3-L1 cells,16 and, in this study, we showed the involvement of miR-148a in XBP1-mediated suppression of Wnt10b. A miR-148a mimic significantly downregulated Wnt10b at both the mRNA and protein levels by binding to the 3′UTR of Wnt10b mRNA in 3T3-L1 cells. Consistent with our findings, Aprelikova et al.22 recently suggested that Wnt10b is a potential target of miR-148a in cancer-associated fibroblasts. These findings provide novel information that XBP1s can downregulate Wnt10b expression through both transcriptional and miR-148a-mediated post-transcriptional mechanisms to facilitate adipogenesis of 3T3-L1 cells. However, the evidence provided in this work is not sufficient to determine the relative contribution of direct and indirect mechanisms to XBP1-induced Wnt10b repression. We may be able to determine such contributions by comparing the temporal expression patterns of XBP1s, Wnt10b and miR-148a during adipocyte differentiation. As mentioned earlier, the XBP1 mRNA level was increased in both the early and late phases, whereas miR-148a was increased only in the late phase of adipogenesis (Supplementary Figure S1). On the basis of these findings, the simplest explanation for this discrepancy in temporal expression patterns between XBP1s and miR-148a is as follows: XBP1 downregulates Wnt10b mRNA by a transcriptional mechanism during the early phase16 and by miR-148a-mediated post-transcriptional mechanism in the late phase of adipogenesis. Moreover, a miR-148a inhibitor increased Wnt10b mRNA levels only in the late phase after induction (Supplementary Figure S3), suggesting that miR-148a may have no effect on Wnt10b mRNA in the early phase. Further research will be needed to reveal the mechanisms involved in the time-dependent and interactive regulation among miR-148a, Wnt10b and XBP1s during adipogenesis.

Wnt10b inhibits adipocyte differentiation by blocking the induction of C/EBPα and PPARγ in the early phase of adipogenesis;30, 31 however, its role in the late phase is relatively unknown. What is the role of miR-148a-induced silencing of Wnt10b in the late phase of adipogenic differentiation? In this study, the miR-148a mimic downregulated the mRNA level of Wnt10b, and it concurrently rescued adipogenic potential and the expression of PPARγ2 in XBP1-KD cells (Figure 4). Therefore, it is plausible that in addition to directly suppressing Wnt10b transcription during the early phase of adipogenesis, XBP1s may intensify adipogenic differentiation by inducing miR-148a, which, in turn, can disinhibit Wnt10b-mediated suppression of PPARγ2 during the late phase of adipogenesis. Moreover, considering that important pro- and anti-adipogenic molecules such as PPARγ2,13 Wnt10b16 and C/EBPα14 are regulated by XBP1s, our data identifying miR-148a as an additional target molecule for XBP1s strongly support the idea that XBP1 plays a critical role in adipogenesis through multiple mechanisms (Figure 5).

Figure 5
figure 5

Diagram of XBP1-mediated adipogenesis. Accumulating evidence has indicated that XBP1 plays a critical role in adipogenesis by regulating important pro- and anti-adipogenic molecules such as PPARγ2, Wnt10b and C/EBPα. In this study, we demonstrate that miR-148a is a novel downstream effector in XBP1-mediated adipogenesis through silencing Wnt10b mRNA. Refer to the Discussion section for a detailed explanation. ACL, ATP-citrate lyase; DNMT1, DNA methyltransferase 1; XBP1, X-box binding protein 1.

Full size image

It is well known that each microRNA can regulate the expression of more than 100 target genes.32 Thus, it is not unlikely that molecular targets other than Wnt10b may be involved in XBP1-mediated adipogenesis. For example, DNA methyltransferase 1 was reported to be a target of miR-148a during adipogenesis,26 indicating that miR-148a might participate in XBP1-induced adipogenesis through Wnt-independent mechanisms (Figure 5). In addition, while we were preparing this manuscript, Wnt1 was suggested as a potential target of miR-148a in human adipose-derived mesenchymal stem cells.7 One notable finding from this study is that, in contrast to our findings, the Wnt10b level was not reduced by miR-148a. In this context, the transcription of only Wnt10b, but not Wnt1, was found to be suppressed by XBP1s in our previous report.16 With the information currently available, we could not explain the differential regulatory mechanisms involved in miR-148a-induced silencing of Wnt1 and Wnt10b. However, part of the answer to this question may be found in several lines of evidence, suggesting that microRNA functions are frequently cell-type-specific. For example, many studies have reported that the inhibitory effects of miR-148a on Wnt10b depend on the cellular context.22, 33, 34 It has also been reported that miR-148a silences DNA methyltransferase 1 in 3T3-L1 cells26 but not in human mesenchymal stromal cells,7 suggesting the presence of species-dependent activities of miR-148a. Further evidence is required to determine whether miR-148a activities are cell-type- or species-specific.

In summary, for the first time, we found that XBP1s could stimulate adipocyte differentiation by directly activating miR-148a-mediated suppression of Wnt10b. With the known regulatory function of XBP1s on the key pro- and anti-adipogenic molecules such as C/EBPα, PPARγ2 and Wnt10b, our data suggesting a novel mechanism involving miR-148a also support our hypothesis that XBP1 may be an essential regulator for adipogenesis.