Is miR‐21 a potent target for liver fibrosis?

Potential conflict of interest: Nothing to report. SEE ARTICLE ON PAGE 2414 Supported by National Institutes of Health grants R01 AA025342, R21AA023599, and R21AA023607. MicroRNAs (miRNAs), small noncoding RNAs, regulate gene expression posttranscriptionally by binding to targeted sites within the 3′ untranslated regions of messenger RNAs (mRNAs) and play a role in a wide range of physiological and pathophysiological conditions, including liver fibrosis. The miRNA miR‐21 has been reported to be up‐regulated in patients1 and mouse models1 of liver fibrosis, and the results of previous studies suggest that its deletion or suppression can reduce fibrosis severity.1 In contrast to these studies (Table 1),1 in the current issue of Hepatology, Caviglia et al.6 report that miR‐21 is not crucial for the activation of hepatic stellate cells (HSCs) or the development of liver fibrosis. To reach this conclusion, the investigators performed rigorous examinations with genetic and pharmacologic manipulation of miR‐21 in multiple models of liver fibrosis. For example, three models of fibrogenesis (carbon tetrachloride [CCl4] injection, multidrug‐resistant 2 knockout [KO] mice, and 3,5‐diethoxycarbonyl‐1,4‐dihydrocollidine diet) were applied to miR‐21 null (miR‐21 KO) and wild‐type mice. None of these models demonstrated significant differences in liver fibrosis and/or HSC activation among these mice as assessed by sirius red staining and hepatic hydroxyproline content as well as markers of HSC activation, including α‐smooth muscle actin protein, Acta2, ColIa1, Lox, and Timp1 mRNA. Furthermore, to address a concern regarding the possible compensatory changes resulting from the absence of miR‐21 throughout development and adulthood, they acutely suppressed miR‐21 immediately before and after the onset of hepatic fibrosis with antisense DNA oligonucleotides (miR‐21 knockdown [KD] antimirs) in four fibrotic models (the previous three models plus bile duct ligation [BDL]). Furthermore, they examined CCl4‐induced fibrosis in a different line of miR‐21 KO mice. All of these in vivo experiments and other studies including in vitro experiments and experiments using human HSC cells and liver slices provided consistent results that led to their conclusion. Table 1 - In Vivo Preclinical Studies Showing miR‐21 Profibrotic Effects Authors In Vivo In Vitro miR‐21 Target Genes Zhang et al.1 Antagomir‐21 in CCl4 model (4 μL/g body weight in 1:3 CCl4/olive oil mixture, 8 doses, twice a week for 4 weeks, intraperitoneally), TAA model (200 mg/kg body weight, 16 doses, twice a week for 8 weeks, intraperitoneally) α‐SMA and Col1a1 mRNA expression were down‐regulated in isolated primary HSCs after miR‐21 knockdown Smad6/7 Wu et al.2 Adenovirus transfected miR‐21 inhibition in CCl4 model (2 μL/g body weight in 2:3 CCl4/olive oil mixture, 24 doses, twice a week for 12 weeks, subcutaneously) Hepatocyte apoptosis (no significant differences) SPRY2 Zhang et al.3 Chemically modified antisense oligonucleotides specific to miR‐21 in mice with hepatocyte‐specific deletion of Pten Liver fibrosis‐related in vitro assays were not performed Spry1/2 Rodrigues et al.4 MCD diet (2 and 8 weeks) to miR‐21 global KO mice Liver fibrosis‐related in vitro assays were not performed PPARα Kennedy et al.5 miR‐21 global KO mice with BDL (1 week) Cholangiocyte proliferation, HSC proliferation/fibrotic reaction (miR‐21 KD) Smad7 Abbreviations: BDL, bile duct ligation; CCl4, carbon tetrachloride; KD, knockdown; KO, knockout; MCD, methionine‐ and choline‐deficient; mRNA, messenger RNA; TAA, thioacetamide. Given their finding on miR‐21, the authors proceeded to examine the importance of miRNAs in HSC activation and liver fibrosis. For this purpose, they generated mice with HSC‐specific deletion of Dicer1. Dicer1 is a key RNAse that generates mature miRNAs. Although deletion of Dicer1 decreased miRNAs including miR‐21, miR‐199a‐3p, and let‐7i in HSCs by at least 50%, HSCs exhibited a normal phenotype, and the effect of the knockout on HSC activation and liver fibrosis was minimal. Furthermore, given that miR‐21 is known as an “oncomir,” the authors investigated the contribution of miR‐21 to liver cancer with miR‐21 KO and KD approaches, multiple liver cancer models, and human cancer cell lines. Although the experiments performed were not as rigorous as those studying liver fibrosis, miR‐21 was shown to be dispensable for hepatic carcinogenesis. What would account for these discrepancies among this study and others? A simple explanation points to differences in experimental design. For example, two previous studies that described a profibrotic effect of miR‐21 on the liver also used CCl4 injection to induce fibrosis,1 but their treatments were much more severe than the one used in Caviglia's study. Whereas Caviglia et al. administered CCl4 by way of oral gavage or intraperitoneally every 3 days for a total of eight doses, with a dose of 0.5 μL/g body weight in a 1:3 ratio of CCl4 and corn oil, one prior study administered a total of 24 subcutaneous injections twice a week for 12 weeks, with a dose of 2 μL/g body weight in a 2:3 CCl4‐to‐olive oil ratio, and the other administered injections eight times intraperitoneally twice a week for 4 weeks, with a dose of 4 μL/g body weight in a 1:3 CCl4‐to‐olive oil ratio. Accordingly, HSC activation and liver fibrosis may be milder in Caviglia et al.'s study than these two preceding studies (e.g., 5% fibrosis by sirius red–positive areas in Caviglia et al. vs. 16% fibrosis in Wu et al.2), possibly suggesting stage‐dependent differential effects of miR‐21 on liver fibrosis. Perhaps miR‐21 does not play a role in early stage fibrosis but may gain importance with fibrosis progression. Therefore, it is interesting to know whether Caviglia et al. would come to the same conclusion if they apply antisense DNA oligonucleotides to more severe cases of liver fibrosis. Different methods used to knock down miR‐21 may have also affected outcomes. Whereas Caviglia et al. used inhibitory antisense DNA oligonucleotides, the two prior studies employed adenoviruses and antagomirs, respectively.1 Experimental environments may also account for some differences. It is reported that the intestinal microbiome influences the development of liver fibrosis7 and that different experimental environments generate different populations of the intestinal microbiome in mice,8 which may affect experimental outcomes in relation to fibrogenesis. These conflicting results, however, may reflect a more fundamental question regarding miRNA regulation. In general, individual miRNAs target many different mRNAs, while individual mRNAs are regulated by many different miRNAs. Thus, there is likely to be functional redundancy of many miRNAs. In many cases, therefore, the effect of a single miRNA on targeted mRNAs may not be decisively crucial. In addition, as Ebert and Sharp (as well as Caviglia et al.) have suggested,9 the function of miRNAs may be mainly to contribute to robustness in biological processes, rather than to control biological processes as primary regulators. Alberti et al.10 also pointed out the moderate regulatory effects of miRNAs on gene expression, given that deletion of a wide range of specific miRNAs seldom causes major defects in the overall organism. They classified miRNAs into two categories: “core miRNAs” and “periphery miRNAs.” The core miRNAs are involved in the early stage of cellular differentiation, and their loss can cause lethal effects on the whole organism. The periphery miRNAs are implicated in the late stage of development with their comparatively subtle effects on the whole organism, and account for approximately 75% of miRNAs. Consistent with this concept, the antifibrotic effects of miR‐21 blockade observed in several studies were mild and were noted to be within a 2‐fold change compared with controls.3 Conflicting results regarding miR‐21 regulation have also been reported in mouse models of cardiac fibrosis, with one demonstrating a profibrotic effect and another showing no effects. Most indices used for assessment of this profibrotic effect were also within a 2‐fold change. Furthermore, the effects of other miRNAs implicated in liver fibrosis were often moderate. For example, miR‐29 was shown to be antifibrotic in mice, and administration of miR‐29a resulted in reduced liver fibrosis. However, most of the scores used to assess the improvement in liver fibrosis were within a 2‐fold change. These observations, including the ones presented by Caviglia et al., may support the notion of moderate biological regulation by miRNAs. The findings of Caviglia et al. strongly support the conclusion that miR‐21 is not a significant regulator of hepatic fibrogenesis under the conditions of their study. As they also note, however, their study does not completely exclude the role of miR‐21 in HSC activation and the development of liver fibrosis. Nevertheless, this is a timely and important study that presents an opportunity to consider the nature of miRNA regulation.