Abstract

Background & Aims

   The paradox of hepatic insulin resistance describes the inability for
   liver to respond to bioenergetics hormones in suppressing
   gluconeogenesis whilst maintaining lipid synthesis. Here, we report the
   deficiency of miR-192-3p in the livers of mice with diabetes and its
   role in alleviating hepatic steatosis.

Methods

   As conventional pre-microRNA (miRNA) stem-loop overexpression only
   boosts guiding strand (i.e. miR-192-5p) expression, we adopted an
   artificial AAV(DJ)-directed, RNA Pol III promoter-driven miRNA hairpin
   construct for star-strand-specific overexpression in the liver. Liver
   steatosis and insulin resistance markers were evaluated in primary
   hepatocytes, mice with diabetes, and mice with excessive carbohydrate
   consumption.

Results

   Functional loss of miR-192-3p in liver exacerbated hepatic
   micro-vesicular steatosis and insulin resistance in either mice with
   diabetes or wild-type mice with excessive fructose consumption.
   Liver-specific overexpression of miR-192-3p effectively halted hepatic
   steatosis and ameliorated insulin resistance in these mice models.
   Likewise, hepatocytes overexpressing miR-192-3p exhibited improved
   lipid accumulation, accompanied with decreases in lipogenesis and
   lipid-accumulation-related transcripts. Mechanistically, glucocorticoid
   receptor (GCR, also known as nuclear receptor subfamily 3, group C,
   member 1 [NR3C1]) was demonstrated to be negatively regulated by
   miR-192-3p. The effect of miR-192-3p on mitigating micro-vesicular
   steatosis was ablated by the reactivation of NR3C1.

Conclusions

   The star strand miR-192-3p was an undermined glycerolipid regulator
   involved in controlling fat accumulation and insulin sensitivity in
   liver through blockade of hepatic GCR signalling; this miRNA may serve
   as a potential therapeutic option for the common co-mobility of
   diabetic mellitus and fatty liver disease.

Lay summary

   The potential regulatory activity of star strand microRNA (miRNA)
   species has been substantially underestimated. In this study, we
   investigate the role and mechanism of an overlooked star strand miRNA
   (miR-192-3p) in regulating hepatic steatosis and insulin signalling in
   the livers of mice with diabetes and mice under excessive carbohydrate
   consumption.

   Keywords: Hepatic steatosis, Diabetes mellitus, MicroRNA,
   Glucocorticoid receptor, Transcription repressor, High carbohydrate
   consumption

   Abbreviations: AAV, adeno-associated virus; CPT, carnitine palmitoyl
   transferase; DEG, differentially expressed gene; DEX, dexamethasone;
   DM, diabetes mellitus; DNL, de novo lipogenesis; FA, fatty acid; FAO,
   fatty acid oxidation; FASN, fatty acid synthase; GCR, glucocorticoid
   receptor; HFD, high-fat diet; HFrD, high-fructose drink; HOMA-IR,
   homeostatic model assessment of insulin resistance; miRNA, microRNA;
   NAFLD, non-alcoholic fatty liver disease; NR3C1, nuclear receptor
   subfamily 3, group C, member 1; NT, non-targeting; OA, oleic acid;
   OGTT, oral glucose tolerance test; SCD1, stearoyl-CoA desaturase-1;
   shRNA, short hairpin RNA; TAG, triacylglyceride/triglyceride; T2DM,
   type 2 diabetes mellitus; VAT, visceral adipose tissue; 3′-UTR,
   3′-untranslated region

Graphical abstract

   graphic file with name fx1.jpg
   [41]Open in a new tab

Highlights

     * •
       Liver-specific knockdown of miR-192-3p recapitulated functional
       loss of the miRNA as in mice with diabetes.
     * •
       This knockdown was characterised by pronounced hepatic
       micro-vesicular steatosis coupled to insulin resistance.
     * •
       In vivo overexpression of miR-192-3p alleviated hepatic steatosis
       in mice with diabetes and wild-type mice with excessive fructose
       consumption.
     * •
       Glucocorticoid receptor (also known as NR3C1) was discovered as the
       immediate target of miR-192-3p in regulating hepatic lipid turnover
       and storage.

Introduction

   Liver, muscle, and adipose tissues are 3 major peripheral effector
   tissues in response to insulin—the essential bioenergetics endocrine
   that governs the entry and storage of glucose in normal starved-fed
   cycles.[42]^1^,[43]^2 Unlike the other 2 insulin-responsive tissues,
   liver is a central hub for nutrient homeostasis, which includes
   synthesising glycogen from glucose after a meal or effectively
   converting excessive glucose to fatty acids (FAs) as precursor for
   triglyceride (TAG) synthesis via de novo lipogenesis (DNL).[44]^3 Liver
   is therefore critical for its unique niche in energy homeostasis and
   the de facto effector in balancing blood glucose levels.

   Liver steatosis is a sign of lipid overload in the organ either when
   there is an oversupply of lipid or when there is a failure in lipid
   breakdown, frequently associated with impaired insulin
   signalling.[45]^4 Lipid overload manifests in a blend of 2 major forms
   (i.e. macro- and micro-vesicular steatoses).[46]^5 The former relates
   to a benign, largely asymptomatic form of steatotic liver. In contrast,
   the latter is adversely associated with metabolic disorder,
   inflammation, and hepatic fibrosis.[47]^6 Intriguingly, there is an
   overt linkage between non-alcoholic fatty liver disease (NAFLD) and
   diabetes, in which NAFLD is prevalent in up to 70% of patients with
   diabetes,[48]^7 and the reported prevalence of non-alcoholic
   steatohepatitis in patients with type 2 diabetes mellitus (T2DM) is up
   to 20%.[49]^8

   Even with the clear close ties, many morbidly obese individuals remain
   non-diabetic. In line with this clinical observation, leptin-deficient
   obese (ob/ob) mice usually exhibited a benign hepatic steatosis,
   whereas leptin-receptor-defective (db/db) mice with diabetes commonly
   developed inflammation upon consumption of high carbohydrates.[50]^9 It
   is concretely confirmed that ob/ob mice developed macro-vesicular,
   while that of db/db mice is predominantly micro-vesicular steatosis
   instead. Indeed, patients with diabetes manifest hepatic
   micro-vesicular steatosis with exacerbated neurovascular
   complications.[51]^10

   MicroRNAs (miRNAs) are naturally occurring non-coding RNAs that bind to
   the 3′-untranslated regions (3′-UTR) of target mRNAs, leading to either
   their degradation or translational repression.[52]^11 Given their
   critical roles exemplified in metabolism control, miRNAs are frequently
   studied as a target for disease interventions (i.e. human
   cancers[53]^11 and metabolic diseases[54]^12). Of note, some miRNAs are
   regulators of liver metabolism (e.g. miR-206, which inhibits lipid and
   glucose production by facilitating insulin signalling[55]^13). In
   addition, miR-192 and miR-193b were presumed as proxy diabetes
   biomarkers, where their abundance was found significantly increased in
   glucose-intolerant mice and in people with prediabetes.[56]^14 Despite
   the circulating levels of miR-192 (5p, the guiding strand) were refuted
   both as a reliable marker for liver damage[57]^15 and diabetic
   nephropathy,[58]^16 its linkage to diabetic liver condition remains
   unclear.

   In the current study, we found that mice with diabetes with apparent
   micro-vesicular nature exhibited a substantial loss of the star strand
   of miR-192 (i.e. miR-192-3p). Overexpression (OE) of the sponge against
   the star strand mimicked the de facto functional loss of
   miR-192-3p-aggravated liver steatosis of micro-vesicular nature in
   wild-type mice given excessive fructose consumption, recapitulating
   conditions of impaired insulin response. Alternatively, we adopted an
   unconventional short hairpin RNA (shRNA) stem-loop construct for the
   specific OE of the star strand miRNA alone. Indeed, liver-specific OE
   of miR-192-3p significantly alleviated hepatic steatosis and improved
   insulin resistance in both mice with diabetes and those given excessive
   carbohydrates. Finally, our study pinpointed nuclear receptor subfamily
   3, group C, member 1 (NR3C1), the glucocorticoid receptor (GCR), as a
   direct target of miR-192-3p to alleviate liver steatosis and insulin
   resistance.

Materials and methods

   Full details of methods can be found in the [59]supplementary
   information.

Human serum sample collection

   Human serum samples were obtained from the Prince of Wales Hospital,
   Hong Kong Special Administrative Region, China. Details of patients are
   listed in [60]Table S1. Studies using human serum were approved by the
   joint Chinese University of Hong Kong/New Territories East Cluster
   Clinical Research Ethics Committee. Informed consent was obtained from
   all participants.

Animal, treatment, and sample collection

   All animals were supplied by the Laboratory Animal Service Centre at
   the Chinese University of Hong Kong. The animals were housed with
   controlled temperature (25 ± 1°C), a 12-h light–dark cycle, and 55 ± 5%
   humidity. All procedures were approved by the Animal Experimentation
   Ethics Committee, Chinese University of Hong Kong under the regulations
   of the Hong Kong Special Administrative Region government.

   A mice fatty liver disease model was established by feeding male mice
   either with a high-fat diet (HFD; rodent diet with 45 kcal% fat;
   D12451; Research Diets, New Brunswick, NJ, USA) for 8∼12 weeks starting
   at the age of 4 weeks, or high-fructose drink (HFrD; water supplemented
   with fructose 30%) for 8∼12 weeks starting at the age of 4 weeks,
   whereas mice in the control group were maintained on a normal chow diet
   (2018SX; ENVIGO, USA). Food and water were provided ad libitum. The
   mice were tail vein injected with adeno-associated virus (AAV) of DJ
   serotype (AAV DJ) expressing miR-192-3p sponge or miR-192-3p OE for
   different periods, and then serum and different tissues were collected
   for further analysis.

Liver-specific AAV delivery of transgenes and in vivo imaging

   To specifically deliver either miR-192-3p or its sponge into mice
   livers, AAV DJ carrying transgenes of either constructs was tail vein
   injected, and repeated injections were performed once every 3 weeks.
   Alternatively, shNR3C1 construct was subcloned into pSicoR-mCherry
   vector (Addgene, USA; #31845). Lentivirus carrying pSicoR-shNR3C1 was
   i.v. injected into 4-week-old mice with diabetes, followed by
   AAV-pgk-Cre (Addgene; #24593) 2 weeks later. These mice were then
   sacrificed after 2 months. Red fluorescent signal intensity was
   detected by the In-Vivo Xtreme Imaging System (Bruker, USA).

Statistical analysis

   Statistical analysis was performed using GraphPad Prism Software®. Data
   derived from cell line or mouse experiments were presented as mean ±
   standard deviation and assessed by a 2-tailed Student's t test or
   one-way analysis of variance test. A p value <0.05 was considered
   statistically significant.

Results

Functional loss of star strand miR-192-3p in the liver manifested as hepatic
micro-vesicular steatosis

   Each pre-miRNA stem loop can produce up to 2 mature miRNAs on either
   strand with relative expression biases in different tissues. Our
   previous work had demonstrated that both mature miR-192 guiding strand
   (miR-192-5p) and its star strand miR-192-3p (miR-192∗) ([61]Fig. 1A)
   were enriched as top candidate differentiation regulators along the
   hepatic lineage.[62]^17 Interestingly, we found that miR-192-3p rather
   than miR-192-5p was lost in the liver of db/db mice with diabetes when
   compared with heterozygous lean control ([63]Fig. 1A). The deregulated
   expression of star strand miRNA instead of the guiding strand in liver
   pinpointed the necessity to showcase their relative expression in
   different organs of mice with diabetes.

Fig. 1.

   [64]Fig. 1
   [65]Open in a new tab

   Hepatic functional loss of miR-192-3p resulted in hepatic
   micro-vesicular steatosis.

   Wild-type C57BL/6 mice were fed on ND or HFrD for 8 weeks. (A)
   Expression of miR-192 and miR-192-3p. (B) Expression ratio of
   miR-192-3p to miR-192-5p in different tissues of db/db mice. (C–E)
   Liver histology showed liver-specific loss of miR-192-3p resulted in
   lipid accumulation and micro-vesicular steatosis in HFrD-fed mice. (F)
   Hepatic triglyceride levels in the indicated groups. (G–I) Hepatic
   knockdown of miR-192-3p led to increased insulin resistance, as
   evidenced by OGTT pattern and HOMA-IR index. Data are represented as
   mean ± SD; n = 5∼8. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. Two-tailed
   Student's t test. H&E, haematoxylin and eosin; HFrD, high-fructose
   drink; HOMA-IR, homeostasis model assessment of insulin resistance; ND,
   normal diet; OGTT, oral glucose tolerance test.

   It was noted that the relative expression bias between the strands
   (ratio of miR-192-3p/miR-192-5p) is within 2-fold. However, there was a
   noticeable decrease in ratio for both liver and kidney, while the ratio
   was increased in mesenteric visceral adipose tissues (VATs)
   ([66]Fig. 1B). Such relative expression biases in liver, kidney, and
   VATs were in line with previous reports, which discussed the potential
   functions in either of these strands.[67]^15^,[68]^18^,[69]^19 In
   addition, the expression of miR-192-3p, albeit much less than its
   guiding counterpart, remained significantly higher in the liver than in
   other organs ([70]Fig. S1A and B).

   Given its described suppressive effect in lipid storage in adipose
   tissues, we therefore proposed that miR-192-3p loss is related to liver
   steatosis, a frequent co-morbidity of diabetes. To assay for the
   effects of functional loss in hepatic miR-192-3p, we transduced
   miR-192-3p sponge (a construct that overexpressed 3 times perfect
   complementary sponge repeats linked to enhanced green fluorescent
   protein) against miR-192-3p by AAV of liver targeting DJ serotype (AAV
   DJ) in vivo ([71]Fig. S1C and D). As a precaution, we checked that such
   liver-specific functional loss of miR-192-3p did not exert visible
   changes in adiposity of liver tissues nor notable changes in oral
   glucose tolerance test (OGTT) in wild-type C57/BL6 mice given normal
   diet ([72]Fig. 1F–H and [73]S1E).

   In contrast, wild-type mice given either HFrD (drinking water
   supplemented with fructose 30%) or HFD (high-calorie diet) demonstrated
   distinctive hepatic lipid content and OGTT patterns. Hepatic lipid
   appeared as patchy droplets in those overexpressed with vector control,
   while animals given miR-192-3p sponge exhibited foamy cytoplasm with
   minute lipid vesicles ([74]Fig. 1C–E). Knockdown of miR-192-3p also
   promoted hepatic lipid storage, as evidenced by the increase of TAG
   levels ([75]Fig. 1F). The homeostatic model assessment of insulin
   resistance (HOMA-IR) was assayed as the surrogate estimate for
   peripheral insulin resistance. Functional loss of miR-192-3p
   demonstrated reduced glucose tolerance, as indicated by the increased
   HOMA-IR upon miR-192-3p sponge OE ([76]Fig. 1G). OGTT profiles also
   showed that depletion of miR-192-3p increased the development of
   glucose intolerance and compensatory hyper-insulinaemia ([77]Fig. 1H
   and I).

   The effect of miR-192-3p loss in mice given HFD was also examined.
   Despite phenotype not as severe as those given fructose drink,
   miR-192-3p sponge also aggravated HFD-induced lipid accumulation and
   insulin resistance ([78]Fig. S1F–I). Taken together, functional loss of
   miR-192-3p promoted hepatic micro-vesicular steatosis—a detrimental
   liver lipid overload as that in patients with diabetes, with an
   aggravated insulin resistance in our mice models.

Overexpressed miR-192-3p altered lipid homeostasis in hepatocytes

   As miR-192-3p was implicated in fatty tissue differentiation, loss of
   this miRNA may impact the role of liver as a key hub for lipid
   metabolism. We screened the expression of some predicted targets of
   miR-192-3p in the mice with diabetes, and observed a remarkably higher
   gene expression of Scd and Vldlr, and a modest increase in Lpin1 in
   their liver ([79]Fig. S2A). Concurrently, FA translocase Cd36/Scarb3
   and DNL enzyme Fasn expression were also found higher ([80]Fig. S2A).

   Subsequently, hepatic normal human cell line L02 was transfected with
   this miR-192-3p sponge and stably selected by puromycin to delineate
   the underlying mechanism of the action of miR-192-3p. Two separate
   RNA-Seq libraries of non-targeting (NT) and miR-192-3p sponge (S) were
   constructed. Of the 1,562 genes differentially expressed, more than 400
   genes were predicted to be direct miR-192-3p targets or their redundant
   genes deduced from TargetScan ([81]Fig. 2A). PI3K-Akt signalling, FoxO
   signalling, adipocytokine signalling, and insulin-resistance-associated
   pathways were enriched by the Kyoto Encyclopedia of Genes and Genomes
   catalogue, reinstating the involvement of miR-192-3p in bioenergetics
   homeostasis ([82]Fig. 2B and C). Gene set enrichment analysis further
   identified processes associated with lipid, FA, steroid, and
   monocarboxylic acid metabolism affected by miR-192-3p functional
   depletion ([83]Fig. 2D and [84]S2B). In addition, lipogenesis- and
   gluconeogenesis-associated genes were also significantly enriched in
   the analysis ([85]Fig. 2E).

Fig. 2.

   [86]Fig. 2
   [87]Open in a new tab

   miR-192-3p altered metabolic homeostasis in hepatocytes.

   (A) DEGs were overlapped with predicted miR-192-3p targets collected
   from the TargetScan database in the Venn diagram. The scatter plots
   indicating relative gene expression profile between groups as
   indicated, with log[2] (fold change) less than –0.585 and larger than
   0.585. (B) Clustering images showing global sample expression profiles
   and relationships analysed by hierarchical clustering. (C–D) KEGG
   pathway enrichment analysis and GSEA of DEGs identified
   lipid-metabolism-associated processes. (E) Volcano plots and expression
   profile of lipogenesis- and gluconeogenesis-associated genes in the S
   and NT groups. (F) Artificial hairpin structure of miR-192-3p
   (miR-192-3p OE) specifically increased the expression of miR-192-3p in
   HEK293T without altering miR-192 expression. (G) The mRNA expression
   levels of lipid-metabolism-associated genes in mouse primary
   hepatocytes transfected with miR-192-3p OE. Data are represented as
   mean ± SD; n = 3. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. Two-tailed
   Student's t test. DEG, differentially expressed gene; GSEA, gene set
   enrichment analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; NT,
   non-targeting control; OE, overexpression; S, miR-192-3p sponge.

   To validate the role of miR-192-3p in liver steatosis, we designed an
   artificial construct resembling shRNAs,[88]^20 with miR-192-3p sequence
   in the anti-sense strand and scrambly manipulated sense strand. This
   miR-192 stem loop was then transiently transfected into HEK293T cells
   and lead to a significant 30-fold upregulation of miR-192-3p, without
   altering the expression of the guiding strand miR-192 ([89]Fig. 2F).

   Furthermore, we transfected miR-192-3p mimics into primary mouse
   hepatocytes and studied genes responsible for FA mobilisation. As
   expected, genes, such as Srebf1, Fasn, Acc, Cd36, and Ffar4, were
   decreased by the mimics ([90]Fig. 2G), suggesting that miR-192-3p might
   have altered the mobilisation of FAs in hepatocytes.

miR-192-3p inhibited lipid accumulation in human hepatic cell line

   Among the major bioenergetic pathways enriched from the panel of
   differentially expressed genes (DEGs) mentioned previously were
   mitochondrial aerobic oxidation and pyruvate metabolism, both of which
   are closely related to FA metabolism. Therefore, we validated the
   action of miR-192-3p on FA oxidation (FAO) by the Seahorse XF Cell Mito
   Stress Test. Stable cell clones overexpressing miR-192-3p (miR-192-3p
   OE) tended to consume more palmitate-bovine serum albumin (PA-BSA) as
   substrate for FAO when compared with an NT cell line ([91]Fig. 3A). The
   increase in oxygen consumption rate in miR-192-3p OE cells suggested
   significantly induced oxidative metabolism, indicating the dissipation
   of lipids by maximising FAO ([92]Fig. 3B and C). This inductive effect
   was further blocked by carnitine palmitoyl transferase (CPT) inhibitor
   etomoxir treatment ([93]Fig. 3B and C).

Fig. 3.

   [94]Fig. 3
   [95]Open in a new tab

   miR-192-3p negatively regulated lipid accumulation in hepatocytes.

   L02 cells were stably overexpressed with miR-192-3p (miR-192-3p OE) or
   NT. (A) Intracellular lipid content in NT and miR-192-3p OE cells using
   PA-BSA as substrate. (B) Fatty acid oxidation by the Seahorse XF Cell
   Mito Stress Test with or without CPT inhibitor etomoxir in NT or
   miR-192-3p OE cells. (C) Quantification of the AUC in each indicated
   group. (D) The mRNA expression of genes related to fatty acid oxidation
   in miR-192-3p OE and NT cells. (E) Representative images of Oil Red O
   and Nile Red staining of NT and miR-192-3p OE cells incubated with OA
   for 8 h. (F) Intracellular triglyceride levels of NT and miR-192-3p OE
   cells stimulated by OA for 8 h. (G–I) The relative mRNA and protein
   expression levels of the indicated molecules in cells of the indicated
   groups. GAPDH served as controls. Data are represented as mean ± SD.
   ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001; n = 3. Two-tailed Student's t test.
   CPT, carnitine palmitoyl transferase; NT, non-targeting control; OA,
   oleic acid; OE, overexpression; PA-BSA, palmitate-bovine serum albumin.

   In addition, miR-192-3p increased the expression of FAO genes encoding
   CPT1 alpha, long-chain acyl-CoA dehydrogenase, acyl-coenzyme A oxidase
   1, hydroxyacyl-CoA dehydrogenase-alpha, and
   peroxisome-proliferator-activated receptor alpha in cells treated with
   PA-BSA ([96]Fig. 3D). With reports showing close ties between
   malfunctioned beta-oxidation to micro-vesicular steatosis,[97]^21 it
   suggested that miR-192-3p OE shall restore fatty liver in a relatively
   benign form.

   We then examined the role of miR-192-3p in preventing lipid
   accumulation. OE of miR-192-3p alleviated oleic acid (OA)-induced lipid
   accumulation in L02 cells ([98]Fig. 3E). Fluorescent staining of
   neutral lipids showed that lipid spot area per cell was significantly
   reduced in miR-192-3p OE cells incubated with OA ([99]Fig. 3E), with an
   obviously reduced intracellular TAG level ([100]Fig. 3F). In addition,
   overexpressing miR-192-3p also suppressed both mRNA and protein levels
   of lipogenesis genes, including FA synthase (FASN), CD36, stearoyl-CoA
   desaturase-1 (SCD1), and SREBP1 ([101]Fig. 3G–I).

Liver-specific expression of miR-192-3p exhibited robust therapeutic effects
on liver steatosis in mice

   To assess the functional contribution of miR-192-3p to the
   hepatosteatosis and hyperglycaemia in mice, a liver-specific miR-192-3p
   expression set-up was generated using AAV DJ. Injection of
   AAV-miR-192-3p into mice led to high levels of miR-192-3p in liver
   other than other tissues ([102]Fig. S3A and B). Early adenoviral
   injection of miR-192-3p to 4-week-old mice with diabetes effectively
   prevented body weight gain ([103]Fig. 4A and B) and reduced the daily
   food intake ([104]Fig. S3C–E) when compared with NT control sacrificed
   2 months later. Such liver-specific OE of miR-192-3p also attenuated
   hepatic lipid accumulation and hyperlipidaemia ([105]Fig. 4C), which
   was further confirmed by the reduction of TAG level in mice treated
   with miR-192-3p ([106]Fig. 4D). OGTT pattern and HOMA-IR indices were
   also improved by the OE of miR-192-3p ([107]Fig. 4E and F).
   Consistently, miR-192-3p significantly suppressed the mRNA and protein
   levels of genes associated with the synthesis and uptake of FA,
   alongside induction in those genes responsible for FAO ([108]Fig. 4G–I
   and [109]S3F).

Fig. 4.

   [110]Fig. 4
   [111]Open in a new tab

   Delivery of miR-192-3p into livers alleviated hepatic steatosis in
   mice.

   Mice with diabetes were injected i.v. with AAV-DJ-miR-192-3p
   (miR-192-3p OE) or the NT and fed on ND for 8 weeks. C57/BL6 mice were
   tail i.v. injected with AAV-miR-192-3p OE or NT, and given HFrD for
   additional 12 weeks. (A–B) Impaired liver and body weight gain of mice
   with diabetes with miR-192-3p OE. (C) H&E and Oil Red O staining of
   liver sections showed miR-192-3p OE alleviated hepatic lipid
   accumulation and steatosis. (D) Hepatic triglyceride levels in the
   indicated groups. (E–F) OGTT pattern and HOMA-IR index of mice in the
   indicated groups were profiled by 8 weeks. (G–I) The mRNA and protein
   levels of lipogenesis-related genes were measured in the mice livers of
   the 2 groups. (J–K) Liver and body weight were impaired in HFrD-fed
   mice after miR-192-3p administration. (L) Histological staining, (M)
   hepatic triglyceride level, (N) OGTT pattern, and (O) HOMA-IR index
   were assayed in mice of miR-192-3p OE and NT groups fed on HFrD. (P)
   The mRNA levels of lipogenesis-related genes were measured in the mice
   livers of the indicated groups. Data are represented as mean ± SD; n =
   5–6. Two-tailed Student's t test and one-way analysis of variance were
   used for statistical analyses. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. H&E,
   haematoxylin and eosin; HFrD, high-fructose drink; HOMA-IR, homeostasis
   model assessment of insulin resistance; ND, normal diet; NT,
   non-targeting control; OE, overexpression; OGTT, oral glucose tolerance
   test.

   Delivery of miR-192-3p into HFrD-fed mice also resulted in significant
   body and liver weight loss ([112]Fig. 4J and K). Consistent with the
   ability of miR-192-3p in turning over lipid content, H&E and Oil Red O
   staining showed that liver-specific OE of miR-192-3p reduced hepatic
   lipid accumulation compared with the NT control ([113]Fig. 4L and M).
   In addition, hepatic miR-192-3p OE mitigated HFrD-induced glucose
   intolerance, as evidenced by the OGTT pattern and HOMA-IR index
   ([114]Fig. 4N and O). The mRNA expression level of lipogenesis-related
   genes was also found to be downregulated by the long-term treatment of
   miR-192-3p ([115]Fig. 4P).

miR-192-3p directly targeted GCR in regulating liver glycerolipid homeostasis

   To identify potential target genes of miR-192-3p, which may be
   responsible for its action in liver, we screened the DEGs according to
   3 criteria:
     * •
       genes must be conserved in the 3′-UTR along predicted domains;
     * •
       gene expression should be regulated following miR-192-3p change;
       and
     * •
       the selected genes are implicated in fatty liver diseases.

   Fifty-four predicted target genes were enriched from the adopted
   prediction databases, including TargetScan, miRDB, DIANA, and
   TargetMiner ([116]Fig. 5A), and then clustered with DEGs from previous
   RNA-Seq ([117]Fig. 5A). Of note, NR3C1, the gene encoding GCR, is of
   high interest, as GCR signalling was enriched as top candidate pathways
   by the Ingenuity Pathway Analysis ([118]Fig. S4).

Fig. 5.

   [119]Fig. 5
   [120]Open in a new tab

   NR3C1 was identified to be negatively regulated by miR-192-3p in liver
   steatosis.

   (A) Combinational analysis of 4 microRNA target prediction databases
   and the DEGs from the RNAseq experiments identified NR3C1 as top
   candidate regulated by miR-192-3p. (B) Complimentary sequences to the
   seed regions of miR-192-3p within the 3′-UTR mRNA of NR3C1 are
   conserved among different species. Luciferase reporter constructs
   containing 3′-UTR of human NR3C1 and miR-192-3p OE construct were
   co-transfected into HEK293T cells; luciferase activity was normalised
   to the activity of β-galactosidase. (C) Immunoblot analysis of NR3C1 in
   L02 and MIHA cells overexpressed with miR-192-3p. (D) Oil Red O and
   Nile Red staining of lipid in L02/miR-192-3p OE cells replenished with
   NR3C1 construct or not. (E) Hepatic triglyceride levels in the
   indicated groups. (F–H) Real-time PCR and Western blotting analysis of
   lipogenesis-related genes in L02/miR-192-3p OE cells transfected with
   NR3C1 OE construct or not. The cells were all incubated with OA for 8
   h. Both the mRNA and protein levels were normalised to GAPDH. Data are
   represented as mean ± SD; n = 3. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001, NT
   vs. miR-192-3p OE; ^#p <0.05, ^##p <0.01, ^###p <0.001, miR-192-3p OE
   vs. miR-192-3p OE + NR3C1. Student's t test and one-way analysis of
   variance with Tukey's t test were used for statistical analyses. DEG,
   differentially expressed gene; GAPDH, glyceraldehyde 3-phosphate
   dehydrogenase; NR3C1, nuclear receptor subfamily 3, group C, member 1;
   OE, overexpression; 3′-UTR, 3′-untranslated region.

   Luciferase report assay in HEK293T showed that miR-192-3p mimics
   significantly reduced luciferase activity of NR3C1 3′-UTR when compared
   with control mimics ([121]Fig. 5B). The mRNA and protein levels of
   NR3C1 were also downregulated in human hepatocytes overexpressed with
   miR-192-3p ([122]Fig. 5C), reinforcing that miR-192-3p is able to
   inhibit expression of NR3C1 by interacting with its 3′-UTR.

   As the aberrant glucocorticoid signalling is tightly related to
   metabolic disorders, including obesity, hyperglycaemia, and insulin
   resistance, we tested whether NR3C1 mediates the role of miR-192-3p in
   regulating liver steatosis. Oil Red O and Nile Red staining of newly
   synthesised lipid indicated that there is a reduction of lipid
   accumulation and TAG level under miR-192-3p OE in L02 cells when
   incubated with OA, which was blocked by forced NR3C1 expression
   ([123]Fig. 5D and E). Furthermore, such lentiviral OE disrupted the
   suppressive role of miR-192-3p on both mRNA and protein levels of
   lipogenesis genes, such as FASN, LIPIN1, SCD1, and SREBP1, suggesting
   that NR3C1 mediates the regulatory role of miR-192-3p in lipid
   homeostasis ([124]Fig. 5F–H).

miR-192-3p target GCR to regulate glycerolipid homeostasis in mice with
diabetes

   To investigate whether NR3C1 mediates the inhibitory effects of
   miR-192-3p on hepatic lipid accumulation, a lentiviral floxed NR3C1
   shRNA was reintroduced to 4-week-old mice with diabetes.[125]^22
   Lentiviral OE of the shRNA in pSicoR was eliminated by the
   AAV(DJ)-pgk-Cre. In vivo imager detected substantial fluorescent signal
   in the liver, lung, and fat, demonstrating an accumulation of
   lentiviral induction in the body driving the transgene. However, in the
   presence of liver-targeting Cre, there was a clear reduction of those
   signals specifically in the liver after Cre recombination
   ([126]Fig. 6A).

Fig. 6.

   [127]Fig. 6
   [128]Open in a new tab

   NR3C1 mediated the inhibitory effects of miR-192-3p on hepatic
   steatosis in mice.

   (A) Diagram of liver-specific knockdown of NR3C1 by Cre-LoxP system.
   Lentiviral of psicoR-shNR3C1 was i.v. injected into 4-week-old mice
   with diabetes; liver-specific reactivation of NR3C1 was achieved by
   i.v. injection of AAV-Cre. Fluorescence signalling was shown in
   different tissues. (B) Representative images of H&E, Oil Red O, and
   Nile Red staining of liver sections in the indicated groups; n = 4. (C)
   Histological analysis of liver sections in mice treated with NT,
   miR-192-3p OE, combinational use of miR-192-3p, and DEX 1 mg/kg/day,
   twice every week for 2 months. (D–F) Body weight gain, hepatic
   triglyceride levels, and HOMA-IR index was assayed in the 3 indicated
   groups. (G–I) Real-time PCR and Western blotting analysis of
   lipogenesis-associated genes in the 3 indicated groups. Both the mRNA
   and protein levels were normalised to GAPDH. Data are represented as
   mean ± SD; n = 5. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001, NT vs. miR-192-3p
   OE; ^#p <0.05, ^##p <0.01, ^###p <0.001, miR-192-3p OE vs. miR-192-3p
   OE + DEX. One-way analysis of variance with Tukey's t test. DEX,
   dexamethasone; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; H&E,
   haematoxylin and eosin; HOMA-IR, homeostasis model assessment of
   insulin resistance; NR3C1, nuclear receptor subfamily 3, group C,
   member 1; NT, non-targeting control; OE, overexpression.

   The constitutive knockdown of NR3C1 before Cre activation prevented fat
   liver accumulation in age-matched mice with diabetes compared with NT
   control. These mice phenocopied miR-192-3p OE by demonstrating
   alleviated liver steatosis ([129]Fig. 6B). Subsequently, mice
   transduced with the shNR3C1 construct were subjected to liver-specific
   Cre to eliminate gene repression caused by the ectopic shRNA
   ([130]Fig. S5A and B). The relief in NR3C1 repression had loosened such
   a layer of control, and therefore restored the hepatic steatotic
   phenotype manifested with glucose intolerance as in untreated,
   age-matched mice with diabetes ([131]Fig. 6B and [132]S5C–E),
   indicating that the interaction between miR-192-3p and NR3C1
   contributed to the pathogenesis of hepatosteatosis.

   To rescue NR3C1 activity in mice injected with AAV-miR-192-3p, mice
   were given additional intraperitoneal injection of dexamethasone (DEX),
   an agonist of NR3C1/GCR. The gene components of the NR3C1-regulated
   lipogenesis in liver of 3 groups (NT, miR-192-3p OE, and miR-192-3p
   OE + DEX) were analysed ([133]Fig. 6C–F and [134]Fig. S5F–H). The
   anti-lipogenesis role of miR-192-3p was partially ablated by DEX, as
   demonstrated by the recovery of hepatic lipid accumulation, TAG level,
   and insulin resistance ([135]Fig. 6E and F). Livers of mice treated
   with miR-192-3p OE had significantly reduced lipogenic gene expression
   while enhancing NR3C1 activity restored their suppressed levels
   ([136]Fig. 6G–I). Taken together, the results showed that miR-192-3p
   reduced lipogenesis by interacting with NR3C1 mRNA, which subsequently
   alleviated liver steatosis in mice with diabetes.

Serum miR-192-3p is detected in patients with diabetes mellitus

   To correlate circulating miR-192-3p level to clinical features, we
   quantified the serum miR-192-3p in obese individuals (body mass index
   >34; [137]Fig. 7A). Among these 8 individuals, only 4 were diagnosed
   with diabetes mellitus (DM). We then performed TaqMan™ miRNA real-time
   PCR assay to screen for the presence of the star strand miRNA in their
   serum samples. Intriguingly, miR-192-3p can be only detected in
   individual diagnosis with DM, indicating the predictive relevance of
   miR-192-3p with DM ([138]Fig. 7B). Accordingly, patients with DM
   exhibited increased levels of fasting glucose ([139]Fig. 7C; p <0.01)
   and haemoglobin A1c ([140]Fig. 7D; p <0.05), decreased HDL level
   ([141]Fig. 7E; p <0.01), and also significantly higher corticosterone
   level ([142]Fig. 7F; p <0.05). The serum TAG, total cholesterol, and
   LDL levels were also simultaneously examined ([143]Fig. 7G–I).

Fig. 7.

   [144]Fig. 7
   [145]Open in a new tab

   Serum miR-192-3p is detectable only in patients diagnosed with diabetes
   mellitus.

   (A) All human samples exhibited obesity with BMI exceeding normal
   range. (B) Serum miR-192-3p levels in patients diagnosed with DM or not
   (N-DM). (C) Fasting glucose, (D) HbA1c, (E) HDL, (F) corticosterone,
   (G) triglyceride, (H) total cholesterol, and (I) LDL in each group.
   Data are represented as mean ± SD; n = 4. Two-tailed Student's t test.
   BMI, body mass index; DM, diabetes mellitus; HbA1c, haemoglobin A1c;
   N-DM, no diabetes mellitus.

Discussion

   The human body relies on glucose as an immediate energy source to deal
   with strenuous movements. Being metabolised less efficiently, FAs are
   the major energy reserve with the prerequisite of a series of oxidation
   for its release.[146]^23 Liver, muscle, and adipose tissues are the
   major effector tissues in response to the bioenergetics endocrine via
   modulating the entry and storage of glucose. Of note, muscle cells are
   privileged in a lack of glucose-6-phosphatase, therefore refrained from
   glycogen breakdown to supply glucose to other somatic cells when
   insulin is low.[147]^24 Adipose tissues respond to insulin by
   synthesising TAG for compact storage. Liver is therefore the only
   metabolic hub, which modulates circulating levels of nutrient supply to
   other tissues. Furthermore, liver is also the major site for
   cholesterol synthesis in response to metabolic demand tightly regulated
   by insulin and glucose at the transcription level,[148]^25 highlighting
   how defective insulin signalling may affect lipid synthesis and
   storage.

   NAFLD is recognised as the predominant form of chronic liver disease
   associated with metabolic disorders, such as T2DM, obesity, and
   hyperlipidaemia.[149]^26 NAFLD was caused by the inhability of insulin
   to suppress hepatic glucose production in obese people with diabetes,
   with NAFLD further promoting diabetes; paradoxically, fat
   overproduction by this runaway hepatic TAG synthesis in the liver
   continues unabated. There is ample evidence to demonstrate impairment
   in hepatic carbohydrate utilisation leading to lipid
   dysregulation.[150]^27 Recent efforts have focused on modulating genes
   along the lipogenic pathway in the liver, which demonstrated marked
   reduction in hepatic steatosis and even confer resistance to
   obesity.[151]^28^,[152]^29

   Our previous study pinpointed that both miR-192-5p and its star strand
   miR-192-3p (∗) were enriched as top regulators along hepatic lineage
   differentiation.[153]^17 The guiding strand was important in inhibiting
   transforming growth factor beta signalling in diabetic
   nephropathy.[154]^18 In our study, only miR-192-3p rather than its
   guiding strand miR-192-5p was downregulated in the liver of mice with
   diabetes. This spontaneous loss of miR-192-3p in mice with diabetes
   further facilitated the runaway hepatic TAG synthesis and promoted
   hepatic steatosis.

   Micro-vesicular steatosis is characterised by functionally defective
   hepatocytes filled up with foamy cytoplasm and small lipid vesicles,
   while macro-vesicular steatosis is manifested with a single, large
   vacuole of lipid that occupies most of the volume in hepatocytes
   coupled with a displaced nucleus, a detrimental form preceding tissue
   inflammation.[155]^6^,[156]^30 Interestingly, our attempts in
   replenishing miR-192-3p in mice with diabetes not only partially
   reduced lipid accumulation, but also effectively mediated the storage
   condition for lipids from the micro-vesicular to the macro-vesicular
   form. Accordingly, we observed an increase of serum tumour necrosis
   factor alpha in HFD-fed mice treated with miR-192-3p sponge,
   implicating the critical role of miR-192-3p along the pathogenesis of
   liver inflammation (data not shown).

   Conventional miRNA OE relies pre-miRNA stem loop; therefore OE only
   produces mature guiding strand. Nonetheless, we believed their strand
   counterparts are useful in targeting an alternative myriad of
   mRNAs.[157][31], [158][32], [159][33] To single out star-strand
   activation, we adopted the shRNA OE system and successfully
   overexpressed mature miR-192-3p in vivo. Further functional study found
   that OE of this miR-192-3p effectively prevented hepatic steatosis
   through restoring insulin sensitivity, which facilitated FAO. To
   further investigate whether this new OE system is applicable for any
   mature miRNA strand, we constructed the miR-206, a guiding strand miRNA
   responsible for lipogenesis,[160]^13 which recapitulated its
   anti-lipogenesis and anti-gluconeogenesis role in hepatocytes
   ([161]Fig. S6A–C).

   Glucocorticoid signalling was central to modulate the activation of
   AMPK and represented an important crosstalk to liver responsiveness to
   insulin.[162]^34 The GCR (also known as NR3C1) was reported to control
   the metabolism of glucose, protein, and lipid homeostasis,
   complementing the action of insulin.[163]^35 NR3C1 (Ad-GcR^–/–)
   knockout mice given HFD exerted attenuated hepatic steatosis and
   insulin resistance.[164]^36 In our study, we found that manipulating
   miR-192-3p levels affected the expression of NR3C1, which in turn
   modulated insulin sensitivity.

   In parallel, we observed that shNR3C1 effectively phenocopied the OE of
   miR-192-3p, while combinational treatment of miR-192-3p and DEX ablated
   the anti-steatotic actions of miR-192-3p. We therefore propose that the
   hepato-protective effect of miR-192-3p/NR3C1 axis may arise from both
   regulation of glucocorticoid signalling and the inter-conversion
   between these lipid storage forms in hepatocytes, which unleashed a
   promising lead for the palliative care of the fatty liver disease in
   patients with diabetes. Notably, miR-192-3p is only detectable in obese
   patients with diabetes, with heightened corticosterone level in their
   serum.

   In summary, the present study establishes the molecular crosstalk
   between miR-192-3p and NR3C1 in the regulation of insulin-regulated
   lipid metabolism and liver steatosis ([165]Fig. 8). Approaches that
   improve lipid metabolism by manipulating a more distal signal
   modulator, such as increasing miR-192-3p OE, might rebalance lipid
   uptake and oxidation through modulating insulin sensitivity, thus
   making miR-192-3p an attractive drug target for carbohydrate
   malnutrition and related metabolic disorders, such as liver steatosis,
   T2DM, and lipogenesis-related diseases.

Fig. 8.

   [166]Fig. 8
   [167]Open in a new tab

   Schematic representation of miR-192-3p in regulating lipid metabolism
   and improving liver steatosis.

   A downregulation of hepatic miR-192-3p was found in mice with diabetes.
   By post-transcriptional regulation of its target glucocorticoid
   receptor (GCR/NR3C1), a master regulator of lipogenesis and
   gluconeogenesis, miR-192-3p significantly alleviates liver steatosis
   and improves insulin sensitivity, through inhibiting lipogenesis- and
   gluconeogenesis-associated pathways. GCR, glucocorticoid receptor;
   HFrD, high-fructose drink; NR3C1, nuclear receptor subfamily 3, group
   C, member 1.

Financial support

   This work was supported by grants from the Health and Medical Research
   Fund, Hong Kong (project number 06170996); Research Grants Council
   General Research Fund, Hong Kong (project number 14103418); Project
   Stem Cell Therapy of Liver Diseases: An Investigation (2015CB964700),
   973 Scheme, Ministry of Science and Technology, China; and Shenzhen
   Virtue University Park Laboratory Support Special Fund (YFJGJS1.0).

Authors' contributions

   Z.W., K.K.M. contributed to the study concept and design, acquisition
   and analysis of data, statistical analysis, and drafting of paper.
   X.Z., A.T.-Y.W. contributed to the acquisition and interpretation of
   data. H.-H.C., G.L. provided technical support. H.-M.L., A.P.-S.K.,
   J.C.-N.C. provided the human samples. W.-Y.C. contributed to the study
   concept and design, revision of the paper for important intellectual
   content, and study supervision.

Data Availability

   The RNA-seq raw data are available from the National Center for
   Biotechnology Information (NCBI) with accession number: PRJNA648925.
   All other data supporting the findings of this study are available from
   the corresponding author on reasonable request.

Conflict of interest

   The authors declare no conflict of interest that pertain to this work.

   Please refer to the accompanying [168]ICMJE disclosure forms for
   further details.

Acknowledgements