Micro-RNA 21 inhibition of SMAD7 enhances fibrogenesis via leptin-mediated NADPH oxidase in experimental and human nonalcoholic steatohepatitis

Micro-RNA 21 inhibition of SMAD7 enhances fibrogenesis vialeptin-mediated NADPH oxidase in experimental and humannonalcoholic steatohepatitis

Diptadip Dattaroy,1 Sahar Pourhoseini,1 Suvarthi Das,1 Firas Alhasson,1 Ratanesh Kumar Seth,1

Mitzi Nagarkatti,2 Gregory A. Michelotti,3 Anna Mae Diehl,3 and Saurabh Chatterjee1

1Environmental Health and Disease Laboratory, Department of Environmental Health Sciences, University of South Carolina,Columbia, South Carolina; 2Department of Pathology, Microbiology, and Immunology, University of South Carolina Schoolof Medicine, Columbia, South Carolina; 3Division of Gastroenterology, Duke University, Durham, North Carolina

Submitted 22 September 2014; accepted in final form 8 December 2014

Dattaroy D, Pourhoseini S, Das S, Alhasson F, Seth RK, NagarkattiM, Michelotti GA, Diehl AM, Chatterjee S. Micro-RNA 21 inhibition ofSMAD7 enhances fibrogenesis via leptin-mediated NADPH oxidase inexperimental and human nonalcoholic steatohepatitis. Am J Physiol Gastro-intest Liver Physiol 308: G298–G312, 2015. First published December 11,2014; doi:10.1152/ajpgi.00346.2014.—Hepatic fibrosis in nonalcoholicsteatohepatitis (NASH) is the common pathophysiological processresulting from chronic liver inflammation and oxidative stress. Al-though significant research has been carried out on the role ofleptin-induced NADPH oxidase in fibrogenesis, the molecular mech-anisms that connect the leptin-NADPH oxidase axis in upregulation oftransforming growth factor (TGF)-� signaling have been unclear. Weaimed to investigate the role of leptin-mediated upregulation ofNADPH oxidase and its subsequent induction of micro-RNA 21(miR21) in fibrogenesis. Human NASH livers and a high-fat (60%kcal) diet-fed chronic mouse model, where hepatotoxin bromodichlo-romethane was used to induce NASH, were used for this study. Toprove the role of the leptin-NADPH oxidase-miR21 axis, mice defi-cient in genes for leptin, p47phox, and miR21 were used. Resultsshowed that wild-type mice and human livers with NASH hadincreased oxidative stress, increased p47phox expression, augmentedNF-�B activation, and increased miR21 levels. These mice andhuman livers showed increased TGF-�, SMAD2/3-SMAD4 colocal-izations in the nucleus, increased immunoreactivity against Col1�,and �-SMA with a concomitant decrease in protein levels of SMAD7.Mice that were deficient in leptin or p47phox had decreased activatedNF-�B and miR21 levels, suggesting the role of leptin and NADPHoxidase in inducing NF-�B-mediated miR21 expression. FurthermiR21 knockout mice had decreased colocalization events ofSMAD2/3-SMAD4 in the nucleus, increased SMAD7 levels, and de-creased fibrogenesis. Taken together, the studies show the novel role ofleptin-NADPH oxidase induction of miR21 as a key regulator of TGF-�signaling and fibrogenesis in experimental and human NASH.

nonalcoholic fatty liver disease; ob/ob; NF-�B; SMAD7; SMAD2/3colocalization; transforming growth factor-�

PATHOGENESIS OF NONALCOHOLIC STEATOHEPATITIS (NASH) is notclearly understood and perceived to comprise an inflammatoryphase of high circulatory leptin, increased oxidative stress,elevated inflammatory cytokines resulting in hepatocellularinjury, and subsequent progression into fibrosis (12). MostNASH pathophysiology is accompanied by late-stage fibrosis(34, 38, 43). Hepatic fibrosis in NASH most likely occurs from

chronic liver inflammation associated with a rise in proinflam-matory cytokines and oxidative stress (27). Fibrosis is closelylinked to accumulation of extracellular matrix (ECM) proteins,mainly type I collagen, which also can occur in many chronicliver diseases (27). The accumulation of ECM proteins distortsthe hepatic architecture by forming a fibrous scar, and thesubsequent development of nodules of regenerating hepato-cytes can lead to condition known as cirrhosis (9). BecauseNASH might arise almost always from a preexisting conditionof obesity, type 2 diabetes, and insulin resistance, the highercirculatory levels of adipose tissue cytokines leptin and TNF-�have been predicted to play a significant role in hepaticfibrogenesis (14). We and others have shown in rodent modelsof NASH that higher leptin is closely associated with fibrosisin NASH (7, 19, 36).

Leptin, a product of the ob gene, is synthesized in the liverand the adipose tissue (30). We have shown that mice fed witha high-fat diet and challenged with low doses of hepatotoxinsdisplay increases in both hepatic and circulatory levels of leptin(3, 35). An early study by Honda et al. (11) showed that leptindeficiency is responsible for resistance to thioacetamide-in-duced fibrogenesis, whereas subsequent studies have high-lighted the role of leptin in stellate cell activation, leptin-induced transforming growth factor (TGF)-� production viaKupffer cells, and leptin signaling-induced collagen productionand ECM formation (3, 8, 11, 33, 39, 40).

Although leptin has been found to play a distinct role in thestellate cell activation and TGF-� production, molecular mech-anisms involving reactive oxygen species, especially the roleof NADPH oxidase, were revealed recently (27). NADPHoxidase, both the phagocytic and nonphagocytic isoforms, hasbeen detected in the liver cell types (26). Hepatic stellate cellshave been found to express the NADPH oxidase isoform 2, anddeletion of one of its cytosolic subunits has resulted in de-creased fibrosis in rodent models of NASH (27). Furtherpublished reports from our laboratory have identified the roleof peroxynitrite, a highly reactive nitroso species formed bysuperoxide and nitric oxide in NASH (3). We further showedthat NADPH oxidase was crucial for peroxynitrite formationthat was again dependent on leptin (3).

It is also critical that high circulatory levels of leptin accom-panied by leptin resistance are common in patients with liverfibrosis (20). Quiescent hepatic stellate cells in humans expressvery low levels of p47phox, a regulatory subunit of NADPHoxidase, but they are highly activated in culture-activatedstellate cells isolated from patients with liver fibrosis (2).

Address for reprint requests and other correspondence: S. Chatterjee, Envi-ronmental Health and Disease Laboratory, Dept. of Environmental HealthSciences, Univ. of South Carolina, Columbia, SC 29208 (e-mail: [email protected])

Am J Physiol Gastrointest Liver Physiol 308: G298–G312, 2015.First published December 11, 2014; doi:10.1152/ajpgi.00346.2014.

0193-1857/15 Copyright © 2015 the American Physiological Society http://www.ajpgi.orgG298

mailto:[email protected]

mailto:[email protected]

Although significant research has been carried out on leptin-induced NADPH oxidase in fibrogenesis, the molecular mech-anisms that connect the leptin-NADPH oxidase axis in upregu-lation of TGF-� signaling have been unclear. Reactive oxygenspecies production by NADPH oxidase, primarily in the formof superoxide or hydrogen peroxide, has been shown to inducenuclear translocation of NF-�B (42). Our unpublished reportsindicate that NF-�B activation led to epigenetic modulations inthe form of upregulation of micro-RNA 21 (miR21) in NASH.miR21 is a small noncoding RNA that has been found to havea distinct role in inflammation, and its regulatory functions inNASH pathophysiology are slowly emerging (25, 44).

miR21, upon induction, targets several proteins either bybinding entirely to the complementary sequence of mRNA ofthe target protein, resulting in a nonfunctional mRNA, orbinding partially to cause a translational defect (28). Regula-tory protein SMAD7, which plays an important role in TGF-�signaling, is a target of miR21 and has been shown to reversethe regulatory effect of this protein (6, 17). Importantly, TGF-�signaling is crucial to the fibrogenesis in NASH and has beenshown conclusively in many studies (13, 41).

In this study, we test the hypothesis that high circulatoryleptin-induced NADPH oxidase-NF-�B activation causesmiR21-mediated SMAD7 inhibition, resulting in increasedTGF-� signaling. The resultant inhibition of SMAD7, a regu-latory SMAD, leads to uninterrupted TGF-� function, causingfibrogenesis in NASH. The results obtained via the use of arodent model of NASH, human tissues, and transgenic micelacking leptin, p47phox, and miR21 show that leptin-inducedNADPH oxidase causes activation of NF-�B and subsequentupregulation of miR21. Lack of miR21 led to significantly lessfibrogenesis, TGF-� production, and downstream signaling ofthe TGF-� pathway.

MATERIALS AND METHODS

Obese mice. Adult male, pathogen-free, 6-wk-old mice withC57BL/6J background (Jackson Laboratories, Bar Harbor, ME) wereused as model for diet-induced obesity. The animals were fed withhigh-fat diet (60% kcal) from 6 wk to 16 wk to develop diet-inducedobesity (DIO). After completion of 16 wk, all experiments wereconducted. Mice that contained disrupted ob gene (leptin) (B6.V-Lepob/J) (Jackson Laboratories) (ob/ob), another group of leptinknockout (KO) mice (ob/ob) treated with leptin (leptin-supplementedgroup, ob/ob � leptin) with disrupted p47 phox gene (B6.129S2-Ncf1tm1shl N14) (Taconic, Cranbury, NJ) (p47 phox KO), and micewith disrupted miR21 gene (B6;129S6-Mir21atm1Yoli/J) (miR21KO) were fed with a high-fat diet and treated identically to DIO mice.The mice were housed one per cage in a temperature-controlled roomat 23–24°C with a 12-h:12-h light/dark cycle with ad libitum access tofood and water. All animals had been treated in strict accordance withthe NIH guidelines for Humane Care and Use of Laboratory Animalsand local IACUC standards. All experiments were approved by theinstitutional review board at NIEHS, Duke University, and the Uni-versity of South Carolina.

Induction of liver injury in obese mice (toxin model). Bromodichlo-romethane (BDCM) (2.0 mmol/kg, diluted in corn oil) was adminis-tered through intraperitoneal injection in 16-wk-old DIO mice (n �6). Liver tissues were collected and pooled after 24-h exposure ofBDCM [DIO � BDCM (24 h)] and after 48-h exposure of BDCM[DIO � BDCM (48 h)].

The other group of DIO mice was administered two doses ofBDCM (1.0 mmol/kg, diluted in corn oil) per week for 1 wk (DIO �BDCM 1 w) and for 4 wk (DIO � BDCM 4 w) (n � 4). Sixteen-

week-old high-fat diet-fed DIO mice, administered with 100 ngGdCl3 per day via intraperitoneal route for 4 wk, were alsoadministered with BDCM (1.0 mmol/kg, diluted in corn oil)through the intraperitoneal route (GdCl3 treated). Sixteen-week-old high-fat diet-fed gene-specific KO mice (ob/ob, p47 phox KO,miR21 KO) were administered with BDCM (1.0 mmol/kg, dilutedin corn oil) through the intraperitoneal route (n � 3). DIO micewere treated with corn oil (diluent of BDCM) to use as a control.After completion of the treatments, all mice were killed for livertissue for further experiments. The tissues were pooled and usedfor the study.

Human tissues. Human liver tissues both from patients with NASHand normal individuals were obtained from NIH repository at Uni-versity of Minnesota and University of Pittsburg. Each experimentwas carried out using three experimental groups and three controlsamples.

Histopathology. Liver sections were collected from each animaland fixed in 10% neutral-buffered formalin. These formalin-fixed,paraffin-embedded tissues were cut in 5-�m-thick sections. Thesesections were deparaffinized using standard protocols and stained withpicrosirius red. Picrosirius red staining of liver sections was done byusing Nova ultra Sirius red stain kit following the manufacturer’sprotocol (IHC World, Woodstock, MD), and they were observedusing �20 objectives under the light microscope.

Immunohistochemistry. Formalin-fixed, paraffin-embedded tissueswere cut in 5-�m-thick sections. Each section was subjected todeparaffinization using standard protocols. To describe briefly, thesesections were washed with two changes of 100% xylene twice for 3min followed by xylene, ethanol (1:1) for 3 min, and rehydratedthrough a series of ethanol (twice with 100%, 95%, 70%, and 50%),twice with distilled water. The sections were finally rinsed twice withPBS. Epitope retrieval of the deparaffinized sections was carried outusing epitope retrieval solution and steamer (IHC World) by follow-ing the manufacturer’s protocol. The primary antibodies anti-3 nitro-tyrosine, anti-TGF-�, anti-connective tissue growth factor (CTGF),anti-extra domain A-fibronectin (EDAFN), anti-�-smooth muscleactin (SMA), and anti-Col1�1 were purchased from Abcam (Cam-bridge, MA) and used in recommended dilutions. Species-specificbiotinylated-conjugated secondary antibody and streptavidin conju-gated with horseradish peroxidase (HRP) were used from VectastainElite ABC kit (Vector Laboratories, Burlingame, CA) to performantigen-specific immunohistochemistry following the manufacturer’sprotocols. 3,3=Diaminobenzidine (Sigma-Aldrich, St. Louis, MO) wasused as a chromogenic substrate. Tissue sections were counterstainedby Mayer’s hematoxylin (Sigma-Aldrich). Phosphate buffer salinewas used for washing thrice between the steps. Sections were mountedin Simpo mount (GBI Laboratories, Mukilteo, WA) and observedunder an �20 objective. Morphometric analysis was done usingCellSens Software from Olympus America.

Immunofluorescence dual-labeling microscopy. Formalin-fixed,paraffin-embedded tissues were cut in 5-�m-thick sections. Eachsection was subjected to deparaffinization using standard protocols.Epitope retrieval of the deparaffinized sections was carried out usingepitope retrieval solution and steamer (IHC World) by following themanufacturer’s protocol. The primary antibodies anti-SMAD2/3, anti-SMAD4, anti-gp-91phox, and anti-p47phox (purchased from Cell SignalingTechnology, Beverly, MA and Santa Cruz Biotechnology, Santa Cruz,CA) were used at recommended dilutions. Species-specific anti-IgGsecondary antibodies conjugated with Alexa Fluor 488, Alexa Fluor633 (Invitrogen, Carlsbad, CA) were used together against anti-SMAD4 and anti-SMAD2/3, respectively. Alexa Fluor 488 and AlexaFluor 568 were used together against anti-p47phox and anti-gp91phox, respectively. The sections were mounted in a ProLonggold antifade reagent with DAPI (Life Technologies, Eugene, OR).Images were taken under �20/�60 oil objectives using an OlympusBX51 microscope.

G299OXIDATIVE STRESS-MIR21 REGULATION OF LIVER FIBROSIS

AJP-Gastrointest Liver Physiol • doi:10.1152/ajpgi.00346.2014 • www.ajpgi.org

Quantitative real-time RT-PCR. Real-time RT-PCR was done tomeasure gene expression (mRNA) levels in the liver tissue samples.Total RNA was isolated from each liver tissue by homogenization inTRIzol reagent (Life Technologies) according to the manufacturer’sinstruction and was purified by using RNeasy mini kit columns(Qiagen, Valencia, CA). iScript cDNA synthesis kit (Bio-Rad, Her-cules, CA) was used to convert 1 �g of purified RNA to cDNAfollowing the manufacturer’s standard protocol.

qRT-PCR was performed with the gene-specific primers by usingSsoAdvanced universal SYBR Green supermix (Bio-Rad) and CFX96thermal cycler (Bio-Rad). Threshold cycle (Ct) values for selectedgenes were normalized against 18S (internal control) values in thesame sample. For each gene and each tissue sample, each reaction wascarried out in triplicate. A liver sample from DIO mice was used as acontrol to compare with all other liver samples in the toxin model ofNASH. The relative fold change was calculated by the 2Ct

method. The sequences for the primer used for real-time PCR are asfollows: human p47phox forward: GTACCCAGCCAGCACTATG,reverse: CCTGGCTTTGCTTTCATCTG; mouse p47phox forward:GGTCGACCATCCGCAACGCA, reverse: TGTGCCATCCGTGCT-CAGCG.

miR21 expression levels in liver tissues. Total miRNA was isolatedfrom liver tissue by homogenization in Qiazol reagent (Qiagen)following the manufacturer’s instructions. The purification was doneby using miRNeasy mini kit columns (Qiagen). Purified miRNA(1,000 ng) was converted to cDNA using miScript cDNA synthesiskit (Qiagen) following the manufacturer’s protocol. qRT-PCR wasperformed with miRNA-specific primers (Qiagen) using miScriptSYBR Green PCR master mix (Qiagen) and CFX96 thermal cycler(Bio-Rad). Ct values for the selected gene were normalized againstRNU6-2 (internal miR expression control) values in the samesample.

Western blotting. Thirty milligrams of each liver tissue werehomogenized in 500 �l of RIPA buffer (Sigma-Aldrich) in thepresence of phosphatase and protease inhibitor (Pierce, Rockford, IL)by using dounce homogenizer. After centrifuging the homogenate, wecollected the supernatant for further experiments. Thirty microgramsof protein from each sample were loaded on 4–12% Bis-Tris gradientgel (Invitrogen) for SDS-PAGE. By using precut nitrocellulose/filterpaper sandwiches (Bio-Rad) and Trans-Blot Turbo transfer system(Bio-Rad), proteins were transferred to nitrocellulose membrane.Blots were blocked with 5% nonfat milk solution. Primary antibodyagainst SMAD7, leptin, and �-actin (purchased from Abcam) wereused at recommended dilutions and incubated overnight at 4°C.Species-specific secondary antibody conjugated with HRP was used.Pierce ECL Western Blotting substrate (Thermo Fisher Scientific,Rockford, IL) was used. The blot was developed by using BioMaxMS Films and cassettes (with intensifying screen, Kodak). Densitom-etry analysis of the images was done using Lab Image 2006 Profes-sional 1D gel analysis software from KAPLEAN Bioimaging Solu-tions (Liepzig, Germany).

NF-�B transcription factor binding assay. Nuclear fractions ofeach liver sample were isolated by using NF-�B assay kit andfollowing manufacturer’s standard protocol (Fivephoton Biochemi-cals, San Diego, CA). DNA binding activity of the transcription factorNF-�B present in those nuclear extracts was detected using the ELISAmethod (Abcam NF-�B p65 Transcription Factor Assay Kit) follow-ing the manufacturer’s standard protocol.

Statistical analyses. In vivo experiments were repeated two to threetimes with an average of three mice per group with other in vivoexperiments having four to six mice per group (n � 3/4/6 whereverapplicable; data from each group of mice were pooled). The statisticalanalysis of data was done by ANOVA followed by Bonferroni posthoc correction for performing intergroup comparisons. Quantitativedata from Western blots, depicted by the relative intensity of the

bands, were analyzed by performing Student’s t-test. *P � 0.05 and#P � 0.01 are considered statistically significant.

RESULTS

Leptin induces NADPH oxidase subunit p47phox and mem-brane association with gp91phox in NASH of both mice andhumans. We have shown previously that leptin inducesp47phox in early steatohepatitis, leading to the formation ofperoxynitrite and corresponding inflammation and Kupffer cellactivation (3). To prove the role of leptin in induction ofp47phox in fibrogenesis of NASH, experiments were per-formed in both a mice model of NASH and human tissuesamples from patients with NASH. Results showed that theDIO � BDCM group that shows typical NASH lesions had a1.6-fold increase in p47phox mRNA expression compared withthe DIO group alone (P � 0.05) (Fig. 1A). Leptin KO (ob/ob)mice showed a significant decrease in p47phox expressioncompared with the DIO � BDCM group (P � 0.05), whereasleptin supplementation in leptin KO (ob/ob � leptin) miceresulted in a significant upregulation of p47phox mRNA com-pared with only leptin KO (ob/ob) mice (P � 0.01). HumanNASH (Hu NASH) tissues showed a significant increasep47phox mRNA expression compared with controls (HuCTRL) (P � 0.05) (Fig. 1B). p47phox subunit of NADPHoxidase requires its association with its membrane counterpartgp91phox for activation of this enzyme. We studied the mem-brane association of these two subunits by immunofluores-cence microscopy in both mice and human tissue samples.Results were analyzed based on the number of colocalization/overlay events (shown by yellow) per 300 cells counted (Fig. 1,C and D). Results showed that the DIO � BDCM group had asignificant increase in the colocalization events compared withthe DIO only group (P � 0.01) (Fig. 1, C and D). Leptin KO(ob/ob) mice had a significant decrease in the number ofcolocalization events compared with the DIO � BDCM group(P � 0.01). Leptin-supplemented leptin KO mice (ob/ob �leptin) had a significant increase in the colocalization eventscompared with leptin KO (ob/ob) mice (P � 0.05) (Fig. 1, Cand D). Human NASH samples showed a significant increasein the p47phox/gp91phox colocalization compared with hu-man controls (Fig. 1, C and D) (P � 0.05). The above resultssuggest that leptin induced p47phox expression in NASHand aided in the membrane association of p47phox andgp91phox that assumes significance in causing oxidativestress in NASH.

Leptin induced NADPH oxidase activation causes peroxyni-trite-mediated oxidative stress in NASH of both mice andhumans. We have previously shown that peroxynitrite-medi-ated oxidative stress plays a crucial role in activating Kupffercells in early steatohepatitic lesions in NASH (3). To prove thatleptin induced activation of NADPH oxidase in causing oxi-dative stress through peroxynitrite formation, experimentswere performed in vivo to estimate 3-nytrotyrosine immuno-reactivity in these tissues (Fig. 2). Results showed that 3-ni-trotyrosine immunoreactivity was significantly increased in theDIO � BDCM group compared with the DIO group (P �0.05) (Fig. 2, A and B). Leptin KO (ob/ob) mice had asignificant decrease in 3-nitrotyrosine immunoreactivitycompared with DIO � BDCM group (P � 0.05). Leptin-supplemented leptin KO (ob/ob � leptin) mice showed a

G300 OXIDATIVE STRESS-MIR21 REGULATION OF LIVER FIBROSIS


significant increase in the 3-nitrotyrosine immunoreactivitycompared with leptin KO (ob/ob) mice (P � 0.05). HumanNASH samples (n � 3) showed a significant increase in3-nitrotyrosine immunoreactivity compared with humancontrols, as shown by morphometry analysis of the slides(P � 0.05) (Fig. 2B).

All morphometric analysis of the stained liver sections wasexpressed as a percentage area of immunoreactivity (Fig. 2B).The results suggested that leptin was significantly associated withan increase in 3-nitrotyrosine immunoreactivity, and peroxynitritemight be a prominent player in causing oxidative stress.

Leptin-induced NADPH oxidase activation mediates NF-�Btranslocation to the nuclei and upregulation of miR21. NA-DPH oxidase-mediated oxidative stress has been shown toinduce NF-�B translocation (42). NF-�B translocation and its

binding to miR21 promoter are crucial for many inflammatoryevents in disease progression (37). To prove that leptin,through its activation of NADPH oxidase in vivo causesNF-�B translocation and binding to DNA, we performedexperiments with liver tissue of both mice and humans. Resultsshowed that NF-�B activation and DNA binding were signif-icantly increased in the DIO � BDCM group compared withthe DIO only group (P � 0.05) (Fig. 3A). Leptin KO (ob/ob)mice showed a significant decrease in the NF-�B activationand binding compared with the DIO � BDCM group (P �0.05). p47phox KO mice, which cannot participate in NADPHoxidase-mediated oxidative stress, also showed a significantdecrease in NF-�B activation and binding to DNA comparedwith the DIO � BDCM group (P � 0.05) (Fig. 3A). HumanNASH samples from liver showed a significant increase in

Fig. 1. Leptin induces NADPH oxidase subunit p47phox and membrane association with gp91phox. A: qRT-PCR analysis of p47phox mRNA expression ofdiet-induced obesity (DIO), DIO � bromodichloromethane (BDCM), ob/ob, and ob/ob � leptin mice samples normalized against DIO. B: qRT-PCR analysisof p47phox mRNA expression of human (Hu) nonalcoholic steatohepatitis (NASH) samples normalized against Hu control (CTRL). C: immunofluorescencedual-labeling microscopy depicting gp91phox (red)-p47phox (green) colocalization (yellow) in DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), HuCTRL (v), and Hu NASH (vi) liver samples taken at �60 (oil) magnification. D: morphometric analysis of colocalization events of C, shown in colocalizationevents/300-cell unit. *P � 0.05 and #P � 0.01 are considered statistically significant.



NF-�B activation and binding to DNA compared with corre-sponding control tissues (P � 0.01) (Fig. 3B). The DIO �BDCM group had a significant increase in miR21 levels(4-fold) compared with the DIO group (P � 0.01) (Fig. 3C).Leptin KO (ob/ob) or p47phox KO mice had a significantdecrease in miR21 levels compared with the DIO � BDCMgroup (P � 0.01) (Fig. 3C). However, leptin supplementationto leptin KO (ob/ob � leptin) mice had a significant increase inmiR21 levels compared with leptin KO (ob/ob) mice and hadcomparable levels of miR21 with the DIO � BDCM group(P � 0.01) (Fig. 3C). Human liver samples that had NASHetiology had significantly higher miR21 levels (4.5-fold) com-pared with human control tissues (P � 0.01) (Fig. 3D). Theabove results suggested that leptin-mediated NADPH oxidaseactivation was significantly associated with increased miR21levels in NASH. The results assume significance becausemiR21-mediated repression of its target proteins can play amajor role in disease pathology.

Leptin-mediated NADPH oxidase activation and its corre-sponding induction of miR21 repress regulatory SMAD proteinin both mice and human NASH. miR21 has been found torepress many functional proteins in disease pathology (31).SMAD7, a regulatory protein in the TGF-� signaling cascade,has been shown to be a target for miR21 (6). We studied therole of the leptin-NADPH oxidase axis in mediating miR21-induced SMAD7 repression. Results showed that protein levelsof SMAD7 were significantly elevated during the initial phasesof NASH development (DIO � BDCM, 48 h) compared withthe DIO group. The protein levels of SMAD7 decreasedsignificantly at 1 wk and at the termination of the study (4 wk)that showed full-blown NASH symptoms and correlated wellwith fibrotic lesions (P � 0.05) (Fig. 4, A and C). miR21 KOmice showed a significant increase in SMAD7 protein expres-sion compared with the DIO � BDCM group at 4 wk (P �0.01) (Fig. 4, B and D). Leptin KO (ob/ob) mice or p47phoxKO mice or DIO � BDCM mice that were depleted withmacrophage toxin GdCl3 did not show any significant proteinlevels of SMAD7 (Fig. 4, B and D). This is not surprising

because these mice did not have oxidative stress and there wasno induction of SMAD7 in the initial phases (data not shown).Human NASH tissues showed a significant decrease inSMAD7 protein levels, as analyzed by Western blot comparedwith corresponding controls (P � 0.01) (Fig. 4, B and D). Theresults suggested that the increased miR21 in mice and humanNASH is strongly associated with decreased SMAD7 protein.The results assume significance because SMAD7 plays a cru-cial role in regulating and inhibiting TGF-� signaling, crucialfor fibrogenesis in NASH.

Leptin-mediated NADPH oxidase activation and corre-sponding induction of miR21 regulate TGF-� levels in miceand human NASH. TGF-� has been shown to be key for NASHfibrogenesis (10). To prove the role of leptin-mediated NADPHoxidase induction of miR21 in regulating TGF-� levels inNASH, experiments were performed with liver slices of miceand human NASH tissues, and we assessed the immunoreac-tivity by using immunohistochemistry. Results showed that theDIO � BDCM group had a significantly increased level ofTGF-� compared with the DIO group (P � 0.01) (Fig. 5, A andB). Leptin KO (ob/ob), p47phox KO, and miR21 KO mice hada significant decrease in TGF-� immunoreactivity comparedwith the DIO � BDCM group (P � 0.01) (Fig. 5, A and B).Leptin supplementation to leptin KO (ob/ob � leptin) miceshowed a significant increase in TGF-� immunoreactivitycompared with leptin KO (ob/ob) mice alone (P � 0.05) (Fig.5, A and B). Human NASH liver slices showed a significantincrease in TGF-� immunoreactivity compared with controltissues, as shown by percentage area covered by the immuno-reactive staining (morphometric analysis) (P � 0.05) (Fig. 5, Aand B). The results suggested that leptin-mediated NADPHoxidase activation and miR21 induction play a crucial role inTGF-� levels in the livers of NASH.

Leptin-mediated NADPH oxidase activation and corre-sponding induction of miR21 in NASH are crucial for regula-tion of SMAD2/3-SMAD4 assembly in the nucleus, a key eventin TGF-� signaling pathway. TGF-� signaling throughSMAD2/3 is crucial for collagen deposition and fibrogenesis

Fig. 2. Leptin-induced NADPH oxidase activation causes peroxynitrite-mediated oxidative stress in NASH of both mice and humans. A: 3-nitrotyrosine (3 N-Tyr)immunoreactivity as shown by immunohistochemistry in liver slices from the DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), Hu CTRL (v), andHu NASH (vi) groups, respectively. Images were taken at �20 magnification. B: morphometric analysis of the immunoreactivity of A. *P � 0.05 is consideredstatistically significant.



(22–24). It has been shown previously that SMAD2/3 bindswith SMAD4 and translocates to the nucleus. The nuclearcolocalization of the SMAD2/3-SMAD4 heterodimer com-plex is a key event in the TGF-� functional pathway. TGF-�induces heteromeric complexes of SMAD2, 3, and 4, andtheir concomitant translocation to the nucleus, which isrequired for efficient TGF-� signal transduction (23). Tostudy the role of leptin-mediated NADPH oxidase activationand miR21 induction in TGF-� signal transduction pathway,experiments were performed using liver tissue slices. Im-munoreactivity to SMAD2/3 and SMAD4 and their nuclearcolocalizations was assessed by immunofluorescence mi-croscopy. Results showed that the number of colocalizationevents in the nucleus were significantly higher in the DIO �BDCM group compared with the DIO group (P � 0.01)(Fig. 6, A and C). The colocalization events in the leptin KO

(ob/ob), p47phox KO, and miR21 KO were significantlydecreased compared with the DIO � BDCM group (P �0.01 and P � 0.05). Leptin supplementation to leptin KO(ob/ob � leptin) mice significantly increased the number ofnuclear colocalization events compared with leptin KO(ob/ob) mice alone (P � 0.05). Human NASH samplesshowed significant increase in SMAD2/3 and SMAD7 nu-clear colocalizations compared with human liver controls(P � 0.05) (Fig. 6, A and C). The results were analyzedbased on the number of nuclear colocalizations per 300 cells(Fig. 6C). Oil micrographs (�60) of the DIO � BDCMgroup in the mouse model and human NASH liver are shownin Fig. 6B. The results show that increased TGF-�-inducedsignaling through SMAD2/3 and SMAD4 is regulated byleptin-mediated NADPH oxidase and its subsequent upregu-lation of miR21.

Fig. 3. Leptin-induced NADPH oxidase activation mediates NF-�B activation and upregulation of micro-RNA 21 (miR21). A: normalized % change ofNF-�B activation in DIO, DIO � BDCM, ob/ob, and p47 phox knockout (KO) samples. B: normalized % change of NF-�B activation in Hu CTRL andHu NASH samples. C: qRT-PCR analysis of miR21 expression of DIO, DIO � BDCM, ob/ob, and ob/ob � leptin mice samples normalized against DIO.D: qRT-PCR analysis of miR21 expression of Hu NASH samples normalized against Hu CTRL. *P � 0.05 and #P � 0.01 are considered statisticallysignificant.



Leptin-mediated NADPH oxidase activation and subsequentmiR21 induction regulate the levels of TGF-� signalingproteins. CTGF and EDAFN were evaluated in a mice modelof NASH and human NASH livers. In fibrotic liver, CTGF isconstantly expressed in activated hepatic stellate cells and actsdownstream of TGF-� to modulate ECM production (18). Onthe other hand, FN is a major component of the ECM andoccurs in two main forms: plasma and cellular FN. The latterincludes the alternatively spliced domain A (EDA) (15).EDAFN has been shown to participate in TGF-�-mediatedfibroblast differentiation and fibrogenesis (1). To prove the roleof leptin-mediated NADPH oxidase activation and subsequentinduction of miR21 in TGF-� signaling, immunohistochemicalanalysis of the mice liver and human liver tissues was per-formed. Results showed that there was a significant increase inthe immunoreactivity of CTGF and EDAFN in the DIO �BDCM group compared with the DIO group (P � 0.01 andP � 0.05) (Fig. 7A, ii, and 7B, ii) (Fig. 7, C and D, formorphometric analysis). On the other hand, leptin KO (ob/ob),p47phox KO, and miR21 KO mice had a significant decreasein the CTGF and EDAFN levels compared with the DIO �BDCM group (Fig. 7A, iii, v, and vi) (Fig. 7B, iii, v, and vi)(Fig. 7, C and D for morphometric analysis) (P � 0.05). Leptinsupplementation in leptin KO (ob/ob � leptin) mice showed asignificant increase in the levels of immunoreactivity of boththese crucial proteins (P � 0.05) (Fig. 7A, iv, and 7B, iv) (Fig.

7, C and D, for morphometric analysis). Human NASH livershad a significant increase in the immunoreactivity of CTGFand EDAFN compared with control livers (P � 0.05) (Fig. 7A,viii, and 7B, viii) (Fig. 7, C and D, for morphometric analysis).The above results show that leptin-mediated NADPH oxidaseactivation and subsequent miR21 inhibition of SMAD7 areclosely associated with increased TGF-� signaling, resulting inhigher CTGF and EDAFN levels in NASH.

Leptin-mediated NADPH oxidase activation and miR21 in-duction cause stellate cell activation and collagen depositionin NASH. Stellate cell activation is a key event for NASHprogression (36). �-SMA is a biomarker for stellate cell acti-vation (5). Importantly, increased fibrogenesis and TGF-�signaling result in higher Col1�1 levels in tissues. To show therole of leptin-mediated NADPH oxidase in stellate cell activa-tion and fibrogenesis, immunohistochemical analyses of miceand human liver tissues were performed. Results showed thatimmunoreactivity to �-SMA and Col1� was significantly highin the DIO � BDCM group compared with the DIO group(P � 0.01) (Fig. 8A, ii, and 8B, ii) (morphometric analysis Fig.8, C and D). Leptin KO (ob/ob), p47phox KO, and miR21 KOmice had significantly decreased immunoreactivity of �-SMAand Col1� compared with the DIO � BDCM group (P � 0.01)(Fig. 8A, iii, v, and vi; 8B, iii, v, and vi) (morphometric analysisFig. 8, C and D). Leptin supplementation in leptin KO (ob/ob � leptin) mice showed a significant increase in the levels of

Fig. 4. Leptin-mediated NADPH oxidase activation and its corresponding induction of miR21 repress regulatory SMAD7 protein both in mice and human NASH.A: Western blot analysis of SMAD7 from liver homogenates of DIO, DIO � BDCM 24 h, DIO � BDCM 48 h, DIO � BDCM 1 wk (w), and DIO � BDCM4 w groups, respectively. B: Western blot analysis of SMAD7 from liver homogenates of DIO, DIO � BDCM 4 w, ob/ob, ob/ob � leptin, p47phox KO,GdCl3-treated sample, miR21 KO, Hu CTRL, and Hu NASH groups, respectively. C: immunoreactive band analysis of SMAD7 normalized against �-actin.y-Axis depicts the SMAD7/actin ratio from DIO, DIO � BDCM 24 h, DIO � BDCM 48 h, DIO � BDCM 1 w, and DIO � BDCM 4 w groups. D:immunoreactive band analysis of SMAD7 normalized against �-actin. y-Axis depicts the SMAD7/actin ratio from DIO, DIO � BDCM 4 w, ob/ob, ob/ob �leptin, p47phox KO, GdCl3-treated sample, miR21 KO, Hu CTRL, and Hu NASH groups. *P � 0.05 and #P � 0.01 are considered statistically significant.



both �-SMA and Col1� compared with the leptin KO mice(ob/ob) group alone (P � 0.05) (Fig. 8A, iv, and 8B, iv)(morphometric analysis Fig. 8, C and D). Human NASH liverexhibited a significant increase in �-SMA and Col1�, shownby increased immunoreactivity to these proteins and morpho-metric analysis of the tissue staining (P � 0.05) (Fig. 8A, viii,and 8B, viii) (morphometric analysis Fig. 8, C and D). Theresults suggested that leptin-mediated NADPH oxidase and thesubsequent induction of miR21 regulated the stellate cell acti-vation and corresponding collagen deposition in both mice andhuman NASH livers.

Leptin-mediated NADPH oxidase activation and miR21 in-duction cause fibrogenesis in NASH. Analysis of picrosiriusstaining for collagen showed increased reactivity in the DIO �BDCM group, especially in the periportal regions, comparedwith the DIO group (Fig. 9A, ii). Leptin KO (ob/ob), p47phoxKO, or miR21 KO mice showed a minimal fibrotic staincompared with the DIO � BDCM group (Fig. 9A, iii, v, andvi). Leptin supplementation sharply increased periportal stain-ing of picrosirius red (Fig. 9A, iv), whereas human tissues forNASH showed higher picrosirius red staining compared with

controls (Fig. 9A, viii). To show a correlation between thelevels of tissue leptin and fibrotic stage, leptin immunoreactiv-ity in mouse liver homogenates at 24 h, 48 h, 1 wk, and 4 wkwas assessed by Western blot followed by normalizationsagainst �-actin (Fig. 9, B and C) and by picrosirius red stainingdepicting fibrosis at 24 h, 48 h, 1 wk, and 4 wk in the DIO �BDCM group (Fig. 9D). Results showed that increased leptinconcentrations correlated well with periportal and bridgingfibrosis at 4 wk following the toxin exposure in an underlyingcondition of steatosis (NASH model).

DISCUSSION

In this study, we report for the first time the role of leptin-mediated NADPH oxidase in inducing miR21 in a preclinicalNASH model as well as in human patients with NASH. Wealso show the mechanistic role of miR21 in repressingSMAD7, a crucial regulator of TGF-� signaling, thus modu-lating fibrogenesis in NASH (Fig. 10). Significantly, the miR21modulation of TGF-� signaling, fibrogenesis, and NASH pro-gression were mediated by NADPH oxidase in our present

Fig. 5. Leptin-induced NADPH oxidase activation and corresponding induction of miR21 regulate transforming growth factor (TGF)-� levels in mice and humanNASH. A: TGF-� immunoreactivity as shown by immunohistochemistry in liver slices from the DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv),p47phox KO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) groups, respectively. Images were taken at �20 magnification. B: morphometric analysisof the immunoreactivity of A. *P � 0.05 and #P � 0.01 are considered statistically significant.



study. The study assumes huge clinical significance becausethe role of NADPH oxidase has been shown in fibrogenesis byus and others, but the link between NADPH oxidase-based oxi-dative stress signaling through posttranscriptional epigeneticchanges and TGF-� signaling remained largely elusive in vivo.

An earlier report from our laboratory showed that leptin-mediated NADPH oxidase activation activated Kupffer cellslargely through the formation of peroxynitrite in an early stageof steatohepatitic injury (3). Through the present study, weadvanced our understanding about the role of leptin-mediatedNADPH oxidase-based oxidative stress in a full-blown NASHmodel and in human NASH livers. The present study used arodent model of NASH that had a hepatotoxin BDCM asa second hit (36). All our investigations were carried out at atime point where NASH symptoms were confirmed usingclinical indicators. NADPH oxidase subunit p47phox inductionand subsequent activation were observed in these liver tissues

(Fig. 1, A–D). The induction and subsequent activation ofNADPH oxidase were dependent on the presence of leptinbecause leptin KO mice (ob/ob) or leptin supplementation (100ng, daily for 4 wk) in leptin KO (ob/ob � leptin) mice hadopposite outcomes (Fig. 1). On the basis of our previousreports in the same model, we also established the nature of theoxidizing species because we observed significant increase in3-nitrotyrosine immunoreactivity, a measure for generation ofhighly reactive peroxynitrite (Fig. 2, A and B).

Highly reactive oxidizing species have been shown to inducechanges in the epigenetic foci and cause increases in theexpression of noncoding RNAs (16). There are several miRNAs thatare known to be induced in an inflammatory condition (31, 37).NASH has been shown to induce several miRNAs with specificfunctions targeted toward its progression (4, 21, 29). miR21 isinduced in an inflammatory microenvironment and has beenshown to be induced in disease pathophysiology following the

Fig. 6. Leptin-mediated NADPH oxidase activation and corresponding induction of miR21 are crucial for regulation of SMAD2/3-SMAD4 assembly in thenucleus. A: immunofluorescence dual labeling depicting SMAD2/3 (red)-SMAD4 (green) colocalization (yellow) in the DIO (i), DIO � BDCM (ii), ob/ob (iii),ob/ob � leptin (iv), p47phox KO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) liver samples, taken at �20 magnification. B: immunofluorescencedual labeling of DIO � BDCM (mouse) (i) and Hu NASH (human) (ii) liver samples depicting SMAD2/3 (red)-SMAD4 (green) colocalization (yellow) takenat �60 (oil) magnification. C: morphometric analysis of colocalization events of A, shown in colocalization events/300-cell unit. *P � 0.05 and #P � 0.01 areconsidered statistically significant.



Fig. 7. Leptin-mediated NADPH oxidase activation and subsequent induction of miR21 regulate TGF-� signaling proteins. Connective tissue growth factor(CTGF) and extra domain A-fibronectin (EDAFN) in mice and human NASH. A: CTGF immunoreactivity as shown by immunohistochemistry in liver slicesfrom the DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), p47phox KO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) groups, respectively.Images were taken at �20 magnification. B: EDAFN immunoreactivity as shown by immunohistochemistry in liver slices from the DIO (i), DIO � BDCM (ii),ob/ob (iii), ob/ob � leptin (iv), p47phox KO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) groups, respectively. Images were taken at �20magnification. C: morphometric analysis of the immunoreactivity of A. D: morphometric analysis of the immunoreactivity of B. *P � 0.05 and #P � 0.01 areconsidered statistically significant.



Fig. 8. Leptin-mediated NADPH oxidase activation and miR21 induction cause stellate cell activation and collagen deposition in NASH. A: �-smooth muscleactin (SMA) immunoreactivity as shown by immunohistochemistry in liver slices from the DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), p47phoxKO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) groups, respectively. Images were taken at �20 magnification. B: Col1�1 immunoreactivity asshown by immunohistochemistry in liver slices from the DIO (i), DIO � BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), p47phox KO (v), miR21 KO (vi), Hu CTRL(vii), and Hu NASH (viii) groups, respectively. Images were taken at �20 magnification. C: morphometric analysis of the immunoreactivity of A. D:morphometric analysis of the immunoreactivity of B. *P � 0.05 and #P � 0.01 are considered statistically significant.



binding of nuclear transcription factor NF-�B to its promoter(37). In the present study, we show that NF-�B binding activityto the DNA was significantly higher in the DIO � BDCMgroup (NASH mouse model) and human NASH livers (Fig. 3,A and B). The activation NF-�B was dependent on leptin andNADPH oxidase activation because leptin KO (ob/ob) miceand p47phox KO mice had significantly lower NF-�B activa-tion (Fig. 3A). Human livers had a significant increase inNF-�B activation, an index of oxidative stress and inflamma-tory signaling pathway (Fig. 3B). Coupled with the NF-�Bactivation, our studies also showed that the miR21 levels in theNASH livers were significantly elevated (Fig. 3, C and D). Thelevels of miR21 were dependent on the NADPH oxidase subunitp47phox and leptin. The results correlate well with observationsreported previously about NF-�B-mediated miR21 induction invitro through a binding of NF-�B to the miR21 promoter region(32). However, the above phenomenon could not be established inthe in vivo models of NASH because of procedural complicationsinvolving transfection in primary cells from the NASH mice atthis point. Thus it can be justified to assume that NF-�B translo-cation in the NASH livers might bind to miR21 promoter, causingits induction, apart from the other functions that it is predicted toperform in NASH (Fig. 3, C and D).

A recent review by Noetel et al. (25) describes the currentknowledge of the role of miR21 in liver fibrosis. Inflammatorymicroenvironments in the NASH livers are predicted to pro-

mote TGF-� signaling, which in turn can stimulate SMAD2/3phosphorylation and their association with SMAD4. Further-more, the above events can upregulate miR21, which bysuppressing SMAD7 can further add to the increased TGF-�-mediated profibrogenic responses, events that are known inother fibrotic mechanisms in the lung and kidney (6, 17). Weare not aware of any reports that establish a direct role ofNADPH oxidase activation-induced miR21 in TGF-� signal-ing and subsequent liver fibrosis. To prove the role of miR21-induced SMAD7 repression and SMAD2/3-SMAD4 interac-tions, we used miR21 KO mice in parallel to p47phox andleptin KO (ob/ob) mice. Results showed that miR21 KO micethat show significantly lower inflammation in similarly usedNASH models (Pourhoseini S, Seth RK, Das S, Dattaroy D,Kadiiska M, Nagarkatti M, Michelotti GA, Diehl AM, andChatterjee S, unpublished data) had higher SMAD7 levelswhen treated identically with high-fat diet and hepatotoxinchallenge compared with the DIO � BDCM group (Fig. 4, Band D). Interestingly, SMAD7 proteins were significantly de-creased in human NASH livers compared with controls (Fig. 4,B and D). Leptin KO (ob/ob) and p47phox KO mice had asmall increase in protein levels compared with the DIO �BDCM group at 4 wk, but the reported increase was notstatistically significant (Fig. 4D). This might be due to thegenetic makeup of these mice and a very low level of inflam-mation that triggers SMAD7 protein expression. Also, it is

Fig. 9. Leptin-mediated NADPH oxidase activation and miR21 induction cause fibrogenesis in NASH. A: picrosirius red (PSR) staining of DIO (i), DIO �BDCM (ii), ob/ob (iii), ob/ob � leptin (iv), p47phox KO (v), miR21 KO (vi), Hu CTRL (vii), and Hu NASH (viii) liver slices. Fibrosis is depicted by red staining.Images were taken at �20 magnification. B: leptin immunoreactivity in mouse liver homogenates at 24 h, 48 h, 1 wk, and 4 wk as assessed by Western blotfollowed by normalizations against �-actin in C. D: PSR staining depicting fibrosis at 24 h, 48 h, 1 wk, and 4 wk in the DIO � BDCM group. *P � 0.05 isconsidered statistically significant.



worth reporting the levels of TGF-� in leptin, p47phox, andmiR21 KO mice (Fig. 5, A and B). Results showed that theseKO mice had significantly decreased TGF-� levels comparedwith the DIO � BDCM group (Fig. 5). The results alsoestablish that TGF-� levels are significantly regulated up-stream of miR21 induction, and leptin, NADPH oxidase, andmiR21 are linked to cause an upregulation of TGF-�.

TGF-� signaling is crucial for fibrogenesis in NASH.SMAD2/3 association with SMAD7 in the nucleus is a crucialevent in this process (22). This important mechanistic associ-ation and its regulation by miR21 have never been shown inliver fibrosis in vivo or in human NASH livers. We show byimmunofluorescence imaging ex vivo the colocalization ofSMAD2/3-SMAD4 in the nuclei of NASH livers (Fig. 6).Murine and human NASH livers had a significant increase inthe number of colocalization events compared with corre-sponding controls, whereas leptin, p47phox, and miR21 KOmice had significantly fewer colocalization events comparedwith NASH groups (Fig. 6C). The results described in thisstudy assume great significance because we show that NADPHoxidase through miR21, not only repressed SMAD7 as a profi-brogenic mechanism, but also contributed partly to enhanceSMAD2/3-SMAD4 association in the nucleus, a key event in liverfibrogenesis. The leptin-mediated NADPH oxidase induction ofmiR21 also played a significant role in the downstream events ofTGF-� signaling pathway by regulating the levels of CTGF andthe intracellular form of FN (EDAFN) (Fig. 7).

Finally, we studied the implications of the leptin-mediatedNADPH oxidase activation and subsequent miR21 upregula-tion in stellate cell activation, collagen levels, and fibrosis, keyevents in NASH pathogenesis. Predictably, our results showeda direct involvement of leptin-mediated NADPH oxidase and a

miR21 role in fibrogenesis because leptin KO (ob/ob),p47phox KO, or miR21 KO mice had significantly decreasedlevels of �-SMA, Col1�, and picrosirius red staining comparedwith mouse and human NASH models (Figs. 8 and 9). Impor-tantly, NASH pathogenesis for proving stellate cell activationand fibrosis in both preclinical and clinical settings relies on thelevels of �-SMA, Col1�, and picrosirius red staining.

In summary, we show that leptin-mediated NADPH oxidaseupregulated the levels of miR21 through NF-�B activationpredictively by binding to the miR21 promoter in vivo, obser-vations that have been reported previously. The resultant up-regulation of miR21 caused fibrogenesis in NASH by SMAD2/3-SMAD4 association in the nucleus and SMAD7 repression.Our results will help advance the current knowledge about themechanisms of NADPH oxidase-mediated fibrogenesis inNASH and will stimulate newer studies that can help developtherapeutic approaches for this clinically silent disease.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the technical services of Benny Da-vidson at the Instrumentation Resource Facility (IRF), University of SouthCarolina, School of Medicine; we also thank Dr. James Carson, Department ofExercise Science, and the IRF at the University of South Carolina, School ofMedicine for equipment usage and consulting services.

GRANTS

This work has been supported by NIH pathway to Independence Award(4R00ES019875-02 to S. Chatterjee), NIH R01 (R01DK053792 to A. Diehl),NIH grants (P20GM103641, R01AT006888, and R01ES019313 to M. Nagar-katti), and the Intramural Research Program of the National Institutes of Healthand the National Institute of Environmental Health Sciences.

DISCLOSURES

No conflicts of interest financial or otherwise are declared by the authors.

Fig. 10. Schematic representation of the role of leptin-induced miR21 in modulating TGF-� pathway in NASH fibrogenesis. SBE, SMAD binding element.



AUTHOR CONTRIBUTIONS

Author contributions: D.D., S.P., S.D., F.A., R.K.S., and S.C. performedexperiments; D.D., S.P., S.D., R.K.S., and S.C. analyzed data; D.D., R.K.S.,M.N., and S.C. interpreted results of experiments; D.D. and S.C. preparedfigures; M.N., G.A.M., A.M.D., and S.C. edited and revised manuscript; M.N.,G.A.M., A.M.D., and S.C. approved final version of manuscript; G.A.M. andS.C. drafted manuscript; S.C. conception and design of research.

REFERENCES

1. Aschner Y, Khalifah AP, Briones N, Yamashita C, Dolgonos L, YoungSK, Campbell MN, Riches DW, Redente EF, Janssen WJ, HensonPM, Sap J, Vacaresse N, Kapus A, McCulloch CA, Zemans RL,Downey GP. Protein tyrosine phosphatase alpha mediates profibroticsignaling in lung fibroblasts through TGF-beta responsiveness. Am JPathol 184: 1489–1502, 2014.

2. Bataller R, Schwabe RF, Choi YH, Yang L, Paik YH, Lindquist J,Qian T, Schoonhoven R, Hagedorn CH, Lemasters JJ, Brenner DA.NADPH oxidase signal transduces angiotensin II in hepatic stellate cellsand is critical in hepatic fibrosis. J Clin Invest 112: 1383–1394, 2003.

3. Chatterjee S, Ganini D, Tokar EJ, Kumar A, Das S, Corbett J,Kadiiska M, Waalkes M, Diehl AM, Mason RP. Leptin is key toperoxynitrite-mediated oxidative stress and Kupffer cell activation inexperimental nonalcoholic steatohepatitis. J Hepatol 58: 778–7784, 2013.

4. Cheung O, Sanyal AJ. Role of microRNAs in non-alcoholic steatohepa-titis. Curr Pharm Des 16: 1952–1957, 2010.

5. Choi SS, Syn WK, Karaca GF, Omenetti A, Moylan CA, Witek RP,Agboola KM, Jung Y, Michelotti GA, Diehl AM. Leptin promotes themyofibroblastic phenotype in hepatic stellate cells by activating the hedge-hog pathway. J Biol Chem 285: 36551–36560, 2010.

6. Chung AC, Dong Y, Yang W, Zhong X, Li R, Lan HY. Smad7suppresses renal fibrosis via altering expression of TGF-beta/Smad3-regulated microRNAs. Mol Ther 21: 388–398, 2013.

7. Copaci I, Micu L, Voiculescu M. The role of cytokines in non-alcoholicsteatohepatitis. A review. J Gastrointest Liver Dis 15: 363–373, 2006.

8. De Minicis S, Seki E, Oesterreicher C, Schnabl B, Schwabe RF,Brenner DA. Reduced nicotinamide adenine dinucleotide phosphate ox-idase mediates fibrotic and inflammatory effects of leptin on hepaticstellate cells. Hepatology 48: 2016–2026, 2008.

9. De Minicis S, Svegliati-Baroni G. Fibrogenesis in nonalcoholic steato-hepatitis. Exp Rev Gastroenterol Hepatol 5: 179–187, 2011.

10. Dooley S, ten Dijke P. TGF-beta in progression of liver disease. CellTissue Res 347: 245–256, 2012.

11. Honda H, Ikejima K, Hirose M, Yoshikawa M, Lang T, Enomoto N,Kitamura T, Takei Y, Sato N. Leptin is required for fibrogenic responsesinduced by thioacetamide in the murine liver. Hepatology 36: 12–21,2002.

12. Ikejima K, Okumura K, Kon K, Takei Y, Sato N. Role of adipocyto-kines in hepatic fibrogenesis. J Gastroenterol Hepatol 22, Suppl 1:S87–S92, 2007.

13. Ikejima K, Okumura K, Lang T, Honda H, Abe W, Yamashina S,Enomoto N, Takei Y, Sato N. The role of leptin in progression ofnon-alcoholic fatty liver disease. Hepatol Res 33: 151–154, 2005.

14. Kamada Y, Takehara T, Hayashi N. Adipocytokines and liver disease.J Gastroenterol 43: 811–822, 2008.

15. Kohan M, Muro AF, White ES, Berkman N. EDA-containing cellularfibronectin induces fibroblast differentiation through binding to �4�7integrin receptor and MAPK/Erk 1/2-dependent signaling. FASEB J 24:4503–4512, 2010.

16. Lin Y, Liu X, Cheng Y, Yang J, Huo Y, Zhang C. Involvement ofmicroRNAs in hydrogen peroxide-mediated gene regulation and cellularinjury response in vascular smooth muscle cells. J Biol Chem 284:7903–7913, 2009.

17. Liu G, Friggeri A, Yang Y, Milosevic J, Ding Q, Thannickal VJ,Kaminski N, Abraham E. miR-21 mediates fibrogenic activation ofpulmonary fibroblasts and lung fibrosis. J Exp Med 207: 1589–1597,2010.

18. Liu Y, Liu H, Meyer C, Li J, Nadalin S, Konigsrainer A, Weng H,Dooley S, ten Dijke P. Transforming growth factor-beta (TGF-beta)-mediated connective tissue growth factor (CTGF) expression in hepaticstellate cells requires Stat3 signaling activation. J Biol Chem 288: 30708–30719, 2013.

19. Marra F, Aleffi S, Bertolani C, Petrai I, Vizzutti F. Adipokines andliver fibrosis. Eur Rev Med Pharmacol Sci 9: 279–284, 2005.

20. Medici V, Ali MR, Seo S, Aoki CA, Rossaro L, Kim K, Fuller WD,Vidovszky TJ, Smith W, Jiang JX, Maganti K, Havel PJ, Kamboj A,Ramsamooj R, Torok NJ. Increased soluble leptin receptor levels inmorbidly obese patients with insulin resistance and nonalcoholic fatty liverdisease. Obesity (Silver Spring) 18: 2268–2273, 2010.

21. Michelotti GA, Machado MV, Diehl AM. NAFLD, NASH and livercancer. Nat Rev Gastroenterol Hepatol 10: 656–665, 2013.

22. Nakao A, Afrakhte M, Moren A, Nakayama T, Christian JL, HeuchelR, Itoh S, Kawabata M, Heldin NE, Heldin CH, ten Dijke P. Identi-fication of Smad7, a TGFbeta-inducible antagonist of TGF-beta signalling.Nature 389: 631–635, 1997.

23. Nakao A, Imamura T, Souchelnytskyi S, Kawabata M, Ishisaki A,Oeda E, Tamaki K, Hanai J, Heldin CH, Miyazono K, ten Dijke P.TGF-beta receptor-mediated signalling through Smad2, Smad3 andSmad4. EMBO J 16: 5353–5362, 1997.

24. Nakao A, Roijer E, Imamura T, Souchelnytskyi S, Stenman G, HeldinCH, ten Dijke P. Identification of Smad2, a human Mad-related protein inthe transforming growth factor beta signaling pathway. J Biol Chem 272:2896–2900, 1997.

25. Noetel A, Kwiecinski M, Elfimova N, Huang J, Odenthal M. mi-croRNA are Central Players in Anti- and Profibrotic Gene Regulationduring Liver Fibrosis. Front Physiol 3: 49, 2012.

26. Paik YH, Brenner DA. NADPH oxidase mediated oxidative stress inhepatic fibrogenesis. Kor J Hepatol 17: 251–257, 2011.

27. Paik YH, Kim J, Aoyama T, De Minicis S, Bataller R, Brenner DA.Role of NADPH oxidases in liver fibrosis. Antioxid Redox Signal 20:2854–2872, 2014.

28. Patel V, Noureddine L. MicroRNAs and fibrosis. Curr Opin NephrolHypertens 21: 410–416, 2012.

29. Pirola CJ, Fernandez Gianotti T, Castano GO, Mallardi P, SanMartino J, Mora Gonzalez Lopez Ledesma M, Flichman D, MirshahiF, Sanyal AJ, Sookoian S. Circulating microRNA signature in non-alcoholic fatty liver disease: from serum non-coding RNAs to liverhistology and disease pathogenesis. Gut. In press.

30. Procaccini C, Galgani M, De Rosa V, Carbone F, La Rocca C,Ranucci G, Iorio R, Matarese G. Leptin: the prototypic adipocytokineand its role in NAFLD. Curr Pharm Des 16: 1902–1912, 2010.

31. Quinn SR, O’Neill LA. A trio of microRNAs that control Toll-likereceptor signalling. Int Immunol 23: 421–425, 2011.

32. Ruan Q, Wang T, Kameswaran V, Wei Q, Johnson DS, MatschinskyF, Shi W, Chen YH. The microRNA-21-PDCD4 axis prevents type 1diabetes by blocking pancreatic beta cell death. Proc Natl Acad Sci USA108: 12030–12035, 2011.

33. Saxena NK, Saliba G, Floyd JJ, Anania FA. Leptin induces increasedalpha2(I) collagen gene expression in cultured rat hepatic stellate cells. JCell Biochem 89: 311–320, 2003.

34. Schuppan D, Schattenberg JM. Non-alcoholic steatohepatitis: pathogen-esis and novel therapeutic approaches. J Gastroenterol Hepatol 28, Suppl1: 68–76, 2013.

35. Seth RK, Das S, Kumar A, Chanda A, Kadiiska MB, Michelotti G,Manautou J, Diehl AM, Chatterjee S. CYP2E1-dependent and leptin-mediated hepatic CD57 expression on CD8� T cells aid progression ofenvironment-linked nonalcoholic steatohepatitis. Toxicol Appl Pharmacol274: 42–54, 2014.

36. Seth RK, Kumar A, Das S, Kadiiska MB, Michelotti G, Diehl AM,Chatterjee S. Environmental toxin-linked nonalcoholic steatohepatitisand hepatic metabolic reprogramming in obese mice. Toxicol Sci 134:291–303, 2013.

37. Sheedy FJ, Palsson-McDermott E, Hennessy EJ, Martin C, O’LearyJJ, Ruan Q, Johnson DS, Chen Y, O’Neill LA. Negative regulation ofTLR4 via targeting of the proinflammatory tumor suppressor PDCD4 bythe microRNA miR-21. Nat Immunol 11: 141–147, 2010.

38. Singh S, Allen AM, Wang Z, Prokop LJ, Murad MH, Loomba R.Fibrosis progression in nonalcoholic fatty liver vs nonalcoholic steato-hepatitis: a systematic review and meta-analysis of paired-biopsy studies.Clin Gastroenterol Hepatol. In press.

39. Tang M, Potter JJ, Mezey E. Leptin enhances the effect of transforminggrowth factor beta in increasing type I collagen formation. BiochemBiophys Res Commun 297: 906–911, 2002.

40. Wang J, Leclercq I, Brymora JM, Xu N, Ramezani-Moghadam M,London RM, Brigstock D, George J. Kupffer cells mediate leptin-induced liver fibrosis. Gastroenterology 137: 713–723, 2009.

41. Yang L, Roh YS, Song J, Zhang B, Liu C, Loomba R, Seki E.Transforming growth factor beta signaling in hepatocytes participates in



steatohepatitis through regulation of cell death and lipid metabolism inmice. Hepatology 59: 483–495, 2014.

42. Yao H, Yang SR, Kode A, Rajendrasozhan S, Caito S, Adenuga D,Henry R, Edirisinghe I, Rahman I. Redox regulation of lung inflam-mation: role of NADPH oxidase and NF-kappaB signalling. Biochem SocTrans 35: 1151–1155, 2007.

43. Yilmaz Y, Younossi ZM. Obesity-associated nonalcoholic fatty liverdisease. Clin Liver Dis 18: 19–31, 2014.

44. Zhang Z, Gao Z, Hu W, Yin S, Wang C, Zang Y, Chen J, Zhang J,Dong L. 3,3=-Diindolylmethane ameliorates experimental hepatic fi-brosis via inhibiting miR-21 expression. Br J Pharmacol 170: 649 –660, 2013.



Micro-RNA 21 inhibition of SMAD7 enhances fibrogenesis via leptin-mediated NADPH oxidase in experimental and human nonalcoholic steatohepatitis

Documents