The nicotinamide adenine dinucleotide phosphate oxidase (NOX) homologues NOX1 and NOX2/gp91phox mediate hepatic fibrosis in mice

Authors


  • Potential conflict of interest: Nothing to report.

  • Supported by grants from the Korea Research Foundation (KRF-2009 E00006) and the Good Health R&D Project, Ministry of Health, Welfare, and Family Affairs, Republic of Korea (A050021). Y. H. P. received a 2010 faculty research grant from Yonsei University College of Medicine (6-2010-0168). E. S. received a Liver Scholar Award from the American Association for the Study of Liver Diseases/American Liver Foundation and a research grant from ABMRF. D. A. B. received National Institutes of Health Grants NIH RO1 DK072237-06, NIH RO1 GM041804-23, and NIH P50 AA011999-11.

Abstract

Nicotinamide adenine dinucleotide phosphate oxidase (NOX) is a multicomponent enzyme that mediates electron transfer from nicotinamide adenine dinucleotide phosphate to molecular oxygen, which leads to the production of superoxide. NOX2/gp91phox is a catalytic subunit of NOX expressed in phagocytic cells. Several homologues of NOX2, including NOX1, have been identified in nonphagocytic cells. We investigated the contributory role of NOX1 and NOX2 in hepatic fibrosis. Hepatic fibrosis was induced in wild-type (WT) mice, NOX1 knockout (NOX1KO) mice, and NOX2 knockout (NOX2KO) mice by way of either carbon tetrachloride (CCl4) injection or bile duct ligation (BDL). The functional contribution of NOX1 and NOX2 in endogenous liver cells, including hepatic stellate cells (HSCs), and bone marrow (BM)-derived cells, including Kupffer cells (KCs), to hepatic reactive oxygen species (ROS) generation and hepatic fibrosis was assessed in vitro and in vivo using NOX1 or NOX2 BM chimeric mice. Hepatic NOX1 and NOX2 messenger RNA expression was increased in the two experimental mouse models of hepatic fibrosis. Whereas NOX1 was expressed in HSCs but not in KCs, NOX2 was expressed in both HSCs and KCs. Hepatic fibrosis and ROS generation were attenuated in both NOX1KO and NOX2KO mice after CCl4 or BDL. Liver fibrosis in chimeric mice indicated that NOX1 mediates the profibrogenic effects in endogenous liver cells, whereas NOX2 mediates the profibrogenic effects in both endogenous liver cells and BM-derived cells. Multiple NOX1 and NOX2 components were up-regulated in activated HSCs. Both NOX1- and NOX2-deficient HSCs had decreased ROS generation and failed to up-regulate collagen α1(I) and transforming growth factor β in response to angiotensin II. Conclusion: Both NOX1 and NOX2 have an important role in hepatic fibrosis in endogenous liver cells, including HSCs, whereas NOX2 has a lesser role in BM-derived cells. (HEPATOLOGY 2011;)

Reactive oxygen species (ROS) function as key secondary messengers in numerous signaling pathways, including transcriptional regulation, differentiation, carcinogenesis, and apoptosis.1 ROS contributes to hepatic fibrosis caused by various injuries, including alcohol abuse, hepatitis C virus infection, iron overload, and chronic cholestasis.2 Activated hepatic stellate cells (HSCs) play a key role in the pathogenesis of hepatic fibrosis by producing extracellular matrix proteins, including type I collagen.3 Free radicals or lipid peroxidation products stimulate the activation of HSCs by inducing c-myb and nuclear factor κB, which are blocked by antioxidants such as α-tocopherol.4 ROS induce activation of the collagen type I gene in HSCs5 and may act as an intracellular signaling mediator of the fibrogenic action of transforming growth factor β (TGF-β).6, 7 Hepatic ROS may be generated by multiple sources, including mitochondrial respiratory chain, cytochrome p4502E1, peroxisomes, and nicotinamide adenine dinucleotide phosphate oxidases (NOXs).2

NOX is a multicomponent enzyme complex originally described in phagocytes.8 The phagocytic NOX consists of the catalytic subunit NOX2 (also known as gp91phox) together with the regulatory subunit p22phox located in the membrane. Other regulatory components p47phox, p40phox, and p67phox and the small guanosine triphosphatase Rac are normally located in the cytoplasm. Upon stimulation with agonists, the cytoplasmic subunits translocate to the membrane-bound complex leading to enzymatic activity.9 Humans have six additional NOX homologues (NOX1, NOX3, NOX4, NOX5, DUOX1, and DUOX2) that may function in nonphagocytic NOX.10 All NOX enzymes are able to catalyze the reduction of molecular oxygen to superoxide, but there are key differences in their activation, subunit composition, localization, and expression. Among the seven members of the NOX family, NOX1 is structurally and functionally similar to NOX2. Although NOX1 is also p22phox-dependent, NOX1 may use NOXO1 (homologue of p47phox) to organize the enzyme assembly and NOXA1 (homologue of p67phox) for enzyme activation.11 NOX1 is abundantly expressed in the colon, but is also detected in the uterus, prostate, stomach, and vascular smooth muscle cells.12

We have demonstrated that HSCs express NOX components, including NOX2 and p47phox.13 Angiotensin II (Ang II) stimulates DNA synthesis, cell migration, procollagen α1(I) messenger RNA (mRNA) expression, and secretion of TGF-β1 and inflammatory cytokines in cultured HSCs. Ang II activates AKT as well as mitogen-activated protein kinase components extracellular signal-regulated kinase, c-Jun N-terminal kinase, and p38 and increases activator protein 1 DNA binding in HSCs. All of these effects were attenuated by antioxidant (N-acetylcysteine) and by diphenylene iodonium, a NOX inhibitor.13 Ang II or leptin increases intracellular ROS and fibrogenic responses in HSCs isolated from wild-type (WT) mice, but not in p47phox-deficient HSCs.13, 14 Hepatic injury and fibrosis is also attenuated in p47phox-deficient mice after bile duct ligation (BDL), demonstrating that NOX is an important mediator in the development of hepatic fibrosis.13

Although NOX plays an important role in HSC activation and hepatic fibrosis, how various NOX homologues expressed in different hepatic cell types contribute to hepatic fibrogenesis is unknown. The purpose of the present study was to investigate the contributory role of NOX1 and NOX2 in hepatic fibrosis. We evaluated the effect of genetic NOX1 and NOX2 inactivation on hepatic fibrosis in two different models of experimental fibrogenesis. We also identified NOX1- and NOX2-expressing cell types in the liver. The functional contribution of NOX1 and NOX2 in endogenous liver cells, including HSCs, and in bone marrow (BM)-derived cells, including Kupffer cells (KCs), to hepatic fibrosis was assessed using either NOX1 or NOX2 BM chimeric mice. Our data indicate that both NOX1 and NOX2 have an important role in hepatic fibrosis in endogenous liver cells, whereas NOX2 has a lesser role in BM-derived cells. These findings may provide new insight into the development of antifibrotic therapy targeting nonphagocytic NOX signaling in the liver without suppression of NOX2-mediated host defense mechanism.

Abbreviations

α-SMA, α-smooth muscle actin; ALT, alanine aminotransferase; Ang II, angiotensin II; BDL, bile duct ligation; BM, bone marrow; BMT, bone marrow transplantation; CCl4, carbon tetrachloride; CM-H2DCFDA, 2′7′-dichlorofluorescein diacetate; DMEM, Dulbecco's modified Eagle's medium; HSC, hepatic stellate cell; KC, Kupffer cell; KO, knockout; mRNA, messenger RNA; NOX, nicotinamide adenine dinucleotide phosphate oxidase; PCR, polymerase chain reaction; ROS, reactive oxygen species; SEC, sinusoid endothelial cell; TGF-β, transforming growth factor β; WT, wild-type.

Materials and Methods

Fibrosis Induction in Mice.

NOX2 knockout (NOX2KO) mice15 with a C57BL/6 background, which lack a catalytic subunit of phagocytic NOX complex, and WT C57BL/6 control mice were purchased from The Jackson Laboratory (Bar Harbor, MA). NOX1 knockout (NOX1KO) mice16 with a C57BL/6 background were a kind gift from John Engelhardt of the University of Iowa. Eight- to 10-week-old male mice were used. Liver fibrosis was induced either by way of intraperitoneal injection of hepatotoxin carbon tetrachloride (CCl4) or by way of BDL. CCl4 (Sigma Aldrich, St. Louis, MO; diluted 1:3 in corn oil) or a vehicle (corn oil) was injected on every third day at a concentration of 2 μL/g body weight 12 times per day. BDL was performed as described, and a sham operation group was used as a control.17 Animals were sacrificed 72 hours after the last CCl4 injection or 3 weeks after BDL and blood and liver samples were collected. All animal studies were approved by The University of California, San Diego Institutional Animal Care and Use Committee (protocol number: S-07088).

BM Transplantation and Chimeric Mice Experiments.

We performed BM transplantation (BMT) experiments as described with slight modifications.18-21 We flushed the tibias and femurs of donor mice to obtain BM. We washed BM cells twice in HBSS and injected 1 × 107 BM cells into the tail veins of lethally irradiated (10 Gy) recipient mice. Mice received an intraperitoneal injection of liposomal clodronate (150 μL intraperitoneally) 2 weeks after irradiation to deplete Kupffer cells. At 8 weeks after BMT, CCl4 was injected on every third day at a concentration of 2 μL/g body weight diluted in corn oil (1:3) (Sigma Aldrich). Animals were sacrificed after 10 intraperitoneal CCl4 injections, and blood and liver samples were collected. Using this protocol, we achieved full replacement of KCs but not HSCs with BM-derived cells in mice transplanted with β-actin promoter-driven green fluorescent protein transgenic BM, as assessed by way of double immunostaining.19 All animal studies were approved by the University of California, San Diego Institutional Animal Care and Use Committee (protocol number: S-07088).

Assessment of Hepatic Fibrosis.

Hepatic fibrosis was assessed by way of morphometric analysis of the sirius red–stained area and measurement of hepatic hydroxyproline content, as described.13, 22 Details are given in the Supporting Information.

Thiobarbituric Acid Reactive Substances Determination.

Hepatic lipid peroxidation was assessed by way of thiobarbituric acid reactive substances formation.23 Details are given in the Supporting Information.

Isolation of Liver Cell Fractions.

Liver cells of WT mice were fractionated into four major cell populations (hepatocytes, KCs, sinusoid endothelial cells [SECs], and HSCs) as described.24 Details are given in the Supporting Information.

HSC and KC Isolation and Cell Culture.

Mouse HSCs were isolated using a two-step collagenase–pronase perfusion of mouse livers followed by 8.2% Nycodenz (Accurate Chemical and Scientific Corp.) two-layer discontinuous density gradient centrifugation as described.13In vivo–activated murine HSCs were isolated from mice that underwent BDL for 3 weeks by way of the same procedure. After isolation, HSCs were seeded on uncoated plastic tissue culture dishes and cultured in Dulbecco's modified Eagle's medium (DMEM) (GIBCO BRL; Life Technologies Inc., Grand Island, NY) supplemented with 10% fetal bovine serum. Mouse KCs were isolated by collagenase-pronase perfusion followed by 15% Nycodenz gradient centrifugation and subsequent positive selection of CD11b-expressing cells by way of magnetic cell sorting, as described.19 Isolated KCs were cultured in DMEM supplemented with 1% fetal bovine serum.

Measurement of ROS Generation in HSCs.

HSCs were isolated from WT mice, NOX1KO mice, and NOX2KO mice and cultured in 96-well black plates with a transparent bottom in phenol red-free DMEM containing 10% fetal bovine serum and antibiotics for 5 days. Cells were then changed to a serum-free media for 24 hours and subsequently loaded with the redox-sensitive dye 2′7′-dichlorofluorescein diacetate (CM-H2DCFDA) (10 μM) diluted in Hank's balanced salt solution for 20 minutes at 37°C. Cells were then rinsed twice with DMEM without phenol red and stimulated with Ang II (10−6 M). CM-H2DCFDA fluorescence was detected at excitation and emission wavelengths of 488 nm and 520 nm.13 ROS formation was measured in a time course of 30 minutes using a multiwell fluorescence scanner (Fluostar Optima; BMG Labtech, Cary, NC). Microscopic detection of ROS generation in HSCs is described in the Supporting Information.

Other Materials and Methods.

Other materials and methods for reagents, serum biochemistry, immunohistochemistry and immunofluorescence, western blotting, quantitative real-time polymerase chain reaction (PCR), and assessment of superoxide production in KCs are described in the Supporting Information.

Statistical Analysis.

Results were expressed as the mean ± SEM. Data between groups were analyzed by a two-tailed Student t test. P < 0.05 was considered statistically significant.

Results

Expression of NOX1 and NOX2 Are Up-regulated in Fibrotic Liver.

We first investigated NOX1 and NOX2 expression in normal and fibrotic liver. Expression of hepatic NOX1 and NOX2 mRNA levels were elevated in the fibrotic liver after CCl4 treatment (Fig. 1A) and BDL (Fig. 1B). We also detected NOX1 and NOX2 protein expression in the fibrotic liver by way of immunofluorescence (Supporting Fig. 1A,B). We further investigated NOX1 and NOX2 expression in HSCs, KCs, and SECs, key cell types in hepatic fibrosis. We performed double immunofluorescence staining between either NOX1 or NOX2 and cell type markers α-smooth muscle actin (α-SMA), F4/80, and CD31 for HSCs, KCs, and SECs, respectively. NOX1 expression in the fibrotic liver overlapped with α-SMA–positive HSCs and some CD31-positive SECs, but there was minimal overlap with F4/80-positive KCs (Fig. 1C). NOX2 staining overlapped with both α-SMA–positive HSCs and F4/80-positive KCs/macrophages in CCl4-treated liver (Fig. 1D).

Figure 1.

Expression of NOX1 and NOX2 are up-regulated in hepatic fibrosis. Livers were obtained from WT mice after 12 intraperitoneal injections of CCl4 (n = 8) or 3 weeks of BDL treatment (n = 10). (A, B) Hepatic mRNA levels of NOX1 and NOX2 were measured after treatment with (A) CCl4 or (B) BDL using quantitative PCR. *P < 0.05. **P < 0.01. (C) Double immunofluorescence staining for NOX1 (red fluorescence) and α-SMA (HSC), F4/80 (KC), or CD31 (SEC) (all green fluorescence). Original magnification ×200. (D) Double immunofluorescence staining for NOX2 (red fluorescence) and α-SMA (HSC), F4/80 (KC), or CD31 (SEC) (all green fluorescence). The liver sections were from CCl4-treated mice. Arrowheads indicate double-positive cells. Data are representative of three independent experiments. Original magnification ×200.

Attenuated Hepatic Fibrosis in NOX1KO and NOX2KO Mice Compared with WT Mice After CCl4 or BDL Treatment.

To investigate the contributing roles of NOX1 and NOX2 in hepatic fibrogenesis, we induced liver fibrosis using two different methods, CCl4 injection and BDL, in NOX1KO, NOX2KO, and WT mice. In the CCl4 treatment group, serum alanine aminotransferase (ALT) levels and the liver/body weight ratio were significantly lower in NOX1KO and NOX2KO mice compared with WT mice (Supporting Fig. 2A). In the BDL experiment group, there was no significant difference in serum ALT levels or total bilirubin levels between the three groups, but the liver/body weight ratio was lower in NOX1KO mice compared with WT mice (Supporting Fig. 2B). We assessed hepatic fibrosis by way of morphometric analysis quantitating the sirius red–stained area in the liver and through measurement of hepatic hydroxyproline content. Hepatic fibrosis was significantly attenuated in NOX1KO and NOX2KO mice compared with WT mice after CCl4 injections (Fig. 2A,B). BDL-induced hepatic fibrosis was also reduced in NOX1KO and NOX2KO mice compared with WT mice (Fig. 2A,C). These results demonstrate that both NOX1 and NOX2 contribute to hepatic fibrosis induced by CCl4 and BDL in mice.

Figure 2.

Hepatic fibrosis is attenuated in NOX1KO and NOX2KO mice after CCl4 treatment or BDL treatment. Livers were obtained from WT mice after 12 intraperitoneal injections of CCl4 (n = 8) or 3 weeks of BDL treatment (n = 10). (A) Representative images of sirius red staining in WT, NOX1KO, and NOX2KO mice after 12 CCl4 injections or 3 weeks of BDL are shown. Original magnification ×100. (B, C) Hepatic fibrosis was assessed by morphometric analysis of the sirius red–stained area in the liver or by the measurement of hepatic hydroxyproline content in CCl4-treated mice (B) or BDL-treated mice (C). *P < 0.05. **P < 0.01.

NOX1KO and NOX2KO Mice Have Reduced HSC Activation and Fibrosis-Related Gene Activation After CCl4 or BDL Treatment.

We next investigated the activation of HSCs in NOX1KO, NOX2KO, and WT mice after treatment with CCl4 or BDL. We assessed α-SMA expression in the liver by way of morphometric analysis and immunoblotting. The fibrotic livers of WT mice showed high levels of hepatic expression of α-SMA, but α-SMA expression was significantly low in NOX1KO and NOX2KO mice (Fig. 3A-C). In addition, the hepatic mRNA levels of fibrosis-related genes, including collagen α1(I), α-SMA, and TGF-β1, were low in NOX1KO and NOX2KO mice compared with WT mice after CCl4 treatment (Fig. 3D) or BDL (Fig. 3E). There was no difference in hepatic expression of M1 or M2 macrophage markers between WT, NOX1KO, or NOX2KO mice (Supporting information Fig. 3A,B). These results suggest that both NOX1 and NOX2 may be directly involved in the activation of HSCs.

Figure 3.

Activation of HSCs and expression of profibrogenic genes are decreased in NOX1KO and NOX2KO mice after CCl4 treatment or BDL. (A) Representative images of α-SMA immunohistochemical staining in WT, NOX1KO, and NOX2KO mice after 12 intraperitoneal injections of CCl4 (n = 8) or 3 weeks of BDL treatment (n = 10). Original magnification ×50. (B, C) Activation of HSCs was assessed by way of morphometric measurement of α-SMA stained area in the liver (B) or immunoblotting for α-SMA (C). (D, E) Hepatic mRNA levels of collagen α1(I) [Colα1(I)], α-SMA, TGF-β, and tissue inhibitor of metalloproteinase 1 (TIMP1) were measured in WT, NOX1KO, and NOX2KO mice after (D) 12 CCl4 injections (n = 8) or (E) 3 weeks of BDL treatment (n = 10) by way of quantitative real-time PCR. *P < 0.05. **P < 0.01.

NOX1KO and NOX2KO Mice Have Reduced Lipid Peroxidation Compared with WT Mice After CCl4 or BDL Treatment.

We measured the lipid peroxidation products 4-hydroxynonenal and malondialdehyde as indicators of oxidative stress in the liver in NOX1KO, NOX2KO, and WT mice after CCl4 or BDL treatment. Immunofluorescence staining showed lower hepatic 4-hydroxynonenal levels in NOX1KO and NOX2KO mice compared with WT mice after CCl4 or BDL treatment (Fig. 4A). Measurement of malondialdehyde using thiobarbituric acid–reactive substances showed that NOX1KO and NOX2KO mice have lower levels of lipid peroxidation compared with WT mice after CCl4 or BDL treatment (Fig. 4B,C), suggesting that both NOX1 and NOX2 play an important role in the generation of hepatic oxidative stress in response to CCl4 or BDL in mice.

Figure 4.

Lipid peroxidation is reduced in NOX1KO and NOX2KO mice compared with WT mice after CCl4 treatment or BDL. (A) Representative images of 4-hydroxynonenal (4-HNE) immunofluorescent staining in WT, NOX1KO, and NOX2KO mice after 12 intraperitoneal injections of CCl4 (n = 8) or 3 weeks of BDL treatment (n = 10). Original magnification ×200. (B) Hepatic malondialdehyde levels were measured using thiobarbituric acid reactive substances (TBARS) assay in WT, NOX1KO, and NOX2KO mice after 12 CCl4 injections (n = 8) or 3 weeks of BDL (n = 10). *P < 0.05. **P < 0.01.

NOX1 and NOX2 in Endogenous Liver Cells Mediate Hepatic Fibrosis.

To characterize the contributory roles of NOX1 and NOX2 in hepatic fibrosis in different liver cell populations, we generated NOX1 and NOX2 BM chimeric mice using a combination of lethal irradiation, KC depletion by way of clodronate injection, and BMT. This combination generates complete substitution of KCs and other BM-derived cells, but not of resident hepatic cell populations, including HSCs and SECs.18, 19 Eight weeks after BMT, hepatic fibrosis was induced by way of CCl4 treatment for 1 month. Serum ALT levels were lower in NOX1 chimeric mice with NOX1-deficient endogenous liver cells (WT BM→NOX1KO and NOX1KO BM→NOX1KO) compared with WT mice transplanted with WT BM (Fig. 5B). As expected, NOX1KO mice transplanted with NOX1KO BM had reduced hepatic fibrosis compared with WT mice transplanted with WT BM. NOX1 chimeric mice that express NOX1 in BM-derived cells but not endogenous liver cells (WT BM→NOX1KO) showed the similar reduction of hepatic fibrosis as mice with complete NOX1 deficiency. However, NOX1 chimeric mice that expressed NOX1 in endogenous liver cells but not BM-derived cells (NOX1KO BM→WT) showed the same levels of fibrosis as WT mice (Fig. 5A,C).

Figure 5.

Chimeric mice with NOX1- or NOX2-deficient endogenous liver cells show decreased hepatic fibrosis after CCl4 treatment. After lethal irradiation, BMT was performed to generate chimeric mice as follows: WT mice with WT BM, WT mice with NOX1KO or NOX2KO BM, NOX1KO mice with WT or NOX1KO BM, and NOX2KO mice with WT or NOX2KO BM. Liposomal clodronate was injected intraperitoneally to deplete resident KCs after 2 weeks of BMT. Mice underwent 10 intraperitoneal injections of CCl4 8 weeks after BMT. (A, D) Representative images of sirius red staining are shown. Original magnification ×100. (B, E) Liver injury was assessed through measurement of serum ALT levels. (C, F) Hepatic fibrosis was assessed through measurement of hepatic hydroxyproline content. *P < 0.05. **P < 0.01.

Serum ALT levels were reduced in NOX2 chimeric mice with NOX2-deficient endogenous liver cells (WT BM→NOX2KO and NOX2KO BM→NOX2KO) compared with WT mice transplanted with WT BM (Fig. 5E). NOX2KO mice transplanted with NOX2-deficient BM had reduced hepatic fibrosis compared with WT mice transplanted with WT BM. NOX2 chimeric mice that expressed NOX2 in BM-derived cells but not endogenous liver cells (WT BM→NOX2KO) showed a reduction in hepatic fibrosis similar to those with complete NOX2 deficiency. NOX2 chimeric mice that expressed NOX2 in endogenous liver cells but not BM-derived cells (NOX2KO BM→WT) showed a modest reduction in fibrosis compared with WT mice (Fig. 5D,F). We confirmed that NOX2 BM chimeric mice harboring NOX2KO BM-derived macrophages (NOX2KO BM→WT and NOX2KO BM→NOX2KO) expressed no NOX2 in F4/80-positive cells in the liver (Supporting information Fig. 4). Taken together, the results of the chimeric mouse experiments suggest that both NOX1 and NOX2 mediate hepatic fibrosis in endogenous liver cells, including HSCs, whereas NOX2 has a lesser role in hepatic fibrosis in BM-derived cells, including KCs/macrophages.

Both Phagocytic and Nonphagocytic NOX Components Are Up-regulated in Activated HSCs Compared with Quiescent HSCs.

To investigate the expression of NOX components in different liver cell populations, we assessed the mRNA levels of NOX components in the four major liver cell fractions (hepatocytes, KCs, SECs, and HSCs) from WT mice. The phagocytic NOX components, including NOX2, p47phox, and p67phox are mainly expressed in KCs. The nonphagocytic NOX components such as NOX1, NOXO1, and NOXA1 are mainly expressed in HSCs and SECs. The mRNA expression of NOX4, another nonphagocytic NOX, was observed in hepatocytes, SECs, and HSCs (Fig. 6A). Next, we investigated the expression of NOX components in quiescent and activated HSCs. mRNAs of the phagocytic NOX catalytic subunit NOX2 and nonphagocytic NOX catalytic subunits NOX1 and NOX4 were up-regulated in in vitro and in vivo (BDL)-activated HSCs compared with quiescent HSCs. Other NOX components, including p40phox, p47phox, p67phox, NOXO1, NOXA1, and Rac1, were also up-regulated in activated HSCs (Fig. 6B). We found that NOX2 and its regulators, including p40phox, p47phox, and p67phox, were strongly up-regulated in in vivo (CCl4)-activated HSCs compared with quiescent HSCs (Supporting Fig. 5). We confirmed that NOX1 and NOX2 proteins were expressed in the activated human HSC line LX-2 (Supporting Fig. 6A,B). These findings provide further evidence that nonphagocytic NOX, including NOX1, as well as phagocytic NOX2 may be involved in hepatic fibrogenesis.

Figure 6.

Expression of NOX components in isolated liver cell fractions. mRNA levels of NOX1, NOX2, NOX4, p22phox, p40phox, p47phox, p67phox, NOXO1, NOXA1, and Rac1 were evaluated by way of quantitative real-time PCR in (A) isolated liver cell fractions from WT mice including hepatocytes, KCs, SECs, and HSCs and in (B) quiescent (1 day in culture) HSCs, in vitro–activated (7 days in culture) HSCs, and in vivo–activated (isolated from WT mice after 3 weeks of BDL treatment) HSCs. Ribosomal 18S was used as an internal control. Data were from two independent experiments performed in triplicate. *P < 0.05. **P < 0.01.

NOX1 and NOX2 Mediate Ang II–Induced ROS Production and Fibrogenic Responses in HSCs.

To identify the NOX components required for ROS generation in HSCs, we assessed ROS generation in HSCs from WT, NOX1KO, and NOX2KO mice. We quantitated the ROS generation in CM-H2DCFDA–loaded HSCs after treatment with Ang II, a known NOX agonist. Cells treated with buffer showed a 3%-4% increase, representing basal ROS production. Ang II induced an 18%-20% increase in ROS production in WT HSCs, a 12%-13% increase in NOX2KO HSCs, and only a 7%-8% increase in NOX1KO HSCs (Fig. 7A). Ang II (10−6 M) treatment induced strong fluorescent signals in both diffuse and dot patterns in the cytoplasm, followed by cell contraction in WT HSCs, weak signals only in the dot pattern in NOX2KO HSCs, and almost no detectable signal in NOX1KO HSCs (Supporting information Fig. 7). These data suggest that both NOX1 and NOX2 contribute to Ang II–induced ROS generation in HSCs, and NOX1 contributes more than NOX2. As a positive control, we also measured superoxide production in isolated KCs from WT, NOX1KO, and NOX2KO mice. After stimulation of phorbol 12-myristate 13-acetate, a NOX activator, WT and NOX1KO KCs produced comparable amounts of superoxide. As expected, NOX2KO KCs failed to produce superoxide (Supporting Fig. 8A,B). These data corroborate the finding that KCs express NOX2 but not NOX1.

Figure 7.

Both NOX1 and NOX2 mediate ROS production and fibrogenic responses in HSCs. (A) HSCs from WT, NOX1KO, and NOX2KO mice were loaded with redox-sensitive dye CM-H2DCFDA (10 μM) for 20 minutes. Cells were then washed twice and subsequently stimulated with Ang II (10−6 M). Fluorescent signals were quantified continuously for 30 minutes using a fluorometer. (B) mRNA expression of collagen α1(I) [Colα1(I)], TGF-β, tissue inhibitor of metalloproteinase 1 (TIMP1), and α-SMA were measured in WT, NOX1KO, and NOX2KO mice 24 hours after Ang II (10−6 M) stimulation by quantitative PCR. Data were from two independent experiments performed in triplicate. *P < 0.05. **P < 0.01.

Finally, we tested whether NOX1 and NOX2 are involved in fibrogenic responses in HSCs. We measured expression of fibrogenic genes [collagen α1(I), TGF-β, tissue inhibitor of metalloproteinase 1, α-SMA] in response to Ang II in HSCs that were isolated from WT, NOX1KO, and NOX2KO mice (Fig. 7B). Ang II induced the up-regulation of these fibrogenic genes except α-SMA in WT HSCs. In contrast, the expression of these fibrogenic genes were not elevated in NOX1KO and NOX2KO HSCs after Ang II stimulation, indicating both NOX1 and NOX2 mediate fibrogenic responses in response to Ang II in HSCs.

Discussion

Several reports have documented that NOX is important in the pathogenesis of hepatic fibrosis.13, 25-27 We previously demonstrated that human HSCs express mRNA for phagocytic NOX components, including NOX2 and p47phox.13 NOXs in HSCs are functionally active in ROS generation in response to agonists such as Ang II, platelet-derived growth factor, and leptin.13, 14, 28 HSCs from p47phox-deficient mice fail to generate ROS in response to Ang II or leptin, and p47phox-deficient mice show decreased hepatic fibrosis, demonstrating that NOXs play an important role in hepatic fibrosis.13, 14, 26 Recently, it was reported that NOX2-deficient mice have reduced hepatic fibrosis after CCl4 treatment.25, 27 However, the contributory role of NOX homologues in different cell types in the liver in the development of hepatic fibrosis is not understood.

Our current study provides compelling evidence that both NOX1 and NOX2 have an important role in hepatic fibrogenesis. Mice deficient for either NOX1 or NOX2 displayed a significant reduction of hepatic fibrosis in two different models of liver injury: CCl4 and BDL. We found that both NOX1 and NOX2 were up-regulated in the fibrotic liver. Through double immunofluorescent staining, we demonstrated that NOX1 is expressed in HSCs and SECs, whereas NOX2 is expressed in HSCs and KCs in the fibrotic liver. Interestingly, NOX1 is expressed in almost all α-SMA–expressing HSCs, but NOX2 is expressed in some HSCs in the fibrotic liver. Recently, Jiang et al.27 reported that phagocytosis of apoptotic hepatocytes directly induced HSC activation and collagen production by NOX2. Perhaps NOX2 expression in HSCs reflects the phagocytic function of HSCs.29, 30 In response to Ang II, we observed minimal ROS generation in NOX1KO HSCs, whereas NOX2KO HSCs generated a decreased but detectable ROS. These data suggest that NOX1 may be a major NOX form in HSCs, and NOX2 may act in specific circumstances such as apoptotic body-induced HSC activation.

The degree of fibrosis reduction in NOX1KO and NOX2KO mice was less than that observed in p47phox-deficient mice after BDL.13, 26 NOX organizer 1 (NOXO1) is a p47phox homologue in the NOX1 complex.11 The NOX organizers (p47phox and NOXO1) share common structures and functional properties, except that NOXO1 lacks a sequence homologous to the C-terminal auto-inhibitory domain of p47phox.9 However, activities of NOX1 as well as NOX2 can be regulated by p47phox in some cell types.31 Studies in vascular smooth muscle cells from normal and p47phox-deficient mice suggest that p47phox participates in an oxidative response that involves NOX1 as the core catalytic oxidase component in these cells.32, 33 Moreover, coexpression of NOX1 with NOXO1 and NOXA1 leads to stimulus-independent, high-level superoxide generation, whereas stimulus dependence of NOX1 was restored when p47phox was used to replace its homologue NOXO1.11 Thus, p47phox appears to involve a functional partnership with both NOX2 and NOX1 in the liver, resulting in hepatic ROS generation and fibrosis.

Compared with WT mice, NOX1KO and NOX2KO mice showed weak hepatic fibrosis after both CCl4 and BDL treatments. However, low serum ALT levels were only observed in CCl4-treated NOX1KO and NOX2KO mice, but not in those treated with BDL. NOX1KO and NOX2KO mice showed low hepatic lipid peroxidation after both CCl4 and BDL treatments. Similar to liver injury, lipid peroxidation in NOX1KO and NOX2KO mice was more evidently reduced after CCl4 treatment than after BDL treatment. We found strong up-regulation of NOX2 and its regulators such as p40phox, p47phox, p67phox in in vivo–activated HSCs by CCl4 compared with quiescent HSCs, suggesting a stronger participation of NOX in CCl4-induced liver fibrosis. Hydrophobic bile acids that accumulate during cholestasis stimulate the generation of ROS in hepatocyte mitochondria through induction of mitochondrial membrane transition.34 NOX-independent ROS such as mitochondria-produced ROS might play a more important role in the generation of hepatic lipid peroxidation in BDL than in CCl4.

Our current study characterizes the functional contribution of different NOX1- and NOX2-expressing cell populations to hepatic fibrosis. Through experiments using NOX1 and NOX2 BM chimeric mice, we demonstrate that NOX1 mediates fibrogenic effects in endogenous liver cells, and NOX2 mediates fibrogenic effects in both endogenous liver cells and BM-derived cells. In this study, NOX2 BM chimeric mice that expressed NOX2 in endogenous liver cells but not BM-derived cells (NOX2KO BM→WT) showed a modest but significant reduction of fibrosis compared with WT mice. These results are consistent with our previous study using p47phox BM chimeric mice. p47phox BM chimeric mice that expressed p47phox in endogenous liver cells but not BM-derived cells (p47phoxKO BM→WT) showed an ≈25% reduction in fibrosis, whereas chimeric mice with WT BM-derived cells and p47phoxKO endogenous liver cells (WT BM→p47phoxKO) showed an ≈60% reduction in fibrosis.26 Taken together, NOX2 in both endogenous liver cells and BM-derived cells contributes to liver fibrosis, with the endogenous liver cells making the greater contribution.

Phagocytic NOX is expressed in all professional phagocytes (macrophage, neutrophils and eosinophils) and also in B and T lymphocytes. The role of NOX expressed in nonphagocytic inflammatory cells such as lymphocytes, natural killer cells and natural killer T cells in hepatic fibrogenesis is unknown. T lymphocytes express a phagocyte-type NOX that functions in T cells to produce ROS in response to stimulation through the T cell receptor.35 When we assessed the expression of M1 and M2 macrophage markers in the fibrotic liver, there was no significant difference between WT and NOX2KO mice, suggesting the less important role of other NOX2-expressing, BM-derived immune cells in hepatic fibrosis.

Analysis of expression of NOX components in isolated liver cell fractions from control mice demonstrate that phagocytic NOX components such as NOX2, p40phox, p47phox, and p67phox are mainly expressed in KCs, whereas nonphagocytic NOX components including NOX1, NOXO1, and NOXA1 are expressed in HSCs and SECs. In addition, both NOX1 and NOX2 components are up-regulated in activated HSCs compared with quiescent cells. We confirmed the expression of NOX1 and NOX2 proteins in mouse HSCs as well as in the human activated HSC line LX-2. We demonstrated that Ang II–induced ROS production and fibrogenic responses in NOX1- or NOX2-deficient HSCs are attenuated compared with WT HSCs, indicating that both NOX1 and NOX2 are important in NOX-mediated ROS generation and fibrogenic responses in HSCs. Taken together, HSCs appear to be the primary cell type for NOX1- and NOX2-mediated hepatic fibrosis. We also found that NOX1 is expressed in the minority of CD31-positive SECs in the fibrotic liver. We suggest that NOX1-mediated low levels of O2.− production in SECs may have some regulatory function in the liver and warrants further study.

ROS has diverse effects with respect to different kinds, concentrations, and cell types. H2O2 and superoxide have different physiological characteristics in that H2O2 can easily diffuse across plasma membrane and throughout the cell, whereas superoxide diffuses poorly across cell membranes.36 Although higher concentration of ROS is cytotoxic, lower concentration of ROS serves as a second messenger during cellular response to a variety of physiological stimuli. A low dose of H2O2 has mitogenic effects and can mimic the function of growth factors.37 Regarding the effects of ROS on HSCs, the contradictory results have been reported: both mitogenic and cell death–inducing properties. Nontoxic levels of ROS or lipid peroxidation products stimulate the activation, proliferation, and collagen production of HSCs, but high concentration of ROS induce HSC death.4, 5, 38 We speculate that NOX2-mediated robust production of superoxide in KCs acts mainly for the host defense, while NOX1- and NOX2-mediated ROS generation in HSCs may act as an important secondary messenger to activate HSCs in hepatic fibrosis.

In conclusion, our study demonstrates the contributory roles of NOX1 and NOX2 in hepatic fibrosis. Mice lacking NOX1 or NOX2 show attenuated hepatic ROS generation and liver fibrosis. Chimeric BM mice demonstrate that both NOX1 and NOX2 have an important role in hepatic fibrosis in endogenous liver cells, including HSCs, whereas NOX2 has a lesser role in BM-derived cells. Activated HSCs have up-regulated expression of components of NOX1 and NOX2, and both NOX1 and NOX2 mediate ROS generation and fibrogenic responses in HSCs. Our study provides the rationale to target specific components of nonphagocytic NOX as novel therapies for hepatic fibrosis without suppression of NOX2-mediated host defense.

Acknowledgements

We thank Karin Diggle for technical assistance and Jung Ho Lee for technical assistance on fluorescent microscopy and helpful discussion.

Ancillary