Introduction

The cytosolic E3 ubiquitin (Ub)-ligase C-terminus of Hsc70-interacting protein (CHIP) along with its cognate E2 Ub-conjugating enzyme H5a (UbcH5a) and its cochaperones Hsc70/Hsp40, participates in the Ub-26S proteasome-dependent endoplasmic reticulum-associated degradation (ERAD) of various proteins including the hepatic drug-metabolizing enzymes, cytochromes P450 (P450s)1,2,3,4,5. CHIP-knockdown in cultured rat hepatocytes stabilizes functionally active P450s CYPs 3A and 2E1 and their inactive ubiquitinated species4 (Fig. S1). Such hepatic stabilization of functionally active P450s upon their ERAD-impairment is clinically relevant to P450-dependent drug metabolism and consequent drug-drug interactions, as human CYP3A4 is responsible for the metabolism of >50% of therapeutic drugs6. Similarly, human liver CYP2E1 catalyzes the biotransformation of many clinically relevant drugs (ethanol, acetaminophen), carcinogens and endogenous ketones and fatty acids (FA)6. Its ability to bioactivate xenobiotics into toxic/reactive intermediates and its high propensity for generating reactive O2 species (ROS) have implicated CYP2E1 in the pathogenesis of toxic liver damage, alcoholic liver disease, nonalcoholic steatohepatitis (NASH), diabetes and obesity7,8,9,10,11. Although CYP2E1 is normally less abundant (≈5–7% of human hepatic P450 content) than CYPs 3A (≈30%)6, its abnormally elevated basal content (>7%) either via transcriptional induction or protein stabilization in these conditions is thought to predispose and/or abet pathogenesis of liver injury7,8,9,10,11. Thus tight regulation of CYP2E1 content is clinically desirable. This regulation involves balanced CYP2E1 protein synthesis and degradation via both ERAD and autophagy2,3,12,13. Indeed, CYP2E1-stabilization upon autophagic disruption enhances ROS-mediated oxidative stress and cytotoxicity, reducing E47 HepG2 cell-viability13.

Although ERAD is also an important determinant of hepatic CYP2E1 content, its physiological relevance and the extent of its dependence on CHIP/UbcH5a/Hsc70/Hsp40-mediated ubiquitination are unknown. CHIP Ub-ligase is physiologically relevant, as its genetic ablation in mice results in premature aging, shortened lifespan and various pathologies including widespread oxidative damage due to disrupted protein quality control14. The markedly increased hepatic lipid peroxidation in 3-month old CHIP−/−-mice relative to that in age-matched wild-type (WT; CHIP+/+) controls, suggests early oxidative liver damage that within 12 months not only spreads to additional tissues, but also compromises hepatic proteasomal function14,15. Our findings that CHIP-knockdown increased functional hepatic P450 content4 prompted us to examine whether the age-dependent oxidative damage in CHIP−/−-livers was due to P450 stabilization.

CYPs 3A and CYP2E1 undergo futile oxidative cycling that generates H2O2 and other ROS (O2- and HO. radicals)16,17. Herein using relatively selective P450 functional probes we document that CYP2E1 largely and CYP3A to a lesser extent, contribute to the age-dependent oxidative damage and proteotoxic stress in CHIP−/−-livers. Additionally, we document that such chronic CYP2E1-elicited oxidative stress in CHIP−/−-hepatocytes is associated with the sustained activation of stress-activated protein kinase (SAPK)/c-Jun NH2-terminal kinase (JNK)-signaling cascades, nuclear factor κB (NF-κB) and inflammatory cytokines and chemokines and the Nod-like receptor P3 (NLRP3)-inflammasome, which may significantly contribute to the age-dependent cellular ballooning and microvesicular steatosis observed in CHIP−/−-livers. However, in spite of this concatenation of nonalcoholic fatty liver disease (NAFLD)/NASH-like events, little evidence of NAFLD/NASH, as signaled by its hallmark macrovesicular steatosis, was actually found in CHIP−/−-livers over the first 8–9 months of life. The significant activation of the hepatic energy- and ROS-sensor 5′AMP-activated protein kinase (AMPK) coupled with the significantly up-regulated expression of epididymal white adipose tissue (EWAT) adiponectin and its hepatic adipoR1/adipoR2-receptors, observed early in 2-month-old CHIP−/−-mice relative to that in age-matched CHIP+/+-controls, indeed presaged the concurrent up-regulation and activation of the adiponectin-AMPK-forkhead box O (FOXO)-signaling axis. In this, FOXO/FKHR transcription factors, critical nodes at the intersection of the JNK- and AMPK-signaling networks18,19,20,21,22,23, play a key regulatory role to insure that the hepatoprotective adiponectin-AMPK-FOXO-signaling largely prevails over the liver injury-promoting JNK1-pathway, thereby counteracting/delaying any pathogenic progression into NASH in CHIP−/−-livers. Our findings reinforce the growing awareness of the beneficial role of adiponectin-AMPK-FOXO-signaling pathway in the pathogenesis of NAFLD/NASH24,25, suggesting that its therapeutic targeting could be exploited as a management strategy.

Results

Hepatic CYP3A and CYP2E1 functional stabilization upon genetic CHIP-knockout induces oxidative stress

Upon Western-immunoblotting (IB) analyses, appreciably higher CYP3A and CYP2E1 protein stabilization was observed in cultured CHIP−/−-hepatocyte lysates than in CHIP+/+-hepatocyte lysates, irrespective of mouse age (Fig. S2B), thereby verifying that CHIP-knockout stabilized both hepatic CYP3A and CYP2E1 content. This increased P450 content in CHIP−/−-hepatocytes was functional, as documented by the CYP3A and CYP2E1 functional probes [7-benzyloxy-4-trifluoromethylcoumarin (BFC) and 7-methoxy-4-trifluoromethylcoumarin (MFC), respectively; Fig. S2C].

Genetic CHIP ablation significantly increased hepatic 15-F2t-isoprostane (15-F2t-IP) and malondialdehyde (MDA) basal levels (Fig. 1A–C). The CYP3A inhibitor ketoconazole (KTZ) and CYP2E1 inhibitor 4-methylpyrazole (4-MP) effectively blocked these increases (Fig. 1A,B). When CHIP+/+- and CHIP−/−-hepatocytes were concomitantly pretreated with both dexamethasone (DEX) and isoniazid (INH) to restore basal CYP3A and CYP2E1 content and then cotreated with KTZ (5 μM) plus 4-MP (2.5 mM), basal 15-F2t-IP levels were decreased by 56% and 80%, respectively (Fig. 1C). On the other hand, MitoTEMPO [(2-(2,2,6,6-tetramethylpiperidin-1-oxyl-4-ylamino)-2-oxoethyl) triphenylphosphonium chloride.monohydrate, 100 μM; a mitochondria-targeted antioxidant probe with superoxide/alkyl radical-scavenging properties]26 attenuated the elevated 15-F2t-IP by 60% in both CHIP+/+- and CHIP−/−-hepatocytes. MitoTEMPO together with 4-MP and KTZ only slightly enhanced this inhibition, thereby revealing the major functional contribution of CYP3A and CYP2E1 to this 15-F2t-IP increase.

Figure 1
figure 1

Enhanced oxidative stress stemming from hepatic CYP2E1 and CYP3A stabilization in CHIP−/−-livers.

Functional contribution of CYP3A (left) and CYP2E1 (right) to hepatic 15-F2t-IP production (A) and MDA-generation (B) in cultured hepatocytes from 2-month-old mice with diagnostic inhibitors KTZ and 4-MP (C). Relative contribution of CYP3A and CYP2E1-generated and/or mitochondrial ROS to 15-F2t-IP-production assessed with KTZ/4-MP and/or MitoTEMPO as probes. (D) IB-analyses of relative HNE-conjugation (left) and oxidized protein-carbonyls (right) in CHIP+/+- and CHIP−/−-liver lysates. -CT refers to the corresponding controls in the absence of Oxyblot-reagents. (E) Confocal immunofluorescence analyses of age-dependent in situ HNE-conjugation in hepatocytes cultured from CHIP+/+- and CHIP−/−-livers. The relative quantification of the HNE-immunofluorescence from 7-month-old CHIP+/+- and CHIP−/−-hepatocytes4 is shown (bottom).

Full size image

Appreciable elevation of 4-hydroxynonenal (HNE)-protein conjugation and oxidized protein side-chain carbonyls provided additional evidence for the relatively enhanced oxidative stress in cultured CHIP−/−-hepatocytes (Fig. 1D). In situ confocal immunofluorescence of CHIP−/−- and CHIP+/+-hepatocytes (Fig. 1E) revealed higher HNE levels in CHIP−/−-hepatocytes that further increased at 4–7 months (Fig. 1E). Together these findings evince that genetic CHIP-ablation functionally stabilizes hepatic CYP2E1 and CYP3A with consequently heightened intracellular oxidative stress that was largely mitigated by P450 functional inhibitors.

Age-related pathological changes in CHIP−/−- and CHIP+/+-hepatocytes: Histological analyses

We characterized the age-related morphological changes in 2-, 4- and 9-month-old CHIP−/−- and CHIP+/+-mouse livers histologically following staining with hematoxylin and eosin (H&E), Oil red O for detection of lipid accumulation and Masson’s trichrome for detection of fibrosis (Fig. 2). No great differences in H&E-based histology were detectable at 2- or 4 months between CHIP−/−-livers and corresponding age-matched CHIP+/+-controls. However with age, hepatocyte ballooning with pyknotic nuclei (characteristic of dying and/or apoptotic cells) was quite marked in 9-month-old CHIP−/−-mice relative to age-matched CHIP+/+-controls (Fig. 2). No evidence of significant inflammation as demonstrated by lymphocytes and neutrophils in portal or lobular areas, characteristic of clinical NASH, could be found. On the other hand, Oil red O-stained sections revealed “microvesicular” steatosis, but not the macrovesicular steatosis characteristic of clinical NASH and rodent NASH-models7,11,27,28, that progressed from 4 to 9 months in CHIP−/−-mice relative to age-matched CHIP+/+-controls (Fig. 2). Trichrome-stained sections from 9-month-old CHIP−/−-livers relative to age-matched CHIP+/+-controls revealed central fibrosis, although the sinusoidal pattern usually associated with clinical NASH was not observed. Rather on examining all liver sections, the examining clinical hepatopathologist (JPG) found striking evidence of “fibrosis due to mild cardiovascular congestion in the central vein stemming from the onset of heart failure”, a plausible cause of death in these prematurely aging CHIP−/−-mice manifesting cardiac hypertrophy and compromised cardiac function15. Thus, although these analyses documented age-dependent hepatic lipid accumulation, the characteristics of injury at this stage were clearly different from those of clinical NASH.

Figure 2
figure 2

Age-related histological alterations in CHIP+/+- and CHIP−/−-livers.

Representative liver sections stained with H&E (left), Oil red O (middle) and Masson’s trichrome (right) at a 10 μ-magnification view are illustrated. Insets correspond to their respective 20 μ-magnification images.

Full size image

Sustained CYP2E1-mediated oxidative stress in CHIP−/−-hepatocytes is associated with the activation of intracellular signaling cascades

To identify any plausible signaling cascades affected by CYP2E1-mediated persistent oxidative stress in CHIP−/−-livers, we screened cultured hepatocytes from 2-, 4- and 9-month-old mice using the PathScan Intracellular signaling array kit. This slide-based antibody array provides a broad snapshot of age-dependent activation of signaling modules and/or proapoptotic processes (Fig. S3A). Several key signaling transducers (Fig. S3) activated early (≈2 months) and in a CYP2E1-dependent manner in CHIP−/−-livers were thus identified: (i) Mitogen-activated protein kinase (MAPK) JNK, activated by pro-inflammatory cytokines and cellular stresses8,11,29,30,31,32,33,34, (ii) energy/metabolic sensor AMPK-α1-subunit (AMPKα), activated via Thr172-phosphorylation by elevated intracellular AMP and/or ROS levels35 and (iii) insulin/insulin receptor substrate (IRS-1/IRS-2)/phosphatidylinositol 3-kinase-dependent activation of the serine-threonine kinase Akt by 3-phosphoinositide-dependent protein kinase-1 (PDK1) via Thr308-phosphorylation36. Insulin-signaling-dependent Akt-activation via Ser473-phosphorylation, on the other hand, was only transiently increased at 4 months, but reverted to basal levels at 9 months in CHIP−/−-hepatocytes (Fig. S3). By contrast to JNK, p38MAPK, another similarly activated kinase and the extracellular-signal-regulated kinase (ERK1/2) were markedly activated in CHIP−/−-hepatocytes but only at 9 months (Fig. S3B).

Concurrent apoptosis in 9-month-old CHIP−/−-hepatocytes relative to age-matched CHIP+/+-controls was documented by the activation of the pro-apoptotic signal transducer and activator of transcription 1 (STAT1)-signaling and marked activation of caspase 3 (critical executor of apoptosis) via endoproteolytic cleavage at Asp214 and subsequent cleavage of its principal target polyADP-ribose polymerase 1 (PARP1), involved in DNA repair (Fig. S3). Such enhanced caspase 3-activation was not 4-MP-sensitive and thus CYP2E1-independent. Most likely, consistent with the pyknotic nuclei in 9-month-old CHIP−/−-livers (Fig. 2; H&E staining), such apoptosis is inherent to the CHIP-null phenotype14,15.

Sustained JNK-activation and steatosis in CHIP−/−-hepatocytes

Given the marked age- and CYP2E1-dependent JNK-activation in CHIP−/−-hepatocytes, we assessed upstream and downstream transducers in the apoptosis signal-regulating kinase (ASK1)-JNK-MAPK-protooncogene c-Jun/activator protein 1 (AP1)-signaling cascade in their native and activated/phosphorylated forms8,10,11,29,30,31,32 (Fig. 3A,B). The levels of ASK1, the first transducer in this cascade and its activated form (pASK1) were increased significantly in CHIP−/−-hepatocytes at 2 months over age-matched WT-controls. Neither increase was CYP2E1-dependent. MAPK-kinase MKK4 levels were comparable in CHIP−/−-hepatocytes and WT-controls; but activated MKK4 (pMKK4) levels were dramatically increased in CHIP−/−-hepatocytes relative to WT-controls, in a 4-MP-sensitive manner (Fig. 3). As previously (Fig. S3), the levels of JNK1 (46 kDa) and its activated species (pJNK1) were also markedly increased in CHIP−/−-hepatocytes relative to WT-controls and this JNK-activation was decisively 4-MP sensitive (Fig. 3). The levels of the pJNK-target c-Jun were slightly decreased, whereas those of activated c-Jun (pc-Jun) were significantly increased in CHIP−/−-hepatocytes relative to WT-controls. Furthermore, this activation was also 4-MP-sensitive (Fig. 3). On the other hand, the levels of the activating transcription factor 2 (ATF2), another pJNK-target37,38,39, were slightly decreased if at all, but its activation (pATF2) was significantly increased in CHIP−/−-hepatocytes relative to WT-controls (Fig. 3). Together these findings reveal that the ASK1-MKK4-JNK-c-Jun- and ASK1-MKK4-JNK-ATF2-signaling cascades were significantly activated upon CHIP-knockout, but only the activation of MKK4, JNK and c-Jun was apparently CYP2E1-dependent. This hepatic pJNK-activation with progressive pc-Jun- and pATF2-activation persisted over 8–9 months (Fig. 3C,D), consistent with the sustained CYP2E1-elicited oxidative stress.

Figure 3
figure 3

Dependence of JNK-activation pathway transducers on CYP2E1 function.

IB-analyses (A) and corresponding densitometric immunoquantification (B) of the parent proteins and their activated/phosphorylated species in hepatocyte lysates from 2-month-old CHIP+/+- and CHIP−/−-mice (N = 3 individual mice), treated with or without 4-MP (5 mM) for 2 h before cell-harvest. IB-analyses (C) and corresponding densitometric immunoquantification (D) of these parameters (N = 3 individual mice) illustrating age-alterations, if any.

Full size image

Up-regulation of hepatic lipogenic genes and pro-inflammatory/inflammatory cytokines/chemokines

The concomitant age-dependent microvesicular steatosis and hepatocyte ballooning observed in CHIP−/−-livers relative to corresponding WT-controls (Fig. 2) led us to determine through quantitative real-time polymerase chain reaction (qRT-PCR) analyses of hepatic mRNA expression, whether lipogenic and pro-inflammatory/inflammatory cytokine/chemokine genes were upregulated and/or antilipogenic genes down-regulated (Fig. S4). Indeed, in spite of the rather weak concurrent AktS473-activation (Fig. S3), the expression of both hepatic antilipogenic “insulin-induced genes” insig-1 and insig-240 was significantly down-regulated at 9 months in CHIP−/−-livers relative to WT-controls, thereby enhancing hepatic sterol regulatory element-binding protein (SREBP)-proteolytic processing and consequent transcriptional activation of SREBP target genes40. Additionally, hepatic srebp-1 c and srebp-2a gene expression was also concurrently upregulated, thus synergistically upregulating lipogenic target genes responsible for FA-synthesis (fas1, scd-1, acc1) and lipid-uptake (ATP-binding cassette transporter abc-a1), consistent with the observed hepatic microvesicular steatosis40.

More importantly, concurrent up-regulation of inflammatory cytokines such as tumor necrosis factor α (TNFα), interleukin 6 (IL-6) and macrophagic chemokine monocyte chemotactic protein 1 (MCP-1) was also significantly detected in CHIP−/−-livers as early as 2 months. Their expression along with that of IL-1β remained upregulated at 9 months. In CHIP−/−-livers, this TNFα-up-regulation is paralleled by their NF-κB-activation profile41. Thus, electromobility shift assays (EMSA) of nuclear extracts from CHIP−/−- and CHIP+/+-livers revealed the very early relative activation of p65/p50 NF-κB-heterodimers in CHIP−/−-livers at 2 and 4 months, that peaked at 8 months but reverted to WT-levels at 12 months (Fig. 4A), possibly due to hepatocyte dropout (see below). Similarly, IL-6 up-regulation was also consistent with CYP2E1-dependent pro-inflammatory activation of STAT3-signaling (Fig. S3). Additionally, a decisive elevation of NLRP3 (cryopyrin), an NLRP3-inflammasome component involved in pro-IL-1β-activation to IL-1β42 was detected at 9 months in CHIP−/−-hepatocytes but not earlier, or in age-matched controls (Fig. 4B). Such up-regulation of inflammatory/pro-inflammatory effectors may stem from the sustained ROS-elicited hepatic JNK-, p38MAPK- and ERK1/2-activation detected in CHIP−/−-mice with age. This activation may also account for the relative up-regulation of grp78, an ER-stress marker.

Figure 4
figure 4

Activation of inflammatory pathways and relative activity of insulin-signaling pathway in CHIP+/+- and CHIP−/−-livers.

NF-κB-activation, monitored through EMSA-analyses of nuclear extracts from 2-, 4-, 8- and 12-month-old intact livers (Experimental Procedures) (A). IB-analyses of hepatic NLRP3-component of the NLRP3-inflammasome with mouse age (B). (C) Age-dependent basal or activated/phosphorylated content of IRS1 and Akt was assessed in cultured hepatocytes via IB-analyses with actin as a loading control. (D) The corresponding densitometric immunoquantification relative to actin from 3 individual mouse livers is illustrated.

Full size image

Relative predisposition to injury of CHIP−/−-hepatocytes

Given this remarkable collective pathogenic profile of CHIP−/−-hepatocytes, we monitored cytosolic alanine aminotransferase (ALT)-leakage into the medium as a hepatocellular-injury marker over a 24 h-period. Although this was slightly, albeit significantly higher in CHIP−/−-hepatocytes from 2-month-old mice relative to those from age-matched WT, little basal cytotoxicity was evident under conditions of routine culture (Williams medium E (WME)/5 days) (Fig. S5). Because all these mice were fed a standard chow-diet, relative predisposition to cell injury conceivably could be elicited upon hepatocyte culture in a reportedly steatogenic methionine-choline deficient (MCD)-WME medium43. Although culture in MCD-WME indeed stimulated extracellular ALT-leakage from CHIP−/−-hepatocytes, this was only nominally higher than that of similarly cultured CHIP+/+-hepatocytes. Comparable findings were observed when cell injury was incited with hepatotoxic acetaminophen concentrations44 (Fig. S5). Surprisingly, these findings revealed that in spite of their persistent oxidative stress and activated JNK1-signaling, CHIP−/−-hepatocytes were no more predisposed to cell injury than their WT-counterparts.

Insulin-signaling in the CHIP−/−-liver

The PathScan arrays revealed that the relative Akt-Ser473/Thr308-phosphorylation ratio, a plausible index of insulin signaling45, actually dropped with age in CHIP−/−-livers (Fig. S3). Thus, on the one hand, Akt-activation critical for insulin signaling was apparently impaired in CHIP−/−-livers, as inferred from the somewhat feeble activation of glycogen synthase kinase (GSK) 3β, an Akt-target46 (see below). On the other, the down-regulation of insulin-regulated hepatic insig-1 and insig-2 genes signaled adequate insulin-availability in CHIP−/−-livers (Fig. S4). These conflicting insulin-dependent responses obfuscated the real status of insulin signaling in CHIP−/−-livers and its plausible impact on their relative NAFLD/NASH-susceptibility. Because of this and the significant basal pancreatic CHIP-expression1, we directly assessed the functional status of insulin signaling in CHIP−/−-livers (Fig. 4C,D). We found that in CHIP−/−-hepatocytes although the basal IRS1-expression (Fig. S4) and protein content (Fig. 4C,D) were comparable to those in CHIP+/+-controls over the first 8–9 months, IRS1-activation (via Tyr895-phosphorylation) was significantly lower than in age-matched controls at 2 months and further declined over 4–8 months (Fig. 4C,D). By contrast, IRS1Ser307-phosphorylation was significantly elevated at 2 months in CHIP−/−-hepatocytes, but reverted to age-matched WT-levels by 8 months (Fig. 4C,D). However, hepatic insulin signaling as reflected by relative AktSer473/Thr308-phosphorylation, although per se not significantly impaired over the first 4 months compared to that in age-matched WT-controls, tended to decline thereafter.

Activation of hepatic adiponectin-AMPK-FOXO1-signaling axis upon CHIP-ablation

For reasons discussed above, we examined whether the significant CYP2E1-dependent AMPK activation (pAMPKα1) was associated with concurrent hepatic activation of the adiponectin-AMPK-FOXO-signaling, an event that could potentially delay NAFLD/NASH-onset. Indeed, the expression of hepatic adiponectin receptors (adipoR1/adipoR2) and the adipocyte adipokine adiponectin (adipoQ) was significantly up-regulated at 2 months in CHIP−/−-livers relative to age-matched WT, but the latter began to decline at 9 months (Fig. 5A). Although the content of the AMPK-activating kinase, liver kinase B1 (LKB1) in CHIP−/−- and CHIP+/+-hepatocytes was comparable, its activated (pLKB1) levels were relatively increased over 2–8 months in CHIP−/−-hepatocytes (Fig. 5B,C). Consistent with this LKB1-activation, a significant enhancement of AMPK-activation (pAMPKα) was concurrently observed in CHIP−/−-hepatocytes relative to age-matched WT-controls (Fig. 5B,C). This AMPKα-activation in CHIP−/−-hepatocytes was also 4-MP-sensitive and thus CYP2E1-dependent (Fig. 5B,C). Predictably, this AMPK-activation was associated with the marked Ser79-phosphorylation of acetyl-CoA carboxylase (ACC2), its diagnostic probe35,47, thereby attesting to its physiological relevance (Fig. 5B,C).

Figure 5
figure 5

Transcriptional up-regulation of adiponectin-AMPK-FOXO and/or insulin signaling genes and activation of adiponectin-AMPK-FOXO-signaling axis in CHIP−/−-livers.

(A) qRT-PCR analyses of total RNA extracted from intact CHIP+/+- and CHIP−/−-livers [adiponectin receptors (adipoR1, adipoR2), acc2, irs-1 and FOXO-regulated genes (pgc1α, acox1, atg14, lpl)] or adipose tissue (adipoQ) at 2 or 9 months. Relative mRNA expression was quantified as described (Methods, Table S1). Statistical significance between the values shown at p < 0.001 (*) or p < 0.005 (§). (B) Age-dependent expression of various transducers in this pathway detected through IB-analyses and (C) corresponding densitometric immunoquantification of individual hepatocytes cultures from N = 3 CHIP−/−- and CHIP+/+-livers. AMPKα-activation via Thr172-phosphorylation was monitored in cultured CHIP−/−- and CHIP+/+-hepatocytes.

Full size image

Remarkably, the basal protein content of AMPKα was distinctly increased in CHIP−/−-hepatocytes (Fig. 5) and this increase was attenuated upon exogenous CHIP-overexpression (Fig. 6A). This suggested that hepatic AMPKα is either a CHIP-substrate, or requires CHIP as a chaperone for its degradation. To examine the first possibility, we co-transfected HEK293T cells with glutathione S-transferase (GST)-AMPKα1-, haemagglutinin (HA)-Ub-, and/or [His]6CHIP-plasmid vectors, singly or in combination (Fig. 6B). GSH-Sepharose pull-down coupled with IB-analyses revealed that AMPKα1 was indeed intracellularly ubiquitinated when all three vectors were cotransfected (Fig. 6B). By contrast, similar coexpression of HA-CHIP and HA-Ub failed to enhance AMPKα2-ubiquitination (Fig. 6C). Significant AMPKα1-ubiquitination was also detected upon cotransfection of just GST-AMPKα1 and HA-Ub, presumably due to endogenous CHIP and/or other putative E3 Ub-ligases (see below). Furthermore, such AMPKα1- but not AMPKα2-ubiquitination could be enhanced upon treatment of the cotransfected cells with the proteasomal inhibitor MG132 (Fig. 6B,C). Incontrovertible evidence was provided by the in vitro ubiquitination of AMPKα1-isoform in a functionally reconstituted CHIP-system (Fig. 6D). Such CHIP-mediated AMPKα1-ubiquitination required both the CHIP-cochaperone-interacting tetratricopeptide repeat (TPR) and Ub-ligase U-box-catalytic subdomains (Fig. 6D). To our knowledge, this is the first evidence that in contrast to AMPKα2, hepatic AMPKα1-isoform is a target of both CHIP-ubiquitination and proteasomal degradation.

Figure 6
figure 6

AMPKα1 but not AMPKα2 is a CHIP-substrate.

(A) The content of AMPKα-subunit and its activated species (pAMPKα) was monitored in the presence or absence of CYP2E1 inhibitor 4MP. In parallel, some CHIP−/−-hepatocytes were also transfected with a CHIP-expression plasmid for 36 h, before harvesting. HEK293T cells were cotransfected with GST-AMPKα1- or [His]6AMPKα2, HA-Ub- and/or [His]6CHIP-expression vectors ± MG132 (10 μM), followed by GSH-Sepharose- (B) or Dynabead-His6- (C) pull-down and subsequent IB-analyses. In vitro AMPKα1- or CYP3A4 (positive control)-ubiquitination (D) in a functionally reconstituted CHIP-system with purified CHIP, its U-box- or TPR-deleted mutant and subsequent IB-analyses with anti-HA antibody and Typhoon visualization4. Color wheel intensity code: Red > orange > yellow > green > blue > indigo > violet. (E) Hepatic nuclear and cytosolic distribution of activated (pS/pT) FOXO1/FOXO3-species in CHIP−/−-livers. Liver lysates were subjected to subfractionation as detailed (Experimental Procedures). Nuclear or cytosolic subfraction was immunoprecipitated (IP) with either FOXO1- or FOXO3-antibody, followed by IB-analyses with an anti-pSer/pThr-antibody. GAPDH and Histone H3 were used respectively as markers of relative cytosolic and nuclear fraction purity.

Full size image

In mammalian cells, activated AMPK phosphorylates FOXO (FOXO1, FOXO3, FOXO4 and FOXO6)-transcription factors18,23, well-established CHIP substrates48 and also increases their expression and protein stability18,23,49. Indeed, we found that CHIP−/−-hepatocytes exhibited not only a relatively higher total cellular phosphorylated FOXO1 and FOXO3 content than CHIP+/+-controls, but also greater nuclear retention of their transcriptionally active phosphorylated species as evident upon cell-subfractionation (Fig. 6E). Such enhanced nuclear retention/segregation apparently protects FOXOs from proteasomal degradation and this protein stability of the AMPK-preferred target FOXO3 was particularly striking (Fig. 6E). As expected50, this enhanced FOXO-stability was associated with the transcriptional up-regulation of peroxisome proliferator-activated receptor-γ-coactivator pgc1α and its target acetyl-CoA oxidase (acox1) (Fig. 5A), consistent with the transcriptional up-regulation of energy, lipid metabolism and oxidative stress-resistance genes upon AMPK-mediated FOXO-activation18,19,20,21,22,23. Such enhanced hepatic FOXO activation was also associated with the up-regulation of their target Atg14 and lpl autophagic/lipophagic genes23,Genoty** of CHIP-knockout mice

Male CHIP+/−- and female CHIP−/−-mice generated as described14,15 were bred and the progeny genotype verified through PCR analyses of mouse-tail genomic DNA and CHIP-protein immunoblotting (IB) analyses of hepatocyte lysates (Fig. S2A). Mice were fed a standard laboratory chow-diet, given water ad libitum and maintained under a normal diurnal light cycle. All animal experiments were carried out strictly by protocols specifically approved by the UCSF/Institutional Animal Care and Use Committee (IACUC) and its care and use of laboratory animal guidelines. CHIP−/−-mice were maintained from birth to ≈12 months, as their median lifespan is <1 year14. We employed 2–9 month-old male mice, with 1 year-olds included whenever feasible.

Mouse hepatocyte culture and oxidative stress analyses

Cells from CHIP+/+-, CHIP−/−-, or CHIP+/−-livers were cultured as described4, with the CYP3A-inducer DEX (10 μM), or CYP2E1-inducer INH (1 mM), supplemented daily in WME for 5 days, to restore basal P450 loss upon culture. Upon harvesting, cell lysates were prepared as described4.

Four commonly employed oxidative stress indices70 were monitored: ROS-triggered membrane lipid-peroxidation 15-F2t-IP-products were assayed in the culture medium by a fluorescent immunoassay as per the manufacturer’s instructions. Levels of MDA, a byproduct of unsaturated FA-oxidation, were monitored in cell lysates as thiobarbituric acid-reactive adducts70. 4-HNE, another reactive unsaturated FA-oxidation byproduct, that covalently binds to protein lysine and/or SH-groups was monitored via cell lysate IB analyses69 and confocal immunofluorescence analyses of cultured hepatocytes in situ with a HNE-specific antibody4. Protein carbonyl-oxidation was assayed via Oxyblot (DNP-IB) analyses.

Fluorescence-based P450 functional assays

CYP3A and CYP2E1 function was assessed by diagnostic assays of BFC 7-O-debenzylation and MFC 7-O-demethylation, respectively, as described4. The P450 functional contribution was assessed by preincubating cultures for an hour with KTZ (5 or 10 μM) or 4-MP (1.25 or 2.5 mM), relatively selective inhibitors of CYP3A and CYP2E1, respectively6, before functional assays.

Nuclear/cytoplasmic extraction and FOXO1/FOXO3-immunoprecipitation (IP)

Nuclear and cytoplasmic subfractions were prepared from cultured hepatocytes washed with chilled phosphate-buffered saline (PBS) for 10 min with nuclear/cytosolic extraction reagents (NER/CER) (Pierce Biotech., Rockford, IL) according to the manufacturer’s instructions. Briefly, cells were mixed with ice-cold CER-I and then incubated on ice for 10 min. CER-II was then added and the cell mixture further incubated on ice for 1 min. Nuclei were harvested by centrifugation (100 × g) and the resulting supernatant was collected as the cytoplasmic extract. Nuclear pellet was suspended with ice-cold NER buffer and nuclear protein was extracted at 4 °C for 1 h, sedimented at 16,000 × g for 10 min at 4 °C to obtain the supernatant (nuclear extract). Total protein (200 μg) from each subfraction was precleared with protein A/G-Sepharose (10 μl; Santa Cruz Biotechnology, Santa Cruz, CA) for 1 h and then immunoprecipitated with anti-FOXO1 or FOXO3 antibody (2 μg) on a rocking platform at 4 °C for 2 h, followed by the addition of protein A/G-Sepharose beads (20 μl) at 4 °C for 16 h as described4. Immunoprecipitated complexes were washed, eluted and subjected to SDS-PAGE coupled with IB analysis against anti-phospho-serine/threonine antibody.

qRT-PCR Analyses

Real-time PCR was performed with total RNA isolated with RNeasy mini-kit (Qiagen), treated with DNase (DNA-free kit, Ambion) and reverse-transcribed with Accupower RT-PCR kit (Bioneer) for cDNA synthesis, in Power SYBR Green PCR Master Mix (Applied Biosystems; final 25 μl-volume) with Agilent Mx3005P System. Adipose tissue adipoQ was similarly analyzed with total RNA extracted using a combined TRIZOL (Invitrogen, Carlsbad, CA) and RNeasy mini-kit protocol. Relative gene expression level was determined by normalizing its threshold cycle (Ct) to that of Gapdh Ct. Primers used are listed (Supplementary Information, Table S1).

IB analyses

Upon harvesting, cultured hepatocytes were washed once with ice-cold PBS for 10 min and lysed in a cell-lysis buffer (Cell signaling Tech., Beverly, MA). The whole-cell lysates were clarified by centrifugation at 12,000 × g for 10 min. Protein concentrations were measured using the bicinchoninic acid (BCA) protein assay reagent (Pierce), 10 μg of proteins were resolved by 4–15% gradient SDS-PAGE, transferred to nitrocellulose membrane (Bio-Rad, Hercules, CA), blocked in 5% skim-milk for 1 h and probed with primary antibodies for 16 h at 4 °C, washed with 1X Tris-buffered saline containing 0.1% Tween 20 for >5 times. Following incubation with anti-mouse IgG HRP-linked antibody or anti-rabbit IgG HRP-linked antibody, bound immunoglobulins were detected using enhanced chemiluminescence (Pierce). Antibodies used for this study and their commercial sources are provided (Supplementary Information, Table S2). Immunoblots were densitometrically quantified by ImageJ (NIH) analyses with available software (http://rsbweb.nih.gov/ij/), using corresponding actin-loading controls for normalization.

Histological analyses

Mouse liver sections were fixed, stained and subjected to light-microscopy and imaging by the UCSF Liver Center Pathology and Gladstone Institute, Histology and Light Microscopy Cores.

Electromobility shift assays (EMSA)

Hepatic nuclear extracts were probed with a biotin-labeled oligonucleotide specific for the NF-κB-consensus sequence, as detailed71.

Statistical analyses

Experiments were generally performed in triplicate. Data were compared by analysis of variance and p values <0.05 were considered statistically significant.

Additional Information

How to cite this article: Kim, S.-M. et al. Chip−/−-Mouse Liver: Adiponectin-AMPK-FOXO-Activation Overrides CYP2E1-Elicited JNK1-Activation, Delaying Onset of NASH: Therapeutic Implications. Sci. Rep. 6, 29423; doi: 10.1038/srep29423 (2016).