Contributed by Lutz Birnbaumer, December 21, 2006 (received for review December 7, 2006)
Heterotrimeric G proteins of the Gi class have been implicated in signaling pathways regulating growth and metabolism under physiological and pathophysiological conditions. Knockout mice carrying inactivating mutations in both of the widely expressed Gαi class genes, Gαi2 and Gαi3, demonstrate shared as well as gene-specific functions. The presence of a single active allele of Gαi3 is sufficient for embryonic development, whereas at least one allele of Gαi2 is required for extrauterine life. Mice lacking both Gαi2 and Gαi3 are massively growth-retarded and die in utero. We have used biochemical and cell biological methods together with in situ liver perfusion experiments to study Gαi isoform-specific functions in Gαi2- and Gαi3-deficient mice. The subcellular localization of Gαi3 in isolated mouse hepatocytes depends on the cellular metabolic status. Gαi3 localizes to autophagosomes upon starvation-induced autophagy and distributes to the plasma membrane upon insulin stimulation. Analysis of autophagic proteolysis in perfused mouse livers showed that mice lacking Gαi3 are deficient in the inhibitory action of insulin. These data indicate that Gαi3 is crucial for the antiautophagic action of insulin and suggest an as-yet-unrecognized function for Gαi3 on autophagosomal membranes.
Heterotrimeric G proteins functionally couple seven-transmembrane cell-surface receptors to a variety of intracellular effector systems, including enzymes and ion channels. The heterotrimeric G protein α-subunits are encoded by 16 genes in humans and mice, and >20 distinct Gα protein isoforms have been identified (1). The Gi family of α-subunits comprises three closely related members, Gαi1–Gαi3, and originally was named for its ability to inhibit adenylyl cyclase activity. G proteins of the Gi class are characterized by their sensitivity to pertussis toxin (PTX). The Gαi1–Gαi3 isoforms share 85–95% of amino acid sequence identity and partially overlapping expression patterns. Although Gαi1 primarily is found in the nervous system, Gαi2 is expressed ubiquitously and represents the quantitatively predominant Gαi isoform. Gαi3, the closest homolog of Gαi1, is hardly detectable in the neuronal system but is broadly expressed in peripheral tissues.
Targeted loss-of-function mutations in mice have been produced for all Gαi isoforms and have provided information about shared as well as gene-specific biological roles. The loss of Gαi1 results in defects of long-term memory (2). Gαi2-deficient mice develop an inflammatory bowel disease with characteristics of ulcerative colitis (3). A gene-specific role of Gαi2 also has been described in the inflammatory response of alveolar macrophages (4). Furthermore, the disruption of the Gαi2 locus leads to increased peri- and postnatal mortality of unclear origin. In contrast, mice constitutively deficient in either Gαi1 or Gαi3 are viable and fertile. The initial analysis of Gαi3-deficient mice did not reveal overt phenotypic defects (5, 6). However, the comparative analysis of Gαi1/Gαi3-double-deficient with Gαi2-deficient mice led to the identification of overlapping as well as gene-specific functions of Gi isoforms in the antiinfectious response of macrophages and splenocytes (7). Recent studies of Gαi3 mutant mice have revealed craniofacial and other skeletal deformities consistent with a specific requirement for Gαi3 in the cranial neural crest and in somites (N. W. Plummer, K.S., J. Malphurs, L.B., unpublished work).
In addition to their established functions for signal transduction across the plasma membrane, heterotrimeric G protein α-subunits localize to intracellular membranes and have been implicated in membrane trafficking and fusion events along the secretory and endocytic pathways, such as vesicle formation by the endoplasmic reticulum, the Golgi/secretory pathway, and vesicle trafficking and fusion (8–11). In particular, Gαi3 has been localized to Golgi membranes in renal and pancreatic cells (12, 13) and to the endocytic compartment in kidney proximal tubule epithelium (14) and in rat hepatocytes (15). Studies in the colon carcinoma cell line HT-29 furthermore have suggested a role for Gαi3 in macroautophagy (16).
Macroautophagy (hereafter referred to as autophagy) is a ubiquitous membrane-trafficking mechanism that sequesters cytoplasmic material for targeted delivery to lysosomes, where the material is degraded and subsequently recycled (17, 18). In contrast to the selective degradation of generally short-lived proteins through the proteasomal degradation pathway, autophagy constitutes the major cellular pathway for the bulk degradation of long-lived proteins and the turnover of organelles. Constitutive autophagy plays an essential role for the maintenance of homeostasis by removing misfolded proteins or damaged organelles. Upon short-term starvation, the initiation of autophagy functions as a temporary survival mechanism by providing an alternative energy source. However, unrestrained autophagy results in cell death (19). Defects in autophagy perturb development, and the deregulation of autophagy has been proposed to play a role in diseases including cancer, neurodegenerative disorders, cardiomyopathy, and muscular diseases (18, 20).
A number of evolutionarily conserved proteins essential for autophagy have been identified, and their molecular and biological functions are beginning to be elucidated (21–25). Nevertheless, little is known currently about the signal transduction pathways that couple extra- or intracellular stimuli to the molecular machinery of autophagosome formation and maturation. The protein kinase mammalian target of rapamycin (mTOR) plays a central role for the transmission of autophagic stimuli. Amino acids alone or in combination with insulin or other growth factors can stimulate the activity of mTOR, leading to an inhibition of autophagy, but TOR-independent pathways of insulin signaling also have been described in the rat liver (26–28). Although it has been suggested that G proteins such as Gαi3 may be implicated in the control of autophagic sequestration in a colon carcinoma cell line (16, 17), supporting in vivo evidence is lacking.
Here, we have used biochemical and cell biological methods together with in situ liver perfusion experiments in Gαi2- and Gαi3-deficient mice to study the biological functions of Gi class G proteins. Our findings show that Gαi3 specifically is required for the insulin-mediated regulation of autophagy and suggest a previously unrecognized function for Gαi3 on autophagosomal membranes.
To distinguish gene-specific functions of the two peripheral Gαi isoforms, Gαi2 and Gαi3, from shared biological roles in vivo, we have intercrossed mice homo- or heterozygously deficient in Gαi2 and/or Gαi3 and analyzed the Gαi2- and Gαi3-deficient progeny. Although Gαi2-deficient mice were born at less than the expected frequency, and animal numbers were reduced by ≈50% at weaning age, i.e., postnatal day 21 (P21) (3), mice mutant for Gαi3 were born at the predicted Mendelian ratio [supporting information (SI) Tables 1–4]. Gαi2/Gαi3 double-deficiency resulted in massive growth retardation and embryonic lethality before embryonic day 10 (E10). The presence of one allele of Gαi3 in Gαi2−/−/Gαi3+/− embryos was sufficient for development and survival up to birth, but animals died perinatally (see SI Table 4). The presence of a single allele of Gαi2 in Gαi2+/−/Gαi3−/− mice maintained survival up to the neonatal period, but postnatal survival was reduced. Representative images of selected Gαi2/Gαi3 mutant embryos at E9 or E10 are shown in SI Fig. 6. Thus, the presence of a single active allele of Gαi3 is sufficient for embryonic development, whereas one allele of Gαi2 is required for extrauterine life. These data not only indicate overlapping roles of Gαi2 and Gαi3 but also suggest the existence of gene-specific functions.
To test the idea of functional compensation at the protein level, we have analyzed the expression of Gαi2 and Gαi3 in membrane preparations obtained from mouse embryonic fibroblasts (MEFs) or organs of wild-type (wt), Gαi2-deficient, and Gαi3-deficient mice. In the absence of Gαi2, steady-state protein levels of Gαi3 were markedly increased compared with wt levels. In contrast, the depletion of Gαi3 did not significantly trigger an up-regulation of Gαi2 protein levels (Fig. 1A). These data suggest that increased levels of Gαi3 may compensate for Gαi2 functions in the case of Gαi2 depletion, whereas wt Gαi2 levels may be sufficient to substitute for Gαi3 functions in a Gαi3-depleted system. In all investigated wt cells and tissues from peripheral organs, including liver, Gαi2 was the predominant Gαi isoform (Fig. 1 A and B and data not shown). We also have performed [32P]ADP ribosylation experiments with PTX to test the functional expression of Gαi2 and Gαi3 in these mice (Fig. 1C). Despite the relatively low abundance of Gαi3 compared with Gαi2 in whole-liver membranes, the amount of PTX-modified Gαi3 was comparable to PTX-modified Gαi2. In contrast, the quantity of [32P]ADP-ribosylated Gαi2/Gαi3 in thymus or spleen was proportional to the relative protein levels of Gαi2 and Gαi3.
Expression pattern of peripheral Gi isoforms. (A) Cell membranes from MEFs or peripheral mouse tissues from wt, Gαi2−/−, or Gαi3−/− mice were analyzed by immunoblotting with a Gαcommon antibody. Equal loading was confirmed by a nonspecific protein stain with Ponceau S. The depletion of Gαi2 induces an up-regulation of the steady-state protein levels of Gαi3, whereas Gαi2 levels remain virtually unchanged in the absence of Gαi3. (B) Liver cell membranes from wt and Gαi3 mutant mice were analyzed for Gi isoform expression as described above. (C) Cell membranes from liver, thymus, or spleen were subjected to a PTX-catalyzed ADP ribosylation in the presence of [32P]NAD+. Shown are representative autoradiographs. The extent of Gαi3 ADP ribosylation is highest in liver. n = 3 for all experiments.
To investigate the function of Gαi3 in liver, we have generated a Gαi3-specific antibody. In its affinity-purified form, this Gαi3 antibody specifically recognized Gαi3 species of the appropriate apparent molecular weight in whole-cell lysates of primary hepatocytes or MEFs of wt and Gαi2-deficient, but not Gαi3-deficient, mice (Fig. 2A). The Gαi3 antibody stained cultured primary hepatocytes from wt and Gαi2-deficient mice but failed to stain hepatocytes from Gαi3-deficient mice (Fig. 2B). We conclude that the Gαi3 antibody specifically detects Gαi3 but not Gαi2 by immunoblotting and immunocytochemistry.
Characterization of a Gαi3-specific antibody. (A) Whole-cell lysates (30 μg) of wt, Gαi2-deficient, or Gαi3-deficient mouse hepatocytes or MEFs were separated on a 6 M urea/SDS/PAGE gel, blotted onto nitrocellulose, and incubated with the affinity-purified polyclonal Gαi3 antibody. The antibody specifically recognizes Gαi3 but not the closely related Gαi2 isoform. Gαi3 typically migrates as a doublet under these conditions (see ref. 36 and references cited therein). (B) Primary mouse hepatocytes from wt, Gαi2-deficient, or Gαi3-deficient mice were fixed and stained with the affinity-purified polyclonal Gαi3 antibody as detailed in Experimental Procedures. The antibody stains wt and Gαi2-deficient, but not Gαi3-deficient, hepatocytes. Images are confocal micrographs (0.5-μm single slices). n = 3 for all experiments.
Gαi3 localizes to Golgi membranes and endosomes in different tissues, including liver (14), and has been involved in the control of autophagic sequestration and intracellular trafficking events in the colon carcinoma cell line HT-29 (16, 17). Membranes derived from the Golgi apparatus and the endoplasmic reticulum have been implicated in the biogenesis of autophagosomal vesicles, and endosomes and lysosomes fuse with autophagic vesicles during their maturation to a lytic compartment (17). To determine the potential role for Gαi3 in autophagy in liver, we first investigated the Gαi3 subcellular localization pattern. To induce autophagy, hepatocytes were subjected to a short-term amino acid starvation and growth factor starvation. Ubiquitin-like light-chain protein 3 (LC3) was used as a specific marker for the autophagosomal compartment (29). LC3 exists in a cytosolic form (LC3-I) and in a membrane-associated, phospholipid-conjugated form (LC3-II). The induction of autophagy leads to the processing of LC3-I to LC3-II and its subsequent targeting to preautophagosomal and autophagosomal membranes (30).
Initially, we ascertained the suitability of a goat polyclonal LC3 antibody for immunocytochemistry in hepatocytes because this antibody does not recognize endogenous or overexpressed LC3 in Western blotting experiments (data not shown). In contrast, a widely used rabbit polyclonal anti-LC3 antibody is able to detect endogenous or overexpressed LC3 by immunocytochemistry as well as in Western blotting experiments, but cannot be used together with the rabbit polyclonal anti-Gαi3 antibody because of cross-reactivity in the secondary detection step. To this end, we first transfected primary mouse hepatocytes with GFP-LC3 and stained for LC3. GFP-LC3 localized to cup-like or ring-like structures typical of autophagosomes, and the same structures were detected by the goat polyclonal LC3 antibody (SI Fig. 7A). Second, we validated the goat polyclonal LC3 antibody by comparing it to a widely used rabbit polyclonal LC3 antibody. Both antibodies recognized the same intracellular structures (SI Fig. 7B). We conclude that the goat polyclonal LC3 antibody is able to reliably detect autophagic vacuoles in cells.
Starvation-induced autophagy triggered a redistribution of Gαi3. In the absence of autophagy, Gαi3 predominantly was found diffusely distributed in the cytosol (see Fig. 2B). In contrast, cells undergoing starvation-induced autophagy were characterized by a local accumulation of Gαi3 on the plasma membrane as well as on LC3-positive autophagosomes and lysosomes (SI Fig. 8). The association of Gαi3 with membranes of autophagic vacuoles was corroborated biochemically by subcellular fractionation of isolated hepatocytes (SI Fig. 9). Gαi3 also distributed to the endoplasmic reticulum and to endosomes upon induction of autophagy (SI Fig. 10). These compartments are required for the biogenesis of autophagosomes (17, 31) and for the maturation of autophagosomes to lytic organelles (17). In particular, a Gαi3-based mechanism has been hypothesized to control the flux of membranes originating from different vacuolar compartments, such as the endoplasmic reticulum, to form a preautophagosomal structure (31). Thus, Gαi3 is targeted to subcellular compartments with known functions in autophagy.
Starvation-induced autophagy is counteracted by hormones, such as insulin, and by nutrients (e.g., amino acids). We therefore tested whether the subcellular localization of Gαi3 is controlled by physiological regulators of autophagy. Fig. 3 shows that the exposure of starved hepatocytes to insulin or phenylalanine led to the disappearance of the vesicular staining of LC3-II, indicating that autophagy was suppressed efficiently. Together with the loss of LC3-positive autophagosomes, the vesicular staining of Gαi3 disappeared. Instead, insulin induced an accumulation of Gαi3 at the plasma membrane. Treatment of cells with phenylalanine also resulted in the loss of autophagosomes and an accompanying disappearance of vesicular Gαi3 staining. Phenylalanine treatment caused a marked flattening of the cells and a diffuse cytosolic distribution of Gαi3, similar to resting cells.
Subcellular localization of Gαi3 upon inhibition of autophagy. Primary mouse hepatocytes were kept in culture medium (resting) or subjected to amino acid and growth factor withdrawal (starved). To inhibit autophagy, cells were treated with insulin (10 nM for 60 min) or with phenylalanine (0.2 mM for 60 min). The inhibition of autophagy is characterized by the disappearance of LC3-positive autophagosomes, paralleled by a loss of Gαi3 from the vesicular compartment and the scattering of lysosomes. Insulin induces an accumulation of Gαi3 in the plasma membrane, whereas phenylalanine leads to a diffuse cytosolic distribution of Gαi3. Note the characteristic membrane ruffling response to insulin and the marked flattening of cells in response to phenylalanine. Images are confocal micrographs. n = 3 for all experiments.
Autophagosomal vesicles mature into a catabolic compartment upon fusion with lysosomes to form autolysosomes (17). Provided that the activity of lysosomal proteases is inhibited by the addition of cell-permeable protease inhibitors, autolysosomes can be visualized with markers for autophagosomal and lysosomal vesicles. In hepatocytes undergoing autophagy (Fig. 3, starved cells) the staining for LC3 and Gαi3 largely overlapped with the lysosomal compartment. These data indicate that, under our experimental conditions, most LC3-positive vesicles already have fused with lysosomes and demonstrate that Gαi3 also can be detected on autolysosomes. Together, these data show that the subcellular localization of Gαi3 depends on the cellular metabolic status and suggest a previously unknown function for Gαi3 on autophagosomal membranes.
To test the relevance of these findings for autophagy in vivo, we have performed in situ perfusion experiments in mouse livers. Autophagy is induced rapidly upon liver perfusion in the absence of added amino acids (32). The system has been used widely to analyze the regulation of hepatic autophagy in rats (28).
Fig. 4A shows that insulin reversibly inhibited the starvation-induced autophagic proteolysis in wt mouse livers by ≈20–25%, comparable to the effect of the hormone in the rat liver (28, 32). Insulin suppresses autophagy to a physiologically maximal extent, even in the absence of added amino acids (17). To test the involvement of Gi proteins in this process, livers were preperfused with PTX. PTX markedly diminished the antiautophagic action of insulin (Fig. 4A). Hepatocytes predominantly express insulin receptors, whereas nonparenchymal liver cells such as Kupffer cells express type I insulin-like growth factor receptors (33). Together, these data suggest a role for Gi proteins downstream of insulin receptor signaling.
The regulation of autophagic proteolysis in vivo requires Gi proteins. (A) The in situ liver perfusion of wt mice with insulin (35 nM for 60 min) but not with perfusion buffer alone (control) leads to a reversible inhibition of proteolysis. The perfusion with PTX (100 ng/ml for 60 min) before the perfusion with insulin (35 nM for 60 min) inhibits the insulin response. (B) The in situ liver perfusion with phenylalanine (2 mM for 30 min) reversibly inhibits autophagic proteolysis. Pretreatment with PTX (100 ng/ml for 60 min) preceding the perfusion with phenylalanine (2 mM for 30 min) blocks the phenylalanine effect. Shown are mean values of duplicate measurements ± SD in three to four independent experiments.
High concentrations of amino acids alone are known to trigger a maximal inhibition of autophagy, even in the absence of insulin (17). Fig. 4B shows that phenylalanine reversibly inhibited autophagic proteolysis in wt mouse livers by ≈25%, which is comparable to the effects of insulin. Likewise, PTX largely abolished the antiautophagic action of phenylalanine.
We observed no significant additional antiautophagic effect of insulin in the presence of maximally effective levels of phenylalanine in four independent experiments compared with insulin alone (n = 3) or phenylalanine alone (n = 4; data not shown). Although insulin inhibited autophagic proteolysis half-maximally after ≈20 min, the effect of phenylalanine was half-maximal after 5–10 min (see Fig. 4). The time course of the inhibition of autophagic proteolysis upon addition of insulin plus phenylalanine was consistent with the combination of a faster and a slower inhibitory phase (data not shown).
The insulin- and amino acid-dependent regulation of autophagy appears to be mediated by distinct signal transduction mechanisms that may converge on mTOR (17, 26, 28). So far, no other signaling components have been identified that are shared by both pathways. Our data show that the antiautophagic action of insulin and phenylalanine can be blocked by PTX, indicating that Gi proteins are crucial for both signal transduction pathways (Fig. 4). However, neither insulin nor amino acids are known to activate G protein-dependent signal transduction. We have demonstrated a localization of Gαi3 to autophagosomal endomembranes that is reversible upon addition of insulin and phenylalanine (see Figs. 2 and 3), suggesting that Gαi3 may function downstream of both insulin- and phenylalanine-mediated regulation of autophagy.
To examine the role of Gαi3 in the insulin- and phenylalanine-mediated regulation in vivo, and to directly compare the potential contribution of Gαi2 and Gαi3, we have analyzed autophagic proteolysis in perfused livers of Gαi2- or Gαi3-deficient mice. Fig. 5A shows that the inhibitory action of insulin was lost almost completely in the absence of Gαi3, comparable to the effect of PTX (see Fig. 4A). In contrast, the insulin response was not detectably impaired in the absence of Gαi2, the predominant Gαi isoform in hepatocytes (see Fig. 1). The loss of Gαi2 or Gαi3 in the respective knockout mice was confirmed by ADP ribosylation (Fig. 5A Inset). Furthermore, the antiautophagic action of phenylalanine was lost in the absence of Gαi3, as shown in Fig. 5B together with a statistical evaluation of the described liver perfusion experiments.
Gαi3 is required for the antiproteolytic action of insulin. (A) Livers from Gαi2-deficient and Gαi3-deficient mice were perfused in situ in the presence of insulin (35 nM for 60 min). Shown are mean values of duplicate measurements ± SD in six Gαi3−/− mice and five Gαi2−/− mice. (Inset) [32P]ADP ribosylation demonstrates the loss of Gαi2 and Gαi3 in the respective gene-deficient mice. Shown is a representative autoradiograph (n = 3). (B) Statistical analysis of in situ mouse liver perfusion experiments. Shown is the mean relative inhibition of proteolysis ± SD in the indicated number of wt, Gαi2-deficient, and Gαi3-deficient mice treated with insulin (35 nM for 60 min; black bars), phenylalanine (2 mM for 30 min; hatched bars), and/or pretreatment with PTX (100 ng/ml for 60 min). Statistical significance was assessed by paired Student's t tests.
In summary, our data clearly demonstrate an obligatory requirement for Gαi3 in the antiautophagic action of insulin and phenylalanine and suggest an essential function for Gαi3 on autophagosomal membranes. To date, no specific G protein-coupled receptors or effectors have been identified on intracellular membranes, and it is unclear whether the classical G protein-coupled receptor/G protein/effector paradigm applies on intracellular membrane compartments. Nevertheless, Gαi3 is known to be regulated intracellularly by the guanine nucleotide dissociation inhibitor GAIP-1 and by the regulator of G protein signaling AGS3, which have been involved in the control of early events in autophagy (17). Future work will be directed toward elucidating the upstream regulation and specific cellular functions of Gαi3 on autophagosomes. It remains to be demonstrated whether Gαi3 is involved in the regulation of autophagy under physiological and pathological conditions in the development of embryos and in adults.
Plasma membranes and autophagosomal membranes were isolated from mouse hepatocytes by subcellular fractionation as detailed in SI Experimental Procedures.
The generation and phenotypic characterization of Gαi2-deficient mice and the generation of Gαi3-deficient mice have been published (3, 34). Knockout mice and their corresponding littermate controls were maintained at the animal facility of the Heinrich-Heine-Universität.
The ADP ribosylation of PTX-sensitive Gi proteins was performed as described in ref. 34. Gi proteins were resolved by electrophoresis on 6 M urea/SDS/PAGE gels, blotted onto nitrocellulose (Hybond C extra; GE Healthcare, Fairfield, CT), and visualized with an FLA 5000 Fuji phosphoimager (Raytest, Straubenhardt, Germany). In parallel, Gi proteins were detected with a Gαcommon antibody (AS 8; ref. 35).
Antibodies were raised against the keyhole limpet hemocyanin-coupled C-terminal 14 aa of human/murine Gαi3, and an additional cysteine was introduced at the N terminus (C-DVIIKNNLKECGLY). The corresponding C-terminal sequence of human/murine Gαi2 (DVIIKNNLKDCGLF) differs only by 2 aa residues from human/murine Gαi3. The internal cysteine of the Gαi3 peptide was replaced with the isosteric α-aminobutyric acid to prevent coupling to activated keyhole limpet hemocyanin. Gαi3-specific IgGs were purified by using ammonium sulfate precipitation, ÁKTA-FPLC affinity chromatography (Amersham Pharmacia, Uppsala, Sweden) on a Gαi3 peptide column and ÁKTA-FPLC gel filtration on a Superdex 200 HR 10/30 column. The Gαi3 antibody was preabsorbed on an acetone powder generated from livers of Gαi3-deficient mice to further reduce nonspecific binding.
The isolation of mouse hepatocytes was performed according to published procedures (28) with modifications and will be described in detail elsewhere (K.K., A.G., B.N., unpublished data). Briefly, the livers of 8- to 12-week-old C57BL/6 or 129Sv male mice were perfused with a Ca2+/SO42−-free perfusion buffer (6 mM glucose, 115 mM NaCl, 25 mM NaHCO3, 5.9 mM KCl, 1.2 mM MgCl2 × 6 H2O, and 1.2 mM NaH2PO4 × H2O, pH 7.5) first without and then with addition of collagenase (0.33 mg/ml; type CLS II, 283 units/ml; Biochrom, Berlin, Germany). Liver tissue was mechanically dissociated, and nonparenchymal cells and cell debris were removed by three centrifugation steps (19 × g for 2 min at 4°C). Sedimented cells were seeded onto collagen (Collagen R; Serva, Heidelberg, Germany) and cultured in serum-free Williams E medium without l-glutamine. The cells essentially were pure hepatocytes with no detectable contamination by other hepatic cell types. Experiments were performed 3 days after hepatocyte isolation.
Hepatocytes and MEFs were lysed in 50 mM Hepes, 150 mM NaCl, 1% Nonidet P-40, 1% SDS, and 1% Triton X-100 (pH 7.4). One Complete EDTA-Free Inhibitor tablet (Roche, Penzberg, Germany) was added per 100 ml of buffer. Proteins (30–50 μg per lane) were separated on 6 M urea/SDS/PAGE gels and blotted onto nitrocellulose. The anti-Gαcommon antibody (AS 8; ref. 35) was diluted 1:150, the anti-Gαi3 antibody was diluted 1:5,000, and the anti-LC3 antibody was diluted 1:750.
Cumulative cellular autophagy was induced by amino acid starvation in Hanks' balanced salt solution (HBSS; Invitrogen/Gibco, Carlsbad, CA) for 2 hours in the presence of 10 μg/ml leupeptin and 10 μg/ml pepstatin. The effects of insulin (0.02–10 nM for 60 min) or phenylalanine (0.2 mM for 60 min) on autophagy were determined in the absence of protease inhibitors. Lysosomes were visualized by incubation with LysoTracker Red DND-99 (75 nM for 30 min; Invitrogen/Molecular Probes, Carlsbad, CA). Hepatocytes were fixed in 4% paraformaldehyde (20 min) and permeabilized with 0.5% Triton X-100 (5 min). Nonspecific binding sites were blocked with 3% BSA. Antibodies were diluted in 1% BSA to the following ratios: anti-Gαi3, 1:1,000; rabbit anti-LC3, 1:300; goat anti-LC3 clone F14, 1:200 (Santa Cruz Biotechnology, Santa Cruz, CA); anti-early endosomal antigen 1 (EEA-1), 1:400 (BD Biosciences, Heidelberg, Germany); and anti-protein disulfide isomerase (PDI), 1:300 (BD Biosciences). Secondary detection was performed with Alexa Fluor-coupled antibodies (Invitrogen/Molecular Probes). Nuclei were stained with DAPI (Sigma, Deisenhofen, Germany), and cells were analyzed on a LSM 510 Meta confocal microscope (Carl Zeiss, Jena, Germany). Single optical slices of ≈0.5 μm are shown.
Livers of 8- to 12-week-old male mice with a body weight of 20–25 g were perfused in situ essentially as described for the rat in ref. 28. Briefly, proteolysis was determined in separate perfusion experiments as 3H-label release from mice that had been injected i.p. with 5 μCi of l-[4,5-3H]leucine 16 h before the perfusion experiment (1 Ci = 37 GBq). Livers from fed mice were prelabeled in vivo by i.p. injection of 5 μCi of [3H]leucine, and 3H-label release into the effluent was monitored as a measure of hepatic proteolysis. Because of different labeling of the animals in vivo, the release of radioactivity was set to 100% during baseline conditions at 100 min of perfusion time. All perfusion experiments were performed with approval of the local animal care and use committee and supervising state authorities.
We (A.G. and B.N.) gratefully dedicate this paper to our former mentor Günter Schultz, Berlin, on the occasion of his 70th birthday. We thank Olga Felda for genotypic analysis and maintenance of the mouse colony, Nicole Eichhorst and Markus Mroz for expert technical assistance with in situ liver perfusion experiments, Dr. Brigitte Anliker for help with the figures, Dr. Yasuo Uchiyama (Osaka University Graduate School of Medicine, Osaka, Japan) for the generous gift of LC3 antibodies, and Dr. Eeva-Liisa Eskelinen (University of Helsinki, Helsinki, Finland) for the GFP-LC3 plasmid. This work was supported in part by Collaborative Research Centers, e.g., SFB575, of the German Research Foundation, and National Institutes of Health Grant DK-19318 (to L.B.).
Author contributions: A.G., R.P.P., L.B., and B.N. designed research; K.K., R.P.P., K.P., S.v.D., K.S., and V.D. performed research; S.v.D. and D.H. contributed new reagents/analytic tools; A.G., R.P.P., S.v.D., D.H., L.B., and B.N. analyzed data; and A.G., L.B., and B.N. wrote the paper.
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/cgi/content/full/0611434104/DC1.
