E XP ER I ME NTAL C E LL RE S E ARCH

] (]]]]) ]]]–]]]

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/yexcr

Research Article

Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs Paula Szalaia, Linda Korseberg Hagenb, Frank Sætrea,b, Morten Luhra, Marianne Sponheimb, Anders Øverbyeb, Ian G. Millsa,c,d, Per O. Seglena,b,n, Nikolai Engedala,nn a

Prostate Cancer Research Group, Centre for Molecular Medicine Norway, Nordic EMBL Partnership, University of Oslo and Oslo University Hospital, N-0318 Oslo, Norway b Cell Biology Section, Institute for Cancer Research, Oslo University Hospital, N-0310 Oslo, Norway c Department of Urology, Oslo University Hospital, N-0310 Oslo, Norway d Department of Cancer Prevention, Institute for Cancer Research, Oslo University Hospital, N-0310 Oslo, Norway

article information

abstract

Article Chronology:

LC3, a mammalian homologue of yeast Atg8, is assumed to play an important part in bulk

Received 29 August 2014

sequestration and degradation of cytoplasm (macroautophagy), and is widely used as an

Received in revised form

indicator of this process. To critically examine its role, we followed the autophagic flux of LC3

9 January 2015

in rat hepatocytes during conditions of maximal macroautophagic activity (amino acid deple-

Accepted 6 February 2015

tion), combined with analyses of macroautophagic cargo sequestration, measured as transfer of the cytosolic protein lactate dehydrogenase (LDH) to sedimentable organelles. To accurately

Keywords: Autophagy Density gradient GABARAP LC3 antibody Sequestration Subcellular fractionation

determine LC3 turnover we developed a quantitative immunoblotting procedure that corrects for differential immunoreactivity of cytosolic and membrane-associated LC3 forms, and we included cycloheximide to block influx of newly synthesized LC3. As expected, LC3 was initially degraded by the autophagic-lysosomal pathway, but, surprisingly, autophagic LC3-flux ceased after
2 h. In contrast, macroautophagic cargo flux was well maintained, and density gradient analysis showed that sequestered LDH partly accumulated in LC3-free autophagic vacuoles. Hepatocytic macroautophagy could thus proceed independently of LC3. Silencing of either of the two mammalian Atg8 subfamilies in LNCaP prostate cancer cells exposed to macroautophagyinducing conditions (starvation or the mTOR-inhibitor Torin1) confirmed that macroautophagic sequestration did not require the LC3 subfamily, but, intriguingly, we found the GABARAP subfamily to be essential. & 2015 Elsevier Inc. All rights reserved.

Abbreviations: 3MA, 3-methyladenine; Baf (or Baf A1), bafilomycin A1; BSA, bovine serum albumin; GABARAP, γ-aminobutyric acid receptor-associated protein; LC3, microtubule-associated protein 1 light chain 3; LDH, lactate dehydrogenase; MEFs, mouse embryonic fibroblasts; MMapp, apparent molecular mass; mTOR, mammalian target of rapamycin; NDK, nucleoside diphosphate kinase; TG, thapsigargin n

Corresponding author at: Centre for Molecular Medicine Norway (NCMM), University of Oslo, P.O. Box 1137 Blindern, N-0318 Oslo, Norway. Corresponding author. E-mail addresses: [email protected] (P.O. Seglen), [email protected] (N. Engedal).

nn

http://dx.doi.org/10.1016/j.yexcr.2015.02.003 0014-4827/& 2015 Elsevier Inc. All rights reserved.

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

2

E XP E RI ME N TAL CE L L R ES E ARC H

Introduction Autophagic sequestration is performed by specialized organelles called phagophores, which are essentially compressed membrane cisterns capable of expanding to envelop pieces of whole cytoplasm (macroautophagy) or individual cytoplasmic elements (selective/exclusive autophagy) for shipment to lysosomes where the sequestered cargo is degraded [1]. The selective autophagies are usually named according to their cargo (mitophagy, pexophagy, reticulophagy, xenophagy, aggrephagy, lipophagy etc.) [2]. Macroautophagy is generally regarded as a nonspecific bulk process, although it has been suggested that it may encompass a selective sequestration of certain cytosolic proteins [3]. In yeast, phagophores are manufactured at distinct cytosolic locations (pre-autophagosomal structures, or PAS) [4], whereas mammalian phagophore assembly has been associated with various organelles such as mitochondria, endoplasmic reticulum or Golgi cisterns [5–7], which may be collectively referred to as phagophore assembly sites (PAS) to conform with the yeast terminology. At these sites, the products of various autophagyrelated genes (Atgs) and other proteins (e.g., WIPIs) [8] cooperate to assemble membrane elements from various sources (endoplasmic reticulum, mitochondria, plasma membrane etc.) and proteins into functional phagophores, a process that culminates in the conjugation of Atg8 (or its mammalian orthologues, LC3s and GABARAPs) to phosphatidylethanolamine (PE) in the phagophore membrane [9]. LC3 was first described as an 18-kDa protein found in preparations of microtubule-associated proteins (MAPs) from bovine brain [10]. Following its recognition as a MAP-1 subunit [11], it was named MAP-1 light chain 3 (LC3). The protein was purified from bovine brain and partially sequenced, and eventually cloned from the rat with a deduced 142-amino acid sequence, a predicted size of 16.4 kDa and a predicted pI of 9.2 [12]. Five isoforms from four genes have been described in humans: LC3A (with the alternative splice variants LC3A1 and LC3A2), LC3B1 and LC3B2 (which are virtually identical in their gene coding region) and LC3C [13,14]. LC3A and LC3B are also found in the rat, both with wide tissue distributions, but only LC3B is significantly expressed in liver tissue [15]. Since LC3B is identical to the previously described LC3, the name “LC3B” is a junior synonym that should be generally discarded unless used to explicitly differentiate it from other LC3 isoforms. When Atg8/Aut7 was identified as a gene required for yeast autophagy and subsequently cloned, it was pointed out that its protein sequence was 28% identical to that of rat LC3 [16]. Soon after, LC3 was shown to be present in animal cells both as a soluble,
18-kDa (MMapp) form (LC3-I), generated by proteolytic cleavage from a slightly larger precursor, and as a structureassociated,
16-kDa (MMapp) form (LC3-II) [17], the latter being derived from LC3-I by lipidation (conjugation to PE) [18,19]. Upon amino acid starvation, the amount of LC3-II associated with autophagic organelles was found to increase rapidly, subject to strong antagonism by autophagic sequestration inhibitors such as 3-methyladenine and wortmannin [17]. LC3-II would thus seem to be suitable as an autophagic organelle marker, and the conversion of LC3-I to LC3-II, detectable by immunoblotting, might conceivably serve as an autophagy assay. However, it should be emphasized that LC3 lipidation is a measure of

] (]]]]) ]]]–]]]

phagophore maturation rather than of autophagic sequestration activity (autophagosome formation); the latter process can only be measured by an appropriate cargo assay [1,20]. Furthermore, since LC3-I displays much weaker immunoreactivity than LC3-II [18], it is not only difficult to detect and quantify, but uncorrected LC3-II/LC3-I ratio values hold no quantitative meaning. It has been suggested that LC3-II values should instead be related to some invariant household protein such as tubulin [20], but the distinction between altered lipidation and changes in total LC3 levels would then be lost. Besides the LC3s, mammalian cells harbor another subfamily of orthologues to yeast Atg8, i.e., the GABARAP (γ-aminobutyric acid receptor-associated protein) family. Its family members GABARAP, GABARAPL1 and GABARAPL2 are, like LC3, subject to constitutive proteolytic processing as well as to enhanced lipidation under autophagy-inducing conditions [18,19,21,22]. The GABARAPs have received much less attention than the LC3s, but both families have been reported to be involved in cargo recognition during selective autophagy [23] and in autophagosome formation [24,25]. In the latter process, LC3 has been associated with phagophore expansion, whereas the GABARAPs have been associated with phagophore closure [25]. However, in contrast to LC3, the GABARAPs have not found general use as autophagy markers. An increase in the steady-state level of LC3-II can result from enhanced phagophore/autophagosome biogenesis, or from an inhibition of autophagosome turnover. Therefore, measurements of LC3 turnover have been proposed to provide a more dynamic picture of “autophagic flux” [26,27], but since LC3 is part of the autophagic “cart” rather than of its cargo, its turnover tells little about actual autophagic activity. The use of fluorescence microscopy to monitor the transfer of LC3 from a diffuse, cytosolic to a dot-like distribution (assumed to represent organelle-associated LC3-II) [20] also has its limitations: only the dots can be quantified, the lipidation step thus being indistinguishable from overall changes in the quantities of autophagic organelles. Clearly, quantitative measurements of both LC3-I and LC3-II, and a correction for their different immunoreactivities [18] would be required for a meaningful analysis of autophagic-lysosomal LC3 flux and its relation to macroautophagic cargo flux. In the present study, we have, therefore, developed an immunoblotting strategy that allows a satisfactory and comparable quantification of both LC3 forms. To simplify the analysis of LC3 dynamics, we have used the protein synthesis inhibitor, cycloheximide, to block de novo influx of LC3, allowing us to measure the turnover of this short-lived protein in isolated rat hepatocytes. By comparing the autophagiclysosomal flux of LC3 with actual macroautophagic sequestration activity, using the enzyme LDH as a cytosolic cargo marker [28], we have been able to demonstrate that macroautophagic-lysosomal cargo flux can take place in the absence of an autophagic-lysosomal LC3 flux. Furthermore, the accumulation of sequestered LDH in LC3-free autophagic vacuoles suggested that macroautophagy may not require LC3 at all. This conclusion was confirmed by an RNAimediated approach in LNCaP prostate cancer cells: Knock-down of the LC3s did not reduce macroautophagic cargo sequestration induced by either starvation for serum and amino acids or by the mTOR-inhibitor Torin1. In contrast, autophagic bulk sequestration of cytosolic cargo was strongly suppressed by selective knock-down of the GABARAPs, indicating that this subfamily of human Atg8 orthologues, but not the LC3 subfamily, serves an essential function in macroautophagy.

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

E X PE R IM EN TA L C ELL R E S EA RC H

Materials and methods Antibodies and chemicals Our in-house polyclonal anti-LC3 antibody LC3abN1, raised in rabbits against a cysteinylated N-terminal 14-amino acid peptide (PSEKTFKQRRSFEQ-C; also used as a blocking peptide to prevent anti-LC3 immunoreactivity), was made to order by BioGenes. The “classical” rabbit polyclonal anti-LC3 antibody [17] was a kind gift from Dr. T. Yoshimori (Mishima, Japan). Rabbit antibodies raised against LC3B (CST #2775), Atg5 (CST #8540), GAPDH (CST #2118) as well as a polyclonal rabbit antibody against cytochrome c oxidase (COX-IV; CST#4844) and anti-rabbit IgG (CST#7074) were from Cell Signaling Technology. Anti-GABARAP (PM037) and antiGABARAPL2 (PM038) antibodies were from MBL Int. Corp., and anti-GABARAPL1 (ab86497) from Abcam. Torin1 was purchased from RnD Systems, dissolved in DMSO to a 0.5 mM stock solution and kept as aliquots at  20 1C. Igepal CA-630 was purchased from Sigma-Aldrich; Iodixanol (Optiprep, #1114542) from Axis-Shield; Purdenz (#AN-6310) from Accurate Chem. Sci. Corp., and Supersignals west Femto chemiluminescent HRP substrate from Thermo Scientific Pierce (#34094). Human recombinant HisAtg4b (#E-400) was from Boston Biochem. The sources of other chemicals were as recently listed [29].

Cell lines Atg5  /  MEFs (RBC2711) and wild-type control Atg5þ/þ MEFs (RBC2710) were a kind gift from Dr. N. Mizushima [30], whereas LNCaP cells were obtained from ATCC (CRL-1740). MEFs were cultured in DMEM (Gibco 41966) and LNCaPs in RPMI 1640 (Gibco 21875), both supplemented with 10% fetal bovine serum (FBS; Gibco 10500) and 5% CO2 at 37 1C. In experiments involving starvation for amino acids and serum, cells were washed with glucose-containing Earle's balanced salt solution (EBSS; Gibco 24010 containing 1 g/L D-glucose) followed by incubation in fresh EBSS.

] (]]]]) ]]]–]]]

3

Measurements of hepatocellular ATP levels and integrity Hepatocellular ATP levels were measured by a luciferin/luciferase assay as previously described [34]. The integrity of the hepatocytes was measured as the retention of total cellular LDH (measured as described below for the autophagy assay) relative to the amount present at the start of incubation.

Cell disruption and subcellular fractionation Hepatocytes were washed twice in electrolyte-free, isotonic (10%) sucrose and 2.4 ml of suspended cells (six pooled 400-μl samples) was electrodisrupted [35] at 0 1C by a single high-voltage pulse (2 kV; 1.2 μF). The total cell disruptate was either lysed directly in SDS buffer or separated into a “cell corpse” fraction that contained all sedimentable cellular components, and a soluble “cell sap” fraction (differing from conventional cytosol by being prepared without prior homogenization) by low-speed centrifugation (4000g for 30 min at 4 1C) above an 8% buffered (isotonic) Nycodenz cushion [28]. All samples were frozen and stored at 70 1C. For subcellular fractionation on isotonic iodixanol (Optiprep) density gradients, freshly prepared cell corpses from
50 mg hepatocytes were Dounce-homogenized with 1.5 ml ice-cold homogenization buffer (HB; 0.25 M sucrose, 10 mM HEPES, 1 mM EDTA, pH 7.3). The homogenate was layered on top of an 11-ml gradient of 0–30% iodixanol (in HB) and centrifuged at 4 1C in a Sorvall SW40 rotor, first for 20 min at 5000 rpm, then for 3 h at 40,000 rpm, and left overnight at 4 1C. Twenty 0.63-ml fractions were collected by upwards displacement by a dense fluid (Purdenz), the fraction densities were determined by refractive index measurements, and 200 μl of each fraction was lysed with 50 μl of proteinase inhibitor-containing 5x lysis buffer to final concentrations of 1% SDS, 5 mM EDTA, 5 mM EGTA, 10 mM sodium pyrophosphate, 0.3 mM leupeptin, 10 μM pepstatin A, 15 μM E-64, 50 μM bestatin, 0.8 μM aprotinin, 3 mM 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF) and 20 mM Tris, pH 7.2. Each lysate was concentrated to 100 μl on a spin column, 20 μl of sample buffer was added, and
8 μl (volume adjusted according to measured protein content) was used for SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting.

Animals and hepatocyte isolation siRNA transfection Hepatocytes were isolated by two-step collagenase perfusion of isolated livers from 18-h starved male Wistar rats (250–300 g; Harlan UK Ltd.) and purified as previously described [31]. The hepatocytes were suspended and maintained at 0 1C in buffered saline (suspension buffer) [31] containing extra Mg2þ (2 mM final conc.), with pyruvate (15 mM) as an energy source. For experiments, 2 ml of this cell suspension (
20 mg wet mass) was incubated at 37 1C in gently shaking 6-cm suspension dishes (Sarstedt #83.1801.002). Unless otherwise indicated, incubations were performed without amino acids (and serum) in the medium in order to induce maximal macroautophagic activity in the hepatocytes. Under such amino acid starvation conditions, cargo sequestration proceeds at a high rate for several hours [28]. In one experiment (Fig. 2C) the medium contained a balanced amino acid mixture [32], previously shown to strongly suppress hepatocytic cargo sequestration and protein degradation [32,33].

Small interfering RNAs (siRNAs) were introduced into LNCaP cells by reverse transfection in 6-well plates (BD Falcon 353046). The plates were coated with poly-D-lysine (Sigma P6407) by incubation with a 2.5 μg/ml solution (in water) for 1 h at 37 1C, followed by two washes with PBS. siRNAs were diluted with Opti-Mem reduced serum medium (Gibco 11058) to a total volume of 332.5 μl (per well), mixed with 7.5 μl Lipofectamine RNAiMax (Invitrogen 13778) and transferred to a well on the poly-lysinecoated plate. After 20–45 min incubation at room temperature,
8  105 LNCaP cells in 2 ml prewarmed (37 1C) complete medium (RPMI1640 with 10% FBS) were added to each well. Individual siRNAs targeting Atg8 family members were used at a final concentration of 10 nM each and mixed with other targeting siRNAs or a non-targeting control siRNA to a final total concentration of 50 nM per well. Experiments were initiated after 48 h of transfection. The following Silencers Select siRNAs (Ambion)

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

4

E XP E RI ME N TAL CE L L R ES E ARC H

were used: Negative Control #1 (4390843), siLC3A (s39156; targets both LC3A1 and LC3A2), siLC3B (s37748), siGABARAP (s22361), siGABARAPL1 (s24332) and siGABARAPL2 (s223228).

Quantitative real-time RT-PCR Cells were harvested with Accumax (Sigma A7089) and washed in cold PBS/0.5% FBS before total RNA extraction using the illustraRNAspin Mini Isolation Kit (GE Healthcare 25-0500-71). Total RNA was reverse-transcribed using the SuperScript VILO Master Mix (Applied Biosystems 11755250). Quantitative real-time RTPCR was performed using TaqMan gene expression assays (Appl. Biosys. 11755250) with TaqMan fast advanced PCR master mix (Appl. Biosys. 4444558). PCR amplification was done in triplicate series with the ABI 7900HT FAST sequence detection system (Appl. Biosys.). The cycling conditions were 50 1C for 2 min, 95 1C for 10 min, 40 cycles at 95 1C for 15 s, and 60 1C for 1 min. Relative transcript levels were determined by the comparative CT method [36] and normalization to the geometric mean CT value of GAPDH and TBP [37]. The following TaqMan gene expression assay probes (Appl. Biosys.) were used: LC3A1 (Hs00261291_m1), LC3A2 (Hs00738808_m1), LC3B1 (Hs00797944_s1), LC3B2 (Hs04195455_u1), LC3C (Hs01374916_m1), GABARAP (Hs00925899_g1), GABARAPL1 (Hs00740588_mH), GABARAPL2 (Hs00371854_m1), GAPDH (Hs99999905_m1) and TBP (Hs99999910_m1).

Atg4 treatment For delipidation of LC3-II by Atg4,
20 mg washed hepatocytes were lysed for 60 min at 0 1C in 0.7 ml Igepal buffer (1% Igepal; 630, protease inhibitors at the concentrations given above, 10 mM Tris–HCl, 150 mM NaCl, pH 7.5). Atg4 was activated by preincubation for 15 min at 37 1C with dithiothreitol (1.75 μM Atg4, 10 mM DTT, 0.5 mM EDTA, 1% Igepal, 130 mM NaCl, 25 mM HEPES, pH 8.0) and added to 25 μl of hepatocyte lysate to a final concentration of 0.1 μΜ. After incubation of this mixture for 10 min at 37 1C, the delipidation reaction was stopped by the addition of buffered SDS to a final concentration of 1%.

Immunoblotting, image analysis and quantification Hepatocyte samples containing 20 μg of protein were separated by SDS gel electrophoresis for 45 min at 200 V, using an SDS running buffer (25 mM Tris, 192 mM glycine and 0.1% SDS) on Criterion™ precast gels (567-1125, Bio-Rad Laboratories ) with an associated midi-tank system (Criterion™ Cell, Bio-Rad Laboratories). Molecular mass standards were included in all gels (Novexs Sharp, LC5800, Life Technologies). The separated proteins were transferred onto PVDF blotting membranes (ISEQ00010, Millipore) using a wet-blotting cell (Criterion™ Bio-Rad Laboratories) and a CAPS transfer buffer (10 mM CAPS buffer, pH 11, with 2% ethanol). Following transfer, membranes for detection of LC3 were fixed in 100% methanol for 3 min and then washed in mqH2O and TBS-T (20 mM Tris, 0.8% NaCl, 0.1% Tween 20, pH 7.6) before blocking. The membranes were blocked by treatment with 5% (w/v) dry milk in TBS-T for 60 min. For immunoblotting, the membranes were first incubated overnight at 4 1C with the respective primary antibodies, usually diluted with TBS-T containing 5% (w/v) dry milk. After washing three times with TBS-T, the membranes were incubated for 1 h at

] (]]]]) ]]]–]]]

room temperature with the respective secondary antibodies conjugated to horseradish peroxidase, usually diluted with TBST containing 5% (w/v) dry milk, and washed three more times with TBS-T. Bound secondary antibodies were visualized using either the Immobilon Western Chemiluminescent HRP Substrate (WBKLS0500, Millipore) or, for increased sensitivity, Supersignals west Femto chemiluminescent HRP substrate. Chemiluminescent signals from immunoblots were detected and quantified using the ChemiGenius 3 imaging system (Syngene, Cambride, UK) and the associated software, GeneSnap and GeneTools. Proprietary image files were converted to TIF-file format for image presentation using the Adobe Photoshop software. With MEFs and LNCaP cells, preparation of protein extracts, immunoblotting and visualization of immunoreactive bands were generally done as previously described [38], with exception of the GABARAPs. For immunodetection of the latter, the membranes were incubated with the primary antibody solutions for 2 h at room temperature rather than overnight at 4 1C. Furthermore, Supersignals west Femto chemiluminescent HRP substrate was used for enhanced visualization of GABARAPL1 bands. The following dilutions were used for antibodies recognizing LC3 (CST #2775), 1/1000; GABARAP, 1/2000; GABARAPL1, 1/1000; GABARAPL2, 1/1000 and GAPDH, 1/4000.

Autophagic sequestration assays With rat hepatocytes, autophagic (macroautophagic) capacity was measured as the ability of the freshly isolated cells to sequester cytosolic LDH into sedimentable (autophagic) organelles when incubated at 37 1C under amino acid/serum-free conditions in the presence of leupeptin (0.3 mM), an inhibitor of intralysosomal degradation [28]. Autophagic (macroautophagic) cargo sequestration activity was measured similarly, with the additions/conditions specified in each experiment. Both autophagic capacity and autophagic activity are given as the net change during the indicated time period (i.e., with subtraction of the background at the time of leupeptin addition), expressed as the percentage of total cellular LDH sequestered per hour. In Fig. 5B the total amounts of sedimentable LDH are shown, i.e., without background subtraction. With MEFs and LNCaP cells, macroautophagic cargo sequestration activity was measured by a downscaled and somewhat simplified modification of the LDH sequestration assay, recently described in detail [39]. Importantly, these cells were incubated with bafilomycin A1 rather than with leupeptin, since the former was found to be a superior inhibitor of LDH degradation in LNCaP cells [39]. Briefly, subconfluent (70%) cells in 6-well plates were incubated in complete medium (with 10% FBS) with or without DMSO (as a vehicle control), or with the addition of Torin1 or incubation in EBSS (subsequent to washing in EBSS) in the presence of bafilomycin A1 (10 nM for MEFs, 200 nM for LNCaP cells) for determination of induced autophagic activity. After the experimental incubation (3–4 h), cells were harvested with Accumax, washed twice with 1 ml 10% sucrose containing 1% BSA, resuspended with 400 μl 10% sucrose/0.2% BSA and electrodisrupted at 2 kV/cm and 1.2 μF in a 1  1  5 cm electrode chamber. Following dilution with 400 μl phosphate-buffered sucrose (100 mM sodium phosphate, 2 mM dithiothreitol, 2 mM EDTA and 1.75% sucrose, pH 7.5), 600 μl of the disruptate was further

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

E X PE R IM EN TA L C ELL R E S EA RC H

diluted with 900 μl resuspension buffer (RSB; 50 mM sodium phosphate, 1 mM EDTA, 1 mM DTT) containing 0.5% BSA and 0.01% Tween-20 and centrifuged for 7 min at 5000g (MEFs) or for 30 min at 20,000g (LNCaP). The supernatant was thoroughly removed by suction. The pellet and the remaining total-cell disruptate (200 μl) were freeze-thawed ( 80 1C) before measurement of sedimentable and total cellular LDH activity, respectively: The pellet was dissolved in 500 μl RSB containing 1% Triton X-405 (TX-405), whereas the total-cell disruptate was diluted with 200 μl RSB/2% TX-405. Following a short centrifugation to eliminate cell debris (5 min at
21,000g), LDH activity was measured by the decline in NADH absorbance at 340 nm in a multi-analyzer (MaxMat PL-II, Erba Diagnostics) using an LDH assay kit (RM LADH0126V, Erba Diagnostics). The percentage of sedimented LDH relative to the total-cell disruptate was calculated for each treatment condition, and the corresponding percentage for untreated cells (considered as background) was subtracted to obtain the percentage of sequestered LDH in the treatment condition. The resulting net value was divided by the incubation time to obtain the rate of LDH sequestration (%/h) during the experimental treatment period.

] (]]]]) ]]]–]]]

5

steps were background correction (Minimum Ranking) and set ting a threshold (Max Entropy), and finally quantifying LC3-positive “dots” with the procedure “Analyze Particles”. Images for presentation were processed with the software ZEN 2008 (Carl Zeiss Micro Imaging GmbH, Jena, Germany) and Photoshop CS4 (Adobe, Mountain View, CA).

Mass spectrometry Protein identification by MALDI-TOF MS (tryptic peptide fingerprinting) was performed in-house as previously described [3]. Identification by electrospray ionization and ion trap tandem mass spectrometry (ESI-IT-MS/MS) was done in-house with an Agilent MSD XCT or with a Thermo Fisher LTQ Orbitrap [40].

Statistical analyses For all experiments where tests of significance were performed, approximate normal distributions of the data were assumed, and two-tailed Student's t-test was employed, using Excel.

Protein measurement Total protein in hepatocyte extracts was measured by the biuret method on a MaxMat PLII (Erba Diagnostics). Absorbance was read at 550 nm wavelength using the accompanying Total Protein kit (RMPROT0125V). Protein concentrations in extracts from MEFs and LNCaP cells were determined by the BCA assay (Pierce 23227).

Fluorescence microscopy Cells were incubated for 4 h at 37 1C, with or without the addition of 100 mM cycloheximide, washed twice in PBS (74g for 5 min at 4 1C) and then resuspended in suspension buffer. Aliquots of
1  104 cells were sedimented (750 rpm for 5 min) onto microscopic slides using a Cytospin cytocentrifuge (Shandon Scientific Ltd., Cheshire, UK) before fixation in 4% formaldehyde (PIER28908, Thermo Scientific) for 15 min at RT. Slides were washed three times in PBS (pH 7.5), rinsed twice in water, dried and stored frozen at 20 1C. For detection of LC3, cells were permeabilized and blocked with 0.3% Triton X-100 in TBS (TBSþ) containing 5% goat normal serum (16210-064, Gibco) for 60 min at RT. Cells were then incubated with antibody against LC3 (2775, Cell Signaling Technology) diluted 1:200 in dilution buffer (TBS with 1% BSA and 0.3% Triton X-100) over night at 4 1C, washed 3  5 min in TBSþ and finally incubated with Alexa Fluor 488tagged secondary antibody diluted to 2.5 mg/ml in dilution buffer (A-11034, Life Technologies) for 2 h at RT. After 3  5 min washes in TBSþ, slides were mounted with Prolong Gold antifade with Dapi (36934, Life Technologies). The cells were observed with a Zeiss LSM 710 confocal microscope (Carl Zeiss Micro Imaging GmbH, Jena, Germany) equipped with an Argon Laser Multiline (458/488/514 nm), a DPSS-561 10 (561 nm), a Laser diode 405-30 CW (405 nm), and a HeNe-laser (633 nm). The objective used was a Zeiss Plan-Apochromat 63x/1.40 Oil DIC M27. Images were analyzed with the open source software ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/, 1997–2014.) where the processing

Results Quantitative measurement of immunoblotted LC3-I and LC3-II In order to accurately study the dynamics of LC3 protein levels in relation to macroautophagy we first examined the identities of the two bands that are generally assumed to represent LC3-I and LC3-II during western blotting for LC3 (bands with apparent molecular masses of
16 and
14 kDa for LC3-I and LC3-II, respectively). To our surprise, we found that our in-house antibody LC3abN1 as well as a commercial anti-LC3 antibody crossreacted with a protein exhibiting the same electrophoretic mobility as LC3-I, identified by mass spectrometry as an isoform of nucleoside diphosphate kinase (described in the Supplement; Suppl. Figs. S1 and S2). Fortunately the cross-staining could be eliminated by omission of the heating step in the sample preparation procedure, allowing specific LC3 staining (Suppl. Fig. S1B). With the aid of chemical signal amplification, satisfactory immunostaining of both LC3 forms (the non-lipidated LC3-I and the lipidated LC3-II) from rat hepatocytes could thus be obtained (see Supplement). However, if these LC3 forms differ in their immunoreactivity, as is probably a general phenomenon regardless of the antibody used [18], a correction for differential staining would be required to enable quantitative studies of LC3 dynamics. To find a suitable correction factor, hepatocyte extracts were incubated with the specific LC3-delipidating enzyme, Atg4 [18] prior to gel electrophoresis and blotting, and the loss in LC3-II staining (quantified by densitometric scanning) was compared with the gain in LC3-I staining. As shown in Fig. 1A, the LC3-II loss exceeded the LC3-I gain by a factor of
4. As a consequence, all LC3-I values are now routinely multiplied by four to obtain a valid quantitative comparison with LC3-II. A calibration curve, based on extract dilution (Fig. 1B), is routinely included in all LC3 blots to ensure that the scan values are within a linear detection range.

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

6

E XP E RI ME N TAL CE L L R ES E ARC H

Extract # 1 - Atg4 23

+ Atg4 35

] (]]]]) ]]]–]]]

Atg4-treated 10 min at 37 ºC

Extract # 2 - Atg4

Extract # 1

+ Atg4

35

47

LC3-I gain: 35 – 23 = 12 LC3-II loss: 64 – 20 = 44 Staining ratio: 44/12 = 3.7

8

LC3-I gain: 57 – 8 = 49 LC3-II loss: 47 – 35 = 12 Staining ratio: 49/12 = 4.1

LC3-I

20

57

RANGE LINEARITY CHECK

LC3-I

LC3-II 100 50 25 12.5 % Extract dilution (% of undiluted)

LC3-II CONCENTRATION (relative value)

64

100

40 R2 = 0.9976

80

30 60 20 40 LC3-I

20

10

LC3-II

20

40

60

80

LC3-I CONCENTRATION (relative value)

Extract # 2

LC3-II

100

EXTRACT DILUTION (% of undiluted)

Fig. 1 – Quantitative immunoblotting of LC3-I and LC3-II. (A) Hepatocytes freshly prepared (extract #2) or incubated for 200 min at 37 1C with 10 mM NH4Cl (extract #1) were lysed and incubated with or without Atg4 (a delipidation enzyme that will convert LC3-II to LC3-I) prior to immunoblotting with our in-house anti-LC3 antibody. The stained bands were quantified by densitometry to yield the (arbitrary) values provided next to the bands in the left panel; all values fell within the linear range indicated by the calibration curve obtained by extract dilution (B). The gain-and-loss calculations in the right panel (yielding LC3-II/I ratios of 3.7 and 4.1 for the two extracts shown, and 4.2 for a third preparation) suggested that LC3-I values should be multiplied by four to be quantitatively comparable to the LC3-II values. (B) Example of a range linearity control: Extracts were diluted and blotted for LC3 as shown in the left panel, and the bands were quantified for calculation of the coefficient of determination (R2) from the linear regression curve (right panel).

Turnover of LC3 after inhibition of protein synthesis with cycloheximide To simplify LC3 pool kinetics, we decided to use the protein synthesis inhibitor, cycloheximide, to block any influx of de novo synthesized LC3. This makes our analysis unaffected by changes in LC3 gene expression, allowing protein turnover to be measured directly as the decline in protein levels, as first shown in perfused rat livers [41,42]. In order to induce maximal autophagic activity, we incubated freshly isolated hepatocytes in an amino acid-free buffer with pyruvate present as an energy source [28,33,34]. As shown in Fig. 2A, treatment of freshly isolated hepatocytes with cycloheximide for up to 4 h had no detectable effect on their viability (structural integrity), which declined only slowly during incubation at 37 1C in intermittently shaken petri dishes. The accompanying slow loss of cellular ATP appeared to be somewhat accelerated by cycloheximide (Fig. 2B), but was still maintained at 80% of the initial value when correcting for the viability loss. Since cycloheximide completely inhibits the incorporation of amino acids into hepatocytic proteins [43], the possibility of a secondary increase in intracellular amino acid concentrations to reach autophagy-inhibitory levels might have been considered. However, such an increase is not observed in hepatocytes [44],

which are functionally programmed to rapidly equilibrate amino acids and other metabolites with the extracellular medium/ plasma [45,46]. Furthermore, we have previously shown that the suppression of hepatocytic protein degradation by added amino acids is rapidly reversible even in the presence of cycloheximide [43]. As shown in Fig. 2C, the conversion of LC3-I to LC3-II by lipidation, an early autophagy-associated event, was effectively blocked by an added amino acid mixture [32], whereas cyloheximide had no such effect. The fact that LC3-II was rapidly degraded in the presence of cycloheximide, but not in the presence of amino acids (Fig. 2C), would additionally make it unlikely that amino acid-mediated cycloheximide effects play any role in the present study. Total LC3 levels (LC3-IþLC3-II) declined during 4 h of incubation both in control cells (Fig. 2D) and in cycloheximide-treated cells (Fig. 2E). The decline was, however, more rapid and sustained in the presence of cycloheximide, with an estimated half-life of
2 h for total LC3, placing it amongst the most shortlived hepatocellular proteins [42,47]. Both LC3-I and LC3-II levels fell over time, but the relatively rapid conversion of LC3-I to LC3-II precluded a net decline in the latter during the first hour of cycloheximide treatment (Fig. 2E). In the control cells, a resurgence of LC3 levels after 3 h was observed

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

INTRACELLULAR ATP (% of initial)

CELLULAR INTEGRITY (% LDH retention)

E X PE R IM EN TA L C ELL R E S EA RC H

100 + CHX

80 CONTROL

60 40 20

0

1

2

3

7

] (]]]]) ]]]–]]]

100 CONTROL

80 60

+ CHX

40 20

0

4

1

2

3

4

5

INCUBATION TIME (h at 37 ºC)

INCUBATION TIME (h at 37 ºC)

+ CYCLOHEXIMIDE

CONTROL LC3-I

15 kDa

LC3-II

AMINO ACIDS + CYCLOHEXIMIDE

AMINO ACIDS LC3-I

15 kDa

LC3-II

0

1

2

3

0

4

1

2

3

4

CELLULAR LC3 LEVELS (% of initial)

INCUBATION TIME (h at 37°C)

100

CONTROL

100

+ CYCLOHEXIMIDE

TOTAL (LC3-I + II)

80

80 60

t½

~ ~

60

2h

40

LC3-II

40

20

20

LC3-I

0

1

2

3

4

0

1

2

3

4

INCUBATION TIME (h at 37°C) Fig. 2 – Effect of cycloheximide on hepatocellular integrity, ATP content, LC3 lipidation and LC3 levels. Hepatocytes were incubated at 37 1C in buffered saline/pyruvate without (○) or with (●) 100 μM cycloheximide (CHX) for up to 5 h, and (A) cellular integrity (LDH retention) or (B) ATP content was measured. (C) Cellular contents of LC3-I (unlipidated form) or LC3-II (lipidated form) were indicated by immunoblotting of extracts from control or cycloheximide-treated hepatocytes, incubated in the absence (upper panels) or presence (lower panel) of an amino acid mixture. Quantified levels of LC3-I (△), LC3-II (○) or total LC3 (IþII; ●) were calculated on the basis of immunoblots from (D) control (mean7S.E./range of 2–5 experiments) or (E) cycloheximide-treated hepatocytes (mean7S.E. of 5–7 experiments). The half-life (t1/2) of LC3 was estimated to be
2 h (E).

(Fig. 2D), possibly reflecting a secondary induction of LC3 expression in response to the stress of amino acid starvation.

Effects of inhibitors of the autophagic-lysosomal pathway on LC3 lipidation and turnover Good immunodetection and quantification of both LC3 forms offer a possibility to distinguish between inhibitor effects on LC3 lipidation (phagophore assembly) and on LC3 turnover. Two time

points (100 and 200 min) were chosen for measurement of hepatocytic LC3-I and LC3-II during autophagy-inducing amino acid starvation in the presence of cycloheximide. As shown in Fig. 3, the two LC3 forms were present at approximately equal levels in freshly isolated hepatocytes. The levels of LC3-I (open bars) then fell rapidly in control cells, partly due to its initial lipidation to LC3-II (dotted bars), resulting in a moderately elevated LC3-II/I ratio at 100 min (from 0.86 to 1.47; top left panel). During the next 100 min the two forms declined

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

8

E XP E RI ME N TAL CE L L R ES E ARC H

CONTROL

3-METHYLADENINE

] (]]]]) ]]]–]]]

THAPSIGARGIN

NH4Cl

100

% OF INITIAL TOTAL LC3

90

(6)

(5)

(4)

**

80 LC3-II

60 50

min at 37°C

** II/I = 0.86 1.47 1.39 0

100

200

BAFILOMYCIN

% OF INITIAL TOTAL LC3

*** ***

***

LC3-I

100 90

*** **

20 10

***

***

40 30

(7)

***

*** **

70

*** **

(2)

*** ***

80 LC3-II

70

0.86 0.82 1.09 0

100

200

LEUPEPTIN

*** ***

0.93

2.53

2.46

0.90

2.70

2.88

0

100

200

0

100

200

PEPSTATIN

(2)

**

(2)

APROTININ

(1)

** *** ** **

60 50 40 30

**

LC3-I

20 10

min at 37°C

II/I = 1.02 2.34 1.65 0

100

200

0.98 3.05 0

100

2.38 200

0.97 1.95 0

100

2.29 200

0.85 1.35 0

100

1.58 200

Fig. 3 – Time-dependent changes in LC3 lipidation and degradation in rat hepatocytes treated with cycloheximide and autophagiclysosomal inhibitors. Hepatocytes were incubated for up to 200 min at 37 1C in buffered saline with pyruvate (15 mM) and cycloheximide (100 μM); without (control) or with autophagic-lysosomal inhibitors as indicated: 3-methyladenine (10 mM), thapsigargin (5 μM), NH4Cl (20 mM), bafilomycin A1 (10 μM), leupeptin (0.3 mM), pepstatin (0.15 mM) or aprotinin (0.1 mM). After incubation, whole-cell extracts were immunoblotted with our in-house anti-LC3 antibody, and the staining of LC3-I (open) and LC3-II (dotted) was quantified. The values in each treatment group are expressed as % of their respective 0-min totals7S.E. or range (error bars extending upwards for total LC3 and LC3-I; downwards for LC3-II) of the number of independent experiments given in parentheses at the top of each 0-min bar. LC3-II/LC3-I ratios (indicated at the bottom of the bars) refer to the time point indicated. **Po0.01; ***Po0.001 vs. control at the same time point according to Student's unpaired t-test.

more in parallel, with maintenance of the LC3-II/I ratio, indicating that LC3-I lipidation was largely at some kind of equilibrium with the autophagic-lysosomal flux and degradation of LC3-II. The overall LC3 turnover in control cells amounted to 63% over the whole 200 min incubation period. With 3-methyladenine (3MA), a well-established inhibitor of PI 3-kinases [48,49] and of autophagic sequestration activity [34], both lipidation (LC3-I decline) and overall LC3 degradation (total LC3 loss) were strongly, but incompletely (
50%) inhibited, indicating that some autophagic-lysosomal flux of LC3 was still maintained. Given the much stronger effect of 3MA on cargo (LDH) sequestration [28], see also Fig. 4B below, it is clear that LC3 turnover is not representative of the overall autophagic-lysosomal cargo flux.

Thapsigargin, a Ca2þ pump inhibitor, suppressed the turnover of LC3-II substantially (
60%), albeit, like 3MA, not as completely as it inhibits autophagic cargo sequestration [38,50]. Furthermore, since thapsigargin did not prevent LC3-I lipidation, it caused a considerable increase in the LC3-II/LC3-I ratio (from 0.93 to 2.53). Given that there is hardly any autophagic cargo sequestration in hepatocytes under these conditions [50], it is obvious that the LC3-II/I ratio is not very useful as an indicator of autophagic activity. Among the late-stage inhibitors of autophagic-lysosomal degradation (Fig. 3), pH-neutralizing agents such as the proton scavenger, NH4Cl [51,52], and the proton pump inhibitor, bafilomycin A1 [53], were clearly more LC3-II-protective than the proteinase inhibitor leupeptin, perhaps indicating proteolytic enzymes other

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

E X PE R IM EN TA L C ELL R E S EA RC H

than the leupeptin-sensitive cysteine proteinases to be involved in LC3 degradation. The aspartic proteinase inhibitor pepstatin and the serine proteinase inhibitor aprotinin had no effect, but without knowledge about the permeability or uptake of these compounds it may be premature to draw any conclusions. Neither NH4Cl nor bafilomycin A1 prevented the early (0–100 min) decline in LC3-I seen in control cells, indicating that they had little or no direct effect on LC3 lipidation. In contrast, the late (100–200 min) LC3-I decline (Po0.01 in the control cells) was significantly prevented by both inhibitors as well as by thapsigargin, perhaps reflecting the operation of some negative feedback mechanism. Since inhibitors of early autophagic sequestration steps (3MA, thapsigargin) as well as of the later processing and degradation steps (NH4Cl, bafilomycin A1, leupeptin) all suppress LC3 turnover, it seems clear that this protein is primarily degraded by the autophagic-lysosomal pathway during the first phase of amino acid starvation in rat hepatocytes.

To investigate how the rapid loss of LC3 under cycloheximide (Fig. 2E) might affect macroautophagic cargo sequestration, the macroautophagic capacity (maximal autophagic activity, measured as the net sequestration of LDH during a 2-h incubation at 37 1C in freshly added amino acid-free buffer with leupeptin present) was measured during consecutive (overlapping) 2-h periods in hepatocytes incubated in the continuous presence of cycloheximide. As shown in Fig. 4A, the autophagic capacity declined in the cycloheximide-treated cells with a “half-life” of
8 h, i.e., much more slowly than the
2 h half-life of LC3 (Fig. 2E). The maintenance of full autophagic sequestration capacity thus seems to require some new protein synthesis, but LC3 is clearly not rate-limiting. Inhibition of autophagic activity with 3MA (10 mM) or thapsigargin (5 μM) during 4 h of preincubation with cycloheximide did not significantly reduce the loss of

CONTROL

(6)

+ CYCLOHEXIMIDE

(8)

+ CHX

3

(12) (5) (5) (5)

t ½ = ~8 h

2 (10)

1 1

AUTOPHAGIC ACTVITY (%/h)

AUTOPHAGIC CAPACITY (%/h)

Macroautophagic sequestration capacity does not correlate well with cellular LC3 levels

CTR

3

2

9

] (]]]]) ]]]–]]]

0

0 0-2

1-3

2-4

3-5

0-2 h 4-6 h 4-6 h +3MA

4-6

0-2 h 4-6 h 4-6 h +3MA

MEASUREMENT PERIOD (h of hepatocyte incubation at 37 ºC)

Fig. 4 – Slow decline in macroautophagic capacity during cycloheximide treatment. (A) Time-dependent maintenance of autophagic capacity was followed by measuring net LDH sequestration during consecutive 2-h periods of leupeptin (0.3 mM) addition to control (CTR) hepatocytes (○; 0–2 and 4–6 h only) or to hepatocytes incubated continuously with CHX (●; 0–2 h, 1–3 h etc.). Each value is the mean7S.E. of the number of experiments given in parentheses. (B) 3MA-sensitivity of the late (4–6 h) LDH sequestration, recorded in the presence of leupeptin without or with 3MA (10 mM) included. Each bar shows the mean7S.E./range of 2–7 experiments.

Table 1 – Effect of lysosomal LC3 retention (reduced degradation) on autophagic sequestration capacity. Incubation with CHX (min)

LC3 at start of sequestration (200 min)

Sedimentable LDH (%) at start

Autophagic capacity (%/h)

0–100

100–200

200–300

LC3-I

LC3-II

LC3 total

(200 min)

(200–300 min)

None NH4Cl NH4Cl NH4Cl

None NH4Cl None 3MA

LPT LPT LPT LPT

15.571.3 (5) 20.871.1 (6) 20.772.4 (6) 28.372.0 (5)

22.271.9 60.073.6 38.772.4 41.072.8

37.871.8 80.874.1 59.374.1 69.274.0

1.8070.17 (3) 6.2370.77 (3) 4.2470.72 (3) 2.9170.55 (3)

1.8270.18 (3) 1.1370.21 (3) 1.9270.29 (3) 1.9470.10 (3)

Hepatocytes were incubated at 37 1C with cycloheximide (CHX; 100 μM) for three consecutive 100-min periods, with washing (at 0 1C) in between. Leupeptin (LPT, 0.3 mM), NH4Cl (20 mM) or 3-methyladenine (3MA, 10 mM) were added as indicated. Autophagic capacity was measured as the net sequestration of LDH in the presence of leupeptin, but in the absence of NH4Cl or 3MA, during the 200–300 min period (200-min LDH background subtracted), and expressed as %/h. LC3 levels in whole cell extracts (given as % of the 0-min values) and LDH levels (backgroundþpre-sequestered) in the sedimentable fraction (given as % of the cellular total) were measured at the beginning of the sequestration period, i.e., at 200 min.

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

10

] (]]]]) ]]]–]]]

E XP E RI ME N TAL CE L L R ES E ARC H

CTR 0’

100

% OF INITIAL TOTAL LC3

90

(8)

LATE ADDITION OF AUTOPHAGICLYSOSOMAL INHIBITORS (after 100 min with cycloheximide)

80 LC3-II

70

CTR 100’

60 (9)

50 40 30

LC3-I

(8)

(5)

(5)

(3)

(3)

20 10 100-200 min

CTR 3MA

TG NH4Cl NH4Cl + 3MA

to preserve LC3-I better (and to bring the level of presequestered protein even further down), but this regimen did not improve the autophagic sequestration capacity (Table 1, last entry line). Although part of the LC3-II in the last experiment might have been nonrecycling and thus non-functional, the LC3-I pool was, after all, almost twice as high as in control cells, without affecting the autophagic sequestration capacity. Thus, once again, no correlation between LC3 levels and autophagic sequestration capacity could be observed.

Absence of autophagic-lysosomal LC3 turnover after extended cycloheximide treatment To examine the effects of autophagic-lysosomal inhibitors on LC3 degradation at a time when both the levels and turnover of LC3 were reduced, hepatocytes were given such inhibitors after 100 min of cycloheximide treatment, and the incubation continued in the presence of cycloheximide and the other inhibitors for another 100 min. Surprisingly, the overall degradation of LC3 from 100 to 200 min was completely unaffected by the inhibitors during the second 100-min incubation (Fig. 5A), indicating that it was not performed by the autophagic-lysosomal pathway. In contrast, the sequestration of autophagic cargo (LDH) was fully inhibitor-sensitive at any time point (Fig. 4B). There was little or no accumulation of LDH in the absence of leupeptin (Fig. 5B), consistent with a continuous autophagic-lysosomal degradation of the sequestered cargo. Since the autophagic capacity was so well maintained beyond 100 min (Fig. 4A), the macroautophagic cargo flux could obviously proceed independently of an accompanying autophagic-lysosomal LC3 flux. The LC3 decline taking place between 100 and 200 min of cycloheximide treatment (from 59.9% to 40.5%; Fig. 5A) would thus have to be due to a non-lysosomal degradation mechanism.

SEDIMENTABLE LDH (% of total cellular LDH)

autophagic capacity (capacities measured, as %/h, in the absence of inhibitors at 4–6 h: control, 1.9470.19; 3MA, 2.3270.05; TG 2.3870.41, mean7S.E. of three experiments), indicating that the capacity loss is not due to autophagic-lysosomal degradation of some rate-limiting factor. The autophagic sequestration activity was fully sensitive to 3MA at 4–6 h both in cycloheximide-treated and control cells (Fig. 4B). To further probe the role of LC3 in macroautophagic sequestration, hepatocytes were pretreated with reversible inhibitors of LC3 degradation to elevate LC3 levels prior to measurements of autophagic capacity (Table 1). Without inhibitors, the starting level of total LC3 at 200 min was 37.8% of the 0-min value, and the autophagic sequestration capacity (200–300 min) was 1.82%/h. After 200 min with ammonia (NH4Cl), the total LC3 level was more than twice as high as without inhibitors, both LC3-I and LC3-II being better preserved, but the autophagic capacity was, nevertheless, lower (1.13%/h) than in the control cells. Although the retained LC3-II could be lysosome-associated and thus inaccessible for continued autophagic sequestration, the autophagic capacity would at least be expected to remain unaltered rather than to decline if any of the LC3 forms were rate-limiting. Possibly, the accumulation of sequestered cytosol-derived protein inside autophagic vacuoles (to a level more than three times higher than in control cells at 200 min) could have congested the pathway to an incapacitating extent. To reduce the level of presequestered protein, ammonia was washed out after 100 min, and the cells incubated for another 100 min in its absence. This treatment reduced the 200-min level of sequestered protein (LDH) and brought the autophagic sequestration capacity up to the control level – but not higher – even though the LC3 values were still well above the controls (Table 1, third entry line). Finally, 3MA was included during the second incubation (following ammonia washout)

7 6 5

+ 2 h LEUPEPTIN

4 3 2 1 0

CONTROL

0 1 2 3 4 5 CYCLOHEXIMIDE TREATMENT (h at 37 ºC)

Fig. 5 – Late addition of inhibitors reveals a halt in autophagic-lysosomal LC3 flux. (A) Hepatocytes were incubated at 37 1C with pyruvate (15 mM) and cycloheximide (100 μM) for two consecutive 100-min periods. 3-Methyladenine (3MA; 10 mM), thapsigargin (TG; 5 μM), NH4Cl (20 mM) or none of them (control; CTR) were included in the second period as indicated (the cells being washed between the incubations). LC3 was measured in total cell extracts at 0, 100 or 200 min (by immunoblotting and blot scanning); all values (mean7S.E. or range of the number of independent experiments given in parentheses at the top of each 0-min bar) are expressed as % of the 0-min total (LC3-IþLC3-II). None of the inhibitors produced effects significantly different from the 200-min control total. (B) Hepatocytes were incubated continuously at 37 1C with pyruvate (15 mM) and cycloheximide (100 μM) for up to 5 h. At the time point indicated, the amount of LDH in sedimentable cell corpses was measured and expressed as % of the total cellular LDH in cells receiving no additional treatment (○) or incubated with leupeptin (0.3 mM) during the preceding two hours (●). Each value is the mean7S.E. of 5–7 experiments. Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

E X PE R IM EN TA L C ELL R E S EA RC H

Both LC3-I and LC3-II levels were reduced, but since these forms can be readily interconverted by lipidation/delipidation, the data give no clue as to which form is the primary target for nonlysosomal degradation. Proteasomes are capable of degrading LC3 in cell extracts, but apparently not in intact cells [54]. In hepatocytes, lactacystin (at a concentration that inhibited the degradation of ubiquitylated cytosolic protein aggregates) was unable to prevent the overall LC3 decline, indicating that proteasomes were not involved (results not shown). Apart from that, the non-lysosomal degradation of LC3 has not been further investigated.

Cycloheximide-induced reduction in size and numbers of endogenous immunofluorescence-labelled LC3 dots Autophagy is frequently monitored as the number of immunofluorescent cytoplasmic LC3 or GFP-LC3 “dots”, although it can be uncertain whether changes in dot numbers reflect alterations in LC3 expression, autophagic activity or autophagic flux (or whether the dots are autophagic organelles at all) [20]. In hepatocytes, a 4-h cycloheximide treatment resulted in significant (Po0.05) decreases in both the number and average size of hepatocytic LC3 dots (Fig. 6A). Of particular note, the largest dots, which are most likely to represent autophagic vacuoles, were practically undetectable after 4 h with cycloheximide. Studies in yeast have shown that cycloheximide treatment as well as a knockout of Atg8 results in smaller autophagosomes [55], which could well be the case in hepatocytes as well. However, since the autophagic sequestration activity was still appreciable at this time (Figs. 4A and 5B), the possibility should be considered that the autophagic vacuoles had not actually disappeared or become smaller, but rather had become “invisible”, i.e., no longer associated with detectable amounts of immunostainable LC3.

Density gradient distribution of autophagic vacuoles and LC3 after cycloheximide treatment To further examine the possibility of autophagic vacuoles gradually losing their association with LC3 under cycloheximide, hepatocytes were allowed to sequester LDH, accumulating in autophagic vacuoles in the presence of leupeptin [28] or bafilomycin A1 [56] from 0–2 h or from 2–4 h after cycloheximide administration. A density gradient fractionation of homogenized cell corpses revealed a heterogeneous distribution of lysosomes after 2 h with leupeptin (Fig. 6B), the lysosomal marker Lamp-1 [57] being present both in a major fraction peaking at 1.11–1.12 g/ml and in minor fractions peaking at 1.08–1.09 and at 1.04– 1.06 g/ml. Autophagically sequestered LDH was also heterogeneously distributed, with a generally more buoyant distribution than either Lamp-1 or LC3-II (Fig. 6B). Notably, a significant fraction of the LDH was present in the region 1.06–1.08 g/ml, where both Lamp-1 and LC3-II levels were minimal. At 4 h after cycloheximide (Fig. 6C), this non-coincidence was even more pronounced, about one-half of the sequestered LDH being found at densities o1.07 g/ml, where little or no Lamp-1 and no LC3-II were present. This LDH would thus seem to reside in LC3-free, non-lysosomal autophagic vacuoles, presumably amphisomes, which in hepatocytes are devoid of Lamp-1 and known to accumulate LDH in the presence of leupeptin [29,57].

] (]]]]) ]]]–]]]

11

When bafilomycin A1 rather than leupeptin was used to suppress LDH degradation (Fig. 6D), sequestered LDH was found to coincide both (and mainly) with the major lysosomal (Lamp-1) peak at
1.08 g/ml and with the minor peak at
1.05 g/ml. This would suggest a predominantly auto/amphilysosomal LDH distribution, consistent with the good maintenance of autophagiclysosomal flux in bafilomycin-treated hepatocytes [56]. Both LDH peaks were strongly diminished if the cells had been incubated in the presence of 3MA, confirming that they were the result of macroautophagic cargo sequestration (Fig. 6D). At 4 h after cycloheximide (Fig. 6E), the minor, buoyant lysosomal peak was strongly reduced in bafilomycin-treated hepatocytes, and although most of the sequestered LDH (and all of the LC3-II) still coincided with the major lysosomal peak, a substantial amount of LDH banded in the more buoyant region of the gradient (o1.06 g/ml), where essentially no LC3 was present. Most of this LDH was in fractions with very little Lamp-1, possibly representing some amphisomal LDH accumulated due to a bafilomycininduced partial delay in the cargo flux from amphisomes to lysosomes [56]. The gradient studies thus indicate that autophagically sequestered LDH can accumulate in LC3-free, non-lysosomal vacuoles, most probably amphisomes. This is observed most clearly with leupeptin (known to suppress amphisome–lysosome fusion) [29,57] and during the period 2–4 h after cycloheximide, when macroautophagy proceeds in an LC3-flux-independent manner. Some evidence for such accumulation can be seen even during the 0–2 h period, probably indicating that LC3-associated and LC3-free phagophores can be simultaneously engaged in macroautophagic cargo sequestration.

Atg5 is required for lipidation of Atg8 orthologues and is also essential for macroautophagic cargo sequestration in mouse embryonic fibroblasts The results shown so far have demonstrated that in rat hepatocytes treated with cycloheximide, the levels and turnover rates of LC3-I and LC3-II are not representative of macroautophagic cargo sequestration and degradation. After exposure to cycloheximide for a couple of hours, macroautophagic cargo flux in fact appeared to be completely independent of LC3. We next asked if this occurs only in cycloheximide-treated hepatocytes, or if macroautophagy may be independent of LC3 also in other cell types, and in the absence of cycloheximide. Atg5 is required for lipidation of LC3 [30,58,59] yet a previous study reported that autophagosome-like structures could form in Atg5  / mouse embryonic fibroblasts (MEFs) [59]. Since no LC3-II was present [59], those results might suggest that LC3 was not required for autophagosome formation. However, to our knowledge, it remains to be formally tested whether Atg5 is required for the lipidation of GABARAPs. Moreover, and importantly, whether the Atg5  / MEFs actually were capable of sequestering cytoplasmic cargo was not examined in the aforementioned study [59]. We therefore decided to compare the macroautophagic sequestration activities of wild-type vs. Atg5  / MEFs in response to starvation for serum and amino acids, using bafilomycin A1 to allow efficient accumulation of sequestered LDH [38,39]. Strikingly, we found that, in contrast to wild-type MEFs, Atg5  / MEFs were virtually unable to sequester LDH (Fig. 7A). Thus, under these conditions, macroautophagic sequestration activity is strictly dependent on Atg5. As

Please cite this article as: P. Szalai, et al., Autophagic bulk sequestration of cytosolic cargo is independent of LC3, but requires GABARAPs, Exp Cell Res (2015), http://dx.doi.org/10.1016/j.yexcr.2015.02.003

12

E XP E RI ME N TAL CE L L R ES E ARC H

NO. OF LC3 DOTS PER CELL PROFILE

LC3

CTR 4 h

50

(