Calmodulin-related CML24 interacts with ATG4b and affects autophagy progression in Arabidopsis


For correspondence (e-mail


Plants encounter environmental stress challenges that are distinct from those of other eukaryotes because of their relative immobility. Therefore, plants may have evolved distinct regulatory mechanisms for conserved cellular functions. Plants, like other eukaryotes, share aspects of both calcium- and calmodulin-based cellular signaling and the autophagic process of cellular renewal. Here, we report a novel function for an Arabidopsis calmodulin-related protein, CML24, and insight into ATG4-regulated autophagy. CML24 interacts with ATG4b in yeast two-hybrid, in vitro pull-down and transient tobacco cell transformation assays. Mutants with missense mutations in CML24 have aberrant ATG4 activity patterns in in vitro extract assays, altered ATG8 accumulation levels, an altered pattern of GFP–ATG8-decorated cellular structures, and altered recovery from darkness-induced starvation. Together, these results support the conclusion that CML24 affects autophagy progression through interactions with ATG4.


Calmodulin is the prototypical calcium (Ca2+) sensor, is highly conserved between plants and humans, and is essential for diverse aspects of eukaryotic cellular function (Chin and Means, 2000). Ca2+ signaling has been implicated in numerous plant processes, and the presence of seven calmodulin and 50 calmodulin-like (CML) genes in the Arabidopsis genome predicts important and diverse roles for Ca2+ signaling in plants (McCormack and Braam, 2003; McCormack et al., 2005). However, the functions of most of the encoded CML proteins remain unknown.

CML24 was originally identified as a gene (called TCH2) whose expression was strongly up-regulated by simple touch stimulation of plants (Braam and Davis, 1990), and encodes a protein that shares 44% sequence identity with calmodulin (Delk et al., 2005). CML24 expression is also up-regulated by darkness, wounding, heat, cold, and a variety of other stresses (Braam, 1992a,b; Polisensky and Braam, 1996; Delk et al., 2005; Lee et al., 2005). To begin to understand CML24 function, plants with reduced CML24 expression and/or point mutations in the CML24 coding region have been isolated and characterized for overt phenotypes. Reduction of CML24 expression results in altered ion sensitivity and abscisic acid sensitivity, and altered timing of the transition to flowering (Delk et al., 2005). Two point mutants, cml24-2 and cml24-4, with Gly67Glu and Glu124Lys missense mutations, respectively, show altered timing of the transition to flowering, aberrant nitric oxide accumulation, and anomalous root growth responses to mechano-stimulation (Tsai et al., 2007; Wang et al., 2011). The opposite phenotypes of the two mutants (for example, cml24-2 is early flowering with low basal nitric oxide and cml24-4 is late flowering with high basal nitric oxide) indicate that these mutations may have opposite effects on CML24 function (Tsai et al., 2007). To shed light on the cellular and biochemical basis for these phenotypes, we initiated this study by identification of a potential CML24-interacting protein. Our evidence indicates that CML24 binds ATG4b, a cysteine protease involved in autophagy progression.

Autophagy is a conserved cellular degradation process that removes cytoplasmic contents and recycles components to promote cellular remodeling and continue growth under nutrient-limiting conditions (reviewed by Klionsky, 2007). In macroautophagy (here called autophagy), bulk components and organelles are enveloped by double membranes, forming autophagosomes that are then transported to and fused with the vacuole, releasing the contents for degradation in the vacuole.

The protein machinery responsible for carrying out the autophagy progress is highly conserved among yeast, animals and plants (Thompson and Vierstra, 2005; Bassham et al., 2006; Klionsky, 2007). Much knowledge about the autophagy process comes from studies with yeast. The ubiquitin-like ATG8 protein plays a critical role in autophagosome formation, membrane expansion and autophagosome closure (Nakatogawa et al., 2007; Sou et al., 2008; Xie et al., 2008; Noda et al., 2009), and the ATG4 cysteine protease is responsible for two modifications of the ATG8 protein (Krisako et al., 2000; Ohsumi, 2001). ATG4 first primes ATG8 by removing the ATG8 C-terminus and exposing a glycine residue. Followed by progressive processing by the E1-like enzyme ATG7, the E2-like protein ATG3, and the E3-like complex ATG12–ATG5, ATG8 is conjugated to the amino group of phosphatidylethanolamine (Phillips et al., 2008; Nakatogawa et al., 2009; Chung et al., 2010). This lipid linkage promotes ATG8 association with autophagosome membranes. The second activity of ATG4 is to release ATG8 from phosphatidylethanolamine, removing the protein from membranes and regenerating free ATG8 (Krisako et al., 2000).

The two activities of ATG4 appear to be antagonistic with respect to regulation of autophagy progression. Whereas the first action of ATG4 in ATG8 maturation promotes autophagy progression, the second action (removal of ATG8 from membranes) may inhibit autophagy progression. The complexity of ATG4 function leads to the prediction that ATG4 activity is probably under regulation, at least in part because excess ATG4 activity may prevent ATG8 association with autophagosome membranes. Consistent with this expectation, human ATG4 activity is inhibited by oxidation (Scherz-Shouval et al., 2007). H2O2 production, stimulated by starvation, induces autophagy. Cys78 of human ATG4b is oxidized in the presence of H2O2, and this oxidation inhibits ATG4b activity (Scherz-Shouval et al., 2007). The substitution of Cys78 by Ser renders ATG4 activity constitutive (Scherz-Shouval et al., 2007). Arabidopsis has two ATG4 genes, ATG4a and ATG4b (Yoshimoto et al., 2004). By sequence comparison, we found that, although plant ATG4s contain the homologous catalytic Cys residue (Cys74 in human ATG4b, Tanida et al., 2004; Kumanomidou et al., 2006), plant ATG4s lack a cysteine analogous to Cys78, which is required for oxidation-dependent regulation of the human protein (Scherz-Shouval et al., 2007). As plant ATG4s appear to lack the ability to be regulated directly by oxidation, we predict that alternative regulatory mechanisms must exist in plants.

Here we show that CML24 binds ATG4b, and that cml24 mutants are affected in ATG4 activity levels, accumulation of ATG8 proteins, patterns of GFP–ATG8 reporter-decorated cellular structures, and starvation recovery. These data indicate that CML24 plays a role in regulation of autophagy progression, most likely through ATG4 binding.


CML24 interacts with ATG4b

To elucidate the biochemical basis for the cml24 mutant phenotypes (Delk et al., 2005; Tsai et al., 2007), we identified CML24-interacting proteins. Our first approach was to perform a two-hybrid screen for genes encoding proteins that bind CML24 in yeast. Using the University of Wisconsin Molecular Interaction Facility (James et al., 1996), 18 million Arabidopsis seedling cDNAs were tested for production of proteins that bind CML24 and thereby promote yeast growth on histidine drop-out plates and generate β-galactosidase activity. Plasmids from the positive clones were rescued, subjected to restriction enzyme digestion, and re-transformed into yeast. Interaction was verified by selective growth of clones in the presence of CML24 bait constructs, not non-bait constructs. cDNA sequencing identified the autophagy cysteine protease ATG4b (At3g59950) as a putative CML24 interactor.

We next obtained further evidence as to whether ATG4b and CML24 interact and whether the interaction is functionally relevant. Although an ATG4b-related protein, ATG4a, exists in Arabidopsis, we focused our analysis on ATG4b because ATG4b was identified as a CML24 interactor in the yeast two-hybrid analysis.

Based on Genevestigator data (Zimmermann et al., 2005, 2004), both CML24 and ATG4b are expressed relatively ubiquitously among diverse tissues and organs and throughout developmental stages, and therefore the encoded proteins may have the opportunity to interact in vivo. Based on the lack of clear subcellular targeting sequences in the ATG4b and CML24 proteins, it is possible that both proteins are localized to the cytosol and have the potential to interact. Indeed, ATG4 is thought to function in the cytosol (Krisako et al., 2000; Ohsumi, 2001).

To obtain additional evidence for CML24 and ATG4b interaction and determine whether binding is detectable in vitro, we generated and purified a glutathione S-transferase (GST)–ATG4b fusion protein and tested its ability to bind purified CML24. GST–ATG4b and CML24 were incubated together in the presence or absence of Ca2+. GST and associated proteins were pulled down using glutathione (GSH)-conjugated Sepharose beads and analyzed by immunoblotting with anti-CML24 antibody (Figure 1a). CML24 was pulled down using the GSH-beads in the presence of GST–ATG4b (Figure 1a) but not GST alone (Figure 1c). Therefore, CML24 binds GST–ATG4b in vitro, and this interaction is dependent upon the presence of ATG4b. Ca2+ had no detectable effect on CML24 binding to GST–ATG4b (Figure 1a); therefore, CML24 binding to ATG4b appears to be Ca2+-independent, at least under these assay conditions. The mutant cml24 isoforms cml24-2 and cml24-4, with the single amino acid substitutions Gly67Glu and Glu124Lys (Delk et al., 2005), respectively, also bound GST–ATG4b in a Ca2+-independent manner under these conditions (Figure 1a). Because of the sequence and structural similarities between CML24 and CaM (Khan et al., 1997; McCormack and Braam, 2003; Delk et al., 2005), we tested whether CaM also bound ATG4b. Purified CaM was incubated with GST–ATG4b, and the GST-associated proteins were analyzed using an anti-CaM antibody. Figure 1(b) shows that, under conditions in which GST–ATG4b interacts with CML24, no binding to CaM was detected. Thus, despite the primary and predicted tertiary structural similarities between CML24 and CaM (Khan et al., 1997; McCormack and Braam, 2003; Delk et al., 2005), the in vitro interaction with GST–ATG4b is specific for CML24.

Figure 1.

CML24 interaction with ATG4b. (a) CML24 binds GST–ATG4b in vitro in a calcium-independent manner. Recombinant GST–ATG4b was used in pull-down assays with CML24, cml24-2 and cml24-4 proteins; proteins were immunoblotted using anti-CML24 antibody. (b) GST–ATG4b was used in pull-down assays with CaM; proteins were immunoblotted using anti-CML24 antibody.(c) GST protein was used in pull-down assays with CML24; proteins were immunoblotted using anti-CML24 antibody. For (a–c), protein interactions were performed in the presence of 1 mm Ca2+ (+) or 1 mm EDTA (−). Proteins pulled down with GSH–Sepharose (‘P’) and proteins that remained soluble (‘S’) were separated by 12% SDS–PAGE before immunoblotting. Similar results were obtained in two independent experiments. (d) Proteins from N. benthamiana leaf tissue, after transient transformation with genes encoding Myc-tagged ATG4b and HA-tagged CML24 or HA-tagged GFP, were subjected to immunoblot analysis directly (‘Input’) or after immunoprecipitation with anti-HA tag antibody (‘Co-IP’). Top panel: Co-IP and input proteins were immunoblotted using anti-Myc epitope antibodies to detect ATG4b–Myc (65.4 kDa). ATG4b–Myc was detected after anti-HA immunoprecipitation (‘Co-IP’) when the gene for CML24–HA was co-transformed but not when the gene for GFP–HA was co-transformed. Bottom panel: control analysis indicating that immunoblotting with anti-HA antibody detects both GFP–HA (33.4 kDa) and CML24–HA (23.9 kDa) in both total protein (‘Input’) and after immunoprecipitation with anti-HA antibodies (‘Co-IP’), showing that both the GFP–HA and CML24–HA proteins accumulate in the transformed cells and were successfully immunoprecipitated with anti-HA antibodies. Twenty times more protein was used for ‘Co-IP’ analysis than loaded directly in the ‘Input’ lanes to enhance detection.

To verify whether the ATG4b–CML24 interaction occurs in planta, we transiently co-transformed tobacco leaf cells with genes encoding Myc epitope-tagged ATG4b and either HA epitope-tagged CML24 or HA-tagged GFP, and performed co-immunoprecipitation analysis of the resulting protein extracts. As shown in Figure 1(d), immunoprecipitation using anti-HA antibodies resulted in co-precipitation of ATG4b–Myc (top panel, ‘Co-IP’) only when cells were co-transformed with genes encoding CML24–HA and ATG4b–Myc. In contrast, ATG4b–Myc was not detectable after anti-HA immunoprecipitation of proteins from cells co-transformed with genes encoding GFP–HA and ATG4b–Myc. Accumulation levels of Myc-tagged ATG4b were similar in both sets of leaves, i.e. leaves co-transformed with either the CML24–HA or GFP–HA expression constructs (Figure 1d, top panel, ‘Input’). Furthermore, as shown in Figure 1(d) (bottom panel), both GFP–HA and CML24–HA accumulated to similar levels in the transformed cells and were successfully immunoprecipitated by the anti-HA antibodies, as demonstrated by anti-HA immunoblot analysis of both total leaf proteins (‘Input’) and anti-HA immunoprecipitated proteins (‘Co-IP’). Therefore, immunoprecipitation of ATG4b–Myc by anti-HA antibodies required the presence of CML24. Together, these results provide evidence for a specific interaction between CML24 and ATG4b in plant cells.

ATG4 activity is altered by darkness and in the cml24 mutants

We monitored ATG4 activity in vitro to test whether the binding interactions between CML24 and ATG4b affect ATG4b cysteine protease activity. ATG4 removes C-terminal amino acid(s) from ATG8, leaving a terminal Gly residue at the C-terminus (Yoshimoto et al., 2004). We generated a substrate to monitor in vitro ATG4 activity; ATG8e was tagged with six copies of histidine (His) and a hemagglutinin (HA) epitope at the N- and C-termini, respectively. Recombinant His–ATG8e–HA protein was produced in Escherichia coli and affinity-purified using Ni-NTA agarose.

To verify that ATG4 activity may be detected using this substrate, His–ATG8e–HA was incubated with or without GST–ATG4b, and His–ATG8e–HA integrity was then analyzed by SDS–PAGE, immunoblotting and anti-HA and anti-His antibody interaction (Figure 2a). In the absence of GST–ATG4b, a single band is detected by both anti-HA and anti-His antibodies (Figure 2a, single asterisk); we interpret this band as representing the full-length, dual-tagged, recombinant His–ATG8e–HA. A lower band, generated only after incubation in the presence of GST–ATG4b, is recognized by anti-His antibodies but not anti-HA antibodies (Figure 2a, double asterisks), as expected for the ATG4 cleavage product His–ATG8e that has lost the C-terminal HA extension. Similar to previous reports with human ATG4 (Tanida et al., 2004), recombinant plant GST–ATG4b protease is able to remove the C-terminal HA extension from a recombinant His–ATG8e–HA in vitro.

Figure 2.

CML24 affects ATG4 activity in plant extracts.(a) Detection of ATG4 activity in vitro. His–ATG8e–HA was incubated with (+) or without (−) GST–ATG4b at 20°C for 2 h, and proteins were subjected to SDS–PAGE and blotting, then detected using anti-HA and anti-His antibodies. The full-length His–ATG8e–HA (single asterisk) and putative His–ATG8e product (double asterisks) are indicated.(b) Relative ATG4 activity in wild-type and mutant extracts. Anti-His antibodies detect His–ATG8e–HA and His–ATG8e after His–ATG8e–HA incubation with extracts derived from wild-type (Col-0), cml24-2, cml24-4 and atg4a4b plants grown under constant light. The lower panel shows the relative band intensity levels of His–ATG8e accumulation between Col-0 and mutants. The hash symbols (#) above bars indicate significant differences compared with Col-0 (< 0.05, Student's t-test, = 3).(c) Darkness-induced changes in ATG4 activity. Anti-His antibodies detect His–ATG8e–HA and His–ATG8e after His–ATG8e–HA incubation with extracts derived from wild-type (Col-0), cml24-2, cml24-4 and atg4a4b plants grown under constant light for 14 days (0) and after 1–12 h of darkness (1–12). The relative levels of His–ATG8e were estimated by band intensity comparisons, normalizing to the ‘0’ band intensity for each genotype. In (b) and (c), total proteins were stained with Coomassie brilliant blue (CBB). The ratio intensity of the His–ATG8e to His–ATG8e–HA was quantified using ImageJ, and total proteins (CBB) were used for normalization. Similar results were obtained in five independent experiments.

To examine whether CML24 affects ATG4 activity, we monitored ATG4-dependent His–ATG8e–HA cleavage in vitro using purified proteins and plant extracts. With these extracts, we cannot distinguish activities of ATG4b from those of ATG4a; these assays probably detect the combined activities of these two enzymes. Assays performed in the presence of recombinant purified CML24 showed only modest or no effect on ATG4 activity (Figure S1). Because post-translational modifications and/or the use of recombinant tagged proteins may influence in vitro activities of E. coli-produced proteins, we also assayed ATG4 activity in extracts derived from wild-type and mutant plants to assess whether CML24 influences ATG4 activity. We generated extracts from wild-type, cml24-2, cml24-4 and atg4a4b, a double mutant with T-DNA insertions in both ATG4a and ATG4b (Chung et al., 2010), to assay ATG4 activity. Extracts were incubated with the His–ATG8e–HA substrate for 2 h, and total proteins were analyzed by SDS–PAGE and immunoblotted with anti-His antibodies. Coomassie brilliant blue (CBB) gel staining was used to assess overall protein levels as a gel loading control. Figure 2(b) compares ATG4 activity detectable in extracts derived from constant light-grown Col-0, cml24-2, cml24-4 and atg4a4b. Although only a single band, corresponding to the full-length His–ATG8e–HA protein, is detected after incubation with the atg4a4b extract (Figure 2b), a second faster-migrating immunoreactive band is present after incubation of the His–ATG8e–HA substrate with the Col-0 protein extract (Figure 2b). The ATG4-dependent appearance of the lower molecular mass band confirms that ATG4 activity is detectable in plant extracts. Therefore, we next compared ATG4 activity in the mutant cml24-2 and cml24-4 extracts. As shown in Figure 2(b), ATG4 activity was detected in both mutant extracts; however, the level of activity was approximately 40% higher in the cml24-2 extract and 50% lower in the cml24-4 extract relative to Col-0 (Figure 2b). This result suggests that mutations in CML24 affect ATG4 activity levels.

To further examine whether mutations in CML24 affect ATG4 activity, we assayed ATG4 activity in plants under conditions in which autophagy levels are known to be altered and CML24 expression is up-regulated. Darkness promotes carbon limitation and enhances autophagy in plants (Brouquisse et al., 1998; Vauclare et al., 2010; Xiao et al., 2010), and darkness induces CML24 expression in Arabidopsis (Braam and Davis, 1990; Delk et al., 2005). Wild-type, cml24-2, cml24-4 and atg4a4b were grown for 2 weeks under constant light and then subjected to incubation in darkness for 1–12 h. ATG4 activity was then examined in the derived tissue extracts by addition of His–ATG8e–HA and monitoring production of the cleaved His–ATG8e product by immunoblot analysis. Over the first 3 h of darkness treatment, Col-0 ATG4 activity increases modestly (Figure 2c). However, after 4 h of darkness, the ATG4 activity in Col-0 extracts decreases below the starting activity level and reaching its lowest levels after 5 and 6 h of darkness and remains low after 12 h of darkness (Figure 2c). These data indicate that the growth conditions, possibly the reduction in carbon fixation resulting from prolonged darkness, lead to changes in assayable ATG4 activity. Extracts from atg4a4b double mutants served as a control; no lower molecular mass band is detected (Figure 2c), indicating that appearance of the faster-migrating lower molecular mass product, interpreted to represent His–ATG8e after C-terminal HA tag removal, depends on ATG4 activity.

We next determined whether ATG4 activity is altered in darkness-treated cml24 mutants relative to wild-type. CML24 is up-regulated in expression within 30 min of darkness treatment (Braam and Davis, 1990; Delk et al., 2005). Similar to the results shown in Figure 2(b), extracts derived from cml24-2 plants grown under constant light have higher ATG4 activity than wild-type, with approximately half of the His–ATG8e–HA substrate being cleaved (Figure 2c), and extracts derived from light-grown cml24-4 display ATG4 activity that is modestly reduced relative to Col-0 (Figure 2c). However, the changes in ATG4 activity levels in extracts from darkness-treated cml24-2 and cml24-4 plants have distinct temporal characteristics when compared to wild-type Col-0 (Figure 2c). Whereas the first 3 h of darkness results in an increase in ATG4 activity in wild-type extracts (Figure 2c), similar treatment leads to decreases in ATG4 activity in cml24 mutant extracts (Figure 2c). After 3 h of darkness, Col-0 ATG4 activity peaks, whereas that of the mutants is very low or undetectable. Furthermore, 12 h of darkness leads to a recovery of detectable ATG4 activity in cml24-2 extracts, although activity levels are not as high as light-grown cml24-2. Darkness results in virtually no detectable ATG4 activity in cml24-4 extracts (Figure 2c). Because of the dual and complex role of ATG4 in influencing autophagy progression, how these extract ATG4 activity levels relate to the regulation of ATG8 modification in planta is not clear at present. However, the magnitude of detectable ATG4 activity in cml24 mutant extracts and temporal differences compared with wild-type are consistent with the hypothesis that CML24 is required for appropriate ATG4 activity regulation.

Loss of ATG4 activity may lead to accumulation of ATG8 proteins (Yoshimoto et al., 2004). If CML24 affects ATG4 activity regulation, we hypothesized that the cml24 mutants may have altered accumulation of ATG8 proteins as a consequence. Therefore, we monitored ATG8 levels in wild-type (Col-0), cml24-2, cml24-4 and atg4a4b. Anti-ATG8a antibodies (Yoshimoto et al., 2004) were used to probe total protein derived from plants grown under 16 h photoperiods and harvested after either the first hour of light or the first hour of darkness (Figure 3). As expected because the anti-ATG8a antibodies cross-react with most ATG8 isoforms (Yoshimoto et al., 2004), multiple protein bands were detected in all plant extracts. Similar protein banding patterns were seen in Col-0 and the cml24 mutants; however, overall ATG8 protein accumulation levels were reduced in cml24 mutants relative to Col-0 harvested after 1 h of light (Figure 3). The ATG8 protein band profile is distinct in atg4a4b, with the highest molecular mass band migrating more slowly than that in wild-type or the cml24 mutants (Figure 3). This higher molecular mass band, detected only in atg4a4b, may represent accumulation of ATG8 proteins with C-terminal amino acid extensions that are not cleaved due to loss of ATG4 activity. The absence of this band in the cml24 mutants indicates that the mutant CML24 proteins do not reduce ATG4 activity to the levels found in atg4a4b. After 1 h of darkness, the ATG8 protein profile changes in Col-0 and the cml24 mutants, and the overall ATG8 protein quantity in wild-type decreased and was more comparable to the levels detected in the cml24 mutants (Figure 3). In contrast, the atg4a4b mutant accumulated ATG8 proteins at much higher levels than Col-0 after darkness treatment (Figure 3), indicating that 1 h of darkness is sufficient to trigger ATG4-dependent changes in ATG8 modifications.

Figure 3.

ATG8 protein accumulation levels in wild-type (Col-0), cml24-2, cml24-4 and atg4a4b.Total protein was extracted from 10-day-old seedlings grown under 16 h photoperiods and harvested either after the first hour of light or after the first hour of darkness. Proteins were subjected to SDS–PAGE with urea, and analyzed by immunoblotting with anti-ATG8a antibodies. Total proteins were stained with Coomassie brilliant blue (CBB) (bottom panel) to confirm that similar protein amounts were loaded. Similar results were obtained in five independent experiments.

The altered levels of ATG8 protein in the light-grown cml24 mutants and darkness-treated atg4a4b mutants may be a consequence of either altered ATG8 synthesis or stability. To obtain insight into the possible mechanism of this regulation, we assessed ATG8 transcript levels. ATG8a, ATG8e and ATG8i are the three most highly expressed ATG8 genes (Yoshimoto et al., 2004; Michael et al., 2008); therefore, we quantified transcript accumulation from these genes by quantitative RT-PCR. ATG8a and ATG8i transcript levels were not significantly altered in plants harvested under light or darkness conditions (Figure S2). In contrast, ATG8e expression increased fourfold in plants treated with 1 h of darkness. ATG8e expression levels in the cml24 mutants were not significantly different from wild-type under both light and darkness conditions. In contrast, atg4a4b had significantly lower ATG8e expression in the light relative to wild-type but comparable ATG8e expression to wild-type under darkness. These results indicate that the cml24 mutants accumulate reduced levels of ATG8 proteins, but that these differences are not reflected in ATG8a, ATG8e and ATG8i transcript abundance. Therefore, altered ATG8 protein levels probably result from post-transcriptional mechanisms.

Mutations in CML24 affect GFP–ATG8 fluorescence patterns

Our findings demonstrate that CML24 interacts with ATG4b and affects assayable ATG4 activity levels in extracts. We next wished to assess whether the cml24 mutants display altered autophagosome accumulation. GFP–ATG8 is a well-characterized tool to monitor autophagosome accumulation in plant cells (Contento et al., 2005), and increased GFP–ATG8 autophagosome detection occurs in plants subjected to darkness treatments (Bassham et al., 2006; Rose et al., 2006). Therefore, we introduced the CaMV35S:GFP-ATG8e transgene into the mutant cml24 backgrounds by genetic crosses (Bassham et al., 2006). Seven-day-old light-grown seedlings of Col-0, cml24-2 and cml24-4 harboring the CaMV35S:GFP-ATG8e transgene were incubated in constant darkness for 16 h in either the absence or presence of concanamycin A (ConA). ConA blocks degradation of autophagic bodies in the vacuole and thereby enhances the ability to monitor successful autophagy progression (Yoshimoto et al., 2004; Thompson et al., 2005). In the absence of ConA, root cells of all three genotypes, Col-0/GFP-ATG8e, cml24-2/GFP-ATG8e and cml24-4/GFP-ATG8e, show few, if any, intracellular particles detectable by differential interference contrast (DIC) or GFP-labeled punctate structures (Figure 4a). These particles localize primarily to the cell edges and therefore may be primarily cytosolic. However, in the presence of ConA, there is an increase in DIC-visible particles and fluorescent GFP punctate structures in all three genotypes (Figure 4a), and, as expected because of ConA-inhibited vacuolar degradation, these structures are probably localized within the vacuole because they are positioned throughout the center of the cells instead of at the cell edges (Figure 4a). When both DIC particles and GFP punctate structures are quantified, cml24-2/GFP-ATG8e cells were found to accumulate significantly more of these structures than Col-0/GFP-ATG8e (Figure 4b). There are also a greater number of larger fluorescent structures in cml24-2/GFP-ATG8e than in Col-0/GFP-ATG8e (Figure 4a). cml24-4/GFP-ATG8e cells have comparable numbers of fluorescent punctate structures to wild-type; however, the intensity of the staining and uniformity among cells are somewhat diminished compared to wild-type (Figure 4a). The enhanced abundance of putative autophagic bodies in cml24-2/GFP-ATG8e relative to Col-0/GFP-ATG8e may be a consequence of the elevated ATG4 activity in cml24-2 extracts (Figure 2). These results suggest that the cml24-2 mutation may enhance autophagy progression, possibly through altered ATG4 activity regulation.

Figure 4.

GFP–ATG8e fluorescence patterns in seedling roots. One-week-old GFP–ATG8e transgenic seedlings of wild-type (Col-0), cml24-2 and cml24-4 were examined after incubation in the dark with or without concanamycin A (ConA) for 16 h.(a) Representative differential interference contrast (DIC) and fluorescence (GFP) images are shown. The white arrowheads indicate cytosolic GFP signals and the pink arrowheads indicate punctate DIC-visible structures and GFP signals in the vacuole. Scale bars = 10 μm.(b) Comparison of the number of DIC and GFP punctate structures per cell. Asterisks indicate a significant difference compared with wild-type (< 0.01, > 20).

Mutations in CML24 affect darkness-induced starvation survival

The cml24 mutants display alterations in ATG4 activity (Figure 2), ATG8 protein accumulation (Figure 3) and GFP–ATG8e-labeled cellular structures (Figure 4), suggesting alterations in the autophagy process. We next wished to test the hypothesis that the cml24 mutations result in altered stress resistance due to autophagy regulation defects. Autophagy plays a role in re-mobilizing nutrients by recycling organelle and protein components in cells in response to starvation conditions (Bassham et al., 2006; Wada et al., 2009). In plants, a failure to undergo autophagy causes accelerated senescence and hypersensitivity to fixed-carbon deprivation conditions (Thompson et al., 2005; Chung et al., 2010). The consequences of elevated autophagy are unclear. To test whether cml24 mutants have altered sensitivity to carbon-limiting conditions, seedlings were grown in nutrient-rich agar medium without sucrose under long-day photoperiods (16 h light/ 8 h dark) for 4 days, and then the plants were subjected to extended darkness followed by a 7-day recovery under long-day photoperiods. Figure 5(a) shows representative seedlings to illustrate the relative leaf chlorosis resulting from increasing durations of darkness. Whereas all three atg genotypes show prominent chlorosis after 7 days of darkness, Col-0 and the cml mutant seedlings maintain substantial green coloration after 10 days of darkness (Figure 5a), and only seedlings harboring the cml24-2 mutation maintained a green color after 15 days of darkness. We quantified survival as another measure of darkness sensitivity (Figure 5b). Most Col-0 plants survived the 10-day darkness treatment, whereas only 20% survived after 12 days of darkness (Figure 5b). atg5-1 and atg10-1, two well-characterized autophagy mutants (Chung et al., 2010), did not survive after 6 days of darkness, and atg4a4b displayed enhanced darkness sensitivity relative to wild-type but was slightly more resistant than atg5-1 or atg10-1; atg4a4b plants did not survive after 8 days of darkness (Figure 5b). cml24-4 also showed an intermediate resistance, maintaining approximately 25% survival after 10 days of darkness (Figure 5b). Remarkably, cml24-2 displayed enhanced tolerance to prolonged darkness, with 25% of plants surviving after being subjected to 18 days of darkness (Figure 5). These results demonstrate that cml24-2 and cml24-4 show altered darkness sensitivity relative to Col-0, consistent with the conclusion that CML24 may play a role in starvation tolerance, possibly through its interaction with ATG4 and autophagy regulation. The opposing phenotypes of cml24-2 and cml24-4 suggest that the distinct point mutations differentially affect protein function, consistent with previous studies (Tsai et al., 2007; Wang et al., 2011).

Figure 5.

Altered sensitivity of cml24 mutants to prolonged darkness. Four-day-old seedlings were transferred to constant darkness for 1–19 days and then allowed to recover for 7 days. (a) Representative images of seedlings after 0, 5, 7, 10 and 15 days to show overt phenotypes. (b) Percentages of plants recovering from extended darkness treatments after a 7-day recovery. Each value is the mean of four replications, and at least 30 plants were used for each replication. Error bars represent standard deviation.


Autophagy is a process aimed at destroying cellular components, and therefore must be tightly regulated to prevent self destruction. The cysteine protease ATG4 has a complex role in autophagy progression because it has two activities. ATG4 primes ATG8 by removal of the C-terminal amino acid(s), enabling subsequent modifications necessary for membrane targeting. The second role of ATG4, removal of ATG8 from lipids (e.g. phosphatidylethanolamine) and thus membranes, may prevent promotion of autophagosomal membrane formation and expansion by ATG8 if too robust. Therefore, regulation of ATG4 activity is expected. For human ATG4, activity is inhibited by oxidation of Cys78 (Scherz-Shouval et al., 2007). This oxidation-dependent regulation of ATG4 activity is required for proper autophagy progression (Scherz-Shouval et al., 2007). Although plant ATG4s share significant sequence identity (28–34%) with human ATG4s, a cysteine analogous to human ATG4b Cys78 is not present in the plant proteins. Therefore, although we predict that plant ATG4s must also be regulated, the regulatory mechanism must be distinct from that of human ATG4s. The data presented here are consistent with the conclusion that CML24, a calmodulin-related protein encoded by a strongly stress-regulated gene (Braam, 1992a; Polisensky and Braam, 1996; Delk et al., 2005; Lee et al., 2005), may act to regulate ATG4b activity and autophagy progression in Arabidopsis because CML24 interacts with ATG4b (Figure 1) and affects ATG4 activity levels (Figure 2), ATG8 accumulation (Figure 3), the pattern of accumulation of GFP–ATG8e-decorated punctate structures (Figure 4) and starvation recovery (Figure 5).

We show that CML24 and ATG4b interact (Figure 1). The protein domains responsible for interaction remain to be defined. The primary sequence of Arabidopsis ATG4b has two predicted CaM-binding sites; one is close to the active site and is similar to a motif that is also found in mammalian ATG4, whereas the other site is distant from the ATG4b active site. Whether these predicted CaM-binding sites are relevant for CML24 interaction is as yet unknown. Because we did not detect CaM binding to ATG4 (Figure 1), these domains may not be active in either CaM or CML interaction. Furthermore, we found that the CML24–ATG4b interaction in vitro was not influenced by the presence of Ca2+. This result suggests that CML24 binding of the potential ATG4b target is not regulated by Ca2+. However, it is possible that, although Ca2+ may not affect CML24 binding to ATG4, Ca2+ binding to CML24 may regulate CML24 activity, and thus ATG4b function. Differentiating among these possibilities will be the aim of future work.

The mechanism by which CML24 affects ATG4 activity levels remains undefined. The finding that CML24 directly binds ATG4b strongly suggests that binding may affect activity; however, recombinant ATG4b activity is not significantly affected by purified CML24 in vitro, at least under our experimental conditions (Figure S1). These results suggest that binding alone may be insufficient for regulatory effect. Alternatively, it is possible that the bacterially produced recombinant proteins lack native protein characteristics, such as post-translational changes or proper folding necessary to mimic protein behavior in vivo. In addition, interaction with CML24 may affect other ATG4b characteristics that affect activity indirectly, such as regulating interaction with other proteins or cellular location. However, the detection of altered ATG4 activity in the mutant plant extracts relative to that in wild-type (Figure 2) indicates that defects in CML24 affect assayable ATG4 activity, consistent with the conclusion that CML24 plays a role in ATG4 regulation. At present, we cannot distinguish between effects on the priming versus de-conjugation activities of ATG4 nor the roles and regulation of ATG4b versus ATG4a. An interesting outstanding problem is to determine whether the distinct ATG4 activities are differentially regulated in cells.

Intriguingly, as has been shown before for the two available mutant alleles of CML24 (Tsai et al., 2007; Wang et al., 2011), the distinct point mutations confer distinct phenotypes. cml24-2 shows the opposite behavior to wild-type when ATG4 activity is measured in plant extracts: ATG4 activity is high in cml24-2 when wild-type activity is typically low, such as under constant light (Figure 2c) and after extended darkness (Figure 2c), and cml24-2 fails to show the wild-type elevation of ATG4 activity during the first 2–3 h of darkness (Figure 2c). cml24-4 has lower ATG4 activity levels than wild-type throughout the time course (Figure 2c). Without a null mutant of CML24, it remains difficult to conclude whether CML24 may act as a positive or negative regulator of ATG4 activity. We favor the hypothesis that CML24 is probably a positive regulator of ATG4 activity because cml24-4 has other phenotypes that resemble plants with reduced expression of CML24 and may therefore be at least a partial loss-of-function allele (Delk et al., 2005; Tsai et al., 2007). However, it is likely that the regulation of ATG4 by CML24 may be more complex because CML24 itself is proposed to be regulated by Ca2+ binding (Khan et al., 1997; McCormack and Braam, 2003; McCormack et al., 2005), and the mutations may not only influence CML24's ability to regulate target proteins, such as ATG4, but also to respond to Ca2+ signals appropriately.

CML24 has also been shown to influence GFP–ATG8e accumulation patterns in root cells following darkness. Figure 4 shows clear accumulation of GFP–ATG8e-decorated punctate structures in root cells of plants subjected to darkness and concanamycin A (ConA), which traps autophagic bodies in the vacuole. Relative to wild-type, cml24-2 has enhanced accumulation of punctate structures, suggesting increased autophagosome generation in the mutant (Figure 4). cml24-2 extracts also have elevated ATG4 activity (Figure 2); one possibility is that the enhanced ATG4 activity in cml24-2 is responsible for the greater autophagosome formation. Higher ATG4 activity may result in an increased availability of primed ATG8 substrates for lipid modification and membrane association. However, the successful production of autophagic bodies in cml24-2 suggests that the higher ATG4 activity in the mutant does not prematurely remove ATG8 from membranes and prevent autophagosome assembly. cml24-4 generally has reduced assayable ATG4 activity compared with wild-type, and the accumulation of GFP–ATG8e fluorescent structures is not statistically distinct from that of wild-type. The cml24-4 mutation may only modestly affect ATG4 function and autophagosome formation.

Remarkably, cml24-2 survives prolonged darkness-induced starvation better than wild-type, whereas cml24-4 has moderately reduced tolerance to extended darkness-induced starvation (Figure 5). The survival data are consistent with the interpretation that cml24-2 has increased autophagy efficiency, perhaps through elevated ATG4 activity, and cml24-4 is moderately defective in autophagy, with reduced but detectable ATG4 activity. These behaviors suggest that elevated ATG4 activity may enhance autophagy efficiency and thereby the ability to survive starvation.

In summary, a novel means of regulation of ATG4 activity and autophagy progression has been identified. The ability of CML24 expression to respond to diverse abiotic stresses may underlie the ability of Arabidopsis to acclimatize to dynamic environmental conditions through alteration in autophagy regulation.

Experimental Procedures

Plant materials

Arabidopsis thaliana seeds were surface-sterilized with 75% ethanol and 6% sodium hypochlorite for 10 min, followed by three 3 min washes with sterilized water. Sterilized seeds were sown in Metro-Mix 200 series soil (SunGro, or on agar plates (0.8% w/v) with 0.5× MS medium (Murashige and Skoog, 1962), and were grown under a 16 or 24 h photoperiod, approximately 60 μmol m−2 sec−1, at 22°C. The T-DNA insertion mutants atg4a4b (Chung et al., 2010), atg5-1 and atg10-1 (Thompson et al., 2005), all generously provided by R. Vierstra (Department of Genetics, University of Wisconsin, Madison, WI, USA), and the tilling mutants cml24-2 and cml24-4 (Tsai et al., 2007), are in the Col-0 ecotype background. Nicotiana benthamiana plants were maintained under 12 h dark/12 h light (approximately 60 μmol m−2 sec−1) at 28°C.

Plasmid construction

To construct E. coli expression plasmids encoding GST–ATG4b, His–ATG8e–HA and CML24, the full-length open reading frames of ATG4b, ATG8e and CML24 were amplified by PCR from Col-0 wild-type cDNA using the following primers: 5′-GggatccATGAAGGCTATATGTGATAGATTTG-3′ and 5′-CCgaattcTCAAAGTAATTGCCAGTCATC-3′ for ATG4b, 5′-GCCcatatgAATAAAGGAAGCATCTTTAAGATGGACGACG-3′ and 5′-TGCggatccCTAAGCGTAATCTGGAACATCGTATGGGTAGATTGAAGAAGCACCGA-3′ for ATG8e, and 5′-catatgATGTCATCGAAGAACGGAGTTG-3′ and 5′-CggatccTCAAGCACCACCACCATTAC-3′ for CML24. The lower-case letters show restriction enzyme sites to facilitate cloning, and the underlined region indicates the HA tag. cml24-2 and cml24-4 open reading frames were amplified from cml24-2 and cml24-4 cDNA using the same CML24 primers. The amplified ATG4b fragment and pGEX2T (GE Healthcare, were digested with BamHI and EcoRI and then ligated together such that GST was fused to the 5′ end of ATG4b. The 3′ end of ATG8e was ligated to an HA tag sequence, amplified by PCR, and inserted into pET15b (EMD Biosciences, via NdeI and BamHI sites. The 6xHis tag encoded in pET15b was cloned in-frame to the 5′ end of the ATG8e–HA gene. CML24, cml24-2 and cml24-4 PCR products were inserted into the cloning site (NdeI/BamHI) of pET21b (EMD Biosciences). All of the constructs were sequenced and transformed into E. coli BL21 (DE3) (Invitrogen,

For transient co-expression in N. benthamiana leaves, pER8-ATG4b-Myc and pER8-CML24-HA were constructed by inserting full-length cDNAs of ATG4b and CML24 at the AscI site of the pER8-Myc and pER-HA vectors, respectively (Kang et al., 2008). The primers used for amplifying ATG4b were 5′-AGGCGCG CCTATGAAGGCTATATGTGATAGATTTG-3′ and 5′-AGGCGCG CCAAGTAATT GCCAGTCATCTTCATGT-3′. The primers used for amplifying CML24 were 5′-AGGCGCGCCTATGTCATCGAAGAACGGAGTTGT T-3′ and 5′-AGGCGCGCCAGCACCACCACCATTACTCATCAT-3′. Constructs were sequenced and transformed into Agrobacterium tumefaciens GV2260 for expression in N. benthamiana.

Generation and purification of recombinant proteins

Expression of the genes encoding the recombinant proteins was induced by adding isopropyl thio-β-d-galactoside and incubating at 37°C for 2 h. After induction, E. coli was lysed using a cell disrupter in TBS buffer (50 mm Tris, pH 7.5, 150 mm NaCl) containing 1 mm phenylmethanesulfonyl fluoride (PMSF) (P7626, Sigma-Aldrich,, 1 mm EDTA, 1 mm dithiothreitol (DTT) and protease inhibitors (P-9599; Sigma-Aldrich).

The recombinant GST–ATG4b proteins were applied to glutathione–Sepharose 4B beads (GE Healthcare) and washed with 25 mm Tris pH 7.5, 150 mm NaCl, 1 mm EDTA containing protease inhibitors. To elute GST–ATG4b proteins, the beads were incubated with elution buffer (10 mm reduced glutathione, 50 mm Tris pH 8.0, 5% glycerol). The eluted proteins were dialyzed using 25 mm Tris pH 7.6, 150 mm NaCl, 1 mm EDTA, 10% glycerol. The recombinant AtATG8e was affinity-purified using a Ni2+-NTA column (Qiagen, equilibrated with MCAC-20 buffer (20 mm Tris pH 7.9, 0.5 m NaCl, 10% glycerol, 1 mm PMSF), 20 mm imidazole) and eluted using MCAC-200 buffer (20 mm Tris pH 7.9, 0.5 m NaCl, 10% glycerol, 1 mm PMSF, 200 mm imidazole). The recombinant CML24, cml24-2 and cml24-4 proteins were incubated with 0.3 mm CaCl2 before loading onto phenyl Sepharose CL-4B columns (Sigma) equilibrated with 25 mm Tris, pH 8.0, 150 mm NaCl, 1 mm DTT, 1 mm CaCl2. CML24, cml24-2 and cml24-4 were eluted using 25 mm Tris pH 8.0, 150 mm NaCl, 1 mm DTT, 1 mm EDTA, and precipitated using 70% ammonium sulfate. The protein pellets were solubilized and dialyzed using 25 mm Tris pH 8.0, 150 mm NaCl, 1 mm DTT, 10% glycerol.

GST pull-down assay

The purified GST-fused ATG4b was incubated with glutathione–Sepharose 4B beads for 30 min at 4°C, and washed with TBS buffer (50 mm Tris, pH 7.5, 150 mm NaCl). CML24, cml24-2 or cml24-4 mutant proteins were added to the beads with 1 mm CaCl2 or 1 mm EGTA, and incubated at 4°C for 1 h. After washing the beads three times with TBS containing 0.05% Triton X-100, 1 mm CaCl2 or EGTA, proteins bound to the beads were directly subjected to SDS–PAGE analysis. GST alone and calmodulin (Sigma) were used as controls.

Immunoblot analysis

Protein samples were mixed with 4× loading buffer (240 mm Tris/HCl pH 6.8, 8% SDS, 40% glycerol, 0.4 m DTT, 0.04% bromophenol blue), and boiled at 100°C for 5 min. Proteins were transferred to 0.2 mm nitrocellulose membrane after 12 or 15% SDS–PAGE separation with or without 6 m urea. Membranes were baked at 65°C for 30 min, and blocked in TBST buffer (50 mm Tris/HCl pH 7.4, 150 mm NaCl, 0.1% Tween-20) with 5% w/v non-fat dry milk, and incubated with mouse anti-His antibody (1:5000, MMS-156R; Covance,, mouse anti-HA (1:1000, 05-904; Millipore,, and MMS-101r; Covance), rabbit anti-ATG8a antibody (1:5000, purified from the antiserum against ATG8a)) or mouse anti-Myc (1:1000, CRL1729; American Type Culture Collection, overnight at 4°C. The antiserum against ATG8a (generously provided by Y. Ohsumi, National Institute of Basic Biology, Okazaki, Japan) was affinity-purified by passing the serum through a His–ATG8e–HA-coupled Ni-NTA column (Yoshimoto et al., 2004). Horseradish peroxidase-conjugated goat anti-mouse IgG (SC-2005; Santa Cruz Biotechnology, or goat anti-rabbit IgG (31460; Pierce, were used as the secondary antibody, and a Pierce Super Signal West Pico chemiluminescent kit (34080; Thermo Scientific, was used to detect the specific interacting secondary antibody.


Co-immunoprecipitation was performed as described by Kang and Klessig (2005). Agrobacterium tumefaciens GV2260, pBin61 carrying the GFP–HA construct, and A. tumefaciens harboring tobacco rattle virus-encoded (TRV) protein (Martínez-Priego et al., 2008), were gifts from Hong Gu Kang (Department of Biology,Texas State University, San Marcos, TX, USA). Two strains of Agrobacteria harboring pER-ATG4b-Myc and pER-CML24-HA constructs or harboring pER-ATG4b-Myc and pBin61-35S-GFP-HA constructs were injected onto the abaxial side of fully expanded tobacco leaves using 1 ml syringes. Agrobacterium harboring TRV was included to suppress potential silencing of the introduced transgenes. Estradiol, at 20 μmol, was sprayed onto the tobacco leaves 1 day after Agrobacterium injection to induce gene expression; leaves were harvested for protein preparation 2 days after estradiol treatment. Leaves were ground in liquid nitrogen and extracted in buffer containing 10% glycerol, 50 mm Tris/HCl pH 7.5, 2 mm EDTA, 150 mm NaCl, 5 mm DTT and 1 mm PMSF. Extracts were passed through NAP-5 columns (GE Healthcare) for exchange with to immuno-binding buffer (25 mm Tris/HCl, 500 mm NaCl, 1 mm EDTA, 0.5% Igepal CA-630 (I8896; Sigma) and 1 mm PMSF). A 50 μl aliquot of the total 1 ml of extract was cleared using agarose resin (1859011; Pierce) and used as the input loading sample. For immunoprecipitation of CML24–HA and GFP–HA protein, cleared extracts were incubated overnight with EZview Red anti-HA affinity gel (E6779; Sigma-Aldrich). The resin was then washed seven times with immuno-binding buffer before SDS–PAGE.

In vitro ATG4 activity analysis

Approximately 60 pmol of recombinant GST–ATG4b (approximately 78 kDa), His–ATG8e–HA (approximately 17 kDa) and CML24 (approximately 17 kDa) were used in vitro activity analysis. GST–ATG4b was mixed with or without CML24 in the presence of 5 mm Ca2+ or 5 mm EDTA at room temperature for 10 min, and then His–ATG8e–HA was added to the reactions. Then, the reactions were incubated at 30°C for 2 h. The samples were subjected to 15% SDS–PAGE, and proteins were detected by Western blotting with anti-His or anti-HA antibody.

For plant extract ATG4 activity analysis, 2-week-old soil-grown plants (Col-0, cml24-2, cml24-4, atg4a, atg4b and atg4a4b) grown under long-day photoperiods were collected during the light period or 1 h after a shift to darkness. Proteins were extracted with KT buffer (25 mm Tris pH 7.5, 50 mm KCl, 1 mm PMSF, 1 mm EDTA). Total protein (50 μg) was mixed with 0.2 μg His–ATG8e–HA as substrate, and incubated at 30°C for 2 h. Samples were subjected to 15% SDS–PAGE, and His–ATG8e proteins were detected by immunoblotting with the anti-His antibody. Gels were stained with Coomassie brilliant blue (CBB) and quantified using ImageJ software ( Immunoblotting signals were quantified using ImageJ software with CBB staining as a loading control.

Fluorescence microscopy

To visualize GFP fluorescence, GFP–ATG8e transgenic seedlings were grown vertically for 7 days and then transferred to medium without or with 1 μm concanamycin A (ALX-380-034; ENZO Life Sciences Inc., for 16 h in the dark. Excised roots were mounted in water and imaged in vivo using an Axioplan II microscope (Zeiss, with a narrow-band GFP filter (41020; Chroma Technology Images were analyzed using MetaMorph software (Molecular Devices Inc.,

Quantitative real-time RT-PCR

Total RNA was isolated using TRI Reagent® (Molecular Research Center Inc., Total RNA (1 μg) was DNase-treated with 10 units of DNase I (Roche Applied Science, at 37°C for 1 h, and heat-inactivated at 75°C for 10 min. Single-strand cDNA was generated by reverse transcription using SuperScript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Each quantitative PCR reaction contained 1× Thermocycler ABI SYBR Green PCR Master Mix (Applied Biosystems,, 500 nm of each forward and reverse gene-specific primer and 80 ng cDNA. The primers used were 5′-CAATTTGTATACGTGGTTCGT-3′ and 5′-AGCAACGGTAAGAGATCCAA-3′ for AtATG8a, 5′-TCTTTAAGATGGACGACGATTTC-3′ and 5′-CTCAGCCTTTTCCACAATCA-3′ for AtATG8e, 5′-TGTCAACAACACTCTCCCTCA-3′ and 5′-AACCAAAGGTTTTCTCACTGC-3′ for AtATG8i, and 5′-CTGTTTCCGTACCCTCAAGC-3′ and 5′-AGGGAAACGAAGACAGCAAG-3′ for TUB4. SYBR Green was detected using an ABI PRISM 7000 (Applied Biosystems). The difference in cycle number where product amplification resulted in a fixed threshold amount of fluorescence was determined by the following equation: ΔCT(sample) = ΔCT(TUB4) − ΔCT(gene of interest). One sample was chosen as a calibrator and ΔΔCT was determined for each sample according to the equation: ΔΔCT(sample) = ΔCT(sample) − ΔCT(calibrator). Relative RNA levels were calculated using 1/log2 [2^ ΔΔ CT(sample)].


We thank Hong Gu Kang (Texas State University, TX, USA) for Agrobacterium tumefaciens GV2260, pBin61 carrying the GFP–HA construct and A. tumefaciens harboring TRV, Dr Yoshinori Ohsumi (National Institute of Basic Biology, Okazaki, Japan) for ATG8a antibodies, Dr Diane Bassham (Iowa State University, Ames, IA, USA) for ATG8e:GFP transgenic plants, and Dr Rick Vierstra (University of Wisconsin, Madison, WI, USA) for the atg4a4b, atg5-1 and atg10-1 mutants – their generosity is greatly appreciated. We are also grateful to Hong Gu Kang (Texas State University, TX, USA) for sharing advice and expertise in tobacco leaf transformation and co-immunoprecipitation, and to the Braam laboratory members for critically reviewing the manuscript. This material is based upon work supported by the US National Science Foundation under grant number MCB 0817976 and by support from the Virginia and L.E. Simmons Family Foundation to J.B.