AtATG18a is required for the formation of autophagosomes during nutrient stress and senescence in Arabidopsis thaliana


  • Yan Xiong,

    1. Department of Genetics, Development and Cell Biology,
    2. Interdepartmental Plant Physiology Program and
    Search for more papers by this author
  • Anthony L. Contento,

    1. Department of Genetics, Development and Cell Biology,
    2. Plant Sciences Institute, 253 Bessey Hall, Iowa State University, Ames, IA 50011, USA
    Search for more papers by this author
  • Diane C. Bassham

    Corresponding author
    1. Department of Genetics, Development and Cell Biology,
    2. Interdepartmental Plant Physiology Program and
    3. Plant Sciences Institute, 253 Bessey Hall, Iowa State University, Ames, IA 50011, USA
      (fax 515 294 1337; e-mail
    Search for more papers by this author

(fax 515 294 1337; e-mail


Vacuolar autophagy is a major pathway by which eukaryotic cells degrade macromolecules, either to remove damaged or unnecessary proteins, or to produce respiratory substrates and raw materials to survive periods of nutrient deficiency. During autophagy, a double membrane forms around cytoplasmic components to generate an autophagosome, which is transported to the vacuole. The outer membrane fuses with the vacuole or lysosome, and the inner membrane and its contents are degraded by vacuolar or lysosomal hydrolases. We have identified a small gene family in Arabidopsis thaliana, members of which show sequence similarity to the yeast autophagy gene ATG18. Members of the AtATG18 gene family are differentially expressed in response to different growth conditions, and one member of this family, AtATG18a, is induced both during sucrose and nitrogen starvation and during senescence. RNA interference was used to generate transgenic lines with reduced AtATG18a expression. These lines show hypersensitivity to sucrose and nitrogen starvation and premature senescence, both during natural senescence of leaves and in a detached leaf assay. Staining with the autophagosome-specific fluorescent dye monodansylcadaverine revealed that, unlike wild-type plants, AtATG18a RNA interference plants are unable to produce autophagosomes in response to starvation or senescence conditions. We conclude that the AtATG18a protein is likely to be required for autophagosome formation in Arabidopsis.


The vacuole (or lysosome) is a major site of macromolecule degradation in eukaryotes, and contains a variety of hydrolases for this purpose. Substrates for vacuolar degradation are taken up into the vacuole from the cytoplasm by autophagy, a pathway that is usually non-selective and is able to sequester large quantities of cytoplasm for degradation, including entire organelles (Klionsky and Ohsumi, 1999). Autophagy is required for responses to nutrient deficiency, heat shock, and other environmental stresses, as well as normal development of some organisms (Klionsky, 2004). Several stages can be distinguished in the breakdown of an organelle or portion of cytoplasm by autophagy. First, the cytoplasm to be degraded is surrounded by a double membrane and thus segregated away from the bulk cytosol to form an autophagosome. Secondly, the autophagosome is delivered to the vacuole, and its contents are deposited inside the vacuole lumen, either by fusion of the outer autophagosome membrane with that of the vacuole, or by engulfment of the autophagosome by the vacuole. Finally, the inner autophagosomal membrane and cytoplasmic components are degraded by vacuolar enzymes.

Morphological studies have demonstrated the presence of an autophagy pathway for uptake of components by plant vacuoles. It can be observed most easily by light and electron microscopy when plant cells are exposed to nutrient-limiting conditions, such as in suspension-cultured cells in the absence of sucrose (Aubert et al., 1996; Chen et al., 1994; Contento et al., 2004; Moriyasu and Ohsumi, 1996). While nutrient starvation causes very extensive autophagy, autophagy also occurs in the absence of stress conditions during specific developmental stages. For example, autophagosomes accumulate during embryo development in an Arabidopsis mutant defective in vacuole biogenesis, and thus unable to degrade the autophagosomes normally (Rojo et al., 2001). In some species such as wheat, storage proteins are delivered to the vacuole by a mechanism that resembles autophagy (Levanony et al., 1992). Non-selective autophagy may also play a role in senescing tissue, where organelles and cytosolic components are degraded to recycle their constituents for use elsewhere in the plant (Matile, 1997).

Genetic studies in yeast have led to the isolation of a number of genes that are required for autophagy (Harding et al., 1995; Thumm et al., 1994; Tsukada and Ohsumi, 1993), and have provided insight into the mechanism of autophagosome formation. Yeast possesses two related autophagy pathways, a regulated pathway induced by nutrient starvation, and the cytoplasm-to-vacuole transport (Cvt) pathway, a constitutive autophagy pathway that delivers certain newly synthesized vacuolar enzymes to the vacuole. These two pathways share many components, and the majority of yeast autophagy mutants are defective in both pathways (Scott et al., 1996). A protein kinase complex, the Atg1 complex, is thought to control the transition from the Cvt pathway to the autophagy pathway during starvation (Scott et al., 2000). Autophagosome formation requires the activity of a phosphatidylinositol-3 kinase complex and two unusual ubiquitin-related protein conjugation systems. An Atg12-Atg5 conjugate is required for autophagosome formation (Mizushima et al., 1998) and associates with the pre-autophagosomal structure (PAS; Suzuki et al., 2001), a structure from which autophagosomes may originate. In addition, Atg8 is conjugated to phosphatidylethanolamine and localizes to autophagosomes themselves (Abeliovich et al., 2000), possibly regulating their expansion. Autophagosomes are then transported to and fuse with the vacuole, in a reaction that depends on components of the vacuolar fusion machinery (Ishihara et al., 2001).

Homologs of many of these yeast autophagy genes are present in multicellular eukaryotes (Yoshimori and Mizushima, 2004). Two reports have demonstrated that Arabidopsis genes with sequence similarity to yeast autophagy genes are likely to function in autophagy. Doelling et al. (2002) identified an Arabidopsis knockout mutant with a T-DNA insertion in an Arabidopsis ATG7-like gene. Yeast Atg7 is an E1-like enzyme that functions in the protein conjugation reaction for both Atg12 and Atg8 (Mizushima et al., 1998). This conjugation process in turn is required for the formation of autophagosomes, and thus a yeast atg7 mutant is defective in autophagy. Likewise, Hanaoka et al. (2002) isolated an Arabidopsis mutant in an ATG9-like gene, again based on homology to the yeast ATG9 gene, which encodes an integral membrane protein required for autophagosome formation (Lang et al., 2000; Noda et al., 2000). Both the atg7 and atg9 Arabidopsis knockout mutants were unable to survive under nutrient-deficient growth conditions, and showed early senescence phenotypes, indicating that the function of these genes contributes both to nutrient stress responses and to regulation of senescence. This is indicative of a function in autophagy, although this has not been assayed directly. In addition, a mutation in AtVTI12, which encodes a vesicle fusion protein, causes similar phenotypes, and AtVTI12 may be required for fusion of autophagosomes with the vacuole (Surpin et al., 2003).

The yeast ATG18 gene is required for both starvation-induced autophagy and the Cvt pathway, apparently at an early stage in autophagosome formation, as atg18 mutants are unable to accumulate autophagosomes in response to starvation (Barth et al., 2001; Guan et al., 2001). The Atg18 protein contains two WD-40 domains and has been predicted to form a β-propeller structure that binds phosphatidylinositol 3,5-bisphosphate (Dove et al., 2004) or phosphatidylinositol 5-phosphate (Stromhaug et al., 2004). Atg18 is associated with the vacuolar surface or with membrane structures close to the vacuole (Barth et al., 2001; Dove et al., 2004; Guan et al., 2001), and is involved in recycling of proteins from the vacuole (Dove et al., 2004) and Atg9 localization (Reggiori et al., 2004). Interestingly, unlike other identified proteins involved in autophagy, yeast contains two additional proteins with sequence similarity to Atg18. Atg21 is required for the Cvt pathway, but not for regulated autophagy, and shows a similar subcellular localization to Atg18 (Barth et al., 2002; Meiling-Wesse et al., 2004). The third ORF, YGR223c, is of unknown function (Barth et al., 2002; Georgakopoulos et al., 2001), but shows a similar localization to and genetic interactions with ATG18 and ATG21 (Stromhaug et al., 2004). Therefore, these three similar yeast genes appear to have related, but not redundant, functions in autophagy.

Previous research into the potential function of Arabidopsis genes in autophagy has focused on single genes, rather than gene families (Doelling et al., 2002; Hanaoka et al., 2002). A comparison of the sequence of yeast ATG18 with the Arabidopsis genome sequence revealed the existence of multiple Arabidopsis ATG18-like genes (Barth et al., 2001). The presence of a small gene family may indicate functional specialization of ATG18-like proteins in Arabidopsis, for example in different stages of the autophagy process, during different stages of development, or under different growth conditions. We now show that the eight members of this family are differentially expressed under various conditions, and that one member is potentially required for the formation of autophagosomes both during nutrient stress and senescence.


AtATG18s are potential WD-40 repeat proteins folded as seven-bladed β-propellers

blast searches (Altschul et al., 1990) of the Arabidopsis genome revealed eight potential encoded proteins related to Atg18, named AtATG18a–AtATG18h (see Experimental procedures for accession numbers). Sequence comparisons of the full-length proteins revealed that the predicted Arabidopsis proteins shared between 39% (AtATG18b) and 21% (AtATG18g) amino acid identity with yeast Atg18 (50–30% similarity, respectively, allowing for conservative substitutions). Based on phylogenetic analysis (Figure 1a), these eight predicted proteins appear to form three major subgroups: AtATG18a, c, d, and e cluster with the yeast protein Ygr223c, AtATG18b is most similar to Atg18 itself, and AtATG18f, g, and h are more divergent, forming a separate clade. None of the Arabidopsis proteins appear closely related to the yeast Cvt pathway-specific protein Atg21. To determine whether the clustering of Arabidopsis genes with yeast Ygr223c or Atg18 is statistically significant, the maximum likelihood (ML) tree was compared with all possible tree topologies containing the three yeast sequences constrained into a single clade. Approximately unbiased analysis indicated that the ML tree was significantly more likely than the best constrained tree (P = 0.996).

Figure 1.

AtATG18 gene family sequence analysis.
(a) Phylogenetic tree of AtATG18 gene family. The maximum likelihood tree for yeast and Arabidopsis Atg18-like proteins was identified using the proml program within the Phylip 3.62 package (Felsenstein, 2002). Numbers show bootstrap values for the indicated clades based on the maximum likelihood distances.
(b) Numbers and locations of WD-40 domains in yeast and Arabidopsis ATG18 family amino acid sequences. The shadings indicate the WD-40 domains. Numbers indicate the number of amino acids in each predicted protein.

Sequence analysis with Pfam (Bateman et al., 2004) indicates that all AtATG18 genes are predicted to encode proteins with one to three WD-40 repeats; the numbers and locations of WD-40 domains in the AtATG18s, compared with the yeast proteins, are shown in Figure 1(b). All of the AtATG18 proteins are predicted by the 3D-PSSM fold recognition server (Kelley et al., 2000; to fold as seven-bladed β-propellers with high similarity scores when compared with known seven-bladed propeller proteins (transducin β-subunit and the C-terminal domain of Tup1). No other conserved functional domains were identified in any of the predicted Arabidopsis proteins.

Expression pattern of AtATG18s

To determine whether all AtATG18 genes have detectable transcript levels, and identify the organs where the genes are expressed, specific primers for each of the AtATG18s were designed and RT-PCR was performed using RNA from root, leaf, inflorescence stem, and flower. Seven of the AtATG18 transcripts were detected in all or some of the organs tested (Figure 2a), with AtATG18a, b, c, and d transcripts detected in all organs, and AtATG18f, g, and h having a more restricted expression pattern. AtATG18e was not detectable in any organ or under any of the conditions tested.

Figure 2.

Expression pattern of AtATG18 genes determined by RT-PCR analysis.
(a) Expression pattern in different organs. R, root; L, leaf; F, flower; S, stem.
(b) Expression pattern during sucrose or nitrogen starvation of Arabidopsis seedlings. One-week-old seedlings were transferred to medium lacking sucrose (−SUC) and incubated in the dark, or to medium lacking nitrogen (−N), for the indicated times.
(c) Expression pattern during senescence of detached leaves. First and second true leaves were detached from 2-week-old seedlings and incubated in the dark for the indicated times. 18S RNA was used as a loading control; AtSEN1 and AtYSL4 were used as senescence controls. At least three independent biological replicates were performed for each experiment, with similar results.

Based on morphological analyses (Aubert et al., 1996; Moriyasu and Ohsumi, 1996) and mutant phenotypes (Doelling et al., 2002; Hanaoka et al., 2002) autophagy is thought to function in nutrient deprivation and senescence in plants. To investigate the expression pattern of the AtATG18 genes during nutrient deprivation, RT-PCR was performed using RNA from seedlings starved for sucrose or nitrogen. Four genes (AtATG18a, f, g, and h) showed an increase in transcript level in both sucrose and nitrogen starvation conditions (Figure 2b) whereas the others show no change under these conditions.

To investigate the gene expression pattern during senescence, detached leaf assays were performed. First and second true leaves from 2-week-old seedlings were detached and incubated in the dark for up to 48 h, and RNA isolated. RT-PCR revealed only one gene, AtATG18a, whose transcript level increases during this artificial senescence process (Figure 2c). AtSEN1 and AtYSL4, two senescence-associated genes (Hanaoka et al., 2002), were used as senescence expression controls. Comparing the expression pattern of the AtATG18 genes during starvation and senescence, only AtATG18a is upregulated in both conditions, suggesting that AtATG18a may have an important role in these two conditions. We therefore decided to focus initially on the functional analysis of this member of the gene family.

Generation of AtATG18a RNAi transgenic plants

To investigate the physiological role of AtATG18a in Arabidopsis, RNA interference (RNAi) was used to generate transgenic plants with a reduced AtATG18a expression level. Arabidopsis thaliana plants were transformed (Clough and Bent, 1998) with an RNAi construct consisting of an inverted repeat of a unique 500 bp region of the AtATG18a gene, with a portion of the GUS gene as a linker (Chuang and Meyerowitz, 2000), driven by the cauliflower mosaic virus 35S promoter. Transformants were screened by RT-PCR for AtATG18a transcript levels, and three RNAi lines with reduced AtATG18a transcript level were identified. The reduction in AtATG18a transcript level was gene-specific, as the other AtATG18s transcript levels were unchanged compared with those of wild-type (WT) plants (Figure 3).

Figure 3.

Expression of AtATG18 genes in AtATG18a RNAi transgenic plants. Total RNA was isolated from 2-week-old seedlings of wild type (WT) and three independent AtATG18a RNAi lines, RNAi-1, RNAi-2, and RNAi-3, followed by RT-PCR analysis. 18S RNA was used as a loading control.

AtATG18a RNAi plants are more sensitive to nutrient deprivation conditions

To investigate the role of AtATG18a during nutrient starvation, growth of the AtATG18a RNAi lines was compared with WT plants under nitrogen or sucrose starvation conditions. For nitrogen starvation, 7-day-old seedlings grown on nutrient Murashige and Skoog (MS) solid medium were transferred to nitrogen-depleted MS solid medium and incubated in long-day light conditions. After 10 days of nitrogen starvation, the AtATG18a RNAi seedlings began to lose their green color and accumulate anthocyanins, whereas WT seedlings remained green (Figure 4a). After 20 days, the AtATG18a RNAi seedlings were brown, compared with the WT seedlings that retained some chlorophyll. Growth under carbon starvation conditions was also analyzed. For carbon starvation, 7-day-old seedlings grown on nutrient MS solid medium were transferred to MS solid medium without sucrose and incubated in 24-h dark conditions. RNAi seedlings were hypersensitive to carbon starvation, as chlorosis was observed earlier in all three AtATG18a RNAi lines compared with WT seedlings (Figure 4c). After 20 days, the cotyledons of AtATG18a RNAi seedlings were yellow whereas WT cotyledons retained a pale green color.

Figure 4.

Phenotype of AtATG18a RNAi plants under nutrient starvation conditions.
(a) One-week-old seedlings of WT and AtATG18a RNAi plants were transferred to MS medium lacking nitrogen and incubated for the indicated times.
(b) Chlorophyll content of WT and AtATG18a RNAi lines was measured at the indicated times after transfer to nitrogen-deficient medium.
(c) One-week-old seedlings of WT and AtATG18a RNAi plants were transferred to MS medium lacking sucrose and incubated in the dark for the indicated times.
(d) Chlorophyll content of WT and AtATG18a RNAi lines was measured at the indicated times after transfer to sucrose-deficient medium and dark incubation. Each experiment was repeated at least three times with similar results. Error bars indicate standard deviation.

To further confirm the observed phenotype, the chlorophyll content of WT and RNAi seedlings was analyzed throughout the nitrogen and sucrose starvation time courses. Total chlorophyll per gram fresh weight of tissue was calculated according to Chapman (1988) and expressed as a percentage of the original chlorophyll content at 0 days. The chlorophyll content of AtATG18a RNAi plants decreased more rapidly during both nitrogen and carbon starvation compared with WT plants (Figure 4b,d). These results indicate that AtATG18a RNAi transgenic plants are more sensitive to nutrient deprivation conditions, and therefore AtATG18a may function in the response of plants to starvation.

Natural senescence and detached leaf senescence are accelerated in AtATG18a RNAi plants

Under normal nutrient medium and soil conditions, the AtATG18a RNAi plants did not exhibit any differences compared with WT plants in rate or efficiency of seed germination, elongation of the root system, seedling development, or flowering time (data not shown). However, after bolting, the rosette leaves of AtATG18a RNAi plants senesced faster than WT plant rosette leaves (Figure 5a). Rosette leaves of RNAi plants were indistinguishable from WT rosette leaves at 30 days after germination. At 45 days, the rosette leaves of AtATG18a RNAi plants began to turn yellow, whereas WT leaves did not begin to senesce until around 60 days. After 75 days, most of the RNAi plant rosette leaves had senesced and were brown, whereas most of the WT rosette leaves remained green.

Figure 5.

AtATG18a RNAi plants exhibit an early senescence phenotype.
(a) Natural senescence. Wild type and AtATG18a RNAi plants were grown in soil under long-day conditions for the indicated times.
(b) Detached leaf senescence. First and second true leaves were detached from 2-week-old WT and AtATG18a RNAi seedlings and incubated in the dark for the indicated times.
(c) Comparison of the expression level of the senescence-associated genes AtSEN1 and AtYSL4 in WT and AtATG18a RNAi-detached leaves after the indicated times.

To better understand the role of AtATG18a during leaf senescence, a detached leaf assay was used (Weaver et al., 1998). The first and second true leaves of 2-week-old WT and AtATG18a RNAi seedlings, grown on nutrient-rich MS medium, were detached and incubated in the dark. Leaves from each of the RNAi lines displayed accelerated senescence when compared with WT leaves, demonstrated by the visible loss of chlorophyll (Figure 5b). The expression of two Arabidopsis leaf senescence marker genes, AtSEN1 and AtYSL4, was also examined during the detached leaf assay time course by RT-PCR (Figure 5c). For both genes, an increase in expression was observed earlier in the RNAi leaves than in WT leaves, with AtSEN1 increasing after 12 h in the RNAi lines compared with 24 h in the control. AtYSL4 was apparently already expressed at high levels in the RNAi lines, even at the 0 h time point, when compared with WT plants. This provides a molecular confirmation of the early senescence phenotype observed.

Autophagosome formation is disrupted in AtATG18a RNAi plants

In yeast, Atg18 is required for autophagosome formation (Barth et al., 2001). Monodansylcadaverine (MDC) is a fluorescent drug that specifically stains autophagosomes in animals (Munafo and Colombo, 2001) and Arabidopsis (Contento et al., 2005). To investigate whether AtATG18a is required for autophagosome formation in Arabidopsis, 7-day old WT and AtATG18a RNAi seedlings were transferred to nitrogen or sucrose starvation conditions for 5 days, and then stained with MDC. Numerous moving MDC-stained autophagosomes accumulated in WT root cells during starvation, observed by fluorescence microscopy. This accumulation was disrupted in AtATG18a RNAi root cells (Figure 6), with no autophagosomes visible after either nitrogen or sucrose starvation.

Figure 6.

MDC staining of root tips. One-week-old WT and AtATG18a RNAi seedlings were transferred to fresh control solid MS medium for 5 days (a), to nitrogen-deficient solid MS medium and incubated under long-day conditions for 5 days (b), or transferred to solid MS medium without sucrose in the dark for 5 days (c), followed by staining with MDC. Insets show an enlargement of the boxed areas. Scale bar = 50 μm for main figures, 10 μm for insets.

To confirm that the lack of labeled structures in the RNAi lines is due to a loss of autophagosomes, rather than a change in their structure that prevents MDC staining, protoplasts from WT and RNAi plants were transformed with GFP-AtATG8e and starved for 2 days. Numerous motile GFP-labeled autophagosomes were seen in WT protoplasts, whereas only diffuse labeling was seen in the RNAi protoplasts (Figure 7), similar to that seen in WT protoplasts prior to starvation (Contento et al., 2005). We conclude that loss of AtATG18a prevents autophagosome formation during starvation.

Figure 7.

GFP-AtATG8e expression in sucrose-starved protoplasts. Protoplasts from WT or AtATG18a RNAi leaves were transiently transformed with GFP-AtATG8e and incubated for 3 days in the presence (a) or absence (b) of sucrose. Scale bar = 20 μm.

To determine the role of AtATG18a during senescence, protoplasts were prepared from WT- and RNAi-detached leaves after 3 days of dark incubation, and then stained with MDC (Figure 8). Whereas MDC-stained autophagosomes were abundant in the WT protoplasts, very few were seen in the AtATG18a RNAi protoplasts, indicating that senescence-induced autophagy also requires AtATG18a. These data imply that AtATG18a is required for autophagosome formation in Arabidopsis during nutrient stress and senescence, and further indicate that the phenotypes of hypersensitivity to nutrient deprivation and early senescence are caused by defects in normal autophagy function.

Figure 8.

MDC staining of protoplasts from senescing detached leaves.
(a) The first and second true leaves of 2-week-old WT and AtATG18a RNAi seedlings were detached, protoplasts were prepared immediately from these leaves followed by staining with MDC. Scale bar = 20 μm.
(b) The first and second true leaves of 2-week-old WT and AtATG18a RNAi seedlings were detached and incubated in the dark for 3 days. Protoplasts were prepared from these detached leaves and were stained with MDC.
(c) Number of protoplasts containing autophagosomes in WT and AtATG18a RNAi leaves 3 days after detaching from the plant. The experiment was repeated three times, and 500 protoplasts were counted each time. Error bars indicate standard deviation.


Autophagy is a key process in the breakdown of cytoplasmic components for survival during environmental stresses, and for cellular remodeling during development. We have identified a family of eight Arabidopsis genes that show sequence similarity to yeast ATG18, and demonstrated that at least one of these genes, AtATG18a, functions in autophagy during nutrient starvation and senescence.

AtATG18a was chosen for initial studies because of its unique expression pattern; of the ATG18 Arabidopsis homologs, AtATG18a was the only gene that displayed increased expression both during nitrogen and carbon starvation and during senescence (Figure 2). In agreement with its expression pattern, a reduction in AtATG18a RNA level by RNAi caused increased sensitivity to nitrogen and carbon starvation and an early senescence phenotype (Figures 4 and 5). These are similar to the phenotypes observed for knockout mutants of other Arabidopsis homologs of yeast autophagy genes, AtATG7 and AtATG9, and appear to be typical characteristics of mutants in the autophagy pathway (Doelling et al., 2002; Hanaoka et al., 2002). However, a direct correlation between autophagy and whole plant phenotypes has not previously been shown. To address this issue, the autophagosome-specific fluorescent dye MDC was used to stain roots of WT and AtATG18a RNAi-intact seedlings. After nitrogen or carbon starvation, MDC-stained autophagosomes are visible in WT roots, whereas AtATG18a RNAi seedlings lack these structures (Figure 6). Similar results are seen in protoplasts from RNAi leaves during sucrose starvation (Figure 7 and Y. Xiong and D.C. Bassham, unpublished data) or senescence (Figure 8). This suggests that the AtATG18a protein is required for the proper formation of autophagosomes in Arabidopsis, although the stage at which autophagosome formation is blocked remains to be determined. Yeast Atg18 is required for cycling of Atg9 through the PAS (Reggiori et al., 2004). AtATG18a may function in a similar manner to ensure correct localization of AtATG9, and thus to allow autophagy to proceed.

Atg18 also appears to be required for recycling of membrane proteins from the vacuole to endosomes in yeast (Dove et al., 2004), even under non-stressed conditions. A role for AtATG18a in vacuolar trafficking may be unlikely, as growth and development of AtATG18a RNAi plants is relatively normal under nutrient-rich conditions. In contrast, unlike in yeast, mutations in vacuolar trafficking components are often lethal in plants (Rojo et al., 2001; Sanderfoot et al., 2001; Surpin et al., 2003). It remains possible that one of the other Atg18 homologs in Arabidopsis functions in vacuolar trafficking or membrane recycling pathways. AtATG18b, c, and d are expressed ubiquitously throughout the plant, and their expression level does not change in response to starvation; these genes are therefore candidates for a role in a constitutive trafficking pathway. Isolation of knockout mutants in these members of the family should allow this possibility to be addressed.

The Arabidopsis autophagy genes that have been studied previously, AtATG7 and AtATG9, are single-copy genes. We are interested in understanding why ATG18 exists as a small gene family in Arabidopsis, and the functions of the remaining members of this family. AtATG18a is not functionally redundant with other members of the family, as the AtATG18a RNAi lines show a clear phenotype. These lines are specific for a reduction in this member of the family, with no effect on expression of the other AtATG18 genes. Whether other members of the gene family share some functional redundancy remains to be seen. We have isolated a T-DNA insertion mutant in AtATG18c, but no visible phenotype has been observed, either under normal growth conditions, or during sucrose or nitrogen starvation. In addition, MDC staining revealed that autophagosome formation proceeds normally in this mutant (see Supplementary Material), indicating that this function is specific to the AtATG18a protein. AtATG18c is very closely related to AtATG18d, and it is possible that AtATG18d can compensate for the loss of AtATG18c in the knockout mutant. Isolation of an Atatg18d mutant, and generation of the Atatg18c/Atatg18d double mutant should allow this to be resolved.

Sequence comparisons indicate that AtATG18a, along with c, d, and e, is in fact more closely related to the yeast Atg18 homolog Ygr223c than to Atg18 itself. Ygr223c shares some properties with Atg18 and Atg21, including lipid-binding and subcellular localization (Dove et al., 2004; Stromhaug et al., 2004), but its function is not yet known. None of the Arabidopsis genes appear closely related to Atg21; this is not entirely unexpected, as in yeast, Atg21 is required only for the Cvt pathway, and not for starvation-induced autophagy (Barth et al., 2002; Meiling-Wesse et al., 2004; Stromhaug et al., 2004). There is no evidence for the existence of a Cvt pathway in multicellular organisms, and the presence of Atg21 may therefore be an adaptation specific to yeast for this unusual trafficking pathway.

It is interesting that, of all of the AtATG18 genes, only AtATG18a is upregulated during senescence (Figure 2c). This could indicate that only this member of the family is involved in senescence. However, care should be taken when drawing conclusions from expression data alone. It should be noted that in yeast, many of the proteins required for autophagy are constitutively expressed under normal growth conditions; only a few, such as Atg8, are induced by starvation (Kirisako et al., 1999). Further analysis is therefore needed to determine whether any other AtATG18 genes also function during senescence.

Three of the Arabidopsis genes (AtATG18f, g, and h) cluster together in the phylogenetic tree and are unusual in that the predicted proteins are much larger than either yeast Atg18 or the other Arabidopsis homologs (Figure 1). While these proteins are more closely related to Atg18 than to any other yeast protein, it is not clear whether they are functionally related to Atg18, or whether they have an alternative function. However, they are all upregulated during nutrient starvation, and therefore may potentially have a role in the response to these stress conditions. In addition, all three genes increase in expression when exposed to other abiotic stresses, AtATG18g under drought conditions and AtATG18f and h in response to a combination of heat and drought stress (Rizhsky et al., 2004), introducing the intriguing possibility that different members of the family may be required during different stress conditions.

In conclusion, we have identified a family of ATG18-related genes from Arabidopsis that are differentially regulated during development and environmental stresses. One member of this family, AtATG18a, is required for the formation of autophagosomes during starvation and senescence. The functions of the remaining members of the family are currently being investigated through a detailed analysis of gene expression under different conditions, including additional abiotic and biotic stresses, and determination of the subcellular localization of the encoded proteins. The isolation and phenotypic characterization of knockout mutants or RNAi lines defective in one or a combination of the AtATG18 genes will allow their specific functions in stress responses and membrane trafficking processes to be elucidated.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana plants were grown under long-day conditions (16 h light) at 22°C. Seeds were surface-sterilized in 33% bleach and 0.1% Triton X-100 solution for 20 min followed by cold treatment for at least 2 days.

For nitrogen starvation experiments, 1-week-old seedlings grown on nutrient solid MS medium [Murashige–Skoog Vitamin and Salt Mixture (Gibco BRL, Gaithersburg, MD, USA), 1% sucrose, 2.4 mm MES (pH 5.7) and 0.8% phytagar] were transferred to nitrogen-depleted MS solid medium as described by Doelling et al. (2002) and incubated under long-day conditions. For sucrose starvation experiments, 1-week-old seedlings grown on nutrient solid MS medium were transferred to MS solid medium without sucrose and incubated in the dark.

For the detached leaf assay, first and second true leaves from 2-week-old seedlings grown on nutrient solid MS medium were excised and incubated on 3 MM paper wetted with 3 mm MES (pH 5.7) in the dark (Doelling et al., 2002). These leaves were used for phenotypic observation, RNA isolation, and protoplast isolation after the indicated times.

Identification of AtATG18 genes

AtATG18 genes were identified by searching the Arabidopsis genomic database with the Saccharomyces cerevisiae Atg18 protein sequence using the blast algorithm (Altschul et al., 1990). Eight AtATG18s were identified as follows, with the corresponding loci in parentheses; AtATG18a (At3g62770), AtATG18b (At4g30510), AtATG18c (At2g40810), AtATG18d (At3g56440), AtATG18e (At5g05150), AtATG18f (At5g54730), AtATG18g (At1g03380), and AtATG18h (At1g54710).

Phylogenetic analysis

The ML tree for our data was identified using the PROML program within the Phylip 3.62 package (Felsenstein, 2002), with the JTT matrix of amino acid substitutions and eight categories of substitution rates. The ML tree was compared to alternative topologies using codeml (Yang, 1997) and consel (Shimodaira and Hasegawa, 2001) programs. Bootstrap values were calculated by ML distance analysis using seqboot (Felsenstein, 2002), tree-puzzle 5.2 (Strimmer and von Haeseler, 1996) and weighbor 1.2.1 (Bruno et al., 2000) programs and the puzzleboot script from Mike Holder and Andrew Roger ( ML distances were calculated based on the JTT matrix, observed frequencies of amino acids, gamma distributed rates with eight categories, and an alpha parameter estimated from the data set.

RT-PCR analysis of AtATG18s, AtSEN1, and AtYSL4 expression

Total RNA was extracted from each sample using the TRIzol RNA isolation method ( with DNase I treatment, and cDNAs were generated using Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA, USA) and an oligo dT primer. Primers used for PCR are shown in Table 1.

Table 1.  Primers used for analysis of AtATG18s, AtSEN1, and AtYSL4 expression
 Forward primersReverse primers

Generation of AtATG18a RNAi transgenic plants

According to the strategy described by Chuang and Meyerowitz (2000), gene-specific sense and antisense fragments of AtATG18a were amplified by RT-PCR using the following primers: sense strand (5′-AGCGGATATCCTTAATTGCGATCCCTTTCG-3′ and 5′-TCGCTCTAGAGATCCGAACCAGAGTACCCTTA-3′), antisense strand (5′-AGCGGAATTCCTTAATTGCGATCCCTTTCG-3′ and 5′-ACGCGGATCCGATCCGAACCAGAGTACCCTTA-3′). Suitable restriction enzyme sites (EcoRV, XbaI, EcoRI, and BamHI, underlined) were introduced in the primer sequences. The sense and antisense DNA fragments were linked with a 1-kb spacer coding for a partial sequence of GUS. The entire fragment containing sense, spacer, and antisense region were cloned into the plant T-DNA binary vector pCGN, driven by the Cauliflower Mosaic Virus 35S promoter. This construct was introduced into Agrobacterium tumefaciens strain GV2260 by electroporation (Mersereau et al., 1990).

The RNAi construct was introduced into A. thaliana Columbia-0 plants by Agrobacterium-mediated transformation using the floral dip method (Clough and Bent, 1998). RT-PCR was performed to determine the expression level of each AtATG18 gene in these transformants. Homozygous T2 transformant seeds with reduced AtATG18a mRNA level were used for further studies.

Isolation of AtATG18c insertion mutant

A T-DNA insertion mutant in the 5′-untranslated region of AtATG18c (Salk_009459) was obtained from the Arabidopsis Biological Resource Center. Homozygous plants were identified by PCR from genomic DNA. Primers used were AtATG18c gene-specific primers LP 5′-AAGAAAACGCAGACACGTGAA-3′ and RP 5′-CTCTCCTCGATTGAGACCAGG-3′, and T-DNA left border 5′-GCGTGGACCGCTTGCTGCAAC-3′ and right border 5′-CCGGACAGGTCGGTCTTGACAA-3′. Reduction in expression of AtATG18c in the mutant line was confirmed by RT-PCR using the primers described in Table 1.

Determination of chlorophyll content

Seedlings were ground in 80% acetone and the absorbance of the supernatant was measured at 663 and 646 nm. Total chlorophyll content per gram fresh weight of tissue was calculated according to Chapman (1988).

Labeling of protoplasts and seedling roots with MDC or GFP-AtATG8e

Protoplasts were prepared from detached leaves according to Sheen (2002) and immediately stained with 0.05 mm MDC (Sigma, St Louis, MO, USA) in phosphate-buffered saline (PBS) supplemented with 0.4 m mannitol for 10 min, then washed twice with PBS and 0.4 m mannitol to remove excess MDC. Seedlings were stained by immersion in 0.05 mm MDC in PBS for 10 min and washed twice with PBS (Contento et al., 2005). After staining, the protoplasts or roots of seedlings were observed using fluorescence microscopy with a DAPI-specific filter. Transformation of WT and RNAi protoplasts with GFP-AtATG8e was performed as described in Contento et al. (2005).


We thank Dr Dennis Lavrov for assistance with the phylogenetic analysis and Dr David Oliver and Dr Robert Thornburg for helpful comments on the manuscript. This research was supported by the Plant Responses to the Environment Program of the National Research Initiative Competitive Grants Program, US Department of Agriculture (grant no. 2002-35100-12034 to D.C.B.) and by the Iowa State University Plant Sciences Institute.

Supplementary Material

The following material is available from

Figure S1. Analysis of AtATG18c mutant phenotype. (a) RT-PCR was performed using RNA extracted from wild-type (WT) and AtATG18c T-DNA insertion mutant (KO) lines with primers specific for AtATG18c, or 18S RNA as a control. (b) One-week-old AtATG18c mutant seedlings were transferred to control medium (+N) or to MS medium lacking nitrogen (−N) and incubated for 5 days, followed by MDC staining. Inset shows an enlargement of the boxed area. Scale bar = 50 μm for main figure, 25 μm for inset.