The Starch Granule-Associated Protein EARLY STARVATION1 Is Required for the Control of Starch Degradation in Arabidopsis thaliana Leaves

Two proteins present in leaf starch granules are important for the control of starch turnover, allowing plants to match the depletion of starch reserves to the length of the night.

Abstract

To uncover components of the mechanism that adjusts the rate of leaf starch degradation to the length of the night, we devised a screen for mutant Arabidopsis thaliana plants in which starch reserves are prematurely exhausted. The mutation in one such mutant, named early starvation1 (esv1), eliminates a previously uncharacterized protein. Starch in mutant leaves is degraded rapidly and in a nonlinear fashion, so that reserves are exhausted 2 h prior to dawn. The ESV1 protein and a similar uncharacterized Arabidopsis protein (named Like ESV1 [LESV]) are located in the chloroplast stroma and are also bound into starch granules. The region of highest similarity between the two proteins contains a series of near-repeated motifs rich in tryptophan. Both proteins are conserved throughout starch-synthesizing organisms, from angiosperms and monocots to green algae. Analysis of transgenic plants lacking or overexpressing ESV1 or LESV, and of double mutants lacking ESV1 and another protein necessary for starch degradation, leads us to propose that these proteins function in the organization of the starch granule matrix. We argue that their misexpression affects starch degradation indirectly, by altering matrix organization and, thus, accessibility of starch polymers to starch-degrading enzymes.

INTRODUCTION

Normal growth rates in many plants are dependent upon the accumulation of starch as a product of photosynthesis during the day and its controlled utilization as a carbon source for growth during the night. In Arabidopsis thaliana leaves, up to half of the photosynthetically assimilated carbon accumulates as starch in leaf chloroplasts. At night, conversion of starch to sucrose proceeds at a near linear rate, such that ∼95% of starch is consumed by dawn (Gibon et al., 2004, 2009; Graf et al., 2010; Sulpice et al., 2014). This pattern of starch biosynthesis and consumption is adjusted in response to changing daylength in a manner that optimizes the allocation of carbon for growth across a wide range of carbon availability. As daylength decreases, more photosynthate is allocated to starch in the day, and its rate of consumption at night decreases so that supplies always last until dawn (Chatterton and Silvius, 1981; Gibon et al., 2009; Sulpice et al., 2014). Adjustments are also made in response to abrupt, unexpected alterations in daylength (Lu et al., 2005). When plants grown in 12-h-light/12-h-dark cycles are subjected to darkness after only 8 h of light, starch degradation is slower than in a normal night so that reserves again last almost precisely until dawn (Graf et al., 2010). Similarly, when the onset of darkness is delayed by 4 h, starch degradation is faster so that reserves are again exhausted almost precisely at dawn (Scialdone et al., 2013). Variation in the amount of starch present at the end of the day also leads to adjustment of starch degradation rates to permit exhaustion of reserves almost precisely at dawn (Graf et al., 2010; Scialdone et al., 2013).

We previously determined through experimental and chemical kinetic modeling approaches that the adjustment of starch degradation to match the length of the night is dependent on both the circadian clock and leaf starch content (Graf et al., 2010; Scialdone et al., 2013). It is not known exactly how leaf starch content is perceived such that it influences degradation rate. However, several potentially important features of leaf starch granules are known to be under tight control. These include the timing of granule initiation during leaf development, granule size, shape, and number per chloroplast (and therefore surface area), and the degree of crystallinity of the constituent glucan polymers (Roldán et al., 2007; Szydlowski et al., 2009; Crumpton-Taylor et al., 2012, 2013; Pfister et al., 2014). These features may combine with more specific control mechanisms that modulate the activities of the enzymes of starch breakdown to confer the observed adjustment of the degradation rate to match the length of the night.

To discover new components that influence starch degradation and mediate adjustment of the rate according to the length of the night, we devised a forward genetic screen to identify mutants defective in this adjustment. The screen made use of the observation that if starch reserves are exhausted prior to dawn, plants rapidly exhibit large transcriptional changes indicative of starvation. This phenomenon is seen in wild-type plants when subjected to an extended night, beyond the point at which starch is exhausted (Gibon et al., 2006), and in the short-period circadian clock mutant cca1 lhy (circadian clock associated1 late elongated hypocotyl), which exhausts its starch reserves ∼3 h before dawn when grown in 12-h-light/12-h-dark cycles (Graf et al., 2010). Mutant plants unable to synthesize or to degrade starch have a transcriptional signature of starvation throughout most of the night. We previously generated a starvation reporter line of Arabidopsis harboring a luciferase gene fused to the promoter of a gene that is expressed only after starch reserves are exhausted. Luciferase expression was very low during the normal day-night cycle but increased rapidly after a 2-h extension of the night (Graf et al., 2010). We reasoned that mutagenesis of the reporter line followed by screening for plants with abnormal temporal patterns of luciferase expression should identify mutants defective in the diel control of carbon availability, including aspects of starch granule biosynthesis and degradation, elements of the circadian clock necessary for correct anticipation of dawn, and components that sense or signal carbohydrate status. In this study, we validated this approach by showing that, as anticipated, the screen identified previously known classes of mutants defective in starch accumulation or with reduced starch crystallinity.

The screen also identified a mutant with inappropriately rapid starch degradation at night, such that starch was exhausted well before dawn. We named the mutant early starvation1 (esv1-1). We established that the causal mutation was in an unannotated gene, encoding a protein with no known or predicted role. The protein is highly conserved and tryptophan-rich and is present in the chloroplast stroma and within starch granules. Based on phenotypic characterization of mutant and overexpressing lines, we propose that the protein has an unrecognized and central function in the organization of glucan polymers within starch granules. We speculate that abnormal levels of the protein alter the starch granule matrix such that the mechanisms controlling the rate of starch degradation no longer function correctly. We show that another related protein is also likely to be involved in organization of glucan polymers within the granule matrix and can indirectly influence the control of degradation. These results are particularly surprising because it has been widely assumed that the formation of the starch granule matrix is a physical process involving self-organization of glucan polymers (Waigh et al., 1998, 2000). Our discoveries suggest that specific proteins are also required for this process.

RESULTS

The esv1 Mutant Has a Low Starch Content at the End of the Night

To identify plants defective in the control of carbon availability at night, we used a forward genetic screen based on an Arabidopsis starvation reporter line (Graf et al., 2010) in which luciferase expression is driven by the promoter of a sugar-repressed gene (At1g10070). Seeds of the reporter line were treated with EMS to induce point mutations. Then, 10-d-old seedlings of the M2 generation were screened for luciferase-induced bioluminescence at the end of the night, when nonmutagenized seedlings of the reporter line showed no bioluminescence. Isolated mutant plants were allowed to self-pollinate, and bioluminescence measurements were repeated on M3 plants to check for reproducibility.

The efficacy of this screen was demonstrated by the identification of two anticipated classes of mutants. Mutants that were shown by iodine staining to contain no starch at the end of the day or the night were not examined further. Such mutants are expected to exhibit symptoms of starvation during the night (Gibon et al., 2006). Mutants that stained reddish-brown rather than blue-black with iodine at the end of the day proved to be defective in ISOAMYLASE1 (ISA1) or ISA2 (Supplemental Figure 1). We showed previously that isa1 and isa2 mutants accumulate soluble phytoglycogen (which stains reddish-brown with iodine) in place of much of their starch during the day and completely degrade this glucan before the end of the night (Zeeman et al., 1998; Delatte et al., 2005).

One mutant line exhibited strong bioluminescence at the end of the night but was shown by iodine staining to have near-normal levels of starch at the end of the day (Figure 1). To test whether the bioluminescence was a consequence of premature exhaustion of starch during the night, rosettes were stained with iodine 2 h before the end of the night. At this point, wild-type rosettes stained a pale color, indicating that some starch remained. Mutant rosettes did not stain, indicating that much less starch remained (Figure 1B). The mutant was named esv1-1.

Figure 1.

Identification of the ESV1 Locus and the Positions and Effects of esv1 Mutations on ESV1 Transcript and Protein.

(A) Bioluminescence of control seedlings (top, unmutagenized reporter line) and the esv1 mutant (bottom) at the end of a 12-h night.

(B) Iodine staining of rosettes 2 h before the end of a 12-h night. Left, control plant (Col-0). Middle, the esv1-1 mutant derived from the forward genetic screen. Right, esv1-2, a T-DNA insertion mutant for At1g42430.

(C) Structure of ESV1 and LESV1 with untranslated regions (light-gray boxes), introns (thick black line), exons (dark-gray boxes), and positions of the EMS-induced mutation and the T-DNA insertions (triangles).

(D) RT-PCR using primers to amplify the full-length cDNA of ESV1 from leaf cDNA preparations. The predicted wild-type cDNA is 1281 bp. M, markers of 1000 and 2000 bp.

(E) Immunoblot of leaf extracts with an antiserum raised against ESV1 (top) or LESV1 (bottom). Each lane contained the same amount of total protein (20 µg). Numbers at the left indicate the positions of molecular mass markers (values in kilodaltons). The predicted masses of ESV1 and LESV proteins are 49 and 66 kD, respectively.

A Splice Site Mutation in At1g42430 Causes the esv1 Phenotype

The mutation underlying the esv1-1 phenotype was identified by a combination of map-based cloning and whole-genome sequencing. The esv1-1 mutant (Col-0 background) was crossed to a Landsberg erecta (Ler) wild-type plant. F1 plants were allowed to self-pollinate. Seedlings of the F1 generation did not show bioluminescence at the end of the night, but the F2 generation segregated for bioluminescence (Supplemental Figure 2A), indicating that the mutation in esv1-1 is recessive. Mapping narrowed the position of the esv1 mutation to a 2.7-Mb region spanning the centromere of chromosome 1 (Supplemental Figures 2B and 2C and Supplemental Table 1).

We used whole-genome sequencing to identify the ESV1 locus. DNA from 500 pooled esv1-1 plants in the F2 generation of the mutant × Ler cross was used as a template. In total, 9000 putative EMS-induced point mutations were identified over the whole genome, two of which were in genes within the target region. One caused an amino acid substitution (glu910lys) in At1g42470, encoding a putative hedgehog receptor located in the plasma membrane. The second was a G-to-A substitution that disrupted the acceptor site at the second intron of At1g42430 (Figure 1C), a gene of unknown function. Amplification of full-length cDNA from At1g42430 showed that the mutation results in synthesis of a shorter transcript in esv1-1 than in wild-type plants (Figure 1D). Thus, disruption of the intron splice site may result in splicing at an alternative site and consequent deletions in the transcript. An antiserum was raised to the recombinant protein product of the wild-type At1g42430 cDNA expressed in Escherichia coli. It recognized a protein of the expected mass on immunoblots of extracts of the wild type but not mutant leaves (Figure 1E). The esv1-1 mutant was backcrossed three times to the starvation reporter line before further characterization to minimize the number of unlinked EMS-induced mutations.

To provide independent evidence that the mutation in At1g42430 was responsible for the early starvation phenotype, we obtained a T-DNA insertion mutant from the GABI-Kat (www.gabi-kat.de) collection (Figure 1C). Genotyping confirmed the presence of the T-DNA insert in the fourth exon and immunoblot analysis confirmed that this insertion mutant (named esv1-2) lacked the protein product (Figure 1E). Rosettes of esv1-2 showed no staining with iodine 2 h before the end of the night (Figure 1B), which is consistent with the phenotype brought about by the original esv1-1 allele. To evaluate starvation responses in the esv1-2 mutant, we measured transcript levels of two sugar-repressed genes, At3g59940 and At1g76410. These genes encode an F-box protein KMD4 and a C3HC4 type zinc-finger protein ATL8, respectively, and were used to monitor starvation responses by Graf et al. (2010).

Like At1g10070, these genes are expressed at very low levels in wild-type plants at the end of both the day and the night. Their expression rises only when sugar levels are abnormally low, for example, during an unexpected extension of the night in wild-type plants and during the normal night in starchless mutants (Bläsing et al., 2005; Osuna et al., 2007; Usadel et al., 2008; Graf et al., 2010). In esv1-2 mutants, levels of transcript of the two genes at the end of the day were similar to those in wild-type plants, but levels at the end of the night were considerably elevated (Supplemental Figure 3).

ESV1 Is Similar to a Second Predicted Arabidopsis Protein: Both Are Conserved Throughout the Plant Kingdom

The predicted ESV1 protein is composed of 426 amino acids, with a predicted mass of ∼49 kD. It contains no previously annotated domains but has a proline-rich region at the C-terminal end (11 proline residues between amino acids 397 and 425) and is enriched in tryptophan and other aromatic amino acid residues in the C-terminal two-thirds of the protein (∼11% of the amino acid residues between amino acids 130 and 380 are tryptophans). Within this region are motifs in which single or paired aromatic amino acids are separated by two or three other amino acids, usually including one or more acidic residues (E or D), for example, WWETW, WTDKW, WEETWW, WYEKWWEKY, and WWEKWGEHY (Figure 2; Supplemental Figure 4).

Figure 2.

Structure and Location of ESV1 and LESV Proteins.

(A) Comparison of domain structures of the ESV1 and LESV proteins.

(B) Transient expression of ESV1-YFP and LESV-YFP in leaves of wild-type and pgm mutant N. sylvestris. For each genotype × construct, top, middle, and bottom panels show the YFP fluorescence, the merged chlorophyll and YFP fluorescence, and the chlorophyll fluorescence on a bright-field image for the same leaf area. All panels are at the same magnification. Bar (bottom right) = 5 μm.

(C) Immunoblots with antisera against ESV1 (upper panel) or LESV (middle panel) of extracts of wild-type, esv1, and lesv Arabidopsis leaves. Crude extracts were separated into soluble (S) and insoluble (I; containing starch) fractions by centrifugation. Each lane contains material from the same fresh weight of leaf (1.35 mg). The lower panel shows, as a loading control, immunoblots of identical gels with an antiserum against plant actin.

(D) Gel (left) and immunoblots with of starch granule-bound proteins from wild-type (Col-0), lesv, and esv1 mutants. Each lane on the left panel contained proteins extracted from 1.3 mg purified starch. Each lane on the middle and right panels contained proteins extracted from 0.3 mg purified starch. Numbers indicate the positions of molecular mass markers (values in kilodaltons). The ESV1 antiserum cross-reacts slightly with LESV (middle panel). On the gel, red asterisks are to the right of LESV bands (upper asterisks) and ESV bands (lower asterisks).

A BLAST search (http://www.arabidopsis.org/Blast/index.jsp) showed that an unannotated protein encoded by At3g55760 has the highest sequence similarity to ESV1: The predicted amino acid sequences are 38% identical. We refer to this second protein as LIKE ESV1 (LESV). It is composed of 578 amino acids with a predicted mass of 66 kD. LESV1 lacks the proline-rich C-terminal region of ESV1 but shares the tryptophan-rich region with aromatic motifs similar to those in ESV1. Within this region, the two proteins have similar numbers of tryptophan residues and the highest level of overall identity (Figure 2A; Supplemental Figure 4A). Unlike ESV1, LESV is represented on commonly used Arabidopsis microarrays. Examination of publicly available microarray data revealed that LESV is expressed in all organs of the plant. In leaves, transcript levels are high at the end of the night and low during much of the day (Supplemental Figure 5). LESV is strongly coexpressed with several genes encoding enzymes of starch metabolism: The top 20 coexpressed genes from Atted-II (www.atted.jp; Obayashi et al., 2009) encode enzymes including ISA3, α-GLUCAN PHOSPHORYLASE2, GLUCAN, WATER DIKINASE (GWD; also known as STARCH EXCESS1 [SEX1]), SEX4, DISPROPORTIONATING ENZYME1 (DPE1), DPE2, STARCH BRANCHING ENZYME3, α-AMYLASE3 (AMY3), and ADP-GLUCOSE PYROPHOSPHORYLASE LARGE SUBUNIT4 (Supplemental Table 2; see Smith [2012] for a discussion of the roles of these enzymes).

Genes homologous to ESV1 and LESV are found in land plants and green algae, including microalgae in the genus Ostreococcus, which have the smallest known genomes among eukaryotes (Supplemental Figure 4B, Supplemental Table 3, and Supplemental Data Set 1). The genes are apparently absent from prokaryotes, red algae, and other eukaryotic life forms. ESV1 was included in a phylogenomic inventory of proteins specific to the plant lineage (GreenCut2; Karpowicz et al., 2011). The high level of evolutionary conservation of ESV1 and LESV protein sequences suggests that both proteins have important functions that are specific to the Viridiplantae.

ESV1 and LESV Are Chloroplastic and Bind to Starch Granules

Chloroplastic locations for ESV1 and LESV were suggested by previous publications on the chloroplast proteome (e.g., Bayer et al. [2011] for ESV1; Kleffmann et al. [2004] and Peltier et al. [2006] for LESV). The ChloroP program predicts a 56-amino acid N-terminal chloroplast transit peptide (cTP) for LESV but no cTP for ESV1. However, the predicted proteins most similar to ESV1 from Glycine max, Prunus persica, Vitis vinifera, cassava (Manihot esculenta), and Populus trichocarpa have putative transit peptides. We sought to confirm the plastidial localization of the Arabidopsis proteins by transiently expressing them in leaves of woodland tobacco (Nicotiana sylvestris) as C-terminal fusions to YFP. For both proteins, YFP fluorescence was exclusively located in chloroplasts and was associated with discrete bodies likely to be starch granules (Figure 2B). Stable expression of these constructs in Arabidopsis gave the same result (Supplemental Figure 6A). To provide more information about the location of the proteins within chloroplasts, we also expressed the fusion proteins transiently in the pgm mutant of N. sylvestris, which lacks chloroplastic phosphoglucomutase and therefore cannot synthesize starch (Hanson and McHale, 1988). In the pgm mutant, YFP appeared as a diffuse signal in the chloroplast stroma for both fusion proteins rather than being associated with discrete structures. This suggests that the fluorescent structures in chloroplasts of wild-type N. sylvestris are indeed starch granules.

ESV1 and LESV were both partitioned between the soluble and insoluble (starch-containing) fraction of leaves. Immunoblotting of extracts of wild-type N. sylvestris leaves transiently expressing ESV1 or LESV as a YFP fusion revealed that the fusion proteins were more abundant in the insoluble fraction (the expected location for starch-bound proteins) than in the soluble fraction. Both proteins were mostly soluble when transiently expressed in pgm leaves, which lack starch (Supplemental Figure 6B). In wild-type Arabidopsis leaves, ESV1 and LESV proteins were also present in both insoluble and soluble fractions (Figure 2C). The proteins were largely soluble at the end of the night when starch levels were low, and they were largely insoluble at the end of the day when starch levels were maximal. ESV1 and LESV were present in proteins extracted from purified starch from wild-type Arabidopsis leaves and were among the most abundant granule-bound proteins in these leaves (Figure 2D). As the ESV1 and LESV genes are conserved throughout the plant kingdom, we examined whether the proteins are present in starches from species of economic significance. Both proteins were present in starch purified from roots of cassava, tubers of potato (Solanum tuberosum), and mature grains of rice (Oryza sativa) and maize (Zea mays; Supplemental Table 4).

esv1 Mutants Cannot Adjust the Rate of Starch Degradation to the Length of the Night

To characterize the defect in starch degradation in the esv1 mutants, we analyzed starch contents in rosettes during a normal 12-h night and in rosettes subjected to darkness after only 8 h of light (an early night) (Figure 3). Under both conditions, as expected, wild-type plants degraded starch at a near linear rate and did not exhaust their reserves before the end of the night. The rates of starch degradation were faster in esv1 mutant plants than in wild-type plants, such that starch reserves were completely exhausted 2 h before the end of the night in normal nights and 4 h before the end of the night in early nights (Figures 3A to 3C). Extrapolation of the near-linear phase of starch degradation for esv1 plants, over the first 8 h of the night, revealed that these rates (estimated from linear regression analysis) would exhaust starch reserves after ∼9 h in both normal (12 h) nights and early nights. Thus, wild-type plants adjusted the rate of starch degradation to the length of the night, but esv1 mutant plants failed to make this adjustment correctly.

Figure 3.

Starch Turnover in Leaves of esv1 and lesv Mutants and Plants Overexpressing ESV1 or LESV.

(A) Starch contents over the day-night cycle in wild-type rosettes (black) and rosettes of esv1-2 (blue), lesv-1 (red), and esv1-2 lesv-1 (orange). Plants were 3 weeks old. Values are means ± se of measurements made on six rosettes.

(B) Starch contents during a normal, 12-h night in wild-type rosettes (black) and rosettes of esv1-1 (blue, closed symbols) and esv1-2 (blue, open symbols) mutants. Plants were 3 weeks old and from a different batch from those in (A). Values are means ± se of measurements on five rosettes.

(C) Starch contents during a 16-h night that commenced 8 h after dawn, in wild-type rosettes (black) and rosettes of esv1-1 (blue, closed symbols) and esv1-2 (blue, open symbols) mutants. Plants were 3 weeks old. Values are means ± se of measurements on five rosettes.

(D) Sucrose levels over the day-night cycle in wild-type rosettes (black) and rosettes of esv1-2 (blue), lesv-1 (red), and esv1-2 lesv-1 (orange). Plants were the same age and from the same batch as those in (A) and (E). Values are means ± se of measurements on six rosettes.

(E) Maltose levels over the day-night cycle in wild-type rosettes (black) and rosettes of esv1-2 (blue), lesv-1 (red), and esv1-2 lesv-1 (orange). Plants were the same age and from the same batch as those in (A) and (D). Values are means ± se of measurements made on six rosettes.

(F) Starch contents over the day-night cycle in wild-type rosettes (black) and rosettes of lesv-1 (red, closed symbols) and lesv-2 (red, open symbols) mutants. Plants were 3 weeks old and from a different batch from those in (A). Values are means ± se of measurements on five rosettes.

(G) Starch contents at the end of the day (white), 10 h into the night (gray), and the end of the night (black) for wild-type plants, esv1 and lesv mutants, and transgenic lines overexpressing either ESV1-YFP or LESV-YFP in a wild-type or an esv1 background as indicated. The inset is a magnification of values for lesv mutants and LESV-overexpressing lines (last three column sets of the main graph). Plants were 3 weeks old. Values are means ± se of measurements made on six rosettes.

To provide more information about the alterations in the control of starch degradation in esv1 plants, we subjected batches of esv1 and wild-type plants to different light levels over a single day to generate plants with very different starch levels at the start of the 12-h night. As expected, the rate of starch degradation in wild-type plants was adjusted such that the starch reserves lasted for 12 h over a wide range of end-of-day starch contents. In esv1 plants, the time at which starch reserves were exhausted depended upon the initial starch content. Reserves were exhausted after 9 h of darkness in plants with 7.5 mg starch g⁻¹ fresh weight at the end of the day, and after 6 h of darkness in plants with 2 mg starch g⁻¹ fresh weight at the end of the day (Supplemental Figures 7A and 7B). Whereas in wild-type plants, the relative rate of starch degradation (fraction of end-of-day starch degraded per unit time) was independent of starch content, in esv1 plants, the relative rate decreased as starch content decreased (Supplemental Figure 7C). Thus, in esv1, the dependence of starch degradation on time until dawn has decreased or been lost.

In most experiments, esv1 mutants had lower rates of net starch accumulation than wild-type plants during the day and, hence, lower starch contents at the onset of darkness (Figure 3). Interestingly, the levels of both sucrose and maltose in esv1-2 plants were higher than in wild-type plants during the day but lower than in wild-type plants toward the end of the night (Figures 3D and 3E). Elevation of sucrose levels during the day is commonly observed in Arabidopsis mutants with defects in pathways of starch biosynthesis or degradation (Chia et al., 2004; Comparot-Moss et al., 2010; Mugford et al., 2014). Elevated levels of maltose are generally indicative of starch degradation (Weise et al., 2004). Taken together, these results indicate that esv1 mutants usually accumulate less starch than wild-type plants because they degrade starch during the day as well as during the night. The low levels of sucrose and maltose toward the end of night in esv1 are consistent with the observed premature starch depletion in these plants.

We investigated whether loss of LESV also affects starch turnover by examining two T-DNA insertion mutants (Figure 1C). The absence of LESV protein from these mutants was confirmed by immunoblotting with a specific antiserum (Figure 1E). Diel patterns of starch accumulation and loss in the lesv mutants were generally similar to those of wild-type plants (Figured 3A, 3F, and 3G). Sucrose and maltose levels in the lesv-1 mutant were also similar to those of wild-type plants over the day-night cycle (Figures 3D and 3E). Starch turnover and sucrose and maltose levels throughout the day-night cycle in an esv1-2 lesv-1 double mutant resembled those of the esv1 parent (Figures 3A, 3D, and 3E).

Lack of ESV1 Affects Starch Contents in Nonphotosynthetic Tissues

We investigated whether ESV1 is important for the control of starch content in nonphotosynthetic parts of the plant (Figure 4). Detailed examination of iodine-stained leaves showed that cells immediately adjacent to veins accumulate less starch during the day in esv1 mutants than in wild-type plants. This netted pattern of starch accumulation in esv1 leaves was particularly pronounced after an extended light period of 24 h (Figure 4A). Staining for starch also revealed lower starch contents in esv1 than in wild-type plants in columella cells of root caps, stems, flowers, and siliques (Figures 4B to 4D).

Figure 4.

Starch Content and Morphology of esv1 and lesv Mutants.

(A) Leaves of the wild-type, esv1-2, and lesv-1 mutants after 24 h light, following decolorization and iodine staining for starch.

(B) Root tips of the wild-type, esv1, and esv1 sex1 seedlings (4 d old), following decolorization and iodine staining for starch.

(C) Young inflorescences of wild-type and esv1 plants, following decolorization and iodine staining for starch.

(D) Longitudinal sections of flower stems of 5-week-old wild-type (Col), esv1, and lesv mutants. Sections from the upper stem of the primary bolt were stained with Schiff’s reagent and propidium iodide and viewed with a confocal microscope. Starch granules appear as white particles in cells (arrowed in left panel). Positions of tissues are noted below the photographs: ph, phloem; en, endodermis; co, cortex; ep, epidermis. Bar = 20 μm.

(E) Appearance of rosettes of wild-type (Col) and esv1-2 plants at 3 weeks old. The photographs are at the same scale.

(F) Appearance of wild-type (Col) and esv1 plants during flowering. The two plants are the same age. Bar = 1 cm.

Plants lacking ESV1 had several developmental and morphological phenotypes that may be attributable to defective starch metabolism. After 24 d of growth, rosettes of esv1 plants were smaller than those of wild-type plants (Figure 4E), and their fresh weights were ∼40% lower. Mutant plants also flowered later than wild-type plants. Other mutants with defects in diel starch turnover in leaves are slow-growing in 12-h-light/12-h-dark cycles, e.g., the essentially starchless mutants pgm and adg1 (lacking the small subunit of ADP-glucose pyrophosphorylase) and the starch degradation mutants sex1 (lacking glucan, water dikinase) as well as bam1 bam3 (lacking β-amylases 1 and 3) (Caspar et al., 1985; Lin et al., 1988; Yu et al., 2001; Fulton et al., 2008), and some exhibit delayed flowering, e.g., sex1 and pgi (lacking plastidial phosphoglucomutase) (Eimert et al., 1995; Corbesier et al., 1998; Yu et al., 2000). The axillary shoots of esv1 flower stems grew with a wider angle from the main stem than those of wild-type plants (Figure 4F). Wide stem angles are characteristic of mutants defective in the stem gravitropic response, which is dependent on the presence of starch granules in endodermal cells (Weise and Kiss, 1999; Fujihira et al., 2000; Tanimoto et al., 2008). Whereas the endodermis of wild-type and lesv stems contained large starch granules, no starch was visible in this tissue in esv1 stems (Figure 4D).

Overexpression of ESV1 and LESV Proteins Alters Leaf Starch Content

To provide more insight into the importance of ESV1 and LESV proteins in controlling starch content in leaves, we generated plants with elevated amounts of these proteins by constitutively expressing them as YFP fusions (Supplemental Figure 8A). Expression of ESV1-YFP in either wild-type or esv1 mutant plants resulted in elevated starch contents. In some lines, starch content was elevated at the end of the night but not at the end of the day, whereas in the line with the highest level of expression of ESV1-YFP (line 3-2), starch content was strongly elevated throughout the day-night cycle (Figure 3G).

We also generated plants in which ESV1 protein without a tag was expressed in an esv1 background. These plants had smaller increases in ESV1 protein than those expressing ESV1-YFP (Supplemental Figure 8B). Nonetheless, as for plants expressing ESV1-YFP, starch contents of leaves were elevated relative to those of wild-type plants at both the end of the day and the end of the night (Supplemental Figure 8C), showing that this phenotype was not an artifact arising from the use of a YFP fusion protein.

Given this striking effect of ESV1 overexpression, we examined whether LESV overexpression affected starch content. Constitutive expression of LESV-YFP in a wild-type background had little impact on starch content at the end of the day, but starch content 10 h into the night and at the end of the night was about 3-fold lower than in wild-type plants (Figure 3G; Supplemental Figure 8A). These data suggested that LESV-overexpressing plants might be starving at the end of the night. Consistent with this expectation, the transcript abundance of two starvation genes was higher in LESV-overexpressing plants than in wild-type plants at the end of the night and was similar to that in esv1 plants at this point (Supplemental Figure 3).

Altered Levels of ESV1 and LESV Affect Starch Granule Morphology, Number, and Composition but Have Only Minor Effects on Amylopectin Structure

Starch granules from leaves of esv1 and lesv mutants and the double mutant esv1 lesv were roughly discoid in shape, like those of wild-type leaves. However, granules of the mutants were less regular in shape than wild-type granules: They consistently had lobed or wavy outlines (Figure 5). ESV1-overexpressing plants had thicker, larger granules than wild-type plants. LESV-overexpressing plants had many more granules per chloroplast than wild-type plants. These granules were highly variable in size and shape, and many were much smaller than granules of wild-type plants.

Figure 5.

Starch Granule Morphology in esv1 and lesv Mutants.

For starch extraction, rosettes were 4 weeks old and were harvested at the end of the day. For microscopy, young leaves were harvested from 3-week-old rosettes in the second half of the light period.

(A) Scanning electron micrographs of purified starch granules from leaves of wild-type (Col), esv1, lesv, and esv1 lesv double mutants. Bars = 2 μm.

(B) Scanning electron micrographs of purified starch granules from leaves of wild-type (Col) and transgenic lines overexpressing ESV1 or LESV (OX lines). Bars = 2 μm apart from the image of LESV OX starch, where the bar = 1 μm.

(C) Light micrographs of leaf sections stained with toluidine blue. Bars = 10 μm.

(D) Transmission electron micrographs of leaf sections. Bars = 5 μm.

The starch of both esv1 and lesv mutants consistently contained ∼60% more amylose than that of wild-type plants (Figure 6). It seemed likely that this change in starch composition was a secondary rather than the primary consequence of the mutations, since the esv1 and lesv mutations have the same effect on amylose content but radically different effects on the rate of starch degradation. To investigate this further, we examined whether loss of amylose affects starch degradation in esv1 mutants by generating a double mutant lacking both ESV1 and granule-bound starch synthase (GBSS), the enzyme responsible for amylose biosynthesis (Seung et al., 2015). As expected, the esv1 gbss double mutant lacked amylose. Its starch content was lower than that of wild-type or gbss plants and comparable to that of esv1 at both the end of the day and 2 h before the end of the night (Supplemental Figure 9A). Thus, loss of ESV1 results in faster starch degradation regardless of whether amylose is present in the starch.

Figure 6.

Effect of ESV1 and LESV on Starch Composition and Structure.

Starch was extracted at the end of the day from rosettes of 4-week-old plants of wild-type (Col), esv1, and lesv mutants and transgenic lines overexpressing ESV1 or LESV (line numbers indicated). ESV1 overexpression is either in a wild-type (lines 3-2 and 4-4) or a esv1 (line 4-6) background; lesv overexpression is in a wild-type background.

(A) Amylose content of starch.

(B) Chain length distribution. Starch was debranched and glucan chains were separated by HPAEC. Peak areas from degree of polymerization 3 to 50 were summed and the relative peak area for each chain length was calculated. Results for each chain length are expressed as the relative percentage of difference from the wild-type (Col) value. Values are means ± se of measurements made on three or four replicates, each prepared from an individual rosette. Top, esv1 (black symbols) and lesv1 (open symbols). Middle, 35S_pro:ESV1-YFP line 4-4 (black symbols) and esv1/35S_pro:ESV1-YFP line 2-5 (open symbols). Bottom, 35S_pro:LESV-YFP line 1-1 (black symbols) and 35S_pro:LESV-YFP line 4-6 (open symbols).

Overexpression of ESV1 and LESV had very different effects on starch composition. In ESV1-overexpressing plants, amylose content was similar to that of esv1 mutants, ∼60% higher than in wild-type plants. Overexpression of LESV drastically reduced amylose content (Figure 6A; Supplemental Figure 9B). The variation in amylose content caused by changes in amounts of ESV1 or LESV was not attributable to variation in the amount of GBSS in the starch. There was no obvious relationship between GBSS protein levels and amylose levels in plants with different levels of ESV1 or LESV proteins (Supplemental Figure 9C). Loss or overexpression of ESV1 caused only minor alterations in the levels of starch-bound phosphate. Changes in LESV content had greater effects: loss of LESV reduced starch phosphate content by ∼25%, whereas overexpression increased it by ∼75% (Supplemental Figure 9D).

We investigated whether altered levels of ESV1 or LESV proteins affect the structure of the major starch polymer amylopectin. The chain length distribution of amylopectin differed slightly between esv1, lesv, ESV1, and LESV-overexpressing lines and wild-type starch (Figure 6B), but these differences were small in comparison to those seen in Arabidopsis mutants deficient in enzymes of starch biosynthesis (Pfister et al., 2014). The chain length distributions of the β-limit dextrin of amylopectin from esv1 and lesv mutants were also similar to those of wild-type plants (Supplemental Figure 9F). These results indicate that neither chain elongation nor branching/debranching during amylopectin biosynthesis is strongly affected by the loss of ESV proteins.

Loss of ESV1 Modifies but Does Not Abolish the Effects of Mutations That Reduce Starch Degradation

We considered the possibility that ESV1 interacts with a specific enzyme of starch degradation and inhibits its activity. To test this idea, we created a series of double mutant plants that lacked both ESV1 and a specific enzyme shown previously to be involved in starch degradation in the chloroplast at night (Figure 7). Interpretation of the phenotypes of mutants lacking enzymes of starch degradation is complicated by redundancy within the pathway: For most enzymes, loss does not prevent starch degradation but rather reduces the proportion of end-of-day starch that is consumed during the night (Scialdone et al., 2013). Nonetheless, if ESV1 inhibits one specific enzyme of starch degradation, the loss of both this target enzyme and ESV1 should give a starch turnover phenotype identical to that given by the loss of the target enzyme alone. The loss of both ESV and any enzyme other than its direct target might be expected to give a phenotype different from that of either parent mutant.

Figure 7.

Characteristics of Starch in esv1 Double Mutants.

(A) Starch contents at the end of the day (white bars) and the end of the night (black bars) in 4-week-old rosettes of wild-type (Col), esv1-1, sex1, and esv1-1 sex1 double mutants. Values for Col, esv1, and sex1 are means ± se from measurements on 6 to 12 individual rosettes. Values for esv1-1 sex1 are means of values from three independently selected double mutant lines. Values for each line are provided in Supplemental Figure 8.

(B) Rosettes of sex1 and esv1-1 sex1 mutants at the end of the day, following decolorization and iodine staining for starch. Red arrows indicate regions of low starch content in the meristem, petioles, and midribs of the esv1-1 sex1 mutant.

(C) Starch contents at the end of the day (white bars), after 10 h of night (gray bars), and at the end of the night (black bars) of wild-type (Col) and esv1-2 rosettes, rosettes of five starch-excess mutants (bam4, bam3, lsf1, sex4, and bam1), and rosettes of double mutants of esv1-2 and starch-excess mutants. Values are means ± se from measurements on five individual rosettes.

(D) As for (B), showing starch contents for wild-type (Col), esv1-2, pwd, and esv1-2 pwd double mutants. Plants were from a different batch from those shown in (B).

(E) As for (B), showing starch contents for wild-type (Col) esv1-2, amy3, and esv1 amy3 rosettes. Values are means ± se from measurements on five individual rosettes. Plants were from a different batch from those shown in (B).

We first investigated the effect of losing ESV1 on the phenotype of the sex1 mutant, which lacks GWD. This enzyme catalyzes the first step in degradation: The phosphorylation of the C6-position of glucose residues in amylopectin (Yu et al., 2001; Ritte et al., 2002). Starch degradation in sex1 mutants is severely impaired and starch accumulates to high levels. It is thought that phosphorylation at the granule surface by GWD disrupts the organization of the starch polymers and facilitates degradation by starch-hydrolyzing enzymes (Blennow and Engelsen, 2010; Hejazi et al., 2008, 2009, 2010). The esv1 sex1 mutants had leaf starch contents that were different from those of either parent. Starch contents were significantly lower than those of the sex1 mutant but much higher than those of esv1 and wild-type plants (Figure 7A; Supplemental Figure 9F). However, the starch phenotype of the double mutant was very different in other parts of the plant. Whereas sex1 mutants accumulate very high levels of starch throughout the plant, including root caps (Caspar et al., 1991; Yu et al., 2001), the esv1 sex1 mutant had very low levels of starch in root cap cells, similar to the esv1 parent (Figure 4B). The esv1 sex1 mutant also had little starch in the major veins and petioles of mature leaves and around the meristem. These regions had very high starch contents in the sex1 parent (Figure 7B).

As for the esv1 sex1 mutant, leaves of other double mutants lacking both ESV1 and a protein necessary for starch degradation had starch turnover phenotypes different from those of their parents (Figures 7C and 7D). Loss of ESV1 reduced the severity of the starch-excess phenotype in all mutant backgrounds examined, but double mutants retained more starch in the rosette at the end of the night than esv1 and wild-type plants. The starch contents of leaves of the esv1 sex4 and esv1 pwd double mutants were reduced relative to that of sex4 (lacking a glucan phosphate phosphatase; Kötting et al., 2009) and pwd (lacking phosphoglucan water dikinase; Baunsgaard et al., 2005; Kötting et al., 2005) by about one-third or less and were always greatly in excess of the starch content of wild-type plants. Mutants lacking ESV1 and either the major chloroplastic β-amylase BAM3, the β-amylase-like protein BAM4 (Fulton et al., 2008), or the glucan phosphate phosphatase-like protein LSF1 (Comparot-Moss et al., 2010) had starch contents that were reduced by 50% or more relative to their starch excess parent at the end of the day and the end of the night. Because loss of ESV1 reduced the severity of the starch excess phenotype in all of the starch degradation mutants, we suggest that its role in starch degradation is not primarily as a negative regulator of any one of the five starch degradation proteins affected in the mutants. However, these results must be interpreted with caution because it remains possible that, as for sex1, the impact of loss of ESV1 on the starch-excess phenotypes is different in photosynthetic and nonphotosynthetic cells of the rosette.

We also investigated whether the high rate of starch degradation in the esv1 mutant results from activation of a degradative enzyme that is redundant with respect to starch degradation in wild-type plants. Accordingly, we crossed esV1-1 with mutants lacking either the β-amylase BAM1 or the α-amylase AMY3. Loss of either of these chloroplastic proteins has no effect on starch turnover in a wild-type background (Yu et al., 2005; Fulton et al., 2008). We found that the esv1 bam1 and esv1 amy3 double mutants had the same very low starch contents at the end of the night as the esv1 parent (Figures 7C and 7E); thus, the high rate of starch degradation in esv1 does not require the activity of either BAM1 or AMY3.

DISCUSSION

This work describes two proteins with central and previously unsuspected roles in plant carbohydrate metabolism. We show that the presence and/or levels of the ESV1 and LESV proteins are crucial for normal patterns of starch biosynthesis and degradation in the Arabidopsis plant and, hence, for normal growth and productivity. Our results lead us to propose that both proteins are involved in determining the conformation of the starch granule matrix and that the perturbations of starch turnover in plants with altered levels of the proteins are indirect consequences of abnormal starch granule structures. Below we present the evidence for this proposal and discuss the wider significance of our findings for the understanding of carbon storage and allocation in plants.

ESV1 and LESV Are Important for Normal Starch Turnover in Leaves

Our starvation reporter screen identified ESV1 as a protein required for normal rates of starch degradation in Arabidopsis leaves at night. Whereas in wild-type plants, starch reserves are used at an essentially linear rate at night such that they are exhausted almost exactly at dawn, in the absence of ESV1, starch reserves are consumed in a nonlinear manner and are exhausted prior to dawn. Accelerated starch degradation at night was previously reported in the double mutant cca1 lhy, which lacks two MYB transcription factors that are central components of the circadian clock and thus anticipates dawn several hours too early (Graf et al., 2010). It seems highly unlikely that the rapid starch degradation in esv1 mutants is brought about by a defective circadian clock. The mutant does not exhibit short-period phenotypes such as early flowering, and the ESV1 protein is located in the chloroplast whereas known clock components are nuclear and cytosolic.

The esv1 phenotype is similar in several respects to that of isa mutants. The isa1 and isa2 mutants also exhaust their glucan reserves prior to dawn and appear to degrade some storage glucan during the day as well as at night (Delatte et al., 2005). The reserve glucans in these mutants are largely soluble rather than in granular form (Zeeman et al., 1998; Delatte et al., 2005). Loss of normal control of degradation is expected in this case. Our chemical kinetic models capable of explaining the dependency of starch degradation on starch content and time until dawn require the interaction of controlling molecules with the solid surface of the starch granule, and they predict that control will be lost if glucans are soluble (Scialdone et al., 2013). esv1 mutants accumulate glucans as starch granules, so loss of control of glucan degradation at night is not due to a radical change in glucan solubility. However, the loss of control may be generally analogous to that in isa mutants in that it may stem from increased accessibility of glucans to starch-degrading enzymes. We discuss this possibility below.

LESV is less important for normal starch turnover than ESV1. The pattern of starch turnover in the lesv mutant is similar to that of wild-type plants. The involvement of LEST in processes underpinning normal starch metabolism is apparent from the significantly altered levels of amylose and of granule-bound phosphate in the lesv mutant and from the effects of LESV overexpression in Arabidopsis leaves. As with esv1 mutants, plants with elevated LESV consume starch too rapidly during the night so that reserves are exhausted before dawn. By contrast, overexpression of ESV1 results in very high starch levels throughout the day-night cycle. Overall, our results show that control of ESV1 and LESV protein levels is essential for normal starch turnover in Arabidopsis leaves.

The ESV1 and LESV Proteins Are Directly Associated with Starch Granules

Experiments employing fluorescently tagged proteins confirmed findings from previous proteomic studies that both ESV1 and LESV proteins are located inside the chloroplast. Together with cell fractionations, these experiments revealed that both proteins are at least in part associated with starch granules. Changes during the day-night cycle in the fraction of the proteins that is granule bound suggest that both are incorporated into granules as they grow during the day and are then released as they are degraded at night. Although LESV transcript levels vary severalfold over the day-night cycle, being low at the end of the day and high at the end of the night, the total amount of LESV protein is similar at these two time points. Thus, it appears that large daily changes in the transcript level of LESV are not reflected at the protein level, a feature that LESV shares with several other proteins of starch metabolism (Skeffington et al., 2014). Equivalent information is not available for ESV1 because it is not represented on commonly used Arabidopsis microarrays. Most of the proteins previously reported to be present in both starch granules and the stroma are enzymes of starch biosynthesis and degradation or proteins that interact directly with these enzymes. They include various isoforms of starch synthase and starch-branching enzyme (Denyer et al., 1993; Mu-Forster et al., 1996; Grimaud et al., 2008) and the enzymes responsible for phosphorylation and dephosphorylation of the starch granule surface (Yu et al., 2001; Kötting et al., 2009; Santelia et al., 2011). The dual location of ESV1 and LESV proteins in the stroma and the starch granule thus indicates a direct role in the biosynthesis, assembly, and/or degradation of starch granules.

Proteins that bind to starch usually possess either carbohydrate binding modules (CBMs), surface binding sites (SBSs), or both. Arabidopsis proteins with one or more of these domains include starch synthases, starch branching and debranching enzymes, amylases, enzymes that phosphorylate or dephosphorylate the starch granule surface, and the recently characterized PTST (PROTEIN TARGETING TO STARCH) protein that facilitates the binding to starch of granule-bound starch synthase (Palopoli et al., 2006; Glaring et al., 2011; Meekins et al., 2013; Seung et al., 2015). Starch binding CBMs fall into distinct classes, for example, GWD and AMY3 both possess two CBM45 domains in tandem, and isoamylases, branching enzymes, and PTST possess CBM48 domains. SBSs are generally highly variable and require experimental definition (Cockburn et al., 2013; Meekins et al., 2014). Neither ESV nor LESV possesses a recognizable CBM, and further work will be required to discover whether they possess SBSs. However, the observation that both proteins contain a large domain strongly enriched in conserved tryptophan residues, and with conserved phenylalanine and tyrosine resides, leads us to speculate that they may be capable of binding to specific glucan structures. Known starch binding CBMs are 90 to 130 amino acids long and typically have four conserved aromatic amino acid residues that align the protein with the nonpolar faces of glucose residues in starch chains (Christiansen et al., 2009). Although they are often found singly, CBMs also occur in tandem; for example STARCH SYNTHASE3 from Arabidopsis has three CBMs (Palopoli et al., 2006). The tryptophan-rich regions of ESV1 and LESV contain over 35 conserved aromatic amino acid residues arranged in motifs. It seems possible that the repeated motifs may mediate binding to numerous glucans or facilitate interaction with long glucans at numerous contact points.

ESV1 Probably Acts Upstream of Enzymes of Starch Degradation

To discover whether ESV1 might inhibit the action of a protein required for starch degradation, we examined the effect of loss of ESV1 on the phenotypes of starch-excess mutants lacking individual proteins involved in starch degradation. Loss of ESV1 reduced but did not abolish the starch excess phenotypes of the sex1, sex4, lsf1, bam3, and bam4 mutants. Loss of ESV1 from plants lacking AMY3 or BAM1, starch-degrading enzymes that are largely redundant in a wild-type background (Yu et al., 2005; Fulton et al., 2008), resulted in esv1-like phenotypes. The pathway of starch degradation is not amenable to straightforward genetic analysis because of multiple redundancies. Nonetheless, these results allow us to make three proposals about the role of ESV1 as follows.

First, ESV1 is unlikely to act through direct inhibition of SEX1, PWD, SEX4, LSF1, BAM3, or BAM4. Second, the accelerated rate of starch degradation in the esv1 mutant is largely catalyzed by the same enzymes and proteins responsible for starch degradation in wild-type plants. SEX1, SEX4, LSF1, BAM3, and BAM4 are all required for the accelerated rate of degradation in the esv1 mutant, and we found no evidence that this accelerated rate requires either of the normally “redundant” enzymes AMY3 and BAM1. Third, in mesophyll cells, ESV1 may act at or prior to the start of the pathway of starch degradation. Loss of ESV1 affected the starch excess phenotypes of mutants lacking both downstream and initial enzymes of the pathway. The observation that it had smaller effects on the starch excess phenotypes of sex1, pwd, and sex4 than on those of mutants lacking β-amylases could indicate that ESV1 function is linked more closely to the phosphorylation/dephosphorylation of the granule surface than to the actions of downstream hydrolytic enzymes. However, as discussed above, there is no evidence that ESV1 directly inhibits the enzymes of phosphorylation/dephosphorylation, and its loss and overexpression have only small effects on levels of starch-bound phosphate.

ESV1 May Play a Role in Determining the Organization of Glucans in the Granule Matrix

Given that ESV1 binds to starch granules and appears to act upstream of the pathway of starch degradation, it seems possible that it influences the accessibility of glucans within the matrix to proteins involved in starch degradation. Any such influence is unlikely to be due to a direct effect of ESV1 on starch polymer composition or structure. Loss or overexpression of either ESV1 or LESV affected the amylose content of starch and had minor effects on amylopectin chain length distribution. However, these changes are highly unlikely to account for the accelerated rate of starch degradation in the esv1 mutant. First, both esv1 and lesv mutants have elevated amylose contents, but only the esv1 mutant has accelerated starch degradation. Second, abolition of amylose from the esv1 mutant (in the esv1 gbss mutant) does not alter its starch degradation phenotype. Third, the alterations in chain length distribution of amylopectin are much smaller than in many mutants that do not exhibit accelerated starch degradation (e.g., starch synthase mutants; Szydlowski et al., 2009; Pfister et al., 2014).

Our data indicate that ESV1 and LESV directly influence the organization of the starch granule matrix rather than the biosynthesis of the starch polymers that comprise it. The esv1 and lesv mutants have irregular granule shapes, and overexpression of ESV1 and LESV dramatically alters granule size, morphology, and, in the case of LESV, abundance. ESV1 overexpressers have large, thick granules, whereas LESV overexpressers have increased numbers of granules of very variable size and shape. These effects are unlikely to be due to altered amylopectin structure in the overexpressing lines: The alterations in amylopectin structure are minor compared with those of other starch mutants that retain normal granule shapes and numbers.

Little is known about how starch polymers are organized to form the starch granule matrix. Waigh and colleagues (Waigh et al., 1998, 2000) showed that many physical properties of starch, including gelatinization, freezing/thawing, and hydration/dehydration, can be explained by modeling it as a chiral side-chain liquid crystalline polymer, a self-organizing structure. By extrapolation, and in the absence of a biological explanation, it has been assumed that starch chains synthesized at the granule surface in vivo assemble into double helices that self-organize through physical processes, forming the semicrystalline lamellar structure of the matrix. The discovery of the ESV1 and LESV proteins raises the possibility that correct assembly of the matrix involves proteins that bind to starch polymers, in addition to physical processes.

A direct role for ESV1 in the correct organization of the granule matrix could explain why its loss perturbs the rate of starch degradation. We suggest that the degree to which hydrolytic enzymes can access starch polymers is in part determined by ESV1. By helping to confer a high level of matrix organization, ESV1 could restrict access by hydrolytic enzymes. If, in the absence of ESV1, the matrix is less organized, it could be much more susceptible to degradation by hydrolytic enzymes than in wild-type plants.

In wild-type plants, control over the rate of starch degradation may be exerted via modulation of the level of starch phosphorylation. Phosphorylation of glucose residues at the granule surface by GWD and PWD is thought to interfere with the regular packing of double helices, thus rendering the starch polymers more accessible to hydrolytic enzymes (Blennow and Engelsen, 2010; Hejazi et al., 2008, 2009, 2010). We recently presented evidence that phosphorylation may be the point at which circadian clock control over starch degradation is exerted. In the absence of PWD, the rate of starch degradation could no longer be correctly adjusted to the length of the night (Scialdone et al., 2013). It seems plausible that an altered level of starch matrix organization in esv1 mutants may render the modulation of phosphorylating activities less effective as a means of controlling the rate of degradation.

Although there is no clear starch turnover phenotype in the lesv mutant, the altered amylose and starch phosphate contents of the mutant starch and the large effects of LESV overexpression on starch degradation and granule size, shape, and number indicate that LESV too may play a role in matrix organization and in modulating the effects of phosphorylation on the accessibility of starch polymers to hydrolytic enzymes. However, ESV1 and LESV appear to have opposite roles in these respects, with ESV1 promoting a high level of organization of the granule matrix and LESV potentially reducing the level of organization. These differences in function may stem from divergent features of the protein sequences. The two proteins share a tryptophan-rich domain, in which many of the aromatic amino acid residues are contained within distinct motifs. In ESV1 sequences, there are 39 conserved aromatic residues within this domain. LESV sequences contain 37 conserved aromatic residues in the same region, 30 of which are in the same amino acid position as in ESV1 (Supplemental Figure 4A). However, motif sequences differ between the two proteins. Comparisons of ESV1 and LESV proteins from a wide range of starch-synthesizing organisms show that many motifs in the domain are conserved within either ESV1 or LESV but not shared between them. ESV1 also possesses a proline-rich terminal domain that is absent from LESV, and LESV has a unique >100-amino acid N-terminal domain. Further information about the roles of these proteins will come from investigation of the starch and glucan binding properties of wild-type and mutant forms combined with detailed physical analyses of starches from plants expressing wild-type and mutant forms at different levels.

The Importance of ESV1 and LESV for Starch Degradation Varies between Organs

Examination of the esv1 mutant indicates that the importance of the ESV1 protein for normal starch turnover varies from one organ and tissue to another. In leaves, loss of ESV1 accelerates starch degradation and also reduces the extent of starch accumulation during the day. The reduction in starch accumulation may be due to the occurrence of starch degradation in the light as well as in the dark: The accumulation of maltose, the main product of starch degradation in the chloroplast (Weise et al., 2004), in esv1 leaves during the day supports this idea. Thus, loss of ESV1 may render leaf starch granules partially accessible to hydrolytic enzymes during the day, as well as increasing access for these enzymes during the night.

In other organs of the plant, loss of ESV1 has a more radical effect on starch turnover. The starch content of root columella cells and of stems in esv1 plants is very much lower in mutant than in wild-type plants. In sex1 mutant plants, loss of ESV1 has little effect on the starch excess phenotype in mesophyll cells of the leaf but abolishes starch accumulation in root cap, vein, and petiole cells. These observations indicate that the importance of ESV1 for normal starch metabolism is dependent on the dynamics of starch turnover. In wild-type leaves, little or no starch degradation occurs during most of the light period: The rates of starch biosynthesis and starch accumulation are the same. In the absence of ESV1, some degradation occurs during the day, but this rate is far lower than the rate of biosynthesis. In other organs of the plant, starch may be subject to simultaneous biosynthesis and degradation at all times. In the embryo, for example, the effects of loss of starch degrading enzymes on starch content in early development reveal that starch biosynthesis and degradation occur simultaneously throughout the period of accumulation and loss of starch (Andriotis et al., 2010). Thus, we anticipate that loss of ESV1 effectively prevents starch storage in organs other than the leaf because it permits faster degradation without a change in the rate of biosynthesis.

Are ESV1 and LESV part of the control mechanism that adjusts the rate of leaf starch degradation according to time of dawn and starch content? Such a role cannot be ruled out, but at present there is no evidence that the actions of these proteins are subject to either transcriptional or posttranslational control on a short-term basis. The wide occurrence and conservation of both proteins, and the importance of ESV1 in many organs of the plant, indicate that they are fundamental components of the starch biosynthesis and turnover machinery in general.

METHODS

Plant Materials and Growth Conditions

All Arabidopsis thaliana mutants were in the Col-0 background. Starch-excess mutants were described previously (amy3, Yu et al., 2005; bam mutants, Fulton et al., 2008; gbss, Seung et al., 2015; isa1, Delatte et al., 2005; lsf1, Comparot-Moss et al., 2010; sex1, Yu et al., 2001; sex4, Kötting et al., 2009; pwd, Kötting et al., 2005). The lines were as follows: amy3-2 (SAIL_613 D12), bam1-1 (SALK_039895), bam3-1 (CS92461), bam4-1 (SALK_037355), gbss (GABI_914G01), isa1 (SALK_042704), lsf1-1 (SALK_100036), sex1-3 (Yu et al., 2001), sex4-3 (SALK_102567), and pwd (SALK_110814). Double mutants were selected from the F2 of crosses between esv1 and a second mutant line. Retention of both T-DNA insertions was checked for all double mutant lines. Primers are listed in Supplemental Table 5.

Unless otherwise stated, Arabidopsis plants were grown on soil in growth cabinets or controlled environment rooms with a 12-h-light/12-h-dark cycle, at 20°C and at 150 to 180 µmol photons m⁻² s⁻¹ (metal halide lamps).

Generation and Mutagenesis of the Starvation Reporter Line

An Arabidopsis starvation reporter line was established by transforming wild-type plants with a construct consisting of a fusion of the starvation responsive promoter from At1g10070 and the luciferase (LUC) gene (Graf et al., 2010). To create M0 seeds, 400 mg of homozygous T4 seeds of the reporter line were incubated in a 50-mL tube in 10 mL EMS solution (0.15% [v/v] EMS and 0.02% [v/v] Tween 20) on a rotating shaker at room temperature for 20 h. Seeds were washed with 10 mL 0.02% (v/v) Tween 20 solution for 10 min on a rotating shaker. This washing step was repeated 12 more times, and the seeds were then combined with 330 mL 0.1% (w/v) aq. agar and pipetted on soil in 121-cm² pots (10 mL per pot). The pots were held at 4°C for 3 d and transferred to a controlled environment room. At 12 d old, five seedlings were transferred to each of 1500 pots (121 cm²) in a greenhouse and allowed to set M2 seed. Seed from each pot was collected as a single pool.

Imaging of Bioluminescence

Seedlings or plants grown on soil were sprayed with luciferin solution (0.8 mM luciferin and 0.01% [v/v] Triton X-100) 24 h before imaging, then sprayed again 60 min before imaging. Plants were transferred to the NightOwl CCD camera system (Berthold Technologies), and bioluminescence was assayed using either Indigo software (imaging settings: 1 min exposure time, 2 × 2 binning) or WinLight software (imaging settings: 1 min exposure time, medium resolution, pixel binning 4 × 4, and single frame accumulation according to the manufacturer’s instructions).

Mapping of the ESV1 Gene

The esv1 mutant was outcrossed to the Arabidopsis accession Ler. About 20 F1 plants were allowed to self-pollinate and set seed. F2 plants were grown for 10 d before identification of individuals showing bioluminescence by the end of the normal night.

DNA was extracted from individual homozygous F2 plants. To obtain a rough map position for the mutation, each plant was genotyped using genetic markers that were distributed over the five chromosomes (Supplemental Table 1). Recombination frequencies for each marker were calculated as the percentage of Ler polymorphisms detected at that locus. To obtain a smaller interval, new polymorphic markers upstream and downstream of the marker indicating the lowest recombination frequency were analyzed. New markers were obtained using the Arabidopsis mapping platform (Hou et al., 2010; http://amp.genomics.org.cn/).

Identification of the ESV1 Gene

The mutation underlying the esv1 phenotype was identified by genome resequencing. About 500 homozygous mutant plants were selected from the F2 of the cross between the mutant and Ler. DNA was prepared from nuclei extracted from leaves. Library construction from 5 µg of RNA-free genomic DNA, cluster generation, and sequencing on one lane on the Illumina GAIIx platform were performed by The Genome Analysis Centre. The raw data were analyzed as follows. Maq v0.71 (Li et al., 2008) was used to align the 35.6 M 100 base paired-end reads against the TAIR8 Col-0 reference sequence, producing ∼28× coverage, thus generating a list of raw single-nucleotide polymorphisms (SNPs). The maq.pl Perl script was employed to filter the SNPs on quality criteria, and the survivors were used as input to a postprocessing script that first eliminated SNPs that corresponded to known Col-0/Ler polymorphisms (http://signal.salk.edu/atg1001/data/) and then retained only EMS candidates from the remainder. The output of this script was a GFF file that was loaded into a local instance of the GBrowse genome browser (Stein et al., 2002), together with the TAIR8 pseudochromosome sequences and gene model annotations, allowing visual inspection through a web browser. By interrogating the GBrowse MySQL database with a Perl script using Bio::DB::GFF methods, a genome-wide list of EMS candidates (G/C→A/T) within annotated gene sequences and inferred to induce either nonsynonymous codon or donor/acceptor splice site mutations was produced, which was then further refined based on chromosomal location.

Quantification of Transcripts for Starvation Marker Genes

RT-qPCR was used to measure transcript levels of starvation marker genes, using the primers and methods described by Graf et al. (2010). Briefly, total RNA was extracted from entire Arabidopsis rosettes harvested at the end of night using an RNeasy Plant RNA purification kit (Qiagen). Following DNase treatment, 2 μg total RNA was used for reverse transcription using RevertAid Reverse Transcriptase (Thermo Fisher), and qPCR analysis was performed using the Fast SYBR Green master mix together with a 7500 Fast Real-Time PCR system (Applied Biosystems). Transcript levels were calculated relative to the YLS8 housekeeping gene. The primer pairs are listed in Supplemental Table 5.

Measurement of Starch and Sugar Contents

Starch and soluble sugars were extracted and quantified as previously described (Critchley et al., 2001; Delatte et al., 2005; Martinis et al., 2014). Briefly, rosettes were harvested into liquid nitrogen and ground to a fine powder in a ball mill. The powder was suspended and agitated in ice-cold 0.7 M perchloric acid. Following centrifugation, the pellet was washed three times in 80% (v/v) ethanol, resuspended in water, heated to gelatinize the starch, digested with α-amylase and amyloglucosidase, and assayed enzymatically for glucose. For sugars, the supernatant was neutralized, passed through sequential cation- and anion-exchange columns (Dowex 50 and Dowex 1), and analyzed by high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD; Dionex).

Visualization of Starch in Tissues

For iodine staining, tissues were decolorized in hot 80% (v/v) ethanol, rinsed in water, stained in Lugol’s iodine solution, and rinsed again. For visualization of starch in the stem, tissue sections were stained using the modified pseudo-Schiff propidium iodide (mPS-PI) staining method described by Truernit et al. (2008). Stained sections were imaged using confocal laser scanning microscopy, as described below.

Analysis of Starch Structure and Composition

The chain length distribution of amylopectin was profiled as described by Streb et al. (2008). Briefly, starch in the pellet from the perchloric acid extraction (see above) was debranched with isoamylase from Pseudomonas sp (Sigma-Aldrich) and pullulanase M1 from Klebsiella planticola (Megazyme). The resulting glucan chains were purified by passage through sequential cation- and anion-exchange columns and analyzed by HPAEC-PAD.

Analysis of Granule Morphology, Amylose Content, and Starch-Bound Phosphate

For the determination of amylose content, starch-bound phosphate, and granule morphology by scanning electron microscopy, starch granules were purified from 4-week-old Arabidopsis rosettes as described by Seung et al. (2015). The apparent amylose content of the starch was determined using the iodine colorimetry-based method described by Zeeman et al. (2002). Starch-bound phosphate was quantified as described by Santelia et al. (2011). Hydrolyzed starch was dephosphorylated with Antarctic phosphatase (New England Biolabs), and the phosphate released was quantified using malachite green. Granule morphology was examined using a Merlin field emission scanning electron microscope (Zeiss).

To visualize granule morphology within chloroplasts, segments from young leaves of 3-week-old plants were fixed in glutaraldehyde followed by osmium tetroxide and embedded in Epon resin as described by Seung et al. (2015). Light microscopy images of toluidine blue-stained sections were acquired on an AxioImager Z2 microscope fitted with a 100× oil-immersion lens with 1.4-numerical aperture and an AxioCam monochrome camera (Zeiss).

For transmission electron microscopy, ultrathin (70 µm) sections were cut with a diamond knife and placed on formvar carbon-coated copper grids, stained with 2% (w/v) uranyl acetate and Reynold’s lead citrate, and imaged with a FEI Morgagni 268 electron microscope. Pictures are representative of sections from two individual plants per genotype.

Expression Vectors for YFP-Fusion Proteins in Planta and Plant Transformation

The coding sequences for Arabidopsis ESV1 and LESV were amplified from the full-length cDNA clones RAFL09-78-O20 and RAFL16-10-H06, respectively (RIKEN Bioresource Centre), using primers flanked with attB recombination sites. Primers are listed in Supplemental Table 5. The amplified inserts were recombined into the Gateway-compatible entry vector pDONR221 (Invitrogen) and then recombined into the expression vector pB7YWG2 (Karimi et al., 2002), downstream of the CaMV 35S promoter and in frame with the C-terminal YFP tag. Expression constructs were transformed into Agrobacterium tumefaciens strain GV3101.

Transient expression of YFP-tagged proteins in Nicotiana sylvestris (wild type and pgm) leaves was achieved by infiltrating Agrobacterium cells into the abaxial epidermis (Seung et al., 2015). Stable transformation of Arabidopsis was also performed using the floral-dipping method, as described by Zhang et al. (2006). Transformants were identified in the T1 generation based on their resistance to the herbicide Basta. Homozygous plants were identified in the T2 generation based on the segregation rates of the Basta resistance gene.

Detection of ESV1 and LESV Proteins by Silver Staining and Immunoblotting

For extraction of total (soluble and insoluble) proteins, two young leaves from individual 4-week-old rosettes were harvested and homogenized using a pestle in microcentrifuge tubes in 300 μL extraction medium (40 mM Tris-HCl, pH 6.8, 5 mM MgCl₂, 2% [w/v] SDS, and Complete Protease Inhibitor [Roche]). Insoluble debris was pelleted at 20,000g. The protein concentration of the supernatant was determined using the Pierce BCA protein assay kit (Thermo Fisher), and the indicated amounts of protein were loaded onto SDS-PAGE gels.

For the fractionation of soluble and insoluble proteins from N. sylvestris leaves, 7-mm leaf discs were collected from transformed leaves at the end of the photoperiod 3 d after infiltration and snap frozen in liquid N₂. Discs were homogenized using a pestle in microcentrifuge tubes in 100 μL extraction medium (40 mM Tris-HCl, pH 6.8, 5 mM MgCl₂, and Complete Protease Inhibitor). Insoluble debris was pelleted at 20,000g. The pellet was washed once in extraction medium, then resuspended in 100 μL SDS-PAGE loading medium (50 mM Tris-HCl, pH 6.8, 3% [w/v] glycerol, 2% [w/v] SDS, 100 mM DTT, and 0.005% [w/v] bromophenol blue) .The suspension was heated at 95°C for 5 min, and insoluble debris was removed by centrifugation. The supernatant was diluted with 10× SDS-PAGE loading medium.

Granule-bound proteins were extracted from purified starch granules (prepared as described above) using the method described by Seung et al. (2015).

Silver staining was performed with the Silver Stain Plus kit (Bio-Rad). For immunoblotting, proteins were transferred onto a PVDF membrane following SDS-PAGE and probed with antisera specific to ESV1 or LESV. Antisera were raised in rabbits against recombinant ESV1 or LESV proteins expressed in and purified from Escherichia coli. YFP-tagged ESV proteins were detected with an anti-GFP antiserum (Clontech). Plant actin was detected with a commercial monoclonal antibody (Sigma-Aldrich A0480). Dilutions of antisera were as follows: anti-ESV1, 1:1000; anti-LESV, 1:3000; anti-GFP, 1:5000; and anti-actin,1:10,000.

Confocal Laser Scanning Microscopy

Confocal laser scanning microscopy was performed on an LSM 780 confocal microscope (Carl Zeiss), with a 40× water-immersion lens (1.1 numerical aperture). For the acquisition of YFP signal, the excitation beam was produced with an argon laser set at 514 nm, and emitted light was captured between 518 and 557 nm. The autofluorescence of chlorophyll was captured between 662 and 721 nm. Images were processed with ImageJ software (http://rsbweb.nih.gov/ij/).

Phylogenetic Analysis

To build the phylogenetic tree, ESV1 and LESV sequences were retrieved from the NCBI and 1000 plants (1KP; Johnson et al., 2012; http://www.onekp.com) databases using BLASTp. The alignment was constructed using the MAFFT server (Katoh and Standley, 2013) with the “Auto” alignment strategy. The tree was built using MEGA software version 6 (Tamura et al., 2013), using an LG model, four gamma categories for rate variation, an SPR level 5 method for heuristic search, and a neighbor-joining tree as starting tree. One thousand bootstrap replicates were used to assess branch support (branch support <60 not shown on the tree). The analysis was run of four threads on a Dell Precision M4700 machine with an Intel Core i7-3520 CPU with dual core.

Accession Numbers

Sequence data from this article can be found in TAIR (www.arabidopsis.org) under the following accession numbers: ESV1 (At1g42430) and LESV (At3g55760).

Supplemental Data

• Supplemental Figure 1. Selection of isa mutants in the starvation screen.

• Supplemental Figure 2. Identification of the ESV1 gene.

• Supplemental Figure 3. Expression patterns of two starvation marker genes in esv1 and LESV-OX lines.

• Supplemental Figure 4. Conservation of ESV1 and LESV proteins in land plants and algae.

• Supplemental Figure 5. Expression of LESV in Arabidopsis thaliana.

• Supplemental Figure 6. Localization of ESV1 and LESV in N. sylvestris and Arabidopsis.

• Supplemental Figure 7. Effect of changing light intensity on starch content and starch degradation rates in esv1.

• Supplemental Figure 8. Levels of ESV1 and LESV in overexpression lines.

• Supplemental Figure 9. Further characterization of changes in starches in lines lacking or overexpressing ESV1 or LESV.

• Supplemental Table 1. Markers used for mapping the ESV1 gene.

• Supplemental Table 2. Arabidopsis genes coexpressed with LESV.

• Supplemental Table 3. Accession numbers of ESV1-like and LESV-like amino acid sequences.

• Supplemental Table 4. Identification of ESV1 and LESV peptides in starch from various species.

• Supplemental Table 5. Primers used in this study.

• Supplemental Data Set 1. Alignment of ESV1 and LESV amino acid sequences used to generate the phylogenetic tree.

Glossary

Ler

Landsberg erecta

CBM

carbohydrate binding module

SBS

surface binding site