Pale-Green Phenotype of atl31 atl6 Double Mutant Leaves Is Caused by Disruption of 5-Aminolevulinic Acid Biosynthesis in Arabidopsis thaliana

Arabidopsis ubiquitin ligases ATL31 and homologue ATL6 control the carbon/nitrogen nutrient and pathogen responses. A mutant with the loss-of-function of both atl31 and atl6 developed light intensity-dependent pale-green true leaves, whereas the single knockout mutants did not. Plastid ultrastructure and Blue Native-PAGE analyses revealed that pale-green leaves contain abnormal plastid structure with highly reduced levels of thylakoid proteins. In contrast, the pale-green leaves of the atl31/atl6 mutant showed normal Fv/Fm. In the pale-green leaves of the atl31/atl6, the expression of HEMA1, which encodes the key enzyme for 5-aminolevulinic acid synthesis, the rate-limiting step in chlorophyll biosynthesis, was markedly down-regulated. The expression of key transcription factor GLK1, which directly promotes HEMA1 transcription, was also significantly decreased in atl31/atl6 mutant. Finally, application of 5-aminolevulinic acid to the atl31/atl6 mutants resulted in recovery to a green phenotype. Taken together, these findings indicate that the 5-aminolevulinic acid biosynthesis step was inhibited through the down-regulation of chlorophyll biosynthesis-related genes in the pale-green leaves of atl31/atl6 mutant.


Introduction
Chlorophyll (Chl) biosynthesis is regulated primarily at the level of 5-aminolevulinic acid (ALA) synthesis. In higher plants, ALA is synthesized from glutamate in the plastid in three enzymatic steps. In the first step, the enzyme glutamyl-tRNA synthetase aminoacylates tRNA Glu with glutamate, generating the substrate for plastid translation and tetrapyrrole biosynthesis. In the following steps, tRNA Glu is reduced by glutamyl-tRNA reductase (GluTR) to glutamate-1semialdehyde, which is converted to ALA by glutamate-1-semialdehyde aminotransferase [1,2].
Of the three enzymes responsible for ALA synthesis, GluTR is regarded as the main target of regulatory mechanisms [3]. The transcription and activity of GluTR is controlled by light, an endogenous clock, cytokinin and development [2][3][4][5]. In Arabidopsis, three HEMA genes encode GluTR. HEMA3 is considered a pseudogene [6], whereas HEMA1 and HEMA2 encode proteins that are 81% identical at the amino acid level [7,8]. Since HEMA1 is highly expressed in photosynthetic tissues of leaves and stems, whereas HEMA2 is constitutively expressed at low levels in all tissues, HEMA1 is considered the dominant form of GluTR for Chl biosynthesis [9,10]. RNA interference (RNAi)-induced down-regulation of HEMA1 expression in Arabidopsis and tobacco resulted in a reduction in plant Chl [5,11]. Furthermore, a hema1 mutant showed a yellowish phenotype, due to a drastically reduced level of Chl with severe growth retardation [12]. Thus, regulating HEMA1 is essential for Chl biosynthesis.
Genes in the Arabidopsis Tóxicos en Levadura (ATL) family encode plant-specific putative RING-type ubiquitin ligases with transmembrane domains. In Arabidopsis, the ATL family is composed of 91 members [13,14]. ATL31 and its closest homolog ATL6 are membraneassociated ubiquitin ligases, shown to be involved in the carbon/nitrogen (C/N) response by regulating the stability of 14-3-3 proteins through their ubiquitination activity [15][16][17]. Plants overexpressing full-length ATL31 or ATL6 (35S-ATL31 and 35S-ATL6) were insensitive to high C/N stress conditions, whereas the single knockout mutants atl31-1 and atl6-1, as well as an atl31-1/atl6-1 double knockout mutant showed increased sensitivity to high C/N stress conditions [15,18]. ATL31 and ATL6 are also involved in the plant immune response. 35S-ATL31 showed enhanced resistance to the bacterial pathogen Pseudomonas syringae pv. tomato DC3000 and powdery mildew Bgh. [18,19]. Moreover, the double mutant, but not the single mutants, showed increased susceptibility to Pst. DC3000. In contrast, the double mutant and wild-type (WT) showed comparable resistance to Bgh entry [18,19]. In addition, the true leaves of the atl31-1/atl6-1 double mutant had a pale-green phenotype during its early developmental stage [15]. However, the mechanism underlying this pale-green phenotype remains unknown.
In this present study, we found that this phenotype was dependent on light intensity. Under low light conditions, the double mutant appeared similar to WT, whereas under medium light conditions, the double mutant showed a decreased level of Chl, an abnormal plastid structure and reduced levels of plastid proteins on their true leaves. Expression analyses revealed that the expression of HEMA1 and GLK1, which encodes key transcription factor for Chl biosynthesis related genes including HEMA1, were markedly lower in the pale-green true leaves of the double mutant than that in WT plants. Application of exogenous ALA could restore the pale-green phenotype of the double mutant to a green phenotype, bypassing the reduction of HEMA1 function. These findings indicate the involvement of ATL31 and ATL6 to Chl biosynthesis by controlling 5-aminolevulinic acid biosynthesis step through the down-regulation of HEMA1 gene.

Measurement of chlorophyll and Fv/Fm
Chl content and the Chl a/b ratio were determined as described [21]. The maximal photochemical efficiency of PSII (Fv/Fm) was measured with a PAM-2000 chlorophyll fluorometer (Walz, Germany) as described [22].

Molecular cloning
To generate an RNAi transgenic plant targeting ATL31 and ATL6, a fragment of complementary DNA encoding a truncated ATL6 was amplified by PCR using pENTRATL6 [18] as the template and the primers shown in S1 Table. The resulting product was cloned into the pENTR/D-TOPO vector (Life Technologies) to generate the plasmid pENTRATL6RNAi. The ATL6RNAi fragment was recombined into the pHellsgate12 transfer-DNA binary vector [23] according to the manufacturer's directions (Life Technologies). All PCR products and inserts were verified by DNA sequencing.

Gene expression analysis
RNA isolation, reverse transcription and PCR were performed as described [24]. RT-PCR was performed using normalized cDNA samples. PCR products were electrophoresed on agarose gels and visualized by ethidium bromide staining. Quantitative real-time PCR was performed with the SYBR Premix Ex Taq (TaKaRa) on an Applied Biosystems 7300 Real-Time PCR system (Applied Biosystems) according to the manufacturer's instructions. All signals were normalized relative to the level of expression of EF1α, with relative gene expression calculated using the ΔΔCT method. The primers used are listed in S1 Table. Transmission electron microscopy To assess plastid structure, first or second leaves from 2-week old WT and atl31-1/atl6-1 plants were fixed in 1% (w/v) glutaraldehyde, 4% (w/v) paraformaldehyde in 0.05 M phosphate buffer, pH 7.4, for 2 h at room temperature, and washed several times with phosphate buffer. The samples were further fixed in 0.5% (w/v) osmium tetroxide, 0.15 M phosphate buffer, pH 7.4, for 2 h at room temperature, washed twice with phosphate buffer, and dehydrated using an acetone series prior to infiltration with Spurr's resin (Polysciences, Inc.). Following polymerization at 70°C for 8 h, the ultra-thin sections were cut using an ultramicrotome (EM UC6, Leica Microsystems) and viewed under a transmission electron microscope (H-7650, Hitachi).
Blue-Native PAGE, immunoblot analysis and CBB staining BN-PAGE and immunoblot analyses were performed as described [25]. Total proteins were extracted from first or second true leaves of 2-week-old WT and atl31-1/atl6-1, with 8 mg of FW used for BN-PAGE and 0.5 mg FW for immunoblotting. For immunoblot analyses, the transferred proteins were detected with antibodies to PsbC (CP43) and PsaA/PsaB (CP1) [26] and with antibodies to PsbD and Lhcb1 (Agrisera, Sweden). For CBB staining, the proteins were visualized using EzStain AQua (ATTO).

Results
Double mutant atl31/atl6 developed light intensity-dependent pale-green leaves h, 100 μmol m -2 s -1 ), with cotyledons similar to those of WT ( Fig. 1 and [15]). These palegreen leaves of mutant plants gradually turned green from the petiole to the leaf apex about 25 days after germination ( Fig. 1), having green-colored leaves at the matured stage, similar to those of WT plants (S1A and S1B Fig.). Efforts to identify the key factor responsible for this phenotype in mutant plants found that the loss of pigment was dependent on light intensity ( Fig. 2A and 2B). Under low light (LL) conditions (40 μmol m -2 s -1 ), the color and Chl contents of mutant true leaves were similar to those of WT. However, as light intensity increased, the chlorophyll contents of mutant leaves were decreased ( Fig. 2A and 2B). The ratio of Chl a to Chl b (Chl a/b) in the mutant was similar to that of WT and independent of light intensity, with leaves showing a pale-green color under medium light (ML: 100 μmol m -2 s -1 ) conditions ( Fig. 2C). In contrast to the double mutants, plants with single mutations in atl31-1 or atl6-1 and those overexpressing ATL31 and ATL6 did not develop any pale-green leaves even under ML conditions (S2 Fig.). To exclude the possibility that this pale-green phenotype is caused by the insertion of other T-DNA sequences, we generated another type of atl31/atl6 double mutant, using different alleles of atl6 mutants and RNAi. The double mutant of atl31-1 and atl6-2, a null-mutant of ATL6 (S3A and S3B Fig.), showed a pale-green phenotype similar to that of the atl31-1/atl6-1 double mutant (S3C Fig.). Moreover, we generated double knock down transgenic plants for atl31 and atl6 by RNAi technique using the RNAi generating vector pHELLSGATE12 [23]. This construct produced a hairpin RNA containing approximately 300 bp of the ATL6 coding region, sharing 75% identity with ATL31. Plants targeted by ATL31/ATL6 also had a pale-green phenotype (S3C and S3D Fig.). These findings indicated that a simultaneous loss of function of the ATL31 and ATL6 genes led to a light-dependent decrease in the amounts of Chl, with this decrease responsible for the pale-green phenotype.
Plastids of atl31/atl6 pale-green leaves had reduced inner membrane structure To investigate the mechanism by which knockout of the ATL31 and ATL6 genes resulted in a pale-green phenotype, we examined plastid ultrastructure by transmission electron microscopy (TEM). Since the pale-green phenotype was gradually recovered from the petiole to the leaf apex ( Fig. 1), we separated a pale-green leaf into 3 sections along the proximal/distal axis to observe the recovering of color of the pale-green leaves (Fig. 3). Under ML conditions, WT plants generated normal crescent-shaped chloroplasts containing plastoglobules, starch grains and well-developed thylakoid structures including stromal thylakoids and grana stacks across sections of the entire leaf ( Fig. 3E-3G). In the same way, the bottom section of an atl31-1/ atl6-1 leaf plastid also showed a structure similar to that of WT under ML conditions (Fig. 3A). However, middle section of atl31-1/atl6-1 leaf plastids was found to contain reduced inner membrane structure (Fig. 3B). Furthermore, leaf plastids in the apex section of atl31/atl6 showed almost total loss of stromal thylakoids and reduced grana stacks (Fig. 3C). These features of the mutant plastid did not resemble the properties of immature leaves or the process of greening or the other pale-green mutants [27][28][29][30][31]. Under LL conditions, however, the structures of the atl31-1/atl6-1 plastids were similar to that of WT even in the leaf apex section ( Fig. 3D and 3H) which is consistent with the same Chl content in WT and atl31-1/atl6-1 (Fig. 2). Lower-magnification views were shown in S4 Fig.. These results suggest that plastid biogenesis in the atl31-1/atl6-1 double mutants was impaired only during early leaf development and in a light intensity-dependent manner.
Double mutant atl31/atl6 pale-green leaves had reduced levels of thylakoid proteins Since the structures of the plastids in atl31-1/atl6-1 resulted in abnormal leaves under ML conditions, we investigated Chl-protein complexes by blue native-polyacrylamide gel electrophoresis (BN-PAGE). Total proteins from the leaves of 2-week-old WT and atl31-1/atl6-1 were separated by BN-PAGE with 4-13% linear gradient gels followed by CBB staining. Although the leaves of the double mutant (atl31-1/atl6-1) had normal levels of expression of Chl-protein complexes, comparable with those of WT under LL conditions, the mutant leaves had markedly reduced levels of Chl-protein complexes under ML conditions, despite these plants having comparable amounts of RubisCO (Fig. 4A).
To confirm the decrease in Chl-binding proteins in atl31-1/atl6-1 under ML conditions, we performed SDS-PAGE followed by Immunoblotting and CBB staining analyses on these  (Fig. 4A). We found that the PSII complex proteins PsbC and PsbD, and Lhcb1, a major LHCII complex protein, were markedly reduced in approximately 10% of that in WT (Fig. 4B). PsaA/B, which are PSI complex protein, were drastically reduced, to approximately 1% of that in WT (Fig. 4B). In contrast, the RbcL level of atl31-1/atl6-1 was normal or somewhat moderately decreased compared with WT (Fig. 4B). Since PSII and LHCII are dominant in the grana stack and PSI is dominant in stroma thylakoids [34,35], the results of immunoblot analyses were consistent with the plastid ultrastructure of pale-green leaves, which showed loss of stromal thylakoids with reduced grana stacks (Fig. 3).
To determine whether the ratio of photo-damaged PSII was increased in the atl31-1/atl6-1 double mutant, we measured Fv/Fm, but found no significant difference between these plants and WT, even when grown under ML conditions (Fig. 5). This result, showing that the ratio of functional to total PSII in atl31-1/atl6-1 was comparable to that of WT, suggested that the pale-green phenotype of atl31-1/atl6-1 was not caused by high light-induced PSII photodamage. This finding, taken together with the normal Chl a/b ratio in atl31-1/atl6-1, suggests that the efficiency of photosynthesis in pale-green leaves of the double mutant was not damaged, although the photoreactive capacity could be reduced under ML conditions. Decreased HEMA1 and GLK1 expression in atl31/atl6 pale-green leaves Using quantitative real-time RT-PCR, we assayed the level of transcripts of photosynthesisrelated genes in WT and atl31-1/atl6-1 (Fig. 6A). Genes assayed included the plastid-encoded RNA polymerase (PEP)-dependent rbcL and psaA genes; the nuclear-encoded-RNA polymerase (NEP)-dependent rpoA and accD genes; and the nuclear-encoded CAB2 and HEMA1 genes whose products are targeted to chloroplasts. Except for HEMA1, all of these genes in WT and atl31-1/atl6-1 showed similar levels of expression in response to different light intensity conditions (Fig. 6A). In contrast, the expression of HEMA1 in atl31-1/atl6-1 was much lower under ML than under LL conditions compared with that in WT (Fig. 6A). Previously, Golden 2-like (GLK) transcription factors GLK1 and GLK2 were identified as positive regulator of nuclear photosynthetic genes including HEMA1 through direct binding to their promoter sequences, resulting in pale-green phenotype of glk mutants [36]. Further analyses of GLK1 and GLK2 transcript levels revealed that GLK1 was specifically down-regulated in atl31-1/atl6-1 under ML condition as well as that of HEMA1 (Fig. 6B). Since HEMA1 is the rate-limiting enzyme of tetrapyrrole biosynthesis [3], these results strongly suggest that the reduced HEMA1 expression in atl31-1/atl6-1 under ML conditions caused a down-regulation of Chl biosynthesis, manifesting as a pale-green phenotype.

ALA application restored the normal phenotype of atl31/atl6
To confirm that Chl synthesis in the atl31-1/atl6-1 double mutant was impaired at the ratelimiting step of ALA synthesis, and that limited ALA synthesis was responsible for the double mutant pale-green phenotype, we tested the effects of the addition of ALA. WT and atl31-1/ atl6-1 plants grown under ML conditions were transferred to medium containing 0, 10, 30 and 100 μM ALA nine days after germination and incubated for three days (Fig. 7A and 7B). The amount of Chl recovery in atl31-1/atl6-1 was dependent on the concentration of ALA, with the highest doses showing leaves similar to those of WT plants (Fig. 7B). Enlarged photographs showed that newly emerging leaves near the petioles were the first to become fully green (S5 Fig.). Taken together of all results indicated that the pale-green phenotype of atl31-1/ atl6-1 was caused by inhibition of 5-aminolevulinic acid biosynthesis through the downregulation of Chl biosynthesis-related genes expression.

Discussion
The pale green phenotype of the atl31/atl6 double mutant was apparently due to a reduction in HEMA1 expression, followed by a reduction in Chl biosynthesis. Our expression analysis showed that HEMA1 expression in atl31/atl6 was markedly down-regulated under ML conditions, whereas the addition of exogenous ALA resulted in the dose-dependent recovery of WT phenotype.
The pale-green phenotype of atl31/atl6 was similar to that of a hema1 mutant recently characterized in Arabidopsis [12]. The plastid of this mutant, with 13% of the Chl present in WT, contains markedly lower numbers of thylakoids but a normal level of RbcL. Similarly, The similar results observed in the hema1 mutant and atl31/atl6 suggests that maturation of the inner membrane structure of both plastid is impaired by the reduction in Chl level.
ATL31 and ATL6 are involved in the regulation of HEMA1 expression maybe through the regulation of GLK1 expression. GLK genes encode a pair of partially redundant transcription factors that are required for the expression of nuclear photosynthesis related-genes including HEMA1 [36,37]. Since GLK regulates expression of Lhcb gene family as well as HEMA1, decreased amounts of Lhcb1 protein in atl31/atl6 double mutant was also consistent to this hypothesis. Present study revealed that the expression of GLK1 was down-regulated in pale-green leaves whereas that of GLK2 was not, ATL31 and ATL6 may regulate the expression of GLK1 specifically by unknown mechanism.
ATL31 and ATL6 function in growth regulation in response to C/N-nutrient availability [15,16] and atl31/atl6 double mutant showed pale-green phenotype due to decreased Chl biosynthesis (this study), implicating close relationship between C/N-nutrient status and Chl biosynthesis. Actually, the primary C and N metabolism interacted at Glu biosynthesis reaction through GS-GOGAT cycle [2,38], which provides Glu-tRNA, the substrate of ALA production by HEMA1. In addition, 14-3-3 proteins, the ubiquitination targets of ATL31 and ATL6, are involved in chloroplast biogenesis. The 14-3-3 protein family consists of 13 expressed isoforms, which bind to a wide variety of target proteins at phosphorylated motifs and function in multiple developmental processes by regulating their activity or sub-cellular localization E3 Ligase ATL31/6 on Chlorophyll Biosynthesis [39,40]. Compared with WT, larger numbers of chloroplasts were found in the roots of 14-3-3μ and 14-3-3ν mutants, depending on the wavelength and intensity of light [41]. These findings, taken together with defects in light sensing and/or response displayed by mutants of 14-3-3μ and 14-3-3ν [42], suggest that these 14-3-3 proteins may be the key integrators between light signaling and chloroplast biogenesis. Since 14-3-3 proteins are also involved in light signaling that regulates HEMA1 transcription [40], ATL31 and ATL6 may be involved in chloroplast biogenesis by regulating the stability of specific isoforms of 14-3-3 proteins in a light intensity-dependent manner. To date, however, the detailed mechanism of the relationship between 14-3-3 proteins and chloroplast biogenesis is still largely unknown.
Other ATL members may also participate in Chl biosynthesis. The pale-green phenotype of atl31-1/atl6-1 was specific to true leaves and transient in younger developmental stages. In addition, the down-regulation of HEMA1 expression only occurred early during true leaf development, followed by a deficiency in plastid biogenesis. These results suggest that ATL31 and ATL6 regulate HEMA1 expression in true leaves of young plants early during development. Since the ATL family in Arabidopsis consists of 91 members [14], other members of this family could be participating in Chl biosynthesis at different developmental stages or in different tissues.
Although, determination of the detailed molecular mechanism by which ATL31 and ATL6 regulate HEMA1 expression is still unknown, further investigation of ATL31 and ATL6 provide us a new aspect of Chl biosynthesis regulation.  Table. Primer sequences for vector constructs and gene expression assays.