The ACT domain in chloroplast precursor–phosphorylating STY kinases binds metabolites and allosterically regulates kinase activity

Protein import of nucleus-encoded proteins into plant chloroplasts is a highly regulated process, requiring fine-tuning mechanisms especially during chloroplast differentiation. One way of altering import efficiency is phosphorylation of chloroplast transit peptides in the cytosol. We recently investigated the role of three serine/threonine/tyrosine (STY) kinases, STY8, STY17, and STY46, in precursor phosphorylation. These three kinases have a high degree of similarity and harbor a conserved aspartate kinase–chorismate mutase–tyrA (prephenate dehydrogenase) (ACT) domain upstream of the kinase domain. The ACT domain is a widely distributed structural motif known to be important for allosteric regulation of many enzymes. In this work, using biochemical and biophysical techniques in vitro and in planta, including kinase assays, microscale thermophoresis, size exclusion chromatography, as well as site-directed mutagenesis approaches, we show that the ACT domain regulates autophosphorylation and substrate phosphorylation of the STY kinases. We found that isoleucine and S-adenosylmethionine bind to the ACT domain, negatively influencing its autophosphorylation ability. Moreover, we investigated the role of the ACT domain in planta and confirmed its involvement in chloroplast differentiation in vivo. Our results provide detailed insights into the regulation of enzyme activity by ACT domains and establish that it has a role in binding amino acid ligands during chloroplast biogenesis.

As a consequence of endosymbiosis, most chloroplast genes were transferred to the nucleus, and the resulting proteins were equipped with N-terminal cleavable transit peptides. A specific subset of these precursor proteins associates with cytosolic compounds, among them 14-3-3 proteins, for efficient relocation from the cytosol to chloroplast translocon prior to the import process (1)(2)(3)(4). Several of these precursor proteins are phosphorylated within the transit peptides at specific serine or threonine residues, leading to their association with 14-3-3 proteins. Although formation of the precursor-14-3-3 complex enhances import rates, dephosphorylation is required prior to translocation to ensure efficient import (5,6). Regulation by phosphorylation has been shown to be especially important in cotyledons and during chloroplast differentiation (7). Precursor phosphorylation is executed by the cytosolic serine/threonine/tyrosine (STY) kinases in Arabidopsis, i.e. STY8, STY17, and STY46. Mutant lines with deletion of STY8 and STY46 accompanied by knockdown of STY17 cause inefficient chloroplast differentiation and delayed accumulation of chlorophyll, hamper photosynthetic capacity, and have a significant reduction in nucleus-encoded chloroplast proteins (7).
A unique feature of these kinases is that they harbor a predicted but uncharacterized ACT 3 domain located upstream of their kinase domains (7). The ACT domain has been found in several enzymes; thus, the name was derived from the first letters of three of the proteins in which it was first identified: aspartate kinase-chorismate mutase-tyrA (prephenate dehydrogenase) (8). ACT domains have been identified in proteins with a plethora of functions, such as control of metabolism, solute transport, and signal transduction (9). The ACT domain ranges from 60 -80 amino acids, and the structural motif comprises four ␤ strands and two ␣ helices organized in a ␤␣␤␤␣␤ fold (10,11). Because the primary sequences are very variable, their identification often requires a combined sequence/structure and also functional analysis (11). Interestingly, ACT domains can vary in number and often mediate oligomerization. The ACT domain has been shown to bind to multiple ligands and generally seems to play a role in allosterically regulating the enzymatic activity of proteins. The ligand types range from amino acids to enzyme cofactors (9,10). In some cases, enzyme activity can be inhibited upon ligand binding, as observed in the Escherichia coli D-3-phosphoglycerate dehydrogenase, where binding to its ligand (serine) causes disruption of the substrate-binding domain (12). In contrast, phenylalanine stabilizes a hydroxylase upon substrate binding in the ACT domain (13). Interestingly, it has been shown that lysine and the cofactor S-adenosylmethionine (SAM) synergistically inhibit aspartate kinase 1 from Arabidopsis by binding to different regions within the ACT domain (14,15).
The STY8, STY17, and STY46 kinases are particularly interesting, as they are the only protein phosphorylating The authors declare that they have no conflicts of interest with the contents of this article. 1  cro ARTICLE enzymes known to contain an ACT domain. We could show previously that deletion of the ACT domain of STY8 causes an increase in autophosphorylation as well as precursor phosphorylation yield in vitro (7). In this study, we examined the role of the ACT domain of the STY kinases using biochemical and biophysical techniques in vitro as well as in planta. We identified Ile as well as SAM as negative regulators acting on kinase autophosphorylation. Thus, this study presents new insights into the regulation of enzyme activity by ACT domains and elucidates its role with respect to ligand binding during chloroplast biogenesis.

The ACT domain regulates STY kinase activity and substrate binding
As a first step, all three STY kinases (STY8, STY17, and STY46) as well as the corresponding ACT domain deletions STY8⌬ACT, STY17⌬ACT, and STY46⌬ACT were expressed in E. coli and purified via affinity chromatography. Autophosphorylation of the six recombinant kinases was investigated using radiolabeled [␥-32 P]ATP. All ⌬ACT deletion mutants displayed hyperautophosphorylation to a similar extent compared with their WT variants (Fig. 1A). To analyze substrate phosphorylation, the small subunit of the chloroplast ribulose-1,5-bisphosphatase (pSSU), which is known to be phosphorylated within the transit peptide, was used as a model substrate (1,16). pSSU was expressed and purified as a His tag fusion protein from E. coli. Phosphorylation of pSSU was also found to be enhanced by all three kinases lacking the ACT domain (Fig. 1A). As a negative control, we used STY8 T439A, STY17 T445A, and STY46 T443A, mutants that exhibit loss of autophosphorylation activity (Fig. 1A). Quantification of three independent kinase assays was performed and is shown in Fig. 1A.
Next we set out to investigate the binding affinities of STY8 to its substrate using microscale thermophoresis (MST). For the MST measurements, pSSU was labeled with a fluorescent dye, and interaction was analyzed applying increasing amounts of recombinant STY8 or STY8⌬ACT. Indeed, strong binding of STY8 to pSSU occurred with a K D of 8.08 Ϯ 1.4 nM. Intriguingly, the binding affinity was even enhanced with STY8⌬ACT, to a K D of 4.20 Ϯ 2.5 nM (Fig. 1B). To investigate whether the autophosphorylation state of the kinase directly affects the binding affinity to pSSU, we also used STY8 T439A. In fact, we obtained the weakest binding affinities of STY8 T439A with pSSU (K D ϭ 708 Ϯ 6.5 nM) (Fig. 1B). These data suggest an important role of the ACT domain not only in the regulation of autophosphorylation but also indicate that autophosphorylation triggers formation of a kinase-precursor complex.

STY8 and as well as the ACT domain form oligomers in vitro
Several ACT domain-containing proteins have been isolated as large multimeric proteins, with ACT domains mediating the formation of dimers, trimers, or even tetramers (10). To elucidate the oligomeric state of the recombinant STY8 protein, we employed size exclusion chromatography (SEC). Our results showed that the majority of STY8, which has a monomeric size of 63 kDa, eluted at 217 kDa, suggesting the forma-tion of a trimer or a tetramer ( Fig. 2A). To investigate the role of the ACT domain in the oligomerization state of STY8, we purified and analyzed only the ACT domain, which has a calculated size of 7.5 kDa. Our results indicate that the purified ACT domain of STY8 likewise elutes as a trimer or tetramer because of the apparent size of 35 kDa, indicating that oligomerization may be mediated by the ACT domain (Fig. 2B). To further investigate the oligomerization status, we applied a nondenaturing native PAGE (Fig. 2C). The results show a band at approximately 230 kDa, also indicating trimer or tetramer formation.

STY8 is negatively regulated by Ile and SAM via its ACT domain
Because amino acids are common ligands of ACT domains, we conducted radioactive kinase assays with STY8 and 20 amino acids at a final concentration of 5 mM. Quantifying the kinase activity via its autophosphorylation levels, we found that Ile had a strong negative effect on kinase autophosphorylation (Fig. 3A). When applying increasing amounts from 0 -5 mM Ile to a STY8 autophosphorylation assay, we observed a decrease in autophosphorylation with increasing amounts of Ile (Fig. 3B, top). However, using STY8⌬ACT, we observed a loss of Ile sensitivity, suggesting that Ile binding indeed occurs within the ACT domain (Fig. 3B, bottom). To analyze the binding of STY8 and Ile on a quantitative level, we employed MST. STY8 and STY8⌬ACT were labeled with a fluorescent dye, and increasing amounts of Ile were used as analytes. Indeed, we observed a binding event using WT STY8 (K D ϭ 311 Ϯ 16 M). However, no binding curve could be fitted using STY8⌬ACT, again supporting the idea of the ACT domain as an Ile-binding site (Fig. 3C).
It has been reported previously that, along with amino acids, ACT domains also interact with various metabolites derived from amino acids. Interestingly, the ACT domain of the Arabidopsis aspartate kinase binds to lysine as well as to SAM. Therefore, we analyzed the effect of SAM on STY8 and discovered that the metabolite down-regulates STY8 autophosphorylation (Fig. 3D). To examine whether SAM also affects the kinase activity via the ACT domain, we incubated increasing amounts of SAM with STY8 and STY8⌬ACT and compared autophosphorylation activity (Fig. 3D). Indeed, SAM reduces autophosphorylation in a manner similar to Ile, and, again, the activity of STY8⌬ACT showed SAM resistance, implying that SAM also binds to STY8 via the ACT domain. To quantify the binding event, we conducted an MST analysis with SAM and STY8 or STY8⌬ACT. Binding affinity could again be calculated for STY8 (K D ϭ 774.7 Ϯ 39 M), but no curve could be fitted for STY8⌬ACT (Fig. 3E). To verify further that metabolite binding is conferred by the ACT domain, we used the recombinant ACT domain to perform MST analysis with Ile and SAM. Interestingly, the K D values were enhanced to 81.48 Ϯ 34 M for Ile and 47.28 Ϯ 25 nM for SAM (Fig. 3F). This result unequivocally confirms that both metabolites are bound by the ACT domain. The differences observed in binding affinities using the full-length proteins versus the isolated ACT domain might indicate that steric hindrance occurs in measurements with the full-length proteins, leading to lower accessibility for the metabolites. Nevertheless, SAM still seems to bind to STY8 with a higher affinity compared with Ile. To confirm the speci-

Regulation of STY kinases by a metabolite binding ACT domain
ficity of the binding event, we included methionine as a negative control. No binding event could be observed for these measurements (Fig. 3F).
Because we observed that the extent of autophosphorylation influences affinity for the precursor pSSU, we wondered whether a reduction in autophosphorylation by adding SAM or Ile has a negative effect on precursor binding. Consequently, we performed an MST analysis with STY8 and pSSU in the presence of Ile and/or SAM. In this case, pSSU was fluorescently labeled, and the final concentration of the labeled pSSU and SAM and/or Ile was kept constant at 1 mM. STY8 was titrated against pSSU, and the thermophoretic mobility was monitored (Fig. 3G). Addition of SAM or Ile had an equal negative effect on the binding of STY8 to pSSU; in both cases, the K D was reduced ϳ4-fold. Interestingly, a 10-fold reduction in binding affinity was observed when SAM and Ile were applied simultaneously (Fig. 3G). To investigate whether STY17 and STY46 were likewise affected by Ile and SAM, we performed autophosphorylation assays in the presence and absence of 5 mM Ile and 5 mM SAM with all three WT kinases as well as kinases lacking the ACT domain. All three kinases behaved in a similar manner, and quantification of the results showed that addition of Ile or SAM resulted in down-regulation of autophosphorylation of ϳ50% in all three kinases, whereas deletion of the ACT domain resulted in metabolite insensitivity (Fig. 3H).

Mutation of a conserved glycine in the ACT domain causes loss of Ile sensitivity
Ligand binding in ACT domains is usually observed at the interfaces between the domains and there appears to be a correlation with the occurrence of a specific glycine residue located in the loop between the first ␤ strand and the first ␣ helix of ACT domains (10). For example, a glycine-to-aspartate mutation of the ACT domain-containing aspartokinase in E. coli revealed a significant loss of amino acid sensitivity (17). To investigate the presence of such a residue in the STY kinases, we performed structure prediction with the ACT domain of STY8 using the iterative threading assembly refinement (I-TASSER) server. Indeed, we could identify a glycine residue at position 197 in the respective loop area in STY8, which also proved to be conserved in STY17 and STY46 as well as in homologs in other plants species (7). The predicted structure, along with an expected scheme of the ACT domain fold, is shown in Fig. 4A. The ␣ helices are depicted in pink, ␤ sheets in blue, and the conserved glycine in green.
To study the role of the glycine residue in question, we substituted the amino acid by aspartate (STY8 G197D) and compared its Ile sensitivity to WT STY8, monitoring autophosphorylation with increasing Ile concentrations (Fig. 4B). Intriguingly, we observed Ile insensitivity of the STY8 G197D mutant, similar to STY8⌬ACT, indicating that the exchanged glycine is indeed involved in Ile binding. To investigate further whether the STY8 G197D mutant is also insensitive to SAM, a kinase assay was performed with STY8 G197D in the presence of increasing concentrations of SAM. Strikingly, in contrast to Ile, we observed SAM sensitivity in both the WT and the STY8 G197D mutant. STY8⌬ACT was again used as a negative control, showing that autophosphorylation was not affected by SAM in this mutant (Fig. 4C). To verify these results, we performed MST with STY8, STY8 G197D, and STY8⌬ACT to analyze Ile and SAM binding. Indeed, the binding affinity of Ile to For quantification of the phosphorylation assays, the WT kinase phosphorylation yields were set to 100%. Data show mean Ϯ S.E.; n ϭ 3. Asterisks indicate statistical significance compared with WT control. B, binding affinities of STY8 variants and pSSU. MST was performed with STY8, STY8⌬ACT, and STY8 T439A mutants. Concentrations from 0.1 nM to 3 M of the kinases were titrated against fluorescence-labeled pSSU (25 nM). The weakest interaction was observed between pSSU and the STY8 T439A mutant compared with WT STY8 (K D of 708 Ϯ 6.5 nM and 8.08 Ϯ 1.4 nM, respectively). The strongest interaction was observed between pSSU and the STY8⌬ACT mutant (K D ϭ 4.20 Ϯ 2.5 nM); n ϭ 3. T/A refers to STY8, STY17 and STY46 T to A substitutions as used in Fig. 1A. * p Ͻ 0.05, ** p Ͻ 0.01, *** p Ͻ 0.001.

Regulation of STY kinases by a metabolite binding ACT domain
STY8 G197D was reduced to almost 1 mM, whereas the binding affinity to SAM was not affected in STY8 G197D (Fig. 4D). Overall, our data indicate that the conserved glycine located in the loop region between the ␣-helix (␣1) and the ␤-sheet (␤2) is only responsible for Ile binding. Although SAM likewise binds to the ACT domain, the binding site must be different. This also explains the additive negative effect observed on STY8-pSSU upon addition of SAM and Ile (Fig. 3G).

Impact of deletion of the ACT domain in planta
To study the role of the ACT domain in planta, we used the previously generated sty8 sty46 double mutant, which shows reduced growth as well as slowed greening, most probably because of inefficient import of precursors (7). Because we could show in the respective study that the growth phenotype of the sty8 sty46 mutant could be fully rescued by overexpression of STY46, we complemented the line with STY46 as well as STY46⌬ACT under control of the 35S CMV promoter (sty8 sty46/ 35 S::STY46 and sty8 sty46/ 35 S::STY46⌬ACT). Expression of STY8 and STY46 in the generated lines was confirmed using RT-PCR. Neither STY8 nor STY46 was detected in sty8 sty46, but STY46 was again expressed in sty46/ 35 S::STY46 and sty8 sty46/ 35 S::STY46⌬ACT (Fig. 5A). Because sty8 sty46/ 35 S::STY46 and sty8 sty46/ 35 S::STY46⌬ACT were compared in the following phenotypic studies, quantitative RT-PCR was performed to show that both STY46 variants were expressed to a similar degree (Fig. 5B).
To investigate whether hyperphosphorylation of the kinase and its substrates could also be observed with sty8 sty46/ 35 S::STY46⌬ACT plant extracts, we performed kinase assays using soluble protein extracts of WT plants, sty8 sty46 double mutants, sty8 sty46/ 35 S::STY46, and sty8 sty46/ 35 S::STY46⌬ACT complementation plants to study the phosphorylationpotentialofrecombinantpSSU.Precursorphosphorylation was reduced in the sty8 sty46 double mutant compared with the WT, as expected (Fig. 5C). Residual phosphorylation in sty8 sty46 is most likely due to the redundant function of STY17. However, the precursor phosphorylation yield was approximately 2.5-fold higher in sty8 sty46/ 35 S::STY46⌬ACT compared with the WT as well as with 35 S::STY46, as shown by quantification of three independent experiments (Fig. 5C). We therefore conclude that the STY46⌬ACT functions in planta in a manner comparable with the recombinant purified protein.
Because we previously observed delayed greening in sty8 sty46, we were interested to see whether deletion of the ACT domain has an effect during etioplast-to-chloroplast transition. WT, sty8 sty46, sty8 sty46/ 35 S::STY46 and sty8 sty46/ 35 S::STY46⌬ACT seedlings were grown in darkness for 6 days and subsequently transferred into light for 4 h to investigate differences during chloroplast differentiation. Greening of the cotyledons was monitored by appearance, and chlorophyll accumulation was determined (Fig. 5, D and E, control). As expected, the greening process was significantly delayed in sty8 sty46. Interestingly, this phenotype could only partially be rescued in sty8 sty46/ 35 S::STY46⌬ACT, as observed by a yellowish appearance as well as a slight reduction in chlorophyll accumulation compared with the WT. sty8 sty46/ 35 S::STY46 plants, in contrast, behaved like the WT.
Further, we analyzed the phenotype of the generated plants under long-day growth conditions for 3 weeks (Fig. 5F). Surprisingly, where sty8 sty46 mutants were smaller compared with the WT, no change in leaf area was observed between the WT, sty46/ 35 S::STY46, and sty8 sty46/ 35 S::STY46⌬ACT.
To investigate the inhibitory effect of SAM and Ile on kinase autophosphorylation in the greening process, WT, sty8 sty46, sty46/ 35 S::STY46, and sty8 sty46/ 35 S::STY46⌬ACT lines were analyzed on MS plates supplemented with 8 M SAM and 100 M Ile (18,19) (Fig. 5D). Interestingly, the WT and sty8 sty46/ 35 S::STY46 both showed reduced chlorophyll accumulation compared with the untreated control, indicating an inhibitory effect of excess SAM and Ile on chloroplast differentiation. However, sty8 sty46/ 35 S::STY46⌬ACT plants showed no reaction to SAM and Ile treatment compared with the untreated plants (Fig. 5E). We therefore propose that the observed negative effect caused by SAM and Ile treatment in the WT is indeed due to kinase inhibition via the ACT domain. The sty8 sty46 mutant also showed a reduction in chlorophyll accumulation upon SAM and Ile treatment, possibly because of STY17 inhibition.

Discussion
Over the years, ACT domains have gained increased recognition as allosteric regulators of various enzymes. Noteworthy, in this study, regulation of a protein kinase by an ACT domain was investigated for the first time. The ACT domain has been suggested to function as an evolutionarily conserved module to mediate various allosteric responses, an idea that is supported by our findings (8,11). Because ACT domains are not very highly conserved on a primary sequence level, they proved to be difficult to identify and classify by simple PSI_BLAST searches. Although structural predictions allow more precise annotation, functional characterization, including their binding of small molecules as well as their regulatory effect on other catalytic domains, was suggested in addition to in silico sequence and structural analysis (11). Indeed, we could provide evidence that the ACT domain in the STY kinases meets these criteria because it binds to typical small molecules, which results in a direct effect on kinase activity. In addition, we could show that the ACT domain mediates oligomerization of STY8, yet another typical feature of the respective domain. We therefore conclude that the investigated domain represents a bona fide member of the ACT domain family.
SAM and Ile were identified as binding partners of the ACT domain in this study and were shown to reduce kinase activity as well as affinity to the substrate pSSU. Interestingly, the precursor of SAM, methionine, as well as Ile are synthesized in the chloroplast. This may indicate feedback regulation of the requirements of the chloroplast and the cytosolic precursorphosphorylating kinases. The exact physiological events and signals, however, remain to be determined. Moreover, we could show that Ile requires a conserved glycine in the first loop region of the ACT domain. This is in contrast to SAM, which apparently utilizes a second binding site within the ACT domain. We hypothesize that binding of SAM and Ile leads to conformational changes rendering an inactive structural state of the kinase. Such allosteric regulations have been observed previously; for example, in 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase, which catalyzes the first step of the shikimate pathway. Upon inhibition by tyrosine, 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase adopts a closed inactive conformation. Although several structures of ACT domains have been obtained, a comparison between structures with and without ligand that would help to unravel conformational switches is often lacking.
The scheme in Fig. 6 summarizes the proposed regulatory role of the ACT domain in kinase activity by autophosphorylation and its consequential role in chloroplast development. We assume that, in WT kinase, modest autophosphorylation takes place, which, in turn, is reflected by modest phosphorylation of precursors. As we have observed previously that greening of young seedlings is delayed in STY kinase mutants, we suppose that phosphorylation of precursors positively influences import efficiency, possibly by stabilizing the precursor and/or enhancing its affinity for the chloroplast translocon. Nevertheless, we also observed that precursor dephosphorylation prior to translocation is vital to ensure efficient import. In this study, we could show that deletion of the ACT domain results in hyperphosphorylation as well as slower greening in planta compared with the WT. According to our previous observations, we argue that the enhanced phosphorylation cannot be compensated by an equivalent activity of the responsible phosphatase, whose nature is still elusive. In turn, hyperphosphorylation caused by lack of the ACT domain leads to slower chloroplast-toetioplast transition (Fig. 6). In line with this, we have shown previously, using precursor phosphomimicking mutants, that constant phosphorylation in the transit peptide likewise results in hampered import and chloroplast defects (5). Apparently, enhanced transit peptide phosphorylation is especially important in cotyledons because mature sty8 sty46/ 35 S::STY46⌬ACT plants are undistinguishable from the WT. This observation also matches our previous results, where we could show that expression of a phosphomimicking variant of the essential Photosystem II assembly factor HCF136 only partially rescued the seedling lethal hcf136 phenotype. Intriguingly, reduced import of HCF136 had a significant effect in cotyledons, whereas mature leaves developed similar to the WT (6). Further, application of SAM and Ile during the greening experiment reduced etioplast-tochloroplast transition in the WT and, consequently, in sty46/ 35 S::STY46 (Fig. 6). sty46/ 35 S::STY46⌬ACT, however, was not affected by SAM and Ile compared with untreated plants, demonstrating that SAM and Ile not only affect . WT STY8 sensitivity to Ile is increased in a concentration-dependent manner in STY8 but not in STY8⌬ACT. A CBB gel shows equal loading of proteins. C, Ile binds to STY8 via the ACT domain. Direct binding of Ile to STY8 and STY8⌬ACT was assessed by MST. The kinases were labeled with Tris-NTA dye. The concentration of the labeled STY8 and its variant was held constant at 25 nM, whereas Ile was titrated in 2-fold serial dilutions against it, and the thermophoretic mobility was monitored. Binding was observed between STY8 and Ile with a K D of 311.4 Ϯ 16 M. No Ile binding was observed using STY8⌬ACT. D, in vitro kinase assay comparing autophosphorylation of recombinant STY8 WT or STY8⌬ACT in the presence of increasing concentrations of SAM (1, 2, and 5 mM). WT STY8 sensitivity to SAM increased in a concentration-dependent manner. A CBB gel shows equal loading of proteins. E, direct binding of SAM to purified STY8 and STY8⌬ACT was assessed by MST as in C. Binding was observed between WT STY8 and SAM with a K D of ϭ 774.7 Ϯ 39 M. No SAM binding was observed using STY8⌬ACT. F, SAM, Ile, and methionine binding to the STY8 ACT domain by MST. Binding was observed between STY8-ACT and SAM (47.28 Ϯ 25 nM) as well as Ile (81.48 Ϯ 34 M). No binding constant could be calculated for methionine. G, SAM and Ile influence pSSU binding to STY8. MST with STY8 and pSSU was performed in the presence of Ile and/or SAM. pSSU was labeled with Tris-NTA dye. The final pSSU and SAM and/or Ile was held constant at 25 nM and 1 mM, respectively. STY8 was titrated in 2-fold serial dilutions against it, and thermophoretic mobility was monitored. H, in vitro kinase assay comparing autophosphorylation of recombinant STY8, STY17, STY46 WT or STY8⌬ACT, and STY17⌬ACT or STY46⌬ACT in the presence of Ile or SAM (5 mM). A CBB gel shows equal loading of proteins. For quantification of autophosphorylation, STY8, STY17, or STY46 WT kinase activity without addition of metabolites was set to 100%. Data show mean Ϯ S.E.; n ϭ 3. Asterisks indicate statistical significance compared with the untreated control. * p Ͻ 0.05, ** p Ͻ 0.01, *** p Ͻ 0.001.

Regulation of STY kinases by a metabolite binding ACT domain
kinase activity in vitro but also chloroplast development, likely via regulation of the STY kinases. We hypothesize that SAM and Ile binding both lead to a conformational change in the STY kinase, resulting in its reduced activity.

Plant materials and growth conditions
Arabidopsis thaliana WT Columbia ecotype (Col-0) and the mutants were grown either on soil or half-strength Murashige and Skoog (MS) medium supplied with 1% sucrose under controlled conditions in a growth chamber. For soil-based pheno-typing analysis, plants were grown under long-day conditions (16 h/8 h light/dark, 22°C, 120 microeinsteins m Ϫ2 s Ϫ1 ). For plate-based analysis, dry seeds were surface-sterilized. Seeds were vernalized at 4°C for 2 days. Mutant plants, SALK 072890 (sty8), and SALK 116340 (sty46) lines have been described previously (7). For generation of complemented sty8 sty46 mutants, WT STY46 cDNA as well as the mutated cDNAs (STY46⌬ACT) were cloned into the binary vector pB7FWG2 with a stop codon (plant systems biology using the Gateway system, Life Technologies). The constructs were introduced into Agrobacterium tumefaciens strain GV3101, and the ho-

Regulation of STY kinases by a metabolite binding ACT domain
mozygous sty8 sty46 mutant plants were transformed by floral dip (20).

Regulation of STY kinases by a metabolite binding ACT domain
fuged for 30 min at 20,000 ϫ g. After centrifugation, the supernatant was incubated with Ni-Sepharose fast flow (GE Healthcare) for 1 h at 4°C. The Sepharose was washed twice with washing buffer (50 mM Tris-HCl, 150 mM NaCl (pH 7.4), and 10 mM imidazole), and recombinant proteins were eluted by increasing the imidazole concentration to 500 mM.

In vitro kinase assays
Recombinant kinase or stroma extract was incubated with recombinant substrate in the presence of 3 Ci of [␥-32 P]ATP and 2.5 M ATP in a total volume of 25 l of kinase buffer (20 mm Tris-HCl (pH 7.5), 5 mM MgCl 2 , and 0.5 mM MnCl 2 ). The reaction was carried out for 15 min at 23°C and stopped by adding 5 l of SDS sample buffer. The proteins were separated on a 12% SDS-polyacrylamide gel followed by autoradiography. Gels were exposed overnight.

MST analysis
MST was performed with purified recombinant protein and different ligands. The proteins were diluted in MST buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 10 mM MgCl 2 , and 0.05% Tween 20). Labeling of proteins was performed using Monolith His Tag Labeling Kit RED-Tris-NTA (NanoTemper, Munich, Germany). Tris-NTA dyes were diluted in PBST buffer (137 mM NaCl, 2.5 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 (pH 7.4), and 0.05% Tween 20) to a final concentration of 100 nM. Increasing concentrations of nonlabeled protein (0.15 nM to 5 M) or metabolites (25 nM to 1 mM) were titrated against the indicated concentrations of labeled proteins and centrifuged for 5 min at 13,000 ϫ g. The MST analysis was performed in Monolith NT.115 glass capillaries using the Monolith NT.115 Red/Green device (NanoTemper) with 20% light-emitting diodes power and 40% MST power. The overall affinity (K D ) was determined using NanoTemper analysis software.

Relative molecular mass estimation by size exclusion chromatography
Proteins were analyzed on a Superdex 200 Increase 3.2/300 SEC chromatography column (GE Healthcare) in 50 mM Tris-HCl and 150 mM NaCl (pH 7.4). The column was calibrated using 20 g of gel filtration standards (thyroglobulin, M r 669,000, void volume; ferritin, M r 440,000; aldolase, M r 158,000; conalbumin, M r 75,000; ovalbumin M r 44,000) (GE Healthcare). The relative molecular masses of the peaks obtained were used to generate a calibration curve using a logarithmic interpolation.

Analysis of proteins by PAGE and immunoblotting
Purified protein samples were separated on 12% SDSpolyacrylamide gels. To analyze oligomerization, the recombinant protein was analyzed by 5%-10% gradient PAGE under nondenaturing conditions, lacking SDS or reducing agents, and using 750 mM 6-aminocaproic acid and 5% (w/v) Coomassie G 250 as loading buffer. NativeMarker TM was obtained from Life Technologies.

Chlorophyll extraction
Chlorophyll determination of Arabidopsis leaves was performed following a method described previously (24). A total of 250 mg of leaf tissue was harvested and incubated in 2 ml of dimethylformamide for 2 h in the dark. Absorbance was measured at 663, 750, and 645 nm. Chlorophyll concentration was calculated as described previously (25).