Magnesium chelatase (Mg-chelatase) inserts magnesium into protoporphyrin during the biosynthesis of chlorophyll and bacteriochlorophyll. Enzyme activity is reconstituted by forming two separate preactivated complexes consisting of a GUN4/ChlH/protoporphyrin IX substrate complex and a ChlI/ChlD enzyme ‘motor’ complex. Formation of the ChlI/ChlD complex in both Chlamydomonas reinhardtii and Oryza sativa is accompanied by phosphorylation of ChlD by ChlI, but the orthologous protein complex from Rhodobacter capsulatus, BchI/BchD, gives no detectable phosphorylation of BchD. Phosphorylation produces a 1-N-phospho-histidine within ChlD. Proteomic analysis indicates that phosphorylation occurs at a conserved His residue in the C-terminal integrin I domain of ChlD. Comparative analysis of the ChlD phosphorylation with enzyme activities of various ChlI/ChlD complexes correlates the phosphorylation by ChlI2 with stimulation of Mg-chelatase activity. Mutation of the H641 of CrChlD to E641 prevents both phosphorylation and stimulation of Mg-chelatase activity, confirming that phosphorylation at H641 stimulates Mg-chelatase. The properties of ChlI2 compared with ChlI1 of Chlamydomonas and with ChlI of Oryza, shows that ChlI2 has a regulatory role in Chlamydomonas.
Magnesium chelatase (Mg-chelatase) catalyzes the first committed step of the chlorophyll biosynthesis pathway by insertion of magnesium into protoporphyrin IX (PPIX). It is also the most complex step of chlorophyll synthesis, employing at least three protein subunits in an ATP-dependent reaction. In anoxygenic bacteriochlorophyll-synthesizing organisms, the protein subunits are known as BchI, BchD, and BchH . In oxygenic chlorophyll-synthesizing organisms, the orthologous proteins are ChlI, ChlD, and ChlH, with an additional accessory protein, GUN4, required for optimal activity [2–6]. BchH/ChlH is the porphyrin-binding subunit and is most likely to have the active site for chelation [7,8]. ChlI is a member of the AAA+ superfamily and has ATPase activity [9,10]. The ChlI subunit in Arabidopsis thaliana, also called CH42 and ChlI1, is a target for thioredoxin and its ATPase activity is dependent on the redox state of the chloroplast . Eudicots, such as A. thaliana, and some algae, like Chlamydomonas reinhardtii, have more than one ChlI subunit gene, whereas in most monocotyledons there appears to be only a single ChlI gene. The ChlI2 of A. thaliana is also a target of thioredoxin and can partly substitute for the ChlI1 [12,13] in CH-42 mutants of A. thaliana. However, ChlI2 from C. reinhardtii cannot substitute for ChlI1 in vivo , indicating that a different function for ChlI2 exists in this organism. The larger ChlD subunit also has an AAA+ domain at the N-terminal and a C-terminal integrin I domain containing a divalent metal ion-dependent adhesion site motif . The AAA+ domain of ChlD, sensor II arginine, and a polyproline region are required for interaction with ChlI in Synechocystis, and inclusion of the C-terminal integrin I domain is still required for enzymatic activity . EM studies revealed that BchI and BchD form a stable ‘motor’ complex of two stacked hexameric rings in the presence of adenosine nucleotides .
Although the precise mechanism of magnesium chelation by Mg-chelatase remains unclear, the insertion appears to involve ATP hydrolysis on the ChlI:ChlD motor complex which drives a conformational change in the ChlH, which has both the porphyrin and magnesium bound [17,18]. A protein phosphate intermediate has been postulated to occur during catalysis based on isotope exchange kinetic data , and exchange of the ChlI/BchI subunit in the motor complex occurs between each catalytic cycle . GUN4 and ChlI subunits have been identified in phosphoproteome studies of A. thaliana and phosphorylation of GUN4 affects Mg-chelatase activity .
In investigating enzymes responsible for phosphorylating GUN4 and ChlI in Chlamydomonas and rice, we discovered that ChlD was also phosphorylated. Purification of this kinase activity from Chlamydomonas identified the ChlI1 and ChlI2 subunits as being responsible for this phosphorylation, and this was confirmed using the purified recombinant subunits. The orthologous subunits from rice also phosphorylate the corresponding ChlD subunit, but the orthologous Rhodobacter capsulatus BchI does not phosphorylate the BchD, suggesting that this activity may be restricted to oxygenic or eukaryotic ChlIs. Comparison of the activities from rice and Chlamydomonas suggests that ChlI2 may have diverged from ChlI1 to regulate the activity of Mg-chelatase.
Materials and methods
Oryza sativa ChlH, ChlD, ChlI, and GUN4 as well as C. reinhardtii ChlH, ChlD, ChlI1, ChlI2, and GUN4
All the recombinant proteins were cloned from cDNA of O. sativa L. Japonica nipponbare  and C. reinhardtii cc124 into expression vector pET 28a and/or pET14b (Merck–Novagen, Darmstadt, Germany) using standard cloning protocols . The Chlamydomonas proteins are differentiated from the O. sativa proteins by the prefixes ‘Cr’ and ‘Os’, respectively. A H641E mutant of the Chlamydomonas ChlD, CrChlD-H641E, was constructed by purchase of a 600 bp G-block containing the H to E mutation from Integrated DNA Technologies. The G-block was subcloned between the MluI and a PstI site in a wild-type (WT) clone of the ChlD and then sequence-verified. This was then subcloned into pET28a.
Protein expression and purification
The proteins were expressed in Escherichia coli strains BL21(DE3) grown at 37°C in 2× YT medium supplemented with 50 mg/l kanamycin (for pET28b) or 100 mg/l ampicillin (for pET14b) as described previously [23,24]. His6-tagged OsChlH, OsChlI, OsChlD, and OsGUN4 proteins were purified as described previously . His6-tagged CrChlI1 and CrChlI2 proteins were purified as for the Oryza proteins, whereas all other proteins were purified as described previously [23,24]. All the proteins apart from OsChlD were desalted into exchange buffer [10 mM Tricine–NaOH (pH 8.0), 10% (w/v) glycerol, 2 mM EDTA, and 2 mM dithiothreitol (DTT)]. OsChlD was desalted into 6 M urea, 10 mM Tricine–NaOH (pH 8.0), 10% (w/v) glycerol, 2 mM EDTA, and 2 mM DTT. R. capsulatus BchI, BchD, and BchH proteins were expressed and purified as described previously . The OsChlD and BchD proteins were refolded by rapid dilution into assay buffer on ice as described previously [22,25].
Protein determinations were performed with Bio-Rad protein assay reagent (Bio-Rad Laboratories Pty Ltd, Hercules, CA, U.S.A.) according to the manufacturer's instructions, with BSA as a standard.
Mg-chelatase enzyme assay
Assays for the various types of Mg-chelatase were performed as previously described [22–25] with the following modifications. The final concentrations of assay components were 50 mM Tricine–NaOH (pH 8.0), 15 mM MgCl2, 2 mM DTT, 2 mM ATP (assay buffer), and 500 nM PPIX. Protein concentrations are given in the figure legends but in general consisted of 25 nM ChlD/BchD, 200 nM ChlI/BchI, 500 nM ChlH/BchH, and 500 nM GUN4.
Assays were based on a modified version of the method of Bayer et al. . Assays were conducted at 22°C for 20 min in a final volume of 10 µl with the following final concentrations: 20 mM Tricine–NaOH (pH 8.0), 10% glycerol, 5 mM MgCl2, ∼100 nM γ-[32P]-ATP (3000 Ci/mmol), 0.8 µg of ChlD/BchD, and variable amounts of ChlI/BchI. Variations to this standard assay are stated in the figure legends. Assays to be run on gels were stopped with the addition of 4 µl of 4× LDS loading dye, including 5 mM EDTA, followed by heating to 95°C for 2 min; a 10 µl aliquot was loaded onto a 4–20% Tris–glycine gel (Bio-Rad or Nu-Sep) and electrophoresed at 200 V for 60 min. The gel was exposed to a Storage Phosphor Screen (Bio-Rad) which was analyzed with a Typhoon Trio Variable mode scanner at 633 nm at 25, 100, or 200 µm. After exposure to the phosphor screen, the gel was stained with the Coomassie brilliant blue dye to visualize the proteins. In analyses of phosphate stability, the phosphorylated ChlD proteins were immediately blotted onto a PVDF membrane using a Bio-Rad semidry blotter at 0.8 mA/cm2 for 45 min using Towbin buffer (pH 8.0) . The membrane was exposed to storage phosphor, then incubated in 2 M KOH for 2 h at 55 °C, washed with 10 mM Tris–HCl (pH 8.0) and re-exposed to storage phosphor, and incubated in 1 M HCl for 2 h at 55°C and re-exposed to storage phosphor .
Phosphoamino acid separation
For both quantification of phosphorylation and amino acid analysis, the assays were stopped by adding 25 mM NaOH and 4 volumes of acetone, and centrifuged at 18 000 g for 2 min. The precipitated protein was washed three times with 0.2 ml of 10 mM Tricine–NaOH (pH 8.0) in 20% (v/v) ethanol and air-dried. This precipitated protein was either solubilized directly in 10% SDS and counted or solubilized in 3 M KOH and hydrolyzed at 100°C for 3 h. The KOH-digested protein was separated in 50 mM triethylamine–CO2 buffer at a pH of 8.25 on Sephadex G10 to remove the KOH. Analysis and synthesis of phosphoamino acids were performed as described previously [29–31], with a few modifications as described below and in figure legends. OPA amino acid analysis was performed on an Agilent UHPLC system with automated precolumn OPA derivatization, separation in a ZORBAX Eclipse Plus C18 rapid resolution HD 2.1 mm × 50 mm 1.8 µm column with diode array and fluorescence detection. The flow rate was 0.5 mL/min with a linear gradient from Buffer A, 0.3% (v/v) tetrahyrdofuran in 7.5 mM sodium phosphate (pH 7.3), to 10% Buffer B, 55% acetonitrile in 7.5 mM sodium phosphate (pH 7.3) at 5 min, and then with a linear gradient to 100% Buffer B at 10 min. Standard amino acid and phosphoamino acid elution times were as follows: 1-P-His, 0.31 min; 3-P-His, 1.31 min; Asp, 1.56 min; Glu, 2.81 min; P-Lys, 3.46 min; Ser, 6.15 min; His, 6.67 min; Ala, 7.77 min; Met, 9.08 min; Lys, 10.46 min. Anion exchange chromatography on a POROS HQ 4.6 mm × 100 mm column was performed at 1.5 ml/min with a gradient from 8 mM potassium bicarbonate to 0.16 M potassium bicarbonate over 5 min and then with a linear gradient to 1.6 M potassium bicarbonate from 5 to 10 min. Inorganic phosphate eluted at 3–3.5 min, and 1-N-phospho-histidine and 3-N-phospho-histidine at 6–7 min. Anion exchange chromatography was also performed on a 2 mm × 50 mm DOWEX 2-X8 column with 1 ml step elution fractions of 0.2 M KHCO3, 0.3 M KHCO3, 2× 0.4 M KHCO3, 0.5 M KHCO3, and 5× 0.8 M KHCO3. Inorganic phosphate eluted in the first 0.4 M KHCO3 fraction. 1-N-phospho-histidine and 3-N-phospho-histidine eluted together in the first two 0.8 M KHCO3 fractions. The position of phospho-His elution on POROS HQ and DOWEX 2-X8 was confirmed by chromatography of 1-N-phospho-His and 3-N-phospho-His standards. The identity of the standards eluting from the ion exchange columns was further confirmed by OPA derivatization and UHPLC analysis of the phosphorylated histidine directly and also of the histidine released after acid hydrolysis in 0.1 M HCl for 30 min at room temperature.
Phosphorylated and non-phosphorylated CrChlD was precipitated as described above and the precipitate was digested with 20 ng of sequencing-grade trypsin at 22°C for 30 min and then 4°C overnight in 0.1 M ammonium formate at a pH of 8.5. The digest was centrifuged at 22 000 g for 10 min. Samples were analyzed directly using a nano-LC–MS/MS coupled to an AB SCIEX QSTAR Elite mass spectrometer (ABSciex, U.S.A.) with positive nanoflow electrospray analysis and an information-dependent acquisition (IDA) mode. MS resolution was set to 120 000. LC gradient conditions were a 30 min gradient from 0.1% formic acid to 90% acetonitrile in 0.1% formic acid at 20°C. In IDA, MS/MS acquisitions of the 20 most intense m/z values, exceeding a threshold >150 counts per second with charge states between 2+ to 5+, were selected for MS/MS analysis following a full MS survey scan and excluded for 20 s to minimize redundant precursor sampling. The raw data files were converted into mzxml and searched using XTandem  against a complete C. reinhardtii protein database that also contained the recombinant protein sequences. Analysis of the mass spectrometry data was also performed using MZmine 2.
Phosphorylation of ChlD by ChlI
Purification of this kinase activity from Chlamydomonas identified the ChlI1 and ChlI2 subunits as being responsible for this phosphorylation (Supplementary Figure S1), and this was confirmed using the purified recombinant subunits (Figure 1 and Supplementary Figure S1). Comparisons of the effectiveness of ChlI/BchI subunits from C. reinhardtii, O. sativa and R. capsulatus, in phosphorylating their cognate ChlD/BchD subunits are shown in Figure 1A. Although the orthologous subunits from O. sativa could phosphorylate its corresponding ChlD subunit, the orthologous R. capsulatus BchI subunit does not phosphorylate BchD (Figure 1A), suggesting that this activity may be restricted to oxygenic or eukaryotic ChlIs. It is clear that CrChlI2 is the most efficient at phosphorylating the CrChlD with a significantly greater extent of phosphorylation activity than the CrChlI1 subunit (Figure 1B and Supplementary Figure S2). The phosphorylation is also strongly dependent on the ratio of ChlI to the cognate substrate ChlD for CrChlI2 (Figure 1D) and OsChlI (Figure 1E) but not for CrChlI1 (Figure 1C).
ChlI phosphorylates ChlD.
Phosphorylation at 1-
The phosphorylation is base stable and acid labile as shown in Figure 2A, indicating that the amino acid phosphorylated is His, Lys, or Arg [28,33]. Chromatography of an aliquot of the hydrolyzed [32P]-phosphorylated CrChlD (pChlD) on a DOWEX 2-X8 column 4 mm × 5 cm with sequential step elution separated the inorganic phosphate from the phosphorylated amino acid, with the radioactivity in fractions 6 and 7 corresponding to elution of phospho-His (Figure 3A). An aliquot of the hydrolyzed phospho-CrChlD sample was also chromatographed on POROS HQ10. The second peak of radioactivity co-chromatographed with authentic 1-N-phospho-His and 3-N-phospho-His that eluted together at 6–7 min (Figure 3B). UHPLC analysis of an aliquot of the hydrolyzed pChlD with precolumn OPA derivatization resulted in all of the radiolabeled amino acid eluting with 1-N-phospho-histidine with ∼60–65% phosphorylation of a single histidine based on relative fluorescence peak areas of OPA derivatives (Figure 3C). All of the radioactivity injected onto these columns was accounted for in the fractions shown in Figure 3A–C. Thus, the phosphorylation occurs at the 1-N position of histidine to produce a 1-N-phospho-histidine within ChlD.
N-phospho-histidine of CrChlD at H641.
Chromatographic evidence for a 1-
N-phospho-histidine released in a total KOH hydrolysate of 32P-phosphorylated CrChlD.
CrChlD and phospho-CrChlD, obtained from CrChlI2 kinase assay, were digested with sequencing-grade trypsin and analyzed by nanoLC-MS/MS to investigate the phosphorylation site. Phospho-histidine-containing peptides are difficult to analyze by MS/MS as the phosphate can be lost during the acidic LC run. Even when molecular ions of phospho-His-containing peptides are observed, the phosphate is lost as a neutral loss of phosphate (79.9663 m/z), phosphate and water (97.9769 m/z), and/or phosphate and two waters (115.9875 m/z) during CID fragmentation [34,35]. There are seven histidine residues in the native CrChlD with six of these His residues: H131, H215, H255, H275, H359, and H641 in C. reinhardtii native sequence numbering, conserved between Synechocystis PCC6803, O. sativa, and C. reinhardtii (Supplementary Figure S3). There were 2822 potential MSMS spectra identified as CrChlD peptide ions by nanoLC-MSMS of phosphorylated and unphosphorylated CrChlD using a maximum mass error of 0.005 for the parental ion and a logE cutoff of −1 for MSMS analysis. From these spectra, putative parental phospho-peptide ion MSMS spectra were only identified for peptides containing H131, H255, and H641, with five, three, and seven different CID peptide ion spectra identified, respectively. No putative phospho-peptide ion spectra were obtained in the unphosphorylated CrChlD sample or for the other histidine-containing peptides from the pChlD. High-resolution parental ion reconstructed chromatograms of either M+2H+ or M+3H+ peptide ions and their phosphorylated derivatives are shown in Figure 2. The H641-peptide ion, LDSLPCGGGSPLAHGLSTAVR, was over an order of magnitude reduced in intensity in the pChlD sample. There was a concomitant appearance of a parental ion for the phosphorylated H641 peptide in the pChlD sample, which was not detected in the ChlD sample. In comparison, the H131- and H255-containing peptide ions were similar in intensity between CrChlD and pChlD. Inspection of the individual CID spectra of the pH641-peptide parental ion confirmed the identity of this parental ion as phosphorylated LDSLPCGGGSPLAHGLSTAVR in the pChlD sample, as the majority of y ions and many b ions were identified with neutral loss of phosphate or phosphate and water from all relevant fragment ions as expected (Supplementary Figure S5).
Aggregation of CrChlI1 and stabilization by ChlI2
It was noted that when both CrChlI1 and OsChlI were stored at 4°C for several days, they would aggregate and become inactive. The aggregation was lower when glycerol was present and did not occur if the samples were snap-frozen in small aliquots at −80°C. If ATP was added to the CrChlI1 and OsChlI, the proteins aggregated more rapidly with visible aggregation of CrChlI1 often observed after several hours on ice, especially at protein concentrations above 8 mg/ml. Size-exclusion chromatography (SEC) of CrChlI1 and OsChlI in the presence of ATP gives a size shift from an apparent molecular mass of 60 kDa to a molecular mass of the main peak at ∼200 kDa as has been reported previously for BchI of R. capsulatus [36,37] and ChlI of Synechocystis, and is used during purification to remove contaminants of similar monomer molecular mass (Supplementary Figure S1). In contrast, the CrChlI2 did not change its mobility on SEC in the presence of ATP (Supplementary Figure S1). Additionally, the ATP-induced CrChlI1 multimers precipitated as large aggregates with almost no soluble protein remaining when stored for extended periods at 4°C (see Supplementary Figure S2). The ATP-dependent precipitation of ChlI1 occurred within hours at higher protein concentrations, >10 mg/ml, and also occurred without the addition of ATP if CrChlI1 was concentrated above ∼20 mg/ml. However, when CrChlI2 was mixed with CrChlI1 in a 1 : 5 molar ratio, the ATP-dependent CrChlI1 multimer remained stable, soluble, and active in Mg-chelatase assays even after 6 days of storage at 4°C.
The relative activities of the various ChlI subunits paired with their cognate ChlD subunits in Mg-chelatase assays are shown in Figure 4A. The O. sativa Mg-chelatase was more active than the best combination of C. reinhardtii ChlI subunits. No Mg-chelatase activity was detectable when CrChlI2 was used in the absence of CrChlI1, and CrChlI1 had lower activity in the absence of ChlI2 under these stopped-assay conditions. However, CrChlI2 stimulated the Mg-chelatase activity more than 10-fold in the presence of CrChlI1. The N-terminal His-tag on CrChlI1 had a significant impact on Mg-chelatase activity with a 50% reduction in the CrChlI2-stimulated activity compared with CrChlI1 without the His-Tag. In comparison, the His-tag on CrChlI2 did not have any effect on its ability to stimulate Mg-chelatase activity in the presence of either CrChlI1 or His-tagged CrChlI1.
Stimulation of magnesium chelatase activity by ChlI2.
A more detailed kinetic analysis of the stimulation of Mg-chelatase activity by CrChlI2 was conducted using continuous assays. Enzyme activity for Mg-chelatase has previously been reported as per nmole or µgram of ChlD or BchD as the ChlI/BchI and ChlH/BchH subunits display saturation kinetics, suggesting that they act as substrates for the ‘ChlD/BchD’ enzyme [25,38]. Activity increases linearly with increasing concentration of BchD up to 30 nM, while the ChlD of Synechocystis displays this linearity up to ∼10 nM after which ChlH appeared to be limiting . The CrChlD behaves differently from both of these with apparent saturation-type kinetics (Figure 4B), and the CrChlI2 stimulated the Mg-chelatase activity at all CrChlD concentrations. This activation of activity by CrChlI2 is titratable against CrChlD at all concentrations tested (Figure 4C,D). The midpoint of the titration activity stimulation with CrChlI2, at every CrChlD concentration, is 36% of the corresponding CrChlD concentration (Figure 4D), suggesting that a slightly less than 1 : 1 CrChlI2 : CrChlD stoichiometry is required for optimal activity.
CrChlD-H641E mutant characterization
The CrChlD-H641E mutant Mg-chelatase activity was significantly lower than the –WT in assays with CrChlD, P < 0.0001, as shown in Figure 5A. Significantly, the Mg-chelatase activity of CrChlD-H641E was not stimulated by CrChlI2, while CrChlI2 clearly stimulates the Mg-chelatase in assays with WT CrChlD.
Comparison of CrChlD-H641E with CrChlD.
It has been unclear what the role of additional ChlI subunit genes for Mg-chelatase in plants and algae is and if each organism with additional genes uses them in the same way. Here, we see that, in C. reinhardtii, ChlI2 is capable of phosphorylating the ChlD subunit on the integrin-I domain and stimulates the enzyme activity. Given that similar ChlI2 : ChlD ratios are required for both stimulation of activity and phosphorylation, it seems most likely that phosphorylation of ChlD and stimulation of Mg-chelatase activity are linked. We have confirmed that the phosphorylation and stimulation of Mg-chelatase activity are linked by characterizing an H641E mutant of CrChlD as shown in Figure 5. The CrChlD-H641E mutant is not stimulated by CrChlI2 and is also unable to be phosphorylated, which confirms that H641 is required both for phosphorylation and for stimulation of Mg-chelatase activity by CrChlI2.
The orthologous subunit BchI from anoxygenic bacteria does not have any detectable kinase activity for its BchD subunit. Thus, the phosphorylation of ChlD may be restricted to oxygenic organisms in regulating Mg-chelatase. As the stimulation of activity by CrChlI2 appears to be greatest at low CrChlD concentrations (Figure 4C), the modulation of activity may be more important when the CrChlD expression is low, for example during light to dark transitions .
The presence of multiple chlI subunit genes found in dicots and algae raises the question of the role of these multiple subunits in chlorophyll synthesis. Investigation into the role of the ChlI2 of A. thaliana shows that it can only partially substitute for the ChlI1 in chlorophyll synthesis as the absence of the chlI1 gene gives a pale green phenotype and prevents seedlings from greening effectively [12,13,40,41]. In contrast, the C. reinhardtii ChlI2 subunit cannot substitute for CrChlI1 even when CrChlI2 is overexpressed , and it was suggested that CrChlI2 acts as a ‘surrogate protein’ to complete a hexameric ring structure, mainly consisting of CrChlI1 units, and that this lowers Mg-chelatase activity to attenuate chlorophyll synthesis . While CrChlI2 may act as a surrogate protein, it does not lower enzyme activity as suggested, but in fact does the opposite and stimulates the Mg-chelatase activity (Figure 4). ChlI1 from both Chlamydomonas and rice also possess kinase activity, although the activity is more limited as both the extent and/or rate of phosphorylation is much lower than for CrChlI2 (see Supplementary Figure S2 and Figure 2). This may be due to ChlI1-forming ATP-dependent oligomers, which bury the ATPase active site at the interface between subunits . The ability of CrChlI2 to prevent aggregation of CrChlI1 suggests that it possibly caps these oligomeric complexes, preventing polymerization. This inhibition of aggregation may also be an important function in vivo where the ratio of CrChlI2 to CrChlI1 is low .
The CrChlI2 has a 43-amino acid C-terminal extension which is highly charged, compared with other ChlIs, and phylogenetic analysis reveals that CrChlI2 appears to be restricted to unicellular algae, with this extension being a characteristic feature . It is possible that this charged domain interacts with the charged domain on CrChlD to better direct CrChlI2 to the C-terminal domain.
The phosphorylation of ChlD by ChlI2 and ChlI1, and the fact that ChlI1 is a target for thioredoxin  indicate a complex interplay between phosphorylation and redox regulation of chlorophyll synthesis, which parallels similar regulatory mechanisms observed in the regulation of photosynthesis .
A.S. and K.K. purified the kinase activity from Chlamydomonas and did the kinase assays. A.S. and S.Z. expressed and purified the recombinant proteins. A.S. and R.D.W. performed the Mg-chelatase assays. R.D.W. conducted the phospho-His analysis and MS analysis. M.L. and S.Z. produced the Cr and Os protein expression clones.
This work was supported by the National Natural Science Foundation of China [Grant No. 30971748] and by a Macquarie University SNS grant.
The Authors declare that there are no competing interests associated with the manuscript.