Catalytic Turnover Triggers Exchange of Subunits of the Magnesium Chelatase AAA+ Motor Unit*

Background: Magnesium chelatase is an AAA+ protein complex involved in chlorophyll biosynthesis. Results: An exchange of subunits occurs during the catalytic cycle. Conclusion: Dissociation of the complex may be part of the reaction mechanism. Significance: Deciphering of the mechanism of AAA+ protein complexes is crucial for our understanding of the catalytic cycle of a common class of molecular machines. The ATP-dependent insertion of Mg2+ into protoporphyrin IX is the first committed step in the chlorophyll biosynthetic pathway. The reaction is catalyzed by magnesium chelatase, which consists of three gene products: BchI, BchD, and BchH. The BchI and BchD subunits belong to the family of AAA+ proteins (ATPases associated with various cellular activities) and form a two-ring complex with six BchI subunits in one layer and six BchD subunits in the other layer. This BchID complex is a two-layered trimer of dimers with the ATP binding site located at the interface between two neighboring BchI subunits. ATP hydrolysis by the BchID motor unit fuels the insertion of Mg2+ into the porphyrin by the BchH subunit. In the present study, we explored mutations that were originally identified in semidominant barley (Hordeum vulgare L.) mutants. The resulting recombinant BchI proteins have marginal ATPase activity and cannot contribute to magnesium chelatase activity although they apparently form structurally correct complexes with BchD. Mixing experiments with modified and wild-type BchI in various combinations showed that an exchange of BchI subunits in magnesium chelatase occurs during the catalytic cycle, which indicates that dissociation of the complex may be part of the reaction mechanism related to product release. Mixing experiments also showed that more than three functional interfaces in the BchI ring structure are required for magnesium chelatase activity.

The ATP-dependent insertion of Mg 2؉ into protoporphyrin IX is the first committed step in the chlorophyll biosynthetic pathway. The reaction is catalyzed by magnesium chelatase, which consists of three gene products: BchI, BchD, and BchH. The BchI and BchD subunits belong to the family of AAA؉ proteins (ATPases associated with various cellular activities) and form a two-ring complex with six BchI subunits in one layer and six BchD subunits in the other layer. This BchID complex is a two-layered trimer of dimers with the ATP binding site located at the interface between two neighboring BchI subunits. ATP hydrolysis by the BchID motor unit fuels the insertion of Mg 2؉ into the porphyrin by the BchH subunit. In the present study, we explored mutations that were originally identified in semidominant barley (Hordeum vulgare L.) mutants. The resulting recombinant BchI proteins have marginal ATPase activity and cannot contribute to magnesium chelatase activity although they apparently form structurally correct complexes with BchD. Mixing experiments with modified and wild-type BchI in various combinations showed that an exchange of BchI subunits in magnesium chelatase occurs during the catalytic cycle, which indicates that dissociation of the complex may be part of the reaction mechanism related to product release. Mixing experiments also showed that more than three functional interfaces in the BchI ring structure are required for magnesium chelatase activity.
Magnesium chelatase is the multisubunit protein complex that conducts the enzymatic catalysis in the first specific step of chlorophyll and bacteriochlorophyll biosynthesis. In this reaction, which requires ATP hydrolysis, Mg 2ϩ is inserted into protoporphyrin IX to form Mg-protoporphyrin IX. Magnesium chelatase requires the presence of three gene products, BchH, BchD, and BchI (often referred to as the H, D, and I subunits), which have molecular masses of 140, 70, and 40 kDa, respectively (1)(2)(3)(4). The H subunit has been shown to bind the porphy-rin substrate with high affinity (3)(4)(5)(6)(7) and has also been predicted to bind the magnesium ion substrate, by analogy with the homologous CobN, the cobalt-binding subunit of cobalt chelatase (8). Based on these observations, the H subunit has been suggested to be the catalytic component of the enzyme (3)(4)(5).
BchI and BchD belong to the structurally conserved family of AAAϩ proteins 2 (ATPases associated with various cellular activities). They form a two-ring structure with six I subunits in one layer and six D subunits in the other layer. Members of this family are known to act as molecular machines that remodel their various target molecules in an ATP-dependent manner (9). In the case of magnesium chelatase, the ID motor unit has the H subunit as a target protein. The I subunit, which contains the characteristic ATP-binding Walker A and Walker B motifs, is the only ATPase-active component of magnesium chelatase (10 -14). The D subunit has an AAAϩ module at its N terminus with distinct homology to BchI (10,15). However, the Walker A and Walker B motifs, which are necessary for ATP hydrolysis, are poorly conserved in the D subunit. Despite this, BchD is still capable of forming oligomeric ring structures, even in the absence of ATP, and has been suggested to function as a platform for the assembly of the ID complex (16). The ID complex has been studied by cryoelectron microscopy (15,17). The complex has a 3-fold symmetry built up from a trimer of homodimers in each ring. Different structures were observed in the presence of ADP, ATP, or a nonhydrolyzable ATP analog AMPPNP, which probably reflect large conformational changes of the complex during the ATP hydrolytic cycle. The BchI ring has three intact ATPase sites within the homodimers when it is in complex with BchD. It is not known how ATP hydrolysis between the subunits is coordinated with functional output. Various mechanistic models have been proposed for other AAAϩ protein family members, including concerted (18,19), sequential or semisequential (20 -22), probabilistic (22), and partly stochastic models (23). We have previously used barley (Hordeum vulgare L.) mutants to study the cooperativity of the I subunits (24,25). The mutant lines Xantha-h.clo161, Xantha-h.clo125 and Xantha-h.clo157 have missense mutations in the barley gene orthologous to Rhodobacter capsulatus bchI and cause the single amino acid exchanges L111F, D207N, and R289K, respectively (25). Recombinant R. capsulatus BchI proteins carrying the amino acid exchanges have a trace of ATPase activity but cannot contribute to magnesium chelatase activity. Magnesium chelatase assays were also performed with mixtures of wild-type BchI and BchI carrying any of the three amino acid exchanges. All mixed assays showed a reduced magnesium chelatase activity compared with assays containing only the wild-type BchI (25). The reduced magnesium chelatase activity correlates with the observation that the mutations Xantha-h.clo161, Xantha-h.clo125, and Xantha-h.clo157 are semidominant, i.e. heterozygous barley mutants carrying both the wild-type and mutant allele have a phenotype with a reduced amount of chlorophyll (24). The structure of BchI has been solved to 2.1 Å resolution (10). It was found that the residue Leu-111 is located on one side of the wedge-shaped BchI structure and that the Asp-207 and Arg-289 residues are located on the opposite side (Fig. 1). Because BchI subunits carrying L111F, D207N, and R289K substitutions are all affected in ATPase activity, it was concluded that the ATPase active site is in the interface between BchI homodimers (25). In the present study, we examined the stability and intactness of the R. capsulatus magnesium chelatase complex. We conclude that an exchange of BchI subunits occurs in magnesium chelatase during the catalytic cycle, which indicates that dissociation of the complex may be part of the reaction mechanism related to product release, specifically, release of the H subunit.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and Purification of BchI Proteins-An expression system of R. capsulatus bchI was previously con-structed in pET3a and not pET15b because an N-terminal His tag added to R. capsulatus BchI resulted in an inactive protein (3). To obtain a one-step purification system, we cloned R. capsulatus bchI in the pET23b expression vector (Novagen), resulting in a recombinant BchI with a C-terminal His tag. The plasmid obtained was verified by DNA sequencing and named pET23bBchI. Plasmid pET23bBchI was used as template in site-directed mutagenesis (the QuikChange method; Stratagene) to construct four different BchI proteins with the modifications L111F, V113G, D207N, and R289K. The resulting plasmids were named pET23bBchI-L111F, pET23bBchI-V113G, pET23bBchI-D207N, and pET23bBchI-R289K. The following primers were used to introduce the mutations: The mismatches introduced are in lowercase letters. The constructs were confirmed by DNA sequencing of the entire gene. Recombinant BchI, with and without the modifications described, were expressed in Escherichia coli BL21pLys. The cells were grown at room temperature, and isopropyl ␤-D-1-thiogalactopyranoside (final concentration 0.5 mM) was added at an A 600 of 0.4. The following day, the cells from 1 liter of culture were harvested by centrifugation at 20,000 ϫ g for 15 min at 4°C and resuspended in 100 ml of binding buffer (20 mM imidazole, 20 mM sodium phosphate, pH 7.4, 500 mM NaCl) prepared from stock solutions (His Buffer Kit; GE Healthcare). The cells were lysed with a cell disrupter (Constant Cell disruption system, Buch & Holm) and aliquoted into eight 12.5-ml portions, one of which was used for purification each time. The protein was loaded onto a HisTrap FF crude column (GE Healthcare) and washed according to the manual using a peristaltic pump at 1 ml/min (Ole Dich, Hvidovre, Denmark). Elution, fractionation, and recording were performed using an ÄKTA FPLC system (GE Healthcare) with a 20-ml isocratic gradient to 100% elution buffer (500 mM imidazole, 20 mM sodium phosphate, pH 7.4, 500 mM NaCl). Purified protein fractions were desalted on NAP-5 columns (GE Healthcare) into 50 mM Tricine-NaOH, pH 8.5, 15 mM Mg-Cl 2 , 5 mM DTT, and 6% glycerol. Protein concentration was determined using Bradford reagent (Bio-Rad) according to the manufacturer's instructions. SDS-PAGE (NuPage 4 -12%; Invitrogen) was performed according to the manufacturer's instructions, with PageBlue Protein staining solution (Fermentas).
Continuous Magnesium Chelatase and ATP Hydrolysis Assays-Assays were performed in 300-l volumes. The protein content was 14 g of BchH, 3.2 g of BchD, and 1.4 g of BchI. The assays also contained 2 mM ATP, 4 M deuteroporphyrin IX (preincubated with BchH), 15 mM MgCl 2, 5 mM DTT, and 50 mM Tricine-NaOH, pH 8.5. For measurement of ATP hydrolysis, the reaction also included MESG (2-amino-6-mercapto-7-methylpurine riboside) and purine nucleoside phosphorylase at concentrations recommended by the manufac- turer (EnzChek Phosphate Assay Kit; Molecular Probes). The two assay systems had no negative effect on each other. All reactions were performed in the dark with a Tecan robot system (Tecan Group) connected to a Magellan M1000 plate reader (Tecan Group) for continuous spectroscopic analysis. Samples were mixed in black 96-well plates with transparent bottoms (BrandTech Scientific). Data points for magnesium chelatase activity were collected every 4th min with top-measured fluorescence (excitation wavelength 408 nm, slit 5 nm, emission wavelength 582 nm, slit 10 nm). ATPase activity was measured simultaneously as an increase in absorbance at 360 nm. The plate reader was adjusted to 22°C, and the plate was shaken between measurements. The plate was removed from the reader at t ϭ 72 min for the addition of extra BchI subunits. Buffer solution was added to control samples to compensate for dilution effects.
Electron Microscopy-Electron microscopy was done as described previously (7). Carbon-coated 400-mesh copper grids (Ax-labs, Vedbaek, Denmark) were first washed for 1 min with a drop of 50 mM Tricine-NaOH, pH 8.5, 15 mM MgCl 2 , 2 mM ATP, 1 mM DTT. Excess liquid was removed by brief contact with filter paper. Protein samples were diluted to 0.01 mg/ml with the wash buffer. A drop of the protein solution was applied to the grid for 1 min, and excess sample was removed. The grid was immediately stained with filtered 2% (w/v) uranyl formate for 30 s, and excess stain was removed. Electron microscopy was performed using a 120-kV Philips CM-10 microscope at 55,000ϫ magnification, giving a sampling size corresponding 4.67 Å/pixel.
Barley Mutants and Quantification of Chlorophyll-Green plants of segregating barley (H. vulgare L.) mutants Xanthah.clo161, Xantha-h.clo125, and Xantha-h.clo157 were crossed with each other in all possible combinations. Yellow seedlings of the resulting F1 generation carried both parental mutations, which were confirmed by DNA sequencing. To determine the amount of chlorophyll, 3 cm of a barley leaf (corresponds to 20 -30 mg) was cut, weighed, and ground in liquid nitrogen. One ml of methanol was added, followed by mixing on a vortex for 1 min. The extract was left for 1 h in the dark and then centrifuged for 5 min to remove debris. Spectra were recorded between 500 and 800 nm (slit width 10 nm) using methanol as blank. The peaks at 646, 663, and 750 nm were recorded. The total amount of chlorophyll (chlorophyll a and chlorophyll b in g/ml) was calculated according to the formula (17.76 ϫ (A 663 Ϫ A 750 ) ϩ 7.34 ϫ (A 646 Ϫ A 750 )) ϫ V total /V sample and normalized to the amount of leaf material used (26).

Mutant-derived BchI and Wild-type BchD Form a Complex-R. capsulatus
BchI proteins with modifications L111F, D207N, and R289K are all affected in the ATP binding site (25). To obtain a BchI protein with a modification outside the ATP binding site, we changed Val-113 to a glycine by site-directed mutagenesis. Val-113 is homologous to Val-92 in the heat shock protein HslU (27). The residue is located in the pore region of the respective AAAϩ complex. In HslU, the V92G modified protein has ϳ80% ATPase activity and can form hexamers but is deficient in substrate recognition (27).
It was previously established that BchI proteins with the modifications L111F, D207N, and R289K are able to interact with wild-type BchI (25). Using immobilized BchD, it was further shown that the modified BchI subunits can associate with BchD (12). In both experiments, the interactions occurred in presence of ATP or ADP but not when AMP was added. To confirm that a structurally comparable ID complex is obtained with modified BchI proteins, BchI-L111F was incubated together with wild-type BchD in the presence of ATP and stained with uranyl formate on a copper grid suitable for electron microscopic analysis. Images revealed clear and defined ring-shaped particles with a diameter of ϳ130 Å (Fig. 2), which is the expected size of an ID complex (15). Thus, BchI-L111F forms a complex with BchD that is similar to that of wild-type BchI in combination with BchD. It should therefore be possible to use BchI-L111F for conclusive analyses concerning the stability and intactness of the magnesium chelatase complex during the catalytic cycle. Because BchI-D207N and BchI-R289K had behavior similar to that of BchI-L111F in our previous experiments (12,25), and from the ability of HslU-V92G to form hexamers (27), we assume that BchI with the modifications D207N, R289K, and V113G would also be suitable for these kinds of studies.
Internal Regulation of ATPase Activity-ATP hydrolysis plays a central role in the magnesium chelatase reaction. In the present study, we developed a continuous assay where we could simultaneously measure ATP hydrolysis and the formation of Mg-deuteroporphyrin IX. Magnesium chelatase activity is directly correlated with ATPase activity, and no Mg-deuteroporphyrin IX is formed without ATP hydrolysis (Fig. 3). In contrast, ATP hydrolysis can occur at a low but detectable rate without formation of Mg-deuteroporphyrin IX, but it is considerably higher during magnesium chelation (Fig. 3). Incubation of 3.2 g (54 pmol) of BchD with 1.4 g (37 pmol) of BchI and 2 mM ATP allowed ATP hydrolysis at a rate of 10 pmol of P i /min, which is approximately 6-fold less than with BchI alone (58 pmol of P i /min) under equal conditions (Fig. 4A). Addition of 15 g (116 pmol) of BchH preincubated with deuteroporphyrin IX to the solution initiated the formation of Mg-deuteroporphyrin IX (1.14 pmol of Mg-deuteroporphyrin IX formed per min) and instantly caused a 20-fold increase in ATP hydrolysis (200 pmol of P i /min) (Fig. 4A). We conclude that the highest ATPase activity is observed when the enzyme is catalyzing the formation of Mg-porphyrin and the lowest ATPase activity is seen when BchI is mixed with BchD. Similar observations have previously been reported from studies with Synechocystis magnesium chelatase (11).
Higher activities could be obtained if more protein was used, but the activities still remained proportional to the wild-type activity. None of the modified proteins could contribute to magnesium chelatase activity. Magnesium chelatase assays containing a mixture of 1.4 g of wild-type BchI and 1.4 g of BchI-L111F, BchI-D207N, or BchI-R289K showed reduced activity compared with assays only containing 1.4 g of wildtype BchI; 38%, 15%, and 24% of wild-type activity for the respective mixture (Fig. 4, B-D).
BchI-V113G had an ATPase activity of 15 pmol of P i /min, which was 25% of wild-type activity (Fig. 4E) and thus considerably lower than the 80% activity reported for the corresponding modification in HslU (27). When combined with BchD and BchH, the ATPase activity measured 35 pmol of P i /min, and the magnesium chelatase activity was 0.26 pmol of Mg-deuteroporphyrin IX formed per min, which was 18 and 23% of the respec-tive wild-type levels (Fig. 4E). It is striking that BchI-V113G can contribute to magnesium chelatase activity whereas BchI-L111F, BchI-D207N, and BchI-R289K are totally deficient even though the ATPase activity of all four proteins is of similar magnitude. The three modifications L111F, D207N, and R289K probably destabilize the conformational changes that are necessary to maintain the chelatase catalytic cycle.
Magnesium chelatase assays containing a mixture of BchI-V113G and wild-type BchI had the effect of stimulating the activity (1.47 pmol of Mg-deuteroporphyrin IX/min) (Fig. 4E); that is, the activity was higher than in magnesium chelatase assays where BchI-V113G was excluded (1.14 pmol of Mg-deuteroporphyrin IX/min, Fig. 4A). This is the first time that we have observed that an ATPase-deficient BchI subunit can stimulate magnesium chelatase activity in an assay in which it is combined with wild-type BchI subunits. BchI-V113G will be studied further to learn about allosteric modifications in the ID complex during the catalytic cycle of magnesium chelatase.
Mixing Experiments with Inactive BchI Subunits-Structural analyses have shown that the BchI ring is formed by six subunits in a trimer of dimers (15). The x-ray structure of BchI shows that residue Leu-111 is located on one side of the wedge-shaped BchI structure and that Asp-207 and Arg-289 are located on the opposite side. However, in a dimer they all share the same ATP binding pocket (Fig. 1). In a subunit-mixing experiment involving BchI-L111F and BchI-D207N or a mixture of BchI-L111F and BchI-R289K, it was possible to form ATP binding sites with no modified residues in the interface between BchI subunits (Fig. 5). However, none of the resulting trimers of dimers contained more than three wild-type subunit interfaces (Fig. 5, E  and F). We performed the mixing experiments to determine whether BchI complexes with only a few wild-type interfaces could contribute to magnesium chelatase activity. However, BchI-L111F mixed with BchI-D207N or BchI-L111F mixed with BchI-R289K could not contribute to magnesium chelatase activity in vitro (Fig. 4, F-H).
Combining Inactive and Active BchI Subunits in Vivo-To detect possible subtle magnesium chelatase activity, we also performed the mixing experiment in vivo.  clo157. The chlorophyll content was analyzed in leaves that had been grown for 7 days. The wild type contained 1.6 mg of chlorophyll/g of fresh leaf weight (Fig.  6). The pale green heterozygote leaves of WT/Xantha-h.clo161, WT/Xantha-h.clo125, and WT/Xantha-h.clo157 contained FIGURE 3. Simultaneous measurement of ATP hydrolysis and magnesium chelatase activity. The activities were followed as formation of P i and Mgdeuteroporphyrin IX, respectively. Reagents for detection of P i (Molecular Probes EnzChek Phosphate Assay Kit) were added to a standard magnesium chelatase assay. A mixture of BchI and BchD subunits showed low but detectable ATPase activity. Addition of the BchH subunit at t ϭ 20 min initiated the formation of Mg-deuteroporphyrin IX and triggered ATP hydrolysis. 0.59, 0.62, and 0.58 mg/g, respectively, which was 36, 38, and 36% of the wild-type level. The amount of chlorophyll in the mutants Xantha-h.clo161/Xantha-h.clo125 and Xantha-h. clo161/Xantha-h.clo157 was less than the detection limit of 1 g/g. Thus, the mutation in Xantha-h.clo161 could not be functionally complemented by the mutations in Xanthah.clo125 or Xantha-h.clo157. The result of the in vivo experiment is congruent with the in vitro data, in that more than three wild-type subunit interfaces in the ID complexes were required for magnesium chelatase activity.
Catalytic Turnover Triggers Subunit Exchange-Dissociation of the proteins making up the ring structure has been suggested as a possible mechanism of AAAϩ proteins releasing their product after catalysis (28). Magnesium chelatase is unusual in the sense that Mg-protoporphyrin IX rather than a protein is the ultimate product of the AAAϩ activity, although the BchH subunit is the target of the BchID complex. Nevertheless, it is essential to understand the stability of the AAAϩ complex during the catalytic cycles to understand the reaction mechanism of magnesium chelatase. Because ATPase-deficient BchI subunits form a complex with wild-type BchI and interfere with the activity, they could be used to monitor the stability of the magnesium chelatase AAAϩ complex. Therefore, we added ATPase-deficient BchI subunits and wild-type BchI at different stages to magnesium chelatase reaction assays that were already running. All assays contained fixed amounts of wild-type BchD and wild-type BchH subunits. The assays also contained either wild-type BchI, BchI-L111F, BchI-D207N, BchI-R289K, or BchI-V113G in the same stoichiometric amounts. The assays were started and continuously monitored for Mg-deuteroporphyrin IX production (Fig. 7) and ATP hydrolysis (data not shown). After 72 min, additional wild-type or modified BchI proteins were added, and the reactions were monitored further for 178 min. The reactions were run at 22°C, which permitted a slow reaction rate, giving time for addition of BchI subunits. Reactions initiated with wild-type BchI were slowed down by the addition of modified BchI subunits. BchI-L111F, BchI-D207N, and BchI-R289K showed an inhibitory effect 22-23 min after addition (Fig. 7A). In the opposite situations, in which the assays were initiated with BchI-L111F, BchI-D207N, or BchI-R289K, the stimulatory effect was first observed 74 -78 min after addition of wild-type BchI (Fig. 7B). Assays initiated with BchI-V113G, which had 18 and 23% of the wild-type ATPase-and chelatase activity, respectively, were stimulated 31 min after addition of wild-type BchI. This was ϳ47 min less than the time required to activate the assays initiated with BchI-L111F, BchI-D207N, and BchI-R289K. It should also be noted that the assay started with BchI-V113G was stimulated by addition of wild-type BchI 9 min later than the wild-type BchI assay was inhibited by addition of BchI-L111F, BchI-D207N, or BchI-R289K. Our results demonstrate that there is an exchange of BchI subunits. Furthermore, the exchange is more pronounced in active assays, which suggests that dissociation and reassembly of the BchI ring is part of the catalytic cycle.

DISCUSSION
In light of our present knowledge of the structure and function of magnesium chelatase, the enzyme should be understood as an AAAϩ system with a dynamic ATPase motor unit (the ID complex) and an AAAϩ-specific substrate (the H subunit). In general terms, the energetically unfavorable reaction of insertion of Mg 2ϩ into protoporphyrin IX by the H subunit is fueled by ATP hydrolysis performed by the ID complex. The process also involves considerable conformational changes within the ID complex (15) as well as within the H subunit (7). The present study supports earlier observations that the I subunit has ATP hydrolytic activity that exceeds the ATP hydrolytic activity of  the ID complex (11). Addition of H subunit triggers ATP hydrolysis and enables enzymatic production of Mg-porphyrin. It is likely that BchD causes an allosteric inhibition of the intrinsic ATPase activity of BchI, which is released when BchH binds to the complex and the reaction starts. Thus, ATP hydrolysis is controlled internally throughout the catalytic pathway. The ID complex is unusual because it consists of a trimer of dimers instead of a hexamer, as found in most other AAAϩ proteins. In experiments involving mixing of inactive BchI subunits, we found that a mixture of BchI-L111F and BchI-D207N or a mixture of BchI-L111F and BchI-R289K could not contribute to magnesium chelatase activity, although 92% of the resulting BchI ring structures contained at least one BchI-BchI interface with no modified amino acid residue. The mixing experiment could also be performed in vivo because barley mutants Xantha-h.clo161, Xantha-h.clo125, and Xantha-h.clo157 correspond to BchI-L111F, BchI-D207N, and BchI-R289K, respectively. The F1 generation of crosses Xantha-h.clo161 ϫ Xantha-h.clo125 and Xantha-h.clo161 ϫ Xantha-h.clo157 resulted in a segregating population where the double mutants Xantha-h.clo161/Xantha-h.clo125 and Xantha-h.clo161/ Xantha-h.clo157 showed no traces of chlorophyll due to an inability to form active magnesium chelatase complexes. Specifically, 7.7, 38, 46, and 7.7% of the I-ring structures in the in vitro and in vivo mixing experiments would contain 0, 1, 2, and 3 intersubunit interfaces with no modified residues, assuming that both kinds of subunits are equally able to be incorporated into the complexes (Fig. 5). We conclude that more than three functional interfaces in the BchI ring structure are required for magnesium chelatase activity. In addition, complexes with up to three functional interfaces are blocked in magnesium chelatase activity rather than showing low activity. It remains to be shown whether complexes containing four or five functional interfaces are able to contribute to magnesium chelatase activity or whether only I-ring structures consisting of six wild-type interfaces are active.
Dissociation of AAAϩ complexes has been suggested to be an important mechanism of product release (28). However, no consensus has been found in terms of stability of AAAϩ complexes during the catalytic cycle, which may reflect the diversity of the AAAϩ protein family. AAAϩ complexes of ClpAP proteins remained associated during hundreds of rounds of ATP hydrolysis (29). In contrast, the AAAϩ chaperone ClpB exchanged subunits fast on a time scale from seconds to minutes (28). In the present study, we developed an assay where we could follow magnesium chelatase activity and ATP hydrolysis simultaneously. The assay allowed us to test the effect of addition of active and inactive BchI subunits to an ongoing reaction. When BchI-L111F, BchI-D207N, or BchI-R289K was added to an active magnesium chelatase assay run with wild-type BchI, it took 22-23 min before a retardation of the reaction was observed. When wild-type BchI was added to a magnesium chelatase assay run with BchI-L111F, BchI-D207N, or BchI-R289K, it took 74 -78 min before a stimulation of the reaction was seen. The experiment showed that an exchange of BchI subunits occurs in the magnesium chelatase complex. Furthermore, the exchange is faster if the reaction has a high turnover. The second observation was strengthened by the partially active BchI-V113G subunit responding faster than BchI-L111F, BchI-D207N, and BchI-R289K to an addition of wild-type BchI. Although the dissociation of magnesium chelatase complexes involving BchI-L111F, BchI-D207N, or BchI-R289K is slow, it still occurs because enzymatic activity could be seen 74 -78 min after addition of wild-type BchI subunits. This time possibly reflects a basic instability of the complex. It also remains to be investigated whether the 5-11% residual ATPase activity seen in BchI-L111F, BchI-D207N, and BchI-R289K is important for the dissociation. Singh et al. (29) reported that dissociation of ClpA and ClpAP is faster in the presence of ATP than in the presence of slowly hydrolyzable ATP␥S, which again suggests that dissociation in AAAϩ complexes is connected to catalytic turnover. The frequency of dissociation might vary considerably between different AAAϩ protein complexes. In the case of ClpA and ClpAP, it was estimated that the dissociation occurs in Ͻ1 in 100 cycles (29). In contrast, hexamers of ClpB, the closest relative of ClpA in E. coli, show fast dissociation as an intrinsic part of the product release mechanism (28). It should be noted that the ring consisting of six magnesium chelatase D subunits is very stable, and no exchange of D subunits could be detected in an ongoing magnesium chelatase reaction, which indicates that the D ring does not dissociate (16). Instead, the D ring was suggested to function as a platform for the reassembly of the ID double ring structure after dissociation of I subunits (16). The formation of the ID complex might be the so-called FIGURE 7. Addition of active or inactive BchI subunits to ongoing magnesium chelatase assays. Activity was followed as formation of Mg-deuteroporphyrin IX. A, at t ϭ 72 min (indicated), inactive BchI subunits were added to assays started with active BchI. B, active BchI was added to assays started with inactive or partially active BchI. Assays started with active BchI slowed down 22-23 min (at t ϭ 94 -95 min) after addition of inactive BchI. Assays started with inactive BchI became activated 74 -78 min (at t ϭ 146 -150 min) after addition of active BchI. Assays started with BchI-V113G became more active 31 min (at t ϭ 103 min) after addition of active BchI. activation step seen as a lag phase at the start of a magnesium chelatase assay (30). From kinetic analysis, it has been suggested that the activation step does not occur in every catalytic cycle (31). The present study clearly demonstrates that catalytic turnover of the magnesium chelatase results in dissociation of I subunits, which may be the mechanism of product release, i.e. the release of the H subunit. We suggest that a transient conformation is formed during catalysis, from which dissociation occurs. The requirement of a complete activation step after product release remains to be investigated.