|Home | About | Journals | Submit | Contact Us | Français|
The genomes uncoupled 4 (GUN4) protein stimulates chlorophyll biosynthesis by increasing the activity of Mg-chelatase, the enzyme that inserts magnesium into protoporphyrin IX (PPIX) in the chlorophyll biosynthesis pathway. One of the roles of GUN4 is in binding PPIX and Mg-PPIX. In eukaryotes, GUN4 also participates in plastid-to-nucleus signalling, although the mechanism for this is unclear. Here, the first crystal structure of a eukaryotic GUN4, from Chlamydomonas reinhardtii, is presented. The structure is in broad agreement with those of previously solved cyanobacterial structures. Most interestingly, conformational divergence is restricted to several loops which cover the porphyrin-binding cleft. The conformational dynamics suggested by this ensemble of structures lend support to the understanding of how GUN4 binds PPIX or Mg-PPIX.
In plants and algae, tetrapyrrole biosynthesis occurs in the chloroplast, and the genes for light-harvesting proteins are encoded in the nuclear genome (Formighieri et al., 2012 ). Retrograde signalling pathways between these two organelles were discovered through the Arabidopsis thaliana genomes uncoupled (GUN) mutants (Susek et al., 1993 ; Willows, 2003 ). Five gun mutants have been reported (gun1–gun5), among which four (gun2–gun5) are on the same retrograde signalling pathway and encode proteins that function in tetrapyrrole metabolism. gun2 codes for haem oxygenase, gun3 for phytochromobilin synthase and gun5 for the H subunit of Mg-chelatase. Mg-chelatase is the enzyme complex which inserts the Mg2+ ion into protoporphyrin IX (PPIX), generating Mg-PPIX, in an ATP-dependent manner, and which constitutes the precursor for chlorophyll biosynthesis. Mg-chelatase consists of three protein subunits: ChlI/BchI (38–42 kDa), ChlD/BchD (60–74 kDa) and BchH/ChlH (140–150 kDa; Gorchein et al., 1993 ; Jensen et al., 1996 ; Papenbrock et al., 1997 ; Wang et al., 1974 ). These subunits are conserved from bacteria to higher plants and are responsible for different functions. The ChlI/BchI subunits are members of the AAA+ superfamily with ATPase activity required for catalysis. They bind Mg2+ and ATP and also function as molecular chaperones for the ChlD/BchD subunits. The ChlD/BchD subunits appear to have an inactive AAA+ ATPase. The ChlH/BchH subunits are the largest within the complex and form the core of the magnesium chelatase enzyme, and they bind both PPIX and Mg-PPIX.
The gun4 gene was the first locus to be identified in the screen for genome uncoupled mutants (Susek et al., 1993 ; Willows & Hansson, 2003 ), and it has since been shown to be involved in both retrograde signalling and post-translational regulation of tetrapyrrole biosynthesis (Larkin et al., 2003 ; Peter & Grimm, 2009 ). The gun4 gene product (GUN4) has also been shown to be required for optimal Mg-chelatase activity through its interaction with the ChlH subunit (Davison et al., 2005 ; Verdecia et al., 2005 ). In doing so, it acts as a regulatory subunit and facilitates delivery of the PPIX substrate to ChlH, and also binds the product (Mg-PPIX; Chen et al., 2015 ; Davison et al., 2005 ; Larkin et al., 2003 ; Verdecia et al., 2005 ). Two crystal structures of GUN4 proteins have been elucidated from the cyanobacteria Synechocystis (PDB entries 1y6i, 4xkb and 4xkc; Verdecia et al., 2005 ; Chen et al., 2015 ) and Thermosynechococcus elongatus (PDB entries 1z3x and 1z3y; Davison et al., 2005 ). The structures with PDB codes 4xkb and 4xkc are with porphyrin bound. Both are predominantly α-helical proteins comprised of a smaller N-terminal domain (five α-helices) connected to a larger C-terminal domain (eight α-helices). Structural conservation is stronger in the C-terminal domains, consistent with the high sequence similarity of the C-terminal domains among GUN4 family members across a broad spectrum of species. The N-terminal domain is more variable and is missing in some prokaryotic family members (Davison et al., 2005 ; Verdecia et al., 2005 ). Here, we report the crystal structure of GUN4 from Chlamydomonas reinhardtii at 3.5 Å resolution and hence reveal the first structural details of an eukaryotic GUN4 protein.
The GUN4 gene from C. reinhardtii was cloned from cDNA into expression vector pET-28a (Merck–Novagen) and was expressed in Escherichia coli strain BL21(DE3) as a fusion with an N-terminal poly-His tag (Table 1 ). The IPTG-induced cells were grown at 37°C in LB medium. His6-tagged GUN4 was purified by immobilized Ni2+-affinity chromatography using a 5 ml HisTrap FF column (GE Healthcare). The supernatant from cleared lysate was loaded onto the column and washed with ten volumes of wash buffer (20 mM Tris–HCl pH 7.9, 0.5 M NaCl, 60 mM imidazole) and the immobilized proteins were eluted in elution buffer (80 mM Tris–HCl pH 7.9, 0.5 M NaCl, 1 M imidazole). The peak protein fractions were determined using the Bradford reagent (Bio-Rad). Proteins in pooled fractions were immediately desalted using a PD-10 column (GE Healthcare) to a final buffer consisting of 20 mM tricine–NaOH pH 8.0, 2 mM β-mercaptoethanol. The GUN4 samples were concentrated to 20–35 mg ml−1 using a 10K Amicon centrifugal filter device (Millipore) prior to crystallization experiments.
The GUN4 protein was screened for crystallization using a robotic liquid-handling dispenser (Phoenix, Art Robbins Instruments) at the University of New South Wales, Sydney, Australia. Crystals were grown using the vapour-diffusion technique and a sitting-drop format, in which equal volumes of protein and well solutions were combined in total drop volumes of 1–2 µl. Crystallization was monitored by a microscope at various time intervals. Numerous crystallization hits (17) were observed from these sparse-matrix screens. The small (100 µm in the longest dimension) rice-grain-like hexagonal crystals used in X-ray diffraction experiments appeared only after a prolonged period of room-temperature incubation, 44 weeks after the trays were established. The crystallization condition was solution B5 from the SaltRx HT screen (Hampton Research), which consisted of 1.0 M ammonium citrate tribasic pH 7.0, 0.1 M bis-tris propane pH 7.0. The protein for this sample was originally supplied at 35 mg ml−1. Harvesting was achieved by simply looping crystals from the mother liquor and then plunge-cooling them in liquid nitrogen with no cryoprotection regime. Crystals were transported to the Australian Synchrotron for data collection. Diffraction data were collected at 100 K on beamline MX2. Dozens of crystals were investigated; however, useable diffraction was only obtained from several crystals using an unattenuated beam and 3 s exposures. Crystallization information is summarized in Table 2 .
Diffraction data from two crystals (44 frames and 22 frames, respectively, each recorded over 1° oscillations with 3 s exposures to an unattenuated beam) were indexed and integrated to 3.5 Å resolution with MOSFLM (Leslie & Powell, 2007 ). The space group (P3221) was confirmed with POINTLESS (Evans, 2011 ). Integrated reflections from these two crystals were scaled and merged with AIMLESS (Evans & Murshudov, 2013 ), accessed via the CCP4 software interface (Winn et al., 2011 ). The data from multiple crystals were combined to maximize the data completeness and multiplicity. Data-collection and processing statistics are summarized in Table 3 .
The structure was determined by molecular replacement with Phaser (McCoy et al., 2007 ). The search model was taken from the T. elongatus wild-type structure (PDB entry 1z3x), which is 40% identical at the amino-acid sequence level over 165 residues. The more divergent first 88 residues of this structure (the N-terminal domain) were excluded from the search model. Molecular replacement initially identified four GUN4 molecules in the asymmetric unit, consistent with the crystal comprising 48% solvent (as calculated from the Matthews coefficient and full-length GUN4). However, incomplete crystal packing and inspection of electron-density maps suggested that additional unaccounted-for molecules might in fact be present. The search model was subsequently trimmed further (a further 18 residues were removed from the N-terminus, with the removal of several loops that were divergent in the T. elongatus and Synechocystis structures and with side chains truncated to the Cβ position), which allowed Phaser to fit six non-clashing molecules in the asymmetric unit (solvent content of 53% for six molecules of ~150 residues). Hence, it appeared that the N-terminal domain of the molecule was missing.
Restrained B-factor refinement was performed with REFMAC5 (Murshudov et al., 2011 ), using local noncrystallographic symmetry (NCS) restraints. Between rounds of refinement, electron-density maps and their fit to the model were examined using Coot (Emsley et al., 2010 ). Amino-acid side chains were added if suggested by difference map electron density. Difficult segments of the model were built with sensible geometry using Buccaneer (Cowtan, 2006 ). Structure validation was performed using the MolProbity web server (Chen et al., 2010 ). Refinement statistics are shown in Table 4 . The coordinates for the final model have been deposited in the Protein Data Bank (PDB entry 4ykb).
Six GUN4 molecules were found in the crystallographic asymmetric unit (chains A–F). All of them are lacking the N-terminal domain (the first resolvable residues are 80–83), and all of them are lacking the last 22–25 residues (the last resolvable residues are 235–238) (see Fig. 1 ). The fold of the resolvable ~150 residues modelled is effectively the same in all chains (r.m.s.d. of 0.14–0.29 relative to chain A; see Fig. 2 ). Owing to the low resolution of the data, between 25 and 30% of the amino-acid side chains in the various chains have not been modelled (Table 4 ). Those unmodelled are predominantly on the surface of the domain, whilst those buried in the hydrophobic core of the domain are generally well ordered.
The fold is largely α-helical, as expected from the T. elongatus and Synechocystis structures (PDB entries 1z3x and 1y6i, respectively). The A chain superposes with the 1z3x structure with an r.m.s.d. of 0.97 Å (over 143 Cα positions) and with the 1y6i structure with an r.m.s.d. of 1.1 Å (over 131 Cα positions) (Fig. 2 ).
The GUN4 molecules resolved in the crystal have quite clearly undergone proteolysis, as confirmed by MSMS proteomic analysis of tryptic digests of a number of crystals. The N-terminus of the resolvable domain immediately abuts a (missing) KKK sequence motif (analogous to the region linking the N- and C-terminal domains in related structures), whilst the C-terminal limit is adjacent to Lys238 (the last residue modelled in chain A). The C-terminal residues missing in our structure contain sites of phosphorylation in eukaryotes, but are not sequence features that are found in the cyanobacterial GUN4 molecules. The extended time taken for crystal growth presumably reflects the time taken for a contaminating protease within the crystallization drop to convert sufficient molecules to the doubly truncated form such that crystal growth could occur.
Whilst the overall fold of the C-terminus of our C. reinhardtii GUN structure is highly similar to the cyanobacterial structures (Fig. 2 ; the Synechocystis structure is coloured light blue and the T. elongatus structure dark blue), the largest differences, and also the largest differences between the cyanobacterial structures themselves, concern two ‘loop’ segments: the loop linking helices 2 and 3 of the Synechocystis structure and the loop linking helices 6 and 7. The α2/α3 loop is actually a short section of helix and the α6/α7 loop is a β-turn in the porphyrin-bound Synechocystis structures (Chen et al., 2015 ), whilst these loops help to cover the hydrophobic ‘greasy palm’ in the unbound Synechocystis structure (Verdecia et al., 2005 ) (Fig. 2 ; α2/α3 and α6/α7 loops). The T. elongatus structure and our C. reinhardtii structure are thus intermediate between the porphyrin-bound and unbound Synechocystis structures (Fig. 2 ).
The α6/α7 loop covers the greasy-palm surface but projects very few residues into the cleft, namely a proline residue (conserved in all three organisms) and a leucine residue (substituted by a threonine residue in the T. elongatus structure and in our C. reinhardtii structure) (Fig. 2 ; proline and threonine residues from our structure are drawn as sticks). The porphyrin-bound Synechocystis structures (Chen et al., 2015 ) clearly show that this α6/α7 loop has a propensity for conformational dynamics to form a β-hairpin, making a more structured cleft which enables porphyrin binding.
We have solved the crystal structure of the GUN4 protein from C. reinhardtii. Proteolysis within the crystallization experiment removed the N-terminal domain and the last ~20 residues, leaving the bulk of the C-terminal domain intact; this domain is involved in binding PPIX and Mg-PPIX. Although the structure is of relatively low resolution, the bulk of the side chains which line the ‘greasy palm’ cleft involved in binding the porphyrin ring can be resolved. When superposed with cyanobacterial GUN4 structures, whilst the α-helical scaffolds are conserved, the α2/α3 and α6/α7 loops previously predicted to undergo movement to accommodate porphyrin binding display markedly different conformations. This diversity of fold captured by multiple crystal structures supports the notion that these loops undergo conformational rearrangement to accommodate the insertion of PPIX or Mg-PPIX into the underlying hydrophobic cleft (Fig. 3 ), as observed in the recent porphyrin-bound Synechocystis structures (Chen et al., 2015 ).