The Arabidopsis glutamyl-tRNA reductase (GluTR) forms a ternary complex with FLU and GluTR-binding protein

Tetrapyrrole biosynthesis is an essential and tightly regulated process, and glutamyl-tRNA reductase (GluTR) is a key target for multiple regulatory factors at the post-translational level. By binding to the thylakoid membrane protein FLUORESCENT (FLU) or the soluble stromal GluTR-binding protein (GBP), the activity of GluTR is down- or up-regulated. Here, we reconstructed a ternary complex composed of the C-terminal tetratricopepetide-repeat domain of FLU, GBP, and GluTR, crystallized and solved the structure of the complex at 3.2 Å. The overall structure resembles the shape of merged two binary complexes as previously reported, and shows a large conformational change within GluTR. We also demonstrated that GluTR binds tightly with GBP but does not bind to GSAM under the same condition. These findings allow us to suggest a biological role of the ternary complex for the regulation of plant GluTR.

Plants synthesize δ -aminolevulenic acid (ALA), the precursor for all tetrapyrrole molecules, from glutamate via a three-step pathway 1 . The first step is ligation of glutamate to tRNA Glu catalyzed by glutamyl-tRNA synthetase. Then glutamyl-tRNA reductase (GluTR) reduces the tRNA Glu -bound glutamate to glutamate-l-semialdehyde (GSA) in an NADPH-dependent manner. GSA is subsequently isomerized to ALA by a vitamin B 6 -dependent enzyme, glutamate-l-semialdehyde aminomutase (GSAM). ALA synthesis is the key regulatory point for the entire tetrapyrrole biosynthetic pathway, and particularly GluTR is subjected to a tight control at the post-translational level 2,3 .
Three mechanisms have been characterized for plant GluTR activity regulation, which are (i) the end-product feedback inhibition by heme 4 , (ii) repression by a membrane protein FLUORESCENT (FLU) 5 , and (iii) formation of complex with a soluble GluTR-binding protein (GBP) 6 . The two inhibitors, heme and FLU, are suggested to concurrently interact with different sites on GluTR 7 . GluTR consists of three domains: an N-terminal catalytic domain, an NADPH-binding domain, and a C-terminal dimerization domain 8 . FLU directly interacts with GluTR's dimerization domain through its tetratricopepetide-repeat (TPR) domain 7,9,10 . Plant GluTRs have an ~30-residue conserved fragment in the N-terminal region, and truncation of this fragment results in resistance to heme inhibition 4 . This putative heme-binding fragment, however, is flexible and hence not observed in the GluTR-GBP complex structure 11 . GBP has been proposed to protect GluTR from FLU inhibition during darkness to ensure heme synthesis when the need for chlorophyll declines 12 , and a membrane anchoring protein specific for GBP has been speculated 13 . Recent structural studies of the GluTR-GBP complex 11 and of FLU's TPR domain (FLU TPR ) complexed with GluTR's dimerization domain 10 have revealed that FLU and GBP bind to different sites on GluTR. These findings indicate that the three post-translational mechanisms of GluTR regulation may function simultaneously.
Transcriptional regulation of enzymes involved in ALA synthesis has been characterized in Arabidopsis thaliana. Among the three GluTR genes (HEMA1, HEMA2 and HEMA3), expression of HEMA1 that encodes

Results
Reconstruction, crystallization and structure determination of the ternary complex. The purified recombinant GluTR, GBP and FLU TPR were mixed at molar ratio of 2:3:3, and the mixture was then subjected to size-exclusion chromatography. A stable FLU TPR -GluTR-GBP ternary complex was obtained with excess amounts of GBP and FLU TPR (Fig. 1A). No complex formation between FLU TPR and GBP was observed. Fractions corresponding to the ternary complex were concentrated for crystallization. Crystals grew under a totally different condition from the GluTR-GBP complex 11 or FLU TPR in complex with GluTR's dimerization domain (GluTR DD ) 10 . The ternary complex crystals belong to space group C2, while the GluTR-GBP binary complex crystals belong to P2 1 2 1 2 1 , and the FLU TPR -GluTR DD binary complex crystals belong to P6 5 22. The ternary complex packs in a symmetric way along its local 2-fold axis, whereas the GluTR-GBP complex arrays in an asymmetric way along the axis (Fig. 1B). The structure of the ternary complex was determined by the molecular replacement method using template coordinates of GluTR-GBP and FLU TPR -GluTR DD , and refined to a resolution of 3.2 Å ( Table 1).
Structure of the ternary complex. The ternary complex resembles the shape of a merged structure of the two binary complexes ( Fig. 2A). However, it does not fit well with either GluTR-GBP or FLU TPR -GluTR DD . The positions of GluTR's C-terminal region are quite different when the ternary complex and GluTR-GBP are superimposed (Fig. 2B). GBP and the remainder of GluTR have no significant change, except that the linker between NADPH-binding domain and the long "spinal" helix of GluTR cannot be traced in the ternary complex. Conversely, the linker between the two C-terminal helices of GluTR that is missing in GluTR-GBP or FLU TPR -GluTR DD can be observed in the ternary complex. Compared with FLU TPR -GluTR DD , there is an extra ionic bond between the catalytic domain of GluTR and the third TPR motif of FLU TPR (Fig. 2C). GluTR's C-terminal region appears more compact in the ternary complex than in FLU TPR -GluTR DD .

Flexibility of GluTR's spinal helix.
The two chains of GluTR in the ternary complex, together with the previous observation in the GluTR-GBP complex 11 , demonstrate the flexibility of GluTR's spinal helix across a large range. When the catalytic domains of the four chains are superimposed, the stem end exhibits maximum shift of approximate 15 Å (Fig. 3A). The two spinal helices in the ternary complex are almost identical, which is reminiscent of Methanopyrus kandleri GluTR 8 . Indeed, the root-mean-square deviation between the two chains of GluTR in the ternary complex is only 0.72 Å for 419 Cα aligned. When the stem of the GluTR dimer in the GluTR-GBP binary complex is superimposed with that in the ternary complex, the angle between the two Y-shaped arms has a difference of ~5 degrees (Fig. 3B).
GluTR's interaction with GSAM and GBP. GSAM is a flexible enzyme undergoing open/close conformational change 21,22 . Synchronized events between GluTR and GSAM are likely required for GluTR-GSAM interaction. A stable GluTR-GSAM complex has been verified for this pair of enzymes from E. coli and C. reinhardtii 18,19 . In contrast, direct interaction between plant GluTR and GSAM has not been reported. We employed ITC to detect such interaction (Fig. 4). No heat change was observed for titration of GSAM to GluTR. Also, no heat change was observed for titration of GSAM to the GluTR-GBP complex. Notably, the GluTR-GBP complex is stable and has a low apparent dissociation constant (K d ). The K d value (41.3 ± 3.7 nM) is about one-fortieth that of FLU TPR and GluTR as measured previously 10 , which indicates that GBP binds significantly more tightly than FLU TPR to GluTR.

Discussion
As the rate-limiting step for the formation of ALA, the common precursor for all tetrapyrrole molecules, the GluTR-catalyzed glutamyl-tRNA Glu reduction by NADPH is a key regulatory point of the tetrapyrrole biosynthetic pathway 2,3 . The membrane-anchored protein FLU were identified as a negative regulator for GluTR 5,7,9 . The soluble protein GBP was initially found in chloroplast stroma 23 , and then later in a thylakoid membrane-bound 300-kDa protein complex 6 . Direct GluTR-GBP interaction has also been found in an interactome screen 24 . With the components of the 300-kDa protein complex remaining unresolved, the FLU TPR -GluTR-GBP ternary complex presented here provides a clue to address this issue. A membrane-bound FLU-containing metabolic complex has been detected by an immunoprecipitation/mass spectrometry study 25 . In this complex, light-dependent protochlorophyllide oxidoreductase (LPOR) is one of the specific FLU-interacting partners. LPOR catalyzes the reduction of the fourth ring of protochlorophyllide, and exists as dimers or tetramers 26 . With a monomeric molecular weight of ~36 kDa, when a LPOR dimer binds to the ternary complex of full-length FLU-GluTR-GBP that has a combined molecular weight of ~224 kDa, the resulting LPOR-FLU-GluTR-GBP quaternary complex might explain the post-translational regulation of ALA synthesis by light 6,27 . Further biological studies are needed to characterize such a macromolecular assembly. How GBP exerts its regulatory role on GluTR activity remains an open question. GBP has higher binding affinity to GluTR compared with FLU TPR . This indicates that GluTR is more prone to bind to GBP than FLU under the same condition. GBP may regulate GluTR activity by the following three mechanisms: (i) to shift GluTR conformation and render GluTR preferable for NADPH accommodation within the NADPH-binding domain, and thus prevent GluTR's esterase activity; (ii) to retain GSA in GluTR's interior before GSAM interaction; (iii) to be involved in chloroplast vitamin B 6 metabolism and hence related to GSAM activation.
The failure to detect GluTR-GSAM interaction using ITC does not preclude the existence of a GluTR-GSAM complex in plants. Nevertheless, such a complex might be less stable than its counterparts from E. coli and C. reinhardtii 18,19 . It should be noticed that, similar to a GluTR dimer, a GSAM dimer has both asymmetric and symmetric states 21,22,28 . A synchronized conformational change of both GluTR and GSAM is likely required for their recognition. In addition, whether and how dissociation of the GluTR-GBP complex is involved in GSAM interaction remains unclear and awaits future biochemical characterization.
The cell pellets expressing GluTR were re-suspended in buffer A (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 1 mM EDTA and 5 mM dithiothreitol) and disrupted by sonication. After centrifugation, the cleared lysate was passed through a maltose binding protein (MBP) affinity column pre-equilibrated with buffer A. The bound protein was eluted with 40 mM maltose in buffer A, and the MBP tag was then cleaved using tobacco etch virus protease. The reaction mixture was then subjected to a HiLoad 16/60 Superdex 200 pg column (GE Healthcare) pre-equilibrated and eluted with buffer A. Protein aggregates and the MBP tag were removed, and the GluTR dimer fractions were collected. Purification of GBP and FLU TPR was described previously 10,11 . GSAM was purified following the same procedure used for GluTR as described above.
Reconstruction of the ternary complex. For preparation of the ternary complex, the GluTR dimer fractions were mixed with GBP and FLU TPR at a molar ratio 2:3:3 and incubated for 1 hour at 4 o C. The mixture was loaded on a HiLoad 16/60 Superdex 200 pg column (GE Healthcare) pre-equilibrated and eluted with buffer containing 20 mM Tris-HCl, pH 7.5, 200 mM NaCl, and 4 mM dithiothreitol. The purified ternary complex was pooled and concentrated to 15 mg ml −1 for crystallization.
Crystallization and data collection. Crystals of the ternary complex were obtained in 0.1 M sodium malonate, pH 7.0, 14.5% (w/v) polyethylene glycol 3,350, 2.5% (v/v) 2-methyl-2,4-pentanediol, 0.5 M lithium chloride by the sitting-drop vapor diffusion method at 16 o C. For data collection, the crystals were transferred step by step into drops of the crystallization liquid supplemented with 5%, 10%, 20% (v/v) ethylene glycol before being flash-frozen in liquid nitrogen. All X-ray diffraction data sets were collected at beamline BL17U of Shanghai Synchrotron Radiation Facility (Shanghai, China) at 100 K, and processed using HKL2000 (HKL Research, Inc.).
Structure solution and refinement. The structural model of the ternary complex was built using molecular replacement with Phaser 30 . The search templates were the GluTR-GBP complex (PDB entry 4N7R) where residues after Arg421 of GluTR were removed, and FLU TPR -GluTR DD (PDB entry 4YVQ). Manual correction was done in Coot 31 according to the 2F o − F c and F o − F c electron density maps. Further refinement was performed with phenix.refine 32 . The diffraction data used for structure refinement was extended to 3.0 Å according to CC 1/2 values 33 , and the final resolution was cut off to 3.2 Å based on traditional restriction. The overall quality of final structure was assessed by MolProbity 34 with 96.7% in favored, 3.0% in general allowed and 0.3% in disallowed regions. Data collection and structure refinement statistics are summarized in Table 1 Control experiments were carried out by injecting protein into the buffer, and the resulting heat of dilution was subtracted. The first injection was discarded, and the data were fitted to a one-site binding model using MicroCal Origin software.