The Non-canonical Tetratricopeptide Repeat (TPR) Domain of Fluorescent (FLU) Mediates Complex Formation with Glutamyl-tRNA Reductase*

Background: FLU negatively regulates glutamyl-tRNA reductase (GluTR) during chlorophyll biosynthesis. Results: The structures of the TPR domain of FLU in its uncomplexed form and the GluTR dimeric domain-bound form were solved. Conclusion: The non-canonical TPR domain of FLU mediates complex formation with GluTR. Significance: The work enables a delineation of the spatial regulation of GluTR. The tetratricopeptide repeat (TPR)-containing protein FLU is a negative regulator of chlorophyll biosynthesis in plants. It directly interacts through its TPR domain with glutamyl-tRNA reductase (GluTR), the rate-limiting enzyme in the formation of δ-aminolevulinic acid (ALA). Delineation of how FLU binds to GluTR is important for understanding the molecular basis for FLU-mediated repression of synthesis of ALA, the universal tetrapyrrole precursor. Here, we characterize the FLU-GluTR interaction by solving the crystal structures of the uncomplexed TPR domain of FLU (FLUTPR) at 1.45-Å resolution and the complex of the dimeric domain of GluTR bound to FLUTPR at 2.4-Å resolution. Three non-canonical TPR motifs of each FLUTPR form a concave surface and clamp the helix bundle in the C-terminal dimeric domain of GluTR. We demonstrate that a 2:2 FLUTPR-GluTR complex is the functional unit for FLU-mediated GluTR regulation and suggest that the formation of the FLU-GluTR complex prevents glutamyl-tRNA, the GluTR substrate, from binding with this enzyme. These results also provide insights into the spatial regulation of ALA synthesis by the membrane-located FLU protein.


The tetratricopeptide repeat (TPR)-containing protein FLU is a negative regulator of chlorophyll biosynthesis in plants. It directly interacts through its TPR domain with glutamyl-tRNA reductase (GluTR), the rate-limiting enzyme in the formation of ␦-aminolevulinic acid (ALA). Delineation of how FLU binds to
GluTR is important for understanding the molecular basis for FLU-mediated repression of synthesis of ALA, the universal tetrapyrrole precursor. Here, we characterize the FLU-GluTR interaction by solving the crystal structures of the uncomplexed TPR domain of FLU (FLU TPR ) at 1.45-Å resolution and the complex of the dimeric domain of GluTR bound to FLU TPR at 2.4-Å resolution. Three non-canonical TPR motifs of each FLU TPR form a concave surface and clamp the helix bundle in the C-terminal dimeric domain of GluTR. We demonstrate that a 2:2 FLU TPR -GluTR complex is the functional unit for FLU-mediated GluTR regulation and suggest that the formation of the FLU-GluTR complex prevents glutamyl-tRNA, the GluTR substrate, from binding with this enzyme. These results also provide insights into the spatial regulation of ALA synthesis by the membrane-located FLU protein.
Fluorescent (FLU), 3 a plastid membrane protein, directly interacts with glutamyl-tRNA reductase (GluTR), the initial enzyme in plant tetrapyrrole biosynthetic pathway (1,2). The mature FLU protein is about 27 kDa and contains an N-terminal hydrophobic region that may form a transmembrane structure, a coiled-coil domain, and a C-terminal tetratricopeptide repeat (TPR) domain. Amino acid sequence analysis predicted that the C-terminal TPR domains of Arabidopsis FLU and Chlamydomonas FLU-like protein consist of two TPR motifs (1,3), whereas barley Tigrina d protein, an ortholog of Arabidopsis FLU, consists of three TPR motifs (4). Mutations in the TPR domain, such as A262V in Arabidopsis FLU (flu1-1) and frameshift at Ala 249 in barley Tigrina d (tigrina d), cause defect in the regulation of chlorophyll biosynthesis (1,4,5). The flu and tigrina d mutants lose their ability to suppress protochlorophyllide (PChlide) accumulation upon a dark/light shift and resemble phenotypes similar to wild-type plants fed with exogenous ␦-aminolevulinic acid (ALA). GluTR is the rate-limiting enzyme for endogenous ALA production (6), and hence the physical interaction between FLU and GluTR suggests an inhibitory role of FLU on GluTR activity (2). Indeed, an up-regulation of ALA synthesis was observed for the flu1-1 mutant under continuous light (7). In a yeast two-hybrid system, it was shown that the C-terminal 43 residues of GluTR were required for interacting with the TPR domain of FLU (7). However, in the crystal structures of GluTR, instead of being exposed, this C-terminal region packs inside the cleft of the V-shaped GluTR dimer (8,9). Therefore, the FLU-GluTR interaction needs to be characterized at the atomic level.
As a hub for the post-transcriptional regulation of tetrapyrrole biosynthesis, GluTR is controlled by multiple mechanisms besides the above mentioned FLU-mediated down-regulation (10,11). Metabolic feedback inhibition is a common strategy to repress GluTR activity, although how the tetrapyrrole metabolites, including heme and, particularly interestingly, PChlide (12,13), affect GluTR awaits to be elucidated. An immunoprecipitation/mass spectrometry study has identified a chloroplast membrane complex containing FLU and several key enzymes devoted to chlorophyll synthesis (14). This complex is proposed to interact with GluTR upon accumulation of fluorescent PChlide. Although the above model explains how FLU mediates GluTR activity, the spatial organization of such a membrane complex is unclear. A delineation of the FLU-GluTR interaction should provide a structural insight into how FLU exerts its regulatory role in this complex.
Here we describe the 1.45 Å structure of TPR domain of FLU (FLU TPR ) consisting of three non-canonical TPR motifs. To crystalize a complex containing the components for direct FLU-GluTR interaction, we constructed GluTR C-terminal truncations that preserve dimerization ability but with different polypeptide size and successfully obtained a 2.4 Å structure of the dimeric domain of GluTR (GluTR DD ) in complex with FLU TPR . The FLU TPR -GluTR DD complex has a dissociation constant similar to that of FLU TPR with the full-length GluTR, which indicates that this complex structure can represent the FLU-GluTR interaction adequately. Together with an enzymatic ALA synthesis assay, these findings allow an elucidation of the molecular basis for FLU-mediated GluTR regulation.

Experimental Procedures
Protein Expression and Purification-The PCR-amplified cDNA fragments containing the TPR domain of Arabidopsis FLU (At3g14110) and the dimeric domain of Arabidopsis GluTR (At1g58290) were inserted into expression vector pET-22b(ϩ) (Novagen). Recombinant N-His 6 -tagged FLU TPR and GluTR DD were overexpressed in Escherichia coli BL21(DE3) and purified to homogeneity by nickel affinity column (Ni 2ϩ -nitrilotriacetic acid, Qiagen) and size-exclusion chromatography (GE Healthcare). For preparation of the FLU TPR -GluTR DD complex, pET-22b(ϩ) vectors containing N-His 6 -tagged FLU TPR and the non-tagged form of GluTR DD were co-transferred into E. coli BL21(DE3) cells. The purification procedure was the same as described above. Protein purity was monitored by SDS-PAGE. Proteins for crystallization were concentrated to ϳ12 mg ml Ϫ1 in buffer containing 20 mM Tris-HCl, 100 mM NaCl, 5% glycerol, and 1 mM DTT (pH 7.5) and then stored at Ϫ80°C.
Isothermal Titration Calorimetry-Experiments were performed on a MicroCal iTC200 calorimeter (GE Healthcare). The purified proteins for ITC measurement were in buffer containing 200 mM NaCl, 20 mM Tris-HCl, 5% glycerol, and 1 mM DTT (pH 7.5) and were degassed before titration. Each titration series consisted of 20 injections of 2 l of FLU TPR into 200 l of GluTR or GluTR DD . Control experiments were carried out by injecting FLU TPR into the buffer, and the resulting heat of dilution was subtracted from the binding isotherm data. The data were analyzed with the Origin software (OriginLab) using a single-site binding model.
Crystallization and Structure Determination-Crystals of FLU TPR and FLU TPR -GluTR DD were grown by the hangingdrop vapor diffusion method. Crystals of FLU TPR were formed within 1 week in 0.2 M NaCl, 0.1 M Bis-Tris (pH 6.5), and 25% (w/v) polyethylene glycol 3350. Crystals of FLU TPR -GluTR DD  were formed over 3 weeks in 0.15 M KBr and 30% (w/v) polyethylene glycol monomethyl ether 2000. Prior to data collection, crystals were cryo-protected in the above crystallization solution supplemented with 20% (v/v) glycerol and flash-frozen in liquid nitrogen. X-ray diffraction data were collected at beamline BL17U of the Shanghai Synchrotron Radiation Facility (SSRF) at a wavelength of 0.9793 Å at 100 K and then processed using HKL2000 (HKL Research). The model of FLU TPR was determined by molecular replacement using the program Phaser (15), and the search model was the TPR domain of the mouse Leu-Gly-Asn repeat-enriched protein (16) (Protein Data Bank (PDB) code 3RO3). The resulting model was rebuilt using ARP/wARP (17). Manual correction was performed in Coot (18) according to the 2F o Ϫ F c and F o Ϫ F c electron density maps, and further refinement was done with phenix.refine (19). The structure of the FLU TPR -GluTR DD complex was determined by molecular replacement using the FLU TPR structure as the search model, and the dimeric domain of GluTR was fitted into the density manually and then refined as described above. The overall quality of the final structural models was assessed by MolProbity (20) with 99.2 and 0.8% in favored and additional allowed regions for the FLU TPR structure, and 99.5 and 0.5% in favored and disallowed regions for the FLU TPR -GluTR DD complex structure (see Table 1). The protein structure figures were prepared using the program PyMOL (31).
ALA Synthesis Assay-The assay was performed as described previously (9), except that an equal volume of buffer was replaced by FLU TPR for measurement with FLU TPR . The molar ratio of FLU TPR :GluTR was kept as 1.2:1. The reaction time was 60 min.

Results
Crystal Structure of FLU TPR -The final model of the recombinant FLU TPR contains 119 residues (Pro 198 -Asp 316 ) and is refined to 1.45 Å resolution (Table 1). FLU TPR consists of three TPR motifs (TPR1-TPR3) and forms dimer through TPR3 (Fig.  1A). When compared with the prototype of TPR motif (21), these three FLU TPR motifs are composed of 40 residues rather than the canonical 34 residues (Fig. 1B). The three non-canonical TPR motifs are structurally conserved and can be well superimposed (Fig. 1C). The flu1-1 mutation (A262V) occurs on the absolutely conserved alanine of TPR2, and the tigrina d mutation, a frameshift at Ala 249 (corresponding to Ala 246 of FLU), results in loss of TPR2 and TPR3. The structure of FLU TPR clarifies the discrepancy among the predicted TPR domains from different species (1, 3, 4).
The FLU TPR -GluTR DD Complex-FLU directly interacts through its TPR domain with GluTR (2). To gain a structural insight into the FLU-GluTR interaction, we purified the fulllength GluTR in complex with FLU TPR (Fig. 2A). However, the FLU TPR -GluTR complex failed to crystallize. We speculated that the two relatively flexible arms of the V-shaped GluTR dimer could prevent crystal formation. Thus, GluTR truncations that constitute only the stem of the V shape were constructed, and a fragment corresponding to the C-terminal 104 residues (Leu 440 -Lys 543 ) was able to form stable complex with FLU TPR (Fig. 2B). The functional integrity of GluTR DD was tested by comparison of its binding affinity to FLU TPR with that of the full-length GluTR (Fig. 3). The K d value, as measured by isothermal titration calorimetry, is 2.03 Ϯ 0.10 M for FLU TPR and GluTR DD , similar to that for FLU TPR and GluTR (K d ϭ 1.66 Ϯ 0.13 M), indicating that GluTR DD maintains binding affinity to FLU TPR .
The crystal structure of the FLU TPR -GluTR DD complex was solved at 2.4 Å resolution ( Table 1). The 2:2 complex resembles two hands holding up a cup (Fig. 4), with two FLU TPR molecules clamping the GluTR DD dimer through their concave face. Because the concave face is a conventional site for TPR-mediated protein-protein interaction (22), the complex structure shows that despite its non-canonical TPR motifs, FLU uses a canonical binding mode for GluTR recognition. Because the GluTR DD structure starts from the middle of a long ␣-helix (Fig.  5), we attempted to crystalize GluTR truncations including this  JULY 10, 2015 • VOLUME 290 • NUMBER 28 whole ␣-helix but did not succeed. Notably, this long ␣-helix could undergo a wobbling motion (8,9), which suggests that the compact GluTR DD structure reported here is likely the stable core for the GluTR dimer.

FLU-GluTR Interactions
The FLU TPR -GluTR DD Interaction-The FLU TPR -GluTR DD complex structure reveals how FLU recognizes GluTR without significant conformational changes. When the structures of FLU TPR in its free and GluTR DD -bound forms are superimposed (Fig. 6A), the root mean square deviation of backbone C␣ atoms is only 0.79 Å. Rigidity is a common property for repeat proteins including TPR proteins (23). Interestingly, detailed analysis of the FLU TPR -GluTR DD structure demonstrates that the TPR2 of FLU TPR does not participate in direct protein-protein interaction (Fig. 6B). This differs from the usual concave binding pocket formed by TPR proteins (24). Only TPR1 and TPR3 are involved in GluTR binding, and the FLU TPR -Glu-TR DD interactions include five salt bridges: Arg 215 -Glu 462 , Glu 284 -Arg 485 , Asp 291 -Arg 450 , Glu 295 -Lys 447 , and Asp 307 -Arg 500 , and a hydrogen bond between Tyr 309 and Arg 450 (Fig.  6B). In a yeast two-hybrid assay, deletion of the C-terminal 43 residues of GluTR (Cys 501 -Lys 543 ) abolished interaction with FLU (7). The FLU-GluTR interaction, as represented by the FLU TPR -GluTR DD structure, does not require this C-terminal region, which is instead involved in GluTR dimerization. Deletion of this region would disrupt the GluTR dimer, suggesting that a 2:2 complex is the functional unit for FLU-mediated GluTR regulation.
FLU TPR Down-regulates GluTR Activity-GluTR is the ratelimiting enzyme for ALA formation (6). The inhibitory role of FLU on GluTR activity has been suggested physiologically in the flu mutant (1, 7). To experimentally address the effect of FLU on GluTR activity, we measured ALA synthesis by an in vitro coupled enzyme assay (Fig. 7A). The rate of ALA formation decreased significantly in the presence of FLU TPR when compared with the rate in the absence of FLU TPR . FLU TPR also repressed ALA formation when GluTR-binding protein (GluBP), a positive regulator of GluTR, was added to the assay. These results show that FLU and GluBP have antagonizing effects on GluTR activity and function independently from each other. Indeed, FLU exclusively binds to the C-terminal dimeric domain, whereas GluBP binds to the N-terminal catalytic domain of GluTR (9). Notably, the surface area of GluTR     that is shielded by FLU TPR is mainly composed of positively charged residues (Fig. 6B). Therefore, it is likely that FLU may prevent the negatively charged tRNA from binding to GluTR due to stereo clash (Fig. 7B).

Discussion
The biosynthesis of tetrapyrroles is vital for plant growth and development (10,11). Its biosynthetic pathway starts with a universal tetrapyrrole precursor, ALA, and branches into heme and chlorophyll pathways after the formation of protoporphyrin IX. The heme pathway requires only one further step, chelation with a ferrous iron, whereas the chlorophyll pathway needs several hallmark steps: chelation with a magnesium ion, formation of the fifth ring, light-dependent reduction of the fourth ring, and attachment of a phytol tail. FLU has been implicated in the formation of the universal precursor ALA and of PChlide, the product of the only light-dependent reaction in the chlorophyll pathway (1)(2)(3)(4)(5).
The non-canonical TPR motifs of FLU make them difficult to predict in previous studies (1,3). The 1.45 Å FLU TPR and 2.4 Å FLU TPR -GluTR DD structures reported here demonstrate how this three-repeat TPR domain recognizes the dimeric domain of the V-shaped GluTR dimer. The FLU-GluTR interaction, as represented by the FLU TPR -GluTR DD complex structure, indicates that the 2:2 complex is the functional unit for FLU-mediated GluTR regulation. In addition, the inhibitory effect of FLU on GluTR activity is confirmed by the in vitro ALA synthesis assay.
FLU has been identified as a negative regulator of chlorophyll biosynthesis over the past decades (1,5). Furthermore, Arabidopsis flu mutant (or tigrina d of barley) has become a strategy to study the singlet oxygen-mediated signaling pathway in plants (25,26). This strategy utilizes accumulation of the photosensitizer PChlide in the flu plants upon a dark/light shift. In the wild-type plants, PChlide does not accumulate because of FLU repression of GluTR. However, the link between PChlide accumulation and FLU-mediated GluTR inhibition remains unclear despite extensive research on PChlide-generated singlet oxygen responses (27). Based on an immunoprecipitation/ mass spectrometry study, a model incorporating enzymes in the chlorophyll biosynthetic pathway was proposed in which magnesium protoporphyrin IX monomethyl ester cyclase and PChlide oxidoreductase form a complex with FLU. This FLUcontaining complex would interact with GluTR upon accumulation of fluorescent PChlide, and thus inhibit ALA synthesis (14). Given the small size of FLU (27 kDa for mature protein) and its membrane-anchoring ability, to unravel its bridging role between membrane and GluTR regulation requires detailed information. The 2:2 FLU-GluTR complex model (Fig. 8) should provide a structural clue of how FLU acts as a bridge between GluTR and other chlorophyll synthesis enzymes. It is possible that the coiled-coil domain of FLU participates in binding of the other enzymes because the coiled coils are a protein motif involved in recognition between proteins (28).
Spatial regulation is a recently recognized strategy for GluTR (29). The way GluTR is anchored to the plastid membrane through FLU has been predicted (30). Based on the two states of FLU TPR (Figs. 1A and Fig. 4) and the inhibitory effect of FLU TPR on GluTR activity (Fig. 7A), the working model proposed by Kauss et al. (14) is further developed here. In this scenario, the FLU-containing complex senses PChlide accumulation and switches FLU from a resting (Fig. 1A) to an active (Fig. 4) state. The active FLU pulls GluTR from the stroma to the proximity of plastid membrane, where binding of the substrate glutamyl-tRNA is inhibited.