Crystallographic analysis and phasing of iron-assimilating protein 1 (FEA1) from Chlamydomonas reinhardtii

In Chlamydomonas reinhardtii, iron levels are strictly controlled by iron-assimilating protein 1 (FEA1). Recombinant FEA1 protein was successfully purified and crystallized by hanging-drop vapor diffusion. The phase problem was solved by the native sulfur SAD method using long-wavelength X-rays (2.7 Å). Laser-cutting technology was used to increase the signal-to-noise ratio and derive an initial structure.


Introduction
The green alga Chlamydomonas reinhardtii is important in many fields of research, including metal metabolism. Metals play essential roles in cells as part of protein cofactors; therefore, their concentrations are tightly controlled as an excess amount can be toxic, while a deficiency leads to inactive metalloenzymes (Hanikenne, 2003;Merchant et al., 2006). As a photosynthetic organism, C. reinhardtii has various irondependent enzymes with vital functions in electron pathways, reactive oxygen detoxification, fatty-acid metabolism and amino-acid biosynthesis (Glaesener et al., 2013).
Rubinelli and coworkers initially characterized the regulation of the expression of HCR1-like protein (H43) (now called iron-assimilation protein 1; FEA1), which is induced by an iron-deficient medium (Rubinelli et al., 2002). FEA1 is a homologue of HCR1 from the marine alga Chlorococcum littorale. Both proteins are highly induced by high CO 2 levels and iron deficiency (Rubinelli et al., 2002;Kobayashi et al., 1997;Baba et al., 2011;Sasaki et al., 1998). Studies have shown that the abundance of FEA1 mRNA and protein is greatly increased under conditions of iron deficiency in C. reinhardtii (Allen et al., 2007;Urzica et al., 2012). Importantly, it has been shown that the FEA1 protein is the major protein secreted into the periplasm by iron-deficient C. reinhardtii cells, and is expressed coordinately with the FRE1 and FOX1 genes (Allen et al., 2007). It has also been shown to be N-glycosylated (Mathieu-Rivet et al., 2013).

ISSN 2053-230X
The exact molecular mechanism of FEA1 in iron assimilation by C. reinhardtii is unknown. Its currently proposed function is to concentrate iron in the periplasmic space in the vicinity of iron transporters in order to increase the efficiency of iron uptake under conditions of iron deficiency (Allen et al., 2007). Although FEA1 is not homologous to classical transferrins, a recent study identified ISIP2A as a phyto-transferrin in diatoms functioning in carbonate-dependent iron assimilation, and phylogenic analysis indicates that FEA1 and ISIP2A are related (McQuaid et al., 2018), potentially offering new hypotheses about the function of C. reinhardtii FEA1.
Structural characterization of FEA1 will be important to gain a deeper understanding of its role in the iron-assimilation pathway. In crystallography, protein structures are mostly solved by molecular replacement; however, proteins with no homology to structures in the Protein Data Bank (PDB) require experimental phasing. One such phasing method is single-wavelength anomalous dispersion (SAD) using selenomethionine (SeMet) or heavy-atom derivatized crystals. However, if derivatization with SeMet or additional heavy metals, such as mercury or platinum, is not possible, the only option left is native SAD using the S atoms of methionine or cysteine (S-SAD), a method that has been under development since its first reported use to determine the structure of crambin in 1981 (Hendrickson & Teeter, 1981). In fact, a number of protein structures deposited in the PDB have been solved by native SAD (Rose et al., 2015). In SAD, using a long wavelength (2.7 Å ) can increase the anomalous signal (f 00 of sulfur is 1.5 e -) relative to shorter wavelengths; however, both the crystal thickness and the solvent surrounding the crystal may cause strong absorption effects prior to data collection at long wavelengths (Basu et al., 2019). Laser-cutting technology might be the answer to this problem: it can improve the data quality in crystallographic data processing, decrease the dispersion of angle-dependent scale factors and increase the signal-to-noise ratio (Kitano et al., 2005;Basu et al., 2019).
We found that FEA1 can only be crystallized in the apo form without bound Fe atoms; therefore, we first attempted experimental phasing using SeMet and heavy-atom derivatization. As these approaches failed, we then used S-SAD to phase the diffraction data from FEA1 crystals. Here, we report the crystallization and successful phasing of FEA1 using S-SAD at a long wavelength of 2.7 Å coupled with laser-cutting technology.

Macromolecule production
The gene fragment encoding the FEA1 protein (Gln19-Ala362 of UniProt Q9LD42) was inserted into plasmid pASK-IBA2 (IBA). The FEA1 plasmid was designed to express the periplasmic protein with a Strep-tag at the C-terminus. Information on plasmid construction and expression is summarized in Table 1. The recombinant plasmid pASK-IBA2 for FEA1 was transformed into Escherichia coli Rosetta 2 (DE3) cells. A preculture was prepared from a fresh colony incubated at 26 C overnight and was used to inoculate the main culture. The cells were grown in Luria-Bertani medium at 26 C and, when the optical density at 550 nm (OD 550 ) had reached 0.5-0.6, protein expression was induced by adding anhydrotetracycline (to a final concentration of 200 ng ml À1 ) and incubating at 26 C for 16-18 h. The cell pellet was collected by centrifugation at 4000g at 4 C for 12 min.
For protein purification, the cells were suspended in 100 mM Tris-HCl pH 8.0, 500 mM sucrose, 2 mg ml À1 polymyxin B sulfate and incubated on ice for 30 min. The periplasmic protein was collected by centrifugation at 149 000g at 4 C for 30 min. The supernatant was applied onto an equilibrated Strep-Tactin column (IBA), contaminants were washed through with 100 mM Tris-HCl pH 8.0, and the FEA1 protein was eluted with 100 mM Tris-HCl pH 7.5, 2.5 mM desthiobiotin. For further purification, the eluted fraction was applied onto an SP HP column (GE Healthcare) and the protein was eluted by a 0-500 mM linear gradient of NaCl in 100 mM Tris-HCl pH 7.5. The purified FEA1 protein was concentrated in 40 mM Tris-HCl pH 7.5 using an Amicon Ultra-15 Centrifugal Filter Unit (10 000 molecular-mass cutoff; Merck Millipore). The protein concentration was estimated by the bicinchoninic acid (BCA) assay.

Crystallization
Initial screening for FEA1 crystals was performed by the sitting-drop vapor-diffusion method using ten commercial screening kits, Crystal Screen, Crystal Screen 2, PEG Rx 1 and 2, PEG/Ion and PEG/Ion 2 (Hampton Research, USA) and Wizard I, II, III and IV (Rigaku, USA), in a 96-well plate using a Mosquito LCP crystallization robot (TTP Labtech). The FEA1 concentration used for crystallization was 10 mg ml À1 in 40 mM Tris-HCl pH 7.5. Droplets consisting of 0.2 ml protein solution and 0.2 ml reservoir solution were equilibrated against 80 ml reservoir solution at 4 and 20 C. Based on the crystals obtained in the initial screening, crystallization was optimized by the hanging-drop vapor-diffusion method using different  precipitating agents and additives around the initial conditions. Ultimately, a mixture of 1 ml protein sample and 1 ml reservoir solution was equilibrated against 150 ml reservoir solution, and crystals of FEA1 were obtained in 200 mM ammonium sulfate, 100 mM imidazole-HCl pH 6.5, 11%(w/v) PEG 3350, 30%(v/v) MPD at 4 C. The crystallization conditions are summarized in Table 2.

Data collection and processing
An FEA1 crystal was picked up by a loop and quickly cooled in liquid nitrogen for data collection. Diffraction data were collected on beamline BL44XU at SPring-8, Harima, Japan using an EIGER X 16M system (Dectris, Baden, Switzerland). S-SAD data were collected on beamline BL-1A at the Photon Factory (PF), Tsukuba, Japan using an EIGER X 4M system (Dectris, Baden, Switzerland). To increase the signal-to-noise ratio, the loop was cut by a laser on station AR-NW12A at the PF. All diffraction images were collected at 100 K and were processed, merged and scaled using XDS/ XSCALE (Kabsch, 2010). The initial model was determined by SAD using autoSHARP followed by autoBUSTER (Vonrhein et al., 2007;Bricogne et al., 2011). It was then used as a search model to solve a native data set collected at a higher resolution on beamline BL44XU at SPring-8, Hyogo, Japan by molecular replacement using Phaser (McCoy, 2017) as part of the CCP4 suite (Winn et al., 2011). Data-collection and processing statistics are summarized in Table 3.

Results and discussion
FEA1 is a periplasmic protein encoded by the h43 gene (UniProt Q9LD42) in Chlamydomonas. Notably, the construct used in this study comprised amino-acid residues Gly19-Ala362 of FEA1 because the first 18 residues of the N-terminal region are considered to be the signal peptide and were thus omitted. FEA1 was constructed in a pASK-IBA2 vector and heterologously expressed in E. coli Rosetta (DE3) cells as a periplasmic protein with a Strep-tag (LEVDL QGDHGLSAWSHPQFEK) attached to the C-terminus. The recombinant protein was purified using a Strep-Tactin column,    followed by cation-exchange chromatography using an SP HP column. The purified protein was checked by 12.5% SDS-PAGE, which showed a single band with a molecular weight of 38.6 kDa (Fig. 1). The purified protein was crystallized by the hanging-drop vapor-diffusion method in 200 mM ammonium sulfate, 100 mM imidazole-HCl pH 6.5, 11%(w/v) PEG 3350, 30%(w/v) MPD at 4 C. The crystals were obtained with good reproducibility and diffracted well to 1.9 Å resolution, as shown in Fig. 2. Because there are no proteins in the PDB with a similar amino-acid sequence, we carried out experimental phasing. Firstly, we tried to express SeMet-substituted FEA1 under conditions of methionine-pathway inhibition. However, the expression level was very low, and the amount of sample obtained was not sufficient for crystallization. Next, several heavy-atom derivatives were prepared, including those with thallous acetate, mercury(II) acetate, potassium tetrachloro-platinate(II), iron(III) chloride and iron(II) sulfate. Unfortunately, none of them were successful in solving the phase problem.
Because FEA1 has 13 S atoms, we next tried S-SAD to determine the initial phase. Diffraction data for S-SAD were collected using an EIGER X 4M system (Dectris, Baden, Switzerland) at the long wavelength of 2.7 Å on beamline BL-1A at the Photon Factory, Tsukuba, Japan. The diffraction data were processed using XDS. 27 unmerged data sets were then analyzed using XSCALE_CLUSTER to find isomorphous data sets. In the first attempt, eight data sets were used from the largest isomorphous cluster from one crystal. The data sets comprised 3600 frames with oscillation angles of 0-360 taken at different crystal positions and kappa angles; however, a native SAD solution was not obtained. In the second attempt, a loop was cut by a laser under the cryogenic conditions to remove the solvent on station AR-NW12A at the Photon Factory, Tsukuba, Japan prior to data collection in order to reduce scattering by the solvent. The crystal of FEA1 was a thin rod-shaped crystal with dimensions of 0.2 Â 0.024 mm as shown by a black dotted line in Fig. 3. The solvent around the crystal was removed as much as possible, and the final distance between the cutting line and the crystal was    about 0.032-0.039 mm. After the laser-cutting process, 29 data sets were collected using the solvent-removed crystals. 16 isomorphic data sets from the first and second data collections were then merged and scaled at 2.6 Å resolution using XSCALE. Finally, the phase was solved by SAD using auto-SHARP followed by autoBUSTER (Vonrhein et al., 2007;Bricogne et al., 2011). There were 1060 residues in seven chains, with 1032 residues of the FEA1 sequence in the initial structure (Fig. 4).
The crystal of FEA1, which diffracted to 1.9 Å resolution, belonged to space group C2, with unit-cell parameters a = 85.75, b = 155.94, c = 129.53 Å , = 102.27 . The Matthews coefficient of the FEA1 crystal was 3.37 Å 3 Da À1 , with three molecules in the crystallographic asymmetric unit and 63.5% solvent content (Matthews, 1968). The initial structure from the SAD solution was used as a search model for higher resolution data sets using Phaser (McCoy, 2017). Model rebuilding and refinement are ongoing.
In summary, we have used S-SAD and laser-cutting technology to solve the phase problem in the crystallographic analysis of FEA1 crystals. This study will lead to further structural studies of FEA1 to understand its function and its links to the iron-assimilation pathway.

Figure 4
Initial structure of FEA1 solved by S- SAD. (a, b) The electron-density map is contoured around the initial structure. (c) The experimental electron-density map from autoBUSTER. The figures were generated using PyMOL.