Crystal structure of the phospholipase A and acyltransferase 4 (PLAAT4) catalytic domain

Phospholipase A and Acyltransferase 4 (PLAAT4) is a class II tumor suppressor, that also plays a role as a restrictor of intracellular Toxoplasma gondii infection through restriction of parasitic vacuole size. The catalytic N -terminal domain (NTD) interacts with the C -terminal domain (CTD), which is important for sub-cellular targeting and enzymatic function. The dynamics of the NTD main (L1) loop and the L2(B6) loop adjacent to the active site, have been shown to be important regulators of enzymatic activity. Here, we present the crystal structure of PLAAT4 NTD, determined from severely intergrown crystals using automated, laser-based crystal harvesting and data reduction technologies. The structure showed the L1 loop in two distinct conformations, highlighting a complex network of interactions likely influencing its conformational flexibility. Ensemble refinement of the crystal structure recapitulates the major correlated motions observed in solution by NMR. Our analysis offers useful insights on millisecond dynamics based on the crystal structure, complementing NMR studies which preclude structural information at this time scale.

Intriguingly, PLAAT4 was recently identified as a restriction factor for intravesicular parasite Toxoplasma gondii infection, as its overexpression induces premature egress from parasitic vacuoles (Rinkenberger et al., 2021).The closely related PLAAT3 (also known as PLA2G16) is an essential host factor facilitating genome transfer during Picornaviridae infection (Staring et al., 2017).PLAAT3 is also important for organelle degradation during eye lens development (Morishita et al., 2021).These activities depend on both the catalytic activity and the recruitment to damaged membranes, which is regulated via their Cterminal hydrophobic domain (CTD) (Morishita et al., 2021;Rinkenberger et al., 2021;Staring et al., 2017).
The PLAATs are part of the LRAT-like sub-cluster in the NlpC/P60 superfamily of papain-like thiol proteases (Anantharaman and Aravind, 2003;Xu et al., 2011).This LRAT-like subcluster contains a circularly permutated NlpC/P60 domain (Anantharaman and Aravind, 2003;Xu et al., 2011) at their N-terminus, followed by a C-terminal hydrophobic domain of about 30 amino acids (CTD), which has been predicted to form a transmembrane region (Golczak et al., 2012;Pang et al., 2012;Uyama et al., 2012Uyama et al., , 2009bUyama et al., , 2009a) ) and plays an important role for both protein localization (Deucher et al., 2000;Sers et al., 1997;Tsai et al., 2007;Wei et al., 2015) and its physiological enzymatic function (Golczak et al., 2012;Morishita et al., 2021;Pang et al., 2012;Staring et al., 2017;Tsai et al., 2007;Wei et al., 2015).While the N-terminal NlpC/P60 domain houses the conserved active site residues, namely the catalytic Cys113 and His23 within the conserved NC-and H-box sequence motifs respectively (Hughes and Stanway, 2000;Uyama et al., 2009b), with His35 helping to orient His23, this soluble domain alone is not sufficient for full enzymatic activity (Golczak et al., 2012;Pang et al., 2012;Uyama et al., 2009b) (residue numbering conserved in PLAAT2-4).In isolation, the recombinant N-terminal domains (NTD) of the family members PLAAT2-4, are enzymatically competent, as evidenced by their self-acylation when exposed to short-chain phosphatidyl choline (Golczak et al., 2012).However, PLAAT4 shows robust phospholipase activity even when truncated (Chatterjee et al., 2021;Golczak et al., 2012), while PLAAT3 and PLAAT2 require the CTD for (appreciable) interfacial phospholipase activity (Pang et al., 2012;Uyama et al., 2009aUyama et al., , 2009b)).Furthermore, studies have shown that the CTD also plays a critical role for the targeting of the enzyme to different intracellular membrane compartments (Deucher et al., 2000;Jans et al., 2008;Morishita et al., 2021;Staring et al., 2017;Uyama et al., 2015;Wei et al., 2015).In undisturbed cells, PLAAT3 and PLAAT4 are distributed in the cytoplasm as well as in punctate structures colocalizing with peroxisomes (PLAAT3) or mitochondria (PLAAT4) (Wei et al., 2015).Truncation of the CTD of PLAAT3 and PLAAT4 abolishes membrane localization as well as the pro-apoptotic activities of the proteins (Wei et al., 2015).The corresponding CTDs (fused to the Cterminus of Blue Fluorescent Protein (BFP)) still target to membranes and are able to induce cell-death in HeLa cells at comparable levels, independent of their catalytic activity that resides in the NTD (Wei et al., 2015).In the context of the full-length proteins however, interactions between the NTD and CTD have opposing effects for PLAAT3 and PLAAT4: for PLAAT4 cell-death inducing ability was higher, while for PLAAT3 it was decreased as compared to BFP-CTD expressed in HeLa cells alone.This loss in cell-death inducing ability observed for PLAAT3 is restored upon truncation of the main L1 loop (residues 38-56; (Wei et al., 2015)), which is also predicted to interact with the membrane and help orient/tether the enzyme during catalysis (Golczak et al., 2012).A recent study of PLAAT3 and PLAAT4 NTD dynamics in solution (Chatterjee et al., 2021) demonstrated that despite structural similarity, these proteins exhibit different dynamics overall, which might help explain the difference in activity of their truncated form.In particular, the differential dynamics of the L2 (B6) loop (residues 100-109), which precedes the active site Cys113 and is in close proximity of the main L1 loop, is a critical determinant of enzymatic activity.Loop-swapped PLAAT3 (with the L2(B6) loop from PLAAT4) gains appreciably in phospholipase activity even in truncated form, while the reverse swap did not inhibit PLAAT4 (Chatterjee et al., 2021).
Existing studies illustrate a complex network of interactions between the different regions of these proteins contributing to their subcellular targeting, and enzymatic activities in cells.Differences in protein dynamics contribute to the regulation of enzymatic activity, possibly by increasing accessibility of the active site.However, structural information on the main L1 loop and its possible interactions with L2(B6) remains lacking.Here we present the crystal structure of the PLAAT4 NTD that was only obtainable through the use of modern automation both in hardware and software.The L1 loop is well ordered in our structure, in contrast with structures of PLAAT2 and PLAAT3 NTD where it is highly disordered.The L1 loop forms a complex network of short secondary structure motifs stabilized by extensive hydrogen bonding both within the loop and with residues from L2(B6).While protein dynamics of our crystal structure are naturally constrained by the crystal lattice, crystallographic ensemble refinement reveals flexible regions and recapitulates the overall dynamics seen in the NMR ensemble.

Recombinant protein production
Human PLAAT4 NTD (Uniprot ID Q9UL19 residues 1-123) was inserted into the pETNKI-6xHis-3C-ORF-amp LIC cloning vector (Luna-Vargas et al., 2011) for bacterial expression, which codes for an N-terminal His 6 -tag, followed by a HRV 3C protease recognition site.All clones were verified by DNA sequencing.Proteins were expressed in E. coli BL21 (DE3) T1 R (Sigma).Cultures were grown at 37 • C to an OD 600 of 0.6 in Luria Bertani medium supplemented with 100 µg ml − ampicillin.Expression was induced with 0.3 mM IPTG and cultures grown for further 18 h at 16 • C. Cells were harvested by centrifugation.Bacterial pellet was resuspended in lysis buffer (20 mM Hepes-NaOH pH7.4,300 mM NaCl, 1 mM TCEP), containing DNase I (Roche) and subjected to a freeze thaw cycle.Cells were lysed by sonication on ice.All subsequent purification steps were performed at 4 • C. The homogenate was clarified by centrifugation at 45,000 rcf for 30 min and affinity purified using chelating Sepharose beads (Cytiva) charged with NiCl 2 .Elution used 400 mM imidazole supplemented to the lysis buffer.Tag removal was by incubation with His 6 -HRV 3C protease overnight at 4 • C during dialysis against 20 mM Hepes-NaOH pH 7.4, 100 mM NaCl, 2 mM TCEP, followed by subtractive affinity purification to remove uncleaved protein, tag and protease from the sample.Samples were concentrated using centrifugal filters (Amicon, 10000 MWCO) and supplemented with 5 mM TCEP prior to size exclusion chromatography on a Superdex S75 column (Cytiva), equilibrated in 20 mM Hepes-NaOH pH7.4, 100 mM NaCl, 2 mM TCEP.All proteins eluted as a monomer and the main peak fractions were pooled.To remove a contaminant, which could not be separated during size exclusion due to their similar size, the protein was subjected to ion exchange chromatography on a ResourceQ column (Cytiva).The buffer of the protein samples was adjusted to mM NaCl, while the buffer system for the ion exchange was 20 mM Bis-Tris-HCl pH 6.4, 2 mM TCEP and 20-1000 mM NaCl.PLAAT4 NTD did not bind to the column, so the flow-through fractions were pooled and concentrated to ~ 24 mg ml − 1 , as judged by Abs 280nm measurements (using a theoretical extinction coefficient for the protein), supplemented with 5-10 mM TCEP, flash frozen in liquid nitrogen and stored at − 80 • C until required.

Crystallization
Following several failed attempts to produce diffraction quality PLAAT4 NTD crystals at our home laboratory using standard protocols and automation robotics (Newman et al., 2005), crystallization trials were performed at the High-Throughput Crystallization laboratory at the EMBL Grenoble outstation (Cornaciu et al., 2021;Márquez and Cipriani, 2013).The protein (~24 mg ml − 1 ) was incubated with 10 mM fresh TCEP for 30 min on ice prior to crystallization trials.Crystals were grown by vapor diffusion in CrystalDirect plates (Cipriani et al., 2012) at 20 • C mixing 100 nl of the protein sample with 100 nl of the mother liquor and equilibrated against 45 µl of reservoir solution.Crystals appeared within 2-6 days in condition 96 of the Morpheus screen (0.1 M amino acids, 0.1 M Tris-bicine pH8.5, 37.5 % v/v MPD PEG 1000 PEG 3350).Crystals were harvested and cryo-cooled automatically using the CrystalDirect technology (Zander et al., 2016) and included a laserbased, crystal-surgery approach for the separation of intergrown needle clusters (see results section).

Structure determination and refinement
Diffraction data for PLAAT4 NTD were collected at ESRF beamline ID30A-1 (Bowler et al., 2015;Svensson et al., 2015).Automated data reduction performed using the autoPROC pipeline (Vonrhein et al., 2011) yielded a 94 % complete dataset extending to 1.73 Å resolution (Table 1).Phasing was performed by molecular replacement in Phaser (McCoy et al., 2007) using the trimmed, refined model of PLAAT3 as a search model (7ZOM, with non-conserved residues truncated at C β positions).Manual model building in Coot (Emsley and Cowtan, 2004;Zander et al., 2016) was alternated with refinement using Refmac (Murshudov et al., 2011) to complete the model and adjust it to the correct sequence.The PDB-REDO webserver (Joosten et al., 2014) was used to determine optimal weights for refinement in Refmac.During later cycles of restrained refinement in combination with TLS refinement in Refmac, the number or TLS groups was increased from one to three groups per chain, based on analysis using the TLSMD server (Painter and Merritt, 2006).The TLS groups were chosen as follows: residues 6-21, 22-41 and 43-123 in chain A and residues 6-37, 38-55 and 56-123 in chain B. No density could be observed for the first five residues of either chain; therefore, the final model encompasses residues 6-123 for both chains, with six residues modelled with side chain alternates and was refined to an Rfree of 20.1 %.It has no Ramachandran outliers, a Ramachandran Z-score of 0.27 (Sobolev et al., 2020) and a Molprobity score of 0.90, ranking in the 100th percentile (Chen et al., 2010).Final refinements statistics are given in Table 1.Model coordinates and structure factors are deposited at the RCSB Protein Data Bank under accession number 7ZOT.To assess the conformational flexibility afforded the protein in the crystal lattice we also performed ensemble refinement using phenix.ensemble_refinement(Burnley et al., 2012) with the default settings, using the MX-PRESTO software stack to access HPC compute resources (https://www.nsc.liu.se/support/presto/MX-PReSTO/index.html).

Determination of oligomeric state in solution
To determine the oligomeric state of PLAAT4 NTD in solution, protein samples were subjected to Size Exclusion Chromatography coupled Small Angle Xray Scattering (SEC-SAXS) and Multi-Angle Light Scattering (SEC-MALS) at beamline EMBL-P12 (PETRA III, DESY, Hamburg, Germany).16 µl protein samples at a concentration of 10 mg ml − 1 were loaded onto an S75 Increase 5/150 analytical SEC column (Cytiva) in 20 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM TCEP, 1 % v/v glycerol with a flow rate of 0.3 ml min − 1 using the BECQUEREL beam line control software.The SAXS intensities were measured as 2400 × 1 s individual X-ray exposures using a Pilatus 6M 2D-area detector from the continuously flowing column eluent.The 2D-to-1D data reduction, i.e., radial averaging of the data to produce 1D I(s) vs s profiles, were performed using the SASFLOW pipeline incorporating RADAVER from the ATSAS 3.0 suite of software tools (Manalastas-Cantos et al., 2021).Data frames were processed using CHROMIXS (Panjkovich and Svergun, 2018) and Guinier analysis and P(r) distributions were fitted using PRIMUS (Konarev et al., 2003) (Supplementary Table 1).The crystal structure, both dimer and monomer, and NMR-models were fitted to experimental data using CRYSOL (Svergun et al., 1995) and SREFLEX (Panjkovich and Svergun, 2015) was used for flexible refinement.
A sEC-MALS experiment was performed by in-line measurements using a Wyatt Technologies Mini-Dawn TREOS multi angle light scattering detector coupled to an OptiLab T-Rex refractometer (RI).A 16 µl sample was injected onto a Superdex 75 Increase 5/150 column (Cytiva) equilibrated in 20 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM TCEP, 1 % v/v glycerol, with a flow rate of 0.3 ml min − 1 .The MALS system was used at the incident wavelength of 659 nm.Concentration estimates obtained from the RI (dn/dc = 0.185 ml g − 1 ), was used to evaluate the molecular weight (MW).The measurements were performed at 25 • C. The MW distribution of species eluting from the column was determined and analyzed using ASTRA7 software (Wyatt Technology).

Results and discussion
The PLAAT4 NTD protein crystallized readily in a range of conditions.However, despite extensive attempts at optimization, crystals recurrently grew in the habit of severely intergrown, thin plates and needles, that were too fragile for manual handling for separation and cryoprotection during harvesting (Fig. 1a).
Ultimately, use of the Crystal direct technology (Cipriani et al., 2012;Cornaciu et al., 2021;Márquez and Cipriani, 2013;Zander et al., 2016) offered at the High-Throughput Crystallization Laboratory at the EMBL Grenoble outstation (https://htxlab.embl.fr/)was critical for obtaining diffraction data.In this approach a laser was used to recover crystals growing on a thin film within a CrystalDirect plate by excising the film around the crystal and attaching it automatically to the tip of a data collection pin.The laser from the CrystalDirect instrument was used also as a surgical tool to separate a larger section of crystals from clusters of smaller and tightly intergrown needles (Fig. 1a-b).As the laser works in photoablation regime, heat transfer is minimized and the quality of the sample is preserved (Zander et al., 2016).Selected crystal clusters were mounted directly without manual intervention and transferred for automated data collection at beamline MASSIF-1 (ESRF ID30-A) (Bowler et al., 2015;Svensson et al., 2015) (Fig. 1b).Despite the "surgical" isolation of the less intergrown part of the crystals cluster, the diffraction data displayed multiple lattices that proved challenging for data processing (Fig. 1c).Uniquely, automated data processing with the Autoproc pipeline (Vonrhein et al., 2011) using 10 % of the initial 98,518 spots for indexing, allowed the integration of diffraction intensities belonging to a single lattice, and yielded a 94 % complete 1.73 Å dataset with good statistics (Table 1).Further structure elucidation proved straightforward, with phasing by Molecular replacement using the core domain of PLAAT3 (7ZOM) as a search model.
PLAAT4 NTD crystallized in space group C2 with two molecules in the asymmetric unit.Compared to the crystal structures of the closely related PLAAT3, where the central loop (L1) encompassing residues 40-57 is disordered, we observed clear electron density which allowed us to model L1 in both chains, through several iterative cycles of manual model building and refinement (Fig. 2a).Both molecules in the asymmetric unit agree well overall with each other (r.m.s.d. of ~ 0.3) as well as with the NMR structure of PLAAT4 1-125 (2MY9, r.m.s.d. of ~ 0.8, Fig. 2c), within the core domain (i.e., excluding L1).As previously described (Wei et al., 2015), the N-terminal half of the protein folds into an antiparallel β-sheet formed by 6 β-strands in total, while the C-terminal half is mainly α-helical and folds against the back of the β-sheet.
The region that diverges the most is L1, which adopts two distinct conformations in the two monomers in the asymmetric unit, probably at least partially due to their distinct crystal lattice environments (Fig. 2d).These two conformations are characterized by the formation of multiple sequential β-turns, most likely favored by the many serines in this stretch of PLAAT4 (Fig. 2e), which enable extensive hydrogen bonding networks stabilizing these short structural motifs (Fig. 2e, Supplementary Fig. 1).L1 is predicted to interact with the lipid membrane during interfacial catalysis and Golczak et al (Golczak et al., 2012) proposed a model for PLAAT3 membrane interaction, where residues 50-54 are predicted to form an amphiphilic α-helix that embeds into the membrane.In our structure of PLAAT4, this stretch forms three sequential and partially overlapping β-turns, further stabilized through several hydrogen bonds with the sidechain of Lys105 of the L2(B6) loop, in molecule A (with main chain carbonyl of Val53 and Ser55 as well as the Oγ of Ser55 in conformation B), while it adopts a 3 10 helix conformation in monomer B, confirming its α-helical propensity (Fig. 2e).These short structural motifs introduce a reduction in degrees of freedom which might contribute to the ordered packing of L1 within the crystal lattice.Taking a closer look at the arrangement of the two molecules in the asymmetric unit reveals that the core domains are related by close to internal 2-fold symmetry (rotation by 178.82 • and a 0.96 Å shift).The L1 loops break this non-crystallographic symmetry, adopting different conformations, as they pack against the β1-β2 hairpin loop, and along the outside edge of the β-sheet along strands β1 and β5 of the other molecule.In total the buried surface is close to 950 Å 2 , with L1 participating in the vast majority of interactions seen in the interface.Given this extensive interaction between the two molecules, it is not surprising that analysis in PISA (Krissinel and Henrick, 2007) indicates either a dimer, or possibly even a tetramer as the likely biological assembly (Fig. 2d).However, analysis of PLAAT4 NTD in solution using both sEC-MALS and sEC-SAXS shows that this protein is monomeric (Fig. 3ab), hinting that the close interaction between molecules in the crystal lattice might just be a way to bury the rather non-polar residues of L1 in the absence of lipid membranes, further promoted by the high concentrations required for crystallization.The dimensionless Kratky plot (Fig. 3c) and P(r) versus r profile (Fig. 3d) indicate that in solution PLAAT4 NTD is mainly well-folded, adopting an extended conformation (Dmax = 7.5 nm).This hints that the L1 loop and N-terminus must be flexible and sampling different conformations, agreeing with what is seen in the NMR ensemble.
In a recent study of PLAAT4 and PLAAT3 NTDs in isolation, Chatterjee and colleagues (Chatterjee et al., 2021) showed that the differences in dynamics of L2(B6) (residues 103-111), preceding the catalytic cysteine (Cys113) are contributing to PLAAT4 activity as a phospholipase in its truncated form, in contrast to truncated PLAAT3 that does not show appreciable catalytic phospholipase activity.Closer analysis of potential correlated interactions and dynamics between L1 and L2(B6) was hindered by the fact that due to motional chemical exchange, the majority of the resonances for the residues in L1 and L2(B6) could not be measured/detected due to line broadening.
Our crystal structure offers two different "static snapshots" with the main loop in two distinct conformations in the different monomers that were captured within the constraints of the crystal lattice.Either structure of the two monomers observed in the crystal, or individual conformations of the NMR ensemble, do not explain the SAXS scattering curve in solution (χ 2 3.004 -9.74; average of the NMR ensemble: χ 2 = 11.401),suggesting that many conformations exist in solution (Fig. 3a).Consistently, flexible modelling with normal mode analysis using SRE-FLEX could yield models that fit the observed scattering curves reasonably well (χ 2 1.001-1.234;Fig. 3a, e-g).
To explore the conformational flexibility afforded to the molecules within the crystal lattice, and see if that would explain the solution structures ensemble, we then employed crystallographic ensemble refinement (Burnley et al., 2012) using default parameters.The resulting ensemble contained 28 structures for each chain and had a final R-factor of 0.1570 and R free = 0.1976 and a mean Molprobity score of 3.49.Comparing the resulting ensemble with the NMR ensemble structure of PLAAT4 1-125 (2MY9 (Wei et al., 2015)) reveals that ensemble refinement can approximate the major motions sampled in solution within the crystal lattice (Fig. 4), and indeed reproduces the fit to the SAXS data equally well as the most representative NMR conformer after SREFLEX modeling (χ 2 = 0.99; after rigid body fitting alone χ 2 was in the range of 24.3-32.2for individual states of each chain in the ensemble, and 64.8 for the average of the ensemble).
As expected, the two molecules exhibit some variation in their conformational dynamics, depending on the crystal environment and lattice interactions.Both molecules recapitulate the flexibility of L1, which samples a wide variety of conformations in both molecules, including a hint of the dynamics seen for L2(B6) in solution, particularly in molecule B. Of particular interest are the dynamics of Tyr106, which might be representative of the dynamics for the whole L2(B6) loop, and which according to previous structural analysis of Papain-like NlpC/P60 proteins by Xu et al (Xu et al., 2011) might function as a fourth catalytic residue in this protein superfamily.In all structures of PLAAT3 NTD reported to date, Tyr106 is facing towards the outside of the molecule, pointing away from the active site, with very little fluctuation seen either between the crystal structures or within the NMR ensemble.In PLAAT4 NTD on the other hand, this loop exhibits concerted fluctuations that propagate to the neighboring catalytic Cys113.While L2(B6) dimer as seen in the asymmetric unit and estimated as the likely biological assembly by PISA.The core domain of chain A is shown in blue and chain B is shown in orange, with L1 depicted as sticks and the surface of the core domain is depicted as a semi-transparent surface.The internal approximate twofold axis of symmetry is depicted as a black diamond (e) Close up of L1 and its packing against the second monomer in the asymmetric unit, coloring scheme as in (d), with L2(B6) shown in black and the sidechain of Lys105 shown as magenta sticks.H-bonds are depicted as dashed lines, in yellow for the interaction between Lys105 from L2(B6) with L1 residues, and H-bonds within L1 depicted in red.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)exhibits large conformational diversity within the NMR ensemble, Tyr106 never adopts the same conformation on the outside of the molecule along α3 as seen in all structures of PLAAT3 NTD to date.This dynamic behavior of the L2(B6) is also reproduced in our ensemble refined structure, though the conformational variability is appreciably smaller within the crystal lattice (Fig. 4).Nevertheless, analysis of the ensemble still allows us a glimpse at the dynamics of these two loops that have been shown to be important for regulating protein activity.To gain a proper understanding how these dynamic loops and the CTD, which is critical for membrane targeting and interfacial activity, modulate enzymatic activity, further structural and biophysical studies in the context of the full-length proteins are needed.We can expect that protein dynamics will play a major role in regulating PLAAT protein activity and localization.The crystal structure reported uncovers dynamic regions even under the constraints of the crystal lattice and can give useful hints towards the dynamics observed in solution, with a smaller investment of compute time.These can be a complementary tool to NMR studies when investigating dynamics in the millisecond time scale, where NMR measurements can be problematic due to line broadening.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.(Krissinel, 2012).R.m.s.d.differences between chain A and the NMR ensemble are plotted as blue triangles, differences between Chain B and the NMR ensemble are plotted as orange rectangles and differences between the two chains in the crystal structure are plotted as black crosses.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 1 .
Fig. 1.Crystal cluster mounted using the CrystalDirect method and its diffraction pattern (a) Crystals grew as intergrown needle and plate clusters.(b) Film area with crystal cluster after automated harvesting (c) Representative multiple lattice diffraction pattern from the crystal cluster, which yielded a 94% complete single lattice dataset used to solve the crystal structure of PLAAT4 NTD.

Fig. 2 .
Fig. 2.An overview of the PLAAT4 NTD structure models highlighting the distinct conformations of the L1 loop between the two monomers in the crystal structure (ab) and the NMR ensemble (2MY9; best representative model shown in c).All models have been aligned to the same orientation and are depicted as cartoon using a modified rainbow coloring scheme with β-strands colored from dark blue to turquoise and α-helices colored from yellow to red from N-to C-terminus.(a-b) Crystal structure monomers A (panel a) and B (panel b) with L1 residues depicted as sticks and the corresponding |(Fo)-(Fc)| omit map for L1 residues 38-56 is shown as green mesh contoured at 3.0 σ.The active site residues His23, His35 and Cys113 are highlighted as green sticks.(d-e) Cartoon representation of the crystallographic Investigation, Visualization.Irina Cornaciu: Investigation, Resources, Visualization.José Antonio Marquez: Resources, Funding acquisition.Anastassis Perrakis: Conceptualization, Funding acquisition.Eleonore von Castelmur: Conceptualization, Investigation, Visualization, Writingoriginal draft, Funding acquisition.

Fig. 3 .Fig. 4 .
Fig. 3. sEC-SAXS and sEC-MALS showing PLAAT4 NTD is monomeric in solution.(a) sEC-SAXS scattering curve and theoretical scattering curves of the crystallographic dimer (red), monomer (chain A, marine, blue) and NMR model (state 1, purple, lilac) with and without modelling of conformational flexibility using normal mode analysis (NMA).The best fit (state 17) after ensemble refinement (ER) and NMA analysis of the crystal structure is plotted in light blue.Apparent molecular weight 15.65 kDa.The Guinier region of the scattering data (inset) is linear and consistent with a monodisperse solution.(b) sEC-MALS analysis of PLAAT4 NTD with Rayleigh ratio shown in red and differential refractive index (dRI) in blue with fitted Mw (15.7 ± 0.3 kDa) plotted as orange dots across elution peaks.The theoretical Mw of monomeric PLAAT4 NTD is plotted as a dashed line (14.0 kDa).(c) Dimensionless Kratky plot and (d) P(r) distribution (D max = 7.5 nm) obtained from sEC-SAXS measurements.Mode vectors illustrate the modeled movement after NMA analysis for (e) crystal structure chain A (f) ensemble refinement crystal structure chain A (state 17) (g) NMR model (state 1).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1
Data collection and refinement statistics.