Structures of Brucella ovis leucine-, isoleucine-, valine-, threonine- and alanine-binding protein reveal a conformationally flexible peptide-binding cavity

Brucella ovis leucine-, isoleucine-, valine-, threonine- and alanine-binding protein structures have a prototypical bacterial periplasmic amino acid-binding protein topology with a conformationally flexible peptide-binding cavity in the absence of peptide.


Introduction
Brucellosis is highly contagious and affects both economically important livestock and wild animals (Ducrotoy et al., 2017;Godfroid, Garin-Bastuji et al., 2013;Godfroid et al., 2011;Megersa et al., 2011;Rossetti et al., 2022).Even when not resulting in significant zoonotic disease, as in the case of Brucella ovis (sheep and rams), brucellosis is economically devastating globally (Peck & Bruce, 2017;Franc et al., 2018).While brucellosis has been eradicated in cattle and small ruminants in some industrialized countries, it remains endemic globally within many animal hosts (Moreno, 2014).Current control approaches for brucellosis include vaccination, education and basic hygiene.However, these strategies have yet to reduce the disease burden successfully due to the high costs, the ineffectiveness of current antibiotics in the latter stages of brucellosis and other factors (Ariza et al., 2007;Franc et al., 2018).Notably, current vaccines are species-specific and are devastating to pregnant livestock, and some animal care practices of rural dwellers and nomadic groups are incompatible with controlling brucellosis in humans and livestock (Ducrotoy et al., 2017;Godfroid, Al Dahouk et al., 2013).
B. ovis is nonpathogenic in humans but is devastating globally to sheep, rams, goats, small ruminants and deer by causing ovine epididymitis (Rossetti et al., 2022).Similarly, B. melitensis, which causes fatal zoonotic disease in humans, also causes ovine epididymitis (Rossetti et al., 2022).Brucella are classified as category B infectious agents that can be aerosolized, and these small Gram-negative, facultative coccobacilli were the first bacterial agent to successfully be developed for biological warfare by the United States (de Figueiredo et al., 2015;Riedel, 2004).There is a continued need for brucellosis treatments in infected people and livestock.New approaches include the rational design or repurposing of small molecules that target proteins that are vital for bacterial survival.Towards these ends, the Seattle Structural Genomics Center for Infectious Disease (SSGCID) has determined the crystal structures of over 120 potential target proteins from different Brucella species.These structures provide a wealth of data for functionally and structurally characterizing Brucella proteins that are potential therapeutic targets and provide insights into fundamental mechanisms that can be used for drug discovery.These structures are used to engage undergraduates in structure analysis and scientific communication (Brooks et al., 2022;Davidson et al., 2022;Maddy et al., 2022;Porter et al., 2022;Beard, Bristol et al., 2022;Beard, Subramanian et al., 2022).Here, we present highresolution crystal structures of B. ovis leucine-, isoleucine-, valine-, threonine-and alanine-binding protein (BoLBP).BoLBP is a putative periplasmic amino acid-binding protein with less than 29% sequence identity to any previously reported structure.We report high-resolution structures of BoLBP in orthorhombic and monoclinic space groups that reveal a prototypical periplasmic amino acid-binding protein.

Crystallization
Purified BoLBP was screened for crystallization in 96-well sitting-drop plates against commercially available screens, including JCSG+ HTS (Rigaku Reagents) and MCSG1 (Microlytic).Vapor-diffusion experiments consisted of equal volumes of protein solution (0.4 ml) and precipitant solution set up at 290 K against an 80 ml reservoir.The crystals were flash-cooled by harvesting and plunging them into liquid nitrogen after passing through Al's oil or soaking in cryo-    soaking solution and 20%(v/v) ethylene glycol before vitrification in liquid nitrogen and data collection (Table 2).Future studies will include co-crystallization and harvesting at different temperatures to identify conditions that may enhance amino-acid binding to BoLBP.

Data collection and processing
For the orthorhombic structure, two data sets were collected: one at 100 K on beamline 21-ID-F at the Advanced Photon Source, Argonne National Laboratory (APS), while the phasing data set was collected on a rotating-anode home source (Table 3).The monoclinic data set was collected on beamline 21-ID-F at APS.All diffraction data were integrated using XDS and reduced using XSCALE (Kabsch, 2010).Raw X-ray diffraction images are available from the Integrated Resource for Reproducibility in Macromolecular Crystallography at https://www.proteindiffraction.org.

Structure solution and refinement
The structure of the monoclinic conformation was phased de novo by single-wavelength anomalous dispersion (SAD) after iodide ion soaks (Abendroth et al., 2011).Iterative refinement cycles with Phenix (Adams et al., 2011) and manual rebuilding using Coot (Emsley & Cowtan, 2004;Emsley et al., 2010) generated the model coordinates and structure factors deposited in the Protein Data Bank as entry 4xfk.The orthorhombic structure was phased by molecular replacement using the monoclinic structure as the search model and the Phaser software (McCoy et al., 2007) from the CCP4 suite of programs (Collaborative Computational Project, Number 4, 1994;Krissinel et al., 2004;Winn et al., 2011;Agirre et al., 2023).After iterative refinement cycles with Phenix (Adams et al., 2011) and manual rebuilding using Coot, orthorhombic coordinates and structure factors were deposited in the Protein Data Bank as entry 7jfn.Both structures were checked using MolProbity (Williams et al., 2018).The final refinement data are reported in Table 4.

Results and discussion
BoLBP resembles a prototypical periplasmic amino acidbinding protein, with a bilobate architecture of two major globular domains forming a Venus flytrap conformation around a large central cleft containing the peptide-binding pocket (Trakhanov et al., 2005).Both domains (domains I and II) have a characteristic �/� fold consisting of a central antiparallel �-sheet flanked by �-helices (Fig. 1).�-Strands from each sheet run towards the central cleft, exhibiting the characteristic LBP left-handed propeller twist that connects domains I and II through three interdomain loops: loop 1 (L1; residues 224-229), loop 2 (L2; residues 384-393) and loop 3 (L3; residues 423-426) (Fig. 1b).Loops 1 and 3 extend from domain I to domain II, while loop 2 traverses in the opposite direction.Loops 1 and 2 are preceded by a �-sheet strand in one domain and succeeded by an �-helix in the other domain, whereas loop 3 spans strands from both domains (Fig. 1b).
An acetate molecule from the crystallization solution sits in the central cleft in the monoclinic structure determined without soaking with amino acids (PDB entry 4xfk; Supplementary Fig. S1b).Threonine does not bind upon soaking the orthorhombic crystals with threonine; instead, a nitrate from the crystallization solution occupies the central binding cavity (PDB entry 7jfn; Supplementary Fig. S1a).Both ligands have well ordered electron density in 2F o À F c maps (Supplementary Figs.S1b and S1c).The two BoLBP structures are similar, with a root-mean-square deviation (r.m.s.d.) value of 1.49A ˚on aligning C � atoms.The main differences in the structures are in domain II, which rotates around the hinge and has a more open central cavity in the orthorhombic structure, with main-chain movements of up to 4.8 A ˚(Fig.1c).The differences in the structures are not as large as the conformational changes that are expected when LBP-like proteins transition from an open to a closed conformation upon binding their peptide ligands (Magnusson et al., 2004).While neither of the BoLBP structures binds an amino acid, both accommodate different ligands from the crystallization solution.
The two structures were compared using DynDom (https:// dyndom.cmp.uea.ac.uk/dyndom/; Lee et al., 2003, Qi et al., 2005).DynDom analysis revealed hinge rotation by a 12 � angle and conformational plasticity of the ligand-binding cavity in the structures, representing a transition from a 'closed' to a 'semi-closed' state.Further details of the hingebending residues and DynDom analysis results are presented in Section S2.
Due to the low sequence similarity of BoLBP to all other reported structures, ENDscript (Gouet et al., 2003;Robert & Gouet, 2014) analysis was used to identify its closest structural neighbor (Fig. 2).The analysis identified the closest structural neighbor of BoLBP to be E. coli LBP (EcLBP; Sack et al., 1989), and despite sharing less than 29% sequence similarity both have a similar overall topology (Fig. 2).Additionally, EcLBP and BoLBP share numerous identical residues (Fig. 2).Interestingly, BoLBP also has key amino-acid insertions resulting in longer helices and additional strands that were not previously observed in EcLBP (Fig. 2).The superposed structures show the Venus flytrap conformation around a large central cleft containing the peptide-binding pocket.Structural alignment reveals that the peptide-binding site is occupied by components of the crystallization buffer (acetate and nitrate) in our BoLBP structures (Fig. 3).The presence of these highconcentration molecules may explain the difficulty of soaking threonine into preformed crystals.ENDscript coil analysis shows that the greatest structural difference in the structures lies in the carboxyl-terminus and hinge (Fig. 3b).Identical residues appear interspersed across both domains (Figs. 2 and  3).Nonetheless, the similarities between BoLBP, EcLBP and other bacterial LBPs present unique opportunities for rational drug discovery based on the existing data.

Conclusion
We report two structures of B. ovis leucine-, isoleucine-, valine-, threonine-and alanine-binding protein (BoLBP).BoLBP is a prototypical bacterial LBP with additional amino acids inserted outside the central cavity and at the carboxylterminus.Both structures exhibit conformational flexibility of BoLBP in the absence of bound amino acids.Despite low sequence similarity, the structures have similarities to bacterial LBPs that can be exploited for future drug-discovery efforts.

Figure 3
Figure 3 Comparison of BoLBP with its closest structural neighbors.(a) The superposed structures show the Venus flytrap motif with a large central cavity.The superposed structures are monoclinic BoLBP (PDB entry 4xfk, gray), orthorhombic BoLBP (PDB entry 7jfn, blue) and EcLBP (PDB entry 1usg, green; PDB entry 2lbp, orange).The acetate (yellow sticks; from PDB entry 4xfk) and nitrate (blue sticks; from PDB entry 7jfn) are shown.(b) ENDscript coil diagram with thinner ribbons representing more conserved regions and thicker ribbons representing less conserved regions; identical residues in the structures are shown in red.(c) ENDscript surface diagram: identical residues in the structures are shown in red.

Table 3
Data collection and processing.APS beamline 21-ID-F

Table 4
Structure solution and refinement.
solution supplemented with 20%(v/v) ethylene glycol (Table2).Two crystallization conditions were used for data collection.The orthorhombic crystal form was obtained at basic pH using CHES-NaOH and 30%(w/v) PEG 3000, and was cryoprotected by passing through Al's oil.Heavy-atom (iodide) phasing was facilitated by the second crystal form, which grew in high salt (1 M LiCl) and 30%(w/v) polyethylene glycol 6000 (PEG 6000).The crystals were subjected to two 30 s soaks in cryo/phasing solution with increasing concentrations of sodium iodide in 20%(v/v) ethylene glycol.A second structure was obtained from soaking crystals grown in polyethylene glycol 3350 (and 200 mM potassium nitrate) overnight with 10 mM threonine in the same buffer.The crystal was briefly dipped into cryosolution comprised of the research communications 196 Graham Chakafana et al. � BoLBP Acta Cryst.(2024).F80, 193-199