Identifying ligand-binding specicity of the oligopeptide receptor OppA from Bidobacterium longum KACC91563 by Structure-based molecular modeling

Background: The OppA receptor as a ATP-binding cassette (ABC) transporter plays key roles in protecting host organism and transport nutrients across the intestine by the oligopeptide transporter from symbiotic bacteria directs maturation of the host immune system. Among lactic acid bacteria, Bidobacterium longum KACC91563, isolated from fecal samples of healthy Korean neonates, has the capability to alleviate food allergy effects. The extracellular OppA receptor from gram-positive Bidobacterium longum KACC91563 translocate nutrients-peptides from the outside environment of intestinal tract to the inside of the symbiotic cell, as a peptide importer. Hence, it was attempting to explicate the relationship between the substrate’s specicity from the OppA importer and the probiotic effects of B. logum KACC91563 in the host intestine. The probiotic effects of B. logum KACC91563 were attributed to the enhancement of the epithelial barrier by several different strain sepcic ways to prevent the strong adhesion of pathogens. The specialized structure-function relationship from the OppA importer could provide the abstract of substrate specicity into unique immunological properties of that the host organism. Results: In the study, we characterized the extracellular OppA importer from B. longum KACC91563 of intestinal microbiome by various protein structure-based modelings in silico. Structural characterization by conserved 5 patches and 4 functional motifs from specic trace residues of the OppA importer. The hydrate surface of the binding site had been decipted by specic trace residues of the OppA that trace residues of Thr58, Lys185, Trp443, and Tyr447, which were characterized in highly exposed hydrophobic binding pocket by its aggregation prones. Therefore, the spatial aggregation propensity in the binding site of the extracellular OppA importer plays a vital role in the specic interaction determinant for peptide binding. In addition, alanine mutation energy values allowed for the virtual determination of the relationships between the energy effects of the peptide binding site mutation on the transporter structural stability, the peptide binding anity, and the specicity into unique immunological prpperties of the host organism. In the current study, we attempted explication of the relationship between the substrate’s specicity from the OppA importer and the probiotic effects of B. longum KACC91563 in the host intestine based on the structure-function perspectives of the OppA importer. Moreover, functional characterization of solute-binding proteins (such as 15 cell wall proteins and 20 extracellular proteins) on the B. longum KACC91563 genome will lead to insight of key switch for substate’s metabolism into reprogramming immune responses in the host intestine.

from the domain itself. Prokaryotic OppA receptors usually have a unique extra domain (e.g., domain III with regular shapes), the substructure of which may also be an important structural factor. This is because the extra domain can have fewer ligand selectivity for the prokaryotic OppA receptor than the other solute transport receptors (such as the ion channel) [2]. Therefore, some prokaryotic OppAs that have the same architecture as an extra domain and hinge region did not only accept peptides with similar lengths. For instance, OppA from Salmonella typhimurium accepts peptides with two to ve amino acid residues [8], whereas L.lactis OppA can capture a wide range of peptides up to 35 residues long with preferences for a range of 9-11 residues [9,10,4]. As a result, the structural architectures and physical properties of an extra domain can also be associated with their transporter's functions by determining the peptide lengths and solute-ligand's types carried from the prokaryotic OppAs, as well as the hingebending motions. This extra domain is the structural and functional determinant from the prokaryotic

OppAs.
In the open conformer of the OppA receptor, the two α/β domains move away from each other on the hydrated state where the peptide-binding pocket is well exposed to water. When any hydrophobic peptide binds, the ligand would push out water molecules from the hydrated cavity to keep on doing hydrophobic interactions with the counterpart residues that protruded into the OppA binding cleft. At the same time, binding the polar residues of the peptide leads to the reordering of the water molecules via H-bonding networks between the polar residues and water. The use of water molecules by the OppA receptor is further linked to the lling of spaces not occupied by the peptide within its voluminous binding pocket. This is responsible for maintaining its H-bonding potentials before and after the ligand-binding [11]. In particular, a broad speci city against its peptide-ligand had been highlighted in the L.lactis OppA receptor, which has a huge binding cavity (almost 4900Å 3 [4,29]). Some residues of the peptide bound should indirectly interact with their counterpart residues on the outer contact surfaces of the hydrated pocket, which is in addition to also getting hands-on interactions inside the pocket. If the L.lactis OppA interacts with the 9-mer peptides as an optimal binding preference, the side-chains of the 9-mer peptides positioned inside the binding pocket (i.e., positions from 1 to 6 side chains on the ligand t into the hydrated pocket) seem to be more selective than the other side chains (i.e., positions 7, 8, 9 on the ligand) located on the outer contacts [10]. On the basis of the side chain's size and characteristics, the peptideligand would be accepted in the hydrated pocket and be deeply tted into the interaction interfaces rather than instantaneous contacts with the OppA receptor, and highly ordered water molecules. Whether the open or closed complex conformer appears is dependent on its peptide-binding a nity to the OppA receptor [4,3,12] in conjunction with producing a structural hinge region and an extra domain. Then more energy compensation through binding both is required than losing the entropy effects from the OppA receptor and peptide-ligand, respectively. In particular, stronger hydrophobic interactions within the OppA/peptide complex are acquired much more from pushing out well-arranged water molecules from the hydrated pocket of the OppA receptor. Because of this, it should be an essential interaction determinant to characterize what are driving forces for satisfying the binding a nity from the prokaryotic OppA/peptide complex structures [10,13].
In the current study, compressed information from prokaryotic OppAs was glued toward the OppA of Bi dobacterium longum KACC91563 and compared to L.lactis OppA. This is because the B. longum KACC91563, a subspecies of the Bi dobacterium genus, belongs to the lactic acid bacteria (LAB) family [14,15]. Currently, the Bi dobacterium genus consists of 56 species and 9 subspecies [81] isolated from human and other mammalian intestinal contents. This includes the 311 bi dobacterial genome sequences that could be retrieved from the National Center for Biotechnology information (NCBI). Bi dobacteria are gram-positive bacteria that possess one lipid biayer memebrane enveloped by a cell well. Notably, the Bi dobacterium longum KACC91563 is a probiotic genus isolated from the feces of neonates (BioSample code SAMN02603656). It is also known to secrete 15 cell wall proteins (0.77%) and 20 extracellular proteins (1.02%) for a total of 1,856 proteins that have a genome size of 2.40 Mb [16] with 87.77% of the Bi dobacterium longum symmetrical identity constituting the bi dobacterial secretome database from the BioProject PRJNA66401. The best characterized member of the Bi dobacteria is Bacillus subtills, whose geneome of 4.21 Mb encodes the putative 77 ATP-binding transport proteins in its 4,100 protein-coding genes [17]. The most common extracellular protein of the LAB family has been identi ed in the OppA receptor belonging to the solute binding proteins of the ATPbinding cassette importers (as represented in Figure S10). Bi dobacterium directly degrades exogenous proteins (such as milk proteins) to peptides by its protease, and then the OppA receptor takes the peptideligand [14,18] up as a nutrient transporter, which is similar to the other LAB members. Based on the structural colorations from both hinge-fragments and an extra domain, we had attempted to nd conserved structural features of LAB OppAs from their determined X-ray structures before centering on sifting the OppA templates of B. longum KACC91563 among them. Therefore, we preformed knowledgebased modeling for an unknown OppA structure from B. longum KACC91563 by projecting its biophysical information into conserved pictures of the OppAs of the LAB family and by making its molecular masking characteristicss distinctive from the LAB member's OppAs from previous experimental studies [14,18,82]. At the prediction stage of the OppA structure from B. longum KACC91563, we dicided not to couple equilibriums between the opened and closed complex conformers related to the ligand's binding a nity from the OppA templates of the LAB members. Practically, L.lactis OppAs were observed in structural differences between the open (PDB ID: 3DRH, bound to 6-mer peptide consisting of all Ala residues with low a nity) and closed (PDB ID: 3DRG, bound to 9-mer peptide being composed of bradykinin like RPPGFSPFA peptide sequence with high a nity) conformers upon peptide-ligand bindings as RMSD (root-mean-squared) of 2.08 Å per 556 residues. Even so, the structural difference between unliganded with counterpart residues (such as Ser472 and Ser474 residues of L.lactis OppA [4] represented in Figure  S14) between the open and closed-liganded complexes. Aside from the addition of the different peptide lengths, all backbones of the peptide-ligands were bound in the same structural patterns (shown in Figure  S6).
The most notable structural feature of the L.lactis OppA is where the mobile loops on the outer surfaces of one α/β domain (residues 301-542 in domain II) are seen to pull up and down between the open and closed-liganded conformers (illustrated in Figure S1). Whereas in the other α/β domain (residues 32-82, 220-300, and 543-570 in domain I) and the extra domain (residues 83-219 in domain III), neither their exible loops on the contact surfaces nor adjacent interfaces into the ligand binding sites are seen to signi cantly differ on the liganded state. As expected, the majority of the two hinge-strands were almost maintained between the complex conformers; the torsion angles of the center residues (Ala299 and Met542 [4]) in the two hinge strands changed little inside the 15° degree centered between the unliganded and liganded states. The functional determinants of the L.lactis OppA as a solute-binding protein (SBP) was therefore re ected into the large movements of the second α/β domain (in particular to residues 301-542 in domain II) that the OppA receptor closed the peptide-ligand up with high binding a nity, instead of incomplete sequence discernments of the peptide itself. As a result, the conformational changes in the peptide-binding sites (between two α/β domains) were less progressive than the mobility caused in the domain II of the L.lactis OppA when it changed from open-liganded to closed-liganded forms. There are still unful lled spaces by the peptide-ligand in the hydrated binding pocket. As if waters were lled instead of the ligand, the L.lactis OppA receptor covers over the bradykinin like peptide (i.e., RPPGFSPFA is only occupied by 30% of the available volumes 4900Å 3 ), which is known to bind with high a nity (dissociation constant K D of 0.1 µM [4]) on the closed complex structure (PDB code of 3DRG).
Because of this, the water molecules contained in the hydrated pocket from template OppAs should additionally be considered during the homology modeling of the query OppA from B. longum KACC91563.
To consider the ligand binding effects from water-mediated H-bonding interactions, water molecules were divided into two groups, in which the rst group was lled into xed water molecules without moving their coordinates in the L.lactis OppA binding pocket from any crystal structure. Then, the second water group collected moving waters by combining with the interaction residues to t into the ligand-binding cavity. In the process of homology modeling for the B. longum KACC91563 OppA, the previous water group would have their initial positions consistent with those located at the place of the hydrated binding pocket through overlaying the liganded conformers from the templates. As a result, the rst water group had less movement within the binding site from the OppA model of B. longum KACC91563, regardless of the binding of its peptide-ligand (on liganded and unliganded states). However, the latter water group can be mobile into the designated cavity depending on special interaction properties (either electrostatic interactions or hydrophobic interactions) connected with their surrounding environment upon ligand binding. Despite the requirements to gather highly energetically compensation for the water rearrangements of the second group, other strong hydrophobic interactions between the counterpart residues in the modelled OppA/peptide complex from the B. longum KACC91563 should induce the waters to stay away from the cavity. Indeed, the second water group has potential possibility of contributing to the determinant factors against the speci city rather than the binding a nity of the peptide-ligand. The buried water effects were known for only the periplasmic OppA in Salmonella typhimurium [11,20], which did not act as a physical barrier to its ligand binding, but seem to affect ligand selectivity in particular, when comparing relative binding a nities between its tripeptide-ligands. Otherwise, the buried water-mediated interactions may mitigate unfavorable interactions (such as charge repulsions) by shielding between the OppA receptor and its peptide-ligand or to replenish lost H-bonding by arbitrating disrupted interactions within the binding pocket. Both roles of the buried water molecules will lead to the capitalization of a peptide-ligand with a broad binding a nity against the OppA receptor, which is unlike highly speci c ligand recognition within other ion channel transporters.

Methods
All molecular modeling and optimization were performed in Discovery Studio, version 2017 R2, from BIOVIA (San Diego, USA) [21].

Selection of template OppAs
In total, bacterial OppAs have 50 crystal structures, of which 39 structures belong to periplasmic OppAs solved in gram-negative bacteria; they includ OppAs from Burkholderia Pseudomallei [22], Escherichia coli [23,24], Salmonella typhimurium [11,25,26], and Yersinia pestis [27]). Otherwise, the other 11 structures of extracellular peptide binding proteins originated from gram-positive bacteria (such as AppA from Basillus subtills [28] and OppA from Lactococcus latis [4,3,29]). Among the known structures of grampositive bacterial OppAs, biological relevance as template OppAs was determined by shared attributes, which are physiologically similar to OppA from L. lactis and AppA from B. subtills; i) the nature of peptide-ligand is fundamentally attendant on lipoprotein derivatives such that in gram-positive bacteria, the OppA transporter consists of a lipoprotein subcomponent distended beyond the extracellular faces of the cell membrane [30]. The additional factor as peptide exporters from L. lactis, B. subtills, and B. longum KACC91563 is accepting proline-rich peptides [18,4,29]. However, the frequencies of the proline are not regular in peptide-ligands bound with both OppAs. Proline-rich peptides that include at least one hydrophobic residue will be bound with high a nity to both the template and the query OppAs from priority selection of peptide composition rather than their exact sequence [14,4]. ii) The homologous to hydrophobicity and cavity size of their binding pockets have closely been related to the characteristic of trace residues derived from these conserved regions on the basis of structural similarities. For instance, templates (like L. lactis OppA) and the query of OppA from B. longum KACC91563 have been classi ed into cluster C of SBPs by R.P. -A. Berntsson et al. [2] based on crystalized structural alignments in PDB. The SBPs of cluster C were furthermore contemplated for their structural and biochemical features between the extracellular OppAs. This is because the applied template SBPs could bind with different solute-ligand types (for example nickel ion [6] or cellobiose [31], which is unlikely to be an oligopeptide within known ligands from the SBP of cluster C) to be dissimilarly functionally characterized. In contrast, all SBPs of cluster C have a unique extra domain, which has differences in its structural skeletons and sizes. Despite these quite little sequence identities (smaller than 20%), their 3D-structual architectures are highly conserved; In the study, the sequence identi es of template candidate SBPs were excited in 15-28% over their equivalent 450-520 residues by searching peptide transporters from the cluster C of SBPs [1,2] via structural homology searches and sequence pro les of iterative protein PSI-BLAST within the UniProt knowledgebase of Swiss-Prot [32]. Among the truncated candidates, the DppA of E. coli [24] and two OppAs of S. typhimurium and of Thermotoga maritima [33] have no clear distinctions of structural features from the searched template SBPs (these also have the same 3D-scaffolds containing an extra domain). Despite this, they were differently being marked in a biochemical overview from each other due to their discrete nature being accessible to the extracytoplasmic receptor OppAs from gram-negative bacteria. From a critical point, periplasmic OppAs in gram-negative bacteria have a limiting size of peptide-ligands (such as peptides from two to ve resides) that can be transported though the outer membrane. In contrast, no such physical constraints exist in extracellular OppAs from gram-positive bacteria. The extracellular OppAs therefore transport longer peptide-ligands depending on the corporeal constraints of the binding pocket's size. Practically, some periplasmic OppAs from gram-negative bacteria (including the S. typhimurium OppA [11] of PDB code of 1B9J) were categorized in cluster C of SBPs based on only their 3D-structural scaffolds (often, ≤ 25% as consequence of their protein sequence alignments to the query OppA). In contrast, the top 5 sorted template's candidates were less frequently listed than in the OppAs of gram-positive bacteria. Even in the gram-positive bacteria of B. substilis, the extracellular AppA was selected into one of the templates, except for its periplasmic OppA. As a result, the OppA templates were carefully picked out, instead of exiguous sequence identities. iii) The query OppA from B. longum KACC 91563 is also an extracellular SBP in cluster C and a LAB member such as L. lactis and B. substilis.
The three considerations mentioned above lead to sorting out templates from the solved gram-positive bacterial OppA structures as being the most suitable for predicting the query OppA structure from the B. longum KACC91563. The hand-picked templates were well matched in their structural alignments, in which the four templates were overlaid with characteristic determinants (as highlighted in Figure S2). There were high scoring matches of extracellular SBPs between the OppA from Bacillus anthracis str. Ames and AppA form B. substilis (the known structures 5U4O and 1XOC are open-unliganded and closedliganded forms, respectively). Both similarities were consistently anticipated from their structures or sequences that exhibit 27% sequence identity and 2.60 Å of RMSD with 467 equivalent positions. This depends on whether the ligand is binding all of the whole structural differences between them, which is over 90% of mapping coverage in e-value of 3e − 50 . In the case of the OppA structure from B. anthracis str. Ames (PDB ID: 5U4O), the structural features have not been known in detail (not published). AppA from B. substilis (PDB ID: 1XOC) [28,34], OppA from B. anthracis str. Ames (PDB ID: 5U4O), and OppA from L. lactis (PDB ID: 3FTO, 3DRG) [3,4] were commonly found in ATP-binding cassette (ABC) transporters for oligopeptide uptake. In contrast, an unexpected template of PreZ from Enterococcus faecalis (E. faecalis) is a pheromone binding protein. The designated fourth template PreZ is a lipid-anchored extracellular SBP and plays as a pheromone receptor from E. faecalis, whose protein sequence is 19% identical to that of the query OppA from B. longum KACC 91563. The E. faecalis PreZ has an external domain like that of the L. lactis OppA, too. Intriguingly, L. lactis OppA was also derived to be homologous to other SBPs that convey sex pheromones in the type of ABC transporters [1,28,35]. The homologous PreZ receptor not only has the high selectivity for hydrophobic 7-mer peptide (present in Fig. 6), but also E. faecalis is a gram-positive bacterium that lives in the gastrointestinal tract of mammals [36]. The favored characteristics of hydrophobic pheromone ligands had been presented from two competitive 7-mer peptides. Both originated from processing cCF10 (LVTLVFV) of the ccfA gene product and the inhibitor peptide iCF10 (AITLIFI) encoded by the icf10 gene. The 7-mer pheromone ligands compete for the same binding site (as shown in Fig. 6) of the E. faecalis PreZ [36,37,38]. These templates all must accommodate their peptide-ligand in a manner of Venus-ytrap mode. We preferentially distinguished their open-unliganded structures (PDB ID: 3FTO, 5U4O) from their closed-liganded forms (PDB ID: 3DRG, 1XOC, 4FAJ), as if all ligands were to be wholly buried into the peptide-binding clefts. A comparison of the templates and the query is given in The templates and the query of OppAs have high structural similarity at the core framework (from domain I to domain III, including the segments of hinge strands shown in Figure S2). However, they have different lengths and conformations for some loops (residues Phe205-Val210, Tyr352-Lys357, and Tyr534-Val539 on the query OppA from B. longum KACC91563) adjoining the ligand-binding sites. In different loop segments, exposed hydrophobic loops to water molecules should affect the structural stability of the OppA receptor itself due to their aggregation prones. Loop segments of the query OppA align best to one template in one loop segment and another template in another loop segment. The loop sections at these positions could not align to those of template structures thoroughly, even if spatial restraints were placed on the conformational similarities (as bonded terms of geometrical features) for residues within the vicinity of these loop spots. There is because any homology restraints of templates cannot be applied in the loop segments of the query OppA models. These parts of loop segments into an entrance of the hydrated pocket were de ned by the MODELER loop re nement based on the sequencestructure alignments between the query sequence and common local structures of the template OppA and AppA (PDB code of 3FTO and 5U4O). After de ning an initial coordination of these loop segments in the OppA model structure, the local loop conformers were additionally optimized according to the CHARM-derived stereochemical and nonbonded restraints, respectively, for statistical preferences of the different residue types and for the different side-chain rotamers in the regions (containing in the residues Phe205-Val210, Tyr352-Lys357, and Tyr534-Val539 within the query of OppA models).
By providing rough OppA models from B. longum KACC91563, we further optimize their local conformers of contiguous segments (i.e., loop and speci c trace residues with high aggregation propensities). Thence, it was indicated whether the residues of the segments are in the desired 3D environment of the open OppA structural states evaluated by the feature energy functions of the MODELER program. In the feature energy functions, geometric features (such as distances and dihedral angles) of the OppA and AppA templates are restrained by setting lower and upper bounds on their allowed values associated with each residue on the speci c contiguous segments. This is in terms of probability density function (PDF) [39] and discrete optimized protein energy (DOPE) [40] along the pre-structural alignment positions of the templates (as depictured in Figure S2). Both feature energy functions of MODELER were also considered in solvation effects by adding the solvation energy term to other intramolecular energy terms. This occurred when water molecules in the best template of L. lactis OppA (PDB ID: 3FTO with openunliganded folds) were reproduced into the OppA models and then were treated as rigid bodies in the OppA models without steric hindrances of the OppA model selves. The solvated OppA model did more optimiziation by conjugate gradient and simulated annealing optimization procedures using the CHARMm force eld. As a result of this, the conserved structural characteristics are well re ected into spatial geometrical restraints beside the homology-derived restraints toward the OppA and AppA templates. If there are hydrophobic residues on the hydrated surface and polar residues in the hydrophobic cores of the query OppA, the higher restraint violations of both features are given to the higher PDF and DOPE for each residue. A smaller PDF energy thus means that the OppA model satis es the homology-derived restraints better. A lower DOPE also indicates a better model. The OppA model that had the lowest PDF and DOPE energies was chosen as the nal model. Then the nal OppA model was optimized in the relationships between the structural features of the templates, by the tness of derived restraints from its current 3D environment and by studying Ramachandran plots [41]. The total PDF energy of the best OppA model was determined to be -14,251.4 kJ/mol in terms of geometric restraints, which are the sum of the scoring function values of all homology-derived pseudo-energy terms and stereochemical pseudo-energy terms. At the same time, the best model had the lowest DOPE score of -50,879.8 kJ/mol as a conformational energy that measures the relative stability of a conformation with respective to other conformations of the loops for optimizing local structures of the query OppA.
To ensure precise of its backbone conformations in the OppA model, the residues on Ramachandran spaces were analyzed by comparing the steric effects of a residue's torsion angles (phi and psi angles) that were derived from how they fold [42][43]. Since the Ramachandran spaces could be classi ed into allowed and disallowed conformations, misfolded model structures were roughly de ned within the disallowed region with bad contacts. The ordered patterns of residues in the OppA model were found to have 469 residues (92.1%) in the favored region, 26 residues (5.1%) in the allowed region, and 14 residues (2.8%) in the outliner region on the open-unliganded scaffolds. The OppA model was thus considered to be a good model structure for which there should be more than 90% of the residues in the favored region of the protein (shown in Figure S5).
Validation of the OppA model structures from B. longum KACC91563 compared to experimentally determined template structures The nal OppA model (from Gly37 to Gln547 residues) was subjected to further validation of the compatibility modes of its own fold and sequence segments on the hydrate environment. Three methods (ERRAT2 [44], Z-score of ProSA [45], and Verify3D [46]) were then employed to measure the overall quality assessment of the OppA model structure for its tness in its current 3D environment by comparing the characteristics of other experimental structures, including the OppA and AppA templates (PDB code of 3FTO and 5U4O in the open-unliganded structural state). The ERRAT2 evaluates the structural errors of a model structure by distinguishing between the correct and incorrect fractions of nonbonded pairwise interactions (i.e., the six types of atom-atom interactions between carbon, nitrogen, and oxygen) within a speci ed distance limit (as predetermined distance cutoffs limits of 3.00 to 4.75 Å) determined from protein X-ray crystal structures. The larger ERRAT2 value then points out the better re ned model (ranges from 0 to 100%) when compared with the score distribution of residues in correct protein structures. This is because error values of residues in the modelled structure higher than 95% were not included in its overall quality. The Z-score of ProSA also evaluates the overall model quality, but is a track recode of only the alpha carbons in the OppA model by measuring the deviations of total energy distributions for all determined proteins (X-ray, NMR in PDB), unlike ERRAT2. Specially, the Z-score better captured the solvent exposed residues within the soluble globular proteins than protein structures containing transmembrane domains. As manifested in Figure S5, the folding features of the hydrated OppA model are well re ected in the Z-score, where there were get-togethers in the middle region of the score scatterplot observed for the experimental protein structures. If the OppA model contained some errors, the Z-score would fall outside the range of characteristic values from known proteins. At the same, both Z-scores appeared in only small differences of relative middle spots to induce the OppA conformation depending on open or closed states. The verify3D was additionally employed to inspect the validity of the OppA model structure by measuring the self-compatibility of its 3D structural pro les with its own protein sequence as the overall quality and local 3D-1D scores of the model structure in a xed-length (typically about 5 to 20 residues) based on experimental data. An overall quality factor of 82.76% was assigned by ERRAT2 for the OppA model, and the Z score of -8.64 in ProSA was within the range of native conformations like other known proteins of similar size (refer to Table S2). On the open-unliganded structural state, the query of OppA model had lower ERRAT2 value and Z-score of ProSA than those of the OppA and AppA templates (OppAs from B. anthracis str. Ames and L. lactis as PDB code of 5U4O and 3FTO), as shown in Table S2.
In the OppA from B. longum KACC91563, a large part of the surface from the binding pocket (listed in Table 1.) was exposed to water rather than the hydrophobic environment like membrane interior. Then the effect of the shielding environment caused from the hydrated binding pocket can be indicated by the hydrophobicity plot (depicted in Figure S7), which shows the average hydrophobicity value of each residue with the neighboring four residues [47] instead of its own hydrophobicity. The uctuations in the hydrophobicity index could be induced by the shielding effects excited from the water. Then the thickset wave was observed in the query OppA of B. longum KACC91563 than the template OppA of L. lactis (PDB ID: 3FTO). This indirectly explained the hydrophobic binding pocket of query OppA that was exposed to the surface too often, albeit it was not radical shielding effects drawn in water. Due to this, in the absence of ligand binding, an open structural state of the OppA model was more unstabile compared to that of the template L. lactis OppA. The uctuations of the hydrophobicity index further showed that the lower ERRAT2 value and Z-score of ProSA were not directly in uenced by the quality of the OppA model, but by the distinct nature of the hydrophobic binding pocket from B. longum KACC91563 compared to that of the OppA and AppA templates. In the case of the open OppA model, the residue hydrophobicity of Thr58 was − 0.70, but 0.72 of its neighboring 5-residues ran an average hydrophobicity. In contrast, Ala60 changed from 1.80 of the residue hydrophobicity to -0.64 of the neighboring 5-residues' running average hydrophobicity. Indeed, the helix's secondary structure of the uctuated region on the query OppA (predicted from such as DSC [48] was based on the solvent exposed patterns of globular proteins) changed into coil that corresponded to those of template PDB structures. If there is a signi cant difference in the hydrophobicity changes and in the underlying structure of residues that lie outside the binding pocket on an open-unliganded state, the gap of values in both hydrophobicity indexes of the residues becomes smaller in the closed-liganded state (in the case of Thr58 shown in Figure S4). The residues with leading deviations between their hydrophobic properties and the hydrated environment were temporally recognized in the binding pocket depending on whether or not the ligand was bound. The signi cant residues were Thr58, Glu59, Ala60 (domain I), and Ala190 (domain III), which are consistent with the speci c trace residues of the query OppA from B. longum KACC91563. Surprisingly, the signi cant residues of Thr58, Glu59, and Ala60 on the query OppA were corresponded to a speci c hydrophobic pocket (denoted Phe(P5) in Fig. 5) binding to the peptide-ligand. In contrast, the other residues of Phe334, Ala396, Val400, Phe426, Leu427, Val430 (domain II), Leu457, and Ala519 (domains I and II) were located in the conserved trace patches of the template and the query OppAs (expressed in Fig. 3). Such interaction residues have its residue characteristics change as a result of the hydrated pocket, which is restricted by the structural compatibility of the query OppA model [49]. In the open states, the shielding effect of water into the OppA model was insu cient to reproduce the characterized structural stability of templates by only the homology-induced geometrical restrictions. This results in different patterns of hydrophilic and hydrophobic interaction residues in the hydrated binding pocket of the OppAs. Therefore, the OppA model structure had lower scores in the validated quality assessments (such as ERRAT2, Z-score of ProSA, and Verify3D) than those of the X-ray determined template OppA and AppA. However, the OppA query model must be a reliable model structure that has its passed standard qualities observed by other experimental protein structures (shown in Table S2 and Figure S5). In another aspect, the Verify3D allows us to rea rm hydrophobic patches on surfaces of the hydrated binding pocket within the OppA model as well as the map of spatial aggregation propensity (in Fig. 4) by connoting whether its resides are in the desired 3D environment. In this context, the hydrophobic residues on the exposed OppA surfaces and the polar residues in the hydrophobic OppA cores were given low We applied the exible docking [50], which allows for the exibility of the open-unliganded OppA receptor from B. longum KACC91563 during the docking of the bradykinin-like peptide (as 9-mer peptide of RPPGFSPFA) ligand in the induced t structure (as the closed-liganded OppA receptor from L. lactis). The side-chains of the speci ed amino acids in 5.0 Å of the active site (listed in Table 1) and in the domain II (residues 294-517 shown in Fig. 1) were allowed to move during the peptide docking by generating ensembles of the OppA receptor conformations. The domain II of the OppA receptor was seen to reel up and down toward the peptide binding site between the open-unliganded and closed-ligand conformers from the template structures of the L. lactis OppA bounded to the 9-mer RPPGFSPFA peptide (PDB code 3DRG), as illustrated in Figure S1. For the varied residues of the OppA receptor, 2,734 of the conformational states were optimized from the lowest energy of -106.49 kcal/mol to the highest energy of -77.95 kcal/mol by CHARM-based scoring functions. However, the backbone and the side-chains of the OppA receptor, which were not speci ed, were xed at their original positions. Also, in the speci ed residues of the OppA receptor, alanine, glycine, proline and cystein in disul de bridges could not be optimized for the conformational ensembles due to the fewer rotatomers of their residues. Subsequently, the initial structure and pose of the 9-mer RPPGFSPFA peptide was coordinated from its X-ray structure (PDB code 3DRG). Then, the sphere at the center of the active site where the peptide interactions were aligned to the site features of the OppA receptor (e.g. polar and apolar or hydrogen bond donor and acceptor) as hotspots was placed into the coordinates 34.57, 2.81, 12.28, and 14.11. The ensemble dockings of the 9-mer RPPGFSPFA peptide poses in the site sphere on 2,734 of OppA receptor conformational states were performed. After that, each docking pose was subjected to the simulated annealing molecular dynamics (heating to 700K for over 4,000 steps followed by cooling to 300K for over 6,000 steps) process under the CHARMm force eld [51] before the complex poses were scored. For each nal complex pose, the CDOCKER interaction energy as the CHARM energy (i.e. the interaction energy Val430 of the OppA from B. longum KACC91563 as criteria of a reasonable shape feature constraint to select and edit the best model among the top 10 pharmacophore models.
Residue alanine mutations and their impact on the protein stability, binding a nity, and aggregation in the OppA/peptide-ligand complex from B. longum KACC91563 We used the structural knowledge of the OppA/9-mer RPPGFSPFA peptide complex from B. longum KACC91563 to consider the effect of the peptide binding site on the peptide substrate selectivity by focusing on the structural stability of the closed-liganded OppA and on the binding a nity of the 9-mer RPPGFSPFA peptide in the complex. We calculated the stability contributions of the 62 key residues (listed in Table 1) in the peptide binding site as the difference between the folding free energy of the Ala mutated structure (i.e., single Ala mutation) and the wild type of the OppA receptor corresponding to the peptide bining site variants.
All interaction energy terms of were cacluated using the CHARMm force eld and a generalized Born implicit solvent model. Van der Waals terms and electrostatic interactions as well as non-polar, surfacedependent solvation energy terms (i.e., the cavitation energy) were determined empirically. The energy function also contained terms for the side-chain and back-bone entropy to allow for room temperaturedependent calculations. Therefore, substituting a polar residue with a relatively non-polar alanine could result in changes in the conformation neighboring residues. After building the OppA mutation structures of the alanine variants, the conformations of the mutated residues and neighbors were optimized further by using the MODELER protocol.
The mutation energy of binding was calculated as the free energy difference of the binding of the OppA receptor from B. longum KACC91563 and the 9-mer RPPGFSPFA peptide that contain the alanine variants of the 62 key residues in the mutated complex structure: The mutation energy for the binding a nity was calculated as the sum of the scaled van der Waals, electrostatic, non-polar, and entropy terms as well as the OppA structural stability. Mutation energy values and the corresponding effects of the complex alanine variants were evaluated, and these values were added as the peptide binding site properties to the different complex variants. This allowed for the virtual determination of the relationships between the energy effects of the peptide binding site mutation on the OppA structural stability, the 9-mer RPPGFSPFA peptide binding a nity and the OppA receptor-related peptide substrate selectivity from B. longum KACC91563.
There was no self-aggregation of the active site on the open-unliganded OppA receptor from B. longum KACC91563 due to the steric effects from the presence of a large extra domain III (residues 84-210 shown in Figure 1) beneath the center of both α/β domains. The result provides more support for the OppA/9mer RPPGFSPFA peptide complex from B. longum KACC91563 in which the self-aggregation cannot occur as a complement to the 9-mer RPPGFSPFA peptide due to direct hydrophobic interactions with the interaction site of the closed liganded OppA receptor. We predicted selective alanine mutation effects and the relactive importance of hydrophobic interactions on the exposed hydrophobic surfaces of the pepide binding site across the the OppA/9-mer RPPGFSPFA peptide complex from B. longum KACC91563 by calculating the spatial aggregation propensity (SAP) based on the pre-calculated solvent accessible area (SAA) of the fully exposed side-chain by the CHARMm force eld. The SAP for the closed-liganded OppA receptor from the open-unliganded was obtained as the speci ed radii from the hydrophobicity scale of Black and Mould [52], which was added as atom and residue properties on the patches of the exposed hydrophobic residues. The hydrophobicity scale was normalized such that glycine had a hydrophobic value of zero. Thus, the amino acids that were more hydrophobic than glycine were positive, while more hydrophilic residues were negative than glycine. Therefore, the aggregation propensity of the OppA receptor conformers for the OppA receptor atom was de ned as: The SAP for each residue on the patches of the exposed hydrophobic residues was obtained as the average of its atomic aggregation scores. High aggregation scores (0.0 < SAP < 0.5) indicated highly exposed regions (in Figure S4). Then an SAP map for the region was generated by red color-coding (in Figure 4), which allowed us to perform target mutations of the peptide binding site to enhance the peptide substrate speci city of the OppA receptor from B. longum KACC91563. Low SAP values (-0.5 < SAP < 0.0) indicated that the exposed surface was a hydrophilic region (blue in Figure 4). This can be expected, as most of the OppA receptor sufaces exposed to water are usually hydrophilic. Conserved functional patterns between four templates and the query OppA were represented by trace residues, which were identi ed from their sequences and mapped to their 3D-structures (shown in Fig. 3).
The trace residues were further characterized by partitioning the conserved functional surface patches into subgroups according to the inferred roles of speci c residues within the query OppA from their structures. In the study, the trace residues were forced to make a direct connection between the conserved residues and their functional importance based on corresponding interaction residues (illustrated as such in Figs. 5 and 6). Then exposed trace residues that were more likely to be responsible for binding activity could be distinguished from the buried trace residues that were more important for maintaining structural integrity on the hydrated binding sites of both the templates and the query OppA. Clustering functional trace residues were indeed applied to de ne the distinct patterns within the query of the OppA structure from those of the other template's structures. As shown in Fig. 3, the conserved patterns were mapped to the query OppA from B. longum KACC 91563 by creating speci c groups of trace residues at a special distance cutoff of 21.0% (based on the protein sequence identities between the templates and the query OppA OppA in Fig. 3 and Table S1) into ve patches. The conserved distances of 21.0% should be depicted as discriminating molecular basis based on either the electrostatic or hydrophobic properties of subgrouping trace residues from the four templates and the query OppA. The conserved trace patches were Asp94-Asn129 (in domain III), Asp201-Ser255 (in domains I and III), His306-Ala347 (domain II), Lys363-Ser432 (domain II), and Leu457-Ser528 (domains I and II) in order of the patch's numbers, respectively, inside the OppA from B. longum KACC91563. Moreover, four functional motifs (in Figure  S13) were identi ed in these ve conserved patches from both the templates and the query of OppA: identi ed by Tam et al [1]. This is because we only subjected class 5 external proteins from a few prokaryotes that were between the OppA and AppA templates and the query of OppA. In view of their pattern motifs on the conserved patches, the pheromone binding protein of E. faecalis PreZ (PDB ID of 4FAJ) was not more distant from other templates and the query OppA, since they were preserved in their 3D-structural topologies and biological properties via the four pattern motifs. The speci c trace residues of OppA from B. longum KACC91563 were Ser46, Glu47, Thr58, Glu59, Ala60, and Tyr76 in domain I, Trp399, Arg440, Trp443, Asp446, Tyr447, and Gln517 in domain II, and Phe139, Val184, Lys185, Gly187, Ser188, His189, Ala190 in domain III of its extra domain located in the interaction interfaces (highlighted in yellow in Fig. 3). Not the speci c trace residues in the interaction interfaces represent the four functional motifs in any conserved patch shown in Fig. 3. In particular, these speci c trace residues would serve to guide the site-directed mutagenesis in silico for studying the OppA protein structure- Speci c volumes for binding pockets range from 1600 to 4900Å 3 in the templates. This is beacuse they all have an extra large domain (domain III) extending the pocket, and also their hinge-region consistists of two β-strands, in which each β-strand is typically observed in 4-5 amino acids as a conserved structural feature (displayed in Figure S2). For these conserved structural traits, their peptide-ligands accommodated in the binding pocket have also been shown to have similar preferences of 7-mer to 9-mer peptide length with a high a nity (dissociation constant K D of µM to pM range [2,13]) in their closedliganded states. Likewisely the OppA models from B. longum KACC91563 have been observed in the little reduced volume of the hydrophobic binding pocket from 2740 to 2200Å 3 by trapping a bradykinin-like peptide (as 9-mer peptide of RPPGFSPFA) may be possible (expressed in Fig. 2 and Fig. 4).
As another characteristic of the binding pocket, the OppA templates were presented as having discernible functional features by superposition of the speci c hydrophobic cavity. The coordinated hydrophobic interfaces from the templates did distinguish their sizes and preferences of the counterpart's hydrophobic residues from any location of the peptide ligands on the hydrated binding pocket (as shown in Fig. 5 and Fig. 6). In the nonbinding ligand, hydrophobic interactions on the uncovered interfaces of the hydrated pocket can be driven in the tendency to aggregate, which may decrease activity and open the structural stability of the OppA receptor's self. By predicting the OppA receptor surface sites that are liable to aggregation, we observed the aggregation propensity of the ve conserved patches and of speci c trace residues, into which the exposed hydrophobic residues were spatially closed. The aggregation propensity is an approximate indication of the equilibrium between the multiple structural states of the OppA receptor in the water solution and may not be an integer. In context, the aggregation propensity is likely to provide an overall tendency to aggregate in the OppA receptor's binding pocket between the openunliganded and closed-liganded states (represented in Fig. 4 and Figure S4). Therefore, the hydrated surface of the binding site had been depicted by speci c trace residues of the OppA that the trace residues of Thr58, Lys185, Trp443, and Tyr447 (in Figure S4), which were located on highly exposed hydrophobic regions in the binding pocket (residues with high aggregation propensity scores are colored red, while those with lower scores were colored blue, as shown in Fig. 4). In Figure S4, the speci c trace residues (Thr58, Lys185, Trp443, and Tyr447) also had higher aggregation scores in the open-unliganded state than the closed-liganded state of the OppA models from B. longum KACC91563. Then highly exposed regions (where the surface is red in Fig. 4) Fig. 4). In particular, the speci c Thr58, Lys185, Trp443, and Tyr447 trace residues on the binding pocket had prominently decreased to act as an aggregation inclination changed in the two structural states, as captured in Figure S4; those aggregation prone scores for the four trace residues can be important for bradykinin like peptide (RPPGFSPFA) binding, as they changed from 0.088, 0.167, 0.203, and 0.207 (at before the ligand-binding) to -0.225, -0.192, -0.058, and − 0.097(after the ligand-binding), respectively.
Oligopeptide binding pocket presented in speci c trace residues Bradykinin (RPPGFSPFR, pfam code of PF06753 and InterPro code of IPR009608) is well documented as a pharmacologically therapy known as an ACE (angiotensin-converting-enzyme, CD143) inhibitor [54,55] to reduce high blood pressure. Together, this bioactive peptide is an in ammatory mediator when kinin B1 receptor (BDKRB1) recruits neutrophil via the chemokine CXCL5 (CD185) [56]. Physiologically, the capability to bind bradykinin for the LAB microbiota (such as L.latics, B.subtilis, and B.longum) did elicit any health effect from a nutrient source, which catches up peptides with antihypertensive activity [57,58].
Bradykinin is well-matched with the favored peptide-ligands from the templates of L.latics OppA and B.subtilis AppA. The oligopeptide ABC importers from the LAB microbiota select preferentially proline-rich peptide, containing at least one branched residue of leucine, isoleucine, and valine [14,28,4,59] with no exact sequence preference. In that case, bradykinin was known for being able to bind the OppA and AppA templates with high a nity (dissociation constant K D of 0.10 µM and 50.30 µM, respectively [59]) in closed-ligand conformation. In essence, the query OppA from B. longum KACC91563 not only has a similar overall structural architecture, but also operates in the organized hydrophobic binding pocket from the template of L. lactis OppA, as if bound with an equivalent bradykinin as a common oligopeptideligand. In this study, we predicted the complex structure of OppA from B.longum KACC91563 bound to a bradykinin like peptide (RPPGFSPFA) by docking the peptide. On the basis of the wailing peptide's speci city, the 9-mer peptide of RPPGFSPFA has a xed proline-rich peptide isoform, of which the backbone does not change like that of the X-ray determined conformer (PDB code of 3DRG) bound to L. lactis OppA. One of the binding factors was de ned by a central hydrophobic pocket (i.e., the counterpart residues to the phenylalanine side chain at position 5) in the OppA interaction interfaces from B. longum KACC91563. The hydrophobic pocket's surface is covered inside Glu47, Thr58, Glu59, Ala60, Gly61, Leu427, and Val430 residues (as presented in Table 1 and Fig. 5). The complex structure shows the partitioned trace residues located in hydrophobic patches as a part of the binding pocket, which correspond to speci c hydrophobic residues of RPPGFSPFA (listed in Table 1 and Fig. 3).
To understand the key residues that contribute to spatial proximity in the contact interface of 5.0 Å, we rst evaluated the mutation energy effects of single Ala mutants on its structural stability and binding a nity within the OppA-peptide complex from B.longum KACC91563. Since the hydrophobic binding site underlies better stability to reduce aggregation, it switches from an open to a closed conformation of the OppA upon the 9-mer RPPGFSPFA binding. The Ala mutation effects tend to cause greater destabilizing structural stabilities of the complex than decreasing binding a nities to the molecular partners, as shown in Fig. 7. The highest energy of Ala mutations was located at residues of Pro54, Tyr76, Lys185, Gly187, Tyr191, Ile271, Phe334, Phe426, Gly442, Trp443, Tyr447, Leu457, Trp515, and Gly536, all of which were designated as structural destabilizing effects (mutation energy > 0.5 kcal/mol). They have mutation energies above 2.0 kcal/mol. In contrast, the Ala mutant of Ser188, Asp446, and Asn452 did alter the stabilizing (mutation energy < -0.5 kcal/mol) complex structure by smaller mutation energies than the − 1.2 kcal/mol. Intriguingly, the Ala mutation effects were well projected into the speci c Thr58, Lys185, Trp443, and Tyr447 trace residues with being decreasing aggregative propensities (shown in Figure S4), wherein the intrinsic moving domain II of the OppA is the most de ning feature of the switching conformation from a part (open conformation) to the two domains I and II together (closed conformation). The four residues were also involved in the interaction interfaces of OppA with the RPPGFSPFA peptide. Thus, their Ala mutants could greatly lead to the unsettling of its complex stability in the presence of 2-to 60-fold mutation energy effects more than its binding a nity. It is noteworthy that a unifying feature in the OppA-RPPGFSPFA complex has been linked to its surface-exposed and nonconserved trace residues at the contact interfaces, which show the largest difference on the complex stability from the patches of exposed hydrophobic residues (highlighted yellow residues in Fig. 3) and its speci city dispensed for peptide-ligand compared to the template complex of L.lactis OppA. There are speci c trace residues of Gly61, Asp67, His269, Ala270, Leu296, Gly444, Asn452, Gly535 and Gly536 in the binding pocket (indicated in Table 1 and Fig. 5) of the OppA from B.longum KACC91563. Taken together, the highly exposed residues of Ser186, Tyr191, and Met192 on the extra domain III, Asn294, and Pro445 on domain II, of which the counter partners are Arg(P1), Pro(P3), and Pro(P7) on the RPPGFSPFA peptide, are prompted by a descending aggregation factor in Figure S4. Among them, the biggest destabilizing Ala mutant effect of polar Tyr191 was envisaged as 5.87 kcal/mol by comparing that of hydrophobic Met191 as 1.77 kcal/mol from other exposed residues on extra domain III. In contrast, the lowest stabilizing Ala mutants were acidic Asp446 and polar Asn452 residues (in domain II) as -9.24 kcal/mol and − 5.57 kcal/mol, relative to Pro(P2) on the corresponding peptide-ligand. This strongly suggests that the stabilizing effects of induced tness on the OppA came from the favoritism of prolinerich (P2, P3, and P7) and hydrophobic Phe(P5), which also have impacted the OppA-RPPGFSPFA interactions for B. longum KACC91563.
If any trace residues have a linked 3D-pharmacophore arrangement to common features of the OppAs between L. lactis and B. longum KACC91563, they are a knowledge-based description of interaction constraints with the RPPGFSPFA. This is more reliable when pharmacophores are depicted as an assemble of essential features (such as hydrophobic, H-bonding acceptor, and H-bonding donor) to trigger the closed-ligand binding OppA. Moreover, the pharmacophores corresponding to the 3D location of key residues from the OppA of B. longum KACC91563 are required for optimal intermolecular interactions with the RPPGFSPFA. They also signify the functional speci city of the trace residues as a binding-site characterization via virtual site-directed Ala mutation against that of the wild type. These pharmacophores of major prolin-rich and hydrophobic phenylalanine (P5) in turn lead to impute featuring conductors of the key trace residues in the interaction interfaces. Two H-bondings are formed in the N atoms of Arg(P1) to the side chain of Gly187 and Ser188 within the OppA of B. longum KACC91563, while it did not form with the OppA of L. lactis (as shown in Fig. 5.). The formation of two H-bondings at Arg(P1) is noted along with the Ala mutation effects of Gly187 and Ser188 matched to -0.57 kcal/mol and − 1.27 kcal/mol with stabilizing binding a nity. This is because the mutation effects lead to stronger H-bonding between the N atoms of Arg(P1) with the backbone of those Ala mutants. The ve-membered ring of Pro(P2) has a hydrophobic point feature (represented as a pharmacophore blue color-code in Ala60 residues of the OppA from B. longum KACC91563, compared to Asn55 and Ala57 in the OppA of L. lactis. The Phe(P5) is inserted into the hydrophobic cavity formed by Glu47, Thr58, Glu59, Ala60, Gly61, Leu427, and Val430 of the OppA from B. longum KACC91563. These hydrophobic interactions may be the grassroot cause of the favorable features with shape constraints (gray colored shape in Fig. 8.), which add to the surrounding assembly of the peptide-ligand's pharmacophores, as the excluded volume is de ned by the cavity shape of the peptide binding site of the OppA. The leading hydrophobic shape constraints to branched hydrophobic residues was enforced rather than the projected hydrophobic point feature of the targeted Phe(P5) for making the interaction sites. In contrast to the Phe(P5), the phenyl ring of Phe(P8) was projected as a hydrophobic interaction point into the Arg440 residues with π-π interaction (shown in Fig. 5). At the same time, the backbone carbonyl group of Phe(P8) as a H-bonding acceptor was achieved by a hydrogen donor from the side chain of Arg440. Then the hydrophobic shape constant of Phe(P8) expanded to surrounding the Phe(P8) by Leu296, Phe334, Ala396, Trp399, and Trp515 residues on the OppA from B. longum KACC91563. The role of Arg440 was well superseded by the Arg416 residue on L.latics OppA. As a result, the con guration of pharmacophores on the complex structure-based is an arrangement of seven chemical features of two H-bond acceptors (HBA) and the features of three H-bond donor (HBD) vectors as well as two hydrophobic(HY) point features with location constraints that represent locations in space within a given radius of 1.6 Å. Then the shape constant adds to the surrounding pharmacophores that were generated from the binding site expanding to 3.5 Å, which corresponds to the OppA-RPPGFSPFA interactions in the docked complex model from B. longum KACC91563. The created pharmacophore model with a 12.42 selectivity score value is shown in Fig. 8. This is physiologically broad peptide selectivity from the OppA transporter itself of the LAB microbiota, which is re ected by its selectivity score of the identi ed pharmacophores on the OppA from B. longum KACC91563. In Fig. 5, the L.latics OppA had exclusive H-bonds with the peptide backbone of RPPGFSPFA via the residues Asn55, Ala57, Arg135, Arg416, Ser472, Ser474, and Ser477 of the OppA, with the exclusion of H-bonding to the side chain of the peptide. When the residues Ser472 and Ser474 of L.latics OppA did consistently interacts with other peptide (e.g., RDMPIQAF) backbones, irrespective of its binding a nity in even opened-ligand conformations [4], as shown in Figure S14. In contrast, the H-bond patterns of L.latics OppA compared to the H-bond register on the OppA from B. longum KACC91563 was formed between the residues Gly187, Ser188, and Arg440 of the OppA and the side chain of Arg(P1) and Phe(P8), in addition to achieving H-bonding to the peptide backbone for the residues Thr58, Ala60, Asp446, and Tyr447 of the OppA. In the only OppA of B. longum KACC91563, the H-bonding register with a side chain of Arg(P1) and Phe(P8) was specially de ned to a speci c determinant for the RPPGFSPFA binding, together with a hydrophobic interaction point (π-π interaction) between the Arg400 and the Phe(P8) to coincide with the shape constant of the targeted interaction sites. This would be caused by those different sizes of the peptide-binding cavity (4900Å 3 vs 2700Å 3 ) to explain the lower size limitation and the spacious speci city of the peptide-ligand in the L.latics OppA than in the OppA from B. longum KACC91563.
To con rm the key features of the OppA from B. longum KACC91563, we screened two similar 9-mer peptides with a central hydrophobic residue of leucine at positions 5 or 6 (Leu(P5) or Leu(P6) represented in Figure S3) by focusing on the assembly of seven pharmacophores based on the interactions with the RPPGFSPFA. The dissociation constants K D of binding to L.latics OppA were known to be 1.2 µM and 4.2 µM for SLSQLSSQS and SLSQSLSQS in the closed-ligand conformers, respectively [53]. Before screening the two peptides, we had docked both peptides into the OppA binding site from B. longum KACC91563, which was similar to docking the RPPGFSPFA peptide from the initial locations of their X-ray crystal structures (PDB code of 3YRA and 3RYB). Then, validating the pharmacophore model as required features allows us to guide the knowledge that these pharmacophores are most likely to be involved in common features related to the peptide binding a nity and contribute to interference t through their peptide speci city. It was observed as a common feature that the geometrical pharmacophores of two Hbonding with the peptide backbone of Leu(P5) on the SLSQLSSQS peptide to the key residues of Thr58 and Ala60 on the OppA. In the case of the SLSQSLSQS peptide, Leu(P6) was perfectly reproduced into the H-bonding record to complement Thr58 and Ala60, as shown in Figure S8. In both peptides, the central leucine residues at positions 5 and 6 were well tted into the hydrophobic shape constant, which In B. subtilis, the AppA of another oligopeptide importer [28] also favors hydrophobic nanopeptide substances with overlapping speci city to its OppA. Figure 6 shows the interaction interfaces of the template structure (PDB code of 1XOC) from B. subtilis AppA with its bound nonapeptide-ligand as the VDSKNTSSW peptide. In comparison of the binding constants in the B. subtilis AppA and the OppA from B. longum KACC91563 (Fig. 5 and Table S3), the peptide backbone of Val(P1) forms two H-bonds to the side chains of Asp154, Thr423, and Asn151 on the AppA. Indeed, the hydrophobic shape constant that surrounds the residues of Trp398, Leu401, Met405, Met407, Trp419 encloses Asn(P5) more through positioning two H-bonds between their backbones of the Asp43, Gly418, and Asn(P5). On the other hand, the Asp(P2) does not match the hydrophobic point constant but Trp(P9) does replace half occupancy in the conjugated hydrophobic shape constant on the side chain of the VDSKNTSSW peptide by the above pharmacophore tness (shown in Figure S9). No proline residues on the VDSKNTSSW peptide were likely to maintain the backbone conformer like the RPPGFSPFA binding within the contact interfaces of AppA through the H-bond network with both main chains. There seems to be no necessary requirement for the AppA to be proline-rich and the branched hydrophobic(P5) on the nonapeptide-ligand to have binding speci city. The AppA could bind to the bradykinin of RPPGFSPFR (the dissociation constants K D binding to B. subtilis AppA is 50.3 µM [59]).
Moreover, the peptide selectivity from E. facalis PreZ like other OppA and AppA templates mainly originated from the hydrophobic shape interactions between the side chain of cCF10 (7-mer pheromone peptide of LVTLVFV with a high binding a nity of 10pM to the PreZ [9]) and the extracellular pheromone receptor (in Table S3 and Fig. 6). Specially, the side chain of Leu(P1) is xed via a salt bridge with the Asp472 residue in the PreZ's hydrophobic pocket corresponding to Val(P2). The side chain of hydrophilic Thr(P3) is lined by the neighboring hydrophilic residues of Asn328, Gly468, and Gln470 on the PreZ with distinct binding constants from the effects of aggregation pores against other OppA and AppA templates. As a common binding constant of the central hydrophobic pore (in Table 1), the widen binding cavity to accommodate the larger peptide-ligands from E. facalis PreZ to L.latics OppA is associated with a steric clash between the domain II and side-chain of the P5 residue on the bound peptide by desceasing the water mediated H-bonding to bury the peptide in the binding pocket. Therefore, a spatial aggregation propensity (SAP) in the binding site of the extracellular SBP importer plays a vital role in the speci c interaction determinant for peptide binding (as that shown in Fig. 4). Lymphocytes are pro table in generating energy in the form of ATP by oxidative glycolysis known as the Warburg effect, unlike most differentiated cells by employing the TCA cycle (tricarboxylic acid cycle) and mitochondrial respiration. In the resting state, lymphocytes became catabolic utilizing autophagy that originates materials essential for protein synthesis and energy. Upon sensing speci c antigens by the Tcell receptor, the T-cell receptor markedly elevated the ability of T-cells to transport neutral amino acids as material sources into their plasma membrane. For example, leucine together with isoleucine and valine is abundant and accounts for more than 40% of the free ensential amino acids (at a protein-unbound state)

Discussion
in the blood plasma [61]. The activated T-cell pulled the trigger of metabolic switch by activating the glycolytic pathway, glutaminolysis, fatty acid synthesis, and pentose phosphate pathway while attenuating fatty acid oxidation [62][63][64][65]. In particular, the activated CD8 + T-cells take up tryptophan from TAT1 (aromatic amino acid as Phe, Tyr, Trp transporter also known as SLC16A10) into the cells by enkinding T-cell receptors and profusing leucines from CD98 (aliases LAT1 of L-leucine transpoter) through animating the mTORC (mechanistic target of rapamycin complex) signaling pathway [66,67]. Actually, the competence of mTORC can lead to control T-cell fate (denoted in Table S4) Figure S11 and S12(a)). Based on their substrate speci city, leucine or tryptophan requisites are quite connected to the xed sequential selectivity of only C-terminal residue from the antigenic peptide on the host TAP-mediated MHC class I presentation for CD8 + T-cell immunity.
Alternately, encountering the T-cell receptor by the host self peptide-MHC complex, leucine, and tryptophan are not indespensible to IL-7-induced CD8 + T-cell survival, but are required for sustaining native CD8 + T-cell size and CD8 + T-cell growth by interaction determinants of the IL-7 and IL-7 receptor (CD127). Consequently, both the leucine and tryptophan uptake of the T-cell are a key switch for metabolism reprogramming immune-activated T-cells, leading to an adjustment in adaptive immune responses. In this regard, in an infection resolution, antigen-speci c CD8 + T-cells de ate anabolism to the catabolism of nutrients from high to low mTOR activity, and thence from the effector T-cell to memory Tcell.
Similar to leucine, T-cell expansions crave tryptophan, by depleting which T-cell proliferation is blocked [69]. Tryptopan is one of the essential amino acids as a biosynthetic precursor of microbiota and host metabolities. This is because tryptopan is not biosynthesized by human and animal cells that rely on intake from exogenous dietary to catabolize bioactive components (such as kynurenine, serotonin, and metatonin). The tryptopan metabolism (described in Figure S12 aryl hydrocarbon receptor (AhR) ligands by intestinal microorganisms [70]. When induced by proin ammatory cytokines, kynurine serves as an AhR agonist and acts on the dendritic cell to inhibit the dendric cell mautation. There is more attunation to immune security via major antigen-presenting cells. Peculiarly, kynurenine can form the intestinal barrier and stimulate the proliferation of regulatory T-cell (Treg) to attenuate in ammation by enhancing the IL-10 generation. Then, decreasing the differentiation of highly in ammatory T H 17 cells reduces the production of IL-22. As the characterized neurotransmitter, serotonin (5-hydroxytryptamine, 5-HT) of another tryptophan metabolite plays neurobiologic functions and brain physiology such as mood stability, pain tolerances, or sleep patterns in the mammalian's central nervous system [71]. In spite of this, more than 90% of the body's serotonins are produced by the intestinal enterochroma n cells of the digestive tract and are also under the in uence of the intestinal microbiota by Streptococcus spp. and E. coli, which can synthesize serotonin themselves from tryptophan. Under the symptoms of serotonin dysbiosis, the peripheral serotonin does not cross the blood-brain barrier, but propagates signals from the intestine to intrinsic neurons. Then host immune cells are coupled with various serotonin receptors (at least fourteen members) on lymphocytes, monocytes, macropharge, T-cells, B-cells, and dendritic cells from the innate to adaptive immune system [72]. In addition, the serotonin-selective reuptake transpoter SERT (encoded by SLC6A4 gene) can remove serotonin from the intestinal epithelial cells to control serotonin bioavailability and to manage reuptake in the brain. Moreover, the intestinal commensal Lactobascillus can catabolize tryptophan to indole 3aldehyde as an AhR ligand to protect against mucosal in ammation by its tryptophanse. Indole is also a signal molecule of bacterial physiology for antibiotic resistance and bio lm formation, whereas in nonindol producing bacteria, indole and its offshoot inhibit quorum sensing and modulate harmfuless factors in intestinal microenvironment. Of note, indol itself further vitalizes enteroendocine L cells to produce glucagon-like peptide-1 (GLP-1) into insulin secretion by pancreatic β-cells.
As an important organ in the host body, the intestine behaves in the shared junction of nutrient digestion and absorption, microbiota colonization, and locates immune cells. The geographical proximate of the intestinal tract contributes a high level of immunity activity to eliminate ingested pathogens from its motility. A true characterization of the molecular interactions might provide valuable insights to decipher how mutualistic symbiosis-host's T-cell communication operates in amino acids for mediating intestinal immunity. In pigs, leucine and its metabolitie glutamine can be strength soruces to surpass mucosal barrier functions, the epithirial villus height in the intestine, and small intestinal growth [73][74]. In particular, only leucine achives in triggering an α-defensin secretion from Paneth cells among the 20 mammalian amino acids, thereby conducting surveillance in the intestine to maintain an intestinal homeostatis response to inner environement facts. In contrast, the isoleucine is prominent in the order of lymphocytes, eosinophils, and neutrophils on the host by inducing β-defensin with a downward trend in TNF-α and IL-10. Isoleucines also promote mucosal immunity and keep up intestinal integrity. Therefore, both leucine and isoleucine stimulate the intestinal SIgA (secretory immunoglobulin A) secretion that is the most abundant antibody to improve the mucosal surface defense in the intestinal mucosa, thereby leading to inhibiting the pathogen introgression into the lamina propria. It is well known that the glutamate and glutamine as leucine metabolities (as Figure S12(a)) comprise almost 70% proteinunbound amino acids (account for 5 ~ 10% total amino acids) in human milk. In fact, the free glutamate level was 40-fold higher in milk compared to plasma [75]. The effects of the glutamate and glutamine on immune parameters may diminish susceptibility relevant for allergic disease and infections in developing neonate over the course of lactation. Since at birth, the increased susceptibility in the neonatal immune system is derived from an immature intestinal barrier and incomplete microbial colonization compared to the adult. Fuctionally, the T-cell response to antigen was warped toward T H 2 immunity than innate CD4 + effector cell functions. In turn, the T H 2 dominant immunity milieu augments the susceptibility of allergenic sensitization while a lowered T H 1 function is conducted to increase the neonatal vulnerability to infection. Conversely, the postnatal period supplementation of glutamine had reinforced the production of T H 1 cytokines IL-2 and IFN-γ, thereby promoting T H 1 cells, whereas the descended secretion of T H 2 cytokines IL-4 and IL-5 is accompained by suppressing T H 2 activity. The glutamate released from dendritic cells has dual immunorgulating roles. Upon speci c antigen presentation, T-cell proliferation is enlivened by glutamate and the production of IL-2, IFN-γ on increasing T H 1 cells, and IL-10 on producing T reg 1 cells. Alternately, in non-speci c antigen presentation, the T-cell activity is obstructed by glutamate.
Neonates used glutamate and glutamine uptaked from the breastmilk by the intestines that have furnished the growth of intestinal epithelial cells and maturation of the intestinal barrier to support protective effects. In infants and weaning piglets, half of the dietary glutamate and glutamine is oxidized by intestinal epithelial cells and immune cells, ultimately leading to energy substrates for glutathione as an antioxidant in the preventive strategy against food allergies [76]. In addition, glutamine, but not gluatamate, can affect mucin synthesis as a precursor the attributable to the increase in the number of mucin-secreting globet cells safeguarding against allergenic sensitization.
As the other causative factor, except for shifting immune responses from T H 2 to T H 1 cells, doublestranded RNA from intestinal commensal lactic acid bacteria (including Lactobacillus and CRISPR family), but not pathogenic bacteria, is a natural ligand for Toll-like receptors TLR3 (CD283) and TLR9 (CD289) as a sensor of the commensal bacteria to trigger IFN-β production by bone marrow dendritic cells in the small intestine [77]. In light of this, CD11c + dendric cells in lamina propia or Peyer's patches may also induce IFN-β production to stabilize protective immunity against virus and pathogenic bacteria.
Besides the secretion of IFN-β, lactic acid bacteria keep their intestinal community by producing lactic acid and by lowering pH related to detergent overgrowth harm of pathogenic bacteria [78]. The ecological success of lactic acid bacteria relies on the substate speci cities of the extracellular proteins represented by the solute-binding proteins (e.g. OppA) of the ABC transport system and the metabolic activities of the glycoside hydrolase enzymes. This is possibly due to the exogenous protective effects accelerating the survival rate of the probiotic cells within the host gastrointestinal tract [79][80]78]. The metabolic substrates of the human milk oligosaccharides enhance the Bi dobacterial probiotic adhesions and colonisations to the host intestinal mucosa, especially during the host's early life [79][80][81]. In particular, members of the Bi dobacterium are among the key bacterial compontents (i.e., B. longum, B. dum, and B. breve species) of the infant intestines. This is because a higher abundance of Bi dobacteria was observed in the infant lactic acid bacteria than in the adult (from the BioProject PRJNA33914 aimed at exploring the vertical transmission of the microbiota from mothers to corresponding infants). Among lactic acid bacteria, Bi dobacterium longum KACC91563, isolated from fecal samples of healthy korean neonates, has the capability to alleviate food allergy effects [82,83]. The ability of B. logum KACC91563 was con rmed in the suppressed allergeic diarrhea from the mouse food allergy model induced by using ovalbumin and alum. Interestingly, B. logum KACC91563 had neither curtailing T H 2 cytokine levels nor in uencing anti-in ammatory cytokine IL-10 by Foxp3 + Treg cell [84]. As opposed to no effect on the Tcell immune responses, B. logum KACC91563 exerts in uence on descending only mast cell numbers via increasing the Annexin V + apoptotic bone marrow-derived mast cell. This results in dramatically decreasing food allergies in the mouse model. Thereupon, B. logum KACC91563-derived extracellular vesicles did much to ameliorate portent food allergy by instigating the mast cell apoptosis [82]. Factually, the extracellular vesicles from B. logum KACC91563 are more internalized in mediating unknown receptors on bone marrow-derived mast cells than phargocyte by dendritic cells. Then the unknown receptor to the extracellular vesicles (with the majority being 60 nm) would be different from some exosome transporter CD63 and CD82 that convey vesicles with virus like properties. Therefore, the extracellular vesicles from B. logum KACC91563 are enriched for the family's 5 solute binding protein, DNA, and lipids inside their lipid bilayer that more effectively transport these molecules than their soluble forms. Speci cally, the extracellular vesicles of B. logum KACC91563 are composed by an important component of OppA importer as a part of the family's 5 solute binding protein. This is because the OppA importer determined the substrate speci city of family's 5 solute binding protein through physiologic routes in the host intestine. The intestinal L. lactis populations and B. logum KACC91563 are also closely connected with selecting favored oligopeptides containing at least one branched residue of leucine, isoleucine, and valine via the OppA importer. This is because both organisms are auxotropic for branched amino acids, and they favor proline-rich caseins in milk media (9.8% of proline residues in milk casein [4]) as a nutrient source of these amino acids to fascilitate their growth. To satisfy the demands of the branched amino acids related to the physiology of both organisms, L. lactis and B. logum KACC91563 proteolytes the exogenous proteins (e.g. α-, β-, and κ-caseins in milk) to oligopeptides, which are transported by the OppABCDF membrane-spanning channel and further metabolized in the cytoplasm.
Their interplay augments the augmenting immune balance in the nutritional amino acid substrate [85] by adjusting for the compotent assistants-host immune crosstalk. On the other hand, the tted leucine or tryptophan in the C-terminal residue on the peptide substrate binding to host TAP redounded upon an allosteric crosstalk between the TMD and the NBD, which subsequently enkindles peptide trasmittion and ATP hydrolysis. In addition, the selected branched residues (e.g. leucine) are precursors of acetate in short-chain fatty acids (such as the acetoacetate shown in Figure S12(a)). The acetate products from B. logum assist the defense function of host epithelial cells [86], which are unliked to kill pathogenic enterobacteria via α-defensin secretion by butyric acid (one of the short chain fatty acids) and leucine in the Paneth cells [87]. The propionic and butyric acid metabolities were known to exert an antagonistic effect on colon cacinomacell proliferation, thereby inducing the apoptosis of the cancer cells [88]. Hence, it was attempting to explicate the relationship between the substrate's speci city from the OppA importer and the probiotic effects of B. logum KACC91563 in the host intestine, as illustrated in Fig. 9. Some factors such as pathogen invasions, proin ammatory cytokine, and toxins contribute to the lowering of the epithelial barrier functions in the host intestines. The probiotic effects of B. logum KACC91563 were stabilizing effects of induced tness on the OppA came from the favoritism of proline-rich (P2, P3, and P7) and hydrophobic Phe(P5), which also have impacted the OppA-RPPGFSPFA interactions for B. longum KACC91563. In addition, the 3D-pharmacophore arrangement to interaction features of the OppA on the complex strucrure-based is an arrangement of seven chemical features of two H-bond acceptors (HBA) and the features of three H-bond donors (HBD) vectors as well as two hydrophobic (HY) point features with location constraints when the leading hydrophobic shape constraints to branched hydrophobic residues was enforced as surrounding assembly of the peptide-ligand's pharmacophores with the excluded volume. In the only OppA of B. longum KACC91563, the H-bonding register with a side chain of Arg(P1) and Phe(P8) was speci cally de ned to a speci c determinant for the RPPGFSPFA binding, together with a hydrophobic interaction point (π-π interaction) between the Arg400 and the Phe(P8) to coincide with the shape constant of the target interaction sites with from 2700Å of the peptide-binding cavity. The intestinal B. longum KACC91563 is auxotropic for branched amino acids, and favors proline-rich caseins in milk media (9.8% proline residues in milk casein) as a nutrient source of these amino acid to fascilitate their growth. To satisfy the demands of the branched amino acids related to the physiology of B. longum KACC91563 proteolysis the exogenous proteins (e.g. α-, β-, and κ-caseins in milk) to oligopeptides, which are transported by the OppABCDF membrane-spanning channel and further metabolized in the cytoplasm. Their interplay arguments the augmenting immune blance in the nutritional amino acid substrate by adjusting for the compotent assistents-host immune crosstalk. The specialized structure-function relationship from the OppA import could provide the abstract of substrate speci city into unique immunological prpperties of the host organism. In the current study, we attempted explication of the relationship between the substrate's speci city from the OppA importer and the probiotic effects of B. longum KACC91563 in the host intestine based on the structure-function perspectives of the OppA importer. Moreover, functional characterization of solute-binding proteins (such as 15 cell wall proteins and 20 extracellular proteins) on the B. longum KACC91563 genome will lead to insight of key switch for substate's metabolism into reprogramming immune responses in the host intestine.

Declarations
Ethics approval and consent to participate Not applicable. The current study was permormed in computer aided molecular modeling in silico.

Consent for publication
Not applicable Avalilability of data and material All structure based modeling data were generated and analyzed during this study are available.
Please contact author for data requests.