Binding Specificity of a Novel Cyclo/Maltodextrin-Binding Protein and Its Role in the Cyclodextrin ABC Importer System from Thermoanaerobacterales

Extracellular synthesis of functional cyclodextrins (CDs) as intermediates of starch assimilation is a convenient microbial adaptation to sequester substrates, increase the half-life of the carbon source, carry bioactive compounds, and alleviate chemical toxicity through the formation of CD-guest complexes. Bacteria encoding the four steps of the carbohydrate metabolism pathway via cyclodextrins (CM-CD) actively internalize CDs across the microbial membrane via a putative type I ATP-dependent ABC sugar importer system, MdxEFG-(X/MsmX). While the first step of the CM-CD pathway encompasses extracellular starch-active cyclomaltodextrin glucanotransferases (CGTases) to synthesize linear dextrins and CDs, it is the ABC importer system in the second step that is the critical factor in determining which molecules from the CGTase activity will be internalized by the cell. Here, structure-function relationship studies of the cyclo⁄maltodextrin-binding protein MdxE of the MdxEFG-MsmX importer system from Thermoanaerobacter mathranii subsp. mathranii A3 are presented. Calorimetric and fluorescence studies of recombinant MdxE using linear dextrins and CDs showed that although MdxE binds linear dextrins and CDs with high affinity, the open-to-closed conformational change is solely observed after α- and β-CD binding, suggesting that the CM-CD pathway from Thermoanaerobacterales is exclusive for cellular internalization of these molecules. Structural analysis of MdxE coupled with docking simulations showed an overall architecture typically found in sugar-binding proteins (SBPs) that comprised two N- and C-domains linked by three small hinge regions, including the conserved aromatic triad Tyr193/Trp269/Trp378 in the C-domain and Phe87 in the N-domain involved in CD recognition and stabilization. Structural bioinformatic analysis of the entire MdxFG-MsmX importer system provided further insights into the binding, internalization, and delivery mechanisms of CDs. Hence, while the MdxE-CD complex couples to the permease subunits MdxFG to deliver the CD into the transmembrane channel, the dimerization of the cytoplasmatic promiscuous ATPase MsmX triggers active transport into the cytoplasm. This research provides the first results on a novel thermofunctional SBP and its role in the internalization of CDs in extremely thermophilic bacteria.

Recently, we reported the entire CM-CD pathway from extremely thermophilic bacteria (T opt ≥ 70 • C) by exploring~246 (meta)genomes from microbial communities living in a wide variety of hot environments on Earth [13]. Sequence analysis revealed that Caldanaerobacter subterraneus ssp., Thermoanaerobacter spp., and Thermoanaerobacterium spp. encoded an exceptional gene cluster of~30 genes (named cld, thm, and thb, respectively) that encrypts the proteins related to the four steps of the CM-CD pathway from Thermoanaerobacterales, involving synthesis, transportation, degradation, and metabolic assimilation of CDs from starch. In the first step, extracellular thermophilic three-domain CGTases convert the surrounding starch substrate to α-, β-, and γ-CDs, as well as linear dextrins and fermentable sugars [13]. According to the sequence analysis of proteins encoded in the cld/thm/thb gene clusters [13], while degradation of CDs in the third step of the pathway occurs in the cytoplasm by the action of the functionally characterized cyclodextrinase (CDase), α-glucan phosphorylase (GP), and glucoamylase (GA) enzymes [28][29][30], the metabolic assimilation in the fourth step follows the typical Embden-Meyerhof-Parnas (EMP) glycolytic pathway, which includes a phosphoglucose isomerase (Pgi), 6-phosphofructokinase (PfkA) and the functionally characterized pyruvate kinase (PykF) [13,31]. Sequence analysis also showed that in the second step of the CM-CD pathway, a mdxEFG cassette that is adjacent to the CGTase-encoding gene, as well as a promiscuous ATPase, MsmX, distally located from the cld/thm/thb gene clusters, complete the entire putative cyclo/maltodextrin ABC importer system, MdxEFG-MsmX, from the Thermoanaerobacterales order [13]. Similar to the sulfur-reducing hyperthermophilic Themoccoci archaea [15,32,33], the synthesis of functional CDs as intermediates of the CM-CD pathway appears to be significant in Thermoanaerobacterales to thrive under extremophilic starch-poor environments [13].
Here, structure-function relationship studies of the novel cyclo/maltodextrin-binding protein MdxE from Thermoanaerobacter mathranii subsp. mathranii A3 (NCBI Taxonomy ID: 583358) are presented. Calorimetric and fluorescence studies of recombinant MdxE using linear dextrins and α-, β-, and γ-CDs showed that the open-to-closed conformational change is solely triggered after αand β-CD binding. Experimental confirmation of MdxE-(α-/β-CD) complex formation, along with docking simulations, homology modeling, and structural bioinformatic analysis of the entire CD MdxFG-MsmX importer system from this extremely thermophilic bacterium, revealed the structural keys that appear to be involved in the binding, internalization, and delivery mechanisms of a novel and intriguing thermophilic ABC importer system for functional CDs.

MdxE Production and Purification
To evaluate the binding affinities for linear dextrins and CDs, as well as the associated open-to-closed conformational changes of SBP MdxE, from T. mathranii subsp. mathranii, a recombinant form was successfully produced in E. coli SHuffle T7. The truncated form of MdxE consists of 380 residues with a calculated molecular mass of 41.5 kDa, including a C-terminal His6-tag sequence without the first 24 residues (Met1-Gly24) of the signal peptide (SP) and the N-terminal intrinsically disordered region (Cys25-Pro46) adjacent to the SP. Hence, to facilitate the recombinant production and functional characterization of MdxE, the conserved Cys25 residue crucial to anchoring the SBPs from G+ to the cytoplasmic membrane outer surface through a covalently bound lipid moiety was deleted in the truncated form [13,34]. Protein purification was performed in three sequential steps: heat treatment, nickel-affinity chromatography, and size-exclusion chromatography (SEC)-dynamic light scattering (DLS) coupled experiments ( Figure 1A), yielding~13 mg of purified MdxE per liter of culture according to the final Bradford quantification assay. The major peak collected after SEC-DLS analysis corresponding to 71.4% of the sample injected showed a calculated polydispersity index (M w /M n ) of 1.04 and a molecular mass of 48.8 kDa (see the inset in Figure 1A), indicating that the biological assembly of MdxE is monomeric (peak 2). The electrophoretic profile in Figure 1B shows that recombinant peak collected after SEC-DLS analysis corresponding to 71.4% of the sample injected showed a calculated polydispersity index (Mw/Mn) of 1.04 and a molecular mass of 48.8 kDa (see the inset in Figure 1A), indicating that the biological assembly of MdxE is monomeric (peak 2). The electrophoretic profile in Figure 1B shows that recombinant MdxE was successfully purified and lies between 37-50 kDa bands, which was expected according to MdxE calculated molecular mass. Note that the MdxE aggregates (4021 kDa, peak 1) correspond to 28.6% of the total injected sample, showing the tendency of MdxE to form aggregates. (B) Coomassie Bluestained SDS-PAGE gel (12%). Lane 1, molecular-weight markers (Bio-Rad, labeled in kDa). Lane 2, insoluble fraction of MdxE production. Lane 3, soluble fraction of MdxE production after a heat treatment procedure (60 °C, 20 min). Lane 4, purified MdxE after Ni 2+ -affinity chromatography. Lane 5, purified recombinant MdxE with optimal monodispersity (Mw/Mn = 1.04, peak 2) after SEC-DLS analysis.

Homology Modeling of MdxE
To obtain insights into the structural basis of MdxE, homology modeling was applied to build the three-dimensional (3D) structure ( Figure 2). The overall architecture of MdxE comprises two globular N-and C-domains, as typically found in SBPs [21,24,35]; both domains consist of a β-sheet core surrounded by α-helices linked by hinge regions that allow switching from an open (sugar-free or nonfunctional sugar-bound) to a closed form (functional sugar-bound) for cellular internalization [21,26]. Structural comparisons of MdxE with the maltodextrin-binding protein from Alicyclobacillus acidocaldarius (AcyMBP, PDB ID: 1URD) [35], which superpose 357 C α with MdxE with an r.m.s.d. of 0.22 Å, allowed the identification of residues involved in the different interactions of both domains. Thus, while the N-domain consists of 46-152 and 302-351 residues, the C-domain consists of 158-296 and 306-425 residues. Hinge regions located at the bottom of the sugar-binding site of MdxE and comprising three small segments, 153-157, 297-301, and 352-355 residues, were identified by structural comparisons with TvuCMBP (PDB ID: 2ZYN) [36], which superpose 320 C α with MdxE with an r.m.s.d. of 2.2 Å. Note that the MdxE aggregates (4021 kDa, peak 1) correspond to 28.6% of the total injected sample, showing the tendency of MdxE to form aggregates. (B) Coomassie Blue-stained SDS-PAGE gel (12%). Lane 1, molecular-weight markers (Bio-Rad, labeled in kDa). Lane 2, insoluble fraction of MdxE production. Lane 3, soluble fraction of MdxE production after a heat treatment procedure (60 • C, 20 min). Lane 4, purified MdxE after Ni 2+ -affinity chromatography. Lane 5, purified recombinant MdxE with optimal monodispersity (Mw/Mn = 1.04, peak 2) after SEC-DLS analysis.

Homology Modeling of MdxE
To obtain insights into the structural basis of MdxE, homology modeling was applied to build the three-dimensional (3D) structure ( Figure 2). The overall architecture of MdxE comprises two globular N-and C-domains, as typically found in SBPs [21,24,35]; both domains consist of a β-sheet core surrounded by α-helices linked by hinge regions that allow switching from an open (sugar-free or nonfunctional sugar-bound) to a closed form (functional sugar-bound) for cellular internalization [21,26]. Structural comparisons of MdxE with the maltodextrin-binding protein from Alicyclobacillus acidocaldarius (AcyMBP, PDB ID: 1URD) [35], which superpose 357 C α with MdxE with an r.m.s.d. of 0.22 Å, allowed the identification of residues involved in the different interactions of both domains. Thus, while the N-domain consists of 46-152 and 302-351 residues, the C-domain consists of 158-296 and 306-425 residues. Hinge regions located at the bottom of the sugar-binding site of MdxE and comprising three small segments, 153-157, 297-301, and 352-355 residues, were identified by structural comparisons with TvuCMBP (PDB ID: 2ZYN) [36], which superpose 320 C α with MdxE with an r.m.s.d. of 2.2 Å.  As seen in other SBPs from bacteria, the MdxE homology model showed the conserved aromatic triad Tyr193, Trp269, and Trp378 in the C-domain ( Figure 2B) that recognizes different sugars through CH-π interactions [36,38]. Accordingly, structural comparisons revealed that the aromatic triad in MdxE is identical to the aromatic triad Tyr175/Trp250/Trp360 in TvuCMBP ( Figure 2C) and Tyr155/Trp230/Trp340 in the wellstudied maltodextrin-binding protein from E. coli (EcoMBP, PDB ID: 1ANF) [39], Figure  2D), strongly suggesting that it has an essential role in sugar recognition. Conversely, the N-domain face at the sugar-binding cleft of MdxE exhibits the aromatic Phe87 residue ( Figure 2B) at the same position as Leu59 and Trp62 in TvuCMBP and EcoMBP, respectively ( Figure 2C,D) [36,39], which have been proposed to play an important role in adopting the closed form in (TvuCMBP/EcoMBP)-ligand complex formation. Hence, while Leu59 in TvuCMBP orients toward the central cavity of CDs for its stabilization [21,36], Trp62 is essential for G2 accommodation in the sugar-binding site of EcoMBP (PDB ID: 1ANF) [37,39]. A similar mechanism has been reported for CGTases, in which a central aromatic residue is crucial for CD formation (cyclization activity) [40,41], as well as cyclodextrinases (CDases), in which an aromatic residue is critical in stabilizing CDs at the active site for hydrolytic activity [21,28]. This information suggests that MdxE has a typical sugar-binding site contour specifically adapted to bind and deliver cyclo/maltodextrins to MdxFG permeases in the second step of the CM-CD pathway from T. mathranii subsp. mathranii. As seen in other SBPs from bacteria, the MdxE homology model showed the conserved aromatic triad Tyr193, Trp269, and Trp378 in the C-domain ( Figure 2B) that recognizes different sugars through CH-π interactions [36,38]. Accordingly, structural comparisons revealed that the aromatic triad in MdxE is identical to the aromatic triad Tyr175/Trp250/Trp360 in TvuCMBP ( Figure 2C) and Tyr155/Trp230/Trp340 in the well-studied maltodextrin-binding protein from E. coli (EcoMBP, PDB ID: 1ANF) [39], Figure 2D), strongly suggesting that it has an essential role in sugar recognition. Conversely, the N-domain face at the sugar-binding cleft of MdxE exhibits the aromatic Phe87 residue ( Figure 2B) at the same position as Leu59 and Trp62 in TvuCMBP and EcoMBP, respectively ( Figure 2C,D) [36,39], which have been proposed to play an important role in adopting the closed form in (TvuCMBP/EcoMBP)ligand complex formation. Hence, while Leu59 in TvuCMBP orients toward the central cavity of CDs for its stabilization [21,36], Trp62 is essential for G2 accommodation in the sugar-binding site of EcoMBP (PDB ID: 1ANF) [37,39]. A similar mechanism has been reported for CGTases, in which a central aromatic residue is crucial for CD formation (cyclization activity) [40,41], as well as cyclodextrinases (CDases), in which an aromatic residue is critical in stabilizing CDs at the active site for hydrolytic activity [21,28]. This information suggests that MdxE has a typical sugar-binding site contour specifically adapted to bind and deliver cyclo/maltodextrins to MdxFG permeases in the second step of the CM-CD pathway from T. mathranii subsp. mathranii.

Determination of MdxE Binding Affinities
To assess the binding affinities of MdxE toward different linear dextrins and CDs for the CM-CD pathway, functional experiments of MdxE-ligand complex formation were performed using isothermal titration calorimetry (ITC) to determine the dissociation constant (K d ) and enthalpy (∆H) for ligand binding, as well as the Gibbs free energy (∆G) contribution to the binding ( Figure 3).

Determination of MdxE Binding Affinities
To assess the binding affinities of MdxE toward different linear dextrins and CDs for the CM-CD pathway, functional experiments of MdxE-ligand complex formation were performed using isothermal titration calorimetry (ITC) to determine the dissociation constant (Kd) and enthalpy (ΔH) for ligand binding, as well as the Gibbs free energy (ΔG) contribution to the binding ( Figure 3). As summarized in Table 1, MdxE showed the highest affinity toward α-CD, with a Kd value of 0.61 µM. Nevertheless, although the Kd values for β-CD and γ-CD were 2.04 and 1.51 µM, respectively, the affinity of MdxE for linear dextrins was directly proportional to the polymerization degree (DP). Hence, while the Kd value for G7 was 0.91 µM, the affinity of MdxE for G3 was much lower (Kd, 3.02 µM), and no binding interaction was observed for G2 (Table 1). In fact, despite titration with G2 resulting in short heat pulses, no sigmoidal model could be fitted (Figure 3), suggesting very low affinity or no binding interaction between MdxE and G2. This behavior contrasts with the results observed for TvuCMBP, AcyMBP, and EcoMBP, which showed a high affinity toward G2 [21,22,39]. However, a similar finding has been reported for CD-binding protein from K. oxytoca (CymE) since the quenching effect of G2 during fluorescence assays to determine Kd was solely marginal compared to CDs and linear dextrins with high DP [14]. Accordingly, since extracellular three-domain ABC CGTases involved in the first step of the CM-CD pathway from Thermoanaerobacterales recycle G2 for disproportionation activity to produce linear dextrins with a DP > 3 [7,13], the latter observations suggest that the inability of MdxE to bind G2 might be related to the activation of a different internalization pathway than CM-CD for residual G2. Indeed, G2, G3, and α-trehalose are internalized by the same catabolite-repressible ABC importer system in Thermoanaerobacter ethanolicus [42]. Similarly, K. oxytoca and B. subtilis express an independent maltose/maltodextrin (mal) system for G2 internalization, as the CM-CD represents a secondary pathway for these As summarized in Table 1, MdxE showed the highest affinity toward α-CD, with a K d value of 0.61 µM. Nevertheless, although the K d values for β-CD and γ-CD were 2.04 and 1.51 µM, respectively, the affinity of MdxE for linear dextrins was directly proportional to the polymerization degree (DP). Hence, while the K d value for G7 was 0.91 µM, the affinity of MdxE for G3 was much lower (K d , 3.02 µM), and no binding interaction was observed for G2 (Table 1). In fact, despite titration with G2 resulting in short heat pulses, no sigmoidal model could be fitted (Figure 3), suggesting very low affinity or no binding interaction between MdxE and G2. This behavior contrasts with the results observed for TvuCMBP, AcyMBP, and EcoMBP, which showed a high affinity toward G2 [21,22,39]. However, a similar finding has been reported for CD-binding protein from K. oxytoca (CymE) since the quenching effect of G2 during fluorescence assays to determine K d was solely marginal compared to CDs and linear dextrins with high DP [14]. Accordingly, since extracellular three-domain ABC CGTases involved in the first step of the CM-CD pathway from Thermoanaerobacterales recycle G2 for disproportionation activity to produce linear dextrins with a DP > 3 [7,13], the latter observations suggest that the inability of MdxE to bind G2 might be related to the activation of a different internalization pathway than CM-CD for residual G2. Indeed, G2, G3, and α-trehalose are internalized by the same catabolite-repressible ABC importer system in Thermoanaerobacter ethanolicus [42]. Similarly, K. oxytoca and B. subtilis express an independent maltose/maltodextrin (mal) system for G2 internalization, as the CM-CD represents a secondary pathway for these microorgan- isms [14,33,43,44]. In agreement, the genome of T. mathranii subsp. mathranii encodes three putative ABC sugar importer systems ( Figure S1), one of which includes a putative SBP (NCBI ID: WP_013149854.1) that shares~50% sequence identity with the characterized G2-binding protein from Thermotoga maritima (tmMBP, PDB ID: 6DTQ) [45].
Compared to MdxE, similar K d values for linear dextrins and CDs have also been determined for various SBPs (Table 1). Nevertheless, except for G2, the ITC analysis cannot determine whether MdxE adopts the open form with a nonfunctional sugar-bound (unproductive complex) or the closed form with the sugars listed in Table 1. An initial approach would be inferring that the magnitude of the binding affinities observed for MdxE in Table 1 is directly proportional to the probability of switching from an open to a closed form to deliver ligands into the MdxFG permeases from the ABC importer system. Nevertheless, the crystallographic structures of TvuCMBP and EcoMBP reveal the opposite [36,39,46]. For instance, although the K d values of EcoMBP are identical for G2 and β-CD (Table 1), EcoMBP complexed with G2 adopts the closed form (PDB ID: 1ANF, Figure 2D) [39], while the open form is observed when complexed with β-CD (unproductive complex, PDB ID: 1DMB, [46]. Likewise, although the K d values of TvuCMBP are similar for linear dextrins and CDs (Table 1), TvuCMBP complexed with α-, β-, and γ-CDs adopts the closed form (PDB IDs: 2ZYM, 2ZYN, 2ZYK, Figure 2C) [21,36], whereas the open form is observed when complexed with linear dextrins (unproductive complex, PDB ID: 2ZYO) [36], despite β-CD even showing lower affinity (K d = 1.2 µM) than linear dextrins. Hence, although ITC experiments confirm when an SBP has no inherent ligand-binding capacity, such as MdxE with G2 (Figure 3), additional experiments are required to determine the functional dynamics associated with binding specificity.

Determination of the Open-to-Closed Conformational Change by MdxE
To evaluate the switching from an open to a closed form of productive MdxE-ligand complexes obtained with the sugars listed in Table 1, changes in the protein surface hydrophobicity (H o ) of MdxE-ligand complexes using the fluorescent probe 8anilinonaphthalene-1-sulfonate (ANS) were determined ( Figure 4). Since H o is governed by the slope generated as the fluorescence response changes upon binding interaction [47,48], increasing MdxE concentrations (2-14 µM) inside a linear range ( Figure S2) were tested using an excess of ligand (350 µM) at a fixed concentration of ANS (20 µM) to assure that MdxE was in its ligand-saturated state based on ITC studies ( Figure 3). First, a notable change in fluorescent response at λ max of 485 nm was observed with increasing MdxEligand concentrations at a constant ANS concentration (10 µM) ( Figure 4A). Since no effect on fluorescent response was observed between ANS and ligands ( Figure 4B), this change could be attributed to the productive sugar-binding interactions that allowed obtaining a MdxE closed form. In fact, docking simulations of MdxE models in open and closed conformations with ANS ( Figure S3) resulted in a striking difference in the total number of ANS molecules binding to MdxE in the open form compared to the closed form. Accordingly, while 18 ANS molecules were observed binding different hydrophobic sites on the MdxE surface in the closed form, only 11 probes were found in the open form. This observation indicates that only upon binding CDs and linear dextrins that the cell will functionally internalize does MdxE undergo a conformational change toward the closed form (functional sugar-bound), exposing more hydrophobic sites for ANS molecules and thus increasing H o . Notably, the slopes obtained from the linear fit model at λ max = 485 nm in Figure 4C shows significant differences (p < 0.05) between the blank, MdxE-ANS (open form), and the CDs and linear dextrins tested, whose H o value associated with the open-toclosed conformational change remains statistically invariant compared to the blank. As expected from the ITC analysis (Figure 3), no MdxE-G2 complex formation was observed by fluorescence measurements (Figure 4). The remaining CDs and linear dextrins tested were grouped according to H o values for further statistical analysis ( Table 2).  Figure 4C shows significant differences (p < 0.05) between the blank, MdxE-ANS (open form), and the CDs and linear dextrins tested, whose Ho value associated with the opento-closed conformational change remains statistically invariant compared to the blank. As expected from the ITC analysis (Figure 3), no MdxE-G2 complex formation was observed by fluorescence measurements (Figure 4). The remaining CDs and linear dextrins tested were grouped according to Ho values for further statistical analysis ( Table 2). According to Tukey's multiple comparisons test (Table S1), there were no significant differences (p < 0.05) between the ligands of the first group, G3-G7/γ-CD, suggesting a similar conformational change as a function of Ho values that might be associated with an inactive semiclosed/open form of MdxE-(G3-G7/γ-CD) complexes (Table 2). Nevertheless, the ligands of the second group, α-and β-CDs, placed in an upper region with significantly different Ho values (p < 0.05) compared to the first group, showed that they have functional effects by triggering the closed form of MdxE for cellular internalization. As expected, the Ho value for MdxE-G2 complex formation was close to that of the blank ( Table 2), confirming that MdxE has no affinity for G2. Although the thermodynamic parameters determined by ITC analysis revealed that the formation of the MdxE-ligand complexes is an exothermic and spontaneous process, the fluorescence analysis showed that the open-to-closed conformational change is solely triggered by α-and β-CDs (Table 2). Since α-and β-CDs are the main cyclization products from the starch substrate of threedomain ABC CGTases from Thermoanaerobacterales [13], these results strongly suggest that the CM-CD pathway is exclusive for transportation, degradation, and metabolic assimilation of α-and β-CDs through the type I ABC MdxEFG-MsmX importer system. According to Tukey's multiple comparisons test (Table S1), there were no significant differences (p < 0.05) between the ligands of the first group, G3-G7/γ-CD, suggesting a similar conformational change as a function of H o values that might be associated with an inactive semiclosed/open form of MdxE-(G3-G7/γ-CD) complexes (Table 2). Nevertheless, the ligands of the second group, αand β-CDs, placed in an upper region with significantly different H o values (p < 0.05) compared to the first group, showed that they have functional effects by triggering the closed form of MdxE for cellular internalization. As expected, the H o value for MdxE-G2 complex formation was close to that of the blank ( Table 2), confirming that MdxE has no affinity for G2. Although the thermodynamic parameters determined by ITC analysis revealed that the formation of the MdxE-ligand complexes is an exothermic and spontaneous process, the fluorescence analysis showed that the open-toclosed conformational change is solely triggered by αand β-CDs (Table 2). Since αand β-CDs are the main cyclization products from the starch substrate of three-domain ABC CGTases from Thermoanaerobacterales [13], these results strongly suggest that the CM-CD pathway is exclusive for transportation, degradation, and metabolic assimilation of αand β-CDs through the type I ABC MdxEFG-MsmX importer system.  Table S1). ND = not detected.

Structural Basis for the Open-to-Closed Conformational Change of MdxE
To identify the critical residues in MdxE that are involved in CD recognition for the CM-CD pathway, docking simulations of MdxE in the open and closed forms with αand β-CDs ( Figure 5, Table S2) were carried out using the coordinates of the MdxE model as a template for calculations (Table S3) (Table S2). This latter observation correlates with the functional studies for αand β-CDs (Figures 3 and 4), suggesting an energetically favorable CD binding by MdxE in the closed form to deliver them into the MdxFG permeases from the ABC importer system.
Regarding the residues from the N-and C-domains involved in the CD-binding site of MdxE, as well as hinge regions at the bottom of the cleft (Figure 5), the docking simulations showed the pair Asn191/Asp270 from the C-domain forming hydrogen bonds with α-, β-, and γ-CDs and the aromatic triad Tyr193/Trp269/Trp378 forming hydrophobic interactions with three contiguous G1 from the three CDs. Owing to the differences in the CD size, while the hydrogen bonds between α-, β-, and γ-CDs and Asp86/Asp109/Asn110 from the N-domain are conserved, the hydrogen bond between Gln84 and α-CD is absent, and a hydrogen bond between Gln88 and γ-CD is gained. Likewise, although the hydrogen bond between α-, β-, and γ-CDs and the main chain of His54 is conserved, the hydrogen bond formed by Ser85 was solely observed for αand β-CDs, but two hydrogen bonds between γ-CD and Asp250/Lys253 were gained. Furthermore, docking simulations of MdxE in the open and closed forms with linear dextrins were also performed ( Figure S4). Thus, while acceptable binding ∆G values for the closed form of MdxE/(G3-G7) complexes were found (Table S2), fewer interactions than those observed for CDs ( Figure 5) between the linear dextrins and the residues of the sugar-binding site were found, suggesting the formation of unproductive complexes between MdxE and linear dextrins that display inactive semiclosed/open forms ( Figure S4, Table S2).    (Table S2). Structural analyses revealed that Glu155 and Gln301 from hinge regions I and II, respectively ( Figure 6), form concerted hydrogen bonds with solely αand β-CDs ( Figure 5A,C), suggesting that they act together to provide part of the driving force to ensure the bending of MdxE and the formation of the closed form. Accordingly, the dynamics of the open-to-closed conformational changes of TvuCMBP crystal structures revealed that the critical factor of this reorganization is centered on the motion of hinge regions I and II at the bottom of the sugar-binding site [36]. Strikingly, amino acid sequence alignment between the hinge regions of TvuCMBP, EcoMBP, and MdxE showed that Glu155 of MdxE ( Figure 5) is found at the exact position as the conserved Glu111 and Glu129 from hinge region I of EcoMBP and TvuCMBP, respectively, which is involved in adopting the closed form of (TvuCMBP/EcoMBP)-ligand complexes ( Figure 6) [36]. Although Gln301 was also observed to form hydrogen bonds with γ-CD ( Figure 5E) and G3/G5/G7 ( Figure S4) in the docking simulations of the closed forms of MdxE, the conserved Glu155 is exclusive for MdxE/(α-/β-CD) complexes, strongly suggesting its structural role in the open-to-closed conformational changes of MdxE. Nevertheless, future X-ray crystal structure determination, site-specific mutants, and functional studies are needed to determine the function of Glu155 in MdxE. Nevertheless, since subtle structural changes in the hinge regions of MdxE-ligand complexes were observed, special attention was focused on the Venus flytrap hinge-bending motion of MdxE by comparing the open and closed forms of all docked structures (Table S2). Structural analyses revealed that Glu155 and Gln301 from hinge regions I and II, respectively (Figure 6), form concerted hydrogen bonds with solely α-and β-CDs (Figures 5A,C), suggesting that they act together to provide part of the driving force to ensure the bending of MdxE and the formation of the closed form. Accordingly, the dynamics of the open-to-closed conformational changes of TvuCMBP crystal structures revealed that the critical factor of this reorganization is centered on the motion of hinge regions I and II at the bottom of the sugar-binding site [36]. Strikingly, amino acid sequence alignment between the hinge regions of TvuCMBP, EcoMBP, and MdxE showed that Glu155 of MdxE ( Figure 5) is found at the exact position as the conserved Glu111 and Glu129 from hinge region I of EcoMBP and TvuCMBP, respectively, which is involved in adopting the closed form of (TvuCMBP/EcoMBP)-ligand complexes ( Figure 6) [36]. Although Gln301 was also observed to form hydrogen bonds with γ-CD ( Figure 5E) and G3/G5/G7 ( Figure S4) in the docking simulations of the closed forms of MdxE, the conserved Glu155 is exclusive for MdxE/(α-/β-CD) complexes, strongly suggesting its structural role in the open-to-closed conformational changes of MdxE. Nevertheless, future X-ray crystal structure determination, site-specific mutants, and functional studies are needed to determine the function of Glu155 in MdxE.

Structural Insights into the Internalization Mechanism of CDs from Thermoanaerobacterales
To propose an internalization mechanism for α-and β-CDs by the CM-CD pathway from Thermoanaerobacterales, homology modeling was applied to build the 3D structures of the extracellular CGTase and the entire MdxEFG-MsmX importer system (Figure 7). Structural comparisons of the putative extracellular three-domain CGTase from T. mathranii subsp. mathranii (ThmA) with the crystallographic structure of the three-domain CGTase from C. subterraneus (CldA, PBD ID: 6WNI, sequence identity: 83%), which superposes 437 C α with ThmA with an r.m.s.d. of 0.40 Å, allowed the identification of the central aromatic residue Phe218, the catalytic triad Asp252/Glu281/Asp353, and the hydrophobic clamp Met283/Trp206 in ThmA, which are essential for producing CDs and linear dextrins from the starch substrate [7,13]. The entire model of the ABC MdxEFG-MsmX importer system from Thermoanaerobacterales (Table S3, Figure 7) was built upon the crystallographic structure of MalEFG-MalK from E. coli (OF, PDB ID: 3PUV) [50], which superposes 938 C α with MdxEFG-MsmX with an r.m.s.d. of 2.9 Å, as it is the only sugar type I ABC importer system well studied thus far. Several residues of MdxEFG-MsmX involved in ligand translocation across the microbial membrane were identified. Accordingly, the spoon loop (P3) in the permease MdxG containing the conserved Gln256 that seems to be involved in removing α-and β-CDs from the MdxE-binding site was found at the exact position as the conserved Gln256 from MalG (PDB ID: 3PUV,

Structural Insights into the Internalization Mechanism of CDs from Thermoanaerobacterales
To propose an internalization mechanism for αand β-CDs by the CM-CD pathway from Thermoanaerobacterales, homology modeling was applied to build the 3D structures of the extracellular CGTase and the entire MdxEFG-MsmX importer system (Figure 7). Structural comparisons of the putative extracellular three-domain CGTase from T. mathranii subsp. mathranii (ThmA) with the crystallographic structure of the three-domain CGTase from C. subterraneus (CldA, PBD ID: 6WNI, sequence identity: 83%), which superposes 437 C α with ThmA with an r.m.s.d. of 0.40 Å, allowed the identification of the central aromatic residue Phe218, the catalytic triad Asp252/Glu281/Asp353, and the hydrophobic clamp Met283/Trp206 in ThmA, which are essential for producing CDs and linear dextrins from the starch substrate [7,13]. The entire model of the ABC MdxEFG-MsmX importer system from Thermoanaerobacterales (Table S3, Figure 7) was built upon the crystallographic structure of MalEFG-MalK from E. coli (OF, PDB ID: 3PUV) [50], which superposes 938 C α with MdxEFG-MsmX with an r.m.s.d. of 2.9 Å, as it is the only sugar type I ABC importer system well studied thus far. Several residues of MdxEFG-MsmX involved in ligand translocation across the microbial membrane were identified. Accordingly, the spoon loop (P3) in the permease MdxG containing the conserved Gln256 that seems to be involved in removing αand β-CDs from the MdxE-binding site was found at the exact position as the conserved Gln256 from MalG (PDB ID: 3PUV, Figure 7A) [50,51]. Likewise, the aromatic triad Trp119/Phe177/Tyr230 in the permease MdxF, which stabilizes the αand β-CDs through CH-π interactions ( Figure 7B), was identified by structural comparisons with the homologous subunit MalF (PDB ID: 3PUV) [51]. Moreover, the coupling helixes (CH) in MalFG, responsible for the transmembrane domain-nucleotide-binding domain (TMD-NBD) coupling that regulates the conformational changes triggered by ATP hydrolysis [52], were identified in MdxFG ( Figure 7C). Additionally, Glu190 in both subunits of MdxFG was found to make an electrostatic interaction with Arg48 in the active site of NBDs ( Figure 7C), suggesting that the interaction could be correlated to the dimerization of the cytoplasmatic promiscuous ATPase, MsmX, as previously found in MalEFG-MalK [53]. Finally, the conserved LSGGQ motif, Walker A and B, as well as several motifs involved in ATP hydrolysis by MsmX ( Figure 7D) [16,18], were identified by structural comparisons with MalK (PDB ID: 1Q12) [54], which superposes 712 C α with an r.m.s.d. of 0.51 Å.
Molecules 2023, 28, x FOR PEER REVIEW 12 of 19 Figure 7A) [50,51]. Likewise, the aromatic triad Trp119/Phe177/Tyr230 in the permease MdxF, which stabilizes the α-and β-CDs through CH-π interactions ( Figure 7B), was identified by structural comparisons with the homologous subunit MalF (PDB ID: 3PUV) [51]. Moreover, the coupling helixes (CH) in MalFG, responsible for the transmembrane domain-nucleotidebinding domain (TMD-NBD) coupling that regulates the conformational changes triggered by ATP hydrolysis [52], were identified in MdxFG ( Figure 7C). Additionally, Glu190 in both subunits of MdxFG was found to make an electrostatic interaction with Arg48 in the active site of NBDs ( Figure 7C), suggesting that the interaction could be correlated to the dimerization of the cytoplasmatic promiscuous ATPase, MsmX, as previously found in MalEFG-MalK [53]. Finally, the conserved LSGGQ motif, Walker A and B, as well as several motifs involved in ATP hydrolysis by MsmX ( Figure 7D) [16,18], were identified by structural comparisons with MalK (PDB ID: 1Q12) [54], which superposes 712 C α with an r.m.s.d. of 0.51 Å. Hence, once α-and β-CDs have entered the PG layer from G+ bacteria, the SBP MdxE, which is anchored to the cytoplasmic membrane outer surface by a lipidic moiety, binds the CD molecule through an open-to-closed conformational change that releases it into the transmembrane channel formed by MdxFG. However, releasing molecules into the cytoplasm requires ATP-mediated conformational changes of the MdxFG-MsmX complex. Therefore, MdxE couples onto the permease subunit MdxFG, triggering a pretrans- The MsmX subunit is shown in gray ribbon representation, as well as signature motifs LSGGQ (red), Walker A (dark green), and D-loop (orange), which along with the Walker B motif, encompass most of the residues (Tyr13, Lys43, Glu160, and His193) needed for ATPase activity and signal transmission between the TMD-NBD. Note that the interaction between Lys43 and the coordinated Mg 2+ atom (gray sphere) is crucial for the cleavage of γ-phosphate from the ATP molecule.
Hence, once αand β-CDs have entered the PG layer from G+ bacteria, the SBP MdxE, which is anchored to the cytoplasmic membrane outer surface by a lipidic moiety, binds the CD molecule through an open-to-closed conformational change that releases it into the transmembrane channel formed by MdxFG. However, releasing molecules into the cytoplasm requires ATP-mediated conformational changes of the MdxFG-MsmX complex. Therefore, MdxE couples onto the permease subunit MdxFG, triggering a pretranslocation state (PTS) to induce the dimerization of the cytoplasmatic ATPase MsmX, obtaining the active form that hydrolyzes an ATP molecule to initiate internalization [52]. This catalytic process triggers a mechanism in which MdxFG-MsmX adopts an outward-facing (OF) conformation where a periplasmic section of this unit will be open, allowing entry of the ligand into the channel through the conserved Gln256 of MdxG, which acts as a spoon to remove it from the sugar-binding site of MdxE ( Figure 7A). The following interaction of CD with MdxFG involves the aromatic triad of the sugar-binding site of MdxF ( Figure 7B), similar to the aromatic triad of MdxE ( Figure 2B). Hence, the CD is trapped in the transmembrane channel before being released into the cytoplasm. Finally, MdxFG-MsmX returns to the initial inward-facing (IF) conformation, releasing the CD molecule into the cell for the next MdxE cycle (Figure 8).
Molecules 2023, 28, x FOR PEER REVIEW 13 of 19 location state (PTS) to induce the dimerization of the cytoplasmatic ATPase MsmX, obtaining the active form that hydrolyzes an ATP molecule to initiate internalization [52]. This catalytic process triggers a mechanism in which MdxFG-MsmX adopts an outwardfacing (OF) conformation where a periplasmic section of this unit will be open, allowing entry of the ligand into the channel through the conserved Gln256 of MdxG, which acts as a spoon to remove it from the sugar-binding site of MdxE ( Figure 7A). The following interaction of CD with MdxFG involves the aromatic triad of the sugar-binding site of MdxF ( Figure 7B), similar to the aromatic triad of MdxE ( Figure 2B). Hence, the CD is trapped in the transmembrane channel before being released into the cytoplasm. Finally, MdxFG-MsmX returns to the initial inward-facing (IF) conformation, releasing the CD molecule into the cell for the next MdxE cycle (Figure 8).

Gene Cloning and Production of Recombinant MdxE
The E. coli codon-optimized N-terminally truncated form of MdxE (NCBI ID: WP_013150585.1) [13] was synthesized by Integrated DNA Technologies (Coralville, IA, USA) and subcloned into a pET-22b(+) expression vector (Novagen, Madison, WI, USA) between the NdeI and NotI restriction sites by Catálisis Biotechnology Company (Morelos, Mexico), resulting in plasmid pMdxE. Restriction analysis and DNA sequencing confirmed the synthesized insert with a sequence coding for six histidines at the C-terminus. The plasmid pMdxE was transformed into E. coli SHuffle T7 competent cells (New England BioLabs) by heat shock. Transformed cells were plated onto Luria-Bertani (LB) agar plates supplemented with 200 µg mL −1 ampicillin at 37 • C. Individual clones of SHuffle T7 harboring pMdxE were cultured in 5 mL LB medium overnight supplemented with 200 µg mL −1 ampicillin at 30 • C and then aliquoted into a sterile solution of 40% (v/v) glycerol and stored at −80 • C. For recombinant MdxE production, an overnight preculture (37 • C, 170 rpm) grown in 250 mL LB medium (200 µg mL −1 ampicillin) was used to inoculate 1 L 2xYT medium (200 µg mL −1 ampicillin) to an initial optical density at 600 nm (OD 600 ) of~0.1. The culture was incubated at 37 • C and 200 rpm until it reached an OD 600 of~0.6 and cooled to 22 • C, and expression was induced by adding a final concentration of 0.1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) to the medium. The induced culture was incubated for 12 h at 22 • C and 170 rpm before harvesting the cells by centrifugation (7500× g, 10 min, 4 • C). The pellet was resuspended in 10 mL of cold lysis buffer A (50 mM sodium phosphate pH 8.0, 500 mM NaCl, 2% (v/v) glycerol, 20 mM imidazole) supplemented with a half mini tablet of EDTA-free complete protease inhibitor cocktail (Roche Molecular Biochemicals) and 1 µg mL −1 DNAse to be disrupted by sonication on ice for 30 min with an amplitude of~29%. The resulting solution was subjected to a heating step for 20 min at 60 • C to precipitate undesirable thermolabile E. coli proteins. Soluble proteins were separated from the insoluble fraction by centrifugation at 19,000× g for 1 h at 4 • C. The supernatant containing thermophilic His6-tagged MdxE was recovered and filtered using a 0.22 µm microfilter (Merck Millipore, Burlington, MA, USA).

Purification of MdxE and SEC-DLS Analysis
Recombinant MdxE was purified by two additional steps, including Ni 2+ -affinity chromatography and SEC-DLS coupled experiments using an ÄKTA Pure 25 M1 FPLC system with UNICORN v1.7.0.1 software (Cytiva, Marlborough, MA, USA) and a Zetasizer µV DLS instrument with OmniSEC v5.12 software (Malvern, UK), respectively. The filtered supernatant was loaded onto a 5 mL Ni 2+ -chelating HisTrap HP column (Cytiva) equilibrated with ten-bed volumes of buffer A. After washing with eight-bed volumes of buffer A to eliminate contaminants, bound MdxE was eluted with a linear gradient of the same buffer A but containing 500 mM imidazole. Elution fractions containing purified MdxE were collected and dialyzed against buffer B (50 mM Tris-HCl pH 7.5, 100 mM NaCl) using an ultrafiltration cell (Amicon Ultracel filter, 30 kDa molecular-weight cutoff, Merck Millipore). SEC-DLS coupled experiments were performed based on the method described previously [13]. A concentrated sample of MdxE at 10 mg mL −1 was filtrated using a 0.22 µm microfilter and resolved on a 120 mL HiLoad 16/600 Superdex 75 pg column (Cytiva) equilibrated with buffer B using a quartz flow cell of 8 µL (Malvern) and bovine serum albumin (BSA, Sigma-Aldrich, St. Louis, MO, USA) as a standard. A major peak corresponding to the MdxE monomer was collected, concentrated, and dialyzed against several volumes of buffer C (10 mM sodium phosphate pH 7.5) using a 30 kDa cutoff ultrafiltration cell for ITC and fluorescence experiments. While SDS-PAGE with Coomassie staining was also used to analyze the purity of MdxE, the protein concentration was determined by the Bradford assay using BSA as a standard [56].

ITC Measurements
Purified MdxE at 4 µM was titrated with 0.8 mM CDs and linear dextrins using a VP-ITC calorimeter (Cytiva) at 55 • C. In all measurements, the sample cell and syringe were filled with protein and ligand solution, respectively, and the reference cell was filled with ultrapure water. The stirring speed of the sample syringe during measurements was 250 rpm. After baseline stabilization and an initial injection of 2 µL, 25 successive injections of 11 µL were carried out with a 5 min equilibration period between injections. Since both protein and ligands were in buffer C, the heat of dilution was neglected in the experiments. The resulting data were analyzed in triplicate (n = 3) using the Origin 7 ITC analysis v7.0 software from Microcal by fitting the titration curves to a single binding site model.

Fluorescence Measurements
The changes in H 0 associated with the open-to-closed conformational changes of productive MdxE-ligand complexes were determined based on the method described previously [47,57], using a Nanodrop ND-3300 fluorospectrometer (Thermo Fisher, Waltham, MA, USA) controlled by ND-3300 v2.8.0 software with ANS as a fluorescence probe at an excitation wavelength of 365 nm and 400-600 nm as the range for emission scan. The fluorescent response was measured in RFU using buffer C for all solutions. Different MdxE concentrations ranging from 2 to 14 µM were tested at fixed ANS concentrations of 20 µM and 350 µM ligand to guarantee saturation of MdxE based on ITC curves. For H 0 measurements, MdxE-ligand complexes were heated at 55 • C for 10 min. Subsequently, ANS was added to form the (MdxE-ligand)-ANS complexes, and the solutions were heated again at 55 • C for 3 min; then, the solutions were cooled to room temperature. Data were analyzed in triplicate (n = 3), and H 0 was estimated as the slope of the linear curve fitted for RFU versus MdxE concentrations plotted at the maximum emission wavelength (λ = 485 nm) using the emtrends() function from emmeans package in R [58,59]. A Tukey's test (α = 0.05) for multiple comparisons was performed to determine significant differences in H 0 between CDs and linear dextrins tested. All fluorescence measurements were made at 22 • C.

Homology Modeling and Docking Simulations
Homology models of MdxE (NCBI ID: WP_013150585.1), MdxF (NCBI ID: WP_012995620.1), MdxG (NCBI ID: WP_012995619.1), and MsmX (NCBI ID: WP_012995743.1) from the sugar ABC importer system, as well as the three-domain ABC CGTase ThmA (NCBI ID: WP_013150586.1) [13], were built on the SWISS-MODEL server [60]. Accordingly, while MdxE models were built using the AcyMBP structure in the closed form (PDB ID: 1URD) [35] and the TvuCMBP structure in the open form (PDB ID: 2ZYO) [36] as templates for calculations, ThmA, MdxF, MdxG, and MsmX models were built using the templates shown in Table S3. The entire model of the MdxEFG-MsmX importer system was assembled manually using Coot [61] upon the entire MalEFG-MalK structure from E. coli (OF, PDB ID: 3PUV) [50]. All the models were subjected to energy minimization using YASARA software v22.9.24 [62] and then validated using MolProbity [63]. Local docking simulations at the sugar-binding site of MdxE models with CDs and linear dextrins were performed using AutoDock Tools 1.5.4 and AutoDock Vina 1.1.0 [64]. Global docking simulations were also performed to visualize the differential binding of the ANS fluorescence probe to the MdxE models in open and closed conformations. All ligand structures were obtained from the PubChem database [65], transformed into 3D structures and minimized using Avogadro v1.95 [66]. The results of MdxE-ligand interactions were selected according to the best ∆G and K d values for subsequent analysis. The molecular docking results were visualized using a BIOVIA discovery studio visualizer [67]. Structure analysis was carried out by manual inspection using UCSF Chimera [68] and Coot [61]. Illustrations were prepared using BIOVIA [67] and UCSF ChimeraX [68].

Conclusions
This work presents structure-function relationship studies of SBP MdxE from the sugar type I ABC importer system MdxEFG-MsmX of the CM-CD pathway from T. mathranii subsp. mathranii. Homology modeling and docking simulations revealed that MdxE possesses the conserved features commonly found in SBPs, such as two structural N-and Cdomains linked by three small hinge regions located at the bottom of the sugar-binding site that allow the transition from an open to a closed conformation upon CD binding. While an aromatic triad Tyr193/Trp269/Trp378 in the C-domain recognizes and stabilizes the ligand through hydrophobic interactions, Phe87 protrudes from the N-domain and assists in stabilizing the CDs by inserting its side chain into the central cavity. ITC measurements showed that, except for G2, MdxE is capable of binding CDs and linear dextrins with high affinity and that the formation of MdxE-ligand complexes takes place in a spontaneous and exothermic interaction for every ligand tested. In contrast, fluorescence measurements and docking simulations showed that despite the binding of CDs and linear dextrins, the open-to-closed conformational change is solely triggered after αand β-CD binding by the concerted action of hinge regions I and II. Structural bioinformatic analysis of the entire CD ABC importer system from T. mathranii subsp. mathranii also revealed the structural keys that appear to be involved in the binding, internalization, and delivery mechanisms of CDs by this nonclassical pathway from Thermoanaerobacterales, which is valuable in the competition for starch availability in extremophilic environments.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/molecules28166080/s1, Figure S1: Gene clusters for alternative carbohydrate internalization pathways involving an ABC transporter and an extracellular glycosyl hydrolase (GH) from T. mathranii subsp. mathranii; Figure S2 Table S3: Template information and statistical validation for homology models from elements involved in CD synthesis and cellular internalization. Data Availability Statement: Data generated in this study are available from the authors upon reasonable request.