Allosteric effects of the coupling cation in melibiose transporter MelB

Reviewer #1 (Public review):

While the structure of the melibiose permease in both outward and inward-facing forms has been solved previously, there remain unanswered questions regarding its mechanism. Hariharan et al set out to address this with further crystallographic studies complemented with ITC and hydrogen-deuterium exchange (HDX) mass spectrometry. They first report 4 different crystal structures of galactose derivatives to explore molecular recognition, showing that the galactose moiety itself is the main source of specificity. Interestingly, they observe a water-mediated hydrogen bonding interaction with the protein and suggest that this water molecule may be important in binding.

The results from the crystallography appear sensible, though the resolution of the data is low, with only the structure with NPG better than 3Å. However, it is a bit difficult to understand what novel information is being brought out here and what is known about the ligands. For instance, are these molecules transported by the protein or do they just bind? They measure the affinity by ITC, but draw very few conclusions about how the affinity correlates with the binding modes. Can the protein transport the trisaccharide raffinose?

The HDX also appears to be well done; however, in the manuscript as written, it is difficult to understand how this relates to the overall mechanism of the protein and the conformational changes that the protein undergoes.

Reviewer #2 (Public review):

This manuscript from Hariharan, Shi, Viner, and Guan presents x-ray crystallographic structures of membrane protein MelB and HDX-MS analysis of ligand-induced dynamics. This work improves on the resolution of previously published structures, introduces further sugar-bound structures, and utilises HDX to explore in further depth the previously observed positive cooperatively to cotransported cation Na+. The work presented here builds on years of previous study and adds substantial new details into how Na+ binding facilitates melibiose binding and deepens the fundamental understanding of the molecular basis underlying the symport mechanism of cation-coupled transporters. However, the presentation of the data lacks clarity, and in particular, the HDX-MS data interpretation requires further explanation in both methodology and discussion.

Comments on Crystallography and biochemical work:

(1) It is not clear what Figure 2 is comparing. The text suggests this figure is a comparison of the lower resolution structure to the structure presented in this work; however, the figure legend does not mention which is which, and both images include a modelled water molecule that was not assigned due to poor resolution previously, as stated by the authors, in the previously generated structure. This figure should be more clearly explained.

(2) It is slightly unclear what the ITC measurements add to this current manuscript. The authors comment that raffinose exhibiting poor binding affinity despite having more sugar units is surprising, but it is not surprising to me. No additional interactions can be mapped to these units on their structure, and while it fits into the substrate binding cavity, the extra bulk of additional sugar units is likely to reduce affinity. In fact, from their listed ITC measurements, this appears to be the trend. Additionally, the D59C mutant utilised here in structural determination is deficient in sodium/cation binding. The reported allostery of sodium-sugar binding will likely influence the sugar binding motif as represented by these structures. This is clearly represented by the authors' own ITC work. The ITC included in this work was carried out on the WT protein in the presence of Na+. The authors could benefit from clarifying how this work fits with the structural work or carrying out ITC with the D59C mutant, or additionally, in the absence of sodium.

Comments on HDX-MS work:

While the use of HDX-MS to deepen the understanding of ligand allostery is an elegant use of the technique, this reviewer advises the authors to refer to the Masson et al. (2019) recommendations for the HDX-MS article (https://doi.org/10.1038/s41592-019-0459-y) on how to best present this data. For example:

(1) The Methodology includes a lipid removal step. Based on other included methods, I assumed that the HDX-MS was being carried out in detergent-solubilised protein samples. I therefore do not see the need for a lipid removal step that is usually included for bilayer reconstituted samples. I note that this methodology is the same as previously used for MelB. It should be clarified why this step was included, if it was in fact used, aka, further details on the sample preparation should be included.

(2) A summary of HDX conditions and results should be given as recommended, including the mean peptide length and average redundancy per state alongside other included information such as reaction temperature, sequence coverage, etc., as prepared for previous publications from the authors, i.e., Hariharan et al., 2024.

(3) Uptake plots per peptide for the HDX-MS data should be included as supporting information outside of the few examples given in Figure 6.

(4) A reference should be given to the hybrid significance testing method utilised. Additionally, as stated by Hageman and Weis (2019) (doi:10.1021/acs.analchem.9b01325), the use of P < 0.05 greatly increases the likelihood of false positive ΔD identifications. While the authors include multiple levels of significance, what they refer to as high and lower significant results, this reviewer understands that working with dynamic transporters can lead to increased data variation; a statement of why certain statistical criteria were chosen should be included, and possibly accompanied by volcano plots. The legend of Figure 6 should include what P value is meant by * and ** rather than statistically significant and highly statistically significant.

(5) Line 316 states a significant difference in seen in dynamics, how is significance measured here? There is no S.D. given in Table S4. Can the authors further comment on the potential involvement in solvent accessibility and buried helices that might influence the overall dynamics outside of their role in sugar vs sodium binding? An expected low rate of exchange suggests that dynamics are likely influenced by solvent accessibility or peptide hydrophobicity? The increased dynamics at peptides covering the Na binding site on overall more dynamic helices suggests that there is no difference between the dynamics of each site.

(6) Previously stated HDX-MS results of MelB (Hariharan et al., 2024) state that the transmembrane helices are less dynamic than polypeptide termini and loops with similar distributions across all transmembrane bundles. The previous data was obtained in the presence of sodium. Does this remove the difference in dynamics in the sugar-binding helices and the cation-binding helices? Including this comparison would support the statement that the sodium-bound MelB is more stable than the Apo state, along with the lack of deprotection observed in the differential analysis.

(7) Have the authors considered carrying out an HDX-MS comparison between the WT and the D59C mutant? This may provide some further information on the WT structure (particularly a comparison with sugar-bound). This could be tied into a nice discussion of their structural data.

(8) Have the authors considered utilising Li+ to infer how cation selectivity impacts the allostery? Do they expect similar stabilisation of a higher-affinity sugar binding state with all cations?

(9) MD of MelB suggests all transmembrane helices are reorientated during substrate translocation, yet substrate and cotransporter ligand binding only significantly impacts a small number of helices. Can the authors comment on the ensemble of states expected from each HDX experiment? The data presented here instead shows overall stabilisation of the transporter. This data can be compared to that of HDX on MFS sugar cation symporter XylE, where substrate binding induces a transition to OF state. There is no discussion of how this HDX data compares to previous MFS sugar transporter HDX. The manuscript could benefit from this comparison rather than a comparison to LacY. It is unlikely that there are universal mechanisms that can be inferred even from these model proteins. Highlighting differences instead between these transport systems provides broader insights into this protein class. Doi: 10.1021/jacs.2c06148 and 10.1038/s41467-018-06704-1.

(10) Additionally, the recent publication of SMFS data (by the authors: doi:10.1016/j.str.2022.11.011) states the following: "In the presence of either melibiose or a coupling Na+-cation, however, MelB increasingly populates the mechanically less stable state which shows a destabilized middle-loop C3." And "In the presence of both substrate and co-substrate, this mechanically less stable state of MelB is predominant.". It would benefit the authors to comment on these data in contrast to the HDX obtained here. Additionally, is the C3 loop covered, and does it show the destabilization suggested by these studies? HDX can provide a plethora of results that are missing from the current analysis on ligand allostery. The authors instead chose to reference CD and thermal denaturation methods as comparisons.

Reviewer #3 (Public review):

Summary:

The melibiose permease from Salmonella enterica serovar Typhimurium (MelBSt) is a member of the Major Facilitator Superfamily (MFS). It catalyzes the symport of a galactopyranoside with Na⁺, H⁺, or Li⁺, and serves as a prototype model system for investigating cation-coupled transport mechanisms. In cation-coupled symporters, a coupling cation typically moves down its electrochemical gradient to drive the uphill transport of a primary substrate; however, the precise role and molecular contribution of the cation in substrate binding and translocation remain unclear. In a prior study, the authors showed that the binding affinity for melibiose is increased in the presence of Na+ by about 8-fold, but the molecular basis for the cooperative mechanism remains unclear. The objective of this study was to better understand the allosteric coupling between the Na+ and melibiose binding sites. To verify the sugar-recognition specific determinants, the authors solved the outward-facing crystal structures of a uniport mutant D59C with four sugar ligands containing different numbers of monosaccharide units (α-NPG, melibiose, raffinose, or α-MG). The structure with α-NPG bound has improved resolution (2.7 Å) compared to a previously published structure and to those with other sugars. These structures show that the specificity is clearly directed toward the galactosyl moiety. However, the increased affinity for α-NPG involves its hydrophobic phenyl group, positioned at 4 Å-distance from the phenyl group of Tyr26 forms a strong stacking interaction. Moreover, a water molecule bound to OH-4 in the structure with α-NPG was proposed to contribute to the sugar recognition and appears on the pathway between the two specificity-determining pockets. Next, the authors analyzed by hydrogen-to-deuterium exchange coupled to mass spectrometry (HDX-MS) the changes in structural dynamics of the transporter induced by melibiose, Na+, or both. The data support the conclusion that the binding of the coupling cation at a remote location stabilizes the sugar-binding residues to switch to a higher-affinity state. Therefore, the coupling cation in this symporter was proposed to be an allosteric activator.

Strengths:

(1) The manuscript is generally well written.

(2) This study builds on the authors' accumulated knowledge of the melibiose permease and integrates structural and HDX-MS analyses to better understand the communication between the sodium ion and sugar binding sites. A high sequence coverage was obtained for the HDX-MS data (86-87%), which is high for a membrane protein.

Weaknesses:

(1) I am not sure that the resolution of the structure (2.7 Å) is sufficiently high to unambiguously establish the presence of a water molecule bound to OH-4 of the α-NPG sugar. In Figure 2, the density for water 1 is not obvious to me, although it is indeed plausible that water mediates the interaction between OH4/OH6 and the residues Q372 and T373.

(2) Site-directed mutagenesis could help strengthen the conclusions of the authors. Would the mutation(s) of Q372 and/or T373 support the water hypothesis by decreasing the affinity for sugars? Mutations of Thr 121, Arg 295, combined with functional and/or HDX-MS analyses, may also help support some of the claims of the authors regarding the allosteric communication between the two substrate-binding sites.

(3) The main conclusion of the authors is that the binding of the coupling cation stabilizes those dynamic sidechains in the sugar-binding pocket, leading to a high-affinity state. This is visible when comparing panels c and a from Figure S5. However, there is both increased protection (blue, near the sugar) and decreased protection in other areas (red). The latter was less commented, could the increased flexibility in these red regions facilitate the transition between inward- and outward-facing conformations? The HDX changes induced by the different ligands were compared to the apo form (see Figure S5). It might be worth it for data presentation to also analyze the deuterium uptake difference by comparing the conditions sodium ion+melibiose vs melibiose alone. It would make the effect of Na+ on the structural dynamics of the melibiose-bound transporter more visible. Similarly, the deuterium uptake difference between sodium ion+melibiose vs sodium ion alone could be analyzed too, in order to plot the effect of melibiose on the Na+-bound transporter.

(4) For non-specialists, it would be beneficial to better introduce and explain the choice of using D59C for the structural analyses.

(5) In Figure 5a, deuterium changes are plotted as a function of peptide ID number. It is hardly informative without making it clearer which regions it corresponds to. Only one peptide is indicated (213-226), I would recommend indicating more of them in areas where deuterium changes are substantial.

(6) From prior work of the authors, melibiose binding also substantially increases the affinity of the sodium ion. Can the authors interpret this observation based on the HDX data?

Author response:

Reviewer #1:

While the structure of the melibiose permease in both outward and inward-facing forms has been solved previously, there remain unanswered questions regarding its mechanism. Hariharan et al set out to address this with further crystallographic studies complemented with ITC and hydrogen-deuterium exchange (HDX) mass spectrometry.

They first report 4 different crystal structures of galactose derivatives to explore molecular recognition, showing that the galactose moiety itself is the main source of specificity. Interestingly, they observe a water-mediated hydrogen bonding interaction with the protein and suggest that this water molecule may be important in binding.

We appreciate the understanding of our work presented in this manuscript by this reviewer.

The results from the crystallography appear sensible, though the resolution of the data is low, with only the structure with NPG better than 3Å. However, it is a bit difficult to understand what novel information is being brought out here and what is known about the ligands. For instance, are these molecules transported by the protein or do they just bind? They measure the affinity by ITC, but draw very few conclusions about how the affinity correlates with the binding modes. Can the protein transport the trisaccharide raffinose?

The four structures with a bound sugar of different sizes aimed to identify the binding motif on both the primary substrate (sugar) and the transporter (MelB_St). Although the resolutions of the structures complexed with melibiose, raffinose, or a-MG are relatively low, the size and shape of the densities at each structure are consistent with the corresponding sugar molecules, which provide valuable data for determining the pose of the bound sugar. Additionally, there is another a-NPG-bound structure at a higher resolution of 2.7 Å. Therefore, our new data support the published binding site with the galactosyl moiety as the main interacting group. The identified water-1 in this study further confirms the orientation of C4-OH. Notably, this transporter does not recognize or transport glucosides where the orientation of C4-OH at the glucopyranosyl ring is opposite. We will provide stronger data to support the water-1.

Regarding the raffinose question, we should have clearly introduced the historical background. Bacterial disaccharide transporters have broad specificity, allowing them to work on a group of sugars with shared structural elements; for example, one sugar molecule can be transported by several transporters. As reported in the literature, the galactosides melibiose, lactose, and raffinose can be transported by both LacY and MelB of E. coli. We did not test whether MelB_St can transport the a-NPG and raffinose. To address this issue and strengthen our conclusions, we plan to conduct additional experiments to gather evidence of the translocation of these sugars by MelB_St.

The HDX also appears to be well done; however, in the manuscript as written, it is difficult to understand how this relates to the overall mechanism of the protein and the conformational changes that the protein undergoes.

Previously, we used HDX-MS to examine the conformational transition between inward- and outward-facing conformations using a conformation-specific nanobody to trap MelB_St in an inward-facing state, as structurally resolved by cryoEM single-particle analysis and published in eLife 2024. That study identified dynamic regions that may be involved in the conformational transitions; however, there was no sugar present. We also solved and published the crystal structure of the apo D59C MelB_St. The sugar-bound and apo states are virtually identical. To address the positive cooperativity of binding between the sugar and co-transport cations observed in biophysical analysis, in this study, we utilize HDX-MS to analyze the structural dynamics induced by melibiose, Na⁺, or both, focusing on the binding residues at the sugar-binding and cation-binding pockets. The results suggest that the coupling cation stabilizes sugar-binding residues at helices I and V, contributing to affinity but not specificity.

Since MelB_St favors the outward-facing conformation, and simulations on the free-energy landscape suggest that the highest affinity of the sugar-bound state is also at an outward-facing state, MelB_St in both the apo and bound states tend to remain in the outward-facing conformation. We will include a section comparing these differences. Thank you to this reviewer for the critical insight.

Reviewer #2:

This manuscript from Hariharan, Shi, Viner, and Guan present x-ray crystallographic structures of membrane protein MelB and HDX-MS analysis of ligand-induced dynamics. This work improves on the resolution of previously published structures, introduces further sugar-bound structures, and utilises HDX to explore in further depth the previously observed positive cooperatively to cotransported cation Na⁺. The work presented here builds on years of previous study and adds substantial new details into how Na⁺ binding facilitates melibiose binding and deepens the fundamental understanding of the molecular basis underlying the symport mechanism of cation-coupled transporters. However, the presentation of the data lacks clarity, and in particular, the HDX-MS data interpretation requires further explanation in both methodology and discussion.

We thank this reviewer for taking the time to read our previous articles related to this manuscript.

Comments on Crystallography and biochemical work:

(1) It is not clear what Figure 2 is comparing. The text suggests this figure is a comparison of the lower resolution structure to the structure presented in this work; however, the figure legend does not mention which is which, and both images include a modelled water molecule that was not assigned due to poor resolution previously, as stated by the authors, in the previously generated structure. This figure should be more clearly explained.

This figure shows a stereo view of a density map created in cross-eye style to demonstrate its quality. We will update this figure with a higher-resolution map, and the density for Wat-1 is clearly visible. This also addresses Reviewer-3’s comment regarding the map resolution.

(2) It is slightly unclear what the ITC measurements add to this current manuscript. The authors comment that raffinose exhibiting poor binding affinity despite having more sugar units is surprising, but it is not surprising to me. No additional interactions can be mapped to these units on their structure, and while it fits into the substrate binding cavity, the extra bulk of additional sugar units is likely to reduce affinity. In fact, from their listed ITC measurements, this appears to be the trend. Additionally, the D59C mutant utilised here in structural determination is deficient in sodium/cation binding. The reported allostery of sodium-sugar binding will likely influence the sugar binding motif as represented by these structures. This is clearly represented by the authors' own ITC work. The ITC included in this work was carried out on the WT protein in the presence of Na⁺. The authors could benefit from clarifying how this work fits with the structural work or carrying out ITC with the D59C mutant, or additionally, in the absence of sodium.

While raffinose and a-MG have been reported as substrates of MelB in E. coli, binding data are unavailable; additionally, for MelB_St, we lack data on the binding of two of the four sugars or sugar analogs. We performed a label-free binding assay using ITC to address this concern with the WT MelB_St. We will also perform the binding assay with the D59C MelB_St, since sugar binding has been structurally analyzed with this mutant, as pointed out by this reviewer. Along with other new functional results, we will prepare a new Figure 1 on functional analysis, which will also address the comment regarding extra bulk at the non-galactosyl moiety with poor affinity.

This D59C uniport mutant exhibits increased thermostability, making it a valuable tool for crystal structure determination, especially since the wild type (WT) is difficult to crystallize at high quality. Asp59 is the only site that responds to the binding of all coupling cations: Na⁺, Li⁺, or H⁺. Notably, this mutant selectively abolishes cation binding and cotransport. However, it still maintains intact sugar binding with slightly higher affinity and preserves the conformational transition, as demonstrated by an electroneutral transport reaction, the melibiose exchange, and fermentation assays with intact cells. Therefore, the structural data derived from this mutant are significant and offer important mechanistic insights into sugar transport. We will provide additional details during the revision.

Comments on HDX-MS work:

While the use of HDX-MS to deepen the understanding of ligand allostery is an elegant use of the technique, this reviewer advises the authors to refer to the Masson et al. (2019) recommendations for the HDX-MS article (https://doi.org/10.1038/s41592-019-0459-y) on how to best present this data. For example:

All authors appreciate this reviewer’s comments and suggestions, which will be incorporated into the revision.

(1) The Methodology includes a lipid removal step. Based on other included methods, I assumed that the HDX-MS was being carried out in detergent-solubilised protein samples. I therefore do not see the need for a lipid removal step that is usually included for bilayer reconstituted samples. I note that this methodology is the same as previously used for MelB. It should be clarified why this step was included, if it was in fact used, aka, further details on the sample preparation should be included.

Yes, a lipid/detergent removal step was applied in this study and in previous studies and this information was clearly described in Methods.

(2) A summary of HDX conditions and results should be given as recommended, including the mean peptide length and average redundancy per state alongside other included information such as reaction temperature, sequence coverage, etc., as prepared for previous publications from the authors, i.e., Hariharan et al., 2024.

We will update the Table S2. Thank you.

(3) Uptake plots per peptide for the HDX-MS data should be included as supporting information outside of the few examples given in Figure 6.

We will prepare the plots in supplementary information.

(4) A reference should be given to the hybrid significance testing method utilised. Additionally, as stated by Hageman and Weis (2019) (doi:10.1021/acs.analchem.9b01325), the use of P < 0.05 greatly increases the likelihood of false positive ΔD identifications. While the authors include multiple levels of significance, what they refer to as high and lower significant results, this reviewer understands that working with dynamic transporters can lead to increased data variation; a statement of why certain statistical criteria were chosen should be included, and possibly accompanied by volcano plots. The legend of Figure 6 should include what P value is meant by * and ** rather than statistically significant and highly statistically significant.

We appreciate this comment and will cite this article on the hybrid significance method. We will include volcano plots for each dataset. We fully acknowledge that using a cutoff of P < 0.05 can increase the likelihood of false-positive identifications. However, given the complexity of the samples analyzed in this study, we believe that some important changes may have been excluded due to higher variability within the dataset. By applying multiple levels of statistical testing, we determined that P < 0.05 represents a suitable threshold for this study. The threshold values were marked in the residual plots and explained in the text. For Figure 6, we have revised it by showing the P value directly.

(5) Line 316 states a significant difference in seen in dynamics, how is significance measured here? There is no S.D. given in Table S4. Can the authors further comment on the potential involvement in solvent accessibility and buried helices that might influence the overall dynamics outside of their role in sugar vs sodium binding? An expected low rate of exchange suggests that dynamics are likely influenced by solvent accessibility or peptide hydrophobicity? The increased dynamics at peptides covering the Na binding site on overall more dynamic helices suggests that there is no difference between the dynamics of each site.

Table S4 was created to provide an overall view of the dynamic regions. If we understand correctly, this reviewer asked us to comment on the effect of solvent accessibility or hydrophobic regions on the overall dynamics outside the binding residues of the peptides that carry binding residues. Since the HDX rate is influenced by two linked factors: solvent accessibility and hydrogen-bonding interactions that reflect structural dynamics, poor solvent accessibility in buried regions results in low deuterium uptakes. The peptides in our dataset that include the Na⁺-binding site showed low HDX, likely due to poor solvent accessibility and structural stability. It is unclear what this reviewer meant by "increased dynamics at peptides covering the Na binding site on overall more dynamic helices." We do not observe increased dynamics in peptides covering Na⁺-binding sites.

(6) Previously stated HDX-MS results of MelB (Hariharan et al., 2024) state that the transmembrane helices are less dynamic than polypeptide termini and loops with similar distributions across all transmembrane bundles. The previous data was obtained in the presence of sodium. Does this remove the difference in dynamics in the sugar-binding helices and the cation-binding helices? Including this comparison would support the statement that the sodium-bound MelB is more stable than the Apo state, along with the lack of deprotection observed in the differential analysis.

Thanks for this suggestion. The previous datasets were collected in the presence of Na⁺. In the current study, we also have a Na-containing dataset. Both showed similar results: the multiple overlapping peptides covering the sugar-binding residues on helices I and V have higher HDX rates than those covering the Na⁺-binding residues, even when Na⁺ is present in both datasets.

(7) Have the authors considered carrying out an HDX-MS comparison between the WT and the D59C mutant? This may provide some further information on the WT structure (particularly a comparison with sugar-bound). This could be tied into a nice discussion of their structural data.

Thanks for this suggestion. Conducting the HDX-MS comparison between the WT and the D59C mutant is certainly interesting, especially given the growing amount of structural and biochemical/biophysical data available for this mutant. However, due to limited resources, we might consider doing it later.

(8) Have the authors considered utilising Li⁺ to infer how cation selectivity impacts the allostery? Do they expect similar stabilisation of a higher-affinity sugar binding state with all cations?

Thanks for this suggestion. We have demonstrated that Li⁺ also shows positive cooperativity with melibiose through ITC binding measurements. Li⁺ binds to MelB_St with higher affinity than Na⁺ but causes many different effects on MelB. It is worth investigating this thoroughly and individually. To address the second question, H⁺ is a poor coupling cation with minimal impact on melibiose binding. Since its pKa is around 6.5, only a small subpopulation of MelB_St is protonated at pH 7.5. The order of sugar-binding cooperativity is the highest with Na⁺, followed by Li⁺ and H⁺.

(9) MD of MelB suggests all transmembrane helices are reorientated during substrate translocation, yet substrate and cotransporter ligand binding only significantly impacts a small number of helices. Can the authors comment on the ensemble of states expected from each HDX experiment? The data presented here instead shows overall stabilisation of the transporter. This data can be compared to that of HDX on MFS sugar cation symporter XylE, where substrate binding induces a transition to OF state. There is no discussion of how this HDX data compares to previous MFS sugar transporter HDX. The manuscript could benefit from this comparison rather than a comparison to LacY. It is unlikely that there are universal mechanisms that can be inferred even from these model proteins. Highlighting differences instead between these transport systems provides broader insights into this protein class. Doi: 10.1021/jacs.2c06148 and 10.1038/s41467-018-06704-1.

The sugar translocation free-energy landscape simulations showed that both helix bundles move relative to the membrane plane. That analysis aimed to clarify a hypothesis in the field—that the MFS transporter can use an asymmetric mode to transition between inward- and outward-facing states. In the case of MelB, we clearly demonstrated that both domains move and each helix bundle moves as a unit, so the labeling changes were identified only in some extramembrane loops and a few highly flexible helices. Thanks for the suggestion about comparing with XylE. We will include a discussion on it.

(10) Additionally, the recent publication of SMFS data (by the authors: doi:10.1016/j.str.2022.11.011) states the following: "In the presence of either melibiose or a coupling Na⁺-cation, however, MelB increasingly populates the mechanically less stable state which shows a destabilized middle-loop C3." And "In the presence of both substrate and co-substrate, this mechanically less stable state of MelB is predominant.". It would benefit the authors to comment on these data in contrast to the HDX obtained here. Additionally, is the C3 loop covered, and does it show the destabilization suggested by these studies? HDX can provide a plethora of results that are missing from the current analysis on ligand allostery. The authors instead chose to reference CD and thermal denaturation methods as comparisons.

Thank this reviewer for reading the single-molecule force spectroscopy (SMFS) study on MelB_St. The C3 loop mentioned in this SMFS article is partially covered in the dataset Mel or Mel plus Na⁺ vs. Apo, and more coverage is in the Na⁺ vs. Apo. In either condition, no deprotection was detected. Two possible reasons the HDX data did not reflect the deprotection are: 1) The changes were too subtle and did not pass the statistical tests and 2) the longest labeling time point was still insufficient to detect the changes; much longer labeling times should be considered in future studies.

Reviewer #3:

Summary:

The melibiose permease from Salmonella enterica serovar Typhimurium (MelB_St) is a member of the Major Facilitator Superfamily (MFS). It catalyzes the symport of a galactopyranoside with Na⁺, H⁺, or Li⁺, and serves as a prototype model system for investigating cation-coupled transport mechanisms. In cation-coupled symporters, a coupling cation typically moves down its electrochemical gradient to drive the uphill transport of a primary substrate; however, the precise role and molecular contribution of the cation in substrate binding and translocation remain unclear. In a prior study, the authors showed that the binding affinity for melibiose is increased in the presence of Na⁺ by about 8-fold, but the molecular basis for the cooperative mechanism remains unclear. The objective of this study was to better understand the allosteric coupling between the Na⁺ and melibiose binding sites. To verify the sugar-recognition specific determinants, the authors solved the outward-facing crystal structures of a uniport mutant D59C with four sugar ligands containing different numbers of monosaccharide units (α-NPG, melibiose, raffinose, or α-MG). The structure with α-NPG bound has improved resolution (2.7 Å) compared to a previously published structure and to those with other sugars. These structures show that the specificity is clearly directed toward the galactosyl moiety. However, the increased affinity for α-NPG involves its hydrophobic phenyl group, positioned at 4 Å-distance from the phenyl group of Tyr26 forms a strong stacking interaction. Moreover, a water molecule bound to OH-4 in the structure with α-NPG was proposed to contribute to the sugar recognition and appears on the pathway between the two specificity-determining pockets. Next, the authors analyzed by hydrogen-to-deuterium exchange coupled to mass spectrometry (HDX-MS) the changes in structural dynamics of the transporter induced by melibiose, Na⁺, or both. The data support the conclusion that the binding of the coupling cation at a remote location stabilizes the sugar-binding residues to switch to a higher-affinity state. Therefore, the coupling cation in this symporter was proposed to be an allosteric activator.

Strengths:

(1) The manuscript is generally well written.

(2) This study builds on the authors' accumulated knowledge of the melibiose permease and integrates structural and HDX-MS analyses to better understand the communication between the sodium ion and sugar binding sites. A high sequence coverage was obtained for the HDX-MS data (86-87%), which is high for a membrane protein.

Thank this reviewer for your positive comments.

Weaknesses:

(1) I am not sure that the resolution of the structure (2.7 Å) is sufficiently high to unambiguously establish the presence of a water molecule bound to OH-4 of the α-NPG sugar. In Figure 2, the density for water 1 is not obvious to me, although it is indeed plausible that water mediates the interaction between OH4/OH6 and the residues Q372 and T373.

Thanks for your comments on the resolution. We will improve the density for the Water 1.

(2) Site-directed mutagenesis could help strengthen the conclusions of the authors. Would the mutation(s) of Q372 and/or T373 support the water hypothesis by decreasing the affinity for sugars? Mutations of Thr 121, Arg 295, combined with functional and/or HDX-MS analyses, may also help support some of the claims of the authors regarding the allosteric communication between the two substrate-binding sites.

The authors thank this reviewer for the thoughtful suggestions. MelB_St has been subjected to Cys-scanning mutagenesis (https://doi.org/10.1016/j.jbc.2021.101090). Placing a Cys residue on the hydrogen bond-donor Q372 significantly decreased the transport initial rate, accumulation, and melibiose fermentation, with little effect on protein expression, as shown in Figure 2 of this JBC paper. Although no binding data are available, the poor initial rate of transport with a similar amount of protein expressed suggested that the binding affinity is apparently decreased, supporting the role of water-1 in the binding pocket for better binding. The T373C mutant retained most activities of the WT. We will discuss the functional characterizations of these two mutants. Thanks.

(3) The main conclusion of the authors is that the binding of the coupling cation stabilizes those dynamic sidechains in the sugar-binding pocket, leading to a high-affinity state. This is visible when comparing panels c and a from Figure S5. However, there is both increased protection (blue, near the sugar) and decreased protection in other areas (red). The latter was less commented, could the increased flexibility in these red regions facilitate the transition between inward- and outward-facing conformations?

Thanks for this important question. We will discuss the deprotected data in the conformational transition between inward-facing and outward-facing states. The two regions, loop8-9 and loop1-2, are located in the gate area on both sides of the membrane and showed increased deuterium uptakes upon binding of melibiose plus Na⁺. They are likely involved in this process.

The HDX changes induced by the different ligands were compared to the apo form (see Figure S5). It might be worth it for data presentation to also analyze the deuterium uptake difference by comparing the conditions sodium ion+melibiose vs melibiose alone. It would make the effect of Na⁺ on the structural dynamics of the melibiose-bound transporter more visible. Similarly, the deuterium uptake difference between sodium ion+melibiose vs sodium ion alone could be analyzed too, in order to plot the effect of melibiose on the Na⁺-bound transporter.

We will analyze the data as suggested by this reviewer.

(4) For non-specialists, it would be beneficial to better introduce and explain the choice of using D59C for the structural analyses.

As response to the reviewer #1 at page 3, “Asp59 is the only site that responds to the binding of all coupling cations: Na⁺, Li⁺, or H⁺. Notably, this mutant selectively abolishes cation binding and cotransport. However, it still maintains intact sugar binding with slightly higher affinity and preserves the conformational transition, as demonstrated by an electroneutral transport reaction, the melibiose exchange, and fermentation assays with intact cells. Therefore, the structural data derived from this mutant are significant and offer important mechanistic insights into sugar transport. We will provide additional details during the revision.”.

(5) In Figure 5a, deuterium changes are plotted as a function of peptide ID number. It is hardly informative without making it clearer which regions it corresponds to. Only one peptide is indicated (213-226), I would recommend indicating more of them in areas where deuterium changes are substantial.

We appreciate this comment, which will make the plots more meaningful. In the previous article published in eLife (2024), we drew boxed to mark the transmembrane regions; however, it generated much confusion, such as why some helices are very short. The revised figure will label the full length of covered positions.

(6) From prior work of the authors, melibiose binding also substantially increases the affinity of the sodium ion. Can the authors interpret this observation based on the HDX data?

This is an intriguing mechanistic question. Based on current data, we believe that the bound melibiose physically prevents the release of Na⁺ or Li⁺ from the cation-binding pocket. The cation-binding pocket and surrounding regions, including the sugar-binding residue Asp124, show low HDX, supporting this idea. Since we lack a structure with both substrates bound, figuring out the details structurally is challenging. However, we have a hypothesis about the intracellular Na⁺ release as proposed in the 2024 JBC paper (https://doi.org/10.1016/j.jbc.2024.107427). After sugar release, the rotamer change of Asp55 will help Na⁺ exit the cation pocket to the sugar pocket, and the negative membrane potential will facilitate the further movement from MelB to the cytosol. We will discuss this during the revision.