Escrever um comentário

Review on “Structural basis of substrate recognition and transport in bacterial ACS transporters”

In this study, the authors investigated the functional roles of six E.coli MFS transorters that were annotated as ACS transporters in previous studies. Building on this, using single knockout gene E.coli library, the authors performed spot assays with putative substrates as sole carbon source to determine if absence of a transporter in E.coli could prevent growth. Results from here were used to assay purified transporters through differential scanning fluorimetry and liposome-based transport assays, where membrane-potential driven transport was assayed. The authors determined an X-ray diffraction structure of one transporter, LgoT, that was identified as an L-galactonate transporter. The structure was used to compare its substrate binding site with that of DgoT, a D-galactonate transporter, to rationalize the basis of stereoselectivity. Overall, the major claims are well supported by the data shown in the study, although the study could benefit from more detailed explanations of methodologies used. Here, the caveats of the experimental setup in liposome-based transport assays, as well as an expanded section for the structural investigations would be helpful. My recommendations are as follows:

Major comments:

1. While measuring the apparent K_d of substrate binding, the authors have compared the shifts in Tm to plot a saturation curve (Figure 2 and SI Figure 4). Thermal shifts are well suited for ligands with fast off rates, but DSF has several pitfalls, including changes in the shape of the unfolding profile itself (not only a shift in Tm), selecting the temperature to report K_d at, and the assumption of reversible, equilibrium unfolding, which rarely holds for membrane proteins. The study would benefit from displaying the thermal denaturation profiles of each substrate titration used to generate the averages shown in SI Figure 4, and it would be worthwhile comparing the first derivative plots of the same, as this would help rule out possible artefacts in K_d,app calculations.

2. Valinomycin-based potential gradient establishment was used to energize the transporter in the liposome assay (Figure 3), but a dependence on pH gradient was not tested. One of the previous studies referred to (Dmitrieva et al., 2024, https://doi.org/10.1038/s44318-024-00279-y) showed through SSM-electrophysiology experiments that pH affects transport activity. Extending the assays across a range of pH values with both buffers composed of KCl instead of one made of NaCl, would help in delineating a possible role of pH and/or pH gradient in the transport activity or inhibition thereof. Further, since the data points shown in panels D and E are averages of two replicates, the authors may consider adding additional technical and best biological replicates to increase statistical robustness.

Minor comments:

1. Determining the structure using LCP at 2.2Å is a significant result and I commend the authors for their work. However, the map quality is not shown in the figures. Structural observations used to comment on the specificity will be bolstered if they are accompanied by electron density maps of the residues in the binding pocket, especially as an addendum in Figure 5 or 6.

2. In lines 211-212, I believe the interpretation of the fluorescence readout could be reworded to emphasize more on the contribution of the changing environments of Tryptophans as the protein melts during a temperature ramp. While Tyrosines do give a faint signal when excited with 280nm excitation, a hydrophobic to a hydrophilic environment shift or vice versa of tryptophan residues account for most of the signal in proteins with a 0.1% (1g/L) A280 of about 1.

3. Error bars are missing on bar plots in SI Figure 5 and on individual data points in SI Figure 4. Having this information would help in assessing the degree of confidence in drawing conclusions from these data.

4. The rationale for partial loss-of-function phenotype for D-galactarate in ΔgarP and loss of function for D-glucarate transport in both ΔgudP and ΔgarP is explained in lines 170-178, but it is difficult to interpret . Expanding this would help readers to interpret this result. For example, if possible, having another supplementary figure similar to SI Figure 1 for proposed D-glucarate and D-galactarate uptake and metabolism, and rephrasing the rationale could clarify the conclusion. One of the points to note would be that while both GarP and GudP can transport glucarate, their expression perhaps is under a cryptic operon, which may have led to the observed phenotype in the D-glucarate supplied plate. Have the authors assayed if the loss of growth phenotype is rescued in the presence of overexpressed GudP and/or GarP?

5. While the authors did not see any growth phenotype in L-rhamnose supplemented growth assay (Figure 1B, 5^th column of spots assays; lines 184-185), this does not suffice to conclude lack of L-rhamnose symport activity in RhmT. If the purification of RhmT is possible, showing nanoDSF based thermal stability signatures for rhamnose interactions would help complete the characterization of the full set of ACS transporters the authors set out to study.

6. In the D-glucarate spot assay (Figure 1B), ΔgudP and ΔgarP show little or no growth like the parental strain, whereas ΔdgoT and ΔlgoT grow better than the parental strain. Provided that D-glucose supplemented plate had almost equal growth phenotype, it stands to reason that growth rates are not intrinsic to the strains and expanding the dilution series to 3 fold instead of 10 and for a longer dilution series may help getting some insights into possible roles of DgoT and LgoT in D-glucarate symport, if any.

7. It would be helpful to include the raw melting curves and their first derivatives alongside the bar graphs in SI Figure 5. Even without a T_m shift, changes in the first-derivative profile such as peak narrowing or altered peak height can indicate how pH and lipids stabilize one conformation over another. Having all the profiles will help in better assessing possible albeit subtle effects of each substrate over the transporters’ thermal stabilities. Also, in-surfo, some transporters may require a more extreme condition to elicit a measureable phenotype than what is observed physiologically, as can be seen in previous literature on distant MFS transporters (Li, J., Li, Y., Koide, A. et al., 2024, https://doi.org/10.1038/s41467-024-48759-3) or in some transporters with completely different folds (Wöhlert, Kühlbrandt, Yildiz, 2014, https://doi.org/10.7554/eLife.03579). Would the authors consider assaying thermal stability at pH 5 or lower?

8. The conclusion in lines 403–406 is partly inferred from predicted models, which could be emphasized in the text. The comparison drawn between putative residues involved in forming the substrate binding site requires high-resolution structural information and stating that these comparisons are drawn from predicted models would help in weighing the inferences drawn in the figure 6 (panel C) better.

9. The method paragraph for design of protein constructs (lines 454-457) does not provide enough detail to reproduce the results, and a more detailed explanation of certain aspects, such as positions of affinity tag would clarify the design of the constructs.

10. The authors hypothesize what structural features of the transporters’ ligand binding pockets may be crucial for selectivity. Specifically, it would be exciting to test some of these predictions by mutagenesis, for example the switch from D- to L-galactonate for DgoT and LgoT. This could also be supported by molecular dynamics simulations.

I reviewed this preprint as part of the Transparent and Accountable Peer Review training program. This report was written with the support of my research mentor.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.