Author response:
Public Reviews:
Reviewer #1 (Public review):
Summary:
This work examines the binding of several phosphonate compounds to a membrane-bound pyrophosphatase using several different approaches, including crystallography, electron paramagnetic resonance spectroscopy, and functional measurements of ion pumping and pyrophosphatase activity. The work attempts to synthesize these different approaches into a model of inhibition by phosphonates in which the two subunits of the functional dimer interact differently with the phosphonate.
Strengths:
This study integrates a variety of approaches, including structural biology, spectroscopic measurements of protein dynamics, and functional measurements. Overall, data analysis was thoughtful, with careful analysis of the substrate binding sites (for example calculation of POLDOR omit maps).
Weaknesses:
Unfortunately, the protein did not crystallize with the more potent phosphonate inhibitors. Instead, structures were solved with two compounds with weak inhibitory constants >200 micromolar, which limits the molecular insight into compounds that could possibly be developed into small molecule inhibitors. Likewise, the authors choose to focus the spectroscopy experiments on these weaker binders, missing an opportunity to provide insight into the interaction between more potent binders and the protein.
We acknowledge the reviewer concern regarding the choice of weaker inhibitors. We attempted co-crystallization with all available inhibitors, including those with higher potency. However, despite numerous efforts, these potent inhibitors yielded low-resolution crystals, making them unsuitable for detailed structural analysis. Therefore, we chose to focus on the weaker binders, as we were able to obtain high-quality crystal structures for these compounds. This allowed us to perform DEER spectroscopy with the added advantage of accurately analyzing the data against structural models derived from X-ray crystallography. Using these weaker inhibitors enabled a more precise interpretation of the DEER data, thus providing reliable insights into the conformational dynamics and inhibition mechanism. However, as suggested by the reviewer, in the revised version, we will perform DEER analysis on the more potent inhibitors to provide additional insight into their interactions.
In general, the manuscript falls short of providing any major new insight into membrane-bound pyrophosphatases, which are a very well-studied system. Subtle changes in the structures and ensemble distance distributions suggest that the molecular conformations might change a little bit under different conditions, but this isn't a very surprising outcome. It's not clear whether these changes are functionally important, or just part of the normal experimental/protein ensemble variation.
We respectfully disagree with the reviewer. The scale of motions seen in this study correspond to those seen in the full panoply of crystal structures of mPPases. Some proteins undergo very large conformational changes during catalysis – such as the rotary ATPase. This one doesn’t, meaning that the precise motions we describe are likely to be relevant. Conformational changes in the ensemble, whether large or small, represent essential protein motions which underlie key mPPase catalytic function. Our DEER spectroscopy data demonstrate the sensitivity and resolution necessary to monitor these subtle changes in equilibria, even if these are only a few Angstroms. For several of the conditions we investigated by DEER in solution, corresponding x-ray structures have been solved, with the derived distances agreeing well with the DEER distributions. This further validates the biological relevance of the structures, including serial time-resolved ones that indicate asymmetry.
The ZLD-bound crystal structure doesn't predict the DEER distances, and the conformation of Na+ binding site sidechains in the ZLD structure doesn't predict whether sodium currents occur. This might suggest that the ZLD structure captures a conformation that does not recapitulate what is happening in solution/ a membrane.
We agree with the reviewer that the ZLD-bound crystal structure does not predict the DEER distances. However, we believe this discrepancy arises from the effect of the bulkiness of ZLD inhibitor, which prevents the closure of the hydrolytic centre. Additionally, the absence of Na+ at the ion gate in the ZLD-bound structure suggests that Na+ transport does not occur, a conclusion further supported by our electrometric measurements. We agree with the reviewer, that the distances observed in the DEER experiments might represent a potential new conformation in solution, which may not be captured by the static X-ray structure, thereby offering insights into the dynamic nature of the protein under physiological conditions. Finally, the static x-ray structures have not captured the asymmetric conformations that must exist to explain half-of-the-sites reactivity.
Reviewer #2 (Public review):
Summary:
Crystallographic analysis revealed the asymmetric conformation of the dimer in the inhibitor-bound state. Based on this result, which is consistent with previous time-resolved analysis, authors verified the dynamics and distance between spin introduced label by DEER spectroscopy in solution and predicted possible patterns of asymmetric dimer.
Strengths:
Crystal structures with inhibitor bound provide detailed coordination in the binding pocket thus useful information for the PPase field and maybe for drug development.
Weaknesses:
The distance information measured by DEER is advantageous for verifying the dynamics and structure of membrane protein in solution. However, regarding T211 data, which, as the authors themselves stated, lacks measurement precision, it is unclear for readers how confident one can judge the conclusion leading from these data for the cytoplasmic side.
We thank the reviewer for acknowledging the advantageous use of the DEER methodology for identifying dynamic states of membrane proteins in solution. We used two sites in our analysis: S525 (periplasm) and T211 (cytoplasm). As we clearly stated in the original manuscript, S525R1 yielded high-quality DEER data, while T211R1 yielded weak (or no) visual oscillations, leading to broad, though different distributions for the several conditions tested. Our main conclusions are based on the S525R1 data. We included the T211R1 data because, although it does not provide definitive evidence, it is consistent with our proposed model and offers additional insights into biologically relevant conditions. Furthermore, the shifts in the centre of mass (Fig EV8D) of the broad T211R1 distributions show a trend that is consistent with our model; although not proving it, it does not exclude it either. Lastly, these data do indeed confirm an important structural feature of mPPase in solution conditions which is the intrinsically high dynamic state of the loop5-6 where T211 is located, and consistent with our previous (Kellosalo et al., Science, 2012; Li et al., Nat. Commun, 2016; Vidilaseris et al., Sci. Adv., 2019; Strauss et al., EMBO Rep., 2024) and current x-ray crystallography data.
The distance information for the luminal site, which the authors claim is more accurate, does not indicate either the possibility or the basis for why it is the ensemble of two components and not simply a structure with a shorter distance than the crystal structure.
We thank the reviewer for pointing out this possibility and alternative interpretation of our DEER data. In the revised version, we will show that our DEER data are consistent with (and do not exclude) asymmetry and rephrase to be inclusive of other possibilities. Importantly, this additional possibility does not affect the current interpretation of the data in our manuscript.
Reviewer #3 (Public review):
Summary:
Membrane-bound pyrophosphatases (mPPases) are homodimeric proteins that hydrolyze pyrophosphate and pump H+/Na+ across membranes. They are attractive drug targets against protist pathogens. Non-hydrolysable PPi analogue bisphosphonates such as risedronate (RSD) and pamidronate (PMD) serve as primary drugs currently used. Bisphosphonates have a P-C-P bond, with its central carbon can accommodate up to two substituents, allowing a large compound variability. Here the authors solved two TmPPase structures in complex with the bisphosphonates etidronate (ETD) and zoledronate (ZLD) and monitored their conformational ensemble using DEER spectroscopy in solution. These results reveal the inhibition mechanism of these compounds, which is crucial for developing future small molecule inhibitors.
Strengths:
The authors show that seven different bisphosphonates can inhibit TmPPase with IC50 values in the micromolar range. Branched aliphatic and aromatic modifications showed weaker inhibition.
High-resolution structures for TmPPase with ETD (3.2 Å) and ZLD (3.3 Å) are determined. These structures reveal the binding mode and shed light on the inhibition mechanism. The nature of modification on the bisphosphonate alters the conformation of the binding pocket.
The conformational heterogeneity is further investigated using DEER spectroscopy under several conditions.
Weaknesses:
The authors observed asymmetry in the TmPPase-ELD structure above the hydrolytic center. The structural asymmetry arises due to differences in the orientation of ETD within each monomer at the active site. As a result, loop5-6 of the two monomers is oriented differently, resulting in the observed asymmetry. The authors attempt to further establish this asymmetry using DEER spectroscopy experiments. However, the (over)interpretation of these data leads to more confusion than any further understanding. DEER data suggest that the asymmetry observed in the TmPPase-ELD structure in this region might be funneled from the broad conformational space under the crystallization conditions.
See also the response below - We respectfully disagree with the reviewer. The asymmetry was previously established using serial time crystallography (Strauss et al., EMBO Rep, 2024) and biochemical assays (e.g. Malinen et al., Prot. Sci., 2022; Artukka et al., Biochem J, 2018; Luoto et al., PNAS, 2013) and also partially seen in one static structure (Vidilaseris et al., Sci Adv 2019). DEER data only show that the previously proposed asymmetry could also be present within the conformational ensemble in solution conditions. Indeed, our data do not (and cannot) exclude this possibility.
DEER data for position T211R1 at the enzyme entrance reveal a highly flexible conformation of loop5-6 (and do not provide any direct evidence for asymmetry, Figure EV8).
Please see relevant response above. We acknowledge that T211 is indeed situated on a highly dynamic loop, which is important for gating and our DEER data confirm its high flexibility. Given we have not observed oscillations of this site, leading to broad distributions, we have stated in the original manuscript that we will not establish the presence of any asymmetry in solution on the basis of T211, rather relying on the S525 site, for which we have acquired high-quality DEER data, as was also pointed out and have been commented on by all reviewers.
Similarly, data for position S521R1 near the exit channel do not directly support the proposed asymmetry for ETD.
The reviewer appears to suggest that we hold the S525R1 DEER data as direct proof of asymmetry; this is combative on the grounds that to directly prove asymmetry would require time-resolved DEER measurements, far beyond the scope of this work. Rather, we have applied DEER measurements to explore whether asymmetry (observed previously via time-resolved X-ray crystallography) is also present (or indeed a possibility) in solution. We simply state that the DEER data are consistent with asymmetry (i.e., that the mean distance increases in the presence of ETD compared to the apo-state). This is a restrained interpretation of the data.
Despite the high quality of the data, they reveal a very similar distance distribution. The reported changes in distances are very small (+/- 0.3 nm), which can be accommodated by a change of spin label rotamer distribution alone. Further, these spin labels are located on a flexible loop, thereby making it difficult to directly relate any distance changes to the global conformation
We thank the reviewer for recognising the high quality of our DEER data for the S525R1, where visual oscillations in the raw traces, as in our case, reportedly lead to highly accurate and reliable distributions, able to separate (in fortuitous cases) helical movements of only a few Angstroms. The ability of DEER/PELDOR offering near Angstrom resolution was previously demonstrated by the acquisition and solution of high resolution multi-subunit spin-labelled membrane protein structures (Pliotas at al., PNAS, 2012; Pliotas et al., Nat Struct Mol Biol, 2015; Pliotas, Methods Enzymol, 2017) as well as it ability in detecting small (and of similar to mPPase magnitude) conformational changes in different integral membrane proteins systems (Kapsalis et al., Nature Comms, 2019; Kubatova et al., PNAS, 2023; Schmidt et al., JACS, 2024; Lane et al., Structure, 2024; Hett et al., JACS, 2021; Zhao et al., Nature, 2024), occurring under different conditions and/or stimuli in solution and/or lipid environment. The changes here are not very small (e.g. ~ 7 Angstroms between the two mean distance extremes (Ca vs IDP)) for DEER’s proven detection sensitivity, and with all other conditions showing changes between those extremes.
These changes are relatively small, but they are expected for membrane ion pumps. Indeed, none of the mPPase structures show helical movements of greater than a half a turn, and that only in helices 6 and 12. There appear to be larger-scale loop closing motions of the 5-6 loop that includes T211, due to the presence of E217 which binds to one of the Mg2+ ions that coordinate the leaving group phosphate. (This is, inter alia, the reason that this loop is so flexible: it can not order before substrate is bound.) Here we have the resolution to detect such subtle differences by DEER, given there are clear shifts in our time domain data and these are reflected in the mean distances in the distributions. Therefore, our study demonstrates the sensitivity and resolution DEER offers in detecting subtle conformational transitions, key in membrane proteins pathways. To further belabour this point, we do not quantify the DEER data (for instance through parametric fitting) to extract populations of different conformational states and we appreciate that to do so would be highly prone to error; however we do (and can, we feel without overinterpretation) assert that the mean distances shift.
The interpretations listed below are not supported by the data presented:
(1) 'In the presence of Ca2+, the distance distribution shifts towards shorter distances, suggesting that the two monomers come closer at the periplasmic side, and consistent with the predicted distances derived from the TmPPase:Ca structure.' Problem: This is a far-stretched interpretation of a tiny change, which is not reliable for the reasons described in the paragraph above.
While the authors overall agree with the reviewer assessment that ±0.3 nm is a small (not a minor) change, there are literature examples quantifying (or using for quantification) distribution peaks separated by similar Δr. (Kubatova et al., PNAS, 2023; Schmidt et al., JACS, 2024; Hett et al., JACS, 2021; Zhao et al., Nature, 2024). In particular, none of the mPPase structures show helical movements of greater than a half a turn (in helices 6 and 12 in particular). There appear to be larger-scale loop closing motions of the 5-6 loop that includes T211, due to the presence of E217 which binds to one of the Mg2+ ions that coordinate the leaving group phosphate. (This is, inter alia, the reason that this loop is so flexible: it can not order before substrate is bound.)
Importantly, we have fitted Gaussians to the experimental distance distributions of 525R1 output by the Comparative Deer Analyzer 2.0 and observed a change in the distribution width in presence of Ca2+, implying the rotameric freedom of the spin label is restricted. However, the CW-EPR for 525R1 indicate that the rotational correlation time of the spin label is highly consistent between conditions (the spectra are almost identical); this cannot be explained simply by rotameric preference of the spin label (as asserted by the reviewer 3), as there is no (further) immobilisation observed from the CW-EPR of apo-state (Figure EV9) to that in presence of Ca2+. Furthermore, in the absence of conformational changes, it is reasonable to assume (and demonstrable from the CW-EPR data) that the rotamer cloud should not significantly change between conditions. However, Gaussian fits of the two extreme cases yielding the longest (i.e., in presence of IDP) and shortest (in presence of ZTD) mean distances for the 525R1 DEER data indicated significant (i.e., above the noise floor after Tikhonov validation) probability density for the IDP condition at 50 Å (P(r) = 0.18). This occurs at four standard deviations above the mean of the ZTD condition, which by random chance should occur with <0.007% probability. Indeed, one can say that to observe 18% probability density at four standard deviations above the mean by random chance would occur on the order of one in 4 x 10^6.
As in previous response the method can detect changes of such magnitude which are not small, but physiologically relevant and expected for integral membrane proteins, such as mPPases. Indeed, even in equal (or more) complex systems such as heptameric mechanosensitive channel proteins DEER provided sub-Angstrom accuracy, when a spin labelled high resolution XRC structure was solved (Pliotas et al., PNAS, 2012; Pliotas et al., Nat Struct Mol Biol, 2015). Despite this is ideal case where DEER accuracy was experimentally validated another high resolution structural method on modified membrane protein and is not very common it demonstrates the power of the method , especially when strong oscillations are present in the raw DEER data (as here for mPPase 525R1), even when multiple distances are present, Angstrom resolution is achievable in such challenging protein classes.
(2) 'Based on the DEER data on the IDP-bound TmPPase, we observed significant deviations between the experimental and the in silico distances derived from the TmPPase:IDP X-ray structure for both cytoplasmic- (T211R1) and periplasmic-end (S525R1) sites (Figure 4D and Figure EV8D). This deviation could be explained by the dimer adopting an asymmetric conformation under the physiological conditions used for DEER, with one monomer in a closed state and the other in an open state.'
Problem: The authors are trying to establish asymmetry using the DEER data. Unfortunately, no significant difference is observed (between simulation and experiment) for position 525 as the authors claim (Figure 4D bottom panel). The observed difference for position 112 must be accounted for by the flexibility and the data provide no direct evidence for any asymmetry.
Reviewer 3 is wrong in suggesting that we are trying to prove asymmetry through the DEER data. That is a well-known fact in the literature (eg Vidilaseris et al, Sci Adv 2019 where we show (1) that the exit channel inhibitor ATC (i.e., close to 525) binds better in solution to the TmPPase:PPi complex than the TmPPase:PPi2 complex, and (2) that ATC binds in an asymmetric fashion to the TmPPase:IDP2 complex with just one ATC dimer on one of the exit channels. We merely use the DEER data to support this well-established fact.
However, we agree that the DEER data in presence of IDP does not provide direct proof for asymmetry; particularly mutant T211R1 yields in silico distributions too short for measurement by DEER. It is possible that the deviations observed (and particularly likely for T211R1) arise from conformational heterogeneity in solution. We will rephrase this paragraph accordingly: “Owing to the broad nature of the T211R1 (cytoplasmic site) distance distributions, we refrain from interpreting shifts in this data. For the 525R1 (periplasmic site) for which we obtained data of high quality (as also pointed out by both reviewers 2 and 3) we observed deviations between the experimental and the in-silico distances derived from the TmPPase:IDP X-ray structure. While this deviation is less pronounced than for the +ZTD condition, the deviation is consistent with an asymmetric conformation in solution.”
(3) 'Our new structures, together with DEER distance measurements that monitor the conformational ensemble equilibrium of TmPPase in solution, provide further solid experimental evidence of asymmetry in gating and transitional changes upon substrate/inhibitor binding.'
Problem: See above. The DEER data do not support any asymmetry.
We feel that the reviewer comments here are somewhat unfounded. The DEER data (and we will limit discussion only to the 525R1 mutant in this regard) satisfy relevant criteria of the white paper (Schiemann et al., 2021, JACS) from the EPR community (signal-to-noise ratio w.r.t modulation depth of > 20 in all cases; replicates have been performed and will be added into the main-text or supplementary; near quantitative labelling efficiency (evidenced by lack of free spin label signal in the CW-EPR spectra); analysed using the CDA (now Figure EV10, this data we will promote to the main-text) to avoid confirmation bias).
While the DEER data do not prove asymmetry, we do not claim proof of asymmetry in the above sentence. We concede to rephrase the offending sentence above as: “Our new structures, together with DEER distance measurements that monitor the conformational ensemble of TmPPase in solution, do not exclude asymmetry in gating and transitional changes upon substrate/inhibitor binding and are consistent with our proposed model.” We feel that this reframed conjecture of asymmetry is well founded; indeed, comparing the experimental apo-state 525R1 distance distribution with in-silico modelling performed on the hybridised asymmetric structure (i.e., comprised of one monomer bound to Ca2+ and another bound to IDP) yields an overlap coefficient (Islam and Roux, JPC B, 2015) of >0.97. This implies the envelope of the modelled distance distribution is quantitatively inside the envelope of the experimental distance distribution. Thus, the DEER data do not exclude asymmetry (previously observed by time-resolved XRC) in solution. While we appreciate that ideally one would measure time-resolved DEER to directly correlate kinetics of conformational changes within the ensemble to the catalytic cycle of mPPase,(and this is something we aim to do in the future), it is beyond the the scope of this study.
Indeed, half-of-the-sites reactivity has been demonstrated in at least the following papers (Vidilaseris et al, Sci Acv. ,2019, Strauss et al, EMBO Rep. 2024, Malinen et al Prot Sci, 2022, Artukka et al Biochem J, 2018; Luoto et al, PNAS, 2013). Half-of-the sites activity requires asymmetry in the mechanism, and therefore asymmetric motions in the active site (viz 211) and exit channel (viz 525). As mentioned above, we have demonstrated this for other inhibitors (Vidilaseris et al 2019) and as part of a time-resolved experiment (Strauss et al 2024). In fact, given the wealth of evidence showing that the symmetrical crystal structures sample a non- or less-productive conformation of the protein, it would be quixotic to propose the DEER experiments - in solution - do not generate asymmetric conformations. It certainly doesn’t obey Occam’s razor of choosing the simplest possible explanation that covers the data.
(4) Based on these observations, and the DEER data for +IDP, which is consistent with an asymmetric conformation of TmPPase being present in solution, we propose five distinct models of TmPPase (Figure 7).
Problem: Again, the DEER data do not support any asymmetry and the authors may revisit the proposed models.
We respectfully disagree with the reviewer. Please see our detailed response above. However, in the revised version, we will clarify that the proposed models are not solely based on the DEER data but are grounded in both current and previously solved structures, with the DEER data providing additional consistency with these models.
(5) 'In model 2 (Figure 7), one active site is semi-closed, while the other remains open. This is supported by the distance distributions for S525R1 and T211R1 for +Ca/ETD informed by DEER, which agrees with the in silico distance predictions generated by the asymmetric TmPPase:ETD X-ray structure'
Problem: Neither convincing nor supported by the data
We respectfully disagree with the reviewer. However, owing to the conformational heterogeneity of T211R1, in the revised version, we will exclude it in the above sentence, to the effect: Please see our detailed response above.
