Deamidation at N53 Causes SOD1 Structural Instability and Excess Zn Incorporation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Review of Zanderigo et al “Deamidation at N53 causes SOD1 structural instability and excess Zn incorporation"

Amyotrophic Lateral Sclerosis (ALS) is a rare neurodegenerative disease characterized by the progressive loss of motor neurons, leading to muscle weakness and paralysis. Approximately 20% of familial ALS (fALS) cases are caused by mutations in Cu/Zn superoxide dismutase (SOD1). In this manuscript, Zanderigo et al. investigate the impact of Asn53 deamidation (N53D) on the structure and stability of Cu/Zn SOD1. Using iPSC-derived motor neurons, biochemical assays, and molecular dynamics simulations, the authors systematically characterize how this post-translational modification influences SOD1 dimerization, metal incorporation, and aggregation propensity.

The results clearly show that N53 deamidation destabilizes the SOD1 dimer, promotes monomer formation, alters Zn/Cu stoichiometry (notably leading to excess Zn incorporation), and enhances aggregation. This work provides valuable mechanistic insight into how post-translational modifications may contribute to the pathophysiology of sporadic ALS. Overall, the manuscript is well written, methodologically sound, and makes a meaningful contribution to the understanding of SOD1-associated ALS mechanisms.

Comments:

1. Line 232: Correct the unit to 6–8 kDa instead of 6–8 kDA.
2. Lines 452–454: Please expand briefly on the rationale for the Zn/Cu quantification assay. For example, clarify that it aims to test Zn/Cu binding preferences and the order of incorporation to better understand the metal-binding regulation mechanism of SOD1.
3. Lines 635–639: The discussion mentions Zn/Zn or Cu/Cu conditions, but these were not experimentally tested. Please clarify whether there was a specific reason for their exclusion (e.g., feasibility, relevance, or redundancy).
4. Figure 2: Consider adding a native-PAGE gel image to directly visualize the dimer-versus-monomer state. If this data is currently in the Supplementary section, it may be more effective to include it in the main figures for clarity.
5. Figure 6: Please provide a higher-resolution version of the figure to improve visual clarity.
6. Computational modeling: Was the N53D mutation modeled using AlphaFold or a similar structure prediction tool? If so, please comment briefly on whether any predicted structural deviations align with the molecular dynamics results—particularly regarding dimer interface stability.
7. Please ensure consistent italicization of all species names (e.g., E. coli) and Latin expressions (e.g., in vitro, in vivo, in silico) throughout the manuscript.

Recommendation

Minor Revision

This is a well-executed and interesting study that merits publication after minor clarifications and figure improvement. The suggested revisions will primarily enhance clarity and provide a more comprehensive biological context.

Author Response

Comment 1: Line 232: Correct the unit to 6–8 kDa instead of 6–8 kDA.

Response 1: Thank you for the proofreading, this correction has been made.

Comment 2: Lines 452–454: Please expand briefly on the rationale for the Zn/Cu quantification assay. For example, clarify that it aims to test Zn/Cu binding preferences and the order of incorporation to better understand the metal-binding regulation mechanism of SOD1.

Response 2: In this specific part of the manuscript, the rationale for Cu/Zn quantification  and the different addition orders were not expanded on, but only mentioned as a reminder of what our WT control was (Zn-first addition order of WT SOD1). However, the reason for our different approaches to metal cofactor addition were only mentioned briefly in the “Metal addition to apo-SOD1” part of the methods section (previously spanning lines 258-269) and have been now expanded on in the discussion with the following statement beginning at line 604:

“After initial experiments with N53D SOD1 showed an increased abundance of Zn and decreased abundance of Cu when using the in vivo metal addition order (Zn-first), we were interested in further testing metal binding preferences or promiscuity by introducing the metal cofactors in the opposite order as well. This would theoretically help identify if the aberrant incorporation of metals was the result of general promiscuity towards metal cofactors introduced by N53D (i.e. Cu-first addition similarly causes increased Cu, decreased Zn) or if instead N53D introduced a specific change in affinities, namely an increased affinity for Zn and/or decreased affinity for Cu (i.e. Cu-first addition does not cause increased Cu, decreased Zn).”

Comment 3: Lines 635–639: The discussion mentions Zn/Zn or Cu/Cu conditions, but these were not experimentally tested. Please clarify whether there was a specific reason for their exclusion (e.g., feasibility, relevance, or redundancy).

Response 3: The study that discusses the impact of Zn/Zn dual-coordination in SOD1 was referenced in line 639 because the measured metal occupancy of Zn-first N53D SOD1 showed a quantity of coordinated Zn exceeding the molar equivalent number of Zn binding sites in the protein while the measured quantity of coordinated Cu was less than the molar equivalent of Cu binding sites. While we did not intentionally produce Zn/Zn dual-coordinated variants of SOD1 in this study, these measurements indicated that a significant portion of the mutant protein was most likely a Zn/Zn dual-coordinated form having Zn in both metal binding pockets. We believe it is much more likely, as it has been reported previously in fALS mutants, that the copper site has become more promiscuous from the introduced mutation than that the mutation developed a novel, stable Zn binding site in the protein.

Comment 4: Figure 2: Consider adding a native-PAGE gel image to directly visualize the dimer-versus-monomer state. If this data is currently in the Supplementary section, it may be more effective to include it in the main figures for clarity.

Response 4: Native PAGE of dimeric WT SOD1 had previously performed in unpublished experiments and unfortunately even the stable WT dimer yielded a smeared band that provided little useful information in initial experiments. We did not find further experiments using non-denaturing PAGE to be practical for this research and furthermore, we expect Native PAGE of monomeric SOD1 to be uninformative based on the established instability and heterogeneity of SOD1 monomers.

Comment 5: Figure 6: Please provide a higher-resolution version of the figure to improve visual clarity.

Response 5: Figure 6 has now been replaced with a higher quality image.

Comment 6: Computational modeling: Was the N53D mutation modeled using AlphaFold or a similar structure prediction tool? If so, please comment briefly on whether any predicted structural deviations align with the molecular dynamics results—particularly regarding dimer interface stability.

Response 6: In preparation for simulations, the N53D mutation was introduced in PyMOL consistent with our previous work (Wells et al https://doi.org/10.1002/pro.4132). We did not do any modeling of the N53D mutation with AlphaFold or similar structure prediction tool. Such tools have at best a mixed track record predicting the effects of point mutations. Furthermore, the N53 site is largely solvent exposed, making tools like AlphaFold even less reliable.

Comment 7: Please ensure consistent italicization of all species names (e.g., E. coli) and Latin expressions (e.g., in vitro, in vivo, in silico) throughout the manuscript.

Response 7: Thank you, italicization has now been applied where appropriate.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a compelling investigation into asparagine deamidation at SOD1 N53 as a motor neuron–enriched PTM with relevance to ALS. The authors employ a rigorous, multi-faceted approach—combining tandem mass spectrometry for PTM detection in iPSC-derived motor neurons, recombinant expression and biophysical assays of the N53D mimetic, and molecular dynamics simulations—to demonstrate that this modification destabilizes SOD1 dimers, enhances monomeric and aggregated forms, and drives non-native excess Zn²⁺ binding with allosteric impacts on Cu²⁺ coordination.  The work is well written, coherent, and methodologically sound, meriting publication in BioChem following minor revisions

1. Please justify the use of 10 mM Zn and Cu salts (>300× over protein) given that standard SOD1 reconstitutions use less than what you have reported. Also, the concentrations used are orders of magnitude above physiological free-ion levels. That’s fine for driving site occupancy in vitro, but it increases the risk of mis-metalation; consider adding dose–response controls to rule out high-Zn artifacts.
2. Page 9 of 21, Methods — Molecular dynamics: “A triclinic box was used with 1.5 Å spacing between the box and the solute.” → Please verify; this likely should be 1.5 nm (15 Å).
3. How does an inter-dimer PTM change the metal ion binding efficiency? Could you please explain it in your discussion? Your central claim—that N53 deamidation (N53D), distant from the catalytic copper pocket, biases metalation (excess Zn and reduced Cu) and destabilizes SOD1—is plausible, but it lacks mechanistic interpretation. Given the long-range nature of the observed effect, readers need clearer mechanistic linkage between the N53 microenvironment and (i) Cu-site access/affinity, (ii) Zn-site tuning, and (iii) maturation kinetics. Please add a concise paragraph to the Discussion explaining a plausible long-range/allosteric mechanism by which N53 deamidation—though distant from the copper pocket—could bias Cu/Zn metalation and stability

4. Related to point #3, add a simple schematic that (i) maps nearby fALS mutations around N53 and the inter-dimer interface, and (ii) illustrates how N53 deamidation could disrupt local interactions at this interface, with a downstream impact on Cu/Zn metal binding.
5. In Figure 4, at high temperatures, the mutant data split into two clusters (an upper band with strong aggregation—very large DLS radii—and a lower band with weaker aggregation), which does not occur for WT at comparable hydrodynamic radii. Please (i) indicate whether this clustering reflects different purification batches/dates and (ii) clarify statistics—state whether error bars are SD or SEM, include n per condition.
6. Figure 6- The current panel appears to be a low-resolution screenshot (pixelation and blurred text). Please replace with a high-resolution image, so all features are clearly legible.

Author Response

Comment 1: Please justify the use of 10 mM Zn and Cu salts (>300× over protein) given that standard SOD1 reconstitutions use less than what you have reported. Also, the concentrations used are orders of magnitude above physiological free-ion levels. That’s fine for driving site occupancy in vitro, but it increases the risk of mis-metalation; consider adding dose–response controls to rule out high-Zn artifacts.

Response 1: The addition of metal cofactors at 10 mM concentrations was based on a previously published protocol from the PI (A.O.), but the current protocol has since been modified in several stages of the purification/maturation (for example, dialysis into MOPS buffer rather than PBS prior to metal additions to avoid metal phosphate precipitation) so the previous procedure was not directly referenced in the manuscript (See reference included below). While the cofactors are indeed added in high excess, we’ve found both in previous unpublished preparations of recombinant WT SOD1 and in the current work that the WT SOD1 metal binding pockets have sufficient specificity to yield consistent proper metal occupancy (1 Zn and 1 Cu per monomer) when added Zn-first. Having observed that this level of excess does not cause aberrant metals in WT SOD1, we found no reason to modify the procedure.

Benkler, C., O'Neil, A. L., Slepian, S., Qian, F., Weinreb, P. H., & Rubin, L. L. (2018). Aggregated SOD1 causes selective death of cultured human motor neurons. Scientific reports, 8(1), 16393. https://doi.org/10.1038/s41598-018-34759-z

 

Comment 2: Page 9 of 21, Methods — Molecular dynamics: “A triclinic box was used with 1.5 Å spacing between the box and the solute.” → Please verify; this likely should be 1.5 nm (15 Å).

Response 2: We thank the reviewer for noticing the error. Page 9 of the manuscript has been corrected to reflect the actual spacing of 1.5 nm.

Comment 3: How does an inter-dimer PTM change the metal ion binding efficiency? Could you please explain it in your discussion? Your central claim—that N53 deamidation (N53D), distant from the catalytic copper pocket, biases metalation (excess Zn and reduced Cu) and destabilizes SOD1—is plausible, but it lacks mechanistic interpretation. Given the long-range nature of the observed effect, readers need clearer mechanistic linkage between the N53 microenvironment and (i) Cu-site access/affinity, (ii) Zn-site tuning, and (iii) maturation kinetics. Please add a concise paragraph to the Discussion explaining a plausible long-range/allosteric mechanism by which N53 deamidation—though distant from the copper pocket—could bias Cu/Zn metalation and stability

Response 3: While we have discussed internally hypotheses as to why N53D presumably has a more promiscuous Cu binding site, predictions about the specific mechanistic cause of increased promiscuity, increased Zn preference, and/or reduced Cu preference in the Cu pocket were not included in the manuscript because such hypotheses are highly speculative without further molecular dynamics studies. N53 deamidation is not exclusively a dimer interface PTM, nor is it necessarily accurate to assume that its influence on the environment of the Cu pocket is “long range”. The paragraph in the introduction that first mentions the location of N53 (originally lines 99-105) indicates that it lies near the dimer interface, but also that it is near the C57 disulfide bridge and within the zinc binding loop which spans residues 49-82 and includes the Zn ligating residues.  This paragraph, however, does fail to mention the proximity of this site to Cu binding residues, and has now been updated to include the following information: "The residue is also located near the Cu binding residues H46 and H48, as well as the Cu/Zn binding residue H63."

N53 lies between these ligating residues in sequence being only +7 residues away from H48, +5 away from H46, and -10 away from H63. The proper folding of the backbone/formation of the native H-bonding network in this region is necessary for the formation of a properly oriented Cu-specific coordination sphere in the Cu pocket. Being situated between these residues, non-native forces introduced by the negative charge on deamidated N53 could certainly disrupt the conformational landscape in several ways, but accurate predictions worthy of inclusion in publication would require evidence.  

Comment 4: Related to point #3, add a simple schematic that (i) maps nearby fALS mutations around N53 and the inter-dimer interface, and (ii) illustrates how N53 deamidation could disrupt local interactions at this interface, with a downstream impact on Cu/Zn metal binding.

Response 4: It is well established that mutations throughout the SOD1 sequence including wild type-like mutations (i.e. A4V, a wild type-like mutation at the dimer interface, much further in space and sequence from the metal binding sites than N53) can indirectly modulate metal cofactor incorporation, although this occurs via inconsistent pathways through a complex allosteric network. This is mentioned in lines 70-72 in the introduction. An additional reference to a previous study looking at metalated species of mutant recombinant SOD1 has now been added. The stability of SOD1 and structuring of its metal binding pockets are the result of a precariously balanced highly structured network that is globally sensitive to local structural perturbations, hence the diversity of the sites and types of fALS mutations. The diversity of allosteric cascades and resulting conformational changes makes it unfeasible to directly compare and propose a specific mechanism for Cu pocket modulation in N53D SOD1 solely based on established effects of fALS mutations at adjacent sites. Further molecular dynamics research beyond the scope of this paper would be necessary to speak on this topic with confidence.

Comment 5: In Figure 4, at high temperatures, the mutant data split into two clusters (an upper band with strong aggregation—very large DLS radii—and a lower band with weaker aggregation), which does not occur for WT at comparable hydrodynamic radii. Please (i) indicate whether this clustering reflects different purification batches/dates and (ii) clarify statistics—state whether error bars are SD or SEM, include n per condition.

Response 5: (i) In the DLS data, the apparent clustering of high radius Zn-first N53D SOD1 at high temperatures was not coupled to independent growths or different dates in a clear/consistent manner. At 70C the higher radius cluster consisted of scans from two independent growths, which in this response I’ll call “A” and “B”. The lower radius cluster included scans from an experimental replicate of one of these growths (B2) and from two replicates of another third independent growth (C1, C2). At 80C and 90C, the replicates that had been in the high cluster at 70C were instead now grouped with the low radius cluster along with an experimental replicate (A, B1, B2), while the 80C and 90C high radius clusters consisted of scans from two experimental replicates of the third independent growth, previously in the lower cluster at 70C (C1, C2). I feel that it is worth noting here that the average radius of the so called “low radius” cluster in N53D Zn-first data was still higher than the highest radius of WT Zn-first data at both 80C and 90C. In Cu-first N53D samples, however, the higher radius cluster at each temperature (70-90C) was consistently derived from scans from one experimental replicate of one independent growth, while scans from an earlier experimental replicate of this same growth were consistently a part of the lower radius clusters. Although these replicates were performed at separate dates, initial radii at lower temperatures in both replicates reflected radii near the expected range of the dimer, so this was not thought to be a result of pre-formed aggregates during refrigerated storage and the full data set was not preemptively eliminated as an outlier using outlier eliminating criteria detailed in the paper. In the absence of the high radius replicate, N53D Cu-First data groups well with WT Cu-first. Although the question was warranted, we felt that discussing these clustering distinctions did not further the data and would not be useful to readers.

(ii) The heated DLS data was analyzed with a 2-way ANOVA with Dunnet’s multiple comparisons back to the Zn-first WT at each temperature. The error bars are 1 standard deviation from the mean. The n is the number of scans from 3-5 experiments using protein from 2-3 independent purifications. This information has been added to the legend.

Comment 6: Figure 6- The current panel appears to be a low-resolution screenshot (pixelation and blurred text). Please replace with a high-resolution image, so all features are clearly legible.

Response 6: Figure 6 has now been replaced with a higher quality image.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have  addressed a relevant topic in ALS research, specifically focusing on SOD1 post-translational modification (PTM) and its implications for protein stability and aggregation. The study design—combining biochemical characterization and molecular dynamics simulations—is appropriate and potentially impactful. However few suggestions need to be considered before publishing the article online.

1. The study lacks functional validation using  assays that would provide quantitative support for the reported conformational and aggregation differences between WT and N53D SOD1.
2. The manuscript reports aggregation tendencies qualitatively but does not present aggregation kinetics. Measuring and comparing aggregation rates (e.g., using ThT fluorescence or light scattering) would strengthen the mechanistic details

3. The detection of Asn53 deamidation in iPSC-derived motor neurons is intriguing, but the physiological or pathological significance remains underexplored. The authors should discuss whether this modification occurs under stress or aging conditions relevant to ALS pathophysiology and how the observed in vitro effects might translate in vivo.

Comments on the Quality of English Language

The written language seems good, do not need any changes.

Author Response

Comment 1: The study lacks functional validation using assays that would provide quantitative support for the reported conformational and aggregation differences between WT and N53D SOD1.

Response 1: While further experiments probing particular conformational and structural details would be worthwhile, this study utilized several quantitative measures aimed at conformational and aggregation properties including measures of monomer/dimer ratios by SEC, secondary structural elements by circular dichroism, relative metal cofactor incorporation (associated with residue orientation in binding pockets), hydrodynamic radii by dynamic light scattering.

Comment 2: The manuscript reports aggregation tendencies qualitatively but does not present aggregation kinetics. Measuring and comparing aggregation rates (e.g., using ThT fluorescence or light scattering) would strengthen the mechanistic details

Response 2: Future experiments probing aggregation to determine kinetics in finer detail are certainly advisable, however heated SOD1 aggregates are heterogeneous with some species not binding amyloidal aggregate stains such as ThT. Further, we would need to heat our samples to high temperatures to induce the start of aggregation and we do not have a plate reader that gets this hot (>70°C). We believe that the dynamic light scattering and CD data serve as sufficient preliminary evidence of increased aggregation propensity, and we agree that future research centered on aggregation kinetics of this SOD1 variant would be valuable.

Comment 3: The detection of Asn53 deamidation in iPSC-derived motor neurons is intriguing, but the physiological or pathological significance remains underexplored. The authors should discuss whether this modification occurs under stress or aging conditions relevant to ALS pathophysiology and how the observed in vitro effects might translate in vivo.

Response 3: SOD1 has a very long in vivo half-life and therefore the accumulation of non-enzymatic damage like deamidation and oxidation seems likely. In our paper, we discuss a recent (2022) study where the authors used tandem mass spectrometry to measure PTMs on SOD1 purified out of ALS patient spinal cords and healthy controls. In this study, they found SOD1 to be “excessively deamidated” at N53 (>10 fold more than controls). Other measured deamidations at Q15, N26, N131 were also found more often in the ALS patient samples. Stronger language has been added to emphasize the disease relevance of the N53 deamidation both in line 101 of the introduction and in line 695 in the conclusion.