Unexpected functional implication of a stable succinimide in the structural stability of Methanocaldococcus jannaschii glutaminase

Protein ageing is often mediated by the formation of succinimide intermediates. These short-lived intermediates derive from asparaginyl deamidation and aspartyl dehydration and are rapidly converted into β-aspartyl or D-aspartyl residues. Here we report the presence of a highly stable succinimide intermediate in the glutaminase subunit of GMP synthetase from the hyperthermophile Methanocaldoccocus jannaschii. By comparing the biophysical properties of the wild-type protein and of several mutants, we show that the presence of succinimide increases the structural stability of the glutaminase subunit. The protein bearing this modification in fact remains folded at 100 °C and in 8 M guanidinium chloride. Mutation of the residue following the reactive asparagine provides insight into the factors that contribute to the hydrolytic stability of the succinimide. Our findings suggest that sequences that stabilize succinimides from hydrolysis may be evolutionarily selected to confer extreme thermal stability.


Originality and interest
The role of the presence of a succinimide in imparting structural stability to a protein has no precedent in the literature. The results highlight a new mechanism for thermostability and are counterintuitive to the well-known mechanism of deamidation, which is one of the potential causes for loss of protein structure at elevated temperatures. Other succinimides that are present in flexible loops of other proteins from thermophilic organisms could also impart rigidity to the proteins by a similar mechanism.
A Data & methodology: validity of approach, quality of data, quality of presentation B Appropriate use of statistics and treatment of uncertainties C Conclusions: robustness, validity, reliability The various methods used are appropriate and the data obtained are sound. Nevertheless, I have a few suggestions to improve the presentation of the manuscript in order to be accessible to non specialists of MS techniques (see major modification). D Suggested improvements: experiments, data for possible revision No experiments needed E References: appropriate credit to previous work? Yes Clarity and context: lucidity of abstract/summary, appropriateness of abstract, introduction and conclusions Major modification One of the main techniques used in this manuscript is mass spectrometry and Nature Communications is a journal of broad scope. Therefore, a figure should explain the different fragmentation patterns of a peptide by MALDI, CID and ETD (p.5 lines 12-13). In particular, as differenciation by ETD-MS/MS of Asp and isoAsp is quite subtile, it is important to include a scheme of the fragmentation to understand how the c+57 and z-57 ions are evidence of the isoAsp amino acid (p.5, 3 lines before the end). Figures 1c and 1d, Figures 2c and 2d: The labels for the z-ions are incorrectly placed under the corresponding fragments. Supplementary Figure 1b Fig 1d, e) p.5 2 last lines: replace c+57 and z-57 by c6+57 and z5-57 and highlight these fragments in Fig.  1d p.7 6 lines before the end: c6+57 and z5-57 fragment ions p.8 6 lines before the end "indicate" instead of "indicates" p.8 3 lines before the end, to avoid redondancy, change to "was followed by tryptophan fluorescence." and delete "revealed that the enzyme...above 4-5M GdmCl." p9 after "in 8M GdmCl." add: "These three mutants possessing no or little succinimide start unfolding above 4-5M GdmCl. In contrast, largely similar..." p.11 line 5: It is not clear that both protein species contain hydrolyzed succinimide. It is better to write : "presence of hydrolyzed succinimide in two species: wild type protein as well as oxidized protein containing two additional oxygen atoms" p.17 Legend  A. Summary of the key results This work has two key aspects: A.1. The "unusual" stability of aspartyl succinimide, e.g., resistance to hydrolysis at high temperature (100 C); A.2. Stabilizing effects of the succinimide on the protein structure, e.g., resistance to denaturation at 6 M guanidinium.
B. Originality and interest: if not novel, please give references B.1. "Unusual" stability of succinimide: as the authors commented, succinimide has been considered labile by many in the field. On the other hand, its stability has also been well documented and recognized. For instance, many bioconjugates contain succinimide thioether resulting from maleimide-thiol reaction. And also noted by the authors, succinimide has been observed in other proteins (see additional examples in Ouellette 2013 and Klaene 2013), albeit not full conversion. Finally, the fact that succinimide in the tryptic peptides was routinely observed in this work clearly demonstrates the relative stability of succinimide. So it is worth addressing some additional questions, such as (1) how much more stable and (2) why.
B.2. The stabilizing effects of the succinimide on the protein structure are noticeable, e.g., by comparing to the mutants that cannot form succinimide. Of course, any mutation is likely to change the structure, perhaps significantly, especially for a well evolved protein. This reviewer is not familiar with the folding of thermophilic proteins, so it would be helpful if the authors could put this into perspective. For example, are the changes in stability observed in this work (WT vs mutants) significantly more than other thermophilic proteins? B.3. The stability of succinimide and protein is coupled to each other. As such, a question this reviewer has is whether the stability of succinimide is due to shielding of water (nucleophile for  hydrolysis) by the protein. This perhaps can be tested by treating the protein with stronger  nucleophiles such as hydroxylamine or hydrazine (see Zhu 2007 and Klaene 2013) under near  neutral pH. B.4. Mechanism and kinetics of succinimide formation. These aspects are not discussed in great details but perhaps are more interesting. For instance, it is noted that re-formation of succinimide from the hydrolysis products were slow in vitro. This raises an interesting possibility that in vivo formation of succinimide may be catalyzed, say be additional enzymes or other factors.
On a related note, the full experimental details for the expression and purification should be included instead of referring to previous papers. This will allow the readers to see how the conditions may contribute to succinimide formation. For example, whether the non-succinimide species were removed during purification.
C. Data & methodology: validity of approach, quality of data, quality of presentation C.1. Overall, the experiments were well designed and data were of high quality.
D. Appropriate use of statistics and treatment of uncertainties D.1. Not applicable. E. Conclusions: robustness, validity, reliability E.1. The findings (data) are robust. As mentioned above, the key is how significant the finding are (e.g., how much more stable and why).
F. Suggested improvements: experiments, data for possible revision F.1. Also provide ESI mass spectra after deconvolution. Easier to see the mass changes. F.2 To provide mechanistic insight and fully assess the unusual factors, it would be helpful to chemically synthesize authentic tryptic peptides and examine their kinetics to form succinimide and the stability of the resulting succinimide. F.3. The authors touched upon minimizing artifacts of asparaginyl deamidation and aspartyl dehydration during sample preparation, which should be explicitly discussed and addressed. Some recent methods to monitor (e.g., 18O labeling, see Du 2012 and Liu 2012) and eliminate (e.g., Glu-C digestion at pH 4, see Liu 2016) such artifacts should be discussed, and if needed, implemented. Again to better asses the potential artifacts, full experimental details for tryptic digestion should be included. F.3. Formation of succinimide can lead to changes in pI, which may be readily detected by isoelectric focusing (IEF). An additional advantage is that IEF is mostly independent of protein conformation. Such data may tease apart the effects of protein structure from the effects of chemical transformation. This paper describes an intriguing finding of a remarkably stable succinimide in an enzyme from a hyperthermophilic archaeon. This is contrary to the belief that succinimide is transiently formed as an intermediate during asparaginyl deamidation or aspartyl dehydration in proteins, and its formation is followed by rapid hydrolysis of this intermediate to aspartyl and isoaspartyl residue. A significant part of the evidence in based on mass spectrometric data presented in Figures 1, 2, 4, and 5, as well as Supplementary Figures 1,2, 3, 4, 6 and 7. As this reviewer is an expert in mass spectrometry, these data drew most of his attention.
The mass spectrometry data please with their abundance, but are not without flaws. These potential flaws are discussed below (not necessarily in the order of significance).
1. First off, the protein spectra (Fig. 1b, e, f; 2 a, b, e, f; etc.) are all taken with different signalto-noise ratios, ranging from ≈5-6 in Fig. 1e to ≈20 in Fig. 2b. This could mean that the spectra were taken at different source conditions (temperature, nozzle-skimmer voltage, source cleanness, etc.). Often, spectra taken at lower S/N show a single peak (e.g., Fig. 1e, 2a), or a dominant peak at a large m/z (Fig. 5c), while those taken at higher S/N show either plurality of peaks (e.g., 1f, 2b and f), or dominant peaks at lower m/z (e.g., 2e, 5b). Could that be an artifact of different source conditions? It is well known in mass spectrometry, that depending upon source conditions, one can obtain protein spectra with different extent of small-molecule losses, such as NH3 and H2O loss, i.e., -17 and -18 Da, respectively. These artificial losses would be indistinguishable in mass from the mass defect due to the succinimide presence.
2. Adding to the above suspicion, the insets are shown with different magnification of the m/z scale, and the latter is never given. The difference in scales (e.g., compare insets in Fig. 2a and 2b) is probably one order of magnitude. Why would so different scales be needed? Even Supplementary Figure 3 shows different and unspecified scales. 3. Contrary to what is customary in protein mass spectrometry, only one charge state is zoomed in, instead of demonstrating the results of neutral mass deconvolution. That is not surprising if the peak ratios change dramatically with the charge state, as would be in the case of gas-phase losses (artifacts), which would be stronger from higher charge states. 4. In general, the mass spectrometric resolution is quite low. Many of today's instruments, including qTOFs, Orbitraps and FT ICR MS can easily resolve isotopic peaks of a 23 kDa protein.
With such isotopic resolution, the difference between the NH3 and H2O losses would be much more clear. Why hasn't been high resolution used in at least most important cases?
1 show that the following residue (D110) was crucial for stabilizing the succinimide (it is formed in the D110G and D100K mutants at acidic pH but rapidly hydrolyzed at neutral pH) as well as the structure of the protein.

Originality and interest
The role of the presence of a succinimide in imparting structural stability to a protein has no precedent in the literature. The results highlight a new mechanism for thermostability and are counterintuitive to the well-known mechanism of deamidation, which is one of the potential causes for loss of protein structure at elevated temperatures. Other succinimides that are present in flexible loops of other proteins from thermophilic organisms could also impart rigidity to the proteins by a similar mechanism.
The various methods used are appropriate and the data obtained are sound. Nevertheless, I have a few suggestions to improve the presentation of the manuscript in order to be accessible to non specialists of MS techniques (see major modification).

One of the main techniques used in this manuscript is mass spectrometry and Nature
Communications is a journal of broad scope. Therefore, a figure

Response to reviewer 2
This referee raises two key questions: (1) as compared to earlier reports, succinimide in MjGATase is how much more stable? (2) Why it is so stable?
These questions are addressed in detail below. # A.1.The "unusual" stability of aspartyl succinimide, e.g., resistance to hydrolysis at high temperature (100 C);

Response
The succinimide in MjGATase is formed by loss of NH 3 from an asparaginyl residue (N109) and not an aspartyl residue. However, we have also observed aspartyl succinimide in In response to the referee's second point as to why succinimide is so stable in MjGATase, we note that the structure of MjGATase appears to play a key role in protecting succinimide from hydrolysis. Rapid hydrolysis of succinimide in the unfolded protein at high pH (Figure 6c), in tryptic peptides (Supplementary Figure 2a) and D110 mutants (Figure 3) strongly supports the role of protein structure in shielding the succinimide from hydrolysis. Further, as suggested by this reviewer we have also examined the potential of the synthetic peptide VYVDKENDLFK to form and retain succinimide. The peptide was found not to form the succinimide. Figure 11). This aspect has been brought out on page 14, lines 343-357.

(Supplementary
In response to the referee's point on the detection of succinimide moiety in the tryptic peptide, it should be noted that it is present only in relatively minor amounts with hydrolysis being more facile in the cleaved product (Supplementary Figure 2a). The line 156 on page 6 has been corrected to emphasize that the succinimide containing peptide in the tryptic digest is present only in small amounts. Further, it should be noted that the trypsin digestion was

Response
Significantly, the absence of succinimidyl moiety in the mutants (MjGATase_N109S, D110G and D110K) results in a T m of 85-90 ο C (Fig. 4a). This relatively high T m suggests that as, in other thermophilic proteins the sequence of Methanocaldococcus jannaschii glutaminase (MjGATase) confers a high degree of thermostability. The spontaneous formation of succinimide in the wild-type enzyme results in a dramatic enhancement of melting temperature, conferring hyperthermostabilty as evident by the absence of melting even at 100 ο C. This aspect is described on pages 9, lines 234-239.
In response to the referee's point on the effect of mutation on the protein stability we would like to highlight that, mutations that perturb the interactions (hydrophobic interactions and salt bridges) in the core of the protein would be expected to alter the structure significantly 1 .
However, in MjGATase the asparaginyl residue (N109) that forms the succinimide is present in a loop that is on the surface of the protein structure. Our studies suggest that the succinimide 9 stabilizes the protein structure possibly; by imparting rigidity through side-chain-backbone cyclization that confines the loop movements (page 18, lines 437-439). This is evident from studies of the mutants (MjGATase_N109S, D110G and D110K) that lack succinimide and exhibit reduced stability (Fig. 4a and b). This inference is further corroborated by the stability of the MjGATase_N109D mutant where a change in the residue side chain from an amide (Asn) to an acid (Asp) has not destabilized the protein as both residues convert to a stable succinimide ( Fig. 4a and b).
We observed that mutations (MjGATase_N109S, D110K and D110G) that do not enable side-chain-backbone cyclization destabilize the enzyme whereas a non-conserved mutation (N109D) that retains this ability is as stable as the WT protein (Fig. 4). Therefore, it is safe to conclude that the difference in the stabilities of mutants, lacking a stable succinimide and those (WT Su MjGATase and N109D Su ) harboring the intermediate is a result of side-chain-backbone cyclization and not due to mutations per se (pages 11-12, lines 284-290).
Thermophilic enzymes exploit a variety of tools to enhance structure stability, with none being universal 1 . This is the first example wherein a stable succinimide enables the enzyme to retain its structure even at 100 °C or in 8 M GdmCl, a feature not reported earlier.  Figure 2a), in unfolded protein (Figure 6c) and D110 mutants (Fig. 3). As suggested by this reviewer, we have examined the stability of the succinimide to NH 2 OH (Supplementary Figure 8d). An appreciable amount of the succinimide modification was present even after incubation for 2 hr at 37° C in a solution containing 2 M NH 2 OH. Figure 8d and page 12, lines 301-308). The appearance of additional species after extensive exposure to NH 2 OH may be assigned to either hydroxymic acid derivative or oxidation of methionine residues present in the protein. However, peptide chain cleavage as reported in earlier studies on peptides and proteins [2][3][4] was not observed in MjGATase, even after long exposure to NH 2 OH. The role of protein structure on the stability of the succinimidyl residue is well brought out by the absence of this modification in the synthetic peptide. It should be noted that hydrolysis of succinimide in MjGATase is achieved only after prolonged exposure (12 hrs at 37 ο C) at high pH (10.5) (Figure 6c). We note that under these conditions the enzyme is also unfolded (Figure 6a and d). The slower rate of succinimide formation from the hydrolyzed product (Asp/isoAsp) in refolded MjGATase could be attributed to the absence of complete reversion to the native fold or presence of high amounts of isoAsP in hydrolyzed product or a combination of both (pages 13-14, lines 334-342 ). The near complete conversion of the hydrolyzed product to succinimide in MjGATase_D110G which was not unfolded by pre-exposure to high pH supports the above inference (Fig.3e). In vitro slower rate of succinimide formation from isoASP as compared to aspartyl or asparaginyl residue has been reported earlier 6 .

(Supplementary
As suggested by the reviewer the full experimental details for the expression and purification of MjGATase has been included in the revised manuscript (methods section; pages 19-20, lines 469-494 ) Briefly, the purification procedure involves incubation of the cell lysate at 70 ο C to remove the thermolabile E. coli proteins. It is unlikely that this step would remove nonsuccinimide proteins as the mutants containing little or no succinimide are fully stable and remain soluble at 70 ο C. After thermal precipitation, the supernatant was subjected to anion exchange chromatography. A salt (NaCl) gradient was used to elute the protein from the column.
Wild type and mutants of MjGATase eluted as a single peak from the column and all fractions under this peak were pooled. Mass spectrum of the protein from this peak, for the wild type and N109S mutant always yielded a single unique mass. The mass spectrum of MjGATase_D110G, which also elutes as a single peak on anion exchange chromatography shows the presence of 2 species; one with succinimide and the other, the hydrolyzed product. Similarly, MjGATase_N109D elutes as a single peak on anion exchange chromatography but shows the presence of two species; a major species with succinimide and a small fraction of native protein.
This clearly indicates that under our purification conditions, the hydrolyzed product or the protein containing the precursor sequence, if present, are not separated. This also supports our conclusion that the wild type protein is only in the form of an extremely stable succinimide.
Further, as purification of MjGATase involved a step of heating and anion-exchange chromatography, an N-terminal (His) 6 -tagged MjGATase was generated and the recombinant protein was directly purified from the cell lysate using Ni-NTA affinity chromatography(methods section; pages 20-21, lines 495-519) pages to rule out artifacts of purification procedure leading to formation or removal of succinimide. Fractions containing MjGATase, when pooled and examined by ESI-MS also showed loss of 17 Da ( Supplementary   Fig.1), suggesting that the purification procedure does not contribute to the formation of succinimide or removal of non-succinimide species. This has been added to the revised draft, page 5, lines 119-126).

Response
Deconvoluted ESI mass spectra are provided in the revised manuscript. 13 #F.3b. Formation of succinimide can lead to changes in pI, which may be readily detected by isoelectric focusing (IEF). An additional advantage is that IEF is mostly independent of protein conformation. Such data may tease apart the effects of protein structure from the effects of chemical transformation.

Response
We agree with the reviewer that formation of succinimide can lead to changes in pI, which may be readily detected by isoelectric focusing (IEF).However, this would be indeed true only if, succinimide was formed from an aspartyl residue in a protein where a unit charge difference may be anticipated between the two forms of protein. In the present case, the succinimide formation takes place at a neutral asparaginyl residue, resulting in no net charge difference between the two forms. The clearest distinction between asparaginyl and succinimidyl form would be evident in their masses. Indeed, our mass spectral studies on intact proteins, tryptic fragments in conjunction with site-directed mutagenesis strongly suggest the transformation of N109 to succinimidyl form. However, as suggested by the reviewer, we performed the isoelectric focusing. IEF followed by SDS-PAGE of MjGATase after preincubation at different pH (7.4 and 10.5) also showed that high pH yields a protein species with a pI distinct from that of the native sample (incubated at pH 7.4) (Supplementary Fig.10).
The lower pI value of the protein sample incubated at pH 10.5 indicates hydrolysis of succinimide to Asp/isoAsp. The IEF results corroborate mass spectrometry and native-PAGE results (Fig.3c, Fig.6c and d).This result has been added to the revised draft (page 13, lines 329-333). Figure 1 Figure 7).

Summary of the key results
This paper describes an intriguing finding of a remarkably stable succinimide in an enzyme from a hyperthermophilic archaeon. This is contrary to the belief that succinimide is transiently formed as an intermediate during asparaginyl deamidation or aspartyl dehydration in proteins, and its formation is followed by rapid hydrolysis of this intermediate to aspartyl and isoaspartyl residue.
A significant part of the evidence in based on mass spectrometric data presented in Figures 1, 2, 4, and 5, as well as Supplementary Figures 1,2, 3, 4, 6 and 7. As this reviewer is an expert in mass spectrometry, these data drew most of his attention. The mass spectrometry data please with their abundance, but are not without flaws. These potential flaws are discussed below (not necessarily in the order of significance).

Response to reviewer 3
#1.First off, the protein spectra (Fig. 1b, e, f; 2 a, b, e, f; etc.) are all taken with different signalto-noise ratios, ranging from ≈5-6 in Fig. 1e to ≈20 in Fig. 2b. This could mean that the spectra were taken at different source conditions (temperature, nozzle-skimmer voltage, source cleanness, etc.). Often, spectra taken at lower S/N show a single peak (e.g., Fig. 1e, 2a), or a dominant peak at a large m/z (Fig. 5c)

Response
Insets to all the figures now have the scale specified in the revised manuscript.

#3.
Contrary to what is customary in protein mass spectrometry, only one charge state is zoomed in, instead of demonstrating the results of neutral mass deconvolution. That is not surprising if the peak ratios change dramatically with the charge state, as would be in the case of gas-phase losses (artifacts), which would be stronger from higher charge states.

Response
In place of the most abundant charge state, result of neutral mass deconvolution has been included in the figures of the revised draft.

Response
We agree with the reviewer that many of the present day Q-TOFs, Orbitraps and especially FTICR yield data with unprecedented resolution. However, we believe our mass spectral data are of sufficient resolution to support our claims. Furthermore, we have validated  Equally remarkable is that this peptide has apparently survived also electrospray ionization to produce an ETD MS/MS spectrum in Fig. 1c. The authors don't comment on this apparent stability of succinimide in a peptide. If it is so stable, why not perform LC-MS/MS analysis and quantify peptide abundances as common in proteomics instead of relying on low-resolution protein mass spectra?

Response
The referee has expressed surprise in the tryptic fragment retaining succinimide under both ESI and MALDI conditions. It should be noted that the succinimidyl derivative in the tryptic peptide is present only in a relatively minor amount (Supplementary Figure 2a), being more susceptible to hydrolysis than in intact protein (Figure 1b) However, the stability of the imide intermediate under mass spectral conditions is not really surprising as succinimide intermediates are stable to both soft ionization, MALDI and ESI. # 6. In all MS/MS spectra, the C-terminal series of fragments (y-and z-) appear to be mislabeled, with a series starting from z2 or y2 ions that have only one amino acid, K.