Comparing Hydrogen Deuterium Exchange and Fast Photochemical Oxidation of Proteins: a Structural Characterisation of Wild-Type and ΔN6 β2-Microglobulin
Hydrogen deuterium exchange (HDX) coupled to mass spectrometry (MS) is a well-established technique employed in the field of structural MS to probe the solvent accessibility, dynamics and hydrogen bonding of backbone amides in proteins. By contrast, fast photochemical oxidation of proteins (FPOP) uses hydroxyl radicals, liberated from the photolysis of hydrogen peroxide, to covalently label solvent accessible amino acid side chains on the microsecond-millisecond timescale. Here, we use these two techniques to study the structural and dynamical differences between the protein β2-microglobulin (β2m) and its amyloidogenic truncation variant, ΔN6. We show that HDX and FPOP highlight structural/dynamical differences in regions of the proteins, localised to the region surrounding the N-terminal truncation. Further, we demonstrate that, with carefully optimised LC-MS conditions, FPOP data can probe solvent accessibility at the sub-amino acid level, and that these data can be interpreted meaningfully to gain more detailed understanding of the local environment and orientation of the side chains in protein structures.
KeywordsHDX FPOP Amyloid β2-microglobulin Protein conformation Protein dynamics
The study of protein structure and dynamics is essential to understand the nature of protein folding, function and mis-folding, the latter of which can result in aberrant protein aggregation associated with numerous amyloid diseases [1, 2]. Owing to its sensitivity and versatility, structural mass spectrometry (MS) is a powerful tool in this field and has been used to characterise individual components within heterogeneous mixtures [3, 4, 5, 6].
HDX-MS is limited, however, by several inherent issues. Firstly, the reversible nature of deuterium labelling makes a certain degree of back-exchange after quenching inevitable, thus reducing the dynamic range of observable differences . Similarly, the quench conditions necessary to reduce back-exchange are not suitable for many proteolytic enzymes, limiting HDX-MS experiments to acid proteases (almost exclusively pepsin, although other enzymes can be used) . Secondly, it is not always possible to obtain residue level HDX-MS information due to the requisite use of non-ergodic peptide fragmentation techniques, such as electron transfer dissociation (ETD), and the careful tuning of MS conditions to reduce scrambling of the deuterium label during the analysis [13, 14].
By comparison with HDX-MS, FPOP is a less commonly used technique in which UV laser irradiation is employed to generate •OH radicals from H2O2 in solution (Figure 1b). These radicals react with solvent accessible amino acid side chains, especially large hydrophobic or sulphur-containing residues, resulting in a variety of side chain oxidations, most notably + 16 Da (+O), + 32 Da (+2O) and + 14 Da (CH2 to CO) mass additions [9, 15, 16]. The reaction is quenched with a reactive scavenger (e.g. methionine) before protease digestion and LC-MS/MS analysis is used to separate the oxidised peptides and identify modified residues. Typically, modifications are quantified by integration of the resulting extracted ion chromatograms (XICs) and are reported as a % modified of the total peptide ion count  (Figure 1b). This method offers a probe of solvent accessibility that labels side chains irreversibly, so no back-exchange of the label is possible, and thus lengthy analytical procedures can be employed to gain maximum information from the resulting data.
FPOP has other characteristics that make it a useful complementary technique to HDX. Firstly, quenching the reaction can be achieved at neutral pH and room temperature [17, 18] and hence a wide range of proteases, or combinations of proteases, can be used to digest the protein. Secondly, FPOP labels side chains on a microsecond-millisecond timescale [9, 19, 20], which are faster labelling times than those typically accessible to HDX-MS experimental setups. In principle, the covalent and residue-specific nature of the FPOP label mean that traditional MS/MS ergodic fragmentation techniques, e.g. collision-induced dissociation (CID), can be used to identify the oxidised side chain to residue level resolution [17, 21, 22, 23]. Like HDX, the side chain of the residue in an FPOP experiment affects its propensity to label . However, compared with the wealth of knowledge available concerning the effect of primary amino acid sequence and solvent conditions on deuterium exchange , the effect of nearby side chains and solution conditions on FPOP labelling are not well understood.
Recently, FPOP and HDX have been used in several studies as complementary methods to study epitope mapping and higher order structure [22, 23, 25, 26]. However, given that FPOP primarily labels hydrophobic residues, we sought here to explore the usefulness of FPOP in the study of aggregation-prone proteins, where surface exposure of hydrophobic side chains is often involved in the unwanted self-polymerisation process [2, 27]. HDX-MS has already proved valuable in this field . Consequently, we decided to use the well-characterised protein β2-microglobulin (β2m), and it is more amyloidogenic variant, ΔN6, as a model system with which to compare the utility of HDX and FPOP to reveal differences in the dynamical properties of these proteins that may be linked to their different propensities to form amyloid.
ΔN6 is considered a structural mimic of the amyloidogenic precursor of β2m amyloidosis, containing a trans His-Pro peptide bond at position 32 instead of the native cis isoform [29, 36]. The structural dynamics of wild-type and ΔN6 β2m have been studied previously by NMR relaxation methods [29, 37], HDX-NMR [38, 39], HDX-MS [34, 40, 41, 42] and various other biophysical techniques, but the mechanism by which truncation of the N-terminal hexapeptide enhances amyloidogenicity remains unclear.
Here, we characterise wild-type and ΔN6 β2m by peptide level HDX-MS, presenting our data using a novel visualisation and processing algorithm developed in-house, and show that our results are consistent with previous studies on β2m dynamics. We then compare these data with residue level FPOP labelling for both proteins and show how FPOP complements the HDX data with information on side chain solvent accessibility. Further, we demonstrate that with optimised LC separation and MS settings, FPOP-LC-MS/MS can be used to quantify modifications at the sub-amino acid level, and that these data can be interpreted meaningfully to attain higher resolution information on protein structure and dynamics.
Protein samples were expressed recombinantly as described previously .
FPOP Followed by LC-MS/MS
The FPOP experimental set-up used was as described previously [17, 18]. Immediately prior to laser irradiation, 1 μl 5% v/v H2O2 was added to 100 μl of protein (10 μM) (final H2O2 concentration 0.05% v/v H2O2), containing 20 mM L-histidine in 10 mM potassium phosphate buffer pH 7.4. The capillary outflow following laser irradiation was collected in an Eppendorf tube containing 20 μl quench solution (100 mM L-methionine, 1 μM catalase in 10 mM potassium phosphate buffer pH 7.4) and placed immediately on ice.
The single disulphide bond in irradiated protein samples was reduced (incubation with 10 mM dithiothreitol (DTT) for 1 h, 55 °C at 500 rpm), and the resulting thiols alkylated (incubation in the dark with 55 mM iodoacetamide for 45 min, 20 °C at 500 rpm) and digested with chymotrypsin (1:50 w/w ratio of enzyme to protein, incubated at 37 °C for 18 h at 500 rpm).
The resulting chymotryptic peptides (1 μl at 0.5 μM peptide concentration) were injected onto an ultra performance liquid chromatography (UPLC) M-Class Acquity system equipped with a C18 column (75 μm × 150 mm, Waters Ltd., Wilmslow, Manchester, UK) and separated by gradient elution of 1–50% MeCN (0.1% v/v formic acid) in H2O (0.1% v/v formic acid) over 60 min at 0.3 μl min−1.
Peptides were analysed using a Q Exactive Plus Orbitrap mass spectrometer (ThermoFisher, Bremen, Germany) in data-dependent acquisition mode. The top five most intense ions were selected for high energy HCD fragmentation. The maximum injection time for MS2 acquisition was set to 300 ms, and dynamic exclusion was reduced from 30 to 3 s to allow fragmentation of isobaric FPOP modifications that elute close together.
XICs were generated by extracting the m/z of the base peak of each peptide isotope distribution, for each observed charge state.
HDX Followed by LC-MS/MS
HDX-MS experiments were carried out using an automated HDX robot (LEAP Technologies, Fort Lauderdale, FL, USA) coupled to an M-Class Acquity LC and HDX manager (Waters Ltd., Wilmslow, Manchester, UK). Thirty microlitres of protein solution containing 8 μM of either wild-type or ΔN6 β2m in equilibration buffer (10 mM potassium phosphate buffer pH 7.4) was added to 135 μl deuterated buffer (10 mM potassium phosphate buffer pD 7.4) and incubated at 4 °C for 30, 60, 120, 1800 or 7200 s. Following the labelling reaction, samples were quenched by adding 50 μl of the labelled solution to 100 μl quench buffer (10 mM potassium phosphate, 2 M guanidine-HCl, 200 mM tris(2-carboxyethyl)phosphine pH 2.2) giving a final quench pH ~ 2.5. Fifty microlitres of quenched sample (ca 24 pmol) was passed through an immobilised ethylene-bridged hybrid (BEH) pepsin column (Waters Ltd., Wilmslow, Manchester, UK) at 500 μl min−1 (20 °C) and a VanGuard Pre-column Acquity UPLC BEH C18 (1.7 μm, 2.1 mm × 5 mm, Waters Ltd., Wilmslow, Manchester, UK) for 3 min. The resulting peptic peptides were transferred to a C18 column (75 μm × 150 mm, Waters Ltd., Wilmslow, Manchester, UK) and separated by gradient elution of 0–40% MeCN (0.1% v/v formic acid) in H2O (0.3% v/v formic acid) over 7 min at 40 μl min−1.
The HDX system was interfaced to a Synapt G2Si mass spectrometer (Waters Ltd., Wilmslow, Manchester, UK). HDMSE and dynamic range extension modes (data independent analysis (DIA) coupled with IMS separation) were used to separate peptides prior to CID fragmentation in the transfer cell .
HDX data were analysed using PLGS (v3.0.2) and DynamX (v3.0.0) software supplied with the mass spectrometer. Restrictions for identified peptides in DynamX were as follows: minimum intensity 1000, minimum products per amino acid 0.3, max sequence length 25, max ppm error 5, file threshold 4/5.
Statistical analysis of the combined means and standard deviations was performed using one way ANOVA followed by post hoc Tukey tests .
Results and Discussion
Analysis of β2m and ΔN6 Dynamics Using HDX LC-MS/MS
Peptide level HDX-MS analyses of wild-type and ΔN6 β2m were first carried out to determine differences in deuterium uptake introduced as a result of the truncation of the N-terminal six amino acids in ΔN6 and to enable comparison of the HDX results with those of FPOP. The proteins were dissolved in 10 mM sodium phosphate buffer and HDX carried out for different lengths of time at 4 °C, pH 7.4 (see Methods). Previous results have shown that the proteins undergo complete HDX within that time at higher temperatures [29, 34, 37, 38, 39, 40, 41, 42].
Following HDX, on-line pepsin digestion of both wild-type and ΔN6 β2m yielded a total of 46 peptides that passed the peptide identification restrictions in both proteins (see Methods), covering ~ 95% of the sequence (Supp. Info. Fig. S1). Peptides identified in only one of the two proteins were not used since for these peptides, a direct comparison of the behaviour of the two proteins could not be made. The N-terminal amino acid of each peptide was not included in the analysis due to unpreventable back-exchange, leaving just two internal residues not covered by any peptide: Tyr-26 and Arg-81. Several peptides showed increased deuterium uptake in ΔN6 relative to the wild-type protein which was apparent after only 30 s of deuterium labelling (Figure 2b–d). Peptide 56–60 (Figure 2b) shows deuterium uptake in ΔN6 reaching a plateau at the earliest labelling time-point (30 s), whereas the wild-type protein had yet to reach the equivalent deuterium uptake after 2 h. Similarly, peptide 26–34 arising from ΔN6 (Figure 2c) reaches a plateau after 2 min incubation with deuterium, while the respective wild-type peptide continues to increase in mass for the duration of the labelling experiment.
Multiple overlapping peptides covering these regions were also identified (Supp. Info. Fig. S2). However, the overlapping nature of the peptides complicates analysis, and makes combining and visualising the data to determine the regions of most significant difference challenging. The DynamX software, used here to curate and assign the exchange data, generates butterfly plots which show uptake differences as a function of peptide rather than amino acid position, and mass rather than relative fractional uptake. These properties can make it difficult to visualise the regions of proteins that undergo differential exchange.
To overcome these difficulties in data presentation, an in-house processing algorithm was developed to take the comma-separated value (csv) output files from the DynamX program and, drawing on previously published processing methods, visualise the data [27, 46]. Named PAVED (positional averaging for visualising exchange data), this process uses the deuterium uptake and standard deviation of deuterium uptake per time-point for each peptide covering an amino acid position in order to determine a combined mean and standard deviation for each position (see Methods, Eqs. 2 and 3). Using ANOVA and post hoc Tukey statistical tests, these combined mean uptake values are then compared in order to determine significant differences in uptake between the different proteins. The advantage of this simple approach is that it consistently uses all the data available from all the peptides detected in an experiment to give a readily interpretable visualisation of the uptake differences identified. No curve fitting is required, and this method avoids the complication of a ‘subtraction analysis’ of overlapped peptides . ANOVA and post hoc Tukey tests are used so that multiple states can be compared on the same plot if required. It should be emphasised, however, that this tool is strictly for visualisation purposes only, the calculations outlined above are not meant to be taken as residue level uptake measurements.
Longer HDX incubation times (30 min and 2 h) begin to show increased deuterium uptake in ΔN6 in the E-F loop (Figure 2e) and the C-terminus, with other small increases (< 5% difference in relative fractional uptake) throughout the length of the protein (Figure 3d; Supp. Info. Fig. S3). This is consistent with previous intact HDX-MS data  and the known difference in stability between the two proteins .
Analysis of β2m and ΔN6 Dynamics Using FPOP LC-MS/MS
Having identified several regions that show differences in dynamic behaviour in wild-type and ΔN6 β2m by peptide level HDX, we next sought to compare these data with results derived from the FPOP labelling method, so that structural and dynamical differences between the two proteins could be evaluated directly at the residue level. After independent oxidative labelling of each protein at pH 7.4 with hydroxyl radicals, wild-type and ΔN6 β2m were each subjected to proteolysis with chymotrypsin and the resulting mixtures of peptides in each case were analysed using LC-MS/MS. Seven chymotryptic peptides, covering 88% of the ΔN6 sequence, were found to be present reproducibly in the digests of both proteins (Supp. Info. Fig. S4) and were subsequently used for FPOP quantification.
Residues in Wild-Type and ΔN6 β2m Labelled by Use of FPOP. Residue Types Are Ordered According to Their Reactivities in Hydroxyl Radical Labelling from Highest (Trp) to Lowest Reactivity (Lys). Detailed Assignment Data for All Residues Can Be Found in Supp. Info. Table S1
Residues modified by FPOP
26, 67, 78
22, 30, 70
13, 31, 51, 84
35 (wild-type only)
19, 48, 94
Analysis of FPOP Structural Isomers
It is interesting to note that the vast majority of FPOP oxidations lead to a decrease in retention time compared with the unoxidised peptide, with the exception of the His oxidations which render the peptide more hydrophobic.
Assigning and Interpreting Structural Isomers Detected by FPOP
We next sought to achieve a more in-depth understanding of the changes in side chain orientation and/or dynamics of wild-type and ΔN6 β2m observed by use of HDX and FPOP, and also to probe the extent to which FPOP structural isomers can be interpreted to gain information on side chain positioning in the two proteins. Using RP-UPLC, the retention time order of the three isomers of oxidised phenylalanine has been reported previously as para < meta < ortho, both as free amino acids and when present in a peptide chain [52, 53, 54, 55, 56]. Assigning the same retention time order to the oxidised Phe peaks observed here, it is possible to extract more detailed structural information from the available data (Figure 7). Phe30 was found to label ~ twofold more in ΔN6 than in wild-type β2m at the residue level (Figures 3a and 4). Interestingly, when quantifying the three isomeric peaks of Phe30 separately, the para isomer (shortest retention time) shows a > fourfold difference in labelling between wild-type and ΔN6 β2m, while the ortho position (longest retention time) shows no statistically significant difference (p = 0.067; Figure 7c). The region surrounding Phe30 is well-defined in the NMR structures of both wild-type and ΔN6 β2m , and is of particular importance as this residue is located directly adjacent to Pro32, which undergoes the cis-trans isomerisation which is thought to be an essential step for β2m aggregation into amyloid [29, 36]. The NMR structure of wild-type β2m shows the para position pointing towards the top of the E-strand, and is almost completely secluded from solvent by the E-strand and D-E loop (Figure 7a). By comparison, the NMR structure of ΔN6 indicates that the Phe30 para position is pointing directly out of the bulk protein into the solvent, in place of the truncated N-terminal six amino acids (Figure 7b). Using all 30 states of the available NMR structures , we calculated the average SASA for each isomer position of Phe30 for wild-type and ΔN6 β2m. The average difference in SASA between the two proteins was 1.51 Å2, 6.98 Å2, and 12.16 Å2 for the ortho, meta and para positions of Phe30, respectively. This trend is in good agreement with the FPOP data observed for each of the oxidisable positions (Figure 7c) and supports the assignment made for each isomer. From these FPOP data, we calculated para:meta:ortho ratios of 0.35:1.00:0.76 compared to 1.19:1.00:0.70 for Phe30 for wild-type β2m and ΔN6, respectively. Neither ratio matches the expected isomer ratio for free Phe oxidation in solution (1.5:1:2 or 2.1:1:2.3 depending on solution conditions ). Similarly, the other two oxidised Phe residues identified (Phe22 and Phe70) do not match this expected para:meta:ortho ratio for the free amino acid. As all three of these Phe residues are, at least partially, buried in the NMR structures of both the wild-type protein and ΔN6 (SASA scores are reported in Supp. Info. Table S2), it is reasonable to conclude that the reduced solvent accessibility of the side chain affects the oxidised isomer distribution, as well as the overall % modified of the residue.
The rotation of the Phe30 side chain in ΔN6 may also offer an explanation of the FPOP behaviour of Val27. All the significantly different FPOP labelled side chains in the region surrounding the N-terminal truncation showed FPOP labelling trends which correlated with the changes observed in calculated SASA from the NMR structures. His84 showed increased labelling in ΔN6 (Figures 3a and 4) and an increase in SASA of ~ 28%. Trp60 and His31 both showed decreased labelling in ΔN6, and, on average, had lower SASAs in the ΔN6 NMR structure (~ 38% and ~ 4% decrease respectively). Val27, however, showed ~ threefold lower FPOP labelling in ΔN6 (Figures 3a and 4), despite a ~ 22-fold higher average SASA. A possible explanation for this is the increased competition for hydroxyl radicals caused by the solvent exposure of the nearby highly reactive Phe30 side chain in ΔN6. Although sequence effects of FPOP labelling have not been well characterised, thermal unfolding experiments have shown that poorly reactive residues undergo less FPOP labelling in unfolded proteins due to increased competition from solvent-exposed reactive side chains .
Trp60, which at the residue level showed a small but statistically significant decrease in oxidative labelling in ΔN6 compared with wild-type β2m (Figure 3a), reveals a more complex picture when quantified to the isomer level. The D-E loop region surrounding Trp60 is ill-defined in the NMR structures of both proteins, suggestive of a dynamic nature and making a detailed structural comparison with the FPOP data challenging. The complexity of these data is such that isomer assignments of this residue could not be made with certainty. However, careful examination of the data permits some tentative assignments to be made.
In the case of ΔN6, peaks 7 and 4 (Figures 7d and 8) both have lower average % modified values, and for peak 7 (+ 16 Da) this is greater than a threefold change. Other covalent labelling MS methods have shown burial of Trp60 in other amyloidogenic variants of β2m relative to the wild-type protein . Similarly, partial proteolysis experiments have revealed that Trp60 in ΔN6 is less accessible to proteases . However, three of the four remaining + 16 Da Trp60 modifications show the opposite trend, with significantly more labelling in ΔN6 relative to the wild-type protein (peaks 9, 8 and 5, Figures 7d and 8). Assuming the previous assignments are correct, these remaining + 16 Da peaks would likely correspond to the four isomers of hydroxytryptophan resulting from benzene ring oxidation (Scheme 2). Combined, these data suggest a rearrangement of the side chain, rather than complete burial where, on average, the pyrrole ring becomes less solvent exposed in ΔN6, and the benzene ring more so. The three remaining + 32 Da peaks (peaks 1–3, Figures 7d and 8) are likely isomers of dihydroxytryptophan, resulting from both pyrrole and benzene ring oxidation .
Together, these results illustrate that FPOP is capable of distinguishing differences in solvent accessibility down the sub-amino acid level, and that the labelling trends of such isomers can be interpreted meaningfully to gain higher resolution information on protein structure.
HDX and FPOP are both powerful techniques for the study and characterisation of protein structure and dynamics. While peptide level HDX allows characterisation of changes in backbone dynamics and hydrogen bonding, typically on greater than millisecond timescales, with broad but low resolution coverage of a protein’s sequence, FPOP is capable of comparing the solvent accessibility of individual residues on much shorter (μs-ms) timescales.
The FPOP data acquired here demonstrate how oxidative labelling of a protein, followed by proteolysis and LC-MS/MS analysis, can lead to the identification of structural isomers associated with oxidation of aromatic amino acid residues. Using affirmed orders of elution of these isomers, their relative abundance can be interpreted to help describe the side chain orientation of such hydrophobic residues within a protein structure, as illustrated by the range of Phe30 structural isomers detected for both wild-type β2m and ΔN6. These side chain rearrangements are undetectable by HDX experiments and further illustrate the complementary nature of the two methods.
While the appearance of structural isomers in FPOP complicates the data analysis significantly, it also provides an exciting opportunity to obtain unique information once methods are developed for the routine characterisation of these modifications in the FPOP workflow. The oxidation products of both Phe  and Trp  are distinguishable by differential UV absorbance spectra. UV detection is routinely implemented into LC-MS workflows. Indeed, online UV absorbance coupled with mass spectrometry has been used previously to characterise unwanted Trp oxidation as degradation products of heat stressed antibodies . This methodology could be employed to confirm the tentative assignments made for Trp60 in this study. Although routine analysis of structural isomers by LC may require long gradients to enable high quality LC separation of peptides, long LC gradients are no barrier in FPOP analysis due to the covalent and irreversible nature of the labelling. Additionally, progress by others has already been made in implementing multi-dimensional LC to improve separation and identification of FPOP peptides . That said, additional studies on other well-characterised systems would be necessary to assess the usefulness of interpreting positional isomers in FPOP. Similarly, detailed data interpretation of structural isomers, and indeed residue level FPOP data, would benefit greatly from a deeper understanding of the effect of nearby side chains, the amino acid sequence and solution conditions on FPOP labelling, and the influence these factors have on the relationship between SASA and the degree of modification.
In terms of the amyloidogenic proteins under scrutiny here, protein aggregation is frequently dependent on a partial unfolding event involving surface exposure of hydrophobic residues . These often subtle structural changes can be difficult to probe by other structural biology methods depending on the amplitude and timescale of the motions involved. As a technique particularly sensitive to labelling hydrophobic residues, and doing so in explicit detail, sub-amino acid level FPOP could prove invaluable to the study of protein assembly mechanisms. Moreover, the potential for detailed FPOP mapping of hydrophobic side chain orientations within aggregation-prone regions offers the promise of using FPOP to aid the design and selection of small molecule inhibitors, in addition to other applications such as the analysis protein-protein and protein-ligand interactions in large and dynamic protein assemblies.
We thank Dr. Nick Bond and Dr. David Lowe of Medimmune PLC, as well as members of the Ashcroft, Radford and Sobott groups in Leeds, for helpful discussions. We would also like to thank Waters Ltd., Wilmslow, Manchester, UK, for their donation of the LEAP HDX sample handling robot.
The Biotechnology and Biological Sciences Research Council (BBSRC) and Medimmune PLC are acknowledged for funding OC (BB/M503459/2). The Synapt G2Si mass spectrometer was purchased with a Research Equipment Initiative grant from the BBSRC (BB/E012558/1) and the Orbitrap Q Exactive Plus mass spectrometer was purchased with funds from the Wellcome Trust (208385/Z/17/Z). SER acknowledges the ERC under the European Union’s seventh framework programme (FP7/2007-2013) (grant number 322408) and the Wellcome Trust (WT204963) for funding.
- 1.Sipe, J.D., Benson, M.D., Buxbaum, J.N., Ikeda, S.-i., Merlini, G., Saraiva, M.J.M., Westermark, P.: Amyloid fibril proteins and amyloidosis: chemical identification and clinical classification International Society of Amyloidosis 2016 nomenclature guidelines. Amyloid. 23, 209–213 (2016)CrossRefGoogle Scholar
- 13.Phillips, J.J., Buchanan, A., Andrews, J., Chodorge, M., Sridharan, S., Mitchell, L., Burmeister, N., Kippen, A. D., Vaughan, T. J., Higazi, D. R. & Lowe, D.: Rate of asparagine deamidation in a monoclonal antibody correlating with hydrogen exchange rate at adjacent downstream residues. Anal. Chem. 89(4), 2361–2368 (2017)CrossRefGoogle Scholar
- 21.Rizzo, A.A., Suhanovsky, M.M., Baker, M.L., Fraser, L.T.C.R., Jones, L.M., Rempel, D.L., Gross, M.L., Chiu, W., Alexandrescu, A.T., Teschke, C.M.: Multiple functional roles of the accessory I-domain of bacteriophage P22 coat protein revealed by NMR structure and CryoEM modeling. Structure. 22, 830–841 (2014)CrossRefGoogle Scholar
- 23.Li, J., Wei, H., Krystek, S.R., Bond, D., Brender, T.M., Cohen, D., Feiner, J., Hamacher, N., Harshman, J., Huang, R.Y.C., Julien, S.rH., Lin, Z., Moore, K., Mueller, L., Noriega, C., Sejwal, P., Sheppard, P., Stevens, B., Chen, G., Tymiak, A.A., Gross, M.L., Schneeweis, L.A.: Mapping the energetic epitope of an antibody/interleukin-23 interaction with hydrogen/deuterium exchange, fast photochemical oxidation of proteins mass spectrometry, and alanine shave mutagenesis. Anal. Chem. 89, 2250–2258 (2017)CrossRefGoogle Scholar
- 26.Li, K.S., Chen, G., Mo, J., Huang, R.Y.C., Deyanova, E.G., Beno, B.R., O’neil, S.R., Tymiak, A.A., Gross, M.L..: Orthogonal mass spectrometry-based footprinting for epitope mapping and structural characterization: the IL-6 receptor upon binding of protein therapeutics. Anal. Chem. 89, 7742–7749 (2017)CrossRefGoogle Scholar
- 27.Dobson, C.L., Devine, P.W.A., Phillips, J.J., Higazi, D.R., Lloyd, C., Popovic, B., Arnold, J., Buchanan, A., Lewis, A., Goodman, J., Van Der Walle, C.F., Thornton, P., Vinall, L., Lowne, D., Aagaard, A., Olsson, L.-L., Ridderstad Wollberg, A., Welsh, F., Karamanos, T.K., Pashley, C.L., Iadanza, M.G., Ranson, N.A., Ashcroft, A.E., Kippen, A.D., Vaughan, T.J., Radford, S.E., Lowe, D.C.: Engineering the surface properties of a human monoclonal antibody prevents self-association and rapid clearance in vivo. Sci. Rep. 6, 38644 (2016)Google Scholar
- 28.Androlewicz, M.J., Ortmann, B., van Endert, P.M., Spies, T., Cresswell, P.: Characteristics of peptide and major histocompatibility complex class I/beta 2-microglobulin binding to the transporters associated with antigen processing (TAP1 and TAP2). Proc. Natl. Acad. Sci. U. S. A. 91, 12716–12720 (1994)CrossRefGoogle Scholar
- 34.Esposito, G., Michelutti, R., Verdone, G., Viglino, P., Hernández, H., Robinson, C.V., Amoresano, A., Dal Piaz, F., Monti, M., Pucci, P., Mangione, P., Stoppini, M., Merlini, G., Ferri, G., Bellotti, V.: Removal of the N-terminal hexapeptide from human beta2-microglobulin facilitates protein aggregation and fibril formation. Protein Sci. 9, 831–845 (2000)CrossRefGoogle Scholar
- 53.Hensley, K., Williamson Kelly, S., Maidt Michael, L., Prasad Gabbita, S., Grammas, P., Floyd Robert, A.: Determination of biological oxidative stress using high performance liquid chromatography with electrochemical detection (HPLC-ECD). J. High Resolut. Chromatogr. 22, 429–437 (1999)CrossRefGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.