To evaluate the utility of the PAVED algorithm, the differences in HDX behaviour of wild-β2m and the ΔN6 truncation variant, being the two well-characterised variants of β2m, were first compared with previously published studies comparing the two
proteins using NMR relaxation methods [78] as well as intact HDX-MS [134].
Despite the initial 100% sequence coverage (Figure 4.10), the removal of the N- terminal residue of each peptide during processing results in the loss of coverage of two internal amides (Tyr 26 and Arg 81), as well as the N-terminal methionine (Figure 4.14b). Figure 4.14c shows the PAVED processed HDX-MS data for wild-type β2m
after 30 seconds of deuteration. As deuterium incorporation is dependent on both the primary sequence and the solvent accessibility and dynamics of each amide, factors which are unique to each protein, these combined relative fractional uptake plots provide a unique protein ‘fingerprint’ of deuterium uptake across the protein sequence, for a given peptide list. Indeed, this ‘fingerprint’ deuterium uptake pattern is remarkably similar for all three β2m variants after 30 seconds of labelling, likely
124
After processing, significant errors in the combined mean relative fractional uptake values are observed across all three variants between residues 35-40 and 65-70 (Figure 4.14c, Figure 4.15). This is due to the positional averaging of peptides with minimal overlap, covering different regions of the protein, which have significantly different relative fractional uptake values, and accurately reflects the uncertainty of the deuterium uptake measurements in these regions (Figure 4.14a and b). The same effect is observed in ΔN6 at the C-terminal region (residues 90-99) where overlapping peptides discussed earlier (peptides 81-99 and 94-99, Figure 4.12) showed differing trends with regards to the deuterium uptake differences relative to the wild-type protein. Here, the positional averaging algorithm has reduced the combined mean relative fractional uptake at the extreme C-terminus to show no statistical significance to the wild-type protein, but significant differences are shown between residues 82- 90 (F-G loop), accurately reflecting the complex overlapping peptide level data for this region (Figure 4.12, Figure 4.15, Figure 4.17).
125
Figure 4.14 Consolidation of HDX-MS data using positional averaging. Relative fractional uptake data taken from DynamX (a) is processed by removing the N-terminal residue from each peptide due to back exchange. Multiple uptake measurements for each residue can then be identified by evaluating the relative fractional uptake value for each peptide, and the residues each peptide covers (b). Finally, Equation 4.3 and Equation 4.4 are used to combine the mean and standard deviations for multiple measurements of the same residue, from the overlapping peptides, to achieve a single uptake measurements and error value, per residue, per time point, per state (c). Data are shown for wild-type β2m for the 30 second time point. Error bars shown in b) are standard deviations associated with
replicate measurements (n=5) and variances between charge states of the same replicate. Error bars shown in c) are combined standard deviations calculated using Equation 4.4.
126
Figure 4.15 Comparing HDX-MS between three β2m variants data after 30 seconds deuteration.
Combined mean relative fractional uptake per residue was calculated using the PAVED algorithm. Shaded regions show combined standard deviation.
Similarly significant differences (p < 0.05, based on combined standard deviation and combined mean relative fractional uptake) between ΔN6 and the wild-type protein are observed in the B-C and D-E loop regions (residues 27-35 and 57-62, respectively), consistent with NMR relaxation experiments [78]. Including significant differences
identified in the F-G loop (residues 82-90), these regions highlight a total of 24 backbone amides with statistically significantly (p < 0.05) increased deuterium uptake in the ΔN6 truncation. This is remarkably close to the 22 ± 1 backbone amides estimated to have decreased protection in this variant from intact HDX-MS data [134].
Although these differences are largely consistent across all time points measured, at later labelling time points (1800 and 7200 seconds), both the ΔN6 truncation variant and the D76N variant show smaller increases in deuterium uptake (2-3%) relative to the wild-type protein, across the length of the β2m sequence (Figure 4.16). This is
likely the result of decreased stability of the native fold in both proteins, and increased propensity of D76N and ΔN6 to populate the unfolded state, relative to the wild-type protein [146, 156], and effectively demonstrates the utility of evaluating each deuteration
127
Figure 4.16 PAVED difference plots of HDX-MS data for wild-type, D76N and ΔN6 β2m. Shaded regions
show combined standard deviation.Wild-type β2m has been set to zero for reference. Incubation time
128
Figure 4.17 Visualising differences in deuterium uptake relative to wild-type β2m. Statistically
significant differences (p < 0.05 based on PAVED processed data) are shown. Increases in deuterium uptake, relative to the wild-type protein, are shown in red. Decreases are shown in blue. Missing residues and the N-terminal truncation are shown in black. For reference, a detailed, annotated structure of b2m can be found in Figure 2.6 in Section 2.2. PDB: 4FXL [157].
129
With a deuterium uptake value for each residue, and statistical analyses determining significant differences, the PAVED processing algorithm was then used to screen data for statistical significance as a robust method of data presentation and visualisation. Residues which show statistically significant differences between states were then coloured according to the magnitude of the difference in relative fractional uptake, and plotted on the β2m structure (Figure 4.17). Structurally, the differences observed
between the wild-type and the ΔN6 truncation variant at the earliest labelling time points, are located proximal to the N-terminal truncation (Figure 4.17), consistent with earlier reports [78].