Table 1.
Fisher’s Kappa statistic test p-values for presence of periodic components in protein sequences.
Figure 1.
Predicted cleavage of tetanus toxin by human cathepsin L and S.
A: Shows the distribution of the distance between successive cleavage probabilities of ≥0.5 for the two cathepsins. λ = expected value (mean) and σ = over dispersion (variance) of the fitted gamma Poisson distribution. B: Cross correlation of cleavage by cathepsin L and cathepsin S cleavage probabilities. A high correlation centered at zero indicates that the two cathepsins have a tendency to cut at the same site within the protein. This is flanked by high correlation of low probability of cleavage at ±5 amino acids of the initial cleavage. The red dashed lines at ±0.04 indicate the 95th percentile confidence limits for non-significant correlation; values outside this band are significant p<0.05.
Figure 2.
Heat diagrams of cross correlation of predicted MHC binding with predicted cathepsin L cleavage in tetanus toxin.
The predicted binding affinity of sequential 9-mers (A: MHC-I) and 15-mers (B: MHC-II) for different human and murine MHC alleles is shown correlated with predicted cathepsin L cleavage sites. As the natural log of MHC binding affinity has been standardized to a zero mean and unit variance by allele within protein, the highest affinity has the lowest numerical value (blue on the thermometer scale). Human cathepsin L cleavage probability ranges from 0–1. The magnitude and sign of the correlation coefficient for each allele is indicated by the thermometer scale. The 95th percentile confidence limits for non-significant correlations is ±0.05. By convention, cleavage is designated as occurring at the P1-P1’ scissile bond; this position is marked. For cathepsin L and S the amino acid at position P2 has a strong tendency to be more hydrophobic than P1. Predicted MHC-I high affinity binding peptides align with their index positions at 10 amino acid positions proximal (toward N-terminus) of the P1P1’and MHC-II at 16 amino acids proximal of P1P1’. The corresponding plot for all 11 proteins is shown in Figure S3.1.
Figure 3.
Parallel plots of cross correlation of predicted MHC binding with cathepsin L cleavage for clusters of alleles in tetanus toxin.
The cross-correlation hierarchies of Figure 2 are shown separated by allele clusters to differentiate their patterns. The blue vertical line marks the P1P1’ cathepsin scissile bond position. The numbering of the X axis reflects amino acid positions proximal of the human cathepsin L cleavage site. The 95 percentile confidence limits differ for each panel, but range from ±0.02–0.05 and are not shown for clarity. Thus the prominent peaks in the graphs are statistically significant but the smaller oscillations of the graphics around zero are not. The corresponding plot for all 11 proteins is shown in Figure S3.2.
Figure 4.
Cross correlation of cathepsin L cleavage probability and B cell epitope probability in tetanus toxin.
Index position zero corresponds to the N-terminal amino acid (P4) of the cleavage site octomer of cathepsin. Hence the scissile bond P1-P1’ occurs at positions 3–4 (solid arrow). The B-cell epitope prediction algorithm evaluates each amino acid in the context of the 4 amino acids each side hence showing the probability that the center amino acid of a 9-mer is a B epitope contact point that will be at index position zero in this graphic. The predictions suggest a strongly negative correlation with cathepsin cleavage to amino acid position running from the predicted cleavage point to -6 (dashed arrow), or that the probability of the peptide whose N terminus is at the position is not favorable for cutting by the peptidase in this region. The dashed lines at ±0.04 indicate the 95th percentile confidence limits for non significant correlation; values outside this band are significant p<0.05.The corresponding plot for all 11 proteins is shown in Figure S3.3.
Figure 5.
Inverse cross correlation of B cell epitope contact positions with N terminal position of predicted MHC binding peptides in tetanus toxin.
Panel A shows (top to bottom) correlation of MHC-I, Class A, Class B, and Murine. Panel B shows correlation of MHC-II, top to bottom DP and DQ, DR, and murine alleles. Each allele is represented by a colored line. The natural log of MHC binding affinity has been standardized to a zero mean and unit variance by allele within the protein and thus the highest affinity has the lowest numerical value. Highest correlation (negative sign is consistent with increased affinity) varies in lag between classes but lies between 3–9 amino acid positions proximal of the N terminus of the MHC binding peptide. The 95 percentile confidence limits are slightly different for each panel, from ±0.03–0.05 and are not shown for clarity. Thus the prominent peaks in the graphs are statistically significant but the smaller oscillations of the graphics around zero are not. The corresponding plot for all 11 proteins is shown in Figure S3.4.
Figure 6.
Cross correlation of the position of MHC-I and MHC-II in tetanus toxin.
An “all against all” cross correlation was conducted for 28 MHC-II HLA against 20 HLA MHC Class I A (Panel A). This was repeated for 17 alleles of Class I B (Panel B). The vertical line indicates the zero lag position (complete correlation of index position). As both the MHC-I and MHC-II affinities are standardized to zero mean and unit variance, a positive number (red) indicates a strong association between the alleles at that position. A negative number (blue) indicates an anticorrelation between the binding affinities of peptides with an N-terminus at that position. The magnitude and sign of the correlation coefficient for each allele can be determined from the thermometer scale beside the heat diagrams. The corresponding paired plots for all 11 proteins is shown in Figure S3.5.
Figure 7.
Conceptual model of an immunologic kernel.
Relationships of the components are shown based on the cross correlations conducted. Two-headed arrows indicate there will be minor positional differences based on the host MHC alleles. Cathepsin cleavage is a requirement at the C-terminal of the MHC peptides; a high frequency of cathepsin cleavage occurs on the proximal side of the B-cell epitope, but no functional requirement for such cleavage has been demonstrated. Each cleavage site comprises an octomer, with the central P1-P1’ scissile bond indicated by the vertical arrow and the octomer amino acids shown as beads. We have characterized a kernel to comprise both B-cell epitope and T-cell epitope components; as shown T-independent and B-independent epitopes comprise subunits of the whole.