The effect of acylation with fatty acids and other modifications on HLA class II:peptide binding and T cell stimulation for three model peptides

Abstract

Immunogenicity is a major concern in drug development as anti-drug antibodies in many cases affect both patient safety and drug efficacy. Another concern is often the limited half-life of drugs, which can be altered by different chemical modifications, like acylation with fatty acids. However, acylation with fatty acids has been shown in some cases to modulate T cell activation. Therefore, to understand the role of acylation with fatty acids on immunogenicity we tested three immunogenic non-acylated peptides and 14 of their acylated analogues for binding to 26 common HLA class II alleles, and their ability to activate T cells in an ex vivo T cell assay. Changes in binding affinity associated with acylation with fatty acids were typically modest, though a significant decrease was observed for influenza HA acylated with a stearic acid, and affinities for DQ alleles were consistently increased. Importantly, we showed that for all three immunogenic peptides acylation with fatty acids decreased their capacity to activate T cells, a trend particularly evident with longer fatty acids typically positioned within the peptide HLA class II binding core region, or when closer to the C-terminus. With these results we have demonstrated that acylation with fatty acids of immunogenic peptides can lower their stimulatory capacity, which could be important knowledge for drug design and immunogenicity mitigation.

Fig 1. Influenza HA, exendin-4 and vatreptacog alfa peptides binding to HLA class II alleles.

Experimental measured binding affinities of (A) Influenza HA, (B) exendin-4 and (C) vatreptacog alfa peptides to 26 alleles. The dotted line indicates IC50 of 1000 nM. The solid black dots indicate alleles that the peptide bind and the open dots indicate non-binding alleles. Comparison of measured and predicted binding data on (D) 13-mer and 15-mer influenza HA epitopes, and (E) exendin-4 and (F) vatreptacog alfa. Affinities are expressed in terms of IC50 nM. The association between the datasets was calculated using a non-parametric two-tailed Spearman rank correlation.

Fig 2. Assessment of binding affinities of acylated and non-acylated influenza HA.

(A) Measured binding affinities of influenza HA and acylated analogues of influenza HA. Solid lines show the geometric mean. P-values have been calculated using Wilcoxon t-test. (B) Show changes in binding affinities of non-acylated influenza HA/ acylated influenza HA for each modification. The dotted line indicates 1, which means no change in binding affinities. Numbers above one indicate increased binding as a result of modification, whereas numbers below one indicate decreased binding of the acylated peptides. DQ alleles have been marked with grey. The mean values shown are the geometric mean.

Fig 3. Assessment of binding affinities of acylated and non-acylated exendin-4 and vatreptacog alfa.

(A) Measured binding affinities of non-acylated and acylated exendin-4 and vatreptacog alfa 15-mer peptides. Solid dots indicate nonacylated binding affinities and empty dots indicate acylated analogues. (B) Changes in binding affinities plotted as a ratio for 26 alleles. Grey dots indicate DQ alleles. The mean values shown are the geometrical mean.

Fig 4. HLA-DRB1 frequency in study population vs. Europeans population.

Comparison of the frequency of HLA-DRB1 allotypes expressed in the study cohort versus the European population. The association between the two datasets was calculated using the Pearson rank correlation.

Fig 5. Cytokine secretion upon re-stimulation with non-acylated peptides and their acylated analogues.

PBMCs (2x10⁶cells/well) were stimulated with individually peptides (5 μg/ml). At day 3, 7 and 10 the media was replenished and IL-2 added. At day 14 the cells were re-stimulated (5 μg/ml) with their corresponding peptides and respective controls for 20 hours. Cytokine levels for IFN-γ, IL-5 and IL-10 was determined by fluorospot analysis. The figure shows the mean number of spot forming cells (SFC) to (A) IFN-g, (B) IL-5 and (C) IL-10 cytokines for each donor. Significance was calculated in relation to WT influenza HA by one-way ANOVA followed by a Dunn’s multiple comparisons test. (D and E) shows a summary of frequency of responding donors and magnitude of response of all cytokines. (D) Shows the frequency of responding donors to all cytokines (eg one positive hit for one cytokine equals a positive donor) and (E) the corresponding magnitude of the response expressed as an average of spot forming cells (SPC) for all three cytokines. The difference in frequency was calculated using a Fisher’s test and the p-values for the magnitude was calculated in relation to WT influenza HA by one-way ANOVA followed by a Dunn’s multiple comparisons test. Data are shown as mean of seven donors ± SD, each donor are represented by a dot.

Fig 6. Summary of frequency of responding donors and magnitude of response of all cytokines in all donors.

(A) the graph shows the frequency of responding donors to all cytokines (eg one positive hit for one cytokine equals a positive donor) and (B) the graph the corresponding magnitude of the response expressed as an average of spot forming cells (SPC) for all cytokines. n = 22. The difference in frequency was calculated using a one-sided Fisher’s test.

Fig 7. Summary of frequency of responding donors and magnitude of response to the vatreptacog alfa peptide of all cytokines in all donors.

(A) the graph shows the frequency of responding donors to all cytokines (eg one positive hit for one cytokine equals a positive donor) and (B) the graph the corresponding magnitude of the response expressed as mean spot forming cells (SFC)/10⁶ cells for all cytokines. n = 22. One-sided Fisher’s test was used to calculate the p value in (A). VA: vatreptacog alfa peptide.