Autoimmune Responses to Soluble Aggregates of Amyloidogenic Proteins Involved in Neurodegenerative Diseases: Overlapping Aggregation Prone and Autoimmunogenic regions

Abstract

Why do patients suffering from neurodegenerative diseases generate autoantibodies that selectively bind soluble aggregates of amyloidogenic proteins? Presently, molecular basis of interactions between the soluble aggregates and human immune system is unknown. By analyzing sequences of experimentally validated T-cell autoimmune epitopes, aggregating peptides, amyloidogenic proteins and randomly generated peptides, here we report overlapping regions that likely drive aggregation as well as generate autoantibodies against the aggregates. Sequence features, that make short peptides susceptible to aggregation, increase their incidence in human T-cell autoimmune epitopes by 4-6 times. Many epitopes are predicted to be significantly aggregation prone (aggregation propensities ≥10%) and the ones containing experimentally validated aggregating regions are enriched in hydrophobicity by 10-20%. Aggregate morphologies also influence Human Leukocyte Antigen (HLA)--types recognized by the aggregating regions containing epitopes. Most (88%) epitopes that contain amyloid fibril forming regions bind HLA-DR, while majority (63%) of those containing amorphous β-aggregating regions bind HLA-DQ. More than two-thirds (70%) of human amyloidogenic proteins contain overlapping regions that are simultaneously aggregation prone and auto-immunogenic. Such regions help clear soluble aggregates by generating selective autoantibodies against them. This can be harnessed for early diagnosis of proteinopathies and for drug/vaccine design against them.

Figure 1

Box and Whisker plots showing average aggregation propensities of the peptide sequences in experimentally validated human T-cell autoimmune epitopes and MHCIInonbind datasets, predicted by (a) TANGO (P_Agg-TANGO) and (b) WALTZ (P_Agg-WALTZ). All epitope classes (HLA-DP, HLA-DQ and HLA-DR) show long tails suggesting that a number of epitopes in these datasets are aggregation prone.

Figure 2. Overlap between aggregation prone and T-cell autoimmune epitope regions in human amyloidogenic proteins.

Human amyloidogenic proteins whose amino acid sequences contain experimentally validated aggregating peptides (bold font) as well as experimentally validated T-cell autoimmune epitope peptides (underlined) are shown. All available experimental data on these proteins is consolidated into the regions highlighted in the sequences. The PDB was queried for available structural information on these proteins and information obtained was used to map overlapping aggregation prone and autoimmune epitope regions in the respective protein structures. These regions are shown in magenta. The magenta regions in structural images correspond to the sequence regions that are simultaneously shown in bold magenta font and underlined. (a) Human Major Prion Protein. Structural information is from the PDB entry 4 KML chain A. The N-terminal residues 1–116 are not present in the structure. (b) Human Amyloid-β peptide 1–42. Structural information is from the PDB entry 2BEG, which shows this peptide in fibrillary form. N-terminal residues 1–16 are not present in the structure. (c) Human Insulin. Structural information was taken from the PDB entry 3EY7. The chains A and B are shown in yellow and blue ribbons. The disulfide bonds are also shown. (d) Human Pmel. No structural information is available for this protein.

Figure 3. Additional examples of overlap between aggregation prone and T-cell autoimmune epitope regions in human amyloidogenic proteins are presented.

This figure is prepared in the same way as the Fig. 2, except that the T-cell autoimmune epitope shown here are the strongly predicted ones. (a) Human α-Synuclein. Structural information is from the PDB entry 1XQ8. This structure is for the micelle bound form of α-Synuclein. In the unbound form, it is an intrinsically disordered protein. (b) Human β2-microglobulin. Structural information is from the PDB entry 2D4F. (c) Human Cystatin C. The structural information is from the PDB entry 1TIJ chains A and B. This PDB entry contains domain swapped form of Cystatin C dimer. Both the chains are shown here in yellow and blue ribbons.

Figure 4. Sequence patterning synergies between the T-cell autoimmune epitopes and aggregation prone regions.

Sequence logos are shown for (a) 100,000 randomly generated 15-residues long peptides, (b) 83,616 random peptides that were not predicted to be T-cell autoimmune epitopes (non-epitopes), (c) 16,384 strongly predicted T-cell autoimmune epitopes, (d) 12,179 random peptides that contain TANGO predicted APRs, (e) 8,559 non-epitopes that contain TANGO APRs, (f) 3,620 T-cell autoimmune epitopes that contain TANGO APRs, (g) 9,582 random peptides that contain WALTZ predicted APRs, (h) 7,261 non-epitopes that contain WALTZ APRs, and (i) 2,331 T-cell autoimmune epitopes that contain WALTZ APRs. Note the enrichment of hydrophobic β-branched and aromatic residues, particularly Val, Ile, and Phe, at initial positions of T-cell autoimmune epitope peptides (see panels c,f,i).