Peptide Library

We identified MHC class I prediction tools through literature searches, and the IMGT link list at http://imgt.cines.fr/textes/IMGTbloc-notes/Immunoinformatics.html#tooMHCbp. Identical tools appearing on multiple websites—most often in combination with proteasomal cleavage/TAP transport predictions—were only included once. Several tools were not available at the time of the study (October/November 2005). One server containing multiple prediction tools (http://www.imtech.res.in/raghava/) could unfortunately not be included, as its terms of use limit the number of predictions to 20 a day.
Several tools allowed making predictions with different algorithms. In cases like this, we retrieved predictions for both, and treated them as separate tools: multipred provides predictions based on either an artificial neural network or a hidden Markov model, which we refer to as multipredann and multipredhmm. Similarly, netmhc provides neural network–based predictions (netmhc_ann) and matrix-based predictions (netmhc_matrix), and mappp provides predictions based on bimas (mapppB) and syfpeithi (mapppS) matrices.
For each tool, we mapped the MHC alleles for which predictions could be made to the four-digit HLA nomenclature (e.g., HLA-A*0201). If this mapping could not be done exactly, we left that allele–tool combination out of the evaluation. For example, HLA-A2 could refer to HLA-A*0201, A*0202, and A*0203, which do have a distinct binding specificity.
For each tool in the evaluation, we wrote a python script wrapper to automate prediction retrieval. The retrieved predictions were stored in a MySQL database. If a tool returns a nonnumeric score such as “–” to indicate nonbinding, an appropriate numeric value indicating nonbinding on the scale of the tool was stored instead.
The algorithms underlying each tool fall in the following categories: arbmatrix, bimas, hla_a2_smm, hlaligand, libscore, mapppB, mapppS, mhcpathway, mhcpred, netmhcmatrix, predbalbc, predep, rankpep, and syfpeithi are based on positional scoring matrices, while multipredann and netmhcann are based on ANNs, multipredhmm is based on a hidden Markov model, pepdist is based on a peptide–peptide distance function, and svmhc is based on a support vector machine. With two exceptions, the tools were generated based on data of peptides binding to or being eluted from individual MHC molecules. The first exception is libpred, which was generated using binding data of combinatorial peptide libraries to MHC molecules, and predep, where the 3-D structure of the MHC molecules was used to derive scoring matrices. References with more detailed description of each tool are indicated in the text.

For each screening experiment, ∼400 mg of the macrocyclic peptide library was swollen in DCM and washed extensively with DMF, ddH₂O and finally incubated overnight at 4°C in 10 mL of blocking buffer (30 mM sodium phosphate, pH 7.4, 150 mM NaCl, 0.05% Tween 20, 3% BSA and 0.1% gelatin). The solution was drained, and the resin was resuspended in 10 mL of blocking buffer containing 500 nM biotinylated Etf-1 for overnight at 4°C. The resin was washed with blocking buffer and resuspended in 10 mL of blocking buffer. M280 streptavidin-coated Dynabeads (Invitrogen; 20 μL) were added to the resin and the mixture was incubated on a rotary wheel for 1 h at 4°C. The magnetic beads were isolated from the bulk by using a TA Dynal MPC-1 magnetic particle concentrator (Invitrogen). The beads were transferred to a Bio-Spin column (0.8 mL; BioRad, Hercules, CA, USA) and incubated in blocking buffer containing 100 nM biotinylated Etf-1 for 4 h at 4°C. The solution was drained, and the resin was quickly washed with blocking buffer to remove any unbound protein. The resin was resuspended in 1 mL of blocking buffer and streptavidin-alkaline phosphatase (SA-AP) conjugate was added to the mixture (1 μg/mL final concentration). After 10 min at 4°C the solution was drained and the beads were quickly washed with 1 mL of blocking buffer (3×) and 1 mL of staining buffer (30 mM Tris, pH 8.5, 100 mM NaCl, 5 mM MgCl₂, and 20 μM ZnCl₂) (3×). The resin was resuspended in 1.5 mL of staining buffer in a Petri dish and 150 μL of a 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) solution (5 mg/mL) was added. After 30 min, 100 μL of 1 M HCl was added to quench the reaction and the intensely turquoise positive beads were manually isolated under a dissecting microscope. Peptide sequences of hit beads were determined using partial Edman degradation-mass spectrometry (PED-MS) as previously described (25 (link), 37 (link)).

The panel of 14 recombinant T-antigens used to characterize antibody reactivity has been previously described [18 (link), 29 (link)]. All T-antigens were expressed in Escherichia coli BL21 (DE3) cells, and N-terminal His₆-tags were removed using recombinant tobacco etch virus (rTEV) prior to use in immunoassays.
The recombinant T18.1 dimer was constructed by inserting the coding sequence for tee18.1 into pProExHTA in tandem such that the vector encoded expression of two T18.1 antigens in which the C-terminal sortase motif of the first monomer is linked to the N-terminus of the following monomer, via a flexible (glycine)₃ linker. The first five N-terminal residues from the second T18.1 monomer (ETAGV) were removed to allow the positioning of the sortase motif of the first monomer near the sortase motif peptide “binding” cleft in the second monomer. Expression and purification of the T18.1 dimer followed published protocols for T-antigen monomers [18 (link)]. Briefly, E. coli BL21 (DE3) cells transformed with the expression plasmid were induced with 0.3 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) and grown at 18°C for 16 h. Cell pellets were resuspended in lysis buffer (50 mM Tris-Cl [pH 8.0], 300 mM NaCl, 10 mM imidazole), lysed using a cell disruptor (Constant Cell Disruption Systems) and the protein enriched from the soluble phase using Ni²⁺-NTA affinity purification. The His₆ affinity tag was cleaved with recombinant tobacco etch virus (rTEV) protease and rTEV-His₆ protease, and uncleaved protein was removed by subtractive immobilized-metal affinity chromatography (IMAC). The unbound protein fraction containing recombinant T-antigen was further purified by size exclusion chromatography on a Superdex S200 10/300 column (GE Healthcare) in crystallization buffer (10 mM Tris-Cl [pH 8.0], 100 mM NaCl).
For generation of biotinylated recombinant T18.1, the tee18.1 sequence [18 (link)] was re-cloned into a pProExHTA vector modified to encode for an N-terminal His-tag and a C-terminal Avi-tag. E. coli BL21 (DE3) were co-transformed with pProExHTA-tee18.1-Avitag and pACYC184BirA, and expression media were supplemented with 20 uM D-Biotin for in cell biotinylation. The biotinylated T18.1 was purified using Ni²⁺-NTA affinity chromatography, the His₆-tag removed using rTEV, and final purification was performed using size exclusion chromatography as described above [18 (link)]. Successful biotin labelling was confirmed by immunoaffinity pulldown with streptavidin-coupled M-280 Dynabeads (Thermofisher) following by SDS-PAGE.
The T18.1 peptide library was designed to cover the entirety of the mature T18.1 monomer. The N-terminal signal sequence was excluded, and the library ended at the conserved threonine residue of the Sortase C recognition site (QVPTG). The tiled peptide library comprised 2915-mer peptides overlapping by five residues (Supplementary Table S1). The peptides were synthesized (GenScript) and purified to >85%.