Biochemical assays with recombinant human MHC II molecules can provide rapid, quantitative insights into immunogenic epitope identification, deletion, or design. Here, a peptide-MHC II binding assay scaled to 384-well plates is described. This cost effective format should prove useful in the fields of protein deimmunization and vaccine design and development.
Biochemical assays with recombinant human MHC II molecules can provide rapid, quantitative insights into immunogenic epitope identification, deletion, or design1,2. Here, a peptide-MHC II binding assay is scaled to 384-well format. The scaled down protocol reduces reagent costs by 75% and is higher throughput than previously described 96-well protocols1,3-5. Specifically, the experimental design permits robust and reproducible analysis of up to 15 peptides against one MHC II allele per 384-well ELISA plate. Using a single liquid handling robot, this method allows one researcher to analyze approximately ninety test peptides in triplicate over a range of eight concentrations and four MHC II allele types in less than 48 hr. Others working in the fields of protein deimmunization or vaccine design and development may find the protocol to be useful in facilitating their own work. In particular, the step-by-step instructions and the visual format of JoVE should allow other users to quickly and easily establish this methodology in their own labs.
Proteins are the fastest growing class of therapeutic agents6, and the rapid expansion of biotherapeutic pipelines has focused increasing attention on the challenges associated with development and use of protein drugs. One unique consideration stems from the fact that, in a healthy and functioning immune system, all extracellular proteins are sampled by antigen presenting cells (APCs). Once internalized by APCs, a protein is cleaved into small peptide fragments, and putative immunogenic segments are loaded into the groove of class II major histocompatibility complex proteins (MHC II). The peptide-MHC II complexes are then displayed on the APC surface, and true immunogenic peptides, termed T cell epitopes, form ternary MHC II-peptide-T cell receptor complexes with cognate CD4 T cell surface receptors7. This critical molecular recognition event initiates a complex signaling cascade, which results in T cell activation, the release of cytokines, CD4 T cell-mediated B cell maturation, and ultimately production of circulating IgG antibodies that bind to and clear the offending exogenous protein. Thus, immunogenic proteins might be deimmunized by identifying constituent T cell epitopes and mutating key residues responsible for MHC II complex formation. It bears noting, however, that T cell epitopes can be numerous and broadly distributed throughout immunogenic proteins, and the majority of epitope-deleting mutations are likely to cause an inadvertent loss of protein function or stability. Therefore, engineering deimmunized biotherapies can be a complex and technically challenging objective, but there exist several examples of successful T cell epitope deletion projects3,5,8-12. Unlike grafting-based "humanization", which is largely restricted to antibody therapeutics, epitope deletion can be applied to essentially any protein target regardless of sequence, structure, function, or the availability of homologous human scaffolds. The first step to implementing such an approach is identification of key peptide epitopes embedded within the target protein sequence.
High throughput biochemical assays using synthetic peptides and recombinant human MHC II molecules can provide rapid preliminary insights into epitope identification and mitigation1,3-5. These ELISA type assays can be a powerful complement to other protein/vaccine design and development tools. For example, one well established experimental approach to epitope mapping relies on time, labor, and resource intensive ex vivo cell proliferation assays 15. Briefly, the primary sequence of a target protein is first divided into a panel of overlapping peptides, often 15-mers with 12 residues overlap between adjacent peptides. The peptide panel is chemically synthesized and the immunogenicity of each peptide is tested in one of several different immunoassays that employ peripheral blood mononuclear cells (PBMC) isolated from human donors13,14. To provide greater confidence in results, peptides are typically tested in replicate with PBMC from 50 or more different donors. In cases where deimmunization is the ultimate objective, the work is compounded further by the need to produce additional panels of mutated peptides and test the new peptide panels in PBMC assays before introducing any deimmunizing mutations into the full length protein for subsequent functional analysis10. While these cellular assays remain the gold standard for assessing immunogenic potential in human patients, the efficiency of such an exhaustive approach might be improved by prefiltering putative immunogenic epitopes using a rapid and high throughput MHC II-peptide binding assay.
Likewise, biochemical peptide-MHC II binding assays can be combined with predictive in silico methods to radically accelerate the epitope identification process. There exist a variety of computational tools for T cell epitope prediction; examples include ProPred16, MHCPred17, SVRMHC18, ARB19, SMM-align20, NetMHCIIpan21 as well as proprietary tools such as EpiMatrix by EpiVax22. Likewise, epitope predictors have recently been combined with other bioinformatics and molecular modeling tools to yield integrated protein deimmunization algorithms designed to mitigate the risk that deimmunizing mutations might disrupt protein structure and function23-26. While several epitope predictors have proven to be reasonably accurate27,28, computational results invariably require experimental validation. Rapid, high throughput, and cost effective experimental methods are best suited as a preliminary filter for in silico epitope predictions.
In a similar vein, epitope predictors can drive antigen selection for reverse vaccinology29,30. For example, advances in bioinformatics have yielded whole genome screens that rapidly identify vaccine candidates in the form of whole proteins or peptide epitopes extracted from pathogen proteomes. While this enabling technology is reshaping discovery and development of protective vaccines, it introduces a new challenge in the form of intractably large lists of immunogenic vaccine candidates. High throughput peptide-MHC II binding assays can guide epitope selection by quantifying peptide binding affinity and binding promiscuity among multiple MHC II alleles. As with protein deimmunization, such experimental methods are ultimately required to validate computational prediction of promising vaccine leads.
Here, a peptide-MHC II binding assay scaled to 384-well format is described. The protocol is highly parallelized and reduces reagent costs by 75% compared to previously described 96-well plate formats1,3-5. Using a single liquid handling robot, this method allows one researcher to easily analyze approximately ninety test peptides in triplicate over a range of eight concentrations and four MHC II allele types in less than 48 hr. This article describes the setup of one 384-well ELISA plate for analysis of seven experimental peptides against one MHC II allele, but spread sheet calculators are provided as supplemental material so as to easily scale the experiment to any number of desired peptides and/or MHC II molecules.
Four major activities comprise the peptide-MHC II binding assay: 1-Binding: Test peptides compete with labeled control peptides for solution phase binding of soluble MHC II proteins. Binding is measured over a wide range of test peptide concentrations. 2-Capture: After the binding reaction approaches equilibrium, peptide-MHC II complexes are captured and separated from unbound peptide and protein by conformation-dependent recognition with an immobilized antibody. 3-Detection: Captured control peptide is quantitatively detected using time-resolved fluorescence. 4-Analysis: Spectroscopic data are processed, plotted and analyzed to ascertain dose-dependent binding properties of test peptide and MHC II proteins.
At various steps in the procedure below, the use of a liquid handling robot is recommended if one is available. Particularly in a 384-well format, automated dispensing and dilution of liquids minimizes user error, results in more consistent well-to-well volumes, and yields narrower 95% confidence intervals for IC50 values compared to manual 96-well assays (data not shown). If a liquid handling robot is not available, the steps annotated with "(liquid handling robot)" may be done by hand. Similarly, if possible, an automated plate washer that is compatible with the 384-well format is recommended. This effectively standardizes the plate washing process. If a plate washer is not available, the steps annotated with "(plate washer)" may be done by hand.
1. Coat the ELISA Plate (Day 1)
2. Make the Test Peptide Dilutions (Day 1)
Dilution Number | Volume to take from prior dilution/stock | Volume Citrate Buffer to add | Conc. of Dilution | Conc. in Binding Reaction | Conc. in Neutralized ELISA |
(μl) | (μl) | (μM) | (μM) | (μM) | |
1 | 0.7 | 35.1 | 195.531 | 97.765 | 48.883 |
2 | 14.3 | 14.3 | 97.765 | 48.883 | 24.441 |
3 | 7.1 | 28.7 | 19.389 | 9.695 | 4.847 |
4 | 14.3 | 14.3 | 9.695 | 4.847 | 2.424 |
5 | 7.1 | 28.7 | 1.923 | 0.961 | 0.481 |
6 | 14.3 | 14.3 | 0.961 | 0.481 | 0.24 |
7 | 7.1 | 28.7 | 0.191 | 0.095 | 0.048 |
8 | 14.3 | 14.3 | 0.095 | 0.048 | 0.024 |
Remove and discard 7.1 μl of peptide solution from dilution number 8. |
Table 1. Preparation of test peptide dilution series.
Figure 1. Plate Map of Peptide Binding Assay. (A) Map of the peptide dilutions in polypropylene 384-well plates. Each polypropylene plate can accommodate three groups of 15 peptides in competition binding reactions against a single MHC II allele. (B) After the binding reaction approaches equilibrium, each peptide group is transferred, in triplicate, to a separate 384-well ELISA plate that has been precoated with anti-MHC II antibody.
3. Prepare the MHC II Master Mix (Day 1)
MHC II Allele DRB1*: | 1501 |
MHC II Stock Concentration (mg/ml) | 1.3 |
MHC II Stock Concentration (µM) | 20 |
Vol. MHC II Stock to Add (µl) | 14.65 |
Vol. Reaction Buffer to Add (µl): | 2885.35 |
MHC II Master Mix Concentration (nM) | 101 |
Table 2. Preparation of MHC II master mix.
4. Prepare the Negative and Positive Controls (Day 1)
5. Add the Appropriate Control Peptide to the MHC II Master Mix (Day 1)
MHC II Allele DRB1* | Biotinylated Control Peptide | (residues) | Sequence |
0101 | Flu-HA-B | (306-318) | Biotin-(Ahx)(Ahx)PRYVKQNTLKLAT-amide |
0301 | Myoglobin | (137-148) | Biotin-(Ahx)-(Ahx)-LFRKDIAAKYKE-OH |
0401 | YAR-B | (1-14) | Biotin-(Ahx)(Ahx)YARFQSQTTLKQKT-OH |
0701 | TetTox-B | (830-843) | Biotin-(Ahx)(Ahx)QYIKANSKFIGITE-OH |
1101 | Flu-HA-B | (306-318) | Biotin-(Ahx)(Ahx)PRYVKQNTLKLAT-amide |
1501 | MBP-B | (84-102) | Biotin-(Ahx)(Ahx)NPVVHFFKNIVTPRTPPPS-OH |
*We recommend ordering 10 mg scale for economy.
Table 3. Control Peptides for Various MHC II Alleles.*
6. Make the Binding Reaction (Day 1)
7. Neutralize and Transfer the Binding Reaction (Day 2)
8. ELISA Development (Day 2 or 3)
9. Data Analysis
The mature peptide sequence of Enterobacter cloacae P99 beta-lactamase (BLA) (Genbank ID# X07274.1) was analyzed with ProPred16 for putative peptide binders to MHC II allele DRB1*1501 (Table 4). ProPred identified 117 nonamer peptides with an obligatory P1 anchor residue (i.e. an M, L, I, V, F, Y, or W at position 1, which is required for MHC II binding31). At a 5% threshold, only peptides with scores greater than or equal to 2.6 are likely binders. Thus, at the 5% threshold, only the top 11 peptides are predicted to bind MHC II DRB1*1501.
Table 4. Top-scoring ProPred predictions for BLA peptides and MHC II allele DRB1*1501.
A representative panel of predicted epitopes was selected for analysis in our MHC II binding assay (Table 4, bold entries). BLA peptide fragments of fifteen residues were chemically synthesized such that putative nonamer MHC II epitopes were embedded within the synthetic protein fragments. While the MHC II binding groove itself accommodates only nine amino acids, evidence suggests that flanking sequences can influence peptide-MHC II interactions, and synthetic peptides of 15-20 residues are therefore commonly employed15. To represent the complexity of overlapping epitopes that might occur in biologically processed protein fragments, we tested synthetic sequences that contained (i) single predicted binders, (ii) single predicted nonbinders, (iii) multiple predicted binders, (iv) multiple predicted nonbinders, or (v) a mixture of predicted binders and nonbinders (Tables 4 and 5).
Table 5. List of Chemically Synthesized Peptides.
The capacity of these synthetic peptides to compete with the biotinylated MBP-B control peptide (Table 3) for binding to MHC II DRB1*1501 was analyzed as described in the protocol above. Competitive binding curves for the synthetic peptides are shown in Figure 2. IC50 values were calculated by fitting the log-transformed data using the one-site competitive binding, nonlinear fit function of Prism (Table 6). As seen in Figure 2, the peptides naturally partition into three groups: strong binders with IC50 < 1 μM (peptides 2 and 10); moderate binders with 1 μM ≤ IC50 < 100 μM (peptides 4, 9, 11, and 24); weak binders with IC50 ≥ 100 μM (peptide 31).
Figure 2. Competitive Binding Curve of BLA Peptides for MHC II allele DRB1*1501. Tight binders are fit with orange lines, moderate binders green lines, and the weak binder a maroon line.
Peptide Rank | IC50 (μM) | 95% C.I. for IC50 (μM) | Quality of Fit (R2) |
2 | 0.1199 | 0.09404-0.1529 | 0.98 |
4 | 15.1 | 11.83-19.28 | 0.87 |
9 | 17.11 | 13.39-21.88 | 0.81 |
10 | 0.2743 | 0.2149-0.3502 | 0.98 |
11 | 10.49 | 8.252-13.34 | 0.98 |
24 | 4.787 | 3.769-6.082 | 0.96 |
31 | 190.6 | 137.7-263.9 | 0.80 |
Table 6. Calculated IC50 values of BLA peptide fragments for DRB1*1501.
Biotherapeutics have established themselves as a cornerstone of modern medicine, representing four of the top five selling drugs in 201232. The biopharmaceutics sector has shown sustained growth for several years6, and the ongoing development of novel agents as well as the emergence of biosimilars has expanded biopharmaceutical pipelines. Looking to the future, assessing and mitigating the immunogenicity of protein therapeutics will become an integral part of early stage biotherapeutic development. To facilitate this process, biotechnologists may avail themselves of computational methods for epitope prediction16-22 as well as integrated protein design algorithms that seek to reduce immunogenicity while maintaining function3,23-26. Likewise, vaccine design and development may capitalize on the power of predictive algorithms4, and the continued maturation of reverse vaccinology is expected to yield protective vaccines for a broad array of infectious agents that have eluded previous efforts33. While these computational tools have the capacity to radically accelerate drug and vaccine development processes, the results of in silico predictions must ultimately be filtered further so as to identify a tractable number of top performing candidates. Recent advances in quantitative peptide-MHC II analysis have included the development of engineered yeast cell surface display systems34 and microbead based Luminescent Oxygen Channeling Immunoassays35, but in vitro ELISA type assays remain the workhorse for quantifying binding between specific MHC II and peptide pairs. Here, a microliter scale procedure for rapid, quantitative, and cost effective analysis of peptide epitope binding to human MHC II immune molecules is described. These methods may be used as an early stage experimental tool in the funneling process.
For the purposes of demonstration, the sequence of BLA was analyzed for putative MHC II binding peptides using the ProPred web server. Seven of these peptides were chemically synthesized and experimentally tested for binding to soluble human MHC II DRB1*1501. Similar to other analyses of predictor accuracy 28, ProPred binding prediction correlated reasonably well with experimentally measured IC50 for the DRB1*1501 MHC II allele. It is also important to note that results are sensitive to variations in concentrations and care should be taken to ensure accurate dilutions of MHC, control peptides, and test peptides. Likewise, a limitation of this method is that the IC50 values are relative to the concentrations of MHC and control peptides. Among the BLA peptide test set, the highest ranked ProPred epitope exhibited the strongest binding to recombinant DRB1*1501. Similarly, all other predicted epitope binders exhibited moderate to high MHC II affinities. As expected, synthetic peptides containing overlapping predicted binding epitopes and nonbinding epitopes were found to bind MHC II in all cases. Peptide 31, which spanned two of the lowest ranked predicted epitopes, proved to be the weakest MHC II binder. Peptide 24, however, was found to bind MHC II with moderate affinity despite the fact that its only highly ranked constituent epitope (epitope 24) was not a predicted binder at the 5% threshold. Interestingly, epitope 24 was identified in a previous study aimed at deimmunizing BLA10. In those experiments, synthetic peptide 24 was shown to be highly immunogenic in human PBMC assays. Thus, in this one case, our MHC II binding assay proved to be predictive of immunogenicity while ProPred did not. In aggregate the results demonstrate the utility of computational predictions in guiding researchers towards likely immunogenic epitopes, but they also underscore the importance of high throughput experimental methods as an integral component of the protein/vaccine design process.
Importantly, this 384-well experimental design is highly parallelized and permits rigorous analysis of up to 15 peptides and one MHC II allele per 384-well ELISA plate. Using a single liquid handling robot, this method allows one researcher to analyze ninety test peptides against four MHC II allele types in less than 48 hours. This design includes eight different test peptide concentrations and triplicate measurement at each concentration. Therefore, the user is assured robust and reproducible quantitative results. Others working in the fields of protein therapeutics or vaccine design and development may find this protocol to be useful in facilitating their work in protein deimmunization or immunogenicity analysis.
Citrate Phosphate Buffer (22.2 mM Citric Acid, 55.6 mM Dibasic Na2HPO4, pH 5.4) |
222 ml 0.1 M citric acid |
278 ml 0.2 M dibasic Na2HPO4 |
500 ml MilliQ H2O |
Mix citric acid, sodium phosphate, and 450 ml of MilliQ H2O. Adjust pH to 5.4 and q.s. to 1 L. |
This solution is stable at room temperature. |
Binding Reaction Buffer |
Citrate Phosphate Buffer, pH 5.4 with |
2% v/v 1 mM PefaBloc |
1.5% w/v Octyl-β-D-glucopyranaside |
Use right away. |
1 mM PefaBloc |
1 ml MilliQ H2O |
25 mg PefaBloc SC (powder) |
Make aliquots and store at -20 ˚C. |
Borate Buffer Formula (12.5 mM Sodium tetraborate decahydrate, pH 8.2) |
1.19 g sodium tetraborate decahydrate (Sigma) |
250 ml MilliQ H2O |
Dissolve sodium tetraborate decahydrate in 225 ml of MilliQ H2O. Adjust pH to 8.2 and q.s. to 250 ml. |
This solution is stable at room temperature. |
PBS-Tween (2.7 mM KCl, 1.5 mM KH2PO4, 136.9 mM NaCl, 8.9 mM Na2HPO4•7H2O, 0.05% Tween, pH 7.4) |
4.5 L MilliQ H2O |
500 ml 10x Dulbecco’s PBS (DPBS) |
2.5 ml Tween 20 |
This solution is stable at room temperature. |
Neutralization Buffer (50 mM Trizma HCl, pH 8.0) |
475 ml MilliQ H2O |
25 ml 1 M Trizma HCl |
Mix and adjust pH to 8.0 |
This solution is stable at 4 °C. |
Table 7. Buffer Recipes.
The authors have nothing to disclose.
This work was supported by NIH grants R01-GM-098977 and R21-AI-098122 to CBK and KEG. RSS was supported in part by a Luce Foundation Fellowship and in part by a Thayer Innovation Program Fellowship from the Thayer School of Engineering.
Sodium Tetraborate Decahydrate | Sigma | 221732 | |
Citric Acid | Sigma | C1901 | |
Dibasic Sodium Phosphate | Sigma | S7907 | |
Trizma HCl | OmniPur | 9310 | |
Tween-20 | Sigma | P7949 | |
100% DMSO | Sigma | D8418 | |
Octyl-β-D-Glucopyranaside | Fischer | 29836-26-8 | |
Pefa Bloc SCF | Roche | 1158591600 | |
Dulbecco’s 10X Phosphate Buffered Saline | Invitrogen | 14200-166 | |
DELFIA Assay Buffer | Perkin Elmer | 4002-0010 | |
DELFIA Enhancement Buffer | Perkin Elmer | 4001-0010 | |
Europium Labelled Streptavidin | Perkin Elmer | 1244-360 | |
L243 anti-HLA-DR antibody | Biolegend | 307602 | |
Biotinylated tracer peptides | 21st Century Biochemicals | Custom Order | |
Test peptides (1-4mg, 85% purity) | Genscript | Custom Order | |
Purified HLA-DRB1 monomers (non-biotinylated) | Benaroya Research Institute* | Custom Order | |
384-well white EIA/RIA plate | Thermo | 460372 | |
Polypropylene 384-well plate | Costar | 3656 | |
Film, AxySeal, 80 µm, ELISA Applicator | Axygen Asicentific | PCR-SP | |
MilliQ Water | N/A | N/A | |
Epimotion Liquid Handler (or similar) | Eppendorf | 5075 | |
Select TS Plate Washer (or similar) | BioTek | 405 | |
SpectraMax Gemini Microplate Reader (or similar) | Molecular Devices | N/A | |
*Recombinant human MHC II molecules can be obtained from the Benaroya Tetramer Core Laboratory. See: https://www.benaroyaresearch.org/our-research/core-resources/tetramer-core-laboratory |