Novel function of complement C3d as an autologous helper T-cell target

doi:doi:10.1038/sj.icb.7100147

Immunology and Cell Biology (2008) 86, 221–225; doi:10.1038/sj.icb.7100147; published online 8 January 2008

Novel function of complement C3d as an autologous helper T-cell target

Paul M Knopf¹, Daniel S Rivera², Si-Han Hai², Julie McMurry², William Martin² and Anne S De Groot^2,3

¹Department of Molecular Microbiology and Immunology, Brown University, Providence, RI, USA
²EpiVax Inc., Providence, RI, USA
³Department of Medicine, Brown University, Providence, RI, USA

Correspondence: Dr PM Knopf, Department of Molecular Microbiology and Immunology, Brown University, Box G-B6, 146 Clifford Street, Providence, RI 02912, USA. E-mail: Paul_Knopf@Brown.edu

Received 7 September 2007; Revised 1 November 2007; Accepted 4 November 2007; Published online 8 January 2008.

Top

Abstract

The C3d fragment of complement component C3 has been shown to enhance immune responses to antigens that lack T-cell epitopes such as bacterial polysaccharides. C3d binds to the B-cell complement receptor 2 (CR2 or CD21); this binding serves as a co-activation signal to the B cell when the polysaccharide antigen portion binds simultaneously to the B-cell receptor (surface Ig). Bringing together receptor-associated signal transduction molecules CD19 and Ig alpha / beta , respectively, results in a lower threshold of activation. Paradoxically, C3d has also been shown to enhance antibody titers in the CD21 knockout (KO) mouse model as well as increase Th1 and Th2 cytokine secretion, suggesting that that an auxiliary CR2-independent pathway of immune activation may exist. We hypothesized that in addition to its molecular adjuvant property that enhances signal 1 during B-cell activation (co-signal 1), C3d also contains T-cell epitopes that are able to stimulate autoreactive C3d peptide-specific helper T cells which we term 'co-signal 2'. Using the EpiMatrix T-cell epitope-mapping algorithm, we identified 11 putative T-cell epitopes in C3d, a very high epitope density for a 302 amino-acid sequence. Eight of these epitope candidates were synthesized and shown to bind a variety of class II HLA-DR molecules of different haplotypes, and to stimulate C3d peptide-specific T cells to secrete pro-inflammatory cytokines in vitro. Further, we demonstrate a C3d-peptide specific increase in CD4⁺ intracellular IFN- gamma ⁺ T cells in peripheral blood mononuclear cells (PBMCs) exposed to C3d peptides in vitro. We believe that the discovery of these autologous T cells autoreactive for C3d provides evidence supporting the 'co-signal 2' hypothesis and may offer a novel explanation of the CD21 KO paradox.

Keywords:

C3, C3d, complement, helper T cell, epitope, vaccine

The efficacy of polysaccharide vaccines has been significantly improved by conjugation with either an immunogenic foreign carrier protein or with complement component fragment C3d.^{1, 2, 3, 4, 5} The adjuvant effect of the carrier protein is attributed to its role in recruiting cognate T help, augmenting the B-cell response to polysaccharide antigens. Polysaccharides usually feature multiple repeats of their sugar subunit residues, which can cause crosslinking of the breakpoint cluster regions (BCRs) and the internalization of the conjugate into the antigen processing compartments of the B cell. For protein antigens or protein conjugates of polysaccharides, this processing step also generates carrier-derived peptides that bind internally to MHC class II molecules and are subsequently displayed on the B-cell surface. Thus, help (in the form of cytokines and differentiation factors) is generated by activation of CD4⁺ helper T cells via their T-cell antigen receptors (TCR). The generation of T-cell dependent (Td) responses may serve to create antipolysaccharide antibodies that are more robust than T-cell independent (Ti) responses by promoting affinity maturation and isotype switching.

The prevailing model of the mechanism responsible for the adjuvant effect of C3d is quite different. In this model, the simultaneous binding of C3d to complement receptor 2 (CR2 or CD21) and the polysaccharide antigen to the B-cell receptor have been shown to provide a co-activation signal to the naive B cell (co-signal 1). However this model does not explain the enhancement of immune responses in studies with CD21 knockout mice, nor does it provide insight into C3d augmentation of Th1 and Th2 immune responses. Moreover, an elegant study comparing the antibody response to a covalent conjugate of the capsular polysaccharide of serotype 14 Streptococcus pneumoniae (PPS14) with C3d in athymic nude mice demonstrated that T cells were required for enhancement of the memory B-cell response after a second injection of PPS14-C3d. We therefore hypothesized that T-cell epitopes contained within C3d may be processed and presented by antigen presenting cells (APCs) to stimulate autoreactive C3d peptide-specific helper T cells. These activated autologous C3d epitope-specific T cells would secrete pro-inflammatory cytokines and provide a co-activation signal (co-signal 2) allowing B-cell maturation.

In a previous immunoinformatics study of the human serum proteome⁶ we mapped T-cell epitopes in complement C3 using the EpiMatrix algorithm and confirmed the presence of multiple HLA–DR binding motifs in the protein. We noted that the C3 component of complement had a significantly higher 'cluster-score' (concentration of HLA binding motifs) than most other serum proteins analyzed.⁷ The high potential for immunogenicity of C3 was found to contrast sharply with other serum protein components including albumin, constant domain of immunoglobulin, transferrin and certain hormones, which scored significantly lower. The C3d fragment of C3 (aa residues 1002–1303 in C3 or 1–302 in C3d)⁸ scored highest in the subsequent EpiMatrix analysis. The present studies were conducted to determine whether the previously observed enhancement of the immune response by C3d might be due in part to the activation of C3d peptide-specific autoreactive helper T cells, and whether the unusually high concentration of T-cell epitopes in C3d reflect an auto-inflammatory role of this particular protein subunit.

Top

Results

Epitope analysis and peptide synthesis

Candidate T-cell epitopes in C3 were identified as described in Methods. Eleven clusters that scored greater than 10 units on an epitope identity scale⁹ were identified in the 302 aa fragment C3d. (Table 1). Compared to other serum proteins, C3d contains a 10- to 100-fold higher T-cell epitope cluster density, with 3 of these 11 clusters having very high scores (>20 units). An additional four clusters have the scores >15. These seven top scoring candidate epitopes are distributed into 2 regions of C3d—Region 1 (aa 10–120) has four of the peptides with scores >15 and Region 2 (aa 175–290) has the remaining three. The amino and carboxyl terminal 10–15 residues and the mid-molecule region aa 120–175 segments were found to be devoid of epitopes, based on EpiMatrix analysis (Figure 1). Eight of the 11 peptides that were initially identified by cluster score using the EpiMatrix algorithm as putative T-cell epitopes were prepared commercially (at greater than 90% purity on an automated Rainen Symphony/Protein Tech synthesizer (SynPep, Dublin, CA, USA and New England Peptide, Gardner, MA, USA). Peptides 32–50, 100–118, 190–209, having three of the lowest four cluster scores, were not selected for synthesis. Epitopes 73–96 and 176–198 had the highest cluster scores but were difficult to synthesize due to high hydrophobicity.

Figure 1.

Map and score of putative C3d T-cell epitopes. Each amino acid within C3d is given a number corresponding to its position (N terminus to C terminus); position 1 of C3d is the position 1002 of C3. Horizontal bars below the sequence correspond to the location of the putative clusters. The color of the bar corresponds to its EpiMatrix cluster score on the scale; the scores are broken down into strength of putative immunogenicity as benchmarked by published epitopes. Scores also shown below bars. Cluster scores of 10 and higher are considered to be potentially immunogenic. The highest scoring clusters were synthesized as peptides and were analyzed for their ability to bind HLA and activate T cells in vitro.

Full figure and legend (176K)

Table 1 - C3d peptide locations, sequences, EpiMatrix cluster scores and in vitro binding affinity.

Full table (67K)

HLA binding assays

In order to validate the in silico predictions, HLA binding assays were performed. All of the peptides were shown to bind to at least one of the four different HLA alleles tested with high affinity (Table 1), confirming the EpiMatrix predictions. Peptide 176–198 was shown to bind very strongly to three alleles and moderately to one allele, peptide 45–65 bound moderately to all four alleles tested. Peptides 223–246 and 73–96 bound to DRB1^*0401 with very high affinity as did peptide 62–80 to DRB1^*1501.

Stimulation of PBMCs by C3d-derived peptides leads to IFN- secretion

To determine if C3d-specific responses could be detected in human subjects, freshly isolated peripheral blood mononuclear cells (PBMCs) from four healthy subjects were stimulated with or without a C3d peptide pool (comprised of 11–26, 45–65, 62–80, 223–246, 248–265 and 269–286) for 7–10 days. PBMCs were then each re-stimulated with the C3d peptide pool or no peptide. Levels of IFN- gamma secretion were determined 3 days post re-stimulation by ELISA or ELISpot assay. The stimulation of PBMCs with the C3d peptide pool led to greater than a threefold increase in IFN- gamma secretion over no peptide stimulation (Figure 2a). To further characterize the C3d peptide responses, PBMCs were stimulated with the C3d peptide pool or no peptide for 7–10 days. PBMCs were then re-stimulated with individual C3d peptides or no peptide. C3d-specific IFN- gamma production was determined by ELISpot. In both subjects tested, dramatic increases in IFN- gamma production (greater than threefold) were seen with peptides 223–246 and 269–286 (Figure 2b).

Figure 2.

(a) Stimulation with C3d peptide pool leads to increased secretion of IFN- gamma . Peripheral blood mononuclear cells (PBMCs) were cultured with or without C3d peptide pool (10 g ml^-1). IFN- gamma secretion was measured by ELISA upon re-stimulation with C3d peptides; all four subjects had greater than twofold increases over a background of IFN- gamma secretion in response to C3d, compared to no peptide. (b) Stimulation with individual C3d peptides leads to increased secretion of IFN- gamma . PBMCs were cultured with or without C3d peptide pool (10 g ml^-1). IFN- gamma secretion was measured by ELISpot upon re-stimulation with individual C3d peptides. Stimulation with peptides 223–246 and 269–286 led to dramatic increases in IFN- gamma secretion over background.

Full figure and legend (78K)

Autologous helper T cells recognize C3d-derived peptides

A key component of our 'co-signal 2' hypothesis relies on the recognition of C3d-derived peptides by autoreactive helper T cells. Cryopreserved PBMCs from an HLA-DR1 subject were stimulated as described above and IFN- gamma secretion was assayed on day 10 by ELISA. Stimulation with the C3d peptide pool led to a nearly 10-fold increase of IFN- gamma over background (Figure 3a). We subsequently evaluated the phenotype of the IFN- gamma secreting cells by surface and intracellular cytokine staining (ICCS). IFN- gamma + cells were found to reside exclusively in the CD4⁺ helper T-cell population (Figure 3b).

Figure 3.

(a) Stimulation of peripheral blood mononuclear cells (PBMCs) from normal healthy human donor with pool of C3d peptides led to secretion of IFN- gamma . PBMCs were stimulated in culture with C3d peptide pool (10 g ml^-1) or no peptide for 7 days. Cells were then washed and re-stimulated with C3d peptides or no peptide. Culture supernatants were collected 3 days later and analyzed by IFN- gamma ELISA. (b) Cells were collected on day 10 and re-stimulated for 4 h at 37 °C with C3d peptides or flu HA (10 g ml^-1 as a negative control) in the presence of GolgiPlug (BD Biosciences). IFN- gamma helper T cells were detected in the C3d re-stimulated cells but not in the negative control. No CD8⁺ IFN- gamma ⁺ cells were detected (data not shown).

Full figure and legend (41K)

Top

Discussion

Initiation of a primary humoral immune response can often be augmented by the formation of complexes composed of the antigen covalently linked to complement component C3d.^{10, 11} This enhanced response has been attributed to the crosslinking of the antigen recognition molecules (Ig) and complement-binding molecules (CD21 also known as CR2), bringing together Ig alpha / beta and CD19, respectively to generate signal 1.^{12, 13} It has been postulated that this crosslinking thereby lowers the threshold for generating signal 1 and for response to signal 1.^{14, 15} The antigen conjugated with C3d can be either a protein or a polysaccharide. This adjuvant effect of C3d has been employed extensively in vaccine design and the engineered attachment of two or three C3d molecules per antigen has been shown to be optimal for amplifying humoral response.^{3, 16} Conjugate vaccines using Ti polysaccharides and C3d have been shown to be as effective as Ti polysaccharides covalently linked to Td protein antigens.²

Conjugation of polysaccharide antigens with C3d has been suggested as a means of harnessing the adjuvant potential of the innate immune system. Published studies have demonstrated that C3d conjugated to PPS14 led to increased immunogenicity and isotype switching from a predominantly (IgM) to an IgG1 response by day 25 following primary immunization. Subsequent immunization with PPS14-C3d caused a booster response and a further increase in the ratio of IgG1 to IgM anti-PPS14;² this boost effect and isotype switching is consistent with our 'co-signal 2' hypothesis that T cells specific for C3d peptides are at least partially responsible for the observations.

In our study, only IFN- gamma was found to be consistently elevated. Whether the variability of other cytokine/chemokine responses is due to the small number of subjects tested, immune state of the subjects or sensitivity of the assay used remains to be determined. The C3d-derived peptides described here exhibit different binding affinities to the HLA molecules. It may be that some haplotypes are more responsive to C3d stimulation than others. The immune state of the individual may also influence the observed responses. Differences in the cytokine profiles of memory and effector helper T cells have been described previously.¹⁷ The basal frequency of C3d autoreactive memory T cells would be expected to be low in most individuals and be tightly regulated. An individual under challenge with a bacterial antigen may have higher levels of C3d autoreactive effector T cells and those cells may display a different cyokine/chemokine profile upon stimulation in vitro, than memory cells. Finally, the assays and methods used to detect a given cytokine of chemokine may not be the best for the measurement of others due to differences in the rates of transcription, translation, secretion, stability and re-uptake.

In this study we have shown that predicted T-cell epitopes derived from the complement fragment C3d, bind to a multiple HLA-DR alleles and stimulate autoreactive helper T cells to produce IFN- gamma . These data are consistent with existence of a CD21-independent pathway by which peripheral helper T cells are activated by self-peptide–MHCII complexes.

The identification of 223–246 as the most active C3d peptide in the stimulation of PBMCs deserves additional comment, since the location of the 13-residue amino segment of 223–246 overlaps with the carboxyl segment of P28, the 28-mer peptide containing the major binding site of C3d for CD21.¹⁸ When conjugated to protein antigens, P28 has been shown to enhance both humoral (antibody isotype switching) and cellular (IFN- gamma , IL-4 secretion) immune responses almost as well as the whole C3d molecule¹⁹ which is in agreement with our hypothesis and findings.

Finally, in addition to elucidating a possible adjuvant mechanism of C3d, this data raises questions relevant to self-tolerance and suggests the possibility of AIRE-independent differentiation into nonpathogenic, self-reactive T cells.

Top

Methods

Immunoinformatics

The C3d protein was screened for potential immunogenicity using previously published EpiMatrix System.²⁰ Briefly, the 302-amino-acid sequence was parsed into overlapping 9-mer frames where each frame overlaps the last by eight amino acids. Each frame was then scored for predicted binding to each of eight common class II HLA alleles (DRB1^*0101, DRB1^*0301, DRB1^*0401, DRB1^*0701, DRB1^*0801, DRB1^*1101, DRB1^*1301 and DRB1^*1501). Due to their prevalence and their difference from each other, these eight alleles cover around 97% of human populations worldwide.²¹ EpiMatrix raw binding score predicted for each 9-mer sequence was normalized with respect to a distribution of scores derived from a very large set (N>10 000) of randomly generated 9-mer sequences. This results in a 'Z' score for each analyzed 9-mer. The Z score determines the position of a 9-mer relative to the distribution of all binding scores generated for the random 9-mer sequences. Any peptide scoring above 1.64 on the EpiMatrix 'Z' scale (approximately the top 5% of the random peptide set) has a significant chance of binding to the MHC molecule for which it was predicted. Peptides scoring above 2.32 on the scale (the top 1%) are extremely likely to bind; most published T-cell epitopes fall within this range of scores. Therefore, the higher the Z score, the higher is the probability that a peptide will be presented to T cells by APCs. Previous studies have demonstrated that EpiMatrix accurately predicts published HLA ligands.^{22, 23} EpiMatrix has been shown to predict HLA-DR epitopes from a therapeutic protein and demonstrated that the reactivity to those epitopes correlated with the strength of the antitherapeutic antibody response.²⁴ Potential immunogenicity is not randomly distributed throughout protein sequences but instead tends to 'cluster' in immunogenic regions. ClustiMer, an ancillary algorithm used with EpiMatrix, maps MHC motif matches along the length of a protein and calculates the density of motifs for several HLA. Typical T-cell epitope 'clusters' range from 9 to roughly 25 amino acids in length and, considering their affinity to multiple alleles and across multiple frames, can contain anywhere from 4 to 40 binding motifs.

Peptide synthesis

Peptides were synthesized (New England Peptide, Gardner, MA, USA) by 9-fluoronylmethoxy-carbonyl (Fmoc) synthesis using an automated Rainen Symphony-Protein Technologies synthesizer (to a purity of 90% by HPLC). The chosen peptides (comprised of amino acids 11–26, 45–65, 62–80, 223–246, 248–265 and 269–286) were based on EpiMatrix analysis described above.

HLA binding assays

Class II HLA binding assay was performed as initially described by Kwok and colleagues²⁵ and adapted for high throughput by EpiVax. Nonbiotinylated test peptide over a wide range of concentrations (0.001–400 M) competes for binding to purified class II molecules (50 nM) against a biotinylated standard peptide at a fixed concentration (0.1 M) for 24 h in 96-well plates at 37 °C. Class II molecules are then captured on ELISA plates using pan anti-class II antibodies (L243, anti-HLA-DR) developed by addition of streptavidin-europium and read on a time-resolved fluorescence (TRF) plate reader. Nonlinear regression analysis is performed and an IC50 value is calculated. Binding assays were performed for four HLA alleles: DRB1^*0101, DRB1^*0401, DRB1^*0701 and DRB1^*1501, which provide a broad representation of HLA class II allele binding pockets.²¹

T-cell assay

PBMCs from four healthy human subjects were isolated by ficoll separation. The purified PBMCs were stimulated with dimethyl sulfoxide (DMSO; negative solvent control) or with a pool of six C3d-derived peptides at a final total peptide concentration of 10 g ml^-1 in RPMI media supplemented with 20% human serum and incubated for 7–10 days at 37 °C, 5% CO₂. PBMCs were then re-stimulated with the appropriate C3d peptide, an irrelevant peptide, no peptide or PHA. Cryopreserved PBMCs (Cellular Technology Limited, Shaker Heights, OH, USA) from an HLA-DRB1^*0101 subject were initially stimulated with no peptide (DMSO negative control) or a pool of six C3d peptides for 7 days. Cells were then re-stimulated with DMSO or a pool of six C3d peptides for 3 days.

Intracelluar cytokine staining

Cryopreserved PBMCs previously re-stimulated with C3d-derived peptides as described above were re-stimulated again for 4 h with GolgiPlug (BD Biosciences, San Jose, CA, USA) and the appropriate peptide (C3d pool of six, Flu HA_306–318 or DMSO solvent control). A total of 1 $times$ 10⁶ PBMCs were stained with fluorescently labeled antibodies to surface proteins CD4 and CD8 (eBioscience, San Diego, CA, USA and BD Biosciences) for 30 min on ice in Flow Staining Buffer (eBioscience) and washed twice with buffer. Following cell surface staining, cells were fixed and permeabilized (eBioscience) and stained intracellularly for IFN- gamma (eBioscience) following the manufacturer's protocol. Cells were run on a FACSCalibur (BD Biosciences) and data was analyzed using the FlowJo software (Treestar).

IFN- ELISA analysis

Supernatants from stimulated cells were evaluated 3 days after re-stimulation by human IFN- gamma ELISA (R&D Systems, Minneapolis, MN, USA).

Multiplex cytokine ELISA analysis

Supernatants from stimulated cells were assayed for levels of multiple cytokines and chemokines by the SearchLight Multiplex ELISA contract service (Pierce Biotechnology, Woburn, MA, USA).

Top

References

Dempsey PW, Allison ME, Akkaraju S, Goodnow CC, Fearon DT. C3d of complement as a molecular adjuvant: Bridging innate and acquired immunity. Science 1996; 271: 348–350. | Article | PubMed | ISI | ChemPort |
Haas KM, Toapanta FR, Oliver JA, Poe JC, Weis JH, Karp DR et al. Cutting edge: C3d functions as a molecular adjuvant in the absence of CD21/35 expression. J Immunol 2004; 172: 5833–5837. | PubMed | ChemPort |
Bower JF, Ross TM. A minimum CR2 binding domain of C3d enhances immunity following vaccination. Adv Exp Med Biol 2006; 586: 249–264. | PubMed |
Jakobsen H, Jonsdottir I. Muscosal vaccination against encapsulated bacteria—new potentials for conjugate vaccines? Scand J Immunol 2003; 58: 119–128. | Article | PubMed | ChemPort |
Test ST, Mitsuyoshi J, Connolly CC, Lucas AH. Increased immunogenicity and induction of class switching by conjugation of complement C3d to pneumococcal serotype 14 capsular polysaccharide. Infect Immun 2001; 69: 3031–3040. | Article | PubMed | ChemPort |
De Groot AS, Goldberg M, Moise L, Martin W. Evolutionary deimmunization: an ancillary mechanism for self-tolerance? Cell Immunol 2006; 244: 148–153. | Article | PubMed | ChemPort |
De Groot AS. Immunomics: discovering new targets for vaccines and therapeutics. Drug Discov Today 2006; 11: 203–209. | Article | PubMed | ChemPort |
de Bruijn MH, Fey GH. Human complement C3: cDNA coding sequence and derived primary structure. Proc Natl Acad Sci USA 1985; 82: 708–712. | Article | PubMed | ChemPort |
De Groot AS, Bosma A, Chinai N, Frost J, Jesdale BM, Gonzalez MA et al. From genome to vaccine: in silico predictions ex vivo verification. Vaccine 2001; 19: 4385–4395. | Article | PubMed | ChemPort |
Fearon DT, Carroll MC. Regulation of B lymphocyte responses to foreign and self-antigens by the CD19/CD21 complex. Annu Rev Immunol 2000; 18: 393–422. | Article | PubMed | ISI | ChemPort |
Carroll MC. The complement system in B cell regulation. Mol Immunol 2001; 41: 141–146. | Article | ChemPort |
Luxembourg AT, Cooper NR. Modulation of signaling via the B cell antigen receptor by CD21, the receptor for C3dg and EBV. J Immunol 1994; 153: 4448–4457. | PubMed | ChemPort |
Fearon DT, Carroll MC. Regulation of B lymphocyte responses to foreign and self-antigens by the CD19/CD21 complex. Annu Rev Immunol 2000; 18: 393–422. | Article | PubMed | ISI | ChemPort |
Mongini PK, Vilensky MA, Highet PF, Inman JK. The affinity threshold for human B cell activation is reduced upon co-ligation of the antigen receptor with CD21 (CR2). J Immunol 1997; 159: 3782–3791. | PubMed | ISI | ChemPort |
Kozono Y, Abe R, Kozono H, Kelly RG, Azuma T, Holers VM. Cross-linking CD21/CD35 or CD19 increases both B7-1 and B7-2 expression on murine splenic B cells. J Immunol 1998; 160: 1565–1572. | PubMed | ISI | ChemPort |
Cherukuri A, Cheng PC, Pierce SK. The role of the CD19/CD21 complex in B cell processing and presentation of complement-tagged antigens. J Immunol 2001; 167: 163–172. | PubMed | ISI | ChemPort |
Wu CY, Kirman JR, Rotte MJ, Davey DF, Perfetto SP, Rhee EG et al. Distinct lineages of T(H)1 cells have different capacities for memory cell generation in vivo. Nat Immunol 2001; 3: 852–858. | Article | ChemPort |
Lambris JD, Ganu VS, Hirani S, Muller-Eberhard HJ. Mapping of the C3d receptor (CR2)-binding site and a neoantigenic site in the C3d domain of the third component of complement. Proc Natl Acad Sci USA 1985; 82: 4235–4239. | Article | PubMed | ChemPort |
Bower JF, Ross TM. A minimum CR2 binding domain of C3d enhances immunity following vaccination. Adv Exp Med Biol 2006; 586: 249–264. | PubMed |
De Groot AS, Jesdale BM, Szu E, Schafer JR. An interactive web site providing MHC ligand predictions: application to HIV research. AIDS Res Hum Retroviruses 1997; 13: 539–541.
Southwood S, Sidney J, Kondo A, del Guercio MF, Appella E, Hoffman S et al. Several common HLA-DR types share largely overlapping peptide binding repertoires. J Immunol 1998; 160: 3363–3373. | PubMed | ISI | ChemPort |
Roberts CG, Meister GE, Jesdale BM, Lieberman J, Berzofsky JA, De Groot AS. Prediction of HIV peptide epitopes by a novel algorithm. AIDS Res Hum Retroviruses 1996; 12: 593–610. | PubMed | ChemPort |
De Groot AS, Sbai H, Jesdale B, Martin W, Saint-Aubin C, Bosma A et al. Mapping cross-clade HIV-1 vaccine epitopes using a bioinformatics approach. Vaccine 2003; 21: 4486–4504. | Article | PubMed | ChemPort |
Koren E, De Groot AS, Jawa V, Beck KD, Boone T, Rivera DS et al. Clinical validation of the 'in-silico' prediction of immunogenicity of a human recombinant therapeutic protein. Clin Immunol 2007; 124: 26–32. | Article | PubMed | ChemPort |
Steere AC, Klitz W, Drouin EE, Falk BA, Kwok WW, Nepom GT et al. Antibiotic-refractory Lyme arthritis is associated with HLA-DR molecules that bind a Borrelia burgdorferi peptide. J Exp Med 2006; 203: 961–971. | Article | PubMed | ChemPort |

Top

Acknowledgements

The initial phases of the work were supported by a grant from EpiVax Inc. to PMK at Brown University. SHH received a summer UTRA fellowship from Brown University and is an undergraduate student at Brown University. All of the other authors are employed by EpiVax Inc. PMK is the Charles A & Helen B Stuart Professor Emeritus of Medical Science, Department of Molecular Microbiology & Immunology at Brown University. DSR is a postgraduate student at the University of Edinburgh, College of Medicine and Veterinary Medicine, UK. ADG is an Adjunct Professor, Brown University School of Medicine. We thank Claire Rodríguez and Christine Malboeuf for their work on this project.

Outstanding Observation