Role of receptor-binding activity of the viral hemagglutinin molecule in the presentation of influenza virus antigens to helper T cells.

The concentration of antigen required to stimulate influenza virus-specific helper T cells was observed to be dependent upon the antigenic form bearing the relevant determinant: intact, nonreplicative virus was needed only in picomolar amounts, while denatured proteins, protein fragments, or synthetic peptides were required in micromolar concentrations for a threshold level of stimulation. Antigenic efficiency of intact virus was found to result from the attachment of virus to sialic acid residues on the surface of the antigen-presenting cell since spikeless viral particles lacking the hemagglutinin molecule were much less efficient antigens for helper T cells and continuous presence of hemagglutination-inhibiting antihemagglutinin antibodies reduced efficiency of stimulation by intact virus approximately 100-fold for both hemagglutinin and internal virion proteins. Influenza virus associated rapidly with antigen-presenting cells; less than 10 min at 20 degrees C was sufficient to introduce virus for a maximal level of T-cell stimulation. This rapid attachment was blocked by antibodies to the hemagglutinin or by pretreatment of the antigen-presenting cells with neuraminidase to remove the cellular virus receptor. Following viral adsorption by antigen-presenting cells, a lag period of 30 min at 37 degrees C was required for the expression of helper T-cell determinants. One early event identified was the movement of the virus to a neuraminidase-insensitive compartment, which can occur at 10 degrees C, but which was not equivalent to expression of helper T-cell determinants. Preincubation of cells with virus at 10 degrees C for 4 h reduced the lag period of helper T-cell determinant expression to 15 min when these cells were shifted to 37 degrees C, suggesting that transition of the virus to a neuraminidase-resistant state is a required step in presentation of T-cell antigenic determinants.     

Membrane interactions of mutated forms of the influenza fusion peptide

We have studied a group of fusion peptides of influenza hemagglutinin in which the N-terminal amino acid, Gly (found in the wild-type peptide), has been systematically substituted with Ala, Ser, Val, or Glu. The activity of the intact hemagglutinin protein with these same substitutions has already been reported. As a measure of the extent of modulation of intrinsic membrane curvature by these peptides, we determined their effects on the polymorphic phase transition of dipalmitoleoylphosphatidylethanolamine. The wild-type peptide is the only one that, at pH 5, can substantially decrease the temperature of this transition. This is also the only form in which the intact protein promotes contents mixing in cells. The Ala and Ser mutant hemagglutinins exhibit a hemifusion phenotype, and their fusion peptides have little effect on lipid polymorphism at low pH. The two mutant proteins that are completely fusion inactive are the Val and Glu mutant hemagglutinins. The fusion peptides from these forms significantly increase the polymorphic phase transition temperature at low pH. We find that the effect of the fusion peptides on membrane curvature, as monitored by a shift in the temperature of this polymorphic phase transition, correlates better with the fusogenic activities of the corresponding protein than do measurements of the isotropic (31)P NMR signals or the ability to induce the fusion of liposomes. The inactivity of the hemagglutinin protein with the hydrophobic Val mutation can be explained by the change in the angle of membrane insertion of the helical fusion peptide as measured by polarized FTIR. Thus, the nature of the interactions of the fusion peptides with membranes can, in large part, explain the differences in the fusogenic activity of the intact protein.     

Removal of carbohydrate from influenza A virus and its hemagglutinin and the effect on biological activities.

Treatment of influenza virus and its purified hemagglutining with glycosidases from Diplococcus pneumoniae, which included beta-galactosidase, beta-N-acetylglucosminidase, and endoglycosidase D, released amino and neutral sugars from the virus and these as well as large oligosaccharides from the purified hemagglutinin. The released glucosamine-containing oligosaccharides were of one discrete size. Large oligosaccharides not removed by the glycosidases were found on the virus as well as the hemagglutinin. Some oligosaccharides on the virus were inaccessible to the enzymes, since they could be removed only from the purified hemagglutinin. Approximately 50% of the hemagglutinin carbohydrates could be removed without effect on hemagglutinating activity. Similarly, removal of 20 to 25% of the carbohydrates from intact virus particles did not alter infectivity.     

A synthetic decapeptide of influenza virus hemagglutinin elicits helper T cells with the same fine recognition specificities as occur in response to whole virus

The immunogenicity of an isolated murine helper T cell determinant was studied. Mice were immunized with a synthetic peptide corresponding to amino acid residues 111-120 of the influenza PR8 hemagglutinin (HA) heavy chain, a region previously identified as a major target of the helper T cell response to the HA molecule in virus-primed BALB/c mice. Lymph node T cells from these mice were fused with BW 5147 cells to produce T hybrids for clonal analysis of their recognition specificities. Three T cell hybridoma clones, obtained from two different mice, responded to the immunizing peptide when presented by syngeneic antigen-presenting cells. All of these clones responded also to antigen provided as intact wild-type PR8 virus. The fine specificity of the peptide-induced T cell hybridomas, in response to a panel of mutant and variant influenza viruses, was indistinguishable from the fine specificities of T cells to the corresponding region of the HA1 chain of the HA molecule which had been generated by priming of mice with intact wild-type virus. These results suggest that an immunogenic determinant is contained within the 111-120 sequence that is able to elicit anti-influenza virus T cells with a similar repertoire to those elicited by immunization with whole virus.     

Recognition of a sequestered self peptide by influenza virus-specific CD8+ cytolytic T lymphocytes

The Ag receptors on CD8+ CTL recognize foreign antigenic peptides associated with cell surface MHC class I molecules. Peptides derived from self proteins are also normally presented by MHC class I molecules. Here we report that an H-2Kd-restricted murine CD8+ CTL clone directed to an influenza hemagglutinin epitope can recognize a peptide derived from the murine mitochondrial aconitase enzyme in association with H-2Kd molecules. Surprisingly, this self peptide is not normally displayed on the cell surface associated with the restricting MHC class I molecule. Several lines of evidence suggest that this self peptide, although requiring association with the Kd molecule for CTL recognition, is not associated with this or other MHC class I allele under physiologic conditions in intact cells. Rather, it is sequestered in the cytoplasm associated with a carrier protein and is released only upon cell disruption. These results suggest a means of restricting the entry of self peptide into the class I pathway. In addition, this finding raises the possibility that self peptides sequestered within the cell can, after release from damaged cells, interact with MHC class I molecules on bystander cells and trigger autoimmune injury by virus-specific CTLs during viral infection.     

Disruption of Myxoviruses with Tween 20 and Isolation of Biologically Active Hemagglutinin and Neuraminidase Subunits

Myxoviruses were disrupted with Tween 20 at high pH, and the major surface antigens were separated in biologically active form. The neuraminidase had a sedimentation coefficient of 10.8S, and the hemagglutinin had a sedimentation coefficient of 8.1S. Electron microscopic examination of negatively stained preparations revealed structures identical in size and morphology to the neuraminidase and hemagglutinin subunits described by others. Inhibition of neuraminidase activity by antibody to the hemagglutinin which occurred with intact viruses (probably for "steric" reasons) did not occur after the viruses were disrupted with Tween 20. Serological assays for neuraminidase were possible in the presence of the mild surfactant, whereas serological assays for hemagglutinin were possible after removal of the reagent. Disruption of myxoviruses with Tween 20 therefore provides a method for the independent study of these antigens during antigenic drift.     

Immunogenicity of synthetic peptides corresponding to flexible and antibody-accessible segments of mouse lactate dehydrogenase (LDH)-C4

The immunological properties of a panel of synthetic peptides that represent the most accessible and mobile segments of the lactate dehydrogenase (LDH)-C4 molecule were characterized. Peptides corresponding to mouse LDH-C4 amino acid sequences: 1-14b, 5-15, 49-58, 97-110, 211-220, 231-243, 274-286, 304-316, and 318-330 were synthesized and compared in terms of binding antibodies raised in rabbits against the intact protein. Six of these sequences were covalently coupled to diphtheria toxoid and used to immunize groups of rabbits. LDH-C4-specific antibodies were detectable in immune sera by enzyme-linked immunosorbent assay, Western blotting, and immunoprecipitation assays. The immunogenicity of the mouse LDH-C4 peptides in rabbits could be ranked in the following order: 5-15, 304-316 greater than 211-220, 274-286 greater than 49-58, 97-110. The immunological properties of these short synthetic peptides did not correlate with features of the mouse LDH-C4 structure except that the most active sequences appeared to be those that differed from the somatic isozymes to the greatest extent. These results have direct bearing on the selection of immunogenic LDH-C4 peptides for contraceptive vaccine studies in humans and non-human model systems.     

The structure of an antigenic determinant in a protein

The immunogenic and antigenic determinants of a synthetic peptide and the corresponding antigenic determinants in the parent protein have been elucidated. Four determinants have been defined by reactivity of a large panel of antipeptide monoclonal antibodies with short, overlapping peptides (7-28 amino acids), the immunizing peptide (36 amino acids), and the intact parent protein (the influenza virus hemagglutinin, HA). The majority of the antipeptide antibodies that also react strongly with the intact protein recognize one specific nine amino acid sequence. This immunodominant peptide determinant is located in the subunit interface in the HA trimeric structure. The relative inaccessibility of this site implies that antibody binding to the protein is to a more unfolded HA conformation. This antigenic determinant differs from those previously described for the hemagglutinin and clearly demonstrates the ability of synthetic peptides to generate antibodies that interact with regions of the protein not immunogenic or generally accessible when the protein is the immunogen.     

Structural characterization of the N-terminal autoregulatory sequence of phenylalanine hydroxylase

Phenylalanine hydroxylase (PAH) is activated by its substrate phenylalanine, and through phosphorylation by cAMP-dependent protein kinase at Ser16 in the N-terminal autoregulatory sequence of the enzyme. The crystal structures of phosphorylated and unphosphorylated forms of the enzyme showed that, in the absence of phenylalanine, in both cases the N-terminal 18 residues including the phosphorylation site contained no interpretable electron density. We used nuclear magnetic resonance (NMR) spectroscopy to characterize this N-terminal region of the molecule in different stages of the regulatory pathway. A number of sharp resonances are observed in PAH with an intact N-terminal region, but no sharp resonances are present in a truncation mutant lacking the N-terminal 29 residues. The N-terminal sequence therefore represents a mobile flexible region of the molecule. The resonances become weaker after the addition of phenylalanine, indicating a loss of mobility. The peptides corresponding to residues 2-20 of PAH have different structural characteristics in the phosphorylated and unphosphorylated forms, with the former showing increased secondary structure. Our results support the model whereby upon phenylalanine binding, the mobile N-terminal 18 residues of PAH associate with the folded core of the molecule; phosphorylation may facilitate this interaction.     

Modification of the N-terminus of membrane fusion-active peptides blocks the fusion activity

The amphiphilic anionic peptides E5 and E5L can mimic the fusogenic activity of influenza hemagglutinin(HA). These peptides induced fusion of egg yolk phosphatidylcholine small or large unilamellar vesicles only at acidic pH in a similar manner to viral HA. Acetylation or acetimidylation of the N-terminus of the peptides drastically reduced the fusion activity of the intact peptides, while C-terminal amidation left the activity unchanged. The binding assay suggested that the interaction of the modified peptides with lipid membranes was almost unchanged in comparison with those of the parent peptides, and the CD spectra showed that these peptides were alpha-helical. The results showed the importance of the N-terminus of the peptides on the membrane fusion activity, although why the N-terminal modifications affect the activity is still unclear.     

Immune responses to the 18-kDa protein of Mycobacterium leprae. Similar B cell epitopes but different T cell epitopes seen by inbred strains of mice.

Antibody responses to the 18-kDa protein of Mycobacterium leprae have been analyzed in different strains of mice. High, intermediate, and low responder strains have been identified and these response patterns show clear linkage to genes encoded in the H-2 complex. Three peptides, residues 1-50, 51-100, and 101-148 have been synthesized, as well as a series of 20-mer peptides, which span the entire 18-kDa protein. Repeated immunization of different strains of mice with the 18-kDa protein resulted in IgG responses to epitopes found on all three synthetic peptides. Immunization of BALB/cJ and B10.BR mice, two high responder strains, with 18-kDa protein resulted in high levels of IgG antibody to epitopes found on peptides 1-20, 16-35, 31-50, 46-65, and 76-95. B10.BR mice also contained IgG that bound peptide 61-80 and BALB/cJ mice produced IgG that bound peptide 91-110. Although B10.BR mice produced IgG that bound the 50-mer peptide 101-148, this IgG was not detected by binding to peptides 91-110, 106-125, 121-140, and 131-148. Immunization of B10.BR mice with individual overlapping 20-mer peptides as Ag revealed that peptides 1-20, 16-35, 31-50, and 76-95 elicited high titers of IgG that bound both the immunizing peptide as well as 18-kDa protein. As these peptides induce antibody synthesis they must contain both B cell and T cell epitopes. By contrast, immunization of BALB/cJ mice with the same 20-mer peptides, all of which contain B cell epitopes for this strain, failed to elicit IgG responses with one exception. Peptide 91-110 induced IgG that bound peptide 91-110, but not the intact 18-kDa protein. We conclude that peptides 1-20, 16-35, 31-50, and 76-95 either lack T cell epitopes for BALB/cJ mice, or activate different T cell subpopulations in the two strains. We suggest that the induction of IgG responses to small peptide Ag is an in vivo assay of the activity of Th2 cell subpopulations.     

T cell responses to cleaved rabies virus glycoprotein and to synthetic peptides

The antigenic structure of the rabies virus glycoprotein has been studied. A limited number of fragments were obtained by cyanogen bromide (CNBr) cleavage of viral glycoprotein, and eight large peptides were isolated by using sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. These were tested for their capacity to stimulate the proliferation of nylon wool-purified T cells obtained from spleens of rabies-immune A/J mice. Three peptides (Cr1, Cr2 plus Cr2A, and Cr3) stimulated antigen-specific proliferation, indicating that at least three T cell determinants of the native molecule are sequential or continuous in nature. Stimulation was also obtained with 27-residue and 13-residue synthetic peptides (designated R21 and R20, respectively) that included sequences towards the carboxy terminal end of Cr1, but not with synthetic peptides that included sequences of Cr2 and Cr3 (which are both glycosylated in virus-derived material). The intact viral glycoprotein and synthetic peptide R21 stimulated T lymphocytes with surface characteristics of helper cells, and induced the production of interleukin 2 by these lymphocytes. Synthetic peptides R20 and R21 also stimulated a minor population of Lyt-2-positive cells, which were not yet identified as either suppressor or cytotoxic T lymphocytes.     

Structural analysis of peptides capable of binding to more than one Ia antigen

The Ia binding regions were analyzed for three unrelated peptide Ag (sperm whale myoglobin 106-118, influenza hemagglutinin 130-142, and lambda repressor protein 12-26) for which binding to more than one Ia molecule has previously been demonstrated. By determining the binding profile of three separate series of truncated synthetic peptides, it was found that in all three cases the different Ia reactivities mapped to largely overlapping regions of the peptides; although, for two of the peptides, the regions involved in binding the different Ia specificities were distinct. Moreover, subtle differences were found to dramatically influence some, but not other, Ia reactivities. Using a large panel of synthetic peptides it was found that a significant correlation exists between the capacity of peptides to interact with different alleles of the same molecule (i.e., IAd and IAk), but no correlation was found with the capacity of peptides to interact with different isotypes within the same haplotype (i.e., IAd and IEd). These data suggest that different alleles of the same MHC molecule may actually recognize closely related structures, whereas different isotypes may recognize unrelated, albeit non-mutually exclusive, structures on an Ag molecule.     

Prediction of T Cell Epitopes from Leishmania major Potentially Excreted/Secreted Proteins Inducing Granzyme B Production

Leishmania-specific cytotoxic T cell response is part of the acquired immune response developed against the parasite and contributes to resistance to reinfection. Herein, we have used an immune-informatic approach for the identification, among Leishmania major potentially excreted/secreted proteins previously described, those generating peptides that could be targeted by the cytotoxic immune response. Seventy-eight nonameric peptides that are predicted to be loaded by HLA-A*0201 molecule were generated and their binding capacity to HLA-A2 was evaluated. These peptides were grouped into 20 pools and their immunogenicity was evaluated by in vitro stimulation of peripheral blood mononuclear cells from HLA-A2⁺-immune individuals with a history of zoonotic cutaneous leishmaniasis. Six peptides were identified according to their ability to elicit production of granzyme B. Furthermore, among these peptides 3 showed highest affinity to HLA-A*0201, one derived from an elongation factor 1-alpha and two from an unknown protein. These proteins could constitute potential vaccine candidates against leishmaniasis.     

Molecular-mass heterogeneity of Griffonia simplicifolia lectin IV subunits. Differences in the oligosaccharide moieties in the N-terminal region.

Lectin IV of Griffonia simplicifolia (Mr approximately 56,000), which has a strong affinity for both the Lewis b and Y blood-group determinants, is a dimeric protein of two subunits, alpha (29 kDa) and beta (27 kDa), separable by SDS/PAGE and containing covalently linked oligosaccharide. After digestion with N-glycanase, the protein migrates as a single band with a mobility identical with that of the beta-subunit. After cleavage with hydroxylamine of 3H-labelled, but otherwise intact, lectin, the radioactively labelled oligosaccharide was found to be associated with two blocked N-terminal peptides separable by h.p.l.c. and having identical amino acid compositions. One of these had three or four glucosamine residues per molecule, whereas the other had only one or two. Sequence analyses of these, as well as of a 21 kDa hydroxylamine-cleaved fragment and of the intact lectin pretreated with pyroglutamate aminopeptidase, have provided a unique sequence for residues 1-62 of the two subunits. Evidence is presented for two sites of N-linked oligosaccharide attachment at Asn-5 and Asn-18. Whereas the alpha-subunit has oligosaccharide linked to both sites, the beta-subunit has carbohydrate associated with only one (Asn-18). Sugar analyses of the whole lectin reveal a monosaccharide composition of (Xyl)3(Fuc)2(Man)10(GlcNAc)6, representing 6.4% of the mass of the molecule. Taken together with the susceptibility of the Asn-5 linkage (but not of Asn-18) to N-glycanase digestion, the observations indicate that the structures of the oligosaccharides at residues 5 and 18 are different.     

Mechanism of antigenic drift in influenza virus. Amino acid sequence changes in an antigenically active region of Hong Kong (H3N2) influenza virus hemagglutinin

During antigenic drift in influenza viruses, changes in antigenicity are associated with changes in amino acid sequence of the large hemagglutinin polypeptide, HA1. In ten variants of Hong Kong (H3N2) influenza virus selected with monoclonal antibodies, the proline residue at position 143 in HA1 changed to serine, threonine, leucine or histidine. In other variants, asparagine 133 changed to lysine, glycine 144 to aspartic acid and serine 145 to lysine. All these changes are possible by single base changes in the RNA except the last, which requires a double base change. Residues 142 to 146 also changed in field strains of Hong Kong influenza isolated between 1968 and 1977 (Laver et al., 1980). The single amino acid sequence changes in HA1 of the monoclonal variants were detected by comparing the compositions of the soluble tryptic peptides from the variants with the known sequences of these peptides from wild-type virus. Two insoluble tryptic peptides, comprising residues 110 to 140 and 230 to 255 in the HA1 molecule, were not examined and we do not know if additional changes occurred in these regions.In order to determine whether sequential changes at the same position occurred during antigenic drift, antibody prepared against the new antigenic site on the variants in which proline 143 changed to histidine or threonine was used to select second generation variants of these variants. In the first case, the glycine residue (144) next to the histidine changed to aspartic acid, and in the second, the threonine residue at position 143 reverted to proline and the virus regained the antigenicity of wild-type.Although monoclonal antibodies revealed dramatic antigenic differences between the variants and wild-type virus, only those variants with changes at position 144 of glycine to aspartic acid or at position 145 of serine to lysine could be distinguished from wild-type virus using heterogeneous rabbit or ferret antisera. The other variants, including those which showed sequence changes in widely separated positions of HA1, could not be distinguished from wild-type with heterogeneous antisera.These findings suggest that sequence changes in the region comprising residues 142 to 146 of HA1 affect an important antigenic site on the hemagglutinin molecule, but how these changes affect the antigenic properties, or whether this region actually forms part of the antigenic site is not known.     

Preclinical and clinical development of plant-made virus-like particle vaccine against avian H5N1 influenza

The recent swine H1N1 influenza outbreak demonstrated that egg-based vaccine manufacturing has an Achille's heel: its inability to provide a large number of doses quickly. Using a novel manufacturing platform based on transient expression of influenza surface glycoproteins in Nicotiana benthamiana, we have recently demonstrated that a candidate Virus-Like Particle (VLP) vaccine can be generated within 3 weeks of release of sequence information. Herein we report that alum-adjuvanted plant-made VLPs containing the hemagglutinin (HA) protein of H5N1 influenza (A/Indonesia/5/05) can induce cross-reactive antibodies in ferrets. Even low doses of this vaccine prevented pathology and reduced viral loads following heterotypic lethal challenge. We further report on safety and immunogenicity from a Phase I clinical study of the plant-made H5 VLP vaccine in healthy adults 18-60 years of age who received 2 doses 21 days apart of 5, 10 or 20 μg of alum-adjuvanted H5 VLP vaccine or placebo (alum). The vaccine was well tolerated at all doses. Adverse events (AE) were mild-to-moderate and self-limited. Pain at the injection site was the most frequent AE, reported in 70% of vaccinated subjects versus 50% of the placebo recipients. No allergic reactions were reported and the plant-made vaccine did not significantly increase the level of naturally occurring serum antibodies to plant-specific sugar moieties. The immunogenicity of the H5 VLP vaccine was evaluated by Hemagglutination-Inhibition (HI), Single Radial Hemolysis (SRH) and MicroNeutralisation (MN). Results from these three assays were highly correlated and showed similar trends across doses. There was a clear dose-response in all measures of immunogenicity and almost 96% of those in the higher dose groups (2 × 10 or 20 μg) mounted detectable MN responses. Evidence of striking cross-protection in ferrets combined with a good safety profile and promising immunogenicity in humans suggest that plant-based VLP vaccines should be further evaluated for use in pre-pandemic or pandemic situations. TRIAL REGISTRATION: ClinicalTrials.gov NCT00984945.     

Direct role of viral hemagglutinin in B-cell mitogenesis by influenza viruses

The mitogenic activity of influenza virus is a function of the hemagglutinin (HA) molecule. Purified HA is mitogenic for murine B lymphocytes but not T lymphocytes. Furthermore, like the intact virus, HA of the H2 (but not H3) subtype is mitogenic only for B cells expressing the class II major histocompatibility complex glycoprotein I-E. Since virus bearing uncleaved HA is as mitogenic as virus bearing cleaved HA, the membrane fusion activity of the HA molecule is not involved.     

A hypoallergenic hybrid molecule with increased immunogenicity consisting of derivatives of the major grass pollen allergens, Phl p 2 and Phl p 6

Allergen-specific immunotherapy is currently based on the administration of allergen extracts containing natural allergens. However, its broad application is limited by the poor quality of these extracts. Based on recombinant allergens, well-defined allergy vaccines for allergen-specific immunotherapy can be produced. Furthermore, they can be modified to reduce their allergenic activity and to avoid IgE-mediated side effects. Here, we demonstrate that the immunogenicity of two grass pollen-derived hypoallergenic allergen derivatives could be increased by engineering them as a single hybrid molecule. We used a hypoallergenic Phl p 2 mosaic, generated by fragmentation of the Phl p 2 sequence and reassembly of the resulting peptides in an altered order, and a truncated Phl p 6 allergen, to produce a hybrid protein. The hybrid retained the reduction of IgE reactivity and allergenic activity of its components as shown by ELISA and basophil activation assays. Immunization with the hybrid molecule demonstrated the increased immunogenicity of this molecule, leading to higher levels of allergen-specific IgG antibodies compared to the single components. These antibodies could inhibit patients' IgE binding to the wild-type allergens. Thus, the described strategy allows the development of safer and more efficacious vaccines for the treatment of grass pollen allergy.     

POPI: predicting immunogenicity of MHC class I binding peptides by mining informative physicochemical properties

MOTIVATION: Both modeling of antigen-processing pathway including major histocompatibility complex (MHC) binding and immunogenicity prediction of those MHC-binding peptides are essential to develop a computer-aided system of peptide-based vaccine design that is one goal of immunoinformatics. Numerous studies have dealt with modeling the immunogenic pathway but not the intractable problem of immunogenicity prediction due to complex effects of many intrinsic and extrinsic factors. Moderate affinity of the MHC-peptide complex is essential to induce immune responses, but the relationship between the affinity and peptide immunogenicity is too weak to use for predicting immunogenicity. This study focuses on mining informative physicochemical properties from known experimental immunogenicity data to understand immune responses and predict immunogenicity of MHC-binding peptides accurately. RESULTS: This study proposes a computational method to mine a feature set of informative physicochemical properties from MHC class I binding peptides to design a support vector machine (SVM) based system (named POPI) for the prediction of peptide immunogenicity. High performance of POPI arises mainly from an inheritable bi-objective genetic algorithm, which aims to automatically determine the best number m out of 531 physicochemical properties, identify these m properties and tune SVM parameters simultaneously. The dataset consisting of 428 human MHC class I binding peptides belonging to four classes of immunogenicity was established from MHCPEP, a database of MHC-binding peptides (Brusic et al., 1998). POPI, utilizing the m = 23 selected properties, performs well with the accuracy of 64.72% using leave-one-out cross-validation, compared with two sequence alignment-based prediction methods ALIGN (54.91%) and PSI-BLAST (53.23%). POPI is the first computational system for prediction of peptide immunogenicity based on physicochemical properties. AVAILABILITY: A web server for prediction of peptide immunogenicity (POPI) and the used dataset of MHC class I binding peptides (PEPMHCI) are available at http://iclab.life.nctu.edu.tw/POPI