Immunosignatures can predict vaccine efficacy

Abstract

The development of new vaccines would be greatly facilitated by having effective methods to predict vaccine performance. Such methods could also be helpful in monitoring individual vaccine responses to existing vaccines. We have developed "immunosignaturing" as a simple, comprehensive, chip-based method to display the antibody diversity in an individual on peptide arrays. Here we examined whether this technology could be used to develop correlates for predicting vaccine effectiveness. By using a mouse influenza infection, we show that the immunosignaturing of a natural infection can be used to discriminate a protective from nonprotective vaccine. Further, we demonstrate that an immunosignature can determine which mice receiving the same vaccine will survive. Finally, we show that the peptides comprising the correlate signatures of protection can be used to identify possible epitopes in the influenza virus proteome that are correlates of protection.

Keywords: antibody repertoire; epitope prediction; immune profile; peptide microarray; systems vaccinology.

Fig. 1.

Outcomes of immunized mice following lethal challenge with influenza A/PR/8/34. Mice were immunized with PBS solution (mock), a live sublethal dose of A/PR/8/34, inactivated A/PR/8/34, the 2006/2007 seasonal influenza vaccine, or the 2008/2007 seasonal influenza vaccine. Mice were challenged intranasally with 5 × 10⁵ pfu per mouse. The average daily percent starting weight is graphed in A, where the average is calculated based on the surviving mice and error bars represent the SD. Survival curves are presented in B and represent the percentage of mice surviving following challenge.

Fig. 2.

Whole virus-specific IgG measured in an ELISA. Before challenge, serum was collected from all mice. The amount of antigen-specific circulating IgG was measured for inactive PR8 and the 2006/2007 and 2007/2008 seasonal vaccines by endpoint titer and is graphed. Error bars are the SD of triplicate measurements of pooled sera.

Fig. 3.

Comparison of the immunosignature to live and inactive Influenza A/PR/8/34. The first analysis of the CIM10Kv3 array data were to compare the immune response to the live and inactive PR8 vaccines. A scatterplot of the fold change ratio of each vaccine to naive mice is shown in A, where the live PR8 is on the x axis and the inactive PR8 is on the y axis. The overlap between peptides that are significantly different (P < 0.05 with Benjamini and Hochberg Multiple Test Correction) above 1.3 fold in each vaccine are presented in the Venn diagram in B. The variance in the immune responses between individuals is shown in the PCA analysis in C for all peptides and (D) for the overlap, where the first two principal components are plotted and individuals colored by vaccine.

Fig. 4.

The immunosignature can distinguish weaker inactive vaccines from a more potent one. The immunosignature on the CIM10Kv3 was compared between the two seasonal vaccines and the killed PR8 vaccine first in an ANOVA in which 55 peptides at a P < 0.0005 (five false positives) were capable of separating the three vaccines. Variance among individuals is represented in a plot of the first and second principal components in A. Comparisons between vaccinated and naive mice were made by using pooled sera on the CIM10Kv1 using a minimum 1.3-fold increase in normalized fluorescence units in sera from immunized over mock and a P value of less than 0.05 using the Benjamini and Hochberg multiple test correction. Overlap between these lists is shown in the Venn diagram in B.

Fig. 5.

The immunosignature predicts that antibody cross-reactivity to NA195-219 was important in protecting seasonal vaccine recipients from the PR8. The whole-virus ELISA endpoint titers for the 2006/2007 vaccine recipients that survived or died following PR8 challenge are shown in A, where the horizontal line represents the group mean and each point represents an individual mouse. The immunosignature was compared between the seasonal vaccine recipients for both years that survived or died of challenge. A Student t test with a P value of 0.005 identified 94 peptides that were significantly different between the two groups. The variance among all individuals receiving an inactive vaccine is presented in B as a plot of the first and second principal components. The 38 peptides at least 1.3 fold less recognized by those that died were used to predict the epitope in NA (C), where the GuiTope score is on the y axis and amino acid on the x axis. Antibody reactivity from pooled sera to the strongest predicted epitope are plotted in D as the mean ± SD of replicate arrays.