High-throughput profiling of the humoral immune responses against thirteen human papillomavirus types by proteome microarrays

Abstract

We have developed microarrays with all eight proteins encoded by 13 different human papillomavirus types associated with anogenital cancer (HPV-16, -18, -31, -33, -35, -45, and -53), genital warts (HPV-6 and -11), or skin lesions (HPV-1, -2, -4, and -5). We analyzed the seroprevalence of antibodies in 546 patients, which had either cervical carcinomas, or precursor lesions, or which were asymptomatic. All patient groups contained sera ranging from high reactivity against multiple HPV proteins to low or no reactivity. Computational analyses showed the E7 proteins of carcinogenic HPV types as significantly more reactive in cancer patients compared to asymptomatic individuals and discriminating between cancer and HSIL or LSIL patients. Antibodies against E4 and E5 had the highest seroprevalence but did not exhibit differential reactivity relative to pathology. Our study introduces a new approach to future evaluation of the overall antigenicity of HPV proteins and cross-reaction between homologous proteins.

Fig. 1. Representative example of the raw data generated by processing HPV protein microarrays with a patient serum

The figure shows a slide with 16 microarrays, each containing duplicates of 104 HPV proteins and additional controls. Brightness of spots correlates with increased sero-reactivity. The upper line of white spots indicating maximal reactivity contains all positive controls. This particular microarray has been probed with serum from a cervical cancer patient. The strongest reactions, indicated, are 15 different proteins from eight HPV types. The uppermost line includes human IgGs and EBNA1 protein as controls.

Fig 2. Complex multi-target as well as low serological activity in patients with HPV associated cancers, in HPV infected asymptomatic patients, and in HPV negative patients

Documentation of the diversity of individual data that led to the heatmap in Fig. 3 and the analyses in Fig. 4 and 5. This figure represents the only data in this paper that compare HPV positive and HPV negative asymptomatic patients, as analyses not discussed here did not reveal obvious serological differences between the two groups.

Fig. 3. Heatmap of 113,568 microarray signals generated by 546 sera targeted at all 104 proteins encoded by 13 HPV types

A color scale identifies signal intensities ranging from strong (red), moderate (black) to weak (green). Columns correlate to samples sorted by subjects, lines correlate to the different proteins. The whole collection of sera is grouped into patients with cervical cancer, HSIL, LSIL and asymptomatic infection.

Fig. 4. Comparison of the seroreactivity of HPV proteins in cancer and asymptomatic patients

The mean and standard error of each antigen on the microarray was determined for 280 healthy subjects (blue bars) and 93 cervical cancer patients (red bars), and the p-value (black line) comparing the two groups was determined. Multiple comparison corrected p-values less than 0.05 are considered significant (below the red horizontal line). A. The antigens were sorted by decreasing average signal intensity and organized to show 16 significantly differentially reactive antigens (left) and 43 cross-reactive antigens (p-value >0.05, right). Proteins with average signal intensity lower than no DNA controls are not shown. B. The antigens were organized according to annotated gene ID. All HPV proteins are shown. One third of the significant proteins (p<0.05) are E7 proteins.

Fig. 4. Comparison of the seroreactivity of HPV proteins in cancer and asymptomatic patients

The mean and standard error of each antigen on the microarray was determined for 280 healthy subjects (blue bars) and 93 cervical cancer patients (red bars), and the p-value (black line) comparing the two groups was determined. Multiple comparison corrected p-values less than 0.05 are considered significant (below the red horizontal line). A. The antigens were sorted by decreasing average signal intensity and organized to show 16 significantly differentially reactive antigens (left) and 43 cross-reactive antigens (p-value >0.05, right). Proteins with average signal intensity lower than no DNA controls are not shown. B. The antigens were organized according to annotated gene ID. All HPV proteins are shown. One third of the significant proteins (p<0.05) are E7 proteins.

Fig. 5. Differential seroreactivity against HPVs early proteins

A. The differentially reactive E antigens comparing 280 subjects without lesions and 93 cervical cancer patients. B. The same antigens were plotted comparing 93 cancer patients with 50 HSIL patients. C. The same antigens were plotted comparing 280 subjects without Lesions with 50 HSIL patients. Multiple comparison corrected p-values less than 0.05 are considered significant (signal extending below the red horizontal line).

Fig. 5. Differential seroreactivity against HPVs early proteins

A. The differentially reactive E antigens comparing 280 subjects without lesions and 93 cervical cancer patients. B. The same antigens were plotted comparing 93 cancer patients with 50 HSIL patients. C. The same antigens were plotted comparing 280 subjects without Lesions with 50 HSIL patients. Multiple comparison corrected p-values less than 0.05 are considered significant (signal extending below the red horizontal line).

Fig. 5. Differential seroreactivity against HPVs early proteins

A. The differentially reactive E antigens comparing 280 subjects without lesions and 93 cervical cancer patients. B. The same antigens were plotted comparing 93 cancer patients with 50 HSIL patients. C. The same antigens were plotted comparing 280 subjects without Lesions with 50 HSIL patients. Multiple comparison corrected p-values less than 0.05 are considered significant (signal extending below the red horizontal line).

Fig. 6. Anti-E7 immune responses exist only in a subset of cancer patients, whose HPV type status was known

The data document that the preferential reactivity of E7 in cancer patients does not mean that the presence of this antigen in all patients induces consistently an anti-E7 immune response.