2Protein Advances Incorporated, Seattle, WA
We have developed an influenza hemagglutinin protein microarray to assess humoral recognition of diverse influenza strains induced by vaccination and infection. Each array consists of controls and 127 hemagglutinin antigens from 60 viruses, spotted in replicates to generate a single array of 1296 spots. Six arrays are configured on a single slide, which in the following analysis was probed simultaneously with 2 isotype-specific fluorescent secondary antibodies yielding over 15,000 data points per slide. Here we report the use of this system to evaluate mouse, ferret, and human sera. The array allows simultaneous examination of the magnitude of antibody responses, the isotype of such antibodies, and the breadth of influenza strain recognition. We are advancing this technology as a platform for rapid, simple, high-throughput assessment of homologous and heterologous antibody responses to influenza disease and vaccination.
Considering the plasticity of the influenza virus, evaluating the ability of vaccines to induce broadly reactive humoral immune responses is critical (1-6). To facilitate this process, a protein microarray of influenza hemagglutinin antigens (HA) was developed by adsorbing picogram amounts of protein onto an activated glass surface via an epoxy attachment method (Arrayit Corporation, Sunnyvale,CA) (7, 8). The resulting covalent interaction yields randomly oriented proteins, promoting exposure of potential antigenic determinants configured in 100 µm size spots, thus allowing for highly dense arrays. We developed an array composed of controls and 127 hemagglutinin antigens from 60 influenza viruses (Protein Sciences Meriden, CT; Sino Biologicals, Bejing, China), including representatives from the circulating H5 and H1 strains that are potential and known causes of pandemic flu (Figure 1A, Supplementary Table S1) (9). Each protein is spotted a total of 9 times in triplicate spots for each of 3 10-fold dilutions. Previously reported arrays were smaller, limiting their breadth of assessment and sample throughput (10, 11).
Execution of the microarray experiment is performed in a manner much like a traditional ELISA, with the exception that fluorescently labeled secondary antibodies mediate detection rather than the HRP conjugates generally used for ELISA. For most samples, a common dilution ranging from 1:10 to 1:1000 is made in a simple buffer of PBS and 1% BSA for a total volume of 300 µL. Each experiment requires 0.3 to 30 µL of sera, which is incubated with the array for 30 min. The protein microarray slide is composed of six partitioned arrays allowing simultaneous testing of analytes and minimizing intra-assay variability (Figure 1A). Arrays are washed, then exposed to secondary antibodies for 30 min and re-washed. Following development, arrays are scanned using a conventional array scanner and software (GenePix 4000b, GenPix6.1, Molecular Devices, Sunnyvale, CA). The experiment is performed in less than two hours.
A protein microarray of influenza hemagglutinin antigens was generated by adsorbing picogram amounts of 127 hemagglutinin antigens from 60 viruses onto a single array. Sera to be evaluated are diluted and incubated with the microarray, followed by simultaneous probing with two isotype-specific fluorescent secondary antibodies. Arrays are then scanned using a conventional gene array scanner.
Adsorption of protein to a solid substrate, as well as protein purification and production processes, could alter the structures of the hemagglutinin antigens on the array. To evaluate the conformational integrity of the antigens, arrays were probed with two broadly reactive, conformation-sensitive monoclonal antibodies, KB2 and 6F12, which react with the stalk region of hemagglutinin, spanning the HA2 and HA1 subunits (a kind gift of Peter Palese and Florian Krammer) (12, 13). KB2 reacts with H1 and H5 viruses, while 6F12 reacts with H1 viruses. A third antibody that reacts with the HA1 subunit of H5N1 viruses, which binds independently of conformation, was also used (7C2, MyBioSource, San Diego, CA). As shown, the stalk-specific antibodies KB2 and 6F12 reacted only with the HA0 (head and stalk) proteins on the array (Supplementary Figure S1A). Examination of the subtypes recognized by the antibodies demonstrated that the array recapitulated the reported specificities of KB2 and 6F12 (Supplementary Figure S1B). While this experiment does not validate each protein on the array or address the concern that some of the proteins may be misfolded, it does demonstrate that the processes used to create the array do not broadly alter the conformation of the antigens to such an extent that conformation-dependent neutralizing antibodies cannot bind. It remains to be determined if the breadth of array recognition correlates to protection.
To test the utility and fidelity of the array, we compared antibody responses induced by vaccination with protein alone or protein combined with different adjuvants. In brief, C57BL/6 mice received intramuscular injections of recombinant protein from the H5N1 A/VietNam /1203/2004 virus (rH5) in saline or in combination with the Th1 IgG1/IgG2c isotype promoting adjuvant, glucopyranosyl lipid adjuvant–stable emulsion (GLA-SE), or the Th2 polarizing adjuvant (IgG1 promoting) stable emulsion (SE) (14-18). Mice were injected at two time points three weeks apart and serum was collected three weeks after the final boost. As expected, GLA-SE elicited antibodies for both IgG1 and IgG2c isotypes in immunized mice, while SE induced primarily IgG1 antibodies (Figure 1B, Figure 2). Within a single reaction, multiple secondary antibodies with distinguishable fluorescent signals were used to resolve antibody isotypes. To demonstrate that combined detection did not interfere with signals, serum from a single mouse was used in three development conditions in which the arrays were probed with anti-IgG1 alone, anti-IgG2c alone, or a combination of the two secondary antibodies. Negligible interference was detected between the conditions, and it is apparent that array to array variability is minimal considering three separate arrays yielded very similar data (Supplementary Figure S2A). We examined the relationship between the amount of protein spotted and the strengths of the signals. The intensity of signal was proportionate to the amount of protein adsorbed, indicating specific reactivity and detection of signal within a linear range (Figure 1C, Supplementary Figure S2B). To verify the accuracy of the microarray, the sera from mice vaccinated with rH5 GLA-SE were subjected to end point titration ELISA and the results were compared with those from the microarray (19). As shown, results were similar between the two assays with regard to the hierarchy of responses as well as identification of antibody isotypes, although the sensitivity of the array was lower than that of the ELISA by a significant factor (Supplementary Figure S2C).
An expanded analysis of the array revealed response differences among vaccination strategies. For many of the viruses represented, multiple forms of the hemagglutinin antigens are included within the array: HA0, globular head and stalk; HA1, globular head (Supplementary Table S1). The responses from the sera of naïve and immunized mice against the HA0 subset of proteins within the microarray are shown in Figure 2. Immunization with the rH5 and saline elicited detectable responses that were largely constrained to the homologous subtype of the immunogen, H5 and were predominately of the IgG1 isotype. Vaccination with adjuvants altered the quality of responses and significantly increased responses to a number of proteins relative to rH5 and saline vaccination. Similar to the rH5 and saline immunization, the Th2-biasing adjuvant, SE, produced mostly IgG1 responses, while the Th1-biasing adjuvant, GLA-SE, produced a more balanced response between IgG1 and IgG2c antibodies (Figure 2).
We also performed experiments on ferret and human samples. We compared the response profiles of serum from an individual ferret before and after infection with the H5N1 A/Viet Nam/1203/2004 virus. Several responses were induced, particularly within the H5N1 subset of proteins (Supplementary Figure S3A). Next, we tested the capacity of the array to detect responses induced by vaccination with the 2012–2013 trivalent vaccine (A/California/7/2009 (H1N1)-like virus, A/Victoria /361/2011 (H3N2)-like virus, B/Wisconsin /1/2010-like virus) within a human subject (Supplementary Fig. S3B). As shown, the array demonstrated pre-existing responses as well as boosted and new responses to H1N1 and H3N2 strains. While the 2012–2013 trivalent vaccine contained no H5 antigens, it appeared that several H5N1 strains were recognized. The only limitation with regard to analysis of species is the availability of secondary antibodies.
Herein, we propose that our hemagglutinin protein microarray provides a significant improvement over comparable measurements of serum antibody levels by techniques such as ELISA or bead based multiplexing. Cost and time savings are tremendous considering that an estimated 200 ELISA end point titration experiments would be required to acquire information equivalent to that from a single microarray slide.
We thank Yeung Tutterow and Vanitha Raman for their thoughtful discussions as well as Patricia Hon and Blarney Hidsee for their contributions. We also thank Winston Wicomb and the vivarium staff at IDRI. We also thank Peter Palese and Florian Krammer for providing antibodies. This work is supported by a contract from BARDA #HHSO100201000039C and by a grant from the Bill and Melinda Gates Foundation #42387.
The authors declare no competing interests.
Address correspondence to Darrick Carter, Infectious Disease Research Institute, Seattle, WA. E-mail: Darrick.Carter@IDRI.org
1.) Song, J.M., N. Van Rooijen, J. Bozja, R.W. Compans, and S.M. Kang. 2011. Vaccination inducing broad and improved cross protection against multiple subtypes of influenza A virus. Proc. Natl. Acad. Sci. USA 108:757-761. 2.) Neirynck, S., T. Deroo, X. Saelens, P. Vanlandschoot, W.M. Jou, and W. Fiers. 1999. A universal influenza A vaccine based on the extracellular domain of the M2 protein. Nat. Med. 5:1157-1163. 3.) Wei, C.J., J.C. Boyington, P.M. McTamney, W.P. Kong, M.B. Pearce, L. Xu, H. Andersen, S. Rao. 2010. Induction of broadly neutralizing H1N1 influenza antibodies by vaccination. Science 329:1060-1064. 4.) Corti, D., J. Voss, S.J. Gamblin, G. Codoni, A. Macagno, D. Jarrossay, S.G. Vachieri, D. Pinna. 2011. A neutralizing antibody selected from plasma cells that binds to group 1 and group 2 influenza A hemagglutinins. Science 333:850-856. 5.) Russell, C.J. 2011. Stalking influenza diversity with a universal antibody. N. Engl. J. Med. 365:1541-1542. 6.) Ekiert, D.C., G. Bhabha, M.A. Elsliger, R.H. Friesen, M. Jongeneelen, M. Throsby, J. Goudsmit, and I.A. Wilson. 2009. Antibody recognition of a highly conserved influenza virus epitope. Science 324:246-251. 7.) Oh, S.J., B.J. Hong, K.Y. Choi, and J.W. Park. 2006. Surface modification for DNA and protein microarrays. OMICS 10:327-343. 8.) Zhou, F.X., J. Bonin, and P.F. Predki. 2004. Development of functional protein microarrays for drug discovery: progress and challenges. Comb. Chem. High Throughput Screen. 7:539-546. 9.) Medina, R.A., and A. Garcia-Sastre. 2011. Influenza A viruses: new research developments. Nat. Rev. Microbiol. 9:590-603. 10.) Gall, A., B. Hoffmann, T. Harder, C. Grund, D. Hoper, and M. Beer. 2009. Design and validation of a microarray for detection, hemagglutinin subtyping, and pathotyping of avian influenza viruses. J. Clin. Microbiol. 47:327-334. 11.) Koopmans, M., E. de Bruin, G.J. Godeke, I. Friesema, R. van Gageldonk, M. Schipper, A. Meijer, R. van Binnendijk. 2012. Profiling of humoral immune responses to influenza viruses by using protein microarray. Clin. Microbiol. Infect. 18:797-807. 12.) Tan, G.S., F. Krammer, D. Eggink, A. Kongchanagul, T.M. Moran, and P. Palese. 2012. A pan-H1 anti-hemagglutinin monoclonal antibody with potent broad-spectrum efficacy in vivo. J. Virol. 86:6179-6188. 13.) Hai, R., F. Krammer, G.S. Tan, N. Pica, D. Eggink, J. Maamary, I. Margine, R.A. Albrecht, and P. Palese. 2012. Influenza viruses expressing chimeric hemagglutinins: globular head and stalk domains derived from different subtypes. J. Virol. 86:5774-5781. 14.) Mosmann, T.R., and R.L. Coffman. 1989. TH1 and TH2 cells: different patterns of lymphokine secretion lead to different functional properties. Annu. Rev. Immunol. 7:145-173. 15.) Coler, R.N., S.L. Baldwin, N. Shaverdian, S. Bertholet, S.J. Reed, V.S. Raman, X. Lu, J. DeVos. 2010. A synthetic adjuvant to enhance and expand immune responses to influenza vaccines. PLoS ONE 5:e13677. 16.) Behzad, H., A.L. Huckriede, L. Haynes, B. Gentleman, K. Coyle, J.C. Wilschut, T.R. Kollmann, S.G. Reed, and J.E. McElhaney. 2012. GLA-SE, a synthetic toll-like receptor 4 agonist, enhances T-cell responses to influenza vaccine in older adults. J. Infect. Dis. 205:466-473. 17.) Fox, C.B., S.L. Baldwin, T.S. Vedvick, E. Angov, and S.G. Reed. 2012. Effects on immunogenicity by formulations of emulsion-based adjuvants for malaria vaccines. Clin. Vaccine Immunol. 19:1633-1640. 18.) Clegg, C.H., R. Roque, N. Van Hoeven, L. Perrone, S.L. Baldwin, J.A. Rininger, R.A. Bowen, and S.G. Reed. 2012. Adjuvant solution for pandemic influenza vaccine production. Proc. Natl. Acad. Sci. USA 109:17585-17590. 19.) Miura, K., A.C. Orcutt, O.V. Muratova, L.H. Miller, A. Saul, and C.A. Long. 2008. Development and characterization of a standardized ELISA including a reference serum on each plate to detect antibodies induced by experimental malaria vaccines. Vaccine 26:193-200.