Molecular signatures of antibody responses derived from a systems biology study of five human vaccines

Journal name:
Nature Immunology
Volume:
15,
Pages:
195–204
Year published:
DOI:
doi:10.1038/ni.2789
Received
Accepted
Published online

Abstract

Many vaccines induce protective immunity via antibodies. Systems biology approaches have been used to determine signatures that can be used to predict vaccine-induced immunity in humans, but whether there is a 'universal signature' that can be used to predict antibody responses to any vaccine is unknown. Here we did systems analyses of immune responses to the polysaccharide and conjugate vaccines against meningococcus in healthy adults, in the broader context of published studies of vaccines against yellow fever virus and influenza virus. To achieve this, we did a large-scale network integration of publicly available human blood transcriptomes and systems-scale databases in specific biological contexts and deduced a set of transcription modules in blood. Those modules revealed distinct transcriptional signatures of antibody responses to different classes of vaccines, which provided key insights into primary viral, protein recall and anti-polysaccharide responses. Our results elucidate the early transcriptional programs that orchestrate vaccine immunity in humans and demonstrate the power of integrative network modeling.

At a glance

Figures

  1. Antibody responses of subjects who received vaccines against meningococcus.
    Figure 1: Antibody responses of subjects who received vaccines against meningococcus.

    (ad) Concentration of total IgG (a), IgG1 and IgG2 (b) and IgM (c) and SBA titers (d) specific to polysaccharides from N. meningitidis serogroup A (MenA) or C (MenC), measured in subjects vaccinated with MPSV4 or MCV4. (e) DT-specific IgG response subjects vaccinated as in a. (f,g) Frequency of ASCs (selected as CD3CD20CD38hiCD27hiCD19+ cells) at various times after vaccination with MPSV4 or MCV4 (f) and quantification of ASCs at day 7 correlated with DT-specific serum IgG at day 7 after vaccination (g). Light lines (b,f) represent individual subjects. *P < 0.05, **P < 0.01 and ***P < 0.001 (unpaired t-test for comparisons between vaccine groups; paired t-test for comparison with baseline in a group; Pearson's correlation in g). Data are from one experiment (mean and s.e.m. of 13 subjects (MPSV4) or 17 subjects (MCV4)).

  2. Analysis of data on the blood transcriptome for five human vaccines.
    Figure 2: Analysis of data on the blood transcriptome for five human vaccines.

    (a) Difference in gene expression at day 3 or day 7 after vaccination compared with baseline at day 0 (horizontal axis) versus results from a paired t-test (vertical axis), for each vaccine and each time point. Red dots indicate DEGs (P < 0.001), numbers in plots indicate number of DEGs. Data are from one experiment. (b) Work flow for comparison of the transcriptome signatures of five human vaccines (interactive figures are in the online data portal).

  3. Analysis of differences in gene expression for five vaccines.
    Figure 3: Analysis of differences in gene expression for five vaccines.

    (a) Example of an online interactive figure showing DEGs shared by vaccine studies ('snapshot' of Fig. 1 in the online data portal at http://www.immuneprofiling.org/papers/meni/). Blue indicates links; when a vaccine is selected, links are shown in green. (b) Expression of selected DEGs encoding molecules involved in B cell development (top) and innate immunity (bottom), for genes upregulated (Up) or downregulated (Down) at day 7 (top) or day 3 (bottom) after vaccination (vaccine, above plots) relative to their expression at baseline (day 0). NS, no significant change in expression. (c) Significance of the enrichment of DEGs, plus 'linker' genes identified by the interactome-bibliome integrative approach and shared by four or more vaccines, in gene-ontology groups (genes found in 'immune system development', Supplementary Fig. 4f). Numbers in bars indicate total DEGs in each category.

  4. BTMs provide a sensitive and robust statistical framework.
    Figure 4: BTMs provide a sensitive and robust statistical framework.

    (a) Construction of BTMs through large-scale data integration (full details, Supplementary Note). GO, gene ontology; TF, transcription factor; KEGG, Kyoto encyclopedia of genes and genomes; PID, pathway-interaction database. (b) Discriminative power of BTMs and canonical pathways (key), or simulated pathways consisting of randomly selected genes (Random), in assessing the transcriptome data for MCV4 at day 3 (relative to day 0), by t-score (additional examples, Supplementary Fig. 9). (c) Analysis of the statistical significance of the correlation of BTMs to antibody data, for the MCV4 transcriptome data comparing day 3 with day 0 (baseline): each module is collapsed to a single activity score (mean value of all member genes), and Pearson correlation to antibody data is calculated across all subjects (red bars); gray shaded curve, distribution of random data generated by permutations of module gene memberships and sample labels. (d) Classification of genes in BTM M156.1 by products encoded (key); each 'edge' (gray line) represents a coexpression relationship learned from public data. (e) Correlation of the activity of module M156.1 and the later DT-specific antibody response to MCV4.

  5. BTM analysis reveals distinct mechanisms of antibody response.
    Figure 5: BTM analysis reveals distinct mechanisms of antibody response.

    (a) Vaccine data sets (six segments in color along perimeter) with ordered list of all BTMs for each (bar code–like bands in innermost ring) adjacent to 'histograms' of modules (ring exterior to innermost ring) significantly correlated to the antibody response (red, positive correlation; blue, negative correlation; plotted on circumferential gray coordinates), with module names inside outermost perimeter. Curved lines (in color) in the center link significant modules that are common between vaccines (as in Supplementary Fig. 12b–d; gray links for modules omitted in Supplementary Fig. 12). An interactive version of this figure is available (Fig. 2 of the online data portal). D3/0, day 3 versus day 0; D7/0, day 7 versus day 0. (b) Illustration of module activity: black boxes indicate membership of genes (top margin) in the corresponding module (left margin). Right, Pearson correlation between module activity and antibody response in each study. Bottom, correlation between module member genes and the antibody response.

  6. The polysaccharide-containing vaccine against meningococcus activates myeloid DCs.
    Figure 6: The polysaccharide-containing vaccine against meningococcus activates myeloid DCs.

    (a) Secretion of IL-6 from human monocyte-derived DCs (top) and myeloid DCs (bottom) treated with medium (Med), lipopolysaccharide (LPS) or various dilutions (horizontal axes) of MPSV4 or MCV4 (key); similar results were obtained for TNF and IL-12p40 (data not shown). (b) Secretion of IL-6 from C57BL/6 wild-type CD11c+ DCs and mutant CD11c+ DCs deficient in TLR4 (Tlr4-KO; top left), TRIF (Trif-KO; top right), MyD88 (Myd88-KO; bottom left) or the adaptor ASC (Asc-KO; bottom right), isolated from mouse spleens by magnetic-activated cell separation and stimulated for 24 h with MPSV4 (dilution, horizontal axes). ND, not detected.*P < 0.05, **P < 0.01 and ***P < 0.0001 (paired t-test (a) or unpaired t-test (b)). Data are representative of four (myeloid DCs) or six (monocyte-derived DCs) experiments (a) or are from one experiment representative of four (TLR4, MyD88 and TRIF) or six (ASC) experiments (b; mean and s.e.m.).

  7. Antibody responses induced by meningococcal vaccines.
    Supplementary Fig. 1: Antibody responses induced by meningococcal vaccines.

    (a) ELISPOT captured day 7 IgG antibody-secreting cells (ASCs) in a subset of vaccinees. Plates were coated with goat anti human IgG antibody, MCV4, DT or MPSV4 and detected with anti human IgG biotin and streptavidin HRP. (b) Two representative subjects are shown. On the left, 1X corresponds to 0.17 million PBMCs. DT = Diphtheria Toxoid. (c) Production of IgG is a robust indicator of antibody response. Day 30 IgG correlates well with data from day 180 and 2 years. Data are shown for both vaccines. The day 30/0 IgG data (serogroups A plus C) were thus used for the later analysis of antibody correlation.

  8. The XBP-1 network is upregulated in the MCV4 transcriptomic response.
    Supplementary Fig. 2: The XBP-1 network is upregulated in the MCV4 transcriptomic response.

    Targets of XBP-1 include genes associated with antibody-secreting cell differentiation (tan shading) and the unfolded protein response (purple shading). MCV4 DEGs are colored in red. Enrichment was assessed by Fisher's exact test (p = 5 x 10-14).

  9. Cytokines and chemokines measured by Luminex assay for meningococcal vaccines.
    Supplementary Fig. 3: Cytokines and chemokines measured by Luminex assay for meningococcal vaccines.

    Left: Day 3 and 7 levels (log2, pg/ml) compared to day 0. Each row represents a cytokine and each column represents a vaccinee. Right: Serum protein levels (log2, pg/ml) that are significantly different after vaccination. * p < 0.05, paired t-test.

  10. Integration of DEGs with interactome and bibliome data.
    Supplementary Fig. 4: Integration of DEGs with interactome and bibliome data.

    (a) Illustration of the network integration method. DEGs from a vaccine study are connected by their links in the Interactome or Bibliome. Additional “linker” genes are then included based on their significant association to the DEGs (see Method for details). (b) Numbers of upregulated DEGs in common to 1 to 5 vaccine datasets. (c) Numbers of genes in the integrated network, and numbers of genes in common to 1 to 5 vaccine datasets. (d) The genes in common between MCV4 network and TIV network are highly expressed in plasma cells. As all genes are ranked by their expression level in sorted plasma cells (the color bar, according to data in Abbas et al 2009, PLoS ONE, 4:e6098), the ranks of MCV4/TIV common genes are indicated by the horizontal bars. (e) A subnetwork that is common between YF-17D and LAIV, enriched for TCR signaling pathway and interferon response genes. Genes in dark gray: DEGs as in (b), light gray: “linker” genes that have p < 0.05 in YF-17D transcriptomic data. (f) Genes enriched in the “immune systems development” GO category from Figure 3c.

  11. Molecular pathways induced by different vaccines.
    Supplementary Fig. 5: Molecular pathways induced by different vaccines.

    Selected pathways from NCI-Nature Pathway Interaction Database whose expression were significantly (p < 0.01) changing at (a) day 3 or (b) day 7 compared to pre-vaccination data (day 0). Genes are ranked by t-score and the significance of pathways is tested by GSEA (see online method for details).

  12. Pathways associated with antibody response.
    Supplementary Fig. 6: Pathways associated with antibody response.

    Selected pathways from NCI-Nature Pathway Interaction Database whose expression at (a) day 3 or (b) day 7 post-vaccination was significantly (p < 0.01) correlated to antibody response. Genes are ranked according to Pearson correlation to antibody response, and the significance of pathways is tested by GSEA (see Method for details). Influenza LAIV is excluded here because this vaccine does not induce serum/plasma antibody responses. MCV4 has two sets of antibody correlation data: one against meningococcal polysaccharide (MCV4-PS) and another against diphtheria toxoid (MCV4-DT).

  13. Examples of protein complexes detected during BTM construction.
    Supplementary Fig. 7: Examples of protein complexes detected during BTM construction.

    (a) Ribosome complex extracted from the master network. (b) Nuclear pore complex and (c) Splicesome complexes extracted from BTM modules. Each edge in the pictures represents a coexpression relationship detected in at least 3 publicly available microarray blood studies.

  14. Examples of BTM modules.
    Supplementary Fig. 8: Examples of BTM modules.

    (a) “T cell differentiation via ITK and PKC” BTM module which contains genes whose functions are supported by literature. This set of genes, however was not found in pathway databases. (b) “Chromosome Y linked” BTM module contains many genes transcribed in Y chromosome.

  15. Discriminative power of BTM modules compared with canonical pathways.
    Supplementary Fig. 9: Discriminative power of BTM modules compared with canonical pathways.

    Discriminative power is shown as t-scores in two-class comparison. The horizontal gray lines show the threshold of confidence level of α = 0.001. By this cutoff, the numbers of significant modules/pathways can be looked up on the X-axis (top two numbers inserted by curves). YF-17D data were from Querec et al 2009, Nature Immunology 10:116; MCV4 this study; Turberculosis from Berry et al 2010, Nature 466: 973; RTS,S from Vahey et al 2010 J Infect Dis 201:580. All testing data sets were never part of the BTM construction process. KEGG, BioCarta and NCI_PID are respective pathway databases. Chaussabel modules were version 2 based on Chaussabel et al 2008 Immunity 29:150. “Random” modules contain randomly selected genes, matched to the sizes of KEGG pathways.

  16. Day 7 activity of plasma cells-immunoglobulin module (M156.1; Figure 4d) in blood transcriptome correlated to day 30 antibody response to TIV.
    Supplementary Fig. 10: Day 7 activity of plasma cells–immunoglobulin module (M156.1; Figure 4d) in blood transcriptome correlated to day 30 antibody response to TIV.

    The TIV 2007 (n=9) and 2008 (n=24) data sets were described in Nakaya et al 2011, Nature Immunology 12:786.

  17. Hierarchical clustering of antibody correlations of BTM modules.
    Supplementary Fig. 11: Hierarchical clustering of antibody correlations of BTM modules.

    Heat map of BTMs (rows) and vaccines (columns) whose baseline-normalized expression at day 3 correlated with baseline-normalized antibody response after vaccination (colors in map indicate Pearson correlation values). Clustering was performed (a) using all 346 BTM modules and (b) using the top 50% modules with highest variance across these 6 datasets.

  18. Distinctive early transcription programs correlated to vaccine antibody response.
    Supplementary Fig. 12: Distinctive early transcription programs correlated to vaccine antibody response.

    (a) Clustering analysis from Supplementary Figure 10b, marked by the three programs: Primary viral response (b), Polysaccharide response (c) and Protein recall response (d). From each data set in (a), the top 20 modules with highest positive correlations and the top 20 modules with highest negative correlations were selected. Among these selected modules, the modules concordant between two datasets (Pearson correlation coefficient above 0.3 or below -0.3 in both datasets) are shown in (b, c, d) and the rest are shown in (e, f, g).

  19. DC-related BTM modules and genes.
    Supplementary Fig. 13: DC-related BTM modules and genes.

    A filled unit in the center grid indicates the membership of the gene (top axis) in the corresponding module (left axis). The heat map on the right shows the Pearson correlation between module activity and antibody response in each study. The bottom heat map shows correlation between module member genes and antibody response.

  20. The DC surface signature module is positively correlated to polysaccharide response.
    Supplementary Fig. 14: The DC surface signature module is positively correlated to polysaccharide response.

    (a) Network of “DC surface” module member genes. (b) Correlation between D3/0 module activity and D30/0 anti-polysaccharide IgG response in MCV4 and MPSV4.

  21. Polysaccharide vaccine stimulates DCs.
    Supplementary Fig. 15: Polysaccharide vaccine stimulates DCs.

    MPSV4 stimulates the phenotypic maturation of human myeloid DC, while MCV4 to a much lesser extent. FACS data were obtained after 24 h (n = 4–6).

  22. MPSV4 stimulation of DCs is dependent on MyD88, TRIF, TLR4 and ASC in vitro.
    Supplementary Fig. 16: MPSV4 stimulation of DCs is dependent on MyD88, TRIF, TLR4 and ASC in vitro.

    Expanded version of Figure 6 showing genes that were required for DC in vitro stimulation with MPSV4. CD11c+ DCs were isolated by MACS from spleens of C57BL/6 mice (WT) or various knock-out mice and stimulated for 24 h; *p<0.05, ***p<0.0001, unpaired t-test was used to compare WT vs. KO (n = 4–6).

  23. Antigen-specific serum antibody and SBA titers induced by MCV4 and MPSV4 in mice in vivo.
    Supplementary Fig. 17: Antigen-specific serum antibody and SBA titers induced by MCV4 and MPSV4 in mice in vivo.

    C57BL/6 mice were immunized sub–cutaneously with one human dose (a-c, n=4-9) or two doses (d, n=5) of MPSV4 (blue) or MCV4 (red) or left untreated. (a) Mice were bled after 7, 14, and 28 days and serum IgM, IgG and IgA levels against meningococcal serogroup C were measured. (b) Serum IgG against serogroups A, Y and W-135 were measured at day 28 and (c) Serum Bactericidal Activity (SBA) titers against serogroups A and/or C were measured after 28 days. (d) Mice were boosted with a 2nd dose at day 30. Each gray line represents an individual animal. * p<0.05, ** p<0.01, Mann-Whitney test.

  24. Cell-type expression patterns in BTM-antibody analysis.
    Supplementary Fig. 18: Cell-type expression patterns in BTM-antibody analysis.

    BTM activities of cell surface signatures (see details in Suppl. Text), and their correlation to antibody response. X-axis shows module activity at D3/0 or D7/0, Y-axis the D30/0 antibody response.

  25. BTM modules whose activity on day 7 is correlated to later antibody response in two or more data sets.
    Supplementary Fig. 19: BTM modules whose activity on day 7 is correlated to later antibody response in two or more data sets.

    Each segment on the circle corresponds to one vaccine dataset. In each segment, the inner circular bands show an ordered list of all BTM modules, layered by histograms of significantly correlated modules, red for positive correlation and blue for negative correlation. Each link inside the circles connects a significant module (top 50 in one dataset by p-value) that is common between different data sets. Modules common to three vaccines are linked in black, and their labels are marked by a black dot. Plasma and immunoglobulin modules are linked in green. An interactive version of this figure is available at online data portal (Interactive Figure 3).

  26. B cell-related BTM modules and genes.
    Supplementary Fig. 20: B cell–related BTM modules and genes.

    A filled unit in the center grid indicates the membership of the gene (top axis) in the corresponding module (left axis). The heat map on the right shows the Pearson correlation between module activity and antibody response in each study. The bottom heat map shows correlation between module member genes and antibody response.

Accession codes

Primary accessions

Gene Expression Omnibus

References

  1. Pulendran, B. Learning immunology from the yellow fever vaccine: innate immunity to systems vaccinology. Nat. Rev. Immunol. 9, 741747 (2009).
  2. Pulendran, B., Li, S. & Nakaya, H.I. Systems vaccinology. Immunity 33, 516529 (2010).
  3. Nakaya, H.I. & Pulendran, B. Systems vaccinology: its promise and challenge for HIV vaccine development. Curr. Opin. HIV AIDS 7, 24 (2012).
  4. Querec, T.D. et al. Systems biology approach predicts immunogenicity of the yellow fever vaccine in humans. Nat. Immunol. 10, 116125 (2008).
  5. Gaucher, D. et al. Yellow fever vaccine induces integrated multilineage and polyfunctional immune responses. J. Exp. Med. 205, 31193131 (2008).
  6. Nakaya, H.I. et al. Systems biology of vaccination for seasonal influenza in humans. Nat. Immunol. 12, 786795 (2011).
  7. Vahey, M.T. et al. Expression of genes associated with immunoproteasome processing of major histocompatibility complex peptides is indicative of protection with adjuvanted RTS,S malaria vaccine. J. Infect. Dis. 201, 580589 (2010).
  8. Bucasas, K.L. et al. Early patterns of gene expression correlate with the humoral immune response to influenza vaccination in humans. J. Infect. Dis. 203, 921929 (2011).
  9. Zak, D.E. et al. Merck Ad5/HIV induces broad innate immune activation that predicts CD8+ T-cell responses but is attenuated by preexisting Ad5 immunity. Proc. Natl. Acad. Sci. USA 109, E3503E3512 (2012).
  10. Obermoser, G. et al. Systems scale interactive exploration reveals quantitative and qualitative differences in response to influenza and pneumococcal vaccines. Immunity 38, 831844 (2013).
  11. Furman, D. et al. Apoptosis and other immune biomarkers predict influenza vaccine responsiveness. Mol. Syst. Biol. 9, 659 (2013).
  12. Oberg, A.L., Kennedy, R.B., Li, P., Ovsyannikova, I.G. & Poland, G.A. Systems biology approaches to new vaccine development. Curr. Opin. Immunol. 23, 436443 (2011).
  13. Trautmann, L. & Sekaly, R. Solving vaccine mysteries: a systems biology perspective. Nat. Immunol. 12, 729 (2011).
  14. Querec, T. et al. Yellow fever vaccine YF-17D activates multiple dendritic cell subsets via TLR2, 7, 8, and 9 to stimulate polyvalent immunity. J. Exp. Med. 203, 413424 (2006).
  15. Geeraedts, F. et al. Superior immunogenicity of inactivated whole virus H5N1 influenza vaccine is primarily controlled by Toll-like receptor signalling. PLoS Pathog. 4, e1000138 (2008).
  16. Koyama, S. et al. Plasmacytoid dendritic cells delineate immunogenicity of influenza vaccine subtypes. Sci. Transl. Med. 2, 25ra24 (2010).
  17. Rouphael, N.G. & Stephens, D.S. Neisseria meningitidis: biology, microbiology, and epidemiology. Methods Mol. Biol. 799, 120 (2012).
  18. Goldschneider, I., Gotschlich, E.C. & Artenstein, M.S. Human immunity to the meningococcus. I. The role of humoral antibodies. J. Exp. Med. 129, 13071326 (1969).
  19. Käyhty, H., Karanko, V., Peltola, H., Sarna, S. & Makela, P.H. Serum antibodies to capsular polysaccharide vaccine of group A Neissera meningitidis followed for three years in infants and children. J. Infect. Dis. 142, 861868 (1980).
  20. Jokhdar, H. et al. Immunologic hyporesponsiveness to serogroup C but not serogroup A following repeated meningococcal A/C polysaccharide vaccination in Saudi Arabia. Clin. Diagn. Lab. Immunol. 11, 8388 (2004).
  21. Findlow, H. et al. Immunoglobulin G subclass response to a meningococcal quadrivalent polysaccharide-diphtheria toxoid conjugate vaccine. Clin. Vaccine Immunol. 13, 507510 (2006).
  22. Borrow, R., Balmer, P. & Miller, E. Meningococcal surrogates of protection–serum bactericidal antibody activity. Vaccine 23, 22222227 (2005).
  23. Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 1554515550 (2005).
  24. Cerami, E., Demir, E., Schultz, N., Taylor, B.S. & Sander, C. Automated network analysis identifies core pathways in glioblastoma. PLoS ONE 5, e8918 (2010).
  25. Nibbe, R.K., Koyuturk, M. & Chance, M.R. An integrative -omics approach to identify functional sub-networks in human colorectal cancer. PLOS Comput. Biol. 6, e1000639 (2010).
  26. Khatri, P., Sirota, M. & Butte, A.J. Ten years of pathway analysis: current approaches and outstanding challenges. PLoS Comput. Biol. 8, e1002375 (2012).
  27. Li, S., Nakaya, H.I., Kazmin, D.A., Oh, J.Z. & Pulendran, B. Systems biological approaches to measure and understand vaccine immunity in humans. Semin. Immunol. 25, 209218 (2013).
  28. Schaefer, C.F. et al. PID: the Pathway Interaction Database. Nucleic Acids Res. 37, D674D679 (2009).
  29. Lau, E. & Ronai, Z.A. ATF2-at the crossroad of nuclear and cytosolic functions. J. Cell Sci. 125, 28152824 (2012).
  30. Chaveroux, C. et al. Identification of a novel amino acid response pathway triggering ATF2 phosphorylation in mammals. Mol. Cell Biol. 29, 65156526 (2009).
  31. Ravindran, R. et al. Vaccine activation of the nutrient sensor GCN2 in dendritic cells enhances antigen presentation. Science (in the press).
  32. Haining, W.N. & Pulendran, B. Identifying gnostic predictors of the vaccine response. Curr. Opin. Immunol. 24, 332336 (2012).
  33. Chaussabel, D., Pascual, V. & Banchereau, J. Assessing the human immune system through blood transcriptomics. BMC Biol. 8, 84 (2010).
  34. Fraser, I.D.C. & Germain, R.N. Navigating the network: signaling cross-talk in hematopoietic cells. Nat. Immunol. 10, 327331 (2009).
  35. Friedman, N. Inferring cellular networks using probabilistic graphical models. Science 303, 799805 (2004).
  36. Segal, E. et al. Module networks: identifying regulatory modules and their condition-specific regulators from gene expression data. Nat. Genet. 34, 166176 (2003).
  37. Basso, K. et al. Reverse engineering of regulatory networks in human B cells. Nat. Genet. 37, 382390 (2005).
  38. Lee, I. et al. A single gene network accurately predicts phenotypic effects of gene perturbation in Caenorhabditis elegans. Nat. Genet. 40, 181188 (2008).
  39. Chaussabel, D. et al. A modular analysis framework for blood genomics studies: application to systemic lupus erythematosus. Immunity 29, 150164 (2008).
  40. Margolin, A.A. et al. ARACNE: an algorithm for the reconstruction of gene regulatory networks in a mammalian cellular context. BMC Bioinformatics 7 (suppl. 1), S7 (2006).
  41. Lefebvre, C. et al. A human B-cell interactome identifies MYB and FOXM1 as master regulators of proliferation in germinal centers. Mol. Syst. Biol. 6, 377 (2010).
  42. Lefebvre, C., Rieckhof, G. & Califano, A. Reverse-engineering human regulatory networks. Wiley Interdiscip. Rev. Syst. Biol. Med. 4, 311325 (2012).
  43. Sales, G. & Romualdi, C. Parmigene–a parallel R package for mutual information estimation and gene network reconstruction. Bioinformatics 27, 18761877 (2011).
  44. Chuang, H.Y., Lee, E., Liu, Y.T., Lee, D. & Ideker, T. Network-based classification of breast cancer metastasis. Mol. Syst. Biol. 3, 140 (2007).
  45. Fredlund, E., Ringner, M., Maris, J.M. & Pahlman, S. High Myc pathway activity and low stage of neuronal differentiation associate with poor outcome in neuroblastoma. Proc. Natl. Acad. Sci. USA 105, 1409414099 (2008).
  46. Vitour, D. et al. Polo-like kinase 1 (PLK1) regulates interferon (IFN) induction by MAVS. J. Biol. Chem. 284, 2179721809 (2009).
  47. Zhang, W. et al. The scaffold protein TANK/I-TRAF inhibits NF-κB activation by recruiting polo-like kinase 1. Mol. Biol. Cell 21, 25002513 (2010).
  48. Zughaier, S.M. Neisseria meningitidis capsular polysaccharides induce inflammatory responses via TLR2 and TLR4-MD-2. J. Leukoc. Biol. 89, 469480 (2011).
  49. Klein, S.L. & Poland, G.A. Personalized vaccinology: one size and dose might not fit both sexes. Vaccine 31, 25992600 (2013).
  50. Nakaya, H.I., Li, S.Z. & Pulendran, B. Systems vaccinology: learning to compute the behavior of vaccine induced immunity. Wiley Interdiscip. Rev. Syst. Biol. Med. 4, 193205 (2012).
  51. Avci, F.Y., Li, X.M., Tsuji, M. & Kasper, D.L. A mechanism for glycoconjugate vaccine activation of the adaptive immune system and its implications for vaccine design. Nat. Med. 17, 16021609 (2011).
  52. Holder, P.K. et al. Assignment of Neisseria meningitidis serogroup A and C class-specific anticapsular antibody concentrations to the new standard reference serum CDC1992. Clin. Diagn. Lab. Immunol. 2, 132137 (1995).
  53. Gheesling, L.L. et al. Multicenter comparison of Neisseria meningitidis serogroup C anti-capsular polysaccharide antibody levels measured by a standardized enzyme-linked immunosorbent assay. J. Clin. Microbiol. 32, 14751482 (1994).
  54. van Gageldonk, P.G., van Schaijk, F.G., van der Klis, F.R. & Berbers, G.A. Development and validation of a multiplex immunoassay for the simultaneous determination of serum antibodies to Bordetella pertussis, diphtheria and tetanus. J. Immunol. Methods 335, 7989 (2008).
  55. Maslanka, S.E. et al. Standardization and a multilaboratory comparison of Neisseria meningitidis serogroup A and C serum bactericidal assays. The Multilaboratory Study Group. Clin. Diagn. Lab. Immunol. 4, 156167 (1997).
  56. Poolman, J.T. et al. Measurement of functional anti-meningococcal serogroup a activity using strain 3125 as the target strain for serum bactericidal assay. Clin. Vaccine Immunol. 18, 11081117 (2011).
  57. Wrammert, J. et al. Rapid cloning of high-affinity human monoclonal antibodies against influenza virus. Nature 453, 667671 (2008).
  58. Moreno, R.L. et al. A murine model for the study of immune memory in response to pneumococcal conjugate vaccination. Vaccine 22, 30693079 (2004).
  59. Cerami, E.G. et al. Pathway Commons, a web resource for biological pathway data. Nucleic Acids Res. 39, D685D690 (2011).
  60. Keshava Prasad, T.S. et al. Human Protein Reference Database–2009 update. Nucleic Acids Res. 37, D767D772 (2009).
  61. Breitkreutz, B.J. et al. The BioGRID Interaction Database: 2008 update. Nucleic Acids Res. 36, D637D640 (2008).
  62. Aranda, B. et al. The IntAct molecular interaction database in 2010. Nucleic Acids Res. 38, D525D531 (2010).
  63. Ceol, A. et al. MINT, the molecular interaction database: 2009 update. Nucleic Acids Res. 38, D532D539 (2010).
  64. Matthews, L. et al. Reactome knowledgebase of human biological pathways and processes. Nucleic Acids Res. 37, D619D622 (2009).
  65. Romero, P. et al. Computational prediction of human metabolic pathways from the complete human genome. Genome Biol. 6, R2 (2005).
  66. Huang, D.W., Sherman, B.T. & Lempicki, R.A. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 4457 (2009).

Download references

Author information

  1. These authors contributed equally to this work.

    • Shuzhao Li,
    • Nadine Rouphael &
    • Sai Duraisingham

Affiliations

  1. Emory Vaccine Center, Emory University, Atlanta, Georgia, USA.

    • Shuzhao Li,
    • Nadine Rouphael,
    • Sai Duraisingham,
    • Carl Davis,
    • Sudhir Kasturi,
    • Mark J Mulligan,
    • Rafi Ahmed,
    • David S Stephens,
    • Helder I Nakaya &
    • Bali Pulendran
  2. Yerkes National Primate Research Center, Emory University, Atlanta, Georgia, USA.

    • Shuzhao Li,
    • Sai Duraisingham,
    • Sudhir Kasturi,
    • Helder I Nakaya &
    • Bali Pulendran
  3. The Hope Clinic of the Emory Vaccine Center, Division of Infectious Diseases, Emory University, Decatur, Georgia, USA.

    • Nadine Rouphael &
    • Mark J Mulligan
  4. Meningitis and Vaccine Preventable Diseases Branch, Division of Bacterial Diseases, National Center of Immunization and Respiratory Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia, USA.

    • Sandra Romero-Steiner,
    • Daniel S Schmidt,
    • Scott E Johnson,
    • Andrea Milton,
    • Gowrisankar Rajam &
    • George M Carlone
  5. Benaroya Research Institute, Seattle, Washington, USA.

    • Scott Presnell,
    • Charlie Quinn &
    • Damien Chaussabel
  6. Baylor Institute for Immunology Research, Baylor Research Institute, Dallas, Texas, USA.

    • Scott Presnell,
    • Charlie Quinn,
    • Damien Chaussabel &
    • A Karolina Palucka
  7. Department of Medicine, Division of Infectious Diseases, Emory University School of Medicine, Atlanta, Georgia, USA.

    • Carl Davis,
    • Mark J Mulligan &
    • David S Stephens
  8. Department of Microbiology and Immunology, Emory University, Atlanta, Georgia, USA.

    • Rafi Ahmed
  9. Department of Pathology, Emory University School of Medicine, Atlanta, Georgia, USA.

    • Helder I Nakaya &
    • Bali Pulendran

Contributions

S.L. did analyses in Figures 4 and 5 and Supplementary Figures 4,5,6,7,8,9,10,11,12,13,14 and 18,19,20; S.L. and H.I.N. did analyses in Figures 2 and 3 and Supplementary Figure 2 and helped with study design and presentation; N.R. organized the clinical study; S.D. did experiments and analyses in Figure 6 and Supplementary Figures 15,16,17; S.L. and S.D. did analysis in Supplementary Figure 3; S.R.-S., D.S.S., S.E.J., A.M., G.R. and G.M.C. did experiments in Figure 1; S.P., C.Q., D.C. and A.K.P. prepared the online data portal and Figure 3a; C.D. did experiments in Supplementary Fig. 1a,b; S.K. assisted with experiments in Figure 6 and Supplementary Figure 16; M.J.M. supervised the clinical study; R.A. supervised the study in Supplementary Figure 1a,b; D.S.S. helped conceive of and design the study; B.P. conceived of the study and designed and supervised the experiments and analyses; and S.L., H.I.N. and B.P. wrote the paper.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Supplementary information

Supplementary Figures

  1. Supplementary Figure 1: Antibody responses induced by meningococcal vaccines. (482 KB)

    (a) ELISPOT captured day 7 IgG antibody-secreting cells (ASCs) in a subset of vaccinees. Plates were coated with goat anti human IgG antibody, MCV4, DT or MPSV4 and detected with anti human IgG biotin and streptavidin HRP. (b) Two representative subjects are shown. On the left, 1X corresponds to 0.17 million PBMCs. DT = Diphtheria Toxoid. (c) Production of IgG is a robust indicator of antibody response. Day 30 IgG correlates well with data from day 180 and 2 years. Data are shown for both vaccines. The day 30/0 IgG data (serogroups A plus C) were thus used for the later analysis of antibody correlation.

  2. Supplementary Figure 2: The XBP-1 network is upregulated in the MCV4 transcriptomic response. (315 KB)

    Targets of XBP-1 include genes associated with antibody-secreting cell differentiation (tan shading) and the unfolded protein response (purple shading). MCV4 DEGs are colored in red. Enrichment was assessed by Fisher's exact test (p = 5 x 10-14).

  3. Supplementary Figure 3: Cytokines and chemokines measured by Luminex assay for meningococcal vaccines. (357 KB)

    Left: Day 3 and 7 levels (log2, pg/ml) compared to day 0. Each row represents a cytokine and each column represents a vaccinee. Right: Serum protein levels (log2, pg/ml) that are significantly different after vaccination. * p < 0.05, paired t-test.

  4. Supplementary Figure 4: Integration of DEGs with interactome and bibliome data. (546 KB)

    (a) Illustration of the network integration method. DEGs from a vaccine study are connected by their links in the Interactome or Bibliome. Additional “linker” genes are then included based on their significant association to the DEGs (see Method for details). (b) Numbers of upregulated DEGs in common to 1 to 5 vaccine datasets. (c) Numbers of genes in the integrated network, and numbers of genes in common to 1 to 5 vaccine datasets. (d) The genes in common between MCV4 network and TIV network are highly expressed in plasma cells. As all genes are ranked by their expression level in sorted plasma cells (the color bar, according to data in Abbas et al 2009, PLoS ONE, 4:e6098), the ranks of MCV4/TIV common genes are indicated by the horizontal bars. (e) A subnetwork that is common between YF-17D and LAIV, enriched for TCR signaling pathway and interferon response genes. Genes in dark gray: DEGs as in (b), light gray: “linker” genes that have p < 0.05 in YF-17D transcriptomic data. (f) Genes enriched in the “immune systems development” GO category from Figure 3c.

  5. Supplementary Figure 5: Molecular pathways induced by different vaccines. (280 KB)

    Selected pathways from NCI-Nature Pathway Interaction Database whose expression were significantly (p < 0.01) changing at (a) day 3 or (b) day 7 compared to pre-vaccination data (day 0). Genes are ranked by t-score and the significance of pathways is tested by GSEA (see online method for details).

  6. Supplementary Figure 6: Pathways associated with antibody response. (303 KB)

    Selected pathways from NCI-Nature Pathway Interaction Database whose expression at (a) day 3 or (b) day 7 post-vaccination was significantly (p < 0.01) correlated to antibody response. Genes are ranked according to Pearson correlation to antibody response, and the significance of pathways is tested by GSEA (see Method for details). Influenza LAIV is excluded here because this vaccine does not induce serum/plasma antibody responses. MCV4 has two sets of antibody correlation data: one against meningococcal polysaccharide (MCV4-PS) and another against diphtheria toxoid (MCV4-DT).

  7. Supplementary Figure 7: Examples of protein complexes detected during BTM construction. (842 KB)

    (a) Ribosome complex extracted from the master network. (b) Nuclear pore complex and (c) Splicesome complexes extracted from BTM modules. Each edge in the pictures represents a coexpression relationship detected in at least 3 publicly available microarray blood studies.

  8. Supplementary Figure 8: Examples of BTM modules. (262 KB)

    (a) “T cell differentiation via ITK and PKC” BTM module which contains genes whose functions are supported by literature. This set of genes, however was not found in pathway databases. (b) “Chromosome Y linked” BTM module contains many genes transcribed in Y chromosome.

  9. Supplementary Figure 9: Discriminative power of BTM modules compared with canonical pathways. (516 KB)

    Discriminative power is shown as t-scores in two-class comparison. The horizontal gray lines show the threshold of confidence level of α = 0.001. By this cutoff, the numbers of significant modules/pathways can be looked up on the X-axis (top two numbers inserted by curves). YF-17D data were from Querec et al 2009, Nature Immunology 10:116; MCV4 this study; Turberculosis from Berry et al 2010, Nature 466: 973; RTS,S from Vahey et al 2010 J Infect Dis 201:580. All testing data sets were never part of the BTM construction process. KEGG, BioCarta and NCI_PID are respective pathway databases. Chaussabel modules were version 2 based on Chaussabel et al 2008 Immunity 29:150. “Random” modules contain randomly selected genes, matched to the sizes of KEGG pathways.

  10. Supplementary Figure 10: Day 7 activity of plasma cells–immunoglobulin module (M156.1; Figure 4d) in blood transcriptome correlated to day 30 antibody response to TIV. (121 KB)

    The TIV 2007 (n=9) and 2008 (n=24) data sets were described in Nakaya et al 2011, Nature Immunology 12:786.

  11. Supplementary Figure 11: Hierarchical clustering of antibody correlations of BTM modules. (590 KB)

    Heat map of BTMs (rows) and vaccines (columns) whose baseline-normalized expression at day 3 correlated with baseline-normalized antibody response after vaccination (colors in map indicate Pearson correlation values). Clustering was performed (a) using all 346 BTM modules and (b) using the top 50% modules with highest variance across these 6 datasets.

  12. Supplementary Figure 12: Distinctive early transcription programs correlated to vaccine antibody response. (468 KB)

    (a) Clustering analysis from Supplementary Figure 10b, marked by the three programs: Primary viral response (b), Polysaccharide response (c) and Protein recall response (d). From each data set in (a), the top 20 modules with highest positive correlations and the top 20 modules with highest negative correlations were selected. Among these selected modules, the modules concordant between two datasets (Pearson correlation coefficient above 0.3 or below -0.3 in both datasets) are shown in (b, c, d) and the rest are shown in (e, f, g).

  13. Supplementary Figure 13: DC-related BTM modules and genes. (251 KB)

    A filled unit in the center grid indicates the membership of the gene (top axis) in the corresponding module (left axis). The heat map on the right shows the Pearson correlation between module activity and antibody response in each study. The bottom heat map shows correlation between module member genes and antibody response.

  14. Supplementary Figure 14: The DC surface signature module is positively correlated to polysaccharide response. (210 KB)

    (a) Network of “DC surface” module member genes. (b) Correlation between D3/0 module activity and D30/0 anti-polysaccharide IgG response in MCV4 and MPSV4.

  15. Supplementary Figure 15: Polysaccharide vaccine stimulates DCs. (470 KB)

    MPSV4 stimulates the phenotypic maturation of human myeloid DC, while MCV4 to a much lesser extent. FACS data were obtained after 24 h (n = 4–6).

  16. Supplementary Figure 16: MPSV4 stimulation of DCs is dependent on MyD88, TRIF, TLR4 and ASC in vitro. (176 KB)

    Expanded version of Figure 6 showing genes that were required for DC in vitro stimulation with MPSV4. CD11c+ DCs were isolated by MACS from spleens of C57BL/6 mice (WT) or various knock-out mice and stimulated for 24 h; *p<0.05, ***p<0.0001, unpaired t-test was used to compare WT vs. KO (n = 4–6).

  17. Supplementary Figure 17: Antigen-specific serum antibody and SBA titers induced by MCV4 and MPSV4 in mice in vivo. (240 KB)

    C57BL/6 mice were immunized sub–cutaneously with one human dose (a-c, n=4-9) or two doses (d, n=5) of MPSV4 (blue) or MCV4 (red) or left untreated. (a) Mice were bled after 7, 14, and 28 days and serum IgM, IgG and IgA levels against meningococcal serogroup C were measured. (b) Serum IgG against serogroups A, Y and W-135 were measured at day 28 and (c) Serum Bactericidal Activity (SBA) titers against serogroups A and/or C were measured after 28 days. (d) Mice were boosted with a 2nd dose at day 30. Each gray line represents an individual animal. * p<0.05, ** p<0.01, Mann-Whitney test.

  18. Supplementary Figure 18: Cell-type expression patterns in BTM-antibody analysis. (277 KB)

    BTM activities of cell surface signatures (see details in Suppl. Text), and their correlation to antibody response. X-axis shows module activity at D3/0 or D7/0, Y-axis the D30/0 antibody response.

  19. Supplementary Figure 19: BTM modules whose activity on day 7 is correlated to later antibody response in two or more data sets. (457 KB)

    Each segment on the circle corresponds to one vaccine dataset. In each segment, the inner circular bands show an ordered list of all BTM modules, layered by histograms of significantly correlated modules, red for positive correlation and blue for negative correlation. Each link inside the circles connects a significant module (top 50 in one dataset by p-value) that is common between different data sets. Modules common to three vaccines are linked in black, and their labels are marked by a black dot. Plasma and immunoglobulin modules are linked in green. An interactive version of this figure is available at online data portal (Interactive Figure 3).

  20. Supplementary Figure 20: B cell–related BTM modules and genes. (208 KB)

    A filled unit in the center grid indicates the membership of the gene (top axis) in the corresponding module (left axis). The heat map on the right shows the Pearson correlation between module activity and antibody response in each study. The bottom heat map shows correlation between module member genes and antibody response.

PDF files

  1. Supplementary Text and Figures (5,785 KB)

    Supplementary Figures 1–20 and Supplementary Note

Excel files

  1. Supplementary Table 1: Differences in gene expression in five vaccine studies. (11 KB)

    Numbers of differentially expressed genes compared to day 0 in five vaccine studies, at varying statistical cutoff by FDR or p-value, without (A) or with (B) gene filtering. Gene filtering is based on Nakaya et al. (2011) Nature immunology 12:786, using probesets with at least 1.25 fold change in > 20% subjects. p-values were based paired t-test. FDR was based on Storey and Tibshirani (2003), PNAS, 100:9440.

  2. Supplementary Table 2: Correlation of BTM activity to vaccine antibody responses. (64 KB)

    Pearson correlation coefficients are reported. For clarity, only absolute values greater than 0.25 are shown.

  3. Supplementary Table 3: GSEA analysis using the BTM modules. (34 KB)

    Genes are ranked according to Pearson correlation to antibody response, and the significance of BTM modules is tested by GSEA (see online method for details). Normalized Enrichment Score values are shown for p-values < 0.05.

Zip files

  1. Data and tutorial package of Blood Transcription Modules. (4,128 KB)

    This package provides BTMs in reusable data formats, and a step-by-step demonstration of their applications.
    [Order of files within the zip package:]
    BTM_application_tutorial.pdf
    README_package_content_description.pdf
    BTM_for_GSEA_20131008.gmt
    monocytes_vs_bcells.txt
    gene_ab_correlation.rnk MCV4_D3v0_probesets.txt
    btm_tool.py
    btm_example_data.py
    btm_annotation_table.xls

Additional data