Cluster analysis of host cytokine responses to biodefense pathogens in a whole blood ex vivo exposure model (WEEM)

Abstract

Background: Rapid detection and therapeutic intervention for infectious and emerging diseases is a major scientific goal in biodefense and public health. Toward this end, cytokine profiles in human blood were investigated using a human whole blood ex vivo exposure model, called WEEM.

Results: Samples of whole blood from healthy volunteers were incubated with seven pathogens including Yersinia pseudotuberculosis, Yersinia enterocolitica, Bacillus anthracis, and multiple strains of Yersinia pestis, and multiplexed protein expression profiling was conducted on supernatants of these cultures with an antibody array to detect 30 cytokines simultaneously. Levels of 8 cytokines, IL-1α, IL-1β, IL-6, IL-8, IL-10, IP-10, MCP-1 and TNFα, were significantly up-regulated in plasma after bacterial exposures of 4 hours. Statistical clustering was applied to group the pathogens based on the host response protein expression profiles. The nearest phylogenetic neighbors clustered more closely than the more distant pathogens, and all seven pathogens were clearly differentiated from the unexposed control. In addition, the Y. pestis and Yersinia near neighbors were differentiated from the B. anthracis strains.

Conclusions: Cluster analysis, based on host response cytokine profiles, indicates that distinct patterns of immunomodulatory proteins are induced by the different pathogen exposures and these patterns may enable further development into biomarkers for diagnosing pathogen exposure.

Figure 1

Scatter plots of 16 cytokine concentrations detected in human blood following ex vivo bacterial exposures. Cytokine concentrations were displayed on a logarithmic scale. The cytokines shown here were detected out of the 30 cytokines in the arrays. The 8 cytokines that were found to be statistically differentially expressed among these samples are highlighted with rectangular boxes. Each mark delineates the average of triplicate exposure samples. Each exposure sample is loaded onto a protein array chip that contains 5 independent measurements per cytokine meaning that fifteen measurements are used to obtain these data.

Figure 2

Concentrations of 8 cytokines in human whole blood after ex vivo exposure to pathogens. The control was a mock-exposed sample. Cytokine concentrations were determined using protein arrays. The bars represent the average of three replicate samples that each contain 5 replicate features per cytokine assay and the lines represent the standard deviation among the three replicates.

Figure 3

Clustering result with HOPACH using the average linkage distance between clusters is shown. The eight pathogen-exposed samples are clustered according to the dendrogram on the left and cluster into five groups, 1) Y. pestis (KIM5, NYC, and India), 2) Y. pseudotuberculosis, 3) Y. enterocolitica, 4) B. anthracis (Ames and Sterne), and 5) Control. Sixteen cytokines (Eotaxin, IL-10, IL-12(p40), IL-15, IL-1α, IL-1β, IL-6, IL-8, IP-10, MCP-1, MIG, TNFα, TRAIL, sCD23, sCD95, and sICAM-1) are also reordered based on their correlations according to the dendrogram on the top. Clusters go from root at top to leaf node for each cytokine. Clusters in between are based on their agglomerative . The branch shows the similarity, the short the branch, the more similar. In addition, the eight rightmost proteins form a cluster that may involve inflammation-related cascades initiated by an innate immune response to these pathogen. Colors represent units of log10 [pg/ml], in ten equally spaced intervals increasing from white to dark red. A key showing the specific log10 values for each interval is shown in the figure.