From transcriptome to cytome: integrating cytometric profiling, multivariate cluster, and prediction analyses for a...

Gene expression studies of peripheral blood cells in inflammatory diseases revealed a large array of new antigens as potential biomarkers useful for diagnosis, prognosis, and therapy stratification. Generally, their validation on the protein level remains mainly restricted to a more hypothesis-driven manner. State-of-the-art multicolor flow cytometry make it attractive to validate candidate genes at the protein and single cell level combined with a detailed immunophenotyping of blood cell subsets. We developed multicolor staining panels including up to 50 different monoclonal antibodies that allowed the assessment of several hundreds of phenotypical parameters in a few milliliters of peripheral blood. Up to 10 different surface antigens were measured simultaneously by the combination of seven different fluorescence colors. In a pilot study blood samples of ankylosing spondylitis (AS) patients were compared with normal donors (ND). A special focus was set on the establishment of suitable bioinformatic strategy for storing and analyzing hundreds of phenotypical parameters obtained from a single blood sample. We could establish a set of multicolor stainings that allowed monitoring of all major leukocyte populations and their corresponding subtypes in peripheral blood. In addition, antigens involved in complement and antibody binding, cell migration, and activation were acquired. The feasibility of our cytometric profiling approach was demonstrated by a successful classification of AS samples with a reduced subset of 80 statistically significant parameters, which are partially involved in antigen presentation and cell migration. Furthermore, these parameters allowed an error-free prediction of independent AS and ND samples originally not included for parameter selection. This study demonstrates a new level of multiparametric analysis in the post-transcriptomic era. The integration of an appropriate bioinformatic solution as presented here by the combination of a custom-made Access database along with cluster- and prediction-analysis tools predestine our approach to promote the human cytome project.