Abstract
Genetic and functional studies have demonstrated that FL cells retain the major features of normal germinal center (GC)-derived B cells, in particular the dependency on an active crosstalk with their specialized microenvironment. In agreement, microarray analyses have recently revealed that FL patient outcome is primarily predicted by molecular characteristics of tumor-infiltrating immune cells instead of tumor cells. However, our knowledge of the crucial interactions between malignant and non-malignant cells in FL remains limited by the use of whole biopsy specimen to perform gene expression profiling (GEP). We thus conducted GEP on both CD19pos B cells and CD19negCD22neg non-B cells purified from lymph nodes of 17 patients with de novo FL & 4 normal donors (CD20pos >94.5%, median=98.2%) and 9 de novo FL patients & 5 normal donors (CD20pos<6.7%, median=0.5%), respectively. Biotinylated cRNA were amplified according to the small sample labelling protocol and hybridized onto HGU133 Plus 2.0 arrays (Affymetrix). Raw data were normalized using GC-RMA methodology (ArrayAssist, Stratagene) and finally, based on a CV>80, 10870 probesets were selected for further analyses. Unsupervised hierarchical clustering (Eisen’s software) allowed the correct classification of the 35 samples into the 4 groups: FL B-cell, Normal B-cells, FL non-B cells, and Normal non-B cells. Supervised analyzes were done using asymptotic non-parametric Mann-Whitney U-test (fold change ≥2, P<0.01) and confirmed by permutation analysis (500 permutations, false discovery rate <5%) using SAM software. We first established the list of the 841 probesets that were differentially expressed between FL and normal B-cells containing, 355 probesets overexpressed in malignant B cells including genes involved in GC B-cell biology (BCL6, MTA3, ID2, CD80, SDC4) and oncogenes as well (BCL2, AURK2) and conversely, 486 probesets downregulated in malignant B cells involving several interferon-stimulated genes for example. We then looked for the FL-specific microenvironment signature and pointed out the 1206 probesets that were differentially expressed between FL and normal non-B cells. Interestingly, all these genes were upregulated in the lymphoma context. Among them, we identified a striking follicular helper T-cell (TFH) signature (CXCR5, ICOS, CXCL13, CD200, PDCD1, SH2D1A) and an activated T-cell signature (IFNG, FASLG, GZMA, ZAP70, CD247). Notably, the TFH and activated T-cell signatures were not merely a surrogate for the number of T cells since many standard T-cell genes (i.e. CD2, CD4, CD7, LEF1, CD8A) were not induced in the FL microenvironment. Finally, in order to draw an overview of the FL-specific synapse between B and non-B cell compartments, we isolated a group of 2323 probesets that were differentially expressed between both compartments in FL and not in normal context. Using Ingenuity Pathway Analysis software we then identified among them FL-specific functional networks, including an IL-4- & an IL-15-centered pathway. Altogether, these data shed new light on our understanding of FL biology and could be a source of new therapeutics targeting the interplay between B cells and their microenvironment.
De NovoTranscriptome Assembly and Differential Gene Expression Profiling of ThreeCapra hircusSkin Types during Anagen of the Hair Growth Cycle