Microarray experiment setup and analysis. (A) The experimental setup of the microarray study. The 28-hpf time point for microarray analysis in both experiments was chosen after the onset of blood circulation in the embryo and to ensure the abundance of myeloid cells of the innate immune system as shown by lcp1 in situ staining of embryos treated with the standard control (SC) morpholino. The lack of lcp1 staining in spi1 morphants confirms the absence of myeloid cells. GFP⁺ myeloid cells were isolated from embryos expressing membrane-targeted EGFP under control of the spi1 promoter. The 2-step approach eliminated false positives because of leaky expression of the spi1:GFP transgene in the brain and allowed determination of the specific effect of spi1 knockdown on myeloid cells. (B) Primary analysis of microarray data. For initial comparative analysis we chose an arbitrary cut-off of absolute fold change larger than or equal to 1.2 and a P value ≤ .01. We found 3551 sequences to be down-regulated and 3684 sequences to be up-regulated in the spi1 knockdown embryos. In the GFP⁺ myeloid cell fraction, 6327 sequences were enriched and 6335 sequences were reduced compared with the GFP fraction. The overlap between the sequences found down-regulated in spi1 morphants and enriched in the GFP⁺ myeloid cell fraction was estimated and filtered against the sequence set found down-regulated in control morpholino-injected embryos. (C) GO term annotation analysis of the 249 genes found in the overlap. Category 1 indicates zebrafish genes with at least one GO term annotation; category 2, genes without any annotation in zebrafish but having a homolog with known GO term annotation in humans; and category 3, genes without any GO term annotation either in zebrafish or in humans.