Many insect vectors of disease detect their hosts through olfactory cues, and thus it is of great interest to understand better how odors are encoded. However, little is known about the molecular underpinnings that support the unique function of coeloconic sensilla, an ancient and conserved class of sensilla that detect amines and acids, including components of human odor that are cues for many insect vectors. Here, we generate antennal transcriptome databases both for wild type Drosophila and for a mutant that lacks coeloconic sensilla. We use these resources to identify genes whose expression is highly enriched in coeloconic sensilla, including many genes not previously implicated in olfaction. Among them, we identify an ammonium transporter gene that is essential for ammonia responses in a class of coeloconic olfactory receptor neurons (ORNs), but is not required for responses to other odorants. Surprisingly, the transporter is not expressed in ORNs, but rather in neighboring auxiliary cells. Thus, our data reveal an unexpected non-cell autonomous role for a component that is essential to the olfactory response to ammonia. The defective response observed in a Drosophila mutant of this gene is rescued by its Anopheles ortholog, and orthologs are found in virtually all insect species examined, suggesting that its role is conserved. Taken together, our results provide a quantitative analysis of gene expression in the primary olfactory organ of Drosophila, identify molecular components of an ancient class of olfactory sensilla, and reveal that auxiliary cells, and not simply ORNs, play an essential role in the coding of an odor that is a critical host cue for many insect vectors of human disease.