Spatiotemporal gene expression is determined by enhancers, which interact with basal promoters to activate transcription. In genes where multiple enhancers and promoters co-exist at the same gene locus, individual enhancers target their specific promoters to activate. However, the underlying mechanism that regulates this enhancer-promoter interaction is unknown. In the current study, we focused on hunchback (hb), which contains two promoters: P1P, a maternal and late zygotic promoter, and P2P, an early zygotic promoter, which is activated by two Bicoid (bcd)-dependent enhancers: P2E and shadow. The shadow enhancer is located 4kb upstream of P2P but is close to P1P. By using a reporter system that mimics the endogenous hb locus and is capable of making a distinction between transcripts from two different promoters, we found that the shadow enhancer bypasses P1P and activates the more remote promoter P2P, and that P2E interacts with P2P, but not P1P. To investigate how these specific interactions between promoters and enhancers are regulated, we tested two hypothetical mechanisms: (i) motifs within the targeted promoter guide each enhancer to its specific promoter, and (ii) chromosome architecture, which facilitates contacts between shadow and P2E with P2P while blocking P1P. By manipulating TATA box sequences and spacing between regulatory elements in our reporter system and using single-molecule FISH (smFISH), we show that inserting or removing the TATA box of the basal promoters cause subtle changes in the relative RNA outputs of the two promoters, and that chromosome architecture plays a dominant role in preventing P1P from being activated. These results suggest that promoter motifs do affect a promoter’s interaction specificity with an enhancer, but whether an interaction happens in vivo depends more on the chromosome environment. Overall, our experiments provided evidence for a double-assurance mechanism that provides robust expression of P2P to achieve normal thoracic development.