AcrAB-TolC is the paradigm resistance-nodulation-division (RND) multidrug resistance efflux system in Gram-negative bacteria, with AcrB being the pump protein in this complex. We constructed a nonfunctional AcrB mutant by replacing D408, a highly conserved residue essential for proton translocation. Western blotting confirmed that the AcrB D408A mutant had the same native level of expression of AcrB as the parental strain. The mutant had no growth deficiencies in rich or minimal medium. However, compared with wild-type SL1344, the mutant had increased accumulation of Hoechst 33342 dye and decreased efflux of ethidium bromide and was multidrug hypersusceptible. The D408A mutant was attenuated in mouse and models and showed significantly reduced invasion into intestinal epithelial cells and macrophages A dose-dependent inhibition of invasion was also observed when two different efflux pump inhibitors were added to the wild-type strain during infection of epithelial cells. RNA sequencing (RNA-seq) revealed downregulation of bacterial factors necessary for infection, including those in the pathogenicity islands 1, 2, and 4; quorum sensing genes; and Several general stress response genes were upregulated, probably due to retention of noxious molecules inside the bacterium. Unlike loss of AcrB protein, loss of efflux function did not induce overexpression of other RND efflux pumps. Our data suggest that gene deletion mutants are unsuitable for studying membrane transporters and, importantly, that inhibitors of AcrB efflux function will not induce expression of other RND pumps. Antibiotic resistance is a major public health concern. In Gram-negative bacteria, overexpression of the AcrAB-TolC multidrug efflux system confers resistance to clinically useful drugs. Here, we show that loss of AcrB efflux function causes loss of virulence in serovar Typhimurium. This is due to the reduction of bacterial factors necessary for infection, which is likely to be caused by the retention of noxious molecules inside the bacterium. We also show that, in contrast to loss of AcrB protein, loss of efflux does not induce overexpression of other efflux pumps from the same family. This indicates that there are differences between loss of efflux protein and loss of efflux that make gene deletion mutants unsuitable for studying the biological function of membrane transporters. Understanding the biological role of AcrB will help to assess the risks of targeting efflux pumps as a strategy to combat antibiotic resistance.