The retrograde transport inhibitor Retro-2 has a protective effect on cells and in mice against Shiga-like toxins and ricin. Retro-2 causes toxin accumulation in early endosomes and relocalization of the Golgi SNARE protein syntaxin-5 to the endoplasmic reticulum. The molecular mechanisms by which this is achieved remain unknown. Here, we show that Retro-2 targets the endoplasmic reticulum exit site component Sec16A, affecting anterograde transport of syntaxin-5 from the endoplasmic reticulum to the Golgi. The formation of canonical SNARE complexes involving syntaxin-5 is not affected in Retro-2-treated cells. By contrast, the interaction of syntaxin-5 with a newly discovered binding partner, the retrograde trafficking chaperone GPP130, is abolished, and we show that GPP130 must indeed bind to syntaxin-5 to drive Shiga toxin transport from the endosomes to the Golgi. We therefore identify Sec16A as a druggable target and provide evidence for a non-SNARE function for syntaxin-5 in interaction with GPP130.