Membrane proteins are targets of most available pharmaceuticals, but they are difficult to produce recombinantly, like many other aggregation-prone proteins. Spiders can produce silk proteins at huge concentrations by sequestering their aggregation-prone regions in micellar structures, where the very soluble N-terminal domain (NT) forms the shell. We hypothesize that fusion to NT could similarly solubilize non-spidroin proteins, and design a charge-reversed mutant (NT*) that is pH insensitive, stabilized and hypersoluble compared to wild-type NT. NT*-transmembrane protein fusions yield up to eight times more of soluble protein in Escherichia coli than fusions with several conventional tags. NT* enables transmembrane peptide purification to homogeneity without chromatography and manufacture of low-cost synthetic lung surfactant that works in an animal model of respiratory disease. NT* also allows efficient expression and purification of non-transmembrane proteins, which are otherwise refractory to recombinant production, and offers a new tool for reluctant proteins in general.