The evolution of new functions has not been fully explained through standard population genetic theory. One alternative hypothesis is based on relaxing the assumption of strict genotype-to-phenotype maps; perhaps populations explore novelty through evolving phenotypic plasticity, allowing new forms and functions. Destabilizing mutations in bacteriophage λ’s receptor binding protein (RBP) cause the protein to fold into a range of confirmations, with differing receptor function. These observations led to a hypothesis for how viruses evolve to use new receptors and shift hosts (Receptor Expansion via Multimorphic Intermediates; REMI). Here we test REMI through the evolution of a new virus, bacteriophage 21 (Φ21), as it evolved to use two new receptors. Multiple lines of evidence were found: First, Φ21 evolves biphasic decay where an isogenic culture of Φ21 particles decays at different rates indicating phenotypic heterogeneity in particle stability. Second, this heterogeneity was found to be associated with the evolution of new functions since particles with novel function decay fastest. Third, we found evidence that this heterogeneity arises during protein folding by manipulating chaperone expression during Φ21 genesis and found that increased chaperones decreased the fraction of particles able to use the new receptors. This last test was not performed on λ, representing a novel test of REMI and is the first to directly link protein folding to the generation of heterogeneity in receptor use. Altogether we provide evidence for REMI in a new virus, lending support to the generalizability of REMI, and we provide a qualitatively new test of the hypothesis.