Energy landscape theory has developed into a dominant view of protein folding, and also provides insights into protein function and conformational changes. One tenant of energy landscape theory is the existence of a folding funnel which gives proteins multiple routes of folding. The principle of minimal frustration is an aspect of funnel theory that supposes energetic traps are very small on the landscape compared to the overall basin. GFP is a unique [Beta]-barrel protein used for fluorescent labeling, as it requires no cofactors and becomes fluorescent by an autocatalytic reaction of its own backbone. Here, we study the unique folding of GFP using a combination of theoretical and experimental techniques, and study an observed hysteresis in folding equilibrium-type experiments. Proline residues in the barrel lid are found to be a novel requirement for chromophore formation, in addition to previously known catalytic residues. In the absence of the chromophore, hysteresis is abolished, suggesting that protein function causes landscape roughness. Theoretical work points to a unique funneled energy landscape of GFP containing two basins, with the "extra" basin representing a loosely-packed intermediate {Niso} where the final strand is slow to pack into the barrel, forming the locked, native form {Nnat}. Collapse into the {Niso} basin occurs on a minute timescale, while the configurational search between {Niso} and {Nnat} occurs over months. As a typical equilibrium experiment is observed over days, between these two timescales, and hysteresis is observed. Further experimental characterization of the {Niso} intermediate shows a structure with heterogeneity in the lid of the barrel, which is evidence of the predicted loose structure. In addition, changes in the chromophore are visible, consistent with changes in chromophore flexibility or isomerization. Due to the bulky chromophore, final packing of the barrel is problematic, with the final step in folding dominated by the configurational search, and potential mispacking of the barrel. During refolding, a threshold of stability is required to capture the native structure. During unfolding, random fluctuations in the barrel may lead to repacking of the chromophore, creating a less stable protein, which immediately unfolds under hysteretic conditions, causing the hysteresis observed