The nuclear pore complex (NPC) is a massive cylindrical structure that bridges the inner and outer membranes of the nuclear envelope and acts as a gatekeeper, preventing the passive diffusion of large biomolecules into and out of the eukaryotic cell nucleus. This 50-MDa, highly symmetric complex is comprised of approximately 30 different nucleoporins, and the enormity of the NPC has meant that there is limited information about the overall structure and relative orientations of the various components. Amlacher et al. previously used a yeast two-hybrid analysis to determine nucleoporin-nucleoporin interactions, but they found that many of the yeast nucleoporins were not stable in vitro, which prohibited the authors from conducting follow-up biochemical and structural studies. For this reason, they turned to Chaetomium thermophilum, a thermophilic fungus whose recombinant nucleoporins were well behaved in vitro. The authors first used recombinant nucleoporins from this thermophile to confirm, in vitro, the protein-protein interactions seen by yeast two-hybrid analysis. They then used electron microscopy to examine the structures of individual nucleoporins and various co-complexes. These analyses revealed that Nup192 and Nup188 have similar S-shaped structures, despite their limited sequence homology; the shape and curvature of half of the 'S' shape is reminiscent of the structures of karyopherin transport receptors such as exportin-t and Crm1. This raises the intriguing possibility that those transport receptors evolved from an ancestral nucleoporin, perhaps one that lost its ability to assemble into the NPC and also gained a nuclear export sequence motif. By integrating the information from their biochemical and structural studies, the authors were able to propose a model for the inner pore ring complex of the NPC. In this model, Nup192 and Nup170, which are fairly large, are linked together through interactions with short, flexible regions of Nup53 and Nic96. Because 73% of the 5,797 Saccharomyces cerevisiae proteins and 44% of the 22,937 human proteins have homologs in C. thermophilum, the authors propose that this thermophilic fungus will be useful for structural studies of other eukaryotic protein complexes, especially when the yeast or human variants are unstable or poorly behaved in vitro. (Cell 146, 277–289, 2011)