Most new proteins are made in the cytosol and then targeted to their mature destinations. In the case of integral membrane proteins (IMPs), this targeting concludes with integration into the target membrane, a process typically facilitated by protein-conducting channels. This thesis examines the two ubiquitous families of protein-conducting channels: SecY and Oxa1.

First, it presents an evolutionary analysis of SecY. This leverages the fact that SecY is a pseudodimer of two duplicated halves, which makes it possible to infer certain characteristics of those halves’ common ancestor, termed proto-SecY. Searches for other proteins structurally similar to proto-SecY detect exceptional similarity with the Oxa1 channels. This is quite surprising, since many studies have examined both channels and none noted this similarity. On the contrary, much has been made of their differences, especially the fact that Oxa1 channels don’t enclose a pore like SecY and other typical channels do. Instead they expose a groove inside the membrane that is open on one side. Nonetheless, the SecY and Oxa1 channels share a structurally similar core, which plays a similar role in a similar translocation mechanism and uses a similar set of amino acids. This indicates that proto-SecY evolved its translocation pore by bringing together the grooves of two Oxa1 proteins. This model provides a new paradigm for understanding how these channels work, and makes several interesting predictions, some of which are tested here. For example, it predicts that extant Oxa1 proteins may use the same interface as proto-SecY to form dimers, and indeed co-evolution analysis indicates that archaeal and eukaryotic Oxa1 proteins heterodimerise with a conserved partner via this interface.

Second, this work presents a structural study of how the eukaryotic SecY channel, Sec61, interacts with the transmembrane domains (TMDs) of nascent IMPs. Sec61 was co-purified with stalled ribosome-nascent chains which encoded TMDs of each orientation, N-in and N-out, that appear biochemically to bind to a recognition site in Sec61. Cryo-EM of these complexes showed that the N-in TMD opens the lateral gate much like a signal peptide (SP) does, despite the TMD being disordered, suggesting that N-in signals engage the channel in a similar way regardless of whether they are SPs or TMDs. Unlike the N-in TMD, the N-out TMD does not occupy the lateral gate, but instead appears bound to the opposite side of Sec61. This site is the closest to the ribosome exit tunnel, suggesting that tension in the nascent chain favours this site. Intriguingly, binding here places the TMD in the lipid-filled cavity formed by TMCO1 and other associated proteins.

Finally, this work presents a structural study of a yeast Oxa1 protein (Get1) bound to its conserved partner (Get2) and a cytoplasmic targeting factor (Get3). Cryo-EM analysis was technically challenging, but a preliminary dataset at least demonstrated particle orientations and per-particle image statistics that would be suitable for high-resolution reconstruction, if a sufficient number of particle images could be obtained. A structure of human Get1/2/3 that was published after this work is briefly reviewed.