Full paper can be found here: http://dx.doi.org/10.1038/s41586-018-0693-y
This post was a joint contribution from Justin C. Deme and Frédéric Lauber.
What do chronic periodontitis in humans and columnaris disease in fish have in common? Both are caused by Gram-negative pathogens of the phylum Bacteroidetes, which have a common but poorly characterized protein translocation pathway termed the type IX secretion system (T9SS). This system is involved in the recognition and secretion of folded protein substrates, many of which are virulence factors, into the extracellular environment. It is also required for the transport and attachment of adhesins to the cell surface that allow bacteria to glide over solid surfaces. At the heart of this bifunctional system was a yet-to-be-identified outer membrane (OM) translocon.
How folded proteins are transported across membrane bilayers, without compromising integrity, has always been a fascinating biological quandary; we sought to understand this dilemma for the T9SS. Some educated guesses later, we hypothesized that the 270 kDa OM protein SprA was an obvious candidate for the job. And based on our study, we now have a structural understanding of this protein’s function - but it was not without challenges!
To understand the function of this protein, we needed to purify it. We proceeded to chromosomally tag SprA and purify it from a native organism. However, it became evident that there were only about 7 copies of SprA per bacteria. This meant that from 16 L of culture, we could purify roughly 30 µg of material. A crystallography target, this was not!
Thankfully the golden age of cryo-electron microscopy (cryo-EM) is upon us. Lo and behold, after countless late nights involving protein preparations, detergent screens and grid optimizations, we were finally imaging clear, vitrified particles. Our excitement, however, was short-lived. Despite the amazing level of detail demonstrated in our 2D class averages, our particles were clearly biased towards certain orientations, which made 3D volume reconstructions impossible.
Back to the drawing board.
Or not. We did, after all, have some tricks up our sleeves. The most successful was doping our purified material with fluorinated detergents just before plunge-freezing. Incredibly, this flipped our particles over almost 90 degrees, which meant that we could now fill enough of projection space to reconstruct SprA in 3D.
It became apparent early into the classification/map reconstruction process that there were two unique SprA states and that both states were too big to be made of only one protein. Based on our proteomics data, we confirmed that SprA was co-purifying with three partner proteins, one of which was PorV, an essential T9SS component involved in substrate maturation after translocation across the OM. With this information in hand and the help of structural homology searches we could assign sequence to the map densities of these partner proteins. However, the biggest hurdle was yet to come: the daunting 2,200 amino acids of SprA that had to be manually and painstakingly built into the EM density. Thankfully the level of detail in our maps (and the high level of connectivity) was sufficient to rationally build sequence into the density for SprA, and after several late night sessions and lots of coffee, we were rewarded with the final structure. And we were in for a treat.
We anticipated SprA to be an impressively large protein, but not to the extent we saw post-model building; it currently is the largest single chain β-barrel protein weighing in with a whopping 36 β-strands. Not only is its size impressive, but the overall complexity of its extracellular domains (appropriately called “cap” domains) make it markedly different to its smaller cousins LptD and FimD. These cap domains extend far beyond the membrane and completely block the extracellular side of the protein. Interestingly, the segments making up this cap are interwoven with the barrel strands, raising a host of questions regarding the biogenesis of this protein.
PorV and Plug SprA complexes, captured by Cryo-EM. In the PorV complex, the periplasmic side of SprA is freely accessible, presumably to allow substrate entry, while the lateral gate is occluded by PorV. In the Plug complex, the lateral gate is now open, tentatively allowing exit of substrate, while the periplasmic opening is sealed by the Plug protein. In both complexes, the PPI protein is attached to the extracellular side of SprA.
Another fascinating feature of SprA is the gaping hole within the 36-stranded barrel, termed the “lateral gate”. This gate is occupied by PorV when bound to SprA, but open in PorV’s absence. We predict this gate permits PorV to reach into the barrel lumen of SprA and recognize incoming substrate; but this is a story for another day.