Gin, with a twist

Robot next to protein dimer
13 March 2018

Much like a food mixer, with interchangeable tools that render it suitable for whipping, beating or liquidising, bacterial RNA polymerase has an exchangeable subunit called a sigma factor which adapts it to different programmes of gene expression. Sigma factors are controlled by other proteins called anti-sigma factors, to make sure the polymerase doesn’t whip when it should be liquidising.

One of these is called Gin (and also CsfB), which acts as a safety lock on two specific sigma factors during the sporulation process. In addition to serving as a premier model system for studies of regulation, understanding sporulation is important for tackling food safety and the ‘hospital superbug’ problem. Recently, the structure of Gin was solved using Nuclear Magnetic Resonance (NMR) spectroscopy. Authors Santiago Martínez-Lumbreras and Rivka Isaacson take up the story:


The course of structure solution by NMR rarely runs smooth and our journey towards publication in ‘Structure’ was particularly convoluted. CsfB (aka Gin) is a small protein with a big job in transforming B. subtilis bacteria into hardy spores when they are threatened with harsh survival conditions.

In order to solve a structure by NMR, its necessary to work out which atoms are close to each other, using something called the Nuclear Overhauser Effect (NOE). Each NOE measurement can be used like a virtual elastic band, pulling two specific atoms near to one another. With hundreds of these elastic bands, the shape of the whole molecule can be determined. Solving structures of homodimers by NMR comes with specific challenges, as we have previously experienced (PDB entries 4ASV and 4CPG), due to the added complication of first needing to know to which molecule the atoms belong. That determines if the elastic band is between two atoms in the same, or different, molecules.


Gin Mixer

One of the ways round this is to make mixed dimers (with one monomer labelled using Carbon-13 and Nitrogen-15) to distinguish between inter- and intramolecular restraints. After killing our protein many times over, we successfully produced a mixed dimer of CsfB by gently warming a labelled/unlabelled mixture to 50 ºC and cooling it slowly.

We completed the NMR experiments, assigned the chemical shifts of the atoms and acquired four through-space experiments to measure the NOEs. We also knew there was zinc ion in there somewhere. Using ARIA software, we ran the first round of structure calculation, letting all those virtual elastic bands pull the structure into shape.


A robot... with a very mysterious finger

Our first results were exciting. They showed a dimer with four cysteines close enough to be coordinating the zinc in a ‘zinc finger’ motif, but the zinc finger was remarkable - three of the cysteines were from one protein molecule and the fourth was from the other. This ‘swapped’ dimer was a highly unusual structure with a strange resemblance to Roberto the Futurama robot! Using these data as our starting point we defined the zinc finger coordination and continued with the structure refinement.

We loved our unusual structure, and began to speculate about the possibilities it held for its function in sigma factor regulation. We had even started designing some experiments with our collaborator, Dr. Amy Camp at Mount Holyoke College, to test it in vivo, but we slowly began to realise that something was not quite right. As we continued refining the data, the stats were just not improving enough. There were no big violations, but the configuration of the ‘swapped’ region was a bit odd; it had poor Ramachandran stats and the expected torsion angles derived from the chemical shifts - TALOS - didn’t agree with what we were calculating.

We were a bit worried that we may be forcing the structure to be swapped when it really wasn't, so we went back to the beginning. We removed the imposed zinc finger coordination and, after carefully examining the spectra by hand, found more NOEs we could use. Although the input had not changed much, it gave rise to a non-swapped version of the structure in which the zinc coordination boringly involved four cysteine residues from the same molecule.


Detail of a swapped zinc finger

Detail of the zinc finger. In red is non-swapped protein where all four cysteines coordinate the zinc.

The swapped dimer is shown in pink and grey. Three cysteines come from the pink protein, but the fourth comes from the grey chain.


Which is right? Only one way to NOE

After this calculation we had two models (swapped and non-swapped) and we genuinely did not know which one was correct. A close analysis of our two possible structures showed that the NOEs describing them were very similar except in the N-terminal region, where some intermolecular NOEs in the swapped version would be intramolecular in the non-swapped.

At this point we performed an exhaustive manual assignment of the NOE data and acquired another, higher quality version at higher magnetic field (thanks to the NMR facility in Oxford). We analysed the structures in depth to find any intermolecular NOEs that should be specific to either case.

We realised then that the swapped structure would show additional intermolecular NOEs that clearly did not appear in the spectra, and also that a few of our NOEs could be only explained by the non-swapped model (these had been rejected in earlier calculations). It became clear that the non-swapped structure matched with the TALOS data and had good Ramachandran statistics. Finally letting go of our quirky structure we fixed the zinc finger in the traditional form and refined our structure which is now deposited in the PDB as 5N7Y.


A musical ending to a cautionary tale

Gin dimer and treble clefs

CsfB adopts a treble-clef zinc finger fold, which is common for zinc finger proteins but our dimer (in red and pink above) looks like two treble clefs stuck together. We found only one other example like this in the PDB, part of a bacterial protease called ClpX. So our structure might not be as exciting as we first thought, but it is still unusual.

We wanted to share our cautionary tale, because it led us on a long wild goose chase, and was a huge part of this work which could obviously not be included in the final paper. Hopefully it will save others some time which could be better spent drinking, rather than studying, Gin…