Sources + sinks = swarmers + stalks: a signaling activity gradient within a bacterial cell

You’re probably familiar with the idea that gradients of signaling molecules determine cell fate in early animal embryos.  Now, a recent paper from the Laub lab (Chen et al. 2010.  Spatial gradient of protein phosphorylation underlies replicative asymmetry in a bacterium, PNAS doi:10.1073/pnas.1015397108) has discovered gradients of active signaling molecules within a single bacterial cell.  The existence of this gradient had been missed until now because the signaling protein itself, CtrA, is evenly distributed across the dividing cell.  But, Chen et al. find, the active phosphorylated form of the protein, CtrA-P, is not.  You’ll have to read a little further to find out why not.

First, the model system.  Caulobacter crescentus is unusual among the familiar bacterial model organisms, in that it divides asymmetrically; it’s used as a model system for studying cell differentiation.  There are two forms, called “stalked” and “swarmer”: the stalked form is able to copy its DNA and divide, producing one stalked daughter and one swarmer daughter; later, the swarmer daughter may differentiate and become stalked, and then have daughters of its own.  The protein that stops swarmers from undergoing replication is CtrA, a transcription factor that, when activated by phosphorylation, binds to the bacterial replication origin and prevents the initiation of DNA synthesis.  So, in theory, one way you could give stalked and swarmer cells different fates is to wait until the cell divides into two (cytokinesis), then selectively dephosphorylate CtrA-P in stalked cells.  By making CtrA-P levels high in swarmer cells and low in stalked cells, you would block division in swarmers but not in stalked cells.

This theory made perfect sense, but turns out not to be true.  The first clue that something else was going on came when Chen et al. blocked C. crescentus cells from dividing using the antibiotic cephalexin.  Instead of having two chromosomes, as you would expect after replicating DNA but not dividing, they found that many of the blocked cells had three chromosomes.  Not four: so of the two chromosomes from the original replication event, only one was able to re-replicate.  What’s more, when they used fluorescent proteins to label replication origins, they found that the third chromosome came from re-replication at the stalked end of the cell over 80% of the time.  So, the dividing cell already has a highly asymmetrical ability to replicate DNA even before the daughter cells divide.

In animal embryos, you make a gradient by producing protein at one end of the embryo and not the other: the diffusion of the protein is what sets up the gradient.  But bacterial cells are way too small for this kind of mechanism to work: a protein can travel from one end of the cell to the other in seconds.  And indeed when Chen et al. measured the distribution of CtrA, using a YFP-labeled reporter, they found that it was uniform across the cell.  But the unphosphorylated form of the protein is not the active form: could it be that the distribution of activity is asymmetrical?

The protein responsible for phosphorylating CtrA,  CckA, is found at both poles of the dividing cell.  It’s not a simple kinase, though. The authors have previously shown that CckA is bifunctional: it can act as a CtrA-P phosphatase as well as a CtrA kinase. A recent paper suggests that CckA requires a second protein, DivL, for its kinase activity.  And DivL turns out to be asymmetrically distributed; it’s found only at the swarmer pole. So the prediction would be that CckA is only active as a kinase at the swarmer pole: what is it doing at the stalk pole?

Looking at this arrangement of proteins, Chen et al. speculated that perhaps the CckA at the stalk pole is active as a phosphatase instead of as a kinase. If that were true, maybe you could make a gradient of CtrA-P by producing CtrA-P at the pole of the cell where CckA/DivL is located, and removing it at the pole of the cell that only has CckA (see figure).  The kinase activity at the swarmer pole provides a “source”, while the putative phosphatase activity at the stalk pole would provide a “sink”.

Would that work, given how quickly proteins can diffuse from one end of the cell to the other?   A reaction-diffusion model suggests that it could: if the lifetime of the phosphorylated molecule is in a similar range to the time required to diffuse from one end of the cell to the other (~1–10 seconds), you would expect the gradient across the cell to be significant.  Chen et al. measured the half-life of CtrA-P, and found that indeed it’s in the right range for gradient formation (~10 seconds).

It’s not easy to measure CtrA-P directly in living bacteria, but it is possible to change the shape of the gradient by modulating the location and activity of the kinase and (putative) phosphatase and asking how that affects the observed asymmetry in DNA replication ability.  If the bias in DNA replication activity towards the stalked pole goes away, that would suggest that the CtrA-P gradient is also gone.  The authors test their hypothesis using a number of temperature-sensitive and other mutants, finding effects that are in line with their theory.  Most convincingly, replacing wild-type CckA with a CckA mutant they previously discovered that has normal kinase activity but decreased phosphatase activity almost completely removes the bias in DNA replication.  In further modeling work, they conclude that almost the only thing you need to create an effective CtrA gradient is a source and sink that work on the same time scale as diffusion (or faster); also one, but not both, of the source/sink activities needs to be located at a pole.  Protein synthesis and proteolysis are too slow to make a difference to the gradient.

Chen et al. use their discussion section to point out the “pressing need” for phosphorylation-state-specific reporters of bacterial proteins — who knows how many other spatially asymmetric, diffusible, transient signals we may be missing?  [And it would certainly have been easier for them to do the work with a more direct assay, not to mention making it easier to deal with reviewers.]  Although bacterial cells are certainly less spatially complex than eukaryotic cells, it seems they still have a number of tricks up their metaphorical sleeves.

Chen YE, Tropini C, Jonas K, Tsokos CG, Huang KC, & Laub MT (2010). Spatial gradient of protein phosphorylation underlies replicative asymmetry in a bacterium. Proceedings of the National Academy of Sciences of the United States of America PMID: 21191097