Written by Hannah Edstrom
In the last post, we introduced the paper and explained the logic behind their methods. Here, we’ll look at what it takes in lab to solve the problem.
Biofilms are a dangerous group behavior of pathogenic bacteria. In humans, biofilms serve to protect the pathogen from antibiotics. Biofilm formation is also often associated with the release of virulent factors. In the case of P. aeruginosa, a pathogen of plants as well as of people, those virulent factors are pyocyanin and elastase and they release it whenever they organize into a biofilm.
This kind of coordinated behavior by bacteria requires some communication between cells. Bacteria communicate via quorum sensing. Pathogens use it as an offensive strategy against their host: if one bacterium releases pyocyanin, the plant is likely to fight it off successfully. But if the bacteria wait until they can sense a large number of kindred microbes, and then become virulent, they up their chances of success against the immune system of their victim.
But chemical warfare is afoot. The authors identified chlorogenic acid and rosmarinic acid (RA) as potential QS signals – but these chemicals are made by plants, not P. aeruginosa or other bacteria, for that matter! Do these plant-based chemicals really have the ability to interfere with bacterial messaging systems? If so, can they, in turn, influence the behavior of their infection?
What did they do in lab to answer their questions?
The first step to identifying the affects of RA and chlorogenic acid is to test if they actually bind to the regulatory proteins present in P. aeruginosa’s cells. They did this with isothermal titration calorimetry, a tool that measures energy differences that result from molecules making & breaking bonds in a chemical reaction.
Working In Vitro
The binding of RA to RhlR (a P. aeruginosa regulatory protein) would result in an increase of negative energy (measured in microcalories, or µcal). The way an microcalorimeter works is by holding two liquid containers next to each other and measuring differences in heat. There is a control container that always remains at the same temperature, and a container holding the chemical reaction. This piece holds your binding protein (RhlR, or LasR), and you inject substrate (RA or chlorogenic acid) at set times (hence the multiple peaks in the top halves of Fig. 1A, 1B, and 1D). Each time you mix substrate in with the protein, if the protein and the substrate bind, the reaction warms or cools depending on the enthalpy of the chemical reaction.
The microcalorimeter senses this heat change, and responds by acting to bring the container back to the same temperature as the control. It is this attempt to change the heat that is measured in the graphs.
Each time you add substrate, you are using up regulatory protein (if it binds). Thus, as you continue adding it, there is less and less regulatory protein available, and fewer reactions occur (you can see this in the decreasing peak size in the Fig. 1 graphs). How fast does the reaction occur? That’s what the second half of Fig. 1A, 1B, and 1D tell you – the faster the substrate gets used up, the steeper the molar ratio curve.
In Fig. 1A, we are seeing the reaction of RhlR with its substrate, the bacteria-produced QS signal, C4-HSL. It binds pretty well. Likewise, Fig. 1B shows the binding of LasR with its bacteria-produced substrate, 3-Oxo-C12-HSL. In Fig. 1D, we are using RhlR as our protein, but here, we are testing to see if it binds with RA. We do see binding, and what’s more, the molar ratio curve (lower half) shows more efficient binding with the protein. So plants produce RA to not only communicate with bacteria, but to communicate better than the bacteria’s intended substrate! (RA was also tested against LasR in Fig. 1E, in which case we see no binding.)
So. We have confirmed that RA binds to RhlR. Does it activate the regulatory protein? If C4-HSL binds to RhlR, after all, it will get transcription started so that the bacteria will start forming biofilms and producing pyocyanin and elastase. RA binds to RhlR – does it also affect transcription? Here, the authors conducted in vitro (that is, in-glass) experiments in which they mixed DNA+RhlR with RA, and DNA+RhlR with C4-HSL. They didn’t just use any DNA, of course – they got an operon controlled by RhlR in P. aeruginosa. Running a transcription assay on a gel (Fig. 3A) showed clear presence of DNA induced by both RA and C4-HSL exposure. But quantification was also needed. In Fig. 3B, we see each the quantified density of DNA produced by in vitro reactions signaled by C4-HSL, RA, and chlorogenic acid.
Chlorogenic acid works the least at both ligand concentrations, but it still causes some RhlR-mediated transcription. C4-HSL is a kind of control: this is how much transcription you would expect if the bacteria itself was using its QS signaling system. Thus, we can see that plant-produced RA stimulates bacterial transcription better than bacteria-based C4-HSL.
Not only is the plant interfering with its pathogen’s QS system, it is doing a really good job at it.
Moving In Vivo
The next step? Move this reaction in vivo – put it in the real thing, P. aeruginosa. The authors decided to do this twice: in E. coli, and in P. aeruginosa. E. coli is our favorite lab-rat species in molecular biology labs, so it’s often chosen to start with.
However, the RhlR/C4-HSL QS system is native to P. aeruginosa, not E. coli. So the whole pathway had to be inserted into E. coli to check for activity. Fig. 4 showcases the data used for this experiment.
First, they inserted a plasmid (called pMULTIAHLPROM) into E. coli. This plasmid contained both RhlI and LacZ. Why these two? The authors stole the promoter from RhlI, which codes for C4-HSL; it’s a promoter that activates under RhlR regulation. And lacZ synthesizes β-galactosidase, an easy chemical to measure in standard assay methods. In other words, now the easily measured β-galactosidase will behave as part of the RhlR QS system.
The second plasmid they inserted was either a control plasmid (an empty vector; no gene present) or a plasmid containing the genes for RhlR expression.
In this way, the idea was: add these genes into E. coli, then expose the cells to RA and see if it stimulates any transcription in the cell. Simply enough, except in an experiment, you’ve got to have plenty of controls to test your work. So the authors also used DMSO (a negative control, predicted to have no affect), C4-HSL (a control depicting what kind of affect the usual bacterial signal would have), and chlorogenic acid (“CA” in Fig. 4, and showing some, but not much, transcription in the previous in vitro experiment).
In Fig. 4A, we see the affects of each potential QS signal – DMSO, C4-HSL, RA, and CA – on β-galactosidase activity. Black bars denote cells carrying the empty vector, and white bars denote cells carrying the whole pathway. Knowing this, we can see that there is no statistically significant affect of DMSO or chlorogenic acid on RhlR-regulated gene activity (which, here, has been manipulated to include β-galactosidase production). But C4-HSL and RA both increase the transcription. In this case, however, C4-HSL has the greater affect.
So the authors repeated the same in P. aeruginosa (except here, they only needed to add pMULTIAHLPROM; the RhlR expression gene was already present, of course). And here, in Fig. 4B, they show that not only does RA stimulate higher β-galactosidase production (all blue-RA bars are higher than the orange-C4-HSL bars), it stimulates production before C4-HSL starts to act (the x-axis here is a time measurement, so RA’s affects are shown to occur before C4-HSL’s.)
This adds more to the RA-as-a-plant-defense-mechanism story! RA can stimulate transcription in the pathogenic P. aeruginosa, and it stimulates more transcription than the bacterial signal, and it stimulates it before the bacterial signal has a chance to act. This is some serious hijacking of the bacterial QS system.
Wrapping up Fig. 4, we can see that the relationship between the affects of RA and C4-HSL are actually dependent on concentration. In low doses, C4-HSL works better (in this case, produces more β-galactosidase) than RA. But once you’ve crossed the 10µM threshold, RA starts strongly outcompeting C4-HSL.
Tying Up Loose Ends
A minor experiment the authors conducted occurred because some of their data contradicted, at first glance, data from some previous studies.
The authors hypothesized that this switch occurs because P. aeruginosa can actually metabolize – that is, it can eat the carbon from – RA. To test this, they grew P. aeruginosa with RA as the only carbon source (none of the usual glucose to feed it!). Here (Fig. 5A), we can see that P. aeruginosa can still grow on RA, indicating that it can metabolize it. But once RA’s concentration gets too high (note: 10 milliMolar, not 10 microMolar), the cells are no longer viable. Previous studies had shown that RA was toxic to P. aeruginosa, however, so to explain their data’s validity in light of previous research, they tested cell survival rates on increments of gradually increased RA concentration.
Cell survival rates was quantified by colony counts on agar, and again, note: the concentrations here are thousands (milli to micro) of times higher than what was used in their transcription studies. RA is apparently not toxic to P. aeruginosa until it reaches concentrations of over 7.8mM (Fig. 5B).
To further confirm their point (again, in the presence of seemingly conflicting data, which is now being reconciled), the authors tested growth rates of P. aeruginosa at concentrations used by the transcription assays (that is, back down to micromolar concentrations). Thus, in Fig. 5C, we can see that neither C4-HSL or RA (or the negative control DMSO) affects growth rates, whether at 1, 10, or 100µM.
Fig. 6 contains quick overviews of studies done for other promoters controlled by the RhlR regulatory protein. By connecting lacZ to the promoters for lasB, rhlA, and hcnA, we can see that RA consistently activates the RhlR-activated QS system, including multiple promoters that the protein is in charge of turning on.
In the next and final installment, we’ll go over the grand finale: the final test that RA is used by plants to confuse its pathogenic bacterial adversaries.
Note on the Figures
All figures for this paper can be found here; sign up for a free account with Science and you’ll be able to view them.
Corral-Lugo, A., Daddaoua, A., Ortega, A., Espinosa-Urgel, M., & Krell, T. (2016). Rosmarinic acid is a homoserine lactone mimic produced by plants that activates a bacterial quorum-sensing regulator. Science Signaling, 9(409), 1–11. https://doi.org/10.1126/scisignal.aaa8271