So when studying bacteria as they cause disease, one is often faced with the question- why do they do things ? why do bacteria suddenly make the decision to become virulent ?
In the simplest terms, bacteria make decisions based on various sensory inputs, such as the presence of other bacteria and chemical signals from them, changes in the environment, food availability, and even the host immune response. How do bacteria take these signals, and direct them into meaningful outputs?
The paper I'm going to talk about contributes an answer to one of these questions- How does a bacterium decide when it can and cannot become virulent ?
In the link to a previous post, I touched on some of the bacterial factors related to expression of the LEE, and it's different roles in cows and in humans. In this paper, we'll be taking a much more in depth look at how the LEE is regulated. In most bacteria, there are a number of virulence regulators used to determine when the bacteria should express certain genes, and when they should not.
The main focus of this paper is a regulator called Hfq (pronounced hayfukyu*), which acts by grabbing sRNA and mRNA and putting them together. There is not much known about these proteins, save for them being important for the regulation of virulence in a number of species of bacteria, including Escherischia coli, Pseudomonas aeruginosa and Staphylococcus aureus (although the last one is debatable). In previous work, LEE expression was noted to increase when Hfq was removed.
Let's see what this paper finds in its results.
One way of determining what Hfq does is removing it from the bacteria, and then looking at what happens. Does taking this one gene out of the network affect the expression of others, and if so, how?
So they performed microarray analysis, which measures the levels of mRNA expression for multiple different genes. and by multiple, I mean ALL of the recorded genes for EHEC. Since the expression of these genes can be affected by multiple factors, such as growth conditions, the authors were careful to ensure that the growth conditions were the same for both the control (which had Hfq) and the knockout (which did not), i.e. they were both grown to late log phase (when LEE genes are usually expressed).
They then collected the bacterial cultures, and extracted mRNA from them, and assessed the relative numbers of mRNA transcripts for each gene, and then compared the differences in expression between the two strains of bacteria. They found that a lot of genes were changed in their expression by this mutation, indicating that the absence of Hfq shows differences in gene expression when they test genes common in lab strains, in other EHEC isolates. The point here is that they looked at a wide variety of genomes, seemingly to ensure that they got coverage of every possible gene in the bacterium. And they found that the deletion of this gene caused the expression of mRNA to change a lot.
Some authors tend to display their mRNA data by listing out the genes which expressed the most change, and the statistics underlying this, so that they can then focus on the genes which either showed the most difference, or genes with hitherto unknown function that were also significantly increased in order to determine their role. Of course, those figures tend to take statistical probability as an output (which in itself can be a questionable practice) , so they've denied me a great opportunity for criticism.
They say that the main genes involved are to do with iron acquisition, Virulence regulation, LEE expression and the Quorum sensing E.coli regulators BC (qseBC). We don't actually get a sense from what they present as to the statistics relating to this expression. Microarray is used as a starting point and provides the impetus for the next sequence of studies where they will analyse these genes in greater detail.
In the previous set of results, it was implied that Hfq impacts the expression of LEE. It is known that the expression of LEE is controlled by a gene called ler, so does hfq change the expression of ler leading to a change in LEE ? Have you gone cross eyed yet?
So they again looked at mRNA transcripts, this time focussing on ler expression in the wild type bacteria, and the knockout. A housekeeping gene standard, rpoB was used to standardise the data for bacterial numbers.
They looked at the ratios of the transcripts at different stages of bacterial growth.
a) shows 86-24 (the wild type) compared to the knockout (86-24 hfq-) and a complemented strain where they added the gene back into the bacteria to show that the knockout didn't interfere with any other genes that could also be responsible for the effect. All this was done at late log (where f you remember LEE is expressed). And it shows that without hfq, there is no ler.
b) shows this is true for mid log, and c) shows it's true for stationary phase,
Since other groups grew their bacteria in LB broth (as opposed to DMEM as this group does) they repeated their results in LB, and this is what is shown in D) for bacteria at mid log.
The bottom line of this figure hints at the main point of interest in this study. This group took a different strain that caused EHEC called EDL933, and a mutant of that strain without the Hfq, and compared the expression of ler at E) mid log and F) late log. The deletion of hfq in this strain had the opposite effect to the mutation of ler in the 86-24 strain, and is more in line with the literature surrounding the subject. They even confirmed this result in LB to ensure that it wasn't some quirk of the media they used G).
So without hfq, there is less ler in the 86-24 strain, but there is more ler when Hfq is removed from EDL933.
Since ler affects the expression of LEE, then you would expect changes in this expression to correspond with it. Table 4 looks in more detail at the fold differences in the expression of genes between the 86-24 mutant (which is now called MK08 to "simplify" things). And indeed, it confirms that the changes you'd expect if ler expression was changed, actually happen.
So we've seen that the change in hfq expression changes the primary regulator of the LEE, but does that constitute actual changes in the LEE genes? Considering how important the LEE is to infection, it is entirely conceivable that this particular strain may have made compensatory changes that would maintain normal LEE expression, which would lead to another interesting avenue of inquiry.
A)-D) Looks at the expression of a number of important LEE genes at different growth phases and in different media. The grey bars are the wild type, and the check bars which are almost completely absent are from the knockout. Suffice to say, this supports the idea that in this particular strain, hfq is necessary for the activation of LEE production.
E) We've been looking mostly at mRNA transcripts, but there is a question of whether these changes are manifested as protein expression. Specifically, they looked at EspA (one of the LEE genes) and using a western blot, found that there was less protein in the hfq knockout.
All right, well we've seen that bacteria without the gene are different, but what does this mean really? Does it have an effect on actual infection ?
To show this, the researchers looked how this strain of bacteria creates attaching effacing (AE) lesions when incubated with HeLa cells using fluorescent microscope, which are shown in the composite images above. Attaching effacing lesions are caused when bacteria stick to the surface of a cell, and through the action of various genes encoded by the LEE. The proteins encoded by the LEE recruit the cytoskeleton of the host cells, causing them to build around the bacteria.
The green in the images above shows the cytoskeleton (which is stained with a fluorescent dye) and how it interacts with bacteria (the smaller red dots). The nuclei of the cells were also stained (they are the bigger red bits) and give you an idea of the relation of the HeLa cells to the bacteria.
The first image is with the wild type bacteria, and you can barely see the bacteria as they have formed lesions of actin in close proximity to themselves. These are the attaching effacing lesions.
But in the second image, you can quite clearly see the bacteria, and the actin are in different parts of the cell. The difference between these two images? The last one does not have hfq, and therefore does not have LEE.
These images confirm that the way 86-24 EHEC regulates itself is quite different to what has been seen in other strains studied in the literature.
Enough about the LEE, what about the other virulence factors used by EHEC... what about Shiga toxin ?
What's Shiga toxin ? It is a deadly toxin produced by EHEC (and Shigella) which kills of cells and is pretty nasty. Will the deletion of hfq have an effect on its production ?
A) So again, they looked at the transcription of mRNA by 86-24, 86-24 hfq knockout, and 86-24 hfq knockout with the hfq returned on a plasmid for reasons that I noted above, and they found that the knockout produced more shiga toxin mRNA.
B) they then looked at the actual presence of shiga toxin by western blotting and showed results that mirrored that of the mRNA- that when hfq is removed, more shiga toxin is made.
C) & D) They compared this with EDL933 strain of EHEC to check that it worked the same way, and it did.
But the point here is that hfq does play a role in the production of the protein, and moreover, it is the same between both the 86-24 and the EDL933. This is in direct contrast with the regulation of the LEE, which is promoted by Hfq in 86-24 and repressed by hfq in EDL933.
In the previous microarray, they found that the qseBC regulatory system was also affected by the removal of hfq.
QseBC is a two component system, with one gene acting as a sensor and the other which acts as a transducer, transcribing genes when the sensor is activated. This particular sensor detects human epinephrine and a substance called auto-inducer-3. These two hormones tend to coincide with the onset of an immune response***, so the qseBC can allow bacteria to detect this response, and respond by regulating it's expression of virulence factors.
But to confirm this properly, they checked out the levels of transcription of this gene in 86-24 hfq, 86-24 hfq knockout, and in the complemented strain. And they confirmed that hfq activates this system in 86-24.
Remember in No Country For Old Men when they killed Josh Brolin off screen randomly, and the rest of the movie became Tommy Lee Jones moaning about how old he was ? This figure is kind of like that. Oh yeah, spoiler alert by the way. This comes out of nowhere, and we lose our main character, Hfq.
This figure focuses on qseBC.
In previous work, they created knockouts of qseC and qseB in the 86-24 strain. In the study where they knocked out qseC, it was replaced with beta-galactosidase, and the strain qseB- had a truncated version of the gene. They decided to look at how the absence of working QseB or QseC affect their own transcription, because this is what had been found in previous work. the implication is that Hfq plays a role in this, but lets see what the data shows,
A) Here, they looked at the mRNA transcripts of the whole qseBC operon (because qseb and qseC are usually transcribed together as one big chunk). This figure shows the effect of the addition of either autoinducer 3, or epinephrine (as previously noted, these are the activators of the qseBC system) when qse C is or is not present. It shows that when qseC is removed, there is an increase in transcription of qseBC, suggesting that qseC acts to block transcription, and that qseB is what promotes it, and is responsive to the addition of epinephrine or autoinducer 3.
B) This is a northern blot with the wild type, the qseC knockout and the qseB knockout to confirm the previous figure. well, I presume it does, but while there are three lanes for each strain, these could just be replicates for all the figure legends tell me.
While this is all very interesting, this last figure feels a bit out of place. The implication of figure 5 is that hfq regulates qseBC transcription, but this last figure doesn't really shed any new light on that. They don't even refer to it in the discussion. Actually, thinking about it, it's more like a big-lipped alligator moment. But for a scientific paper.
So here in this paper, we have a strain of EHEC which just won't behave the way you've expect. It regulates LEE in the opposite way to other strains, but regulates shiga toxin in the same way. They also found that this has an effect on the qseBC system.
The EDL933 and the 86-24 bacteria were both isolated in the 80's during two different disease outbreaks. The 86-24 was more severe, which makes sense considering how much it loves to make attaching effacing lesions. I've been struggling to pick holes in individual experiments but there are some lingering questions that I do have-
We've established that hfq acts differently in 86-24, but why is it acting differently ?
Is there a mutation in it perhaps? I haven't seen sequencing data between the strains, so I as a reader cannot rule it out. But I think there is a far more interesting question here.
The main thing that you need to remember with Hfq is that it doesn't act alone. The main way it functions is through using sRNA's. Focusing just on Hfq is only going to get us half the story, the other half lies in how the sRNA's are regulated, and which ones are working with Hfq to do its job.
The surprising finding here is that bacterial strains can show a lot of variation in the community, although I suspect that as labs work on greater numbers of clinical isolates, we'll see more variation like this, and hopefully we can use these differences to further understand how these bacteria are evolving.
* by me only. Sorry for making you scroll down so far for just three words.
**I'm starting with table 3, because the previous two tables simply listed out the various strains/ primers used in this study, so that other scientists can replicate and expand on the work performed here, and I don't feel a powerful need to comment on them.
*** and a helluva lot of other stuff that I don't want to get into here.