When we picture bacteria, it is tempting to imagine millions of tiny individual cells, each going about its own business.
That picture is increasingly wrong.
Bacteria interact. They compete for food, cooperate with their neighbours, respond to chemical messages and build remarkably complex communities. Sometimes they even enlist viruses in the process.
Understanding these microbial communities was at the heart of my PhD.
I studied the bacterium Clostridioides difficile — better known as C. difficile — and in particular how it forms communities called biofilms and interacts with other bacteria living in the gut.
What began as a question about a single bacterial gene ended up involving chemical communication, microbial warfare, viruses that infect bacteria and the ecology of the human gut.
A difficult bacterium
C. difficile is an opportunistic pathogen of the human intestine.
For most of us, a diverse community of microorganisms already occupying the gut makes it difficult for C. difficile to become established. This phenomenon is known as colonisation resistance.
Antibiotics can disrupt that ecosystem.
When competing microorganisms disappear, ecological opportunities open up. C. difficile spores can germinate, the bacterium can multiply and toxins released by the growing population can damage the intestinal lining.
One particularly difficult feature of C. difficile infection is its ability to return after apparently successful treatment.
There are several possible explanations for this. Its extremely resistant spores are undoubtedly important. But when I began my PhD, researchers were also becoming interested in another possibility: biofilms.
A bacterial city
A biofilm is a community of microorganisms attached to a surface and surrounded by material produced by the community itself.
Dental plaque is one familiar example.
Rather than individual bacteria floating freely through their environment, cells within a biofilm live together inside a matrix containing substances such as proteins, polysaccharides and extracellular DNA.
You can think of the matrix as something between the scaffolding and infrastructure of a microbial city.
Living this way can have major advantages.
The environment inside a biofilm is very different from the world outside it. Nutrients and oxygen can vary over tiny distances, different cells can adopt different behaviours, and the community can become considerably more tolerant of environmental stresses.
For a pathogen, that can include increased tolerance to antibiotics.
C. difficile was known to form biofilms in the laboratory, but much less was understood about how it decided to build them.
That brought me to a gene called luxS.
Can bacteria talk?
Many bacteria communicate using a process known as quorum sensing.
Individual cells release small chemical molecules into their surroundings. When there are very few bacteria, the molecules remain at low concentrations. As the bacterial population grows, more of the signal accumulates.
Eventually its concentration can cross a threshold.
The bacteria can therefore use the concentration of the chemical as a rough indication of how many other cells are nearby.
Once that threshold is reached, the population can alter its behaviour.
In different bacterial species, quorum sensing helps coordinate processes including toxin production, movement, sporulation and biofilm formation.
One particularly intriguing signalling molecule is known as autoinducer-2, or AI-2.
An enzyme called LuxS is involved in its production.
AI-2 attracted considerable scientific interest because LuxS is found in many different bacteria. This led to the intriguing suggestion that AI-2 might not merely allow members of one bacterial species to communicate with each other, but could potentially contribute to communication between different species.
There was, however, a complication.
LuxS also has an ordinary metabolic role inside bacterial cells. So when researchers remove the luxS gene and see the bacterium behave differently, it isn't automatically proof of bacterial communication.
Have you removed its telephone?
Or have you damaged its metabolism?
My PhD tried to tease those possibilities apart in C. difficile.
What happens when C. difficile loses LuxS?
One of the clearest results was also one of the most striking.
Normal C. difficile produced substantial biofilms.
A mutant in which luxS had been disrupted produced dramatically less.
Importantly, the number of viable bacterial cells recovered from the two populations was similar.
The mutant wasn't simply incapable of growing.
Something about the structure of the community had changed.
That pointed towards LuxS playing a role in producing the extracellular material that turns a collection of bacterial cells into a biofilm.
But that still didn't tell us whether the important factor was bacterial signalling or the metabolic function of LuxS.
So we tried giving the signal back.
Giving the bacteria their message back
AI-2 is formed from a molecule called DPD — 4,5-dihydroxy-2,3-pentanedione.
When we added chemically produced DPD to the luxS mutant, something interesting happened.
At an appropriate concentration, biofilm formation was partially restored.
It didn't completely return the mutant to normal, and higher concentrations could actually inhibit biofilm formation.
But partial restoration was important.
If removing LuxS had affected the biofilm entirely because of its intracellular metabolic function, simply putting an extracellular signalling molecule back into the environment shouldn't necessarily have rescued the phenotype.
The result therefore provided evidence that C. difficile wasn't merely producing AI-2 as an accidental metabolic by-product.
It appeared capable of responding to it.
There was one rather large mystery.
We didn't know how.
The well-characterised AI-2 receptors found in some other bacteria weren't obvious in C. difficile.
Something was listening to the message, but we didn't know what the receiver looked like.
And then the RNA sequencing produced an even stranger result.
The viruses hiding inside the bacterial genome
We compared gene expression in normal C. difficile with the luxS mutant while the bacteria were beginning to form biofilms.
A surprisingly coherent group of genes behaved differently.
Many belonged to prophages.
Bacteriophages — usually shortened to phages — are viruses that infect bacteria.
Some phages immediately hijack a bacterial cell and ultimately destroy it. Others can insert their genetic material into the bacterium's chromosome and remain there, potentially for generation after generation.
This integrated viral DNA is called a prophage.
In our luxS mutant, genes belonging to two prophage regions were expressed at lower levels than in normal C. difficile.
We were also able to find evidence of intact phage particles associated with C. difficile biofilms.
That suggested an intriguing possibility.
Perhaps the relationship between LuxS, AI-2 and biofilm formation wasn't straightforward at all.
Perhaps viruses were involved.
Building a community by sacrificing some of its members
One important ingredient in many bacterial biofilms is extracellular DNA.
That may sound unusual. We normally think of DNA as something carefully protected inside a cell.
But DNA released into a biofilm can become a structural component of its extracellular matrix.
One way to release it is obvious: break some bacterial cells open.
Our results led us to propose a model.
As a population of C. difficile grows, the concentration of AI-2 around it increases.
Once that concentration reaches an appropriate level, an AI-2-dependent pathway could stimulate prophage activity in a proportion of the population.
Those phages could cause some bacterial cells to lyse — to burst apart.
Their contents, including DNA, would then spill into the surrounding environment.
That extracellular DNA could help strengthen the developing biofilm.
In other words, one possibility emerging from the work was extraordinary:
chemical communication within a bacterial population might trigger viruses to kill some of its members, and the remains of those cells could help build a protective community for the survivors.
It is a compelling idea.
But it is important to distinguish the model from what we actually demonstrated.
We showed differences in prophage gene expression associated with LuxS and found evidence of phage in the biofilm. We did not complete every experiment required to demonstrate the entire causal chain from AI-2 signalling to phage activation, cell lysis, extracellular DNA and increased biofilm formation.
That would have been one of the obvious next stages of the research.
But bacteria don't live alone
The second major part of my PhD moved beyond interactions between C. difficile cells.
The human intestine contains an enormous microbial ecosystem. Studying a pathogen entirely by itself therefore risks missing much of what determines whether it succeeds or fails.
We chose another gut bacterium, Bacteroides fragilis, and grew it together with C. difficile.
The result was striking.
When the bacteria were growing together as biofilms, B. fragilis strongly inhibited C. difficile.
The number of viable C. difficile cells fell dramatically, while B. fragilis did considerably better in the mixed community.
Yet the same clear inhibition wasn't seen when the organisms were simply growing together as free-floating cells.
Whatever was happening depended on the ecological circumstances in which the two organisms met.
Even more intriguingly, normal C. difficile was inhibited more strongly than the luxS mutant.
LuxS therefore appeared to influence not just how C. difficile interacted with members of its own species, but also what happened when it encountered another member of the gut microbiota.
Microbial warfare — or simply a better competitor?
We tried to identify what B. fragilis was doing.
One obvious possibility was that it secreted an antibiotic-like substance that killed C. difficile.
But when we removed the B. fragilis cells and exposed C. difficile to the remaining liquid, the inhibitory effect disappeared.
That made a simple freely secreted toxin less convincing.
Our sequencing experiments instead revealed substantial changes in metabolism when the organisms were grown together.
This opened another possibility: competition.
The two bacteria could be competing for the same resources, with B. fragilis simply being better adapted to acquire some of them.
One interesting candidate was iron.
Genes involved in obtaining iron were activated in B. fragilis, and our analysis suggested that it possessed an iron-acquisition system that C. difficile lacked.
If B. fragilis could strip a shared environment of an essential resource more effectively than C. difficile, no microbial poison would be necessary.
Starving your competitor can be just as effective as attacking it.
But again, the experiments didn't allow us to close the case. Other mechanisms remained possible, including close-contact interactions or molecules whose functions hadn't yet been identified.
That uncertainty is part of the science rather than a failure of it.
From pathogens to ecosystems
Looking back, one of the things I find most interesting about the work is how quickly a seemingly simple question expanded.
I started with a gene involved in bacterial biofilm formation.
Following the results led from that gene to chemical signalling, from signalling to viruses, and from individual bacterial species to competition within the wider gut ecosystem.
And that is increasingly how microbiology has to be understood.
A pathogen doesn't encounter a sterile host.
It enters an ecosystem.
Whether it succeeds can depend upon the organisms already present, the resources available, chemical signals released by neighbouring cells, viruses hidden inside bacterial chromosomes and the physical communities microorganisms construct around themselves.
Even something as apparently straightforward as asking why one bacterium forms a biofilm can therefore reveal an entire world of interactions.
The questions I would ask next
My PhD finished with considerably more questions than it started with.
I would want to establish whether changing prophage activity really alters the amount of extracellular DNA in C. difficile biofilms, and whether the LuxS-dependent effects seen in laboratory biofilms also occur during infection.
I'd also want to determine precisely how C. difficile detects AI-2.
For the interaction with B. fragilis, the obvious challenge would be identifying exactly why C. difficile was being inhibited and testing whether the same phenomenon occurs with other members of the Bacteroidetes.
Most importantly, I would move towards increasingly complex microbial communities.
Instead of asking how C. difficile behaves alone, we could ask what happens when it encounters communities representing a healthy gut and compare those with communities disrupted by antibiotics.
Because perhaps the most important lesson from the research is also one of the simplest:
microbial life makes far more sense when we stop thinking about microorganisms as isolated cells and start thinking about the worlds they build together.