How Bacteria Evolve by getting DNA from Family and Friends - Professor Zamin Iqbal

The Director of the Milner Centre for Evolution, Professor Turi King, talks to Professor Zamin Iqbal about his research that uses computers to look at the genetic makeup of disease-causing bacteria to understand how they evolve and how they're becoming resistant to antibiotics.

Episode Transcript

(00:02):
Hello and welcome. You are listening to a podcast by the Milner Centre for Evolution,
at the University of Bath. I'm Professor Turi King, your host, and today I'm talking to Zamin
Iqbal, professor of algorithmic and microbial genomics here at the centre. Zam uses computers
to look at the genetic makeup of disease-causing bacteria to understand how they evolve and how

(00:24):
they're becoming resistant to antibiotics.Zam, lovely to chat to you. I want to start
with the fact that I know you have a slightly unusual career history, so tell me about that.
Yes, I do have a slightly long and winding road to get here. I started out as a mathematician,
once upon a time, I did a degree and PhD in maths, at which point I just decided I

(00:48):
was really tired and needed a change.So, I went and worked in the software
industry for about 8 or 9 years. I worked in mobile phones. I worked on the first
operating systems for the first phones that had cameras in them, about which I was very,
very sceptical because I didn't see any reason why anyone would want cameras in their phones. So
yeah. Anyway, luckily, I was wrong about that.And at some point, about eight years in, I

(01:13):
realized I was really missing science and started learning about biology again, which I'd been
interested in as an undergraduate, but I've never really followed up and started looking for whether
or not there were jobs where I could contribute.I thought at the time that maybe the right thing
to do would be, just be a programmer helping biologists. And so, I got a job on something

(01:33):
called a Thousand Genomes Project, which was just when Illumina technology was taking off,
and they were going to sequence 2500 human genomes. I basically got to sit in rooms with
a lot of famous people and listen to them debate about things and learn about basic algorithms
people used to study genomes.So, what were you working on?

(01:53):
Well, at the time, our idea was, and this is not me, this is a global consortium,
the idea was to sequence 2500 human genomes, which was a massive statement back then. You know,
we had two human genomes up to then, and the idea was to try and understand the genetic variation
that we all share, and which separates us.So, we took 2500 people from 26 different

(02:18):
populations, sequenced their genome in a certain sense, learned about the common variation between
them all. And one of the funny things about how we study genomes is we try and describe
everyone in terms of how they differ from a standard. This is the standard human, and,
you know, Turi is just like the standard, but she differs here and here and here.
And that works if everybody is very close to the standard, but in some bits of the genome that just

(02:41):
doesn't work very well. It's a bit like describing some route around the London Underground in terms
of always starting and ending on one particular line. You know, sometimes there are chunks of DNA,
which are very different between you and me, and sometimes it's just not helpful
to describe that in terms of a reference.So, I started thinking about ways to compare

(03:04):
people's DNA directly, instead of introducing a third party, saying, you know, Turi and I both
differ from the standard in this way. We could say, how do we differ from each other? And I
developed some methods to do that and looked at the human populations we've been sequencing. And,
you know, we discovered that there's quite a lot of variation that was completely missing
from the reference, you wouldn't know it was there, but which did differ between people.

(03:26):
And the most obvious stuff is to do with the immune system. And that wasn't a surprise,
I mean, people would have expected that.And that led me down a path towards other species,
to be honest, because I got excited about weird and wonderful ways that genomes evolve, and
humans are fine, but there's so much more other stuff going on in other species and particularly
in bacteria. In fact, your and my genomes are basically 99% the same, even if we differ at some

(03:50):
mutations, we have the same genes, we have the same chunks of DNA. But if I took two bacteria
from the same species, they might only be 50% in common and they have huge amounts of what we call
mosaicism, so they can absorb lots of different blocks, and you make this kind of patchwork quilt
of a genome. And that's super interesting.So, you come back into academia,

(04:10):
but what leads you down to the path to the research that you're doing now?
Well, I guess what I did was, I was trying to find ways to understand how different genomes
vary. And sometimes with these things, you need to frame the question the right way. If you frame
it the right way, everything becomes a lot easier.So, the way we were framing things at the time was

(04:31):
describing everyone in terms of how they differ from a standard genome, and that works when
everyone's very close to a standard genome, but when they're not, actually, it becomes easier to
compare everybody against everybody. So, I develop algorithms and methods for building a kind of,
well what we call a pan-genome. So, something that represents all of human genetic variation.

(04:54):
And then using that as a structure to look at all of the variation within the species.
And that was super interesting for humans, and we built the first human pan-genome,
well more than ten years ago now, and looked at how, which bits of the genome were varying in
different populations. But the whole methodology was super interesting, and actually the concept of
a pan-genome came about ten years before that, it was discovered in bacteria.

(05:19):
People were trying to develop vaccines, and they thought the sensible thing to
do was to look at the proteins that would be generated by a particular bacterial species.
And then they were bit shocked when they took a few genomes from that species to discover that
they were not as similar as they expected.They expected everyone to have more or
less the same genes, but actually it turns out that in bacterial species,
one of the ways in which they're able to adapt super quickly to new environmental conditions,

(05:43):
is passing huge chunks of DNA around between each other. And those chunks of DNA code for
entire functional units that give them new, I keep saying superpowers, sorry it's a bit childish,
but you might be able to metabolize something new, you might be able to photosynthesize,
you can find… they do crazy things. You can fire harpoons at adjacent bacteria and blow them

(06:03):
up. You can release toxins, all kinds of things like that. And those are all passed around very
quickly. And so, the pan-genome of a bacterial species can be ten, 100 times bigger than the
actual genome of any individual bacterium.And that's really important because as
you're saying, bacteria evolve really quickly and they're slightly weird in terms of evolution. So,
the way we think of evolution normally is like so, you know, we passed down half of our DNA from each

(06:27):
parent goes to each child. It's about going down through the generations. But bacteria are weird
in that they can swap bits of DNA back and forth between one another. So, they're unusual that way.
You're right. Every bacterial cell has two types of DNA. It's got the chromosome, which is passed
down to their children, and that is inherited sort of vertically through the family line. And

(06:51):
they have other DNA, sometimes it's actually specifically separated into circular rings
called plasmids. And they may get passed on to the children, but they also may escape the cell, move
to a different cell. And so now their evolutionary history is moved through unrelated cells,
may move to a totally different species. So that means very rapid and unexpected things can happen.

(07:13):
You know, maybe it doesn't work in the new place, maybe that new bit of DNA is unhelpful,
and that cell will die. But sometimes that's the chance for incredible novelty, a new function
that wasn't there before, and so on. So, there are lots of bacteria who have acquired the ability to
move between motile and sessile or can photosynthesize or can kill things or have

(07:35):
an immune system, all kinds of things like that are passed around very quickly, sort of sideways.
Yeah. So, I'm a human geneticist, so I know the very basics really about bacteria, but I
know that plasmids are really important. So, what are plasmids doing? I mean I like the fact that
you kind of get these like patchwork quilts of plasmids because you get one bacteria going, oh, I

(07:56):
have this snazzy little superpower I can do this, here you can have it. So, what are plasmids doing?
Well, that's interesting. They do all kinds of things. They seem to mostly provide
functions that are locally adaptive. So, they're an advantage temporarily.
Yep.If they were permanently
always needed, then I think generally you would find the functions would be absorbed into the

(08:18):
chromosome and they would be there forever.So, what are the challenges bacteria face?
They've been around 2 billion years. The things that they've always faced are viruses,
which predate on them and each other and things like temperature and environment.
So, plasmids provide a variety of functions. If you look across a generalist species, you
will find a wide variety of different plasmids. You know, in the different cells. You can think

(08:41):
of that as a kind of bet hedging, I think. If there's some plasmid that's super critical,
even if you lose some members of the species that you know, the others will survive, and also
potentially that plasmid could spread again.So that's the kind of thing they do. I mean,
we notice them and care about them a lot at the moment because they often carry drug resistance,
or they carry genes that allow the bacteria to survive when they're treated with antibiotics.

(09:04):
That makes a big difference. And since we've introduced antibiotics,
they've spread super-fast across the world.And often people who are genomic epidemiologists,
so people who track down how the virus is spreading. If you're in a hospital and you've
got several people who are ill and they've got the infections resistant to a certain
set of antibiotics, then you might think, okay, well, is there a single bacterial species that’s

(09:27):
spreading? Is there something that's very fit and it's moving through my ward? Or sometimes
what happens is there's actually a plasmid that's moving between different species.
There are very famous examples where you have, you know, more than a decade of an
outbreak in a hospital. And the thing that is in common is either a plasmid or even

(09:47):
something smaller than a plasmid jumping between plasmids. So yeah, it's chaos.
Yeah. So, what are you doing?I try and invent ways to think about these things,
and if you find the right way to think about these things, everything becomes easier. And ways to
think about things means computer programs. So, like a simple question would be, if two people

(10:09):
are ill and they both have E. coli, is this the same E. coli that’s infected them both,
you know, maybe they share the same infection.And that's an easy question to answer because
we can look at the sort of family tree of the bacteria and try and see how closely
related they are. And if they're very, very closely related, almost the same,
then it becomes quite credible that actually there's been an infection passed between them.

(10:31):
Whereas if they're like thousands and thousands of mutations apart, well, there's no way that
happened within the two weeks that they were both in the same ward. So, they've got two different
infections. That's sort of well understood.But with plasmids, because they actually
change their genomes very dramatically. They will eject chunks of DNA, they will rotate bits around,
they will jumble things up. It's a bit like the Ship of Theseus or Trigg's broomstick,

(10:54):
if anyone remembers 1980s comedies, it's hard to know if these two plasmids are the
same plasmid. Is it credible that this one has changed, you know, thrown out two kilobases of
DNA and gained another five kilobases of DNA.So, what we look for, are computational methods
for looking at two chunks of DNA and saying, well, given how plasmids change, which is a certain kind

(11:17):
of structural changes, is it credible that there's not too many steps to go from one to the other
and sort of inventing ways to do that is sort of one of the things we've been working on recently.
So, it's like you're kind of tracing sideways family trees on most of plasmids and looking
how they are chatting to each other and swapping bits and how they're related. And I'm guessing,

(11:41):
given that you're using these huge kind of computational methods to kind of interrogate these
plasmid genomes on a really big scale, you're looking is it globally now that you're looking?
Yes, we look globally. I mean, one of the things we're trying to do is if you're trying to look
at how plasmids or other things move across the world and across species, it's not enough to just,

(12:01):
you know, get the bacteria from your fridge or from your local hospital,
you want to look very broadly, and you can't do that yourself, because even if you exclude the
expanse, I mean, where would you sample from?It's actually much more effective to look at
everything that everyone else has ever sequenced. So, there are huge DNA archives where people share
DNA of bacteria that have been sequenced. So, in principle, we can just go and look in there

(12:23):
and see where are these plasmids? The trouble is that the process of constructing a genome
from raw experimental data is imperfect, and the data that's in the archives is sort of a
hotchpotch of different methods of different quality control systems.
Most commonly, you find a super exciting thing and that actually, well, either it's an error in your

(12:44):
code or, you know, all of these things that you think of exciting they were all done by the same
person, and they all made the same mistake, or they've done something slightly unusual.
So, what we've done is we've started a large global consortium of people taking
all of the DNA of bacteria that have ever been sequenced and reprocessing it all,
building high quality genomes. And yes, we will be looking through all of that and all

(13:07):
the plasmids in that. That's quite exciting, we've got 2.5 million genomes at the moment.
And I know one of the things you're looking at is antimicrobial resistance. So, we've touched on it
already the fact that bacteria are now becoming resistant to antibiotics. And that's something
else that you're looking at the moment, isn't it?Yeah. That's an ongoing theme. So, one of the big

(13:28):
challenges for us is how to deal with infectious diseases. And treatment of infectious diseases is
actually an evolutionary challenge. So, it's not like a kitchen where you just clean it,
and the bugs go. When you treat a patient with a drug, the bacteria inside the patient are growing,
they're reproducing, they're mutating, they're evolving during the therapy.

(13:51):
So, during the first ever clinical trial of streptomycin for tuberculosis,
and during the trial, patients got better and got worse because there was resistance that
evolved within the patients during the lifetime of the trial. And that's an ongoing problem for
us. It's a problem per patient because we wanted them to survive, and it's a problem

(14:13):
across the population because small numbers of bacteria can survive that therapy and spread.
So, for tuberculosis, it's a problem because a lot of the people who get ill from it are
poor. They don't have access to health care, they live in greatly overcrowded conditions,
and you can be infected and be asymptomatic for a while, you don't know that you're infected.

(14:36):
Actually, I did a back of the envelope calculation that I thought six months in, you might have over
a billion bacilli in your lungs, which means actually, you know, every position in the genome
could have mutated.Wow.
So, if we assume that drug resistance can be caused by a mutation in the genome,
that sort of means that every type of resistance is already there before you start therapy,

(14:56):
which is why standard therapy for TB is actually multiple drugs at the same time.
It's much harder for the bacteria to develop resistance to multiple things at the same time.
So, this has been an ongoing challenge for decades. And the problem for us is that we develop
new drugs at a very slow rate. So, it's exciting times in TB in some ways, because we've discovered

(15:18):
that we can have slightly shorter therapy regimens that are just as successful, but we don't have an
armoury of, you know, new antibiotics ready to treat. So, it is inevitable that there will be
resistance to any antibiotic that we generate. And so, there's not a very good answer to that, apart
from the fact that we need very good diagnostics.So, is your work trying to work out what the

(15:43):
genetic changes are that are making this bacteria resistant to the antibiotics? And are you trying
to kind of like track those changes, I suppose kind of through time and geographical location,
because once you know about that, if you kind of go, oh, okay, they've got this particular version

(16:03):
of the bacteria that's going to be resistant to this antibiotic, hit it with this one. Is
that the kind of stuff that you're trying to do?That is the kind of stuff we do, yes. Basically,
there are drugs where there's a lot of resistance out there already. And for those, you can try and
work out which mutations cause resistance by a statistical process. Collect a lot of samples

(16:24):
from patients, measure them, mix the bug in the drug and see if it kills it, essentially,
and measure the level of resistance in them all, and then get the genomes of each of those bacteria
and then each of those genomes you can break that down into which mutations they have, and then
you can work out which mutations correlate with resistance. That is not a totally trivial process.

(16:46):
So, we spent five years actually collecting about 15,000 samples from 23 countries,
measuring drug resistance to 13 different drugs. That's a consortium called the Cryptic Consortium.
And what we did manage to do is actually build a very comprehensive catalogues of resistance
mutations, which got absorbed by the W.H.O.The problem is with new drugs, you want to

(17:08):
know as quickly as possible what's causing resistance. But if you want to spot that fast,
you don't have many samples, and if you don't have any samples, you don't have much evidence.
If I got a new TB sample from a patient, there would probably be, you know, 1000 or
2000 mutations in it and which one is causing the resistance, I wouldn't know, not upfront.

(17:28):
So, one of the things you can do is go to places where they first started therapy. So bedaquiline
is one of our latest and greatest drugs, and I say latest and greatest drugs it went into 20 years
ago and go somewhere where they started treating them earlier than the rest of the world, so South
Africa and track resistance that appeared there.And so, when we did and when we went to look

(17:51):
essentially there are two options. Either we would see one strain or small number of strains of TB
that were somehow becoming very successful and spreading. And we'd see the same thing in many,
many patients. And that's not what we saw. Or we might see drug resistance evolving in slightly
different ways, again and again and again in different sort of genetic backgrounds. And

(18:12):
that is what we saw. And so that means that that pressure to evolve the resistance is
there again and again in different patients.We were using data from during the pandemic,
and there's more data has come through since then. So, we're going to track forward and try
and understand whether or not anything specific is spreading or you know what's going on.
So, what's the hope, is the hope that eventually it might be something that clinicians can use

(18:37):
very quickly in the clinic. It’s like, right, we're going to get a sample here,
you've got this particular bacteria, it's got this particular sequence. We know not to bother hitting
it with this, we're going to hit it with this. Is that the hope that eventually it'll become
something that clinicians can use pretty quickly?Yeah, that's absolutely right. So, there are two

(18:59):
things you need, you need diagnostics to tell you what to do, and you need therapies. And what
we're talking about here is the first half. So, if you want good diagnostics and maybe to tie in
with epidemiology, to know who's infecting who, then whole genome sequencing offers that if it
can be made rapid enough and easy enough to use.And so, there's a strong push to do all of that.

(19:20):
I mean, you need everything from how to get the sample from the patient prepared in the right way
to go in the sequencing machine… actually, that's the hardest bit now, the algorithms
and the computational bit is all reasonably straightforward now, especially for TB. So,
the tricky bit is getting sputum from a patient, most of that is human cells which we are not that

(19:41):
interested in sequencing. So, you need to sort of narrow down to the bacterial cells
and sequence just those.So, what's next for you?
Well, I guess there are two major things that are happening in the field technologically.
One is the volume of data that we're generating, keeps doubling. And that presents big challenges
for computational methods, especially if, as like me, you're interested in scanning through all of

(20:05):
the data that's ever gone before. Quite often you want to see has this been seen before. Is
this a strain of E. coli that’s been seen before? Is this a plasmid that's been seen before? So,
you need kind of search indexes. So, we develop algorithms to do that,
but it's very hard to keep up with data that keeps doubling. So definitely working on that.
And the other thing is that sequencing data itself is improving. And in our search for

(20:27):
antibiotic resistance genes and tracking them through bacteria, we often find that bacteria,
they're not satisfied with just having one copy, they’ll double up or triple up. They'll have
lots of copies in the genome sometimes, and they copy paste themselves in funny ways. And because
they've copy pasted themselves in funny ways, it actually becomes harder for us to reconstruct
the genome. If you've got a jigsaw, and more and more and more of it is copies of zebras,

(20:52):
it becomes harder and harder and harder to work out, you know, which zebra goes where.
But it's actually often quite important to know, you know, not just where the zebras are,
but almost like what their identities are, because different versions of an antibiotic resistance
gene can be subtly different. So, we develop algorithms to try and do all this jigsaw work.
And it turns out the new technologies are actually much better for that. So that's pretty exciting.

(21:14):
So, what's the ultimate goal of your research? So, what would make you happy,
if you get to the end of your career and you think, I did that, that's what I've
been trying to solve. Is there something?I can give you an answer now. I think it's
probably true if you asked me every year over the last decade, I'd give you slightly different
answers every year. But right now, I think if we could come up with a way of thinking

(21:40):
about evolutionary trees of bacteria and the evolutionary trajectories of plasmids
together, I think that would be amazing.Zam, thank you so much for talking with me.
This was a podcast by the Milner Centre for Evolution at the University of Bath.
I'm Turi King and thank you for listening. If you have any thoughts or comments on this

(22:01):
or any other episodes, please contact us via our X channel @MilnerCentre.
For more information about the Milner Centre for Evolution, you can visit our website.