at the University of Tennessee at
Knoxville did some time as a post back

at the do furry laboratory and then his
PhD at Penn in biology he postdoc first

in Michigan with Julianne Adams after
which he was hired on to the faculty at

the University of Idaho spent eight
years Idaho before me

medical school at the University of
Florida and after about three years of

that it's to Missoula Montana which was
where he was when I met him I've

actually been in England he was a
Montana for 15 years before it came to

Georgia Tech about two and a half years
ago all right Thank You Matthew

really appreciate that and thank you
guys for coming out I know that it's

fall break and consequently there is
sort of a diminished student population

or at least the undergraduates and thank
you also for the staff people who showed

up for this I interact with you guys on
a weekly basis and it's a special

pleasure for me to be able to tell you
guys a little bit about what we're up to

in the lab so my lab group is broadly
interested in some basic questions would

even say in evolutionary biology but in
biology related to adaptation and

speciation thanks to Matt heron who is a
research faculty leading the charge on

this project we're interested in the
evolution of multicellularity and in

various ways both supported through the
NAI as well as through exobiology

projects were interested in in the
origins of bio complexity sort of writ

large so the common denominator in terms
of how we approach these fundamental

issues is experimental evolution and for
those who were reading the news last

week there were people in chemistry and
biochemistry at Cal Tech Missouri and

Cambridge who won the Nobel Prize for
using this general approach of

experimental evolution to direct the
evolution of proteins so we use this to

direct the evolution of microorganisms
we can specify the selection pressures

in our experiments we can do controls
and replicate trials to the limits of

our patients or our students patients or
our budgets

and because we use model organisms that
have short generation times and are easy

to manipulate genetically we can do
things that are what really have been

inconceivable of more than a generation
ago for those of us who are using micro

organisms to do this evolutionary
biology in the laboratory we have the

additional power of being able to cryo
preserve a living fossil record so that

for any evolutionary series we can go
back to our freezers or to our Dewar

flasks and there liquid nitrogen and
pull out populations and pull out clones

from those populations for detailed
genetic and physiological analysis so

the talk for today is exploring the
emergence of complexity in clonal

populations and the the Latin title for
this which is actually the prequel to

one of the papers that I'll talk a
little bit about today see if I can make

this happen okay is explore ace so this
was actually the title given to this by

a colleague fellow nai colleague at the
University of Illinois

Nigel golden Feld and what this means is
out of one comes many and this is kind

of a plate on our national motto
eeep Laura bus in them out of many comes

one and it struck me this morning as I
was looking at this title that perhaps X

you know pleurisy is a sad commentary on
our political situation because it seems

that out of one we are becoming of many
many many tribes not a good thing all

right so so what do we mean by clonal
reproduction let's just do a quick

review for everybody in the room to sort
of get up to speed we mean binary

fission in bacteria we mean budding and
yeast this is a mitotic process which is

whose molecular biology is fundamentally
different from

we also mean reproductive mechanisms
like budding and planaria we also mean

budding in other groups such as these
marine invertebrates the bryozoans and

the cnidaria here represented by
anemones and of course clonal

reproduction oh it there we go
includes rhizome formation which is

responsible for our strawberry gardens
but it's also responsible for these

beautiful patches of Aspen that you see
out on the west that are largely clonal

aside from the few somatic mutations
that creep into the roots all right

so what else do we mean by clonal
reproduction we mean mitosis to make a

sand dollar body so in order to make a
this complex multicellular organism

starting from a zygote you have to have
a series of cell divisions and in

addition you have to have mitosis to
make a million-dollar baby this is this

is early human embryological development
leading to an adult of that species this

going through indirect development that
going through direct development I'm

going to get the hang of this soon
mitosis is also required to make liver

cancer or to make lung cancer and so
these are evolutionary experiments that

take place in our own bodies in fact I
would say everybody in the room has one

of these experiments ongoing at the
present it's just that our immune system

is doing a good job of surveillance over
these likewise that you have binary

fission that can make chronic bacterial
infections that result in long term

debilitating effects so how then do
clonal populations evolve so here you

have an initial founder cell and if you
don't have any sort of mutation creeping

into the population then in fact you
don't have evolution all of the

individuals are genetically out
: so I'm sort of suspending

consideration of epigenetic changes that
might creep into this otherwise clonal

population so is there where's Randy
third is there a delay on this always so

we have come on I really don't want to
have to do this let's see if we can make

this go all right so if you have
mutation then of course you can then we

become concerned with the proliferation
of this strain so here's a mutation

occurring in the population if this
mutant is not a lethal mutant and is

able to reproduce then we become
concerned with the fate of that new

mutation in the adapting population bear
with me for a moment folks that was

probably not a good idea come on alright
so how do clonal populations evolve

they've long been thought to be governed
by two related principles one ecological

and one population genetic the
ecological one is that of competitive

exclusion and this is work that goes
back into the early 30s the work of

Gauss and his famous experiments showing
that if you have a pair of species that

have very close nutritional and
physiological requirements that you try

to grow them together in the laboratory
that one eventually displaces the other

and so the the the term that Garrett
Hardin used to describe this in a famous

paper bin science back in the in the
early 60s was complete the competitors

cannot coexist a nice piece of
alliteration this is a supremely

aggravating is Randy here so we're
having a little bit of trouble with the

advance
okay so how do clonal populations evolve

the population genetic corollary to this
is that of clonal replacement and this

is a concept that was developed
theoretically by the great geneticist

Hermann Muller also back in the early
1930s and so what Muller conceived of if

you have this graph of individuals and
the frequency of particular individuals

as in an evolving asexual population
that you should have a new mutant

arising on the background of the
founding population which itself is

displaced when a new adaptive mutant
comes in and on the background of this

population that has swept you have a new
adaptive mutant that sweeps that

population so in this scenario you have
large asexual populations where you have

a succession of fitter mutants that
complete competitively displace one

another over time via periodic selection
so do a sexual populations evolving in

the laboratory actually adhere to these
principles these this ecological

principle and this competitive this
population genetic one so here is work

that was done by my postdoctoral mentor
back in the in the in the mid to late

80s and what he and his colleagues did
was follow the fluctuation in the

frequency of a neutral marker something
that you wouldn't expect

therefore to be under selection all
right and that neutral market was a

single gene that when mutated confers
resistance to bacteria phage t5 and the

model organism that they were working
with is just is E coli all right so this

is an experiment ecoli under glucose
limitation and this is on the y-axis the

frequency of these fade-resistant cells
and of course here is time measured in

generations of these ecoli growing where
glucose is the limiting reagent and so

what they observed was here is the
increase in fade resistance

background mutation rate the deflection
point was taken as an indication that

you'd had a new adaptive mutant arising
in the founder population whose

expansion pushes these down and then you
have its subsequent increase again at

the background mutation rate and then a
new adaptive mutant arises on the

background of a and so forth and so on
and so counting up these deflection

points which in fact was a time-honored
way of measuring mutation rate even back

in the 50s with no vikins the Lord they
concluded that there had been the

fixation of eight adaptive mutations in
this population so this suggested that

indeed we had empirical evidence for
periodic selection so let's bear in mind

that this is 1987 and then let's go
forward a generation and instead of

monitoring the incidence of a particular
mutant one gene one mutation in a

population my colleague Gavin Sherlock
and his student Dan Kavita BTech did

population monitoring using whole
genomes so now we're looking at every

gene in the genome of the simple
eukaryotic organism baker's yeast and

what he saw again looking at the percent
on the y-axis and generations on the

x-axis these are in fact called Muller
diagrams in honor of Hermann Muller what

he saw were so many beneficial mutations
coming into the population that they

actually interfered with each other's
clonal expansions and they interfered

largely because they're Fitness's in
this highly selective glucose limiting

environment were very similar to one
another so this is the first instance

where something that had been
theoretically predicted namely clonal

interference was shown to actually occur
in an evolving population and what is

clonal interference it is in fact
nothing less than a kind of a battle

royale where one fittest genotype rarely
prevails

largely because their Fitness's are very
similar in the evolving population and

something else to sort of take away from
this as an aside is that we're sort of

trained in bio 101 to think of
deleterious mutations are our and

neutral mutations is mainly what you see
in evolutionary biology in fact the rate

of beneficial mutations is is pretty
high okay so can an asexual population

evolve into multiple forms that coexist
under these models of periodic selection

and clonal interference it would seem
that the times in which you had multiple

clones stable coexisting within the
population was those those were very few

okay so but it would be interesting if
that was indeed the case and so we were

curious under what conditions this
arises so I'm going to use a term freely

over the next couple of minutes and that
is polymorphism simply meaning multiple

forms and and polymorphism here in this
context and this talk will refer to

genetic polymorphism but bear in mind
that we'll be talking about the

physiological polymorphisms that are
emergent from those genetic differences

okay so stable genetic polymorphism can
arise in temporally varying environments

and a classic example of this is the
ongoing Lenski experiments at Michigan

State University so what's going on
there you've had over the course of now

since 1991 rich primarily but also his
students coming in everyday and taking a

stationary phase bacterial culture ecoli
culture putting it into fresh medium and

letting it grow another 24-hour cycle
again taking a small volume and then

popping it like this and what you have
then are a seat or seasonal environments

where the cells initially experience a
replete nutrient environment go through

rapid growth and then inter stationary
phase

so the bottom line is that they go over
a 24-hour cycle from plenty of nutrients

to very few nutrients and dani rosen and
dominique schneider and others have

shown that you can get stable
polymorphism existing at this in these

environments because you get mutants
that do particularly well in one of the

other of these growth phases another
classic example not here necessarily or

only temporal variation but also spatial
variation so Paul rainy and travisano a

nature published a very important paper
20 years ago and what they were able to

show is that if you don't have a stirred
environment but simply go from beaker to

beaker and let this beaker sit over the
period of growth of these bacteria that

you quickly arrive at a number of
different Morpha types within the poly

within the population a polymorphic
population that stable over long periods

of time so then it's clear that these
sort of batch cultures if they're not

stirred becomes spatially structured
with respect to oxygen tension and this

sets up selective conditions for
different morphs to persist together in

the same vessel in the same evolving
population over long periods of time so

stable polymorphism can arise in the lab
it can also arise and be perpetuated in

nature if there's some sort of
differential selection in time or space

if there's temporal seasonal variation
or spatial variation another way of

looking this from the population
geneticists perspective is if there's

some sort of balancing or frequency
dependent selection on the population

that gives at different times or
different places

selective advantages to one or the other
morph but this sort of thing of

polymorphism should not be stable in a
constant environment and with one

limiting resource so about the simplest
though not in terms of how you set up

the vessel but in
what's going on in the vessel about the

simplest experimental arena that you can
devise is that of a chemist at this is

not something that I made up the great
Nobel laureate from France Jacques mano

was the first who pioneered this back in
the late 30s and 40s the idea here is

that you have all nutrients but one
present in excess and one president

limiting amounts and you feed this and a
carefully controlled manner to a

bioreactor this establishes in this
bioreactor which has an e flux it

establishes conditions where you have
physiological steady-state

and the rate at which the cells
reproduce in the bioreactor is exactly

equal to the rate at which they're being
washed out so you have a careful control

over growth rate in the experiments I'll
talk about we are conducting these at

low growth rates under glucose
limitation and we're aerating these

cells also so this is a well-mixed
aerated glucose limited environment

instead of these fluctuations in a
chemist at to set the number of cells is

constant over as many generations as you
have the patience to perpetuate the

experiment so you have then no steady
you have no temporal variation if

steady-state conditions it's well mixed
so you have no spatial variations

so in principle those conditions in the
lab and a nature that promotes stable

polymorphism are are averted and with
the use of a chemist at and yet you have

a number of examples from Julian's lab
from my work from that of other

laboratories where you do see stable
polymorphism different genotypes

persisting for long periods of time in
this simple environment so X you know

place out of one comes many for the next
five or 10 minutes I'm going to talk

about strains that was isolated in a in
a paper in genetics by helling it al in

1987 the bottom line
is that these guys thought that they had

a contaminated culture from about 300
generations on to this

they had cells in there that were large
and small

they had cells in there that were Epis
ill and resistant some that were

ampicillin sensitive and at the end of
the experiment they isolated some

colonies from that experiment that they
could differentiate in this manner and

effort and analyze them further and then
I came in a few years later and did some

other work on these the the main thing
that they reported in this paper is that

this ancestral strain had this growth
rate and yield and batch culture and all

of the evolved strains have
differentiated from that ancestor in

interesting ways all but one were had
higher growth rates or higher yields but

one of them in fact this III strain had
a lower growth rate and lower yield not

in the chemist at but when it was grown
in one of these Lensky style batch

cultures my contribution to this was
showing that you can take this simple

community represented by these different
clones and you can take them out of the

freezer and pop them into the chemist at
and they predictably form a community of

the following numbers where e1 is always
dominant in the chemist at even though

it's growth rate and yield in batch
culture is less than these other clones

so stable polymorphism of this nature
arises repeatedly under nutrient

limitation so in a number of parallel
experiments conducted with strains that

are closely related to this ancestor
anytime they ran the chemist at more

than 100 generations they saw these
small variants that look like this see

v103 strain and so the question then and
this is where I came in was if you have

a single limiting nutrient then both
Miller and Gauss predict that you should

not be able to have clones stable
coexisting over long periods of time so

there must be something else in the
medium that is supporting the growth of

these other strains and so what I did
was consider all the various

fermentation products of e.coli as well
as all of the diffuse' belén Tirmidhi

'its that are present in the glycolytic
pathway as well as the TCA cycle and

analyzed all of those in the spent media
and discover that in fact there were

some of these overflow metabolites and
some of these intermediates this is sort

of a summary of work from that paper we
showed that there was increased glucose

uptake almost twice as great glucose
uptake in this III clone but it left

plenty of acetate it also left glycerol
and Lissa all three phosphate one of the

Clones has come to specialise on that
acetate and another of the clones is

able to access preferentially those
glycerol and glycerol 3-phosphate so

this led us to propose this model that
you have stable polymorphism being

perpetuated in this experimental
environment through cross feeding so

starting out with an ancestral strain
and over many generations ending up with

these others that are interacting in
this fashion through these secondary

metabolites well it turns out that
stable polymorphism repeatedly involves

these secondary metabolites
so in Along Came David Travis a number

of years later and in those instances
where these small colony variants

reminiscent of the ìiî type we're seeing
the majority of cases you also saw these

acetate scavengers are arising in
parallel and so this led us a couple of

years ago to propose that in addition to
periodic selection and in addition to

clonal reinforcement that there's
another way

of looking at evolution of asexual
populations particularly microorganisms

and it's like this that you have a
phenomenon that we call clonal

reinforcement which means that you may
have periodic selection taking place you

may have clonal interference taking
place but both of these can take place

on the background of a situation where
you have novel variants arising within

lineages whose basic physiology is
intact but these lineages are sort of

wedded or welded together by biochemical
interactions okay so this is this is

what we're proposing as an alternative
view to clonal replacement and clonal

interference and we're not proposing it
as something that is necessarily to

supplant those alternative views but we
suggest that in fact clones may evolve

this way and incorporate these other
models so it turns out that clone will

replace the reinforcements can be it's
pretty much synonymous with something

another term that has arisen in the
literature enter clonal cooperativity

and this term is generally used in the
discussion of chronic infections caused

by bacteria and fungi as well as in
cancer and there is new evidence coming

forth even over the last year that these
sorts of interactions positive

interactions amongst the clones in a
chronic infection positive interactions

amongst the clones and a genetically
heterogeneous tumor are important for

their diagnosis as well as for their
prognosis and advising effective drug

treatments so what are these guys anyway
and and when I say these guys we're

talking about this simple system we
evolved in the lab but I would invite

you to think more broadly to think about
cells that might be in a persistent

infection say in the law
the urinary tract as well as evolving

tumors within the body are they
co-operators are they competitors

are they something in between and so
Margie Kenner's Lee and other

collaborators have been interested in
the transcriptional patterns that

underlie these simple communities and
the genetic differences that help

explain those patterns all right
so here you I want to walk you through

this carefully on the top is a heat bar
indicating the transcript levels for a

particular gene and a treatment relative
to the ancestor so it's always this

comparison we can grow each of these
clones by themselves in monoculture or

we can grow them together in a community
right and so again each of these

comparisons is relative to the ancestor
this means that the transcript level for

a particular gene is high
this means that relative to the ancestor

that transcript level is low so the
point then of this slide is that for

many genes the transcript levels differ
when the evolved clones are grown by

themselves these are the same as when
they are grown as a community and just

to pull a couple of out here is AG like
upon Lambie that's responsible for

getting glucose through the outer
membrane and here is MGL BAC which is a

glucose transporter which is on the
inner membrane as you can see it across

the board these genes are very highly
expressed relative to the ancestor

however this is not universally the case
for other genes the transcript levels

differ and we know that this community
is about 80% III and yet the

transcriptional profile of the
consortium or the community

recapitulates II 1 e 5 and e 6 and in
important ways differs from that of the

predominant clone when that clone is
grown by itself so what's going on the

inference that Margie came to
was that the transcriptome of this III

clone the dominant clone is different
when it's grown by itself as compared to

when it's grown as part of a community
and so how might that be and what that

what might they say about this sort of
inter clonal cooperativity it turns out

it's pretty straightforward to explain
so this dominant clone is consuming

glucose it's very avid for this primary
this limiting substrate but it's

wasteful in its use and we know from our
experiments that it's producing lots of

acetate it also it turns out is unable
to assimilate this low levels of acetate

through this pathway acid to you Co a
synthetase and that'll come up in a

moment the bottom line is that we know
that just like we have product

inhibition through lactate in our own
cells yeast have product inhibition

through their fermentation product
ethanol and the same thing goes for

e.coli they have product inhibition
through acetate well it turns out that

acetate product inhibition is mediated
through its immediate it's its previous

step in glycolysis or in fermentation
acid to a phosphate this compound which

basically backs up in the cell in its
phosphorylated form is able to serve as

a phosphate donor to a couple of key
transcription factors CP XR and op R and

the phosphorylated form of these
effector

genes is in fact able to influence the
expression of a wide variety of genes in

the e coli genome and guess what it is
just exactly those genes that are over

represented in that sort of interesting
situation where III transcriptome

differed when it was grown by itself as
opposed to when it was grown in the

presence of the other clones
thus III when it's part of the community

has the benefit
that accrues from having II one present

with it consuming this acetate it
relieves product inhibition you don't

have this unusual record or this
regulation of these CP XR and op our

genes through phosphate and it
essentially you have a community looking

exactly alike across the way e1 e3 and
so forth so in what order did this

community arise and what is the genetic
basis for the expression differences but

you know before I sort of launch into
that I'd like to to point out that it

not only is e one getting a benefit from
the e3 strain by virtue of its

consumption of acetate II one is also
conferring a benefit on e3 in enabling

it to continue to scavenge glucose at
great rates in the absence of product

inhibition so it's really a cooperative
arrangement they've come to alright

another paper by a Margie Kindersley who
did whole genome sequencing with Jared

winger at Stanford University and their
experiments indicated that major lineage

is diverged early in this experiment so
remember these are simply clones that

were taken out at the end of the
experiment that are representative or

what Julian and Bob and Chris thought
were representative of these these

Morpho types but these Morpha types
clearly diverged early in the experiment

and they've been Cote evolving over long
periods of time so it might be obvious

to you that this III clone which was
taken out at the same time as the e1 and

the e6 clone has many more mutations
this is in fact the I can make this

thing work this is the result of the
whole genome sequencing where Margie

found unexpectedly high levels of
nucleotide diversity almost 600 total

mutations distinguish these strains and
more than two-thirds up or up almost

two-thirds of them are missense and
nonsense mutations and there's a strong

bias towards transverse
versions in fact 99% of these mutations

are transversions indicating that this
is probably being driven by

hypermutation so if you have a defect in
DNA mismatch repair that not only

elevates the mutation rate it also
predisposes the the cell to accumulate

transversions and so III turns out to
have almost two to three orders of

magnitude higher mutation rate than the
canonical wild-type k12 but it also

turns out that the ancestral strain was
also a mutator by virtue of another

mutation in DNA mismatch repair a
nonsense mutation in mud Y so X in a

place from one committee but how and so
we're interested in which are the

mutations out of these 584 that are
actually the drivers of this process of

stable polymorphism and which are merely
the passengers in the that are

hitchhiking along but these very
beneficial mutations and one of these

days so here's the ancestral strain this
is j1 22 and this is a depiction of the

important steps that we believe are
involved in establishing this

polymorphism and here is the e3 clone
and here are the the driver mutations

that we think are essential to the
establishment of this the simple

community so this evolved strain III has
accumulated a number of mutations that

affect glucose transport both across the
outer membrane and the inner membrane as

well as mutations in glycolysis and the
TCA cycle importantly in Lippo mi

dehydrogenase which is a member of
several multi enzyme complexes

importantly the pyruvate dehydrogenase
complex here and suck a be down in the

see a cycle work not done by ourselves
but by others in sort of regular

mutational screenings have shown that
the particular mutations that arose in

LPD in this stream have previously been
shown to diminish flux through the TCA

cycle and to diminish the growth rate of
the cell phenotypes that we both observe

in previous previous work in our
laboratories alright so it turns out

then that you can increase flux through
the glycolytic pathway so you have more

acetate being produced by e3 it turns
out that III retains an ancestral

mutation that essentially restricts its
access to that acetate through a

mutation in acetyl co a synthetase which
is the high affinity acetate transporter

in e.coli this acetyl genic fermentation
actually creates a drain on nad plus in

in the glycolytic pathway because you
are not regenerating it the way you do

in yeast through alcohol dehydrogenase
so in fact this creates a redox balance

in cells that are growing anaerobically
or fermentated li and that places a

premium on anything that can regenerate
NAD+ that enables glycolysis to continue

and so one of the ways in which you can
regenerate NAD+ is simply by making

glycerol so we believe that that's has
been one of the responses by the sell-by

III that result in the production of
other secondary metabolites another key

mutation in this III strain in fact to
missense mutations are in pts i which is

known to have an inhibitory effect these
mutations on literal kinase so that

restricts the access of this III strain
to this secondary metabolite glycerol so

along comes a 1 and a 1 has
a an insertion element that has landed

upstream of a CS this results in
constitutive over expression of this key

gene to access acetate and then the e6
strain is able to have access to a

glycerol and glycerol 3-phosphate by
virtue of retaining the the the parental

capacity to assimilate glycerol alright
so I made this point several times that

under anaerobic conditions that this e3
clone is behaving like an anaerobic cell

so here you have the ancestor grown
anaerobic here you have the ancestor

grown aerobic and you can see that
because the denominator is lower the

redox ratio of NADH to NAD+ is like this
under anaerobic conditions and under

aerobic conditions II one growing an
aerobic chemostat looks more like the

aerobic ancestor III looks more like the
anaerobic ancestor

so we're believed that trade-offs here
drive the evolution of complexity you

have one string that's best able to
scavenge limiting glucose and favoring

fermentation over respiration and
creating a redox imbalance putting

pressure on the cell to regenerate NAD+
and this is supremely an aggravating III

then has restricted access to acetate
and III has restricted access to

glycerol so then the primary resource
specialist is a wasteful scavenger it's

avid for the primary nutrient but it's
wasteful in its metabolism so now we

have three models full of clonal
evolution this of clonal replacement the

old periodic selection model this of
clonal interference shown very nicely by

Gavin and by many others and
a clonal reinforcement that incorporates

elements of the other two what which is
really based on the notion that you have

you have different lineages interacting
with one another biochem chemically over

long periods of time in ways that are
favorable to their perpetuation so are

there any ecological or evolutionary
advantages to this arrangement and this

is where a talented postdoc and
undergraduate came into the laboratory

and attacked this problem
so Ashley Alexander who's here and

dongdong who's probably back in the
laboratory integrated GFP protein into

these strains for the with the idea of
doing competition experiments and here

you have a pairwise competition between
the marked ancestor and the unmarked

ancestor and that essentially gave you a
competition coefficient of one because

what the ancestor only barely displaces
the mark strain of the ancestor and so

this selection coefficient is accounted
for in all of the data that I'll show

you here presently so now here you have
each of the evolved clones grown by

themselves that have much higher Fitness
than the ancestor perhaps not

surprisingly and then you have the
various combinations of III one and III

six that can be grown together and the
highest fitness of all is that of the

entire community ricans reconstituted as
III 1/6 this is reflected not only in

the Fitness differences amongst these
different strains but also in

significant differences in their
productivity in the same general order

increasing from the ancestral to the
individual strains grown in monoculture

in these pairs and the whole community
the point then is that even under the

simplest of conditions biodiversity
builds upon itself and this biodiversity

building upon itself has biomedical
implications

patience so if you imagine in the panel
a where you have a homogeneous tumor and

you have some sort of selection pressure
such as chemotherapy and if you have

chosen wisely your chemotherapy and are
able to eliminate this homogeneous tumor

then in fact you have killed the cancer
if in fact you have a genetically

heterogeneous tumor and that the some of
these cells these clones are actually

interacting with one another in a
mutually beneficial way then it means

that you might be able to kill some but
not all of the members of this

genetically polymorphic population and
those that are resistant to this

selection pressure the chemotherapy are
in fact those cells that are responsible

for the the relapse so I'll call your
attention to just read off the bottom

here from a recent paper in Nature revue
cancer cancer cells behave as

communities and increasing attention is
now being directed towards the

cooperative behavior of subclones that
can influence disease progression and so

I submit to you that the sort of work
that we're doing has direct bearing not

only on infectious disease where we know
that bacteria can cross talk with one

another but also in cancer so if these
things if these communities form how

stable are they and our collaborator
Exeter University Ivana gudelj divided a

simple model to address this question
that's really based on ATP generation

and glycolysis versus the TCA cycle and
the velocity of the processes of

glycolysis the production of extra
cellular intermediates the import and

the and the TCA cycle and what she was
able to show is that the stability of

these consortium or these communities
depends on initial population density

and the frequency of the secondary
specialists so in green a primary

resource specialist in red a secondary
specialist like the e1 or

six strains and here are these
intermediate areas of resort of

population density and the frequency of
the secondary specialist where you have

can expect cross feeding to emerge in a
chemist at this y-axis here is really

the fitness of the mutation giving the
differential access to this secondary

resource and this is the amount of
limiting resource that you're actually

putting into the chemist and so then
clone reinforcements may not inevitably

arise and it might not even be
sustainable when it does arise and so we

did recently a repeat or a Redux of the
heling @al experiment starting with the

same ancestor J 1:22 except and running
it under exactly the same conditions

glucose limitation slow growth aerobic
metabolism except now instead of picking

out a few clones and analyzing them in
exhausting detail we did whole genome

sequencing every 50 generations at 1,000
X coverage and so what this then gives

you is recovery of every allele that has
the new allele that has risen in the

population to a frequency of about 1% in
order to get a handle on the linkage

relationships amongst these different
alleles how they're Bend in different

genomes we did whole genome sequencing
on 96 clones from each one of these

populations at the point where we
determined that allelic diversity had

reached its asymptote which typically
was towards the end of the experiment so

then this is I'll just rip through some
of this genomic data the rate at which

these new alleles appear so this is the
percent of a particular allele this is

the number of different mutations that
are accumulating within the whole

population the rate at which they
increase is about the same this may look

shallower but bear in mind this is only
300 generations

again a large fraction of these are
nonsense and

since mutations that one could construe
would have some sort of impact on

Fitness the population sequencing
recovered about 3,700 snips that rose to

this 1% frequency at at least one time
point across the experiment again

starting with this same ancestor
majority of them are trans versions the

clone sequencing recovered about 11,000
mutations again most of them trans

versions in this slide on that on the
y-axis we have the clone frequency for a

particular allele here we have the
population frequency so the frequency

with which the allele appears in the
population itself and you can see for

each of these experiments there's a very
good agreement across the line between

the clone frequency and the population
sequence of frequency what this means is

we can have confidence and our ability
to construct the linkage relationships

using the whole genome sequence of these
hundred clones that we pulled out of

each experiment so now we can construct
our own Muller diagrams and here are

some representative ones in the paper we
actually have a lineage for all the top

50 lineages that you can scroll through
and it's really quite quite amazing to

me have witnessed going from one gene to
being able to look at whole genomes

again molar diagrams with percent or
frequency on the y-axis generations on

the x-axis and the take home from this
is that you have some mutations like

mg/l be upstream that arrives early and
sweep the population and you have others

like hfq that arrives later and are kind
of fine-tuning the Edit adaptation of

these cells to glucose limitation so
certain genes are hit multiple times in

independent lineages and so here you
have our based on the target size of the

gal s gene or hfq this is the number of
expected mutations we actually see many

more than that
very significant differences

interestingly we never saw any ACS
mutations we saw no PT si mutations we

saw no LPD mutations these Keystone
mutations that we believe are the

drivers of the establishment of stable
polymorphism in these other experiments

never saw this at all
what we did see is that most of the top

hits impact glucose assimilation and
they do so in interesting ways you have

many of these new alleles arising in
different lineages within the same

population so evolution is really
exploring a variety of different

solutions to the problem of this glucose
limitation so we're pitching this as a

new way a complementary way to classical
sort of molecular genetics to doing

analysis of protein structure function
as well as understanding the wiring

diagram of a simple cell and so here is
how glucose gets into an e coli cell

under glucose limitation through lamby
the glycol pouring and through MGB mg/l

BAC and here you have all of these top
hits these are the ten top hits and you

can see that they all affect the
expression of these genes in interesting

ways and I'm not going to I'm gonna just
take you through a couple of these and

show what we've done additionally with
with our experiments so let's just take

a look at gal s you have gal s or to the
binding side of gal s upstream of MgO

BAC operon you have like forty different
de novo alleles that reached at least 1%

frequency in the population well boom
here they are well coated with respect

whether shape which chemo staff they
came from and the different type whether

it's missense nonsense etc and where
they land in the gal as gene 0 to a 100

that's the frequency with which they're
observed in the population and what you

can see
is all but most of these never rose

above 1% this missense mutation in
chemists at one was fixed and these

operator mutations upstream of MgO ABC
were all fixed

here's another lamby
so lamby expression is key to glucose

import and one of the things that
controls lamby expression is hfq this is

a an RNA chaperone that controls
translation of the of Mik a which is an

inhibitor of lamby so anything that will
screw up that interaction is likely to

increase lamby expression here are the
hfq mutants

so again zero to 100% a few reaching
high frequency most at low frequency but

what's interesting is if you take these
mutations and layer them paint them if

you will onto the protein you see
something really quite curious here's

the hfq hexamer so it forms a it's a
multi merit protein there's the distal

face where you have RNA binding on the
distal face and now we're going to take

the mutations and we're going to paint
them on the molecule and what you see is

that missense mutations in red here we
have nonsense mutations in blue they're

all located on this distal face and on
the rim side of the six mer these are

the sites of RNA binding and they almost
certainly impact that interaction

between the RNA target and this hfq
molecule which incidentally has a number

of other partners
aside from McKay so let's consider this

interaction which influences ultimately
lambie expression ball k is an inhibitor

of mal T which activates expression of
this of these genes and so here we have

what is it something on the order of 22
mal cavitations they most of them go to

low frequency here's one that has swept
the chemists at a

and if you paint these onto Malky you
see that all of these mutations in fact

localized to the region where maokai and
mountie interact so evolution has

basically unlocked the key by knocking
out the capacity of mal k to interact

with this partner protein so the
takeaways then from this helling it out

Redux is that deep population sequencing
plus whole genome sequencing on hundreds

of clones shows no evidence for this
clonal reinforcement that I've been sort

of banging my shoe on the table about
for the last 50 minutes instead we see

widespread clonal interference we see
parallelism in the sense that we see the

same genes being hit in different
experiments the same genes being hit and

different the different lineages in the
same experiment we do not see any of

these key cross feeding mutations in ACS
PT si and LPD and what this really I

will have you take home from this is
this is the importance of historical

contingency that either the presence or
the absence of certain mutations in an

evolving population can actually favor
or preclude the evolution of other or I

won't say the evolution of but at least
other genes going to fixation in that

population and then the last point to
take away from this is that adaptive

genetics is complementary to molecular
genetics and we can use that to

eliminate a cell's
wiring diagram as well as to gain

additional insight into protein
structure function by doing this

business of painting adaptive mutations
mutations that we know are if it goes to

from one cell out of 10 billion to one
percent frequency in the population so

even if it winks out of existent at for
some period of time that is an adaptive

mutation and we can paint these onto
3-dimensional structures and get

interesting insights
into the structure functional relations

and how proteins bind with their
partners okay in the interest of time

we've been slowed down a little bit with
the AV but I will stop it here and say

that the people who did the work where
Margie Kenner's Lee Karen Schmidt at

University of Montana kochia Jared and
Gavin at Stanford always a shout out to

my postdoctoral mentor who's been a
important figure in my life

Eugene crow Ashley Alexander dong dong
yang all working in the lab here at

Georgia Tech and our laboratory manager
Emily cook who is truly an indispensable

member of our team I thank you for your
patience and I'll take any questions or

I'll take any comments matthew rhys two
different ways I think I think the same

question
which is first do you see this clonal

reinforcement as a different process
from negative frequency dependent

selection and the other way of asking
the question is how do these clones

behave in reciprocal invasion
experiments I will start with the second

one and I will say that in the
reconstruction experiments that I did

we always biased I always biased it
wasn't anybody else I biased the result

in trying to put them in at the
frequency that they were that they were

observed there were some experiments
where I would start them at equal

frequencies and they would sort
themselves out but I never did the

experiment of say establishing III and
then throwing a 1 in there and see how

we know how well it does they were
always put in there together so actually

it's an interesting experiment you know
could a 1 evolve an e 3 population my

prediction is that it would be able to
very easily because we know that e 3 is

leaving living a lot of acetate in the
medium but the experiment has not been

done is it different from negative
frequency dependent selection is that

that was question number one I actually
used that term in my in that 94 paper

and in fact explored that possibility by
changing the amount of substrate so you

can play with the equilibrium abundance
of these different clones by changing

the amount of glucose that you feed in
you can also change it by putting in

acetate or putting in glycerol and so I
think that I think they're very it's

very similar

yes sir where the mutation rates to the
changing the growth rate so if you treat

your key was done
speed everything up where we getting my

restrictor
so the question is is mutation rate

sensitive to growth rate so the the the
immediate answer to that is I haven't

done that experiment and I'm trying to
think if anybody has done so with the

explicit purpose of seeing what mutation
rate would be under those different

circumstances one so here I'm going to
venture into the land of speculation

which is a fun place to be so it's it's
pretty well established that that you

have an increase in the incidence of
mutations under under stress right and

in fact when minoo first invented the
chemist at he called it a mutation

machine right so the idea he was trying
to get a handle on mutation rate but

also wanted something other than just UV
light or hydrogen peroxide to actually

make interesting mutants and he had the
insight to make interesting meetings in

a very directed way by placing these
under some sort of nutrient limitation

but paying attention to which nutrient
you know was being limited so to the

extent that say so these were conducted
at point two per hour which means that

the whole population has to double every
five hours in order to sort of keep up

with the delivery of the medium if if
dialing it down to say 0.05 is in fact

more stressful to the cell and I think
we could argue that it would be because

they would be you know even even less
you know nutrient is available because

there is this relationship between
growth rate and the in the equilibrium

nutrient
concentration then my speculation would

be that you might see a higher mutation
rate on the other hand let's sort of

argue it the other way so so a variation
of this sort of continuous delivery

system is something called a turbid
estat a turbid estat is where all of the

nutrients are present in super abundance
and you are diluting the the vessel at a

very high rate and close to the maximum
specific growth rate of whatever the

cell is that you're playing with and on
one hand you might say well these guys

are in Fat City they should be just fine
on the other hand we know that that that

that transcription itself can have an
influence on mutation rate and to the

extent that these cells are carrying out
transcription at certain low side just

you know basically as fast as they can
because they're reproducing as fast as

they can then you might have a high
mutation rate under those sort of me max

conditions I will have a colleague at
Colorado who does turbid estat

experiments she doesn't do chemists and
experiments I'd I would I would I'll ask

her to sort of weigh in on that so I
could argue it either way that you might

see an increase in mutation rate sort of
when you approach the the MU max and

then you might see an increase in
mutation rate when you get close to out

an out stationary phase in a chemist and
this is that's a that's the best hand

waving that I I can do here but it's a
great question

Ashley

the model of that
would you speculate that in a tumor

environment if excess glucose

more likely for a cross meeting
event to average about your environment

and conversely if you were to limit the
amount of glucose

yes so that's the answer to that
question is yes so as as many of you

guys know it has been known since the
1930s and 40s that most tumors exhibit

you know kind of a Warburg effect in the
sense that like yeast they are very avid

for glucose but even in the presence of
oxygen they carry out fermentated

metabolism it's a it's a curious thing
that so many cells achieved their

maximum growth rate on glucose under
fermentated conditions and it turns out

that most cancer cells exhibit this as
sort of a it's like a blueprint of a

cancer cell what this leads to and this
again has been known for decades is that

for say a solid tumor the pH of the
environment in a solid tumor is very

different from the pH of the surrounding
tissues and this is due to nothing more

or less than the production of our
overflow metabolite lactate all right so

the presence of so much lactic acid
building up locally in a solid tumor

leads to very very different pH profile
I think what you're suggesting Ashley is

do you get inter clonal cooperativity
arising within solid tumors where there

is the possibility for lactate and other
overflow metabolites to be locally

confined providing a secondary resource
for primarily oxidative cells to consume

those I think that I would say yes to my
knowledge no one has you know

specifically addressed that question but
then you ask the other question would

glucose limitation have an effect
overall on cancer and I think there

there actually a lot of people who are
trying to to look at this question we

know that caloric restriction is is the
only way

that you can extend the chronological
lifespan of any cell alright this has

been shown you know from yeast you know
all the way up to primates whether this

also has an whether part of the equation
at least in higher eukaryotes is staving

off cancer by being on a near starvation
diet I don't know that anybody has sort

of connected the dots there but it is it
is an it is an interesting correlation

and I know that for example there's a
sunfish Jenny is interested in is

basically starving cancer cells as a way
of at least making them or making them

more vulnerable to chemotherapy

yes sir just one condition Oh what
initial substrate

but or using a different limiting
nutrient like phosphate or hundred can

it happen with multiple nutrients at
once well you've asked that question in

the right setting so our colleague over
here Matt heron did published a paper in

PLoS Biology about four years ago and
this was a serial dilution experiment

but he actually was doing a dual
nutrient experiment with acetate and

glucose also using e.coli but with a
different ancestor Matthew would you

care to comment

now you might ask the question is there
a substrate on which you would not

expect this to occur and it seems to me
that the the key to this particular type

of interaction and maybe to interaction
in genetically heterogeneous tumors is

having a fermentable carbon source in
other words you have almost like the the

Greek mythological creature you know the
harpy you have something that is that is

devouring everything at the table but is
excreting it you know almost as fast as

it sits as it's eating it and so that
creates opportunities for these other

clones to specialize but if you had us
say a simple if you had a simple carbon

source if you were limiting on a simple
carbon source like acetate or glycerol

with which just about all you can do
with that is to turn it into carbon

dioxide and some biomass and you know
get a few ATP's and reducing equivalents

out of it I would say it would diminish
the likelihood that you get this sort of

interaction established just because
there's not that many ways that you can

push metabolism if that's the only thing
that you're eating there's not many ways

that you can push metabolism to produce
you know secondary resources does that

answer your question yet now if it now
you might say well okay let's forget

about a simple carbon source what about
we limit on ammonium right you can limit

on ammonia more nitrogen or sulfur again
I think it depends on I think it depends

on the carbon that's being fed to the
cells of whether it's fermentable or non

fermentable that sets up the possibility
for these sorts of cross feeding

interactions yes sir chance

that's a great question so the question
is in these particular experiments a

complicating factor is the is the fact
that the the ancestor is a mutator that

has about an order of magnitude higher
than wild-type and then this clone that

appears or this the lineage that is as
driving the the cross feeding has an

even higher mutation rate restate your
question again I'm sorry I was right

gotta drive get it yes thank thank you
so unfortunately in that experiment we

don't know exactly when that mutt in
mutation arose what I can say is so so

Margie for one of her papers did go back
and so we have these driver mutations

that we think are important to the
establishment of this community and what

she did was go back to the point in the
experiment where these small variants

first arose and I will say at that at
that point the driver mutation when I

say arose that means they had reached a
certain frequency in the population

where you just see him on a plate
alright so they were pretty abundant at

that point so at that point you had the
LPD mutation you also had the mutator

mutation so at least halfway through the
experiment we know that this defect in

mismatch repair arose accounting almost
certainly for the fact that you see so

many more mutations in that lineage than
you do in the other two

okay well thank you all very much I
appreciate your time good luck