Highlight text to annotate itX
Hi, I'm Sharon Long,
from Stanford University,
and I'm here to give my lecture on nitrogen-fixing symbiosis.
Today, I'll be talking about the regulation of the genes in the
bacterium itself and how those genes are controlled.
So just as a review, nitrogen fixation happens in root nodules.
These nodules form as a result of action of
soil bacteria interacting with a specific plant.
Here, you see nodules growing on the root system of a soybean.
This is a complicated series of tissue and cell interactions,
a complex developmental process,
and there's specificity between the bacteria and the plant.
I'll talk about one side of that specificity a little later today.
In the early stages for sure, and I'll make the case that it's
in the later stages as well, bacteria and plant exchange
signals in order to coordinate with each other.
Now, let's take the standpoint of the bacterium.
It's making this huge transition from the free-living bacteria,
through invasion, in which the bacteria are able to form
an infection thread and penetrate through individual cells,
and then into multiple layers of the root, and finally,
nitrogen fixation, where the bacteria have become endosymbionts,
inside the cytoplasm of a plant cell,
and they persist there in a differentiated form, almost like an organelle.
All of these changes are accompanied by carefully regulated
changes in gene expression.
Here's an outline of what I'd like to tell you today.
I'll begin by talking about a few tools that we use to study
the genetics and the gene expression of the bacteria.
And then I'd like to say something about the control of the nodulation genes.
This is probably the most studied of any of the gene regulation
events in symbiosis. I'll review some of the background,
and then I'll talk primarily about the protein activator NodD.
I'll then tell you some of our recent work on RNA polymerase
and especially the sigma factors and how they may be
playing a role in changing and guiding gene expression during symbiosis.
I'll finish with several glimpses of how particular sets of
genes are controlled, and a preview of all of the factors
yet to find in terms of how many genes are active in nodulation.
So, the first tool to remind you about is my favorite, and that's genetics.
We can take bacteria and, through mutagenesis and then
blind screens, we can ask what are the most
important genes for symbiosis.
Now, there's a couple of different symbiotic phenotypes
that we have observed. In contrast to a wild-type situation,
where the bacteria growing in the free-living state are able to
interact with the correct host plant and make functional nodules,
shown here by having a red color reflecting the
leghemoglobin that's inside the plant.
The nodules harbor bacteria that fix nitrogen, the plants are able to
use that to make chlorophyll and protein, they flourish,
and they're nice and green.
But, if the bacteria have a mutation in a gene that's
required for the formation of the nodule, the result
will be that the plant forms no nodules.
Therefore, it's not getting any fixed nitrogen, and it starves,
it's yellow, and it doesn't grow very well.
We would call this a nod gene in the bacterium
because mutations in that gene result in the so-called
Nod- (no nodule) phenotype.
The other outcome, again comparing to the wild type,
would be that a bacterium might be able to form a nodule,
but those nodules don't function, and what I've shown here
is I've symbolized that by these empty, white circles.
Those show nodules which are not fixing nitrogen,
they don't have leghemoglobin, and so we would call
this Fix- phenotype,
and the corresponding bacterial gene is going to be a fix gene.
Another approach is to look at the genome.
Bacteria tend to have genomes between about one million
for the very smallest pathogens, all the way up to 10 million bases.
Rhizobium is on the large size, over 6.5 million bases,
divided into three replication units: a chromosome, 3.65 million;
and two very large so-called megaplasmids.
They're called megaplasmids because they're more than
one million bases each.
pSymB, 1.68 million, and pSymA, 1.35 million.
The nodulation and nitrogen-fixation genes are located on pSymA,
but in addition, there are symbiosis genes on pSymB and
on the chromosome. So to some extent,
the symbiosis genes are dispersed.
Now, we can take a genomic approach to look at transcription,
and one way in which we have approached that
is to do global assays of transcription
using a custom-designed Affymetrix chip.
On this chip, we place the genome of the bacterium
together with 10,000-some genes from the plant
as follows: The entire Sinorhizobium meliloti genome is there,
every one of the 6000+ open reading frames and the intergenic spaces,
and 10,000 tentative consensus sequences from cDNAs
that were sequenced from the plant.
Now you'll see some experiments later in which
we're looking at the expression of genes in Sinorhizobium meliloti,
and in one case, we're going to be simultaneously looking
at both the bacterium and the plant.
And we can do that from a single RNA prep, as shown here.
So, if you grind up a nodule,
so you're starting with a nodule over here...
of course, this is going to be full of many plant cells,
each one of those plant cells has bacteria,
and those bacteria shown here have differentiated.
So a nodule lysate is going to have a
mixture of all of those materials.
We have a method which allows us to capture both bacteria and plant
DNAs at the same time, and that can be shown here.
For reference, in the right-hand lane,
we have isolated RNA from a plant root,
that shows you the large and small ribosomal RNAs here, for reference.
On the left-hand lane, we have bacteria grown in culture.
You can see again, the bacterial RNAs here.
Our prep for nodule RNAs captures both of those, as you can see.
And this, by being labeled and hybridized onto our Affy Chip
allows us to get a snapshot of both partners at the same time,
which is very useful, especially when you're analyzing mutants,
and useful because perfect synchrony of nodulation
is hard to achieve experimentally.
So, having introduced those tools,
now I'd like to tell you about the control of the nodulation genes.
Just a little bit of review:
Nodulation genes here are the genes
that encode enzymes that synthesize Nod Factor.
I talked about Nod Factor, and I'll show you a cartoon in a moment.
It is the substance created by the bacteria that's a
molecular signal that causes the plant to form the nodule,
and it also causes transcription in a number of other events in the plant.
So, powerful signal molecule.
The genes in the bacterium that are responsible for producing
Nod Factor include the common nod genes, nodABC.
The transcription of nodABC is controlled by a promoter
that includes what's shown here in this oblong, the so-called nod box,
a highly conserved motif that's in the DNA
upstream of the nodulation genes, both this set of common nod genes,
and others that are coordinately regulated.
All of these genes, as I mentioned before,
are located on pSymA.
Now, this cartoon introduces a couple of themes.
First, it reminds us, from my first lecture,
that the early events of nodulation include a conversation
between the plant and the bacteria.
The plant sends a signal out to the bacteria,
the bacteria send a signal back.
In the case shown on the left, we're taking a look at alfalfa,
This plant produces a molecule, coded in yellow here,
and you'll see the structure in a moment.
That signal, from the plant to the bacterium,
is able to trigger gene expression, and we'll talk about that in detail.
The result of that gene expression, as I mentioned before,
is the production of a Nod Factor called NF-Sm.
That's the Nod Factor from Sinorhizobium meliloti.
It has a backbone of N-acetylglucosamines, which is modified,
in the case of Sinorhizobium meliloti,
with a 16-carbon N-acyl substitution here,
an acetyl, and a sulfate, each shown by a color code.
Now, on the right, we have the contrasting example of a cousin,
a related legume, pea, or Pisum sativum.
Here, we've got the root of the pea plant.
It's creating a flavonoid signal whose structure is slightly different,
and I've shown that just by the color coding.
That signal triggers Rhizobium leguminosarum,
which is the symbiont for pea, to transcribe its nodulation genes,
and those encode enzymes that make a Nod Factor,
but it's a slightly different Nod Factor.
The primary difference, which I'm showing here,
which is not the only difference, but for the sake of simplicity...
A primary difference is the nature of the N-acyl substitution,
it's eight carbons instead of 16, and it has four unsaturations.
So what we see is some specificity.
A signal that goes out from the plant from alfalfa here,
or a slightly different one from pea,
and the corresponding bacteria which respond to this signal
make a slightly different Nod Factor, as shown here.
So, this conversation, plant to bacteria and back again,
already has some of the hallmarks of the host specificity.
We're going to talk only about the first part of that conversation.
Here I'm showing again the host plant producing a flavonoid signal,
and that is causing transcription in the bacteria.
Now, I'd like to focus in on the interaction of the flavonoid
with a protein encoded by this gene, NodD.
NodD is an activator protein, a part of the LysR family.
Now, what are all the elements we're going to have to look at?
Of course, at a promoter, there's going to be an RNA polymerase,
there's going to be a DNA motif to interact with,
in this case there's an activator, and we wonder,
how do these plant signals interact with all of these
components in order to achieve transcription?
And how does this occur in a specific way?
So, I'll take a look now first, and most specifically,
at the protein NodD.
I'll put that here in the context of all of those other components.
The red here is showing DNA, all right, so here's the DNA,
and the object of this regulation is going to be to
achieve transcription of the nodulation genes,
so that they can then make the proteins that synthesize Nod Factor.
At the DNA, we have a promoter,
and that's upstream of the nod boxes.
We know that NodD binds to the promoter and bends the DNA.
I've shown here a model of NodD as a tetramer;
we don't know exactly what its form is,
but we suspect that it's most likely to be a tetramer,
and more work in the future, we hope,
will elucidate that more precisely.
We've shown through genetics that the ability of NodD
to cause gene expression in bacteria in response
to a plant signal requires the chaperonins GroES and GroEL.
Also, we found through genetics that this entire system
requires the "stringent response," that being a response,
classically to starvation, where a protein, here RelA,
causes the synthesis of guanosine tetra- or pentaphosphate,
shown here as this star.
All of that has to interact with RNA polymerase,
the machine that's going to interact with DNA
and cause template-driven transcription of an mRNA
off of the DNA template.
So, these are some of the things that we know,
but there's a lot of unknowns.
Let's ask, for example, how the inducer works with the NodD.
Now I'm not going to show you the data on our GroEL work,
but I will just mention that we have a combination of
genetic data and in vitro biochemical studies
that show how GroEL is required for NodD and luteolin to be active.
In one of the experiments that follows,
I'll be referring to that some more.
Let's now just ask about flavonoids,
including the particular flavonoid luteolin, and NodD.
So in this experiment, we're doing the following:
We're growing Sinorhizobium either without an inducer
or with an inducer, so it either has just culture medium
or it has the plant signal in it.
From those cultured cells, we purify NodD,
and we ask, using electrophoretic mobility shift assay,
does that NodD bind to the nod box promoter or not?
So in pairs here, you're going to be seeing,
in the open boxes, how much DNA binding is accomplished
by this amount of NodD onto the nod box promoter
(the nodulation genes' upstream region).
And this is the amount of binding when the NodD in question
came from cells grown without inducer.
But you can see here that, if the NodD had been purified from
cells grown with inducer, the same amount of NodD
does much more DNA binding.
So it appears that the affinity of NodD for its DNA target
is increased if the NodD was purified from a
cell exposed to the plant inducer luteolin.
Now, this is an in vitro study.
In this case, all of the NodD is being purified from
cells grown with no inducer.
That purified NodD is then incubated with buffer only
or with the plant inducer luteolin.
So we're going to have "minus" for no luteolin and
"plus" for added flavonoid luteolin.
We found early on in data not shown here,
that if this were done with NodD only, there was no result.
However, if we carried this experiment out in a system where the
NodD was also presented with GroEL/GroES,
magnesium, ATP, and inducer, there was a profound effect.
And that's shown here.
The primary data are as follows:
In each case, we've got purified NodD that was incubated
just with magnesium, ATP, and GroES/GroEL here,
or with those same components and luteolin;
just chaperonins and ATP here, or with luteolin;
components only, or with luteolin, and so forth.
And what you can see is that the presence of the luteolin,
the flavonoid inducer, is able to enhance the ability of purified NodD
to bind to its target DNA sequence in the nodulation promoter.
So, what we've got here, then, is a demonstration that,
in the presence of chaperonins
and in the presence of the flavonoid inducer,
again we see enhanced ability of NodD to bind to DNA.
And that's quantified here, if you look at the fraction bound:
absence of luteolin in the incubation mixture,
presence of luteolin.
Now, this was exciting to us because we were starting
to get at the question of how the flavonoids
affect the bacteria's ability to express genes.
But recall, I told you that one of the hallmarks of this system
is that it is specific.
We know that different plants make different flavonoids.
Some of them work in vivo, and some don't.
Is that due to DNA?
Knowing that we should expect any in vitro regulatory system
to explain specificity as well as other aspects of NodD activity,
we carried out an in vivo experiment based on some earlier data
from some years ago, that showed a correlation
between what species a NodD protein came from,
and what kinds of flavonoids a cell would respond to.
So in this experiment, genetically,
we've got a responder cell that's going to report for us
whether nodulation genes are transcribed or not.
It's got a nod gene promoter fused to beta-galatosidase.
But that genetic construct has its own nodD taken away.
There's a deletion for nodD, but there's a nod-lac fusion.
Into that strain, we're going to place one or the other
of these two plasmids.
These plasmids have the same vector, so the same backbone.
They also have a promoter that's exactly the same;
it's a constitutive promoter that's highly expressed.
What's different in them is that, in this strain,
we've got a gene that represents the trimmed open reading frame
for NodD from Sinorhizobium meliloti, which is the symbiont of alfalfa.
Now in this other plasmid, we have the trimmed open reading frame
of NodD from Rhizobium leguminosarum trifolii,
which is a symbiont of clover.
So everything is the same except the actual amino acid sequence
of the two NodDs.
Now we're going to place those into a reporter strain
that has no NodD of its own,
and ask whether nod genes are expressed or not.
So in the left-hand column here, what you're seeing
is whether or not nod genes are expressed in response to luteolin.
And what you can see is that when the cell includes
this plasmid that has the Sinorhizobium meliloti nodD,
then yes, there is nod gene expression.
However, if the strain has the nodD gene that came from Rhizobium trifolii,
then it's kind of iffy, there's not very much expression.
So this suggests that this NodD can induce
in response to this molecule, but this NodD cannot induce very well.
But now let's ask:
How do those reporter strains respond to a completely
different flavonoid, shown here?
You can see for example it has only one hydroxyl,
whereas luteolin has four.
Well, now, if we take the strain containing the Sinorhizobium meliloti
NodD, we see that there's no expression at all.
Whereas the NodD from the Rhizobium trifolii shows good activity.
So this shows that the amino acid sequence of just NodD
is sufficient to account for a different response
to one flavonoid structure versus another flavonoid structure.
We wanted to put that together with the data that I just showed you,
that demonstrated that NodD can bind better to its
nod box target if it's got luteolin present.
But what we found was something of a surprise.
Now what I'm showing here on the top are the
structures of the flavonoid that I've introduced before,
that's luteolin, and a set of flavonoids here
that in vivo do not cause gene expression in Sinorhizobium meliloti.
So we would infer from that that the NodD of Sinorhizobium meliloti
cannot interact productively with any of these,
because they don't work in vivo.
However, if we carried out experiments similar to
those I showed you before with each of these,
we got the following:
Here is a set of data, and here's another.
This for daidzein, that's the molecule up here.
This one for eriodictyol, which is the molecule there.
And what you can see here is that,
if you compare NodD protein incubated with components such as
GroEL, magnesium, ATP,
or those same components plus the flavonoid in question,
these flavonoids, even though they don't work in vivo,
they still enhance the binding of NodD to its target promoter.
That's true for daidzein, that's true for eriodictyol.
And what this shows is that DNA binding is caused
even from flavonoids that are not active.
And some of the examples of the data
and a quantification are shown here.
This shows a gel shift of NodD from Sinorhizobium meliloti
presented with buffer only,
with daidzein, luteolin, and naringenin, three inducers,
of which only the luteolin is really active in vivo.
Nonetheless, all of them cause gel shifting,
and that's quantified here.
What you can see in this graph is NodD with no inducer
of any kind added, and NodD with the three different inducers,
some of which are active in vivo and some of which aren't.
So right now, our working model is that NodD
is able to bind DNA with an increased affinity
no matter what flavonoid is there.
And in fact, the flavonoids that do not work in vivo
may actually bind to NodD and inhibit the ability
of the correct flavonoid to induce,
by causing unproductive DNA binding.
So DNA binding does not explain the specificity of NodD
for the flavonoid inducers that come from different plants.
That leaves us with a series of new experiments
to do in the future, and among them,
what we want to ask is:
Does the presence of the correct flavonoid
somehow affect the interaction of NodD with RNA polymerase?
And that's led us in turn to ask about the components
of RNA polymerase, and in the end,
we hope also to work in vitro with the guanosine tetraphosphate
and ask about all of these components together.
But in this next section, I'll just like to talk about RNA polymerase.
In particular, I want to talk about the sigma factors.
Now just as a review, RNA polymerase in prokaryotes has
a core made of two α subunits (two copies of the α subunit),
a β and β' subunit, and in addition, a sigma factor,
which has multiple domains, and they're shown here in this cartoon,
which is derived from E. coli sigma factor.
And what we now know from work in many labs over the years,
is that there are many kinds of sigma factors.
Originally, the one that was best characterized
was the housekeeping, or rpoD, σ70 type of sigma factor.
And that's the form of sigma factor that reconstitutes
with RNA polymerase for the transcription of most genes
in E. coli and in other organisms.
The sigma factor here has various alpha-helices
and other subdomains; that's what these letters are for.
But this is one polypeptide as it interacts with different parts
of the β', β, and α subunits.
But in addition to the rpoD,
there are other sigma factors, a stress sigma factor rpoS,
which is very prominent in survival responses of E. coli;
heat shock rpoH;
in some enterics, there's also rpoN,
which is important for nitrogen metabolism and for other functions
such as nitrogen fixation in Klebsiella;
and a number of other sigma factors,
and I'm going to come back in a few minutes to one in particular, the rpoE.
In Sinorhizobium meliloti, our annotation showed
that there's at least double that number of sigma factors.
One of the striking differences with enterics
is that there's no copy of rpoS.
However, there is a housekeeping sigma,
it looks very comparable to the σ70 of E. coli.
There are two copies of the rpoH or so-called heat-shock sigma.
There are ten copies at least of rpoE,
in addition to the fecI originally identified for relationship to iron control,
and the rpoN that's important for nitrogen fixation.
In the following section, I'd like to tell you some of our
new work on rpoH and refer briefly to
some emerging work on rpoE.
We wanted to know what do those two different copies of rpoH do?
Are they important for heat shock?
Are they important for symbiosis?
Let's start with a look at how these are expressed.
On the left-hand column, you can see how the two different copies,
rpoH1 and rpoH2, are expressed in free-living cells.
The rpoH1 tends to be expressed mostly
in late log (late exponential) and stationary,
but rpoH2 is expressed mostly in stationary cells.
If you grind up nodules and ask whether
these rpoH genes are transcribed,
you can see that rpoH1 is transcribed, yes, it's present in nodules;
but rpoH2 was not detectable in our assays.
That leads to surprise in a minute, as I'll tell you.
Knocking out one or the other or both of these rpoH genes
leads to some phenotypes in the free-living state.
The rpoH1 mutant is very sensitive to high temperature.
rpoH2 does not appear to have much of a phenotype,
but the double mutant, similar to the rpoH1,
is affected by high temperature.
Here's the surprise about the symbiosis.
What we found was that a mutation in rpoH1 was Fix-.
rpoH2, if it's knocked out only by itself,
does not appear to have any symbiotic defect.
However, if you knock out both rpoH1 and rpoH2,
and despite the fact that rpoH2 doesn't seem
to be detectable as a transcript in nodules,
remarkably, that's Nod-.
That means that one or the other of those rpoH sigma factors
has to be functioning in order for the symbiosis
to proceed past some of the very earliest stages.
So that's a mystery that we are beginning to look at in more detail.
Meanwhile, we've also asked about the spectrum of genes
that are expressed.
In the following experiment, we used the Affymetrix chip
that I described earlier.
And what we're going to be doing here is taking a look
at the transcripts in wild-type cells, and in each of the two single mutants,
and the rpoH1H2 double mutant, as we grow them in culture,
give them a heat shock, and then harvest the
RNA to assess global transcription.
Now, in the following, I'll show you summaries of our data.
Remarkably, we found that, when you heat shock wild-type rhizobia,
almost a third of the genes in the genome, over 2000 genes,
are affected one way or another.
One thousand of them go up, another thousand go down.
Another intriguing feature of this transcription picture is shown here.
Now, if you take the entire genome of Sinorhizobium and you look
at the distribution of open reading frames,
you can see that more than a half of the open reading frames
are in the chromosome, and that's shown in this here,
so that's the chromosome.
pSymB accounts for this many, as a fraction of the whole,
and this is the number that are in pSymA.
That's the total genome annotation,
but if you look at those 2000 genes that are changed
one way or the other by heat shock,
what you see is that the chromosome is underrepresented,
and the two Sym plasmids are overrepresented.
So that, in and of itself, was kind of intriguing,
especially when you put it together with the fact that the
double mutant of those sigma factors is very defective in symbiosis.
Here was another surprise:
If you take a look in this picture only at the genes
that are increased in heat-shocked cells, and that's, as I said,
over a thousand, what we find is that the large majority
of those are increased by heat shock
even if the two rpoH sigma factors are both knocked out.
And that's shown here.
These are the increased genes by heat shock in a wild-type,
and this blue circle shows those genes
that do not get enhanced in a mutant.
These are the genes that do not get enhanced
by heat shock in an rpoH1H2 double mutant.
And the overlap is only about 200.
So, only a minority of the heat-shock genes are apparently
controlled by the rpoH sigma factors.
So, they're doing other things,
and some other regulatory system is taking the job that,
in E. coli, would be done just by rpoH.
Another surprise that needs to be followed up.
So I'd now like to ask the question:
How about the other sigma factors?
As we move forward looking at rpoH,
we know that there are more sigma factors to
account for in Sinorhizobium meliloti.
Now one that's important for sure, that I won't say more about,
is obviously the rpoN σ54 sigma factor known to be
required for the transcription of nitrogenase and a
number of the other genes in the late differentiated state.
The rpoE family presents some other very intriguing puzzles.
Now, rpoE was identified in E. coli as a sigma factor
responsible for genes that were triggered by denaturation
of proteins in the periplasm.
They're controlled post-transcriptionally by an anti-sigma factor,
which captures them in an inactive form,
unless the anti-sigma factor is somehow triggered
to release the rpoE sigma,
which can then go and associate with core polymerase
and turn on a large number of genes.
So we began to look genetically at these rpoE genes.
We found that the single mutants in each of the rpoE
open reading frames don't have any particular phenotype,
and that's shown here. Right?
Single rpoEs have some slight membrane sensitivity, but basically,
they seem normal in culture and they're normal for symbiosis.
We've now also gone through and made all of the double mutants
for each of those sigma factors.
What we find there is that some of these rpoE sigma factor
double mutants have a symbiotic defect.
They make nodules, but nodules do not fix nitrogen.
That suggests that somewhere in the transition between the
beginning of nodule formation and the differentiation of the bacteria,
essential genes require the activity of some of the rpoE sigma factors.
It also points out the likelihood that these are at least partially redundant;
that's why none of the single mutants had a phenotype,
but some of the doubles do.
So, in our future work, we are now asking,
which of the genes are controlled by each of these rpoEs,
and can we use transcriptional analysis of the global picture
of rpoE control to understand the redundancy
and understand where they're required in symbiosis.
So, from that view of the RNA polymerase,
I'd like to go to a few more case histories.
But we need to keep in mind that whatever stages we're looking at,
we may need to consider the global picture of RNA
polymerase and of other regulatory contexts
to understand how specific regulatory circuits
are able to function in symbiosis.
Now, my next topic is about the control of
EPS, extracellular polysaccharides, are required for symbiosis.
And I'll tell you in this section about how these are
regulated by a family, that includes a periplasmic regulator
and a two-component regulatory system.
Now, the EPS in Sinorhizobium meliloti has been studied
in detail by Graham Walker and his colleagues at MIT,
over a number of years.
The name of this exopolysaccharide is succinoglycan
because it is modified by acidic groups,
including succinyl groups.
And they have shown that mutants which are not able to
synthesize succinoglycan are able to provoke nodules
to form on alfalfa as a host, but they cannot invade.
So what's shown here, which is a root hair with bacteria invading,
and these bacteria are all labeled with green fluorescent protein,
allowing you to follow their invasion into the plant.
This process will not occur unless you have the
Now here's some mutants that were found by Graham Walker
and his colleagues that were able to make polysaccharide,
but they don't make the right amount.
In fact, they make too much.
So what you're seeing here are bacteria that are streaked
on a solid medium including a dye called calcofluor.
Calcofluor in solution is not fluorescent,
but when it binds to certain beta-glucans, it becomes fluorescent.
It's used sometimes as a laundry brightener because,
among other glucans, calcofluor can bind to cellulose,
which is the constituent of cotton.
And when it binds and becomes fluorescent,
that means that any cotton fabric that has calcofluor
will fluoresce slightly in sunlight, because there's a little bit of UV in sunlight.
Picks up the UV, fluoresces bright blue,
that makes the laundry look whiter than white.
Now in the case of Sinorhizobium,
it's lucky for us that the succinoglycan also binds calcofluor.
So a wild type, which you can see here, has a light,
white glow when it's grown on calcofluor medium
and exposed to UV light.
But two mutants, here exoR mutant that's defective,
and here an exoS mutant that's constitutively active
(that is, it's hypermorphic), each of these
makes too much polysaccharide.
And actually, it turns out that's not good for symbiosis either.
If a bacterium makes too much polysaccharide,
it doesn't do very well at invasion.
So what's going on here?
Why does this mutant
make too much polysaccharide when it's overexpressed,
and this mutant makes too much polysaccharide
when it's defective?
Well, we would guess that that's because this is a negative regulator
and this is a positive regulator.
But we took these and found, fortuitously,
some unexpected phenotypes that went beyond polysaccharide.
Here's the first of those:
And that is that mutants defective in exoR
or overproductive of/overactive in exoS are not motile.
They are not able to swarm on a motility plate, and that's shown here.
This is wild type; you can see that on this particular agar,
it's able to swim around, it makes a big, wide colony.
But the exoR- mutant or the exoS overexpression mutant,
both of these are not able to swarm.
That's because they're not motile.
This plate also lets me introduce to you two genetic constructs
that give us clues.
In the first one, over here, we've taken this exoS mutant,
and we've taken a wild-type ExoR (that's what we
infer to be an inhibitory factor)...
we've taken ExoR and we're overexpressing it.
If we overexpress that ExoR in the presence
of this overexpressed ExoS,
it apparently allows the cells to be normal again.
So that suggests to us that ExoR actually inhibits ExoS,
and even if you have a hyperactive ExoS,
if you then add more and more ExoR,
you're able to get things back to the wild-type balance.
Now, the second clue was also very interesting.
Here's exoR, and in a population of exoR mutants,
which are, as you can see, nonmotile,
a suppressor arose.
That suppressor allows the exoR to be motile again.
So when we tracked down the gene and sequenced it,
we found that that suppressor was actually in a gene called chvI.
ChvI is a DNA-binding protein that's linked to ExoS,
which is a response regulator.
ExoS and ChvI are a two-component system.
And without even asking about ExoS,
what we did was to mutate ExoR and find a suppressor
that was in the substrate of ExoS.
Now, motility appeared to show us a linkage of the exoS
and the exoR with the chvI,
but how about other phenotypes?
Well, one of the other dramatic changes that happens
in an exoR- or exoS constitutive bacterium
is that genes which are normally upregulated in a wild type
are very low in expression in those mutants.
And you can see that here, if you take a look here,
these blue panels.
These are actually two duplicates of an exoR-
and an exoS constitutive, showing a whole set of genes
that are not appropriately regulated if exoS
and exoR are in those defective conditions.
The two suppressors that I showed you,
however, the ones that reverse motility?
They also reverse the transcription defects
and restore them to wild-type levels,
and that's shown in these panels on the right,
where you can see that coordinate restoration
of a wild-type function from the same suppressors.
So, what we find is that the motility defect
and in the transcription profiles, exoS and exoR and chvI
all appear to be targeting coordinated functions.
We believe, based on some biochemical data,
that the system works as follows:
We've shown biochemically that ExoR is in the periplasm
in between the cytoplasmic membrane and the outer membrane.
So as a negative regulator that's present in the periplasm,
it's unique at this point.
There's no model for how this might be working,
because there's no other proteins of that type yet reported.
ExoS and ChvI are the two-component system,
with ExoS as a histidine kinase and ChvI is the
response regulator that binds to DNA.
And through ChvI, all of these functions,
including motility, transcription, and we've also shown biofilm formation
and other functions... they're all controlled.
Now, this presents us with a lot of mysteries.
For example, his-kinases (sensor kinases) are presumably
responding to some kind of change in the environment,
some signal. What is it?
Is the plant sending something?
Are other bacteria sending something?
What are the factors in the
environment that control the activity of ExoS?
ExoR as well, how does that act?
Does that also respond to the environment?
So these are some of the mysteries that we hope to
work on at the molecular level,
and one of the other directions we're going right now
is to look very specifically for the targets of the ChvI DNA-binding protein.
Now, ExoR and ExoS, as I showed you, control a lot of genes.
There are more genes out there.
Now we already know that this regulon is required for exopolysaccharide,
it has to be just right in order to get invasion,
you saw that there are many genes that are controlled.
But we also think that, as you go from invasion
all the way to differentiation,
there are many more genes than that.
First, in a promoter trap screen, genetically,
we found several dozen bacterial promoters
that appear to turn on in between the beginning of invasion
and the final differentiation.
So, we're looking at those genes to ask,
what is it that controls them?
Furthermore, we've recently carried out more genetic screens
of the type that I described in the beginning --
putting random bacterial mutants on plants and asking whether
those bacteria are able to cause nodules to form
and whether those nodules are functional --
to find more nodulation mutants and fixation mutants.
We have already found several new genes
not previously recovered in earlier screens,
for transporters for example,
which suggests that the bacteria must be able to import or perhaps
export materials in order to accomplish nitrogen fixation
and differentiation. And in addition,
we've taken the transcriptional approach as follows:
So based on our Affymetrix transcription studies,
we were able to show that hundreds of bacterial genes change,
if you compare cultured bacteria to bacteria in nodules.
Over 300 genes increase in expression, including the obvious suspects,
such as nitrogenase and fixation genes.
But many more than that, 900-some genes, go down.
These decreased genes include biosynthetic functions
such as amino acid synthesis and others.
So the bacteria is having a major change in its gene expression profile.
The plant changes as well.
Now, several hundred genes change,
of which 250 or so increase in the expression,
and many plant genes decrease in their expression as well.
Now what we wanted to do was ask something about
the strategy of how the plants and the bacteria
control their transcription in these nodules.
So what we're going to do in the next set of experiments
is compare the transcripts in a wild-type nodule, such as this one here,
to a nodule that's being established on a wild-type plant
by a bacterium that's mutated.
Now what mutant are we going to use?
We're going to use a mutant called fixJ.
Now the protein FixJ is required for bacteria in
the nodule to transcribe the nitrogenase, or nif genes.
And that works as follows:
Leghemoglobin, which is present in the plant cytoplasm,
shown in this cartoon, buffers oxygen
so that the free oxygen tension is very low
(although oxygen is delivered at a high rate by leghemoglobin).
Now, the low oxygen is able to interact with
the two-component regulatory system, FixL and FixJ,
that activates a signal transduction cascade
that results in nitrogen fixation.
So, here's the plant cytoplasm
and the bacterium responding to the plant's signal.
If you have no FixJ in the bacterium, then there's no nitrogen fixation.
So the bacterium provokes a nodule, it gets into the plant,
but then it doesn't work.
So the RNA transcription studies we're doing now
are going to compare a nodule made with wild-type bacteria
and a nodule that's made and infected
with bacteria that then don't fix nitrogen.
And here are results.
Let's take a look at the wild type first.
As I mentioned... and by the way,
we're looking only at the transcripts that are going to go up.
The plant has about 250 or so transcripts that are upregulated in a nodule.
The bacterium has somewhat over 300 transcripts that increase.
This is the wild type.
Now, if we ask,
what happens if we look at the transcripts
made in a nodule that was created by the mutant?
We see that a nodule made by the fixJ mutant
upregulates very few transcripts.
The vast majority of these 300-some transcripts that are
increased in bacteria when they're wild-type
apparently requires the FixJ system.
So I would say from this that the bacterial transcription profile
is responding to some signals right in time,
at the point that nitrogen fixation is occurring.
We found a contrasting result when we ask these same nodules
that have so few bacterial transcripts,
how many plant transcripts do they have?
And that's shown here.
What you can see is that almost all of the wild-type transcripts
shown in this outer box are also made in the nodule
that resulted from the bacteria that is mutated for nitrogen fixation.
So I would say that suggests a couple of things.
First, that the plant isn't making its decision about
whether to transcribe these genes at the last minute.
It made that decision to upregulate these genes early on,
before the bacterium makes the decision about
whether or not to fix nitrogen, right?
So most of these genes, I would guess,
are being made to build the nodule in the first place,
that the strategy of the plant is to decide on the morphogenesis.
But it also shows that, yes, there are some genes,
and they're shown in this outer part that's not
covered by the overlap,
there are some genes that do depend on bacterial nitrogen fixation,
in order for the plant to express them.
So we want to find out in future studies
how those are controlled, as well as how some
of these are controlled irrespective of the bacterial signaling.
And that brings me to the final topic,
in which we've been able to use a combination of
bacterial genetics and plant genetics,
as a way of elucidating new signals.
In a screen of plant mutants, we asked,
can we find plant mutants that fail to accomplish nitrogen fixation,
fail to accomplish correct symbiosis?
Just as a baseline, let's take a look here at wild type.
Now, in a wild-type plant with wild-type bacteria,
you get nice nodules, here's those nodules.
If you do a section through a nodule,
you can see all of the infected cells.
So here in the corner is the vasculature of the plant,
so the root is going like this.
Here's the nodule coming out and all of these dark-stained cells
are packed full of successful bacteria fixing nitrogen.
Using a bacterial strain that has a glucuronidase fusion
in the nif genes, but is still wild-type,
we're able to see that, if you take this kind of section of a nodule
and you stain it for glucuronidase activity,
you get this deep-blue stain, and that means that the nitrogen fixation
genes are being transcribed.
But now let's take a look at a plant mutant.
So this plant mutant is called dnf,
that stands for "defective in nitrogen fixation."
If you see here, the bacterial action (wild-type bacteria)
on this plant yield tiny, little, white nodules.
They're not functioning, they're not fixing nitrogen.
A look inside those nodules shows that the bacteria do get into cells.
Here's the vasculature again, here's the root,
here's the nodule.
All these dark-staining cells are packed full of bacteria.
However, if you put in the indicator for nif to glucuronidase
and you stain for glucuronidase activity,
you can see here that the nif genes (nitrogenase genes)
are not being expressed. There's no blue stain.
So something about this plant mutant
is not allowing nitrogen fixation to occur.
So we can say that the bacteria need whatever it is that DNF1 does
in order to differentiate and fix nitrogen.
Having carried out the crosses and mapping necessary,
we have cloned and identified DNF1.
We found that DNF1 encodes a plant protein
that is a nodule-specific signal peptidase.
Now, using this illustration from Molecular Biology of the Cell,
I can just go through roughly what we would
expect the signal peptidase to do doing.
In the plant cytoplasm, you're going to have the endoplasmic reticulum.
Some plant proteins are going to be encoded in the nucleus,
they're going to be translated such that they go
into the lumen of the endoplasmic reticulum,
and that is made possible by a signal peptide.
The signal peptidase cleaves off the signal peptide as so,
releasing the mature polypeptide product.
Now, what do those do?
I'll get to that in a moment.
But, what we just conclude is that there's going to be something
about the action of DNF1 that's going to result
in the production of a mature protein that might go to the cell wall,
or it might go into vesicles.
So whatever is being produced by the DNF1 signal peptidase
may be purposed for secretion or for vesicle delivery.
So what are the substrates?
And here, our work was able to intersect
with the work of Eva Kondorosi and Peter Mergaert
and their colleagues.
They had discovered a family of nodule-specific plant proteins
that they called NCR, "nodule-specific cysteine-rich proteins."
Using our mutant, they have been able to show that these NCR proteins
are actually processed by our DNF1 signal peptidase.
Therefore, what we can envision is that the lumen
of the endoplasmic reticulum, as a donor compartment,
is going to be full of these NCR peptides after they're processed.
And these may be then delivered to a target compartment;
we might envision that as having the Rhizobium.
So, once again, what can consider here
is that we've got NCR proteins from the plant
being processed into the endoplasmic reticulum, put into vesicles,
and then delivered into a compartment where
they actually go into the Rhizobium and
cause those Rhizobium to differentiate.
This model can be tested in a number of ways,
and those are underway at present.
What we can envision though, is the following:
We know that these bacterial cells,
which are here shown in a closeup of an infected cell in a nodule,
these bacterial cells must be getting a lot of signals from the plant.
So if we envision this as a cartoon of a plant cell,
and here's the cytoplasm of the plant cell,
we already know from previous work that the leghemoglobin
that I mentioned earlier, that gives the nodule its pink color,
helps to buffer oxygen.
It keeps free oxygen low while delivering it at a fast rate,
and we know from earlier work that there's a signal transduction
cascade that triggers the activity of NifA and
eventual transcription of nitrogenase, due to the FixL/FixJ
oxygen-sensitive two-component system.
But now what we know is that something else is needed.
Unless you've got DNF1 processing the NCRs
and maybe doing other jobs as well,
unless you've got DNF1, you still won't get nitrogen fixation.
So, a completely new kind of process.
Deep inside the nodule, there is a new signaling process
going that we've now discovered.
And this raises the possibility that specific
DNF1-processed peptides may be triggering differentiation.
As Mergaert and Kondorosi have shown,
these peptides are also able to carry out control
of bacterial cell division and DNA endoreduplication.
So there are many hundreds of NCRs,
and tracking down what each of them does
is a big challenge for the future.
From our standpoint, as we look at bacterial differentiation,
we want to know, are these effects direct or indirect?
Does an NCR peptide actually go into the bacterium
and interact with a receptor, or is it causing some condition
in the plant that then causes gene transcription
to change in the bacterium?
So, our purpose is right now to characterize the mechanism
in the plant of how these DNF1 signal peptidase components
are working to ask whether there are other substrate
proteins as well, and to find out their mechanism in symbiosis.
So what I've shown you today is an outline of how
we're taking a look at bacterial genes from the early
nodulation processes all the way through the differentiation
of the bacteria that makes it possible for them to fix nitrogen.
And I think in particular that the use of plant mutants,
as we did with DNF1, to look for plant factors
that control bacteria may be an especially promising
direction for the future.
They're going to be many levels of regulation,
from external signals to sigma factors.
I think that the combination of genetics and in vitro transcription
studies are important.
And a lot of our effort in the future is going to be directed towards
this double study of plant mutants and bacterial genes,
in order to try to dissect out the signals and controls.
So with that, I'll conclude this part of the lecture,
and my third lecture is going to be on how the plant responds
to the bacterial signals at the various stages of nodulation.