Sharon Long (Stanford) Part 3: Plant genes and cell response in nitrogen-fixing symbiosis

Hi, I’m Sharon Long, from Stanford University, and I’m here today to talk about our recent on the plant side of the nitrogen-fixing symbiosis, specifically about genes and cell responses By way of review, I’ll mention again that the Rhizobium symbiosis happens in organs called root nodules And within these nodules, bacteria are able to fix nitrogen and provide that to the plant for nutrition This is complex developmental process, it goes through a number of stages, and you’ll see some of those today It is species specific, which I’ve discussed in some of my earlier talks And importantly, for today’s lecture, bacterium and plant each respond to signals from the other An outline of my talk today is shown here At the beginning, I’ll give you some introduction, reviewing some of the dynamics of how bacteria and plants interact Then I’d like to talk about some recent work in our lab, first about infection and about of group of proteins called flotillins, and also about a Nod Factor receptor from the plant Then I’d like to talk about two plant mutants, dnf1 and dnf2, which we’ve identified as being defective in late stages of nodulation And by describing the genes that we’ve now cloned from those mutants, I can share with you some of our ideas on how the symbiosis is working So, here I’m showing a review of the basic idea of signal exchange in the early phases of nodulation The plant secretes a flavonoid signal, this flavonoid acts as a trigger in the bacterium, and that causes the transcription of genes called nod genes, which encode enzymes. Those enzymes are able to synthesize this second signal, the so-called Nod Factor, down here, sometimes abbreviated as NF in some of the slides to come This is a modified chitin fragment, you can see there’s four residues of N-acetylglucosamine These are modified at the reducing end with a sulfate, and at the non-reducing end with both an acetyl and an N-acyl group And in one of the slides yet to come, I’ll be talking about the differences between the plant response to chitin to a true Nod Factor So Nod Factors are very powerful, and this gives you an idea of just how powerful they are On the top, we have an alfalfa root, and on this root, Sinorhizobium meliloti, the symbiont, has established a nodule, and this is an intact nodule It’s been cleared, and you can see the deep brown color, which is from the metals that are in the various cytochromes and enzymes of the bacteria while they’re fixing nitrogen Now at the bottom, we see a form that looks very similar, and yet this has no bacteria in it This entire structure on an alfalfa root was caused just by a small droplet of Nod Factor So this one chemical is able to produce an entire organ on the plant Here is another one of the characteristic early events in Rhizobium-legume interactions This shows root hairs on the root of a plant, and here on the right, you can see a normal root hair, which grows straight And here you see a root hair that has been provoked by Rhizobium to grow in a such a way that it tops over and forms a curl. And here in the crook is where the bacteria are trapped, and these deformed root hairs of various kinds are very characteristic of the effect of the correct Rhizobium on its compatible host Now, the bacteria are then able to travel by proliferating and burrowing their way into the plant cell itself And this white arrow shows an infection thread; this is formed by the plant in response to the bacteria Within the infection thread, bacteria are proliferating, and they are invading through the root hair cell, and as you’ll see, eventually down into further cell layers So there’s the infection thread, and you’ll see more about that in a little while I’d now like to show you something about the dynamics of how Rhizobium affect plants Let’s start by looking at normal root hairs So, on a root, the youngest part of the root is over here, and you can see that they don’t have any root hairs yet, and then the root hairs get longer And over here, they’re full sized So, root hairs get to a particular size, and then they stop Now let’s watch in this movie, to see how root hairs elongate normally, that’s without Rhizobium So here you see root hairs They start, they grow, and then at some point they reach a mature length, and they just stop, but they grow fairly straight during this entire process

Now we’re going to look at root hairs on a plant that has been treated with its Rhizobium symbiont So here, the root hairs are in the process of growing, and if you watch what happens to them as they grow, you’ll see how different it is from unperturbed growth So here, for example, we see root hairs that are curling Over on this side, if you look at this root hair, you can see that it’s growing, and then it pauses, see that pause? And then it emerges, and now it pauses and branches again So there’s two things going on One is this characteristic pause, so right here you can see there’s a pause right before it starts to top over And the other is that you’re getting branching or curling, other kinds of deformation So Rhizobium has a profound effect on the morphogenesis of this particular plant cell Now, we’ve seen a little bit now about root hair deformation, and I’ll mention along the way that Rhizobium can also cause roots to engage in specific transcription, which we study with microarrays, and they provoke cell division But for the moment, I just want to look a little bit more about what happens to these root hairs, because if you think about it, root hairs are on the surface of the plant, they’re single cells, they have a high surface-to-volume ratio it makes sense that these are going to be exposed to the signals from the outside, and they might be the place where signal transduction begins So, we might ask, by cell biology and by genetics, are there receptors, and what is the signal transduction pathway? And ask whether we can study root hairs as a way of capturing those Now, signal transduction is often a complex and fairly rapid multiprotein process, so we set out to look for events that occur in root hairs that are cell autonomous and fairly rapid One of the earliest studies that we did in this area was simply to look at the electrochemical potential across the plant membrane So in this, we’re taking a single root hair here, and by putting in a microelectrode and measuring the difference in potential between the interior of the cell and exterior of the cell, we were able to find that untreated root hairs have a stable, very negative potential at about -130 to -140 mV However, if we treated a root hair with Rhizobium, what we found was that the Rhizobium and its Nod Factor are able to cause a depolarization of that plasma membrane, so that the potential change across the membrane here diminishes And that happens actually within a minute In our studies and others’, this has been shown to be accompanied by ion fluxes (ion currents) near the tip Now one ion in particular is very interesting to us, and that is calcium. We’ve used different techniques, but here’s one of them that we have used On the left, you can see a root hair, and into this root hair we’re delivering a mixture of two different fluorescent proteins One is fluorescent irrespective of the calcium concentration, the other increases its fluorescence when calcium is high Using those two together, we can take a ratio and get the value of calcium corrected for the concentration of the cytoplasm And then we can track that over time So now we’re going to follow calcium in this slide Now, this is a pseudo-color representation of the fluorescence corrected for cytoplasmic concentration Cool colors mean low calcium, warm colors mean high calcium If you take a look at the root hairs here, they’re all blue, that means low calcium As you move in time, and these are ten second intervals, you see the warmer color, yellow, appearing from the tip of the root hair, so calcium is getting high at the tip As you continue the time series up here, you can now get to a point where you see this extremely high calcium, shown by red, and this happens to be in the part of the cell represented by the nucleus So, what can we say, and I’ll quantify that with the following graphs First, if you compare an untreated cell, that’s here, and here’s the baseline of the calcium, with the presentation of one nanomolar of Nod Factor, what you find is that about ten minutes after the presentation of the Nod Factor, you get these sharp upswings in calcium And that’s what’s going on here This is the calcium spiking represented there Now, number two, if you put in a higher amount of Nod Factor,

10 nanomolar, now you get something more complex You get a faster response, and this turns out to be the tip flux of calcium coming in, and then calcium spiking happens at the same time Now the other data that I’ll show you here is that, if you take chitin alone Now remember I pointed out that Nod Factor is like an oligomer of chitin N-acetylglucosamine resides, with some modifications. But what if you take the modifications off? It’s not Nod Factor anymore, it’s just chitin If you add very, very large amounts of chitin, we’ve got one micromolar here, so a thousand-fold higher, you get some spiking, but you need much more chitin than Nod Factor, and it’s not completely normal So for that, we can say that the plant is exquisitely tuned to the Nod Factor Finally, we do know that the calcium flux and calcium spiking are not just separated in space, but they’re really independent in that you can add high Nod Factor after calcium spiking has started, and you’ll get the big influx So they really do appear to be two separate events Now a third response that we were able to document very early in nodulation is this one: Using a whole seedling assay and with an indicator that shows the presence of hydrogen peroxide, we can follow normal roots over time, and we can assess how much peroxide are they producing And in a normal, untreated root hair, that’s shown here They don’t produce zero, they produce a modest amount, and here’s what it looks like over a period of about 1.5 hours Now what if we treat those with Nod Factor? What we find, very intriguingly, is that Nod Factor causes the rate of evolution to be diminished So a lower level of hydrogen peroxide is produced in the presence of Nod Factor Now what if we add an “elicitor,” pathogenic elicitor? We would expect an elicitor to cause the plant to mount a defense response, and sure enough, if you look at the amount of hydrogen peroxide coming in the plant after the treatment with the elicitor, it’s accelerated So it seems that roots are able to elaborate hydrogen peroxide, an example of reactive oxygen species, that Nod Factor diminishes that, which would suggest it’s lowering its defenses, and an elicitor increases it, consistent with an increase in defenses Now, this is only over the couple of hours, we’re interested here only in very early events In fact, the defense reactions appear to have a very interesting and complex role later on in nodulation, but this is not related to those This is just in the first two hours or so So now we can fill in a bit what we know about the early responses of the plants to the rhizobial Nod Factor We know that it causes nodules to form, cell divisions in the plant We now can fill in that, in addition to the overall root hair curling, which we could see in the microscope, other kinds of assays demonstrate that there’s a rapid depolarization across the plant plasma membrane in response to Nod Factor This is accompanied by calcium flux Slightly later there’s calcium spiking in the cytoplasm and a suppression of the rate of reactive oxygen production Now one other topic that I’ll just mention briefly is transcription We’ve been able to assess transcription during nodulation, at various stages, with a specialized approach, which is shown on the next slide We, following our work on determining the complete genome sequence of the bacterium, we then constructed an Affymetrix chip in which we had the complete bacterial genome, plus about 10,000 sequences representing probable genes from EST libraries in the plant, and we put the two genomes on the same Affymetrix chip So we call this our SymbioChip, and through analysis of RNA species from the nodules, we can actually get a readout of both bacteria and the plant at the same time

And through that, we were able to show that, within the first 24 hours after treatment with Nod Factor or after treatment with bacteria, a characteristic set of plant genes are upregulated or are downregulated Some four dozen or so sequences, and I’ll talk a little bit more about that in a moment when I get to the mutant analysis But now, let me put all of those events on a timeline for you So I’m going to start here on the left Now, we know that in the very early stages, within a minute, we get calcium flux, suppression of reactive oxygen, we get a depolarization There’s also calcium spiking at about 10 minutes There are morphological changes You saw in the film that the root hairs pause and swell a little bit right before they start to curl Then there’s morphogenesis of the root hair, forming branches and curls and both And we can see the cell divisions. Finally, the transcription Now here’s the timeline for that happening If we start at time zero, and this is where the bacteria are being added, those first events, such as the calcium flux and the depolarization, happen within a few minutes, one to a few minutes Calcium spiking at an average of 10 minutes The gene expression that I mentioned, that we can assess with our Affymetrix chip, that’s happening as early as a few hours and goes on for about 24 hours Now the actual curling of the root hair, the deformation of the root hair, takes a while to express Of course we have to keep in mind that this is limited by how fast the root hair can grow We don’t know when the decision to curl is made, but the actual outcome, the mechanics of curling, take somewhat longer And then, here around 20 hours, we’re going to start to get infection The production of the infection thread and the ability of the bacteria now to penetrate in through that root hair And then the cell division here leading to the production of the nodule, that’s all happening over this period of a day to two days Now through a set of genetic mutageneses and screens, our group and others have identified many different plant mutants that arrest at one or the other of these stages And having these subcellular assays to do allowed us to distinguish between the different kinds of plant Nod- mutants. So here’s an example Here, on the left, is a gene called NFP for “Nod Factor perception.” If that gene is mutated, you have a block right here, so nothing happens at all, there’s no depolarization, there’s no calcium flux, no change in the morphology And then, here are two mutants that are present in such a way that they allow the plant to have calcium flux, and it has suppression of reactive oxygen species and so forth However, it doesn’t show calcium spiking So these two mutants were able to distinguish between the early calcium flux and the calcium spiking itself They are not only separated in the cell in geography and in time, but they can be separated genetically Here, we found one or more mutants that allow calcium spiking but do not allow any gene expression And so forth and so on down the line So with all of this, we are able to set a timeline and a developmental sequence for how signals are transduced And among other things, we were able to use that transcription assay to distinguish between root hair curling genes and early calcium transduction genes And we were able to show that all of those plant genes that are expressed come on at the same time It’s not the case that you can get a few genes here, and a few more here, and a few more after that You get no genes expressed at all, up until this point here And we were able to show that by a combination of mutant analysis and transcription analysis So that means we’re going to be able to say that the series of events of shown here — calcium flux, suppression, growth arrest (the pausing) — can cluster, and we can see that one particular gene called NFP, which we believe encodes a receptor, is responsible and required for those events Then there’s a set of two genes, DMI1 and DMI2,

that are required in order to get to calcium spiking One of these is an ion channel, the other is a putative receptor kinase Intriguingly, just downstream from calcium spiking is a mutant, DMI3, and another mutant, CYCLOPS; these are both required in order for any of the later events to occur, such as transcription So we believe that these genes help interpret the calcium spiking signal and transduce it to the next set of genes, NSP1 and NSP2 These are required for transcription, and indeed, they turn out by sequence to be predicted transcription factors More transcription factors are also required in order for nodulation to occur. Now all of these events on the right require the one pathway, this is called the signaling pathway, starting with the signaling receptor NFP1, and going through transcription factors But I haven’t said anything about infection, and I’d like to mention that It turns out that the plants appear to have not one receptor, but two The second receptor is called the “stringent” or “entry” receptor, and the name of that gene is LYK3, in the case of Medicago truncatula This, and one of the transcription factors, are required in order for infection to occur So you can have these other events happening, but no infection This additional set of factors is required for infection So back to the outline I’ve gone through a description of some of the plant cell responses to Nod Factor, and told you about how we have used plant mutants, plus those cell phenotypes, in order to create an ordered set of steps that we believe represents the signal transduction pathway in early nodulation In the next part of the talk, I’m now going to move, as I say, beyond nodulation: I’ll be talking about infection and also about later stages in nodulation So, as we think about the symbiosis, remember, everything that I described in terms of early signal transduction is over here. That’s just the beginning There’s so much more to come There’s as the bacteria infect and eventually form the nodule So now let’s take a look at some of the specific topics that are interesting. First, infection We’ve already seen that more genes are needed for infection than just for signaling And the infection thread is a remarkable structure It penetrates not just through the root hair, but then through multiple layers of plant cells, on its way in to find target cells Now, as this proceeds, the plant elaborates an entire nodule, and all the dark cells in this photo are packed full of bacteria And if you look close up at those, each bacterium is surrounded by an envelope of plant membrane, and within that, the bacteria are going to differentiate So, using different approaches, we’re going to take a look at how the plant manages to support these really remarkable events We’ll start with infection. As I mentioned, infection occurs beginning in the root hairs, and what we wanted to ask is, how does the inside of the plant cell cope with the invasion of the bacterium? How does it reorganize itself? We decided to take a look at some candidate genes called flotillins These are genes that are originally identified in animals I’ll give a brief review, and then I’ll show you our data that demonstrate that plant flotillins are active in nodulation There’s two copies of flotillins that are specifically upregulated in nodules, their transcription is regulated along with other nodulation genes, they’re required for infection, they’re required for elongation, and remarkably, the proteins of these flotillins have very striking localizations that we believe relate to the mechanism of infection. So, what are flotillins? They’ve been studied widely in animal cells This diagram shows an animal cell with plasma membrane Flotillins are shown here. There’s two of them, they’re outlined in red They have a domain that is affiliated with the membrane, and a tail that is more cytoplasmic And it is thought that they relate to signaling, to endocytosis, and to activity of the actin network within the cell

It’s also been demonstrated that these flotillins occur in what are sometimes termed “membrane microdomains,” also sometimes called lipid rafts, although the terminologies are actively being discussed these days Now plants have flotillins, too, although they have not been widely studied In Arabidopsis, the sequence shows that there are three flotillins. In our analysis, we found more than seven in the legume Medicago truncatula Of these seven, we found that two are uniquely expressed in nodules We found that the proteins are, as with animals, flotillins present in small domains, or puncta And I’ll show you that one of the two has a very special localization during infection Now, the predicted structures of the flotillins in plants looks very much like the animals’, with a head domain and a tail domain, so we might expect that it’s going to have some of the same properties So, as I mentioned, Arabidopsis has only three flotillins, but Medicago truncatula has more than seven, and here’s one of the really interesting things for us, is that there are a great number of these that are actually all linked together Now you’ll see here that there’s flotillin 3 and 1, then there’s down here flotillin 4 and flotillin 2 We believe that those are active flotillins One of the others, shown here flotillin 5, is in the same genomic regiion, but we believe it’s not an active gene Now, we took a look at the activity of the promoters for each of those flotillins, and what you can see here is that the top, flotillin 1, its promoter is active in the vasculature Here it’s expressing a GUS fusion Flotillin 3 is also active in the vasculature But if you take a look at flotillin 2 and flotillin 4, these are being expressed in nodules We can also quantify the transcript, and the same story emerges We can see here that flotillin 1 and 3 are at a low level throughout nodulation, but both flotillin 2 and flotillin 4 are strongly upregulated in the first day after roots are presented with bacteria Flotillin 4 comes down, although flotillin 2 stays fairly well expressed during at least a couple of weeks of nodulation So in terms of their expression, these appear to be especially associated with nodule development Does that relate to any of the formalities that we know right now about the genetics of early nodulation? So remember that there are a set of genes that I introduced earlier, and we have mutations in these various steps for nodule signal transduction, and that transcription occurs only after a whole set of steps have taken place. So what about flotillin 2 and 4? And as you can see in this bar graph, those two flotillins actually require the same signal transduction pathway in order to be expressed Here is a wild type on the left, we’re seeing flotillin 2 and 4 And then here are a couple of the nodulation mutants LYK3, NSP2, NIN-1, those are here NSP and NIN, here’s LYK3 for a receptor for Nod Factor These appear to be required in order for the flotillins to be upregulated We could also show that some of the downstream genes are not required Now, we know that these flotillins are controlled in the same way nodulation genes are controlled, but what about their function? Through using RNAi approaches, we can diminish the activity of the various flotillins, and here’s one particular experiment, and these results have been borne out in others’, though the absolute numbers tend to vary So if we take a look at a wild-type plant for example, we can see that it’s got about six nodules per plant If you knock out flotillin 1 and 3, there’s again about six But if you knock out flotillin 2, there’s many fewer nodules Flotillin 2 and flotillin 4, if knocked out together, is even more But flotillin 4 knocked out by itself does not appear to have much an effect on the number of nodules

If you look at the percent of plants with nodules, you don’t see a major effect In fact, if anything, you seem to get a few increases of the percentage of plants that have some kind of growth on them But now if we look at the percentage of nodules that are actually Fix+, which is indicated by the pink color from leghemoglobin, you can see that about, in this particular experiment, almost 30% of the wild-type nodules appear to be functioning, about the same for flotillin 1 and 3 knockouts But if flotillin 2 is knocked out, then there’s a striking diminution of the number of functional nodules, and a less striking but still significant change for flotillin 4 And the double mutant almost completely wipes out any effectiveness of the nodules that do form So, we would conclude from that that flotillin 2 and flotillin 4 are both required for nodulation, and that they are not redundant, because the double mutant has a more severe phenotype than either of the single mutants alone We’ve also now gone to take a look at the proteins, and what we found is the following: Let’s begin on the left with flotillin 2 This is a view of epidermal cells, so we’re looking down at the surface of the root here And you can see that flotillin 2 is present in little dots, or puncta, and there’s a very striking localization of flotillin 2 at the polar end of the cell In the root hairs, the flotillin 2 is very strikingly punctate, it’s present in these little dots, and this kind of distribution remains the case if the plants are inoculated, and that’s shown in the lower part of the left-hand side So, uninoculated and inoculated, flotillin 2 is punctate, it remains fairly stable, although there are some subtle changes in density and in polarity But we got a very striking result with flotillin 4, and I’ll show you that here. Now let’s start on the top again The epidermal cells and root hair cells, and it’s very striking that the flotillin 4 signal (a GFP fusion signal) is present in puncta. But even more striking was that, after inoculation, what starts out as an even distribution of flotillin 4 in root hairs becomes a highly polar distribution, and that’s shown with these red arrows, where you can see that at the tips of these inoculated, bacteria-treated root hairs, flotillin 4 has migrated to the tip So, that suggests that flotillin 4 has a very active participation in something having to do with what goes on at root hair tips So we began to look specifically at infection, using wild type as control, where you can see bacteria invading These blue, lac-stained bacteria show infection threads for bacteria on a wild-type plant Here’s a nice infection thread growing through a root hair on the wild type. But in a flotillin 4 “minus” mutant, things are not working very well What you’re seeing here is that there’s no significant penetration deep into the nodule Furthermore, even if you look at a root hair, the infection threads lack integrity, they’re not well structured, and we find in fact that they don’t succeed, they don’t penetrate So it appears that flotillin 4 function is important for infection threads, and we also now have evidence that shows that flotillin 4 is associated with infection threads Flotillin 2 is shown here, bright green fluorescence around the cell This particular root hair, which is curled, has been infected, and it has bacteria that fluoresce red. Those bacteria are shown here But you can see that the flotillin 2 is around the membrane of the outer cell, it’s not affected, not close by to the red fluorescence of the bacteria However, if we look at flotillin 4, that’s shown here Flotillin 4-GFP is around the outside, but it’s also here, see? Around the infection thread. Green fluorescence, and here’s the red bacteria inside the fluorescence So that suggests that flotillin 4 is affiliated with the infection thread membrane Now you’ll recall that, in the sequence of genes that are important for nodulation signal transduction,

one set is called the signaling pathway, here, but more genes are needed in order to get infection, including the entry receptor LYK3 Now that we’ve got some clue that flotillins are involved in infection, we might want to ask whether we can place flotillins in any kind of relationship to the genes that have been previously found to be important for early signaling and infection Now, in the following photos, I’m looking at the puncta density of flotillin 4 Here’s a wild type, and the puncta density is the number of bright spots per square micron, so you might have a little square micron there, and then you plot the number And I’m going to compare that wild type to three different mutants Each of these is a mutant in the putative entry receptor LYK3 lyk3-1, which is the most severe, and two other alleles, lyk3-2 and 3-4 The lyk3-1 is predicted to have a protein with a dead kinase domain, and what we saw looking at this particular index of puncta density is that, if you compare the wild-type number here to the number in the lyk3-1 mutant, the puncta density for FLOT4 is greatly diminished, significantly lower We don’t draw any mechanistic conclusion from this, but it did give us the sense that maybe LYK3 itself, together with FLOT4, would be worth looking at So we want to ask whether LYK3 and flotillin 4 interact We’re going to need to study LYK3, and we were fortunate to collaborate with Doug Cook, Brendan Riely, and their colleagues at UC Davis, and thanks to their contributions, we have the following tools for study: First and most importantly, a stable transgenic of Medicago truncatula that carries a GFP fusion to LYK3, under the control of its own promoter It’s known that this is functional because it complements a lyk3-1 mutant Because it’s under its own promoter, we have more reassurance about its correct position and localization Now into this stable transgenic, we’re now going to introduce a second marker, we’re going to create a transformant that also has flotillin 4 linked to the fluorescent mCherry protein, so we’ll have green fluorescence for LYK3, red fluorescence for FLOT4 And our particular tool is going to be a spinning-disk confocal at the Carnegie Institution, in collaboration with David Ehrhardt We were able to visualize the LYK3 putative receptor in root hairs So the results are shown here We have LYK3-GFP, it’s present in the root hair, you can see that it’s punctate And when we took a look at the LYK3 plant with infections, we found the following: Here’s a curled root hair, and you can see going down through the curled root hair is an infection thread, and the green fluorescence of that LYK3 is all the way along the infection thread membrane, so we can see that the LYK3 putative receptor is localizing the same way that the flotillin 4 appears to do along the infection thread, as it moves into the cell Now, an important control is shown here also Root hairs and other parts of plant cell walls have autofluorescence as well, and we wanted to ask whether the green fluorescence we saw here was really due to the LYK3, or was it autofluorescence The results of that are shown in this other panel Now, this is a fusion of LYK3 to a nonfluorescent protein And again, there are bacteria that are fluorescing red Well, what you can see is that it’s a good idea to do this control because the crook of the root hair is very highly autofluorescent, so if we were to see this and say it’s GFP fluorescence, we’d be wrong, because that’s just what root hair cell wall is doing You can see here the infection thread is moving down with the red fluorescence, and there is no green fluorescence around it So infection threads are not intrinsically fluorescent; it’s only that the LYK3-GFP is providing that signal So now we know that flotillin 4 and LYK3 appear to localize to the same place Can we find out anything else about them? So in this, we have started to use that double transgenic

So what’s going on here is that this plant has LYK3-GFP It also has flotillin 4-mCherry Now, these are closeups below, you can see the LYK3-GFP, flotillin 4-mCherry And these are uninfected root hairs And if you merge the two images, it’s possible to see here that the green and red are distinct So that suggests that they are not co-localizing within the distance where their fluorescence would mix and make a yellow color Now, taking a look at that more statistically, we can do a correlation plot, as shown here, where on one axis, on the x-axis, that’s the fluorescence intensity of one, on the y-axis is the fluorescence intensity of the other And in fact, there’s a distribution all the way around, it doesn’t particularly correlate If you have a high fluorescence of mCherry, that could be either low or high fluorescence of GFP But what happens when we treat root hairs with bacteria? And that’s shown here Now, again we’ve got the GFP, here’s a closeup We’ve got the mCherry for the flotillins, and a closeup But now, when we do the co-localization, you can see that there appears to be a merging of the green and red fluorescence, to present a higher number of puncta that look yellow So we would say there appears to be an increase in the correlation of their location, and here’s another way of looking at that The correlation plot now shows a correlation coefficient that’s more than 0.5 That means that more than half the time, you’re getting a correlation of the intensity of fluorescence of one and the other So that suggests to us that whether or not flotillin 4 and LKY3 are associated depends on whether the bacteria have interacted with the root hair So we pursued that a little more Another way that we have started to look at that is to examine the dynamics, not just the location, but dynamics of what these proteins are doing So in this experiment, we’re going to be looking at a root hair which has a LYK3-GFP fusion, so that’s the only fluorescence you’re going to see, it’s just the receptor LYK3 And what you’re going to be seeing in the micrographs is as if you took a plane like this, right? Through the root hair and, where you see this bright point, that’s what is going to be shown now on the films Now the left-hand we’re going to look at LYK3 in an uninfected root hair So one of the things you can see is that it’s very dynamic If you try to focus on a point and follow the LYK3, it’s moving around too much, you can’t do that But now let’s look at a root hair that has been treated with bacteria What a difference Now you’re looking at the LYK3 fluorescence, and it’s really behaving itself. It’s sort of staying place, and these two arrowheads, for example, show places that you can focus in, and you can see a LYK3 signal that’s not moving, that’s fairly stable So that suggests to us that it’s not just where LYK3 is, it’s how fast it’s moving and perhaps shuttling around, that is being changed by the presence of the bacteria, which is remarkable, and that led us to another way to look at LYK3 together with flotillin 4 In the following, we won’t be looking at movies, but we’ll be representing the dynamics over time through what’s called a kymograph Now what you’re seeing on the left is a micrograph of the receptor LYK3-GFP fluorescence, flotillin 4 fluorescence, and the merge Likewise, here for bacterially treated root hairs, LYK3-GFP, flotillin 4-mCherry, and then the merge Now, at each point in time, we don’t look at the whole root hair, we’re going to just be looking at a transect shown here, and also shown over here, by these blue arrows And at any one point in time, just that line is going to be represented as follows Here, if we take this line going across the uninfected LYK3-GFP, then we can see at one 10-second period whether a particular position was light or dark

We can follow that over time and we can see that the correlation is rather loose. However, flotillin 4 is very stable; if you look at the flotillin 4, 10 seconds, 20 seconds, and so forth if it’s fluorescent at one 10-second interval, it’s highly likely to be fluorescent in the next one, and so you get this set of lines going through Now, if we take a look at root hairs that are treated with bacteria, we find the following Again, this is what the root hairs look like, but if we follow them over time, you can now see that the LYK3-GFP is very stable. If that transect shows a bright point at one 10-second interval, it’s probably going to be bright the next 10-second interval as well, so the LYK3 has settled down, the flotillin 4 is still very stable, and when you take a look at the merge, you can see that there appears to be a correlation of the brightness So, as we summarize, we can say the following: That the puncta density is changed for FLOT4-GFP in a genetic background where LYK3 is mutated We can see that LYK3 itself localizes in puncta that suggests it’s in membrane microdomains Also, it also localizes to infection threads We found that in uninfected root hairs, the receptor and the flotillin do not have much overlap, and also they are very different in their motility, in their dynamics However, after inoculation with Sinorhizobium meliloti, the receptor and the flotillin co-localize, and their dynamics remain similar Not known is whether this is direct or indirect What is the nature of the protein-protein interactions? Because fluorescence in and of itself is relatively loose, it does not have to have precise molecular adjacency So further fluorescence studies with FRET and biochemical studies will be necessary, and we think that following the combination of flotillins and receptors is going to be a very exciting way to ask how the plant is mobilizing itself to accept infection So back to the outline I’ve finished talking a little bit about how a candidate gene approach took us from the study of flotillins to the study of infection In this next segment, I’d like to talk about some mutants that we’ve isolated called the dnf mutants These have identified plants genes necessary for the final stages of symbiosis And I’ll talk about two of those, DNF1, which encodes a signal peptidase, and DNF2, a putative phospholipase C So looking again at the sequence of nodulation, we’re now taking a look at an even later stage, where bacteria have penetrated into the nodule, and they get released into the cells, and they’re able to fix nitrogen So the dnf mutants, called “defective in nitrogen fixation,” were isolated out of a screen that we did of mutants that we generated by fast neutron bombardment This often creates deletions, it’s an ionizing radiation, and so these are severe mutants When we screen them, we found a number of mutants which were like this. Instead of being nice, pink nodules here, the dnf mutants are white, and they’re small. They don’t have any nitrogenase activity, and that’s shown in this slide by assessment of acetylene reduction, shown here Wild type is able to convert acetylene to ethylene, which an indicator for nitrogenase enzyme But the dnf mutants, as you can see, have very low or even no ability to fix nitrogen Now, they also don’t seem to have completely normal development One of the mutants, dnf1, is shown here, and we got two alleles of that, they’re very similar A wild-type nodule has cells filled with bacteria The bacteria do get infected into the dnf1 nodules as well; nonetheless, they are not as large, and they do not seem to be able to fix nitrogen So we took a look further at dnf1, and one of the ways in which we studied this was as follows Here’s a wild-type nodule shown in section, and this nodule, in this case, has been established by the action of a bacterium

carrying a glucuronidase fusion into the promoter for the nitrogenase gene. That means that if the bacteria are expressing their nitrogenase, then the nodule will turn blue in the presence of a glucuronidase substrate So, what we see here is that the nifH (which is nitrogenase)-glucuronidase fusion is turned on That’s good, that means that the bacteria are expressing their nitrogen fixation genes However, what we found with the dnf1 was the following: Taking a look at these nodules with the same exact bacterium carrying a glucuronidase fusion to the nif promoter, we found no activity You could see here with the red arrow; it’s pointing toward the nodule, but there’s no blue stain So we conclude from this that the bacteria are not able to express the genes for nitrogen fixation So we can say that the bacteria need the plant’s DNF1 in order for the bacteria to fix nitrogen, or even to express their nitrogen fixation genes So what is dnf1? This gene in the plant encodes one of the subunits of what turns out to be a nodule-specific signal peptidase And I’ll tell you about nodule-specific and the other components in a moment But first I’ll just review a little bit what signal peptidase is In a eukaryotic cell, we’ve got the endoplasmic reticulum, and proteins that are extruded into the lumen may get there by a signal peptide, and if the signal peptidase cleaves off that signal peptide, and that allows the production of the mature protein, which would be shown here Now, once that mature protein is formed, after its signal peptide is taken off, then it may be targeted for secretion or for vesicle trafficking So what we found is that DNF1, required for bacterial gene expression and differentiation, has something to do with protein processing Now let’s take a look at the expression of this gene If you take a look at the vegetative parts of the plant, such as the leaf and stem or flowers, what you find is that expression of the DNF1 signal peptidase subunit is very low, but it’s quite high in nodules It’s also present to some extent in developing seeds, although at a much lower level Taking a look at the time course, what we can see is here The DNF1, if you plot days of nodule growth and development, and DNF1 expression, you can see that DNF1 goes up early in nodule development You can also track it during seeds and find that it’s relatively steady during the maturation of seeds So, DNF1 is most highly expressed in nodules and it goes up very early in those nodules Now, let’s take a look not just at DNF1, but using transcription, let’s take a look at what genes, out of all the microarray data, tend to be expressed whenever DNF1 is expressed, and those are shown here Now, on the left, you can see the names of the proteins This is what they do, and what you can see is that there’s the other subunits in the signal peptidase complex, are also regulated up in early nodules the same way the DNF1 is In addition, the signal peptide peptidase, which is needed for completion of the degradation of the signal peptide, that’s also upregulated And finally, this very intriguing protein, SYP132, syntaxin 132 Now what are syntaxins? This is a reminder of how they have been characterized in animal cells; syntaxins are important proteins for vesicle targeting, and in fact, SYP132 here in the plant probably marks the last step of protein secretion, and empirically, it has been observed on plant plasma membranes and also on what’s called the symbiosome membrane That’s the membrane that surrounds the bacterium once it’s inside the cell So if we take a look at all of these genes, we can see that they are co-regulated So we’ve got a whole signal peptidase complex, and what is it doing? One of the most important questions that could be asked is, what are the substrates? What proteins does this signal peptidase help to mature? And the answer came from work in the labs in Gif-sur-Yvette and in Szeged, Hungary, led by Eva Kondorosi and Peter Mergaert, and their colleagues And what they have described is a set of proteins

called “nodule cysteine-rich peptides.” These are characteristic of nodules, they are not expressed anyplace else in the plant And they are rich in cysteine and bear some resemblance to defensins, which are proteins made by plants and animals in response to microbes They were able to show using our dnf1 mutant, that one of these, for example, NCR peptide #1, in a wild-type plant, gets processed to a smaller molecular weight But unless the DNF1 peptidase is there, it does not get processed, because here’s our mutant, and you can see that the protein, shown here in immunoblot, is larger It’s because its signal peptide was never taken off They have also been able to show that, with antibodies, that in wild-type cells, if you use a green dye to indicate the bacteria and a red fluorescence for the antibody to the NCR peptide, they co-localize Bacteria and NCR peptides co-localize in a normal wild-type cell Here, they showed an example of what happens in a mutant dnf1 cell You can see that the bacteria are all over the place in green, but that this NCR peptide, which we know is not being processed, right? Because it’s large molecular weight? That NCR peptide is stuck someplace, probably in the endoplasmic reticulum So, their conclusions, together with our work on the DNF1, are that the NCR peptides are substrates of the DNF1 complex In wild type, these are delivered to the symbiosome, and in dnf1, they’re retained in the ER They’ve shown in other work that these mature NCR peptides have profound effects on bacteria, such as causing them to cease cell division and changing their membrane permeability, so that’s an ongoing story about the activity of the NCRs To complete the model so far, we might ask, what does DNF1 do, and one likely activity is that it processes NCR proteins These end up in the lumen of the ER, and they are targeted, perhaps using SYP132 as part of the mechanism for targeting, into vesicles, which come over and fuse with the symbiosome, in which Rhizobium are going to be differentiating Now, we also have questions about DNF1 beyond the NCR proteins For example, DNF1 is present even in legume plants that do not have the NCR proteins So what is it doing there? Another question is, in the dnf1 mutant, the bacteroids do not differentiate, but they also stop dividing, so it seems that even without DNF1, there may be other signals that are helping these bacteria to stop dividing. So we hope that we’ll be able to work on those, as well as on the actual mechanism of DNF1 in the next period of our work I’d now like to finish by telling you about the other mutant, dnf2 And this, as you recall, as shown right there, that also had no nitrogen fixation And we’ve done some work to characterize what’s going on with dnf2 and also to figure out what the gene is Here is a picture in which you can see, from the root part to the distal part of the nodule, fluorescently labeled bacteria, and they are inside We can tell that the bacteria get inside the dnf2 mutant nodules However, they don’t express nitrogenase, and they don’t express many other bacterial genes either So, somehow, DNF2 is again required for the bacteria to differentiate and, in fact, to express genes Now we know that DNF2 is expressed only in nodules, even more strictly than DNF1 This shows you some promoter-GUS fusions where you can see that the early part of the nodule is expressing DNF2, and here you can see it at the tip, and then it diminishes just slightly as the nodules get somewhat older What is DNF2? It appears to be similar to a phospholipase C, but it’s not a canonical phospholipase C the way we’re used to seeing Here would be the classic PLC You can see it’s got the inositol trisphosphate binding, EF-hands, catalytic domains, calcium-binding domains, and so forth

The DNF2 is only homologous to one part of this large, complex phospholipase C However, we do find that it is very, very similar to a phospholipase C whose model is shown here This is actually a bacterial phospholipase C, and if you take this in red and then superpose DNF2, it does appear to be very, very similar in its 3D structure So we want to know what’s the biochemical activity; sequence alone doesn’t tell us whether it really acts to process phospholipids or as a signaling But we found one other thing that’s really surprising to us First, as I mentioned, it’s only expressed in nodules, but secondly, it’s not expressed in the cells of the plant that have Rhizobium It’s expressed in the noninfected cells of the nodule And that’s really a first There are known metabolic enzymes in soybean, for example, that are present in uninfected cells But there has never been found a regulator that appears to be active in the noninfected cells of the root nodule So, what we’re trying to ask ourselves is, does the DNF2 protein mediate the infection process directly or indirectly? Is it acting in some way that allows the noninfected cells to send signals or perhaps to avoid signals? So those are some of the questions that lie ahead for us So winding up the combination of the studies that I’ve shown you today, I think that the future of our field is going to be very exciting as we take the results of the plant mutants and we study the proteins that are defined by those mutations, and look at how those work, both plant and bacteria together, not looking just at position, but looking at time We’re interested in the great questions of: What the signals are, and how the signals are transduced Given that so much complicated cell biology has to happen between the beginning of the infection thread and the maturation of the symbiosome, we want to know not just the regulation, but what is the machinery of this infection, and we believe that both the proteins and molecular probes that we can obtain can be used to study the individual steps of nodulation as they move forward So that concludes our journey through the plant side of symbiosis