Azotoreview

Azotobacter vinelandii is known for its amazing ability to fix nitrogen, converting nitrogen gas into biological forms like protein even in the presence of oxygen. This is apparent even in its name—Azotobacter—which translates approximately to "nitrogen bacteria."

But the nitrogen-fixing process and machinery is pretty complicated and difficult to study in a reductionist fashion. Many components don't work the same outside of the context of the rest of them or outside of the cell itself.

So in this study, scientists (some of whom were involved in sequencing the first A. vinelandii genome) took a broad approach to the subject, by looking at the expression of all the genes in the organism, comparing their expression when the cells were fixing nitrogen compared to when they weren't.

In addition, they examined the expression of genes when the bacteria were grown with or without certain metals needed for the various nitrogenase versions: molybdenum, vanadium, etc. More specifically:

• To study non-nitrogen-fixing cells: they grew cells with ammonium
• Cells using the molybdenum nitrogenase: they grew cells with molybdenum and no ammonium
• Cells using the vanadium nitrogenase: they grew cells with vanadium and no Mo or ammonium
• Cells using the iron-only nitrogenase: they grew cells with iron and no other metals or ammonium

Simple enough.

Then, to measure levels of expression, they extracted RNA from the cells in each condition, converted the RNA sequences to DNA (called cDNA or complementary DNA, converted from RNA messengers), and then these chunks of DNA were sequenced using a high-throughput technology called SOLiD.

SOLiD (or Sequencing by Oligonucleotide Ligation and Detection) is one of the main kinds of next-generation sequencing, alongside Roche's 454 pyrosequencing and Illumina. I had to look it up. Apparently one machine these days can sequence 5 trillion bases per day (for reference, the human genome is about 3.2 billion bases long, so SOLiD could sequence more than 1500 human genomes per day). Of course, this is not cheap: that much sequence data would cost about $500,000.

The way it works is by cutting a piece of DNA into short sequences, binding them to tiny magnetic beads so there's one sequence per bead. Then the beads are mixed into an emulsion of oil so that on average, each bead is encased in a small bit of water in a sea of oil, along with reagents needed for polymerase chain reaction (PCR). This allows the DNA on each bead to be copied many times, all at once yet individually, so that many reactions can be done in the volume of liquid that would normally allow only one. Very cool. This is the same technology used to prepare samples for 454 pyrosequencing.

In pyrosequencing, the sequencing is done with DNA polymerase, which is what is normally used to copy DNA, and each base is added one at a time, so each bead will incorporate only the base that fits with the sequence bound to the bead. When a base is added, a tiny bit of light is given off, which a camera in the machine detects and registers it as the base that was added at that time.
However, SOLiD works a bit differently: instead of DNA polymerase, the enzyme is DNA ligase, which links together two strands of DNA. And instead of single bases, short DNA sequences called probes are added, with the two bases at one end known. When one of these matches the bead's sequence, ligase links it up. The probes are labeled with fluorescent molecules, so the next two bases in a sequence can be determined from the color of the fluorescent probe. Then this is cleaved off and another two bases are added. It's somewhat confusing; this site might help a little.
This system avoids some of the problems that pyrosequencing experiences, such as with accuracy, but has some problems of its own (especially price), so it's useful in some cases and not as much in others.

So what good is sequencing all the cDNA? The number of copies of one cDNA, relative to the copies of others, shows how much the cell is transcribing that gene, which can be an indicator of expression levels (transcribing more may mean that the gene is translated into protein more, so there may be more of that specific protein in the cell). It doesn't always work quite proportionally, since there are mechanisms other than transcription for regulating cellular protein levels, but it can usually provide some interesting data. So they sequenced all the cDNA in the cell and compared the number of copies for each gene to see which ones were present in higher or lower numbers in different growth conditions.

Then after analyzing all this data (a large undertaking in itself) and finding genes that seemed to be expressed at higher or lower levels in the different conditions, the scientists confirmed the most interesting findings using real-time quantitative PCR, which is a more sensitive way of measuring the same information. It works by doing PCR on a gene but adding some kind of fluorescent molecule to detect exactly how much of that sequence is present over time in the PCR. Ideally PCR should double the number of sequences in every round of the reaction, but this doesn't always work exactly, so the most accurate measure is to determine the point at which the fluorescence becomes bright enough that the PCR machine can detect it, and then extrapolate back to figure out how much of the sequence was present at the beginning. It's another way to compare transcription levels.

Ok, finally on to results. The authors found that almost 30% of A. vinelandii's genes were affected when fixing nitrogen compared to when not doing so. Many of these were affected regardless of which nitrogenase the cells were using. Mo nitrogenase growth affected the most genes on its own compared to the other two, but the two alternative nitrogenases (V and Fe) together affected more genes than any nitrogenase on its own. So overall, compared to non-fixing conditions, the using the alternatives affected many more genes than using the Mo version; but compared to each other, using the alternatives didn't change many genes. Apparently using the alternatives involves a large shift in the cell's gene regulation, compared to using the main Mo nitrogenase. I wonder why.

Genes Expressed When Using Mo Nitrogenase
The nif genes that make up the Mo nitrogenase are found in two clusters in the genome, one near the beginning (relative to the origin of replication) and one near the end. Some of these genes form the actual enzyme structure itself, some help to put it together with its metal-containing cofactor and such, and some (especially nifA) regulate the process.

Not surprisingly, the main structural genes increased their expression greatly when the cells switched to nitrogen-fixing mode, between 50 and 150 times higher. The primary dinitrogenase reductase, NifH, increased the most, which fits in with previous observations that a high ratio of this protein to the Mo-containing dinitrogenase allows higher nitrogenase activity.

More surprisingly, other nif genes in the major cluster only increased expression up to about 14 times more. This could be because not much of their proteins is needed, or possibly that they were already expressed at high levels and their regulation is mostly post-transcriptional, so not much change would be visible in transcript levels. In the minor cluster, some genes necessary for making the metal cofactor increased around 20-fold.

Other than these expected increases, lots of other genes changed as well; not surprising, considering that nitrogen fixation is essential for growth in low-nitrogen environments but is also very energy-intensive. The most significant changes were in type IV pilus genes. These pili, little hairlike projections from the cell, are involved in lots of things: motility, sensing the environment, attachment to surfaces, etc. It's not entirely clear what they're doing in this context, but apparently something.

Another important factor for nitrogen fixation is protecting the nitrogenase from oxygen. A. vinelandii seems to do this by consuming a lot of carbon in order to reduce whatever oxygen is present, transforming it to water. Its genome has many electron-transporting proteins such as oxidoreductases and terminal oxidases, some of which did appear to be somewhat upregulated in nitrogen-fixing conditions. This could also be useful for producing more energy to power the nitrogenase. The genes for the uptake hydrogenase, which recovers hydrogen produced by the nitrogenase and regenerates some energy from it, also showed increased expression.

There also seems to be a change in some genes associated with iron and sulfur organization, something else that is important for nitrogenase because it contains multiple atoms of these elements. Also, not very surprisingly, the genes related to molybdenum gathering increased also.

Genes Expressed When Using Alternative Nitrogenases
Obviously, the genes that encode the alternative nitrogenases themselves (vnf genes for the vanadium-containing nitrogenase and anf genes for the iron-only nitrogenase) are going to be upregulated when these are in use. The V (vanadium) nitrogenase is used when molybdenum is not present but vanadium is, and the Fe nitrogenase is used when neither of these metals is available.

But these alternative systems don't have alternative versions of all of the necessary enzymes for fixing nitrogen, only the main ones, so they share some of the proteins that the Mo nitrogenases uses. This is especially true of enzymes involved in assembling the nitrogenases and their cofactors, such as NifUSVMB.

When vanadium was present, vnf genes were upregulated, as I said, but in this case, the vnfH gene encoding the vanadium dinitrogenase reductase wasn't as high relative to the other V nitrogenase components as was the case with the molybdenum nitrogenase. Not sure why. There were some other differences, especially that vnf homologs of nif proteins involved in cofactor synthesis were expressed in different proportions, so the process of V-containing cofactor synthesis might be different somehow.

In the case of the iron-only nitrogenase, the nifH and other components' homologs (anfH, etc.) were upregulated in the same ratio as the nif genes, distinct from the vnf homologs: that is, anfH was expressed much higher than anfDK, around four- to five-fold higher.

Of the genes that don't have anf homologs, some nif genes were upregulated (nifUSVMG again), but in other cases the vnf versions were preferred (vnfENXY). vnfH was also upregulated, even though there is a separate anfH. This is in agreement with other previous studies (003), and may be because vnfH has some kind of role in regulating gene expression.

Genes Related to Electron Transport
All of the nitrogenases require electron transport machinery, since the nitrogenase functions by putting electrons (and protons) onto nitrogen gas (N₂) to make ammonia (NH₃). This takes at least eight electrons for each molecule of nitrogen: six for two molecules of ammonia, and two for one molecule of hydrogen as a byproduct. The alternative nitrogenases produce more molecules of hydrogen, so they need even more.

Some of the genes involved are nifF and vnfF, which encode proteins called flavodoxins that transport electrons. They may not be necessary to fix nitrogen, but presumably they're helpful. When the cells were using the Mo nitrogenase, nifF was upregulated, and both were higher when Mo was absent (though vnfF much more so).

Some other genes that seemed involved included rnf1 genes, whose products are membrane-bound and also help to transport electrons to nitrogenase; they also seem to be important for the iron-sulfur cofactor of dinitrogenase reductase. And fix genes also seem important for electron transport. All of these were expressed more when fixing nitrogen in all conditions, but when Mo was absent, fix genes were much higher than rnf1 genes.

Regulatory Genes
Clearly A. vinelandii's nitrogenase system has a lot of regulation going on, so regulatory genes are important. vnfA and anfA are necessary to use the alternative nitrogenases, as is nifA for the primary nitrogenase, and these regulatory genes increased whenever their respective isozyme was in use, though low levels of them were present constantly. Regulation of these genes is likely to be how the cells turn on and off the alternative nitrogenases.

There are a few other homologs of nifA and vnfA that show similar patterns, but may fine-tune the regulation somehow (how is not yet known).

Other Differences in Global Expression
As mentioned, the transcriptional profile when using the Mo nitrogenase is very different from when using V or Fe nitrogenases, probably because the latter are less efficient. The most apparent difference in this study was in the hutU gene, for urocatanase hydratase, which increased greatly when Mo was absent. This gene is necessary to degrade histidine, one of the 20 common amino acids, which makes sense because cells using a less efficient nitrogen-fixing enzyme might want to get nitrogen from other places too, like breaking down some less essential proteins. Similar results have been seen in other diazotrophs.

Also quite interesting, genes for a putative soluble hydrogenase discovered when the genome was sequenced were upregulated, especially when using the Fe nitrogenase. This may be a backup system for recycling the extra hydrogen molecules that these nitrogenases produce, to recover the valuable energy that would otherwise escape.

Some other genes increased also, but the function of their products is as yet unknown. Might be worth investigating.

Comparing expression when using the V nitrogenase vs. the Fe nitrogenase, there was at least one interesting point: there were a few genes near a vnf operon upregulated during V nitrogenase growth, seemingly related to a transporter system, so they're probably a vanadate transporter.

Evolution of Nitrogenase
There's some debate about which came first in history: the Mo nitrogenase or the alternatives. Since the alternatives are less efficient, it would make some sense if they came first and the Mo nitrogenase just improved on them, especially since Mo and possibly V were probably difficult to find before there was much oxygen in the atmosphere. But no one has discovered a species that has alternative nitrogenases and not the Mo nitrogenase, though there are plenty that have only the latter, and not many that have all three. And this study seems to show that the alternatives evolved from the Mo nitrogenase to allow cells to thrive in environments where Mo is absent, which also makes sense.

So these results are very interesting and potentially useful, and seem to tell us a lot about what's going on inside the cells in different conditions!


Citation: Hamilton, T. L. et al. Transcriptional Profiling of Nitrogen Fixation in Azotobacter vinelandii. J. Bacteriol. 193, 4477–4486 (2011).

Remember the uptake hydrogenase? A byproduct of nitrogen-fixation is hydrogen gas, which allegedly would be a waste of energy, except Azotobacter vinelandii has an enzyme that oxidizes it and converts it into ATP to recoup some of the energy cost of nitrogen-fixation. That's the common story, anyway. But in science you can't always rely on common stories, so Kerstin Linkerhägner and Jürgen Oelze decided to test that hypothesis.

Actually Yates and others had already tested to see if hydrogenase-knockout mutants of Azotobacter chroococcum were impaired, and found that wild-type strains outcompeted them, at least in some conditions such as when carbon was limiting in the medium (019). But these results may be questionable because the wild-type could consume both its own hydrogen and that produced by the mutant when they were mixed together. This wouldn't often be the case in the wild though.

So Linkerhägner and Oelze grew A. vinelandii strain DJ (relatively wild-type) and a hydrogenase-knockout strain in carbon-limited chemostats. They tried two different concentrations of oxygen, low and high. In terms of biomass, the biomass levels from DJ at each oxygen concentration over a range of dilution rates matched rather well with the results these authors saw in 066 (two years later), though not perfectly. Interestingly, the mutant strain matched its parent perfectly. As D increased, the amount of biomass they produced increased. Presumably this would've leveled off at the lower oxygen concentration if they had tested higher D values, as in 066, but they stopped.

Their next figure is somewhat conceptually challenging, but I'll try to break it down:

Figure 2, Linkerhägner and Oelze 1995

It shows the inverse of the yield, Y (which is, in this case, grams of biomass produced per mole of substrate [glucose] consumed) on the y-axis, and the inverse of the dilution rate D on the x-axis. The different symbols are the two strains and two oxygen concentrations, but the strains are virtually identical, so all you need to know is that the top line is higher oxygen and the bottom is lower.

Biological nitrogen fixation (turning N₂ gas into usable forms for protein and such) is an energetically expensive process, requiring large amounts of resources the cell could devote to other purposes; however, if that's the only available source of fixed nitrogen, it's worthwhile, because the alternative is paralysis, essentially.

However, being such an expensive process also means that cells will try to regulate its use very tightly, making sure only to use it when it is absolutely necessary. This is different for different organisms; some regulate the same way all the time, some photosynthetic microbes turn everything off or on depending on available light, etc.

Assembly of the nitrogenase enzyme is a very complex process that requires complex regulation as well. This study looks at the regulation of the molybdenum-containing nitrogenase, the primary one, especially the nifB and nifEN genes. NifB is a protein that seems to help synthesize the Mo-containing cofactor essential for the nitrogenase; it also seems to work for the vanadium- and iron-containing cofactors of the alternative versions. The cofactor that NifB makes transfers to the NifEN complex, which adds the Mo for the nitrogenase. This NifB process seems to be a key point for regulating the entire process.

The nitrogenases each have an activator that helps regulate them: nifA, vnfA, and anfA, and these all influence nifB production. But there are probably other genes involved. For example, Azotobacter vinelandii has a gene called clpX2 in between two other nitrogenase-related genes; clpX encodes a common protease that breaks down proteins that are deformed or no longer useful, but clpX2, while seemingly related, is different. It's sometimes found in nif gene clusters in other species, and knocking it out doesn't disrupt nitrogen fixation; rather, it may increase it.

This study looks more specifically at ClpX2's role in regulating nitrogen fixation in A. vinelandii. To do this, they made new strains with modified genes involved in this process:

• UW233: NifB only works when chemical called IPTG is present; can't fix nitrogen otherwise
• UW238: nifB is IPTG-inducible; nifENX is deleted
• UW295: nifB is IPTG-inducible; nifA is deleted
• UW318: clpX2 fused to lacZ; produces more yellow color from ONPG when clpX2 expressed
• UW319: clpX2 fused to lacZ and nifA is deleted
• UW322: lacks clpX2 gene

UW233 allowed them to control when NifB was produced. They found that when cells were growing with ammonium (a source of fixed nitrogen), not fixing nitrogen, they accumulated higher levels of NifB. This is probably because the cells consume NifB when fixing nitrogen; after removing IPTG from the cells' medium, they stopped producing NifB, but those growing in ammonium still had fairly high levels of NifB even after a few hours, while those fixing nitrogen lost most of theirs.

In UW238, NifB accumulates to higher levels regardless of whether or not ammonium is present, so the NifENX proteins seem to be involved in NifB's regulation. In UW295 when nifA is missing and all the major nif genes are silent, NifB disappears more quickly in both conditions; it seems that whatever is degrading NifB isn't activated by NifA.

Using UW318, the authors discovered that clpX2 was expressed more when fixed nitrogen was absent and the cells were fixing nitrogen, so ammonium seems to downregulate it. UW319 revealed that NifA was not necessary for clpX2 expression either; in fact, expression was higher when nifA was deleted. Why is unclear. Semi-quantitative RT-PCR confirmed these results.

So then the question is, how is ClpX2 involved in regulation of NifB and NifEN? So of course they deleted the clpX2 gene to get UW322. The main difference in this strain was that levels of NifB and NifEN were higher than usual, much higher; it seems that ClpX2 plays a big role in their turnover.

However, deletion of clpX2 comes with a price. When fixed nitrogen was present, the cells grew fine, but when fixing nitrogen they slowed down a little, and the initial setting up of nitrogen fixing was slower too.

One interesting specific effect requires a bit of explanation: the Mo nitrogenase has two main components, which are the main part that contains the Mo cofactor and does the actual reaction with nitrogen, called the dinitrogenase; and the dinitrogenase reductase, which kinda recharges the dinitrogenase and prepares it for the next round of reactions. Having these present in different ratios can affect the overall rate of the process.

But what the authors found in UW322, with clpX2 missing, was that there was much more of the Mo-cofactor-containing dinitrogenase than there was normally, while levels of dinitrogenase reductase remained the same. So maybe ClpX2 holds in check the production of dinitrogenase somehow, so the ratios of the two components are optimized.

The authors hypothesized that ClpX2 might provide an advantage when iron is scarce, because the proteins it regulates are both involved in directing a lot of the cell's iron into nitrogenase cofactors, so they tested in low-iron conditions. UW322 seemed to have a slightly greater disadvantage when fixing nitrogen with limited iron than when fixing nitrogen with sufficient iron, but I'm not sure it looks that significant. Could be.

So here's the figure they made to explain their results, showing the regulation pathways:

Figure 10, Martinez-Noël et al. 2011

It doesn't really specify how ClpX2 might repress those proteins (i.e. by their degradation probably); nor is it clear how ClpX2 itself is regulated. But it is interesting.

Citation: Martínez-Noël, G., Curatti, L., Hernandez, J. A. & Rubio, L. M. NifB and NifEN protein levels are regulated by ClpX2 under nitrogen fixation conditions in Azotobacter vinelandii. Molecular Microbiology 79, 1182–1193 (2011).

One good way to learn a lot about a bacterial species, or at least to get a lot of hints about what it might do or be capable of doing, is to sequence its genome. So that's what a bunch of people decided to do with Azotobacter vinelandii. It makes sense, since this organism is one of the better-studied ones and has interesting capabilities, such as nitrogen fixation.

The strain they chose was called DJ, a variant of the wild-type strain CA. DJ is supposed to be easier to manipulate genetically than its parent. So they sequenced its entire genome, but focused mainly on the surprising amount of oxygen-sensitive enzymes they found in an obligately aerobic organism.

The method of sequencing, for those who care, was plain shotgun Sanger dye-terminator sequencing after generating a clone library. (Apparently, for some reason, Monsanto did a lot of the work.) I guess this was before the next-generation sequencing technologies were available or affordable. And teams of undergrads did much of the work labeling genes and such.

Overall, the genome is pretty similar to that of pseudomonads, especially Pseudomonas stutzeri (another nitrogen-fixing soil microbe). Some of their genes have been rearranged compared to each other, though, and A. vinelandii has almost 1,000 more genes.

In terms of energy-generating systems, A. vinelandii's genome has all the genes needed for aerobic metabolism but seems to lack any complete system for anaerobic respiration or fermentation. It is well-equipped for aerobic respiration though, which it seems to use to consume large amounts of oxygen that would otherwise damage its nitrogenase and other enzymes. The other mechanism it has to protect its nitrogenase is called the FeSII or Shethna protein, which can temporarily deactivate the nitrogenase when oxygen is too high, protecting it from damage.

The sequence showed the precise location of each set of nitrogenase genes relative to each other. They're somewhat spread out. It also located the mod genes for molybdenum (Mo) transport and the hox genes of the uptake hydrogenase (which are pretty close together). Though it turns out there is a second set of genes similar to the original mod operon elsewhere in the genome, that may be a second Mo transport system. Possibly even a third set right next to the first, but it's not certain what it does.

Somewhat interesting is a set of genes that are similar to something called carbon monoxide dehydrogenase (CODH) that is present in some anaerobic organisms. This can convert CO to CO₂ and H₂, effectively using it as an energy source instead of something toxic. But it's not certain whether this is functional in A. vinelandii at all. It may be related to some genes that seem to be related to soluble hydrogenases in other organisms, but their function isn't clear either.

A. vinelandii, some strains of it at least, is well-known for producing certain polymers: polyhydroxybutyrate (PHB), which can be used to make a kind of bioplastic; and alginate, and kind of mucusy stuff that has various uses too. The strains that make alginate are rather slimy and hard to work with, and supposedly this provides a further barrier against oxygen poisoning, but strains CA and DJ don't make it, and this sequence revealed why: a transposon inserted itself in the middle of a regulatory gene, inactivating it. That's all it takes.
The genes for PHB synthesis seem to be intact though.

Knowing the sequence of an organism is very helpful; if you want to check for new capabilities, you can just check the genome. So this is a good study.

Citation: Setubal, J. C. et al. Genome Sequence of Azotobacter vinelandii, an Obligate Aerobe Specialized To Support Diverse Anaerobic Metabolic Processes. J. Bacteriol. 191, 4534–4545 (2009).