 |
MB 451 Microbial Diversity |
 Audio recording of this lecture |
Readings for this lecture:
- Beja O, et al. 2000 Bacterial rhodopsin: Evidence for a new type of phototrophy in the sea. Science 289:1902-1906.
- Manefield M, Whitely AS, Griffiths RI and Bailey MJ 2002 RNA stable isotope probing, a novel means of linking microbial community function to phylogeny. Appl. Environ. Microbiol. 68:5367-5373
Linking phylotype & phenotype: Identifying the environmental contribution of an uncultivated specie
Microbial ecologists have the tools to quantitatively decipher the processes taking place in microbial communities. They can follow nitrogen, sulfur, carbon, reducing equivalents, energy, &c, &c, as they are transformed from one form to another. For example, it is reasonable to measure the steady-state levels of all of the forms of nitrogen in a system, determine the inputs and outputs, and the flux rates between all of the forms in a ecological community. The limiting factor in microbial ecology these days is that essentially all we know about the organisms that make up these communities comes from the study of the less than 1% of microbes that are readily cultivated. I hope it is clear at this point that even taking a basic quantitative census of a microbial population is not straightforward, but it is possible, with much work, to get a semi-quantitative assessment of the makeup of the population, especially those organisms that are abundant in the population. It is much harder to link these bits of information together - to identify the organisms responsible for a particular step in an ecological transformation. For example, suppose you have an environment in which a certain process is taking place, and you have a phylogenetic census of the organisms in a sample from this environment, how would you determine which organisms in your census are the ones that carry out the process you're interested in? Or, coming from the other direction, suppose you know an organism is abundant in an environment, how do you determine what it's doing there, i.e. what it's ecological niche is? This week we'll be discussing two approaches to this problem: genomics, and stable-isotope probing (SIP).
The genomic approach
This paper addresses the problem of linking microbial processes to specific organisms using the genomic approach; fishing out big peices of DNA that contain both phylogenetic information (a copy of the ssu-rRNA gene) and phenotypic information (in thius case, a gene for phototrophy). Because both types of genes come from single peices of DNA, they must be from the same organism.
Question : What is the ecological role of the SAR86 rDNA sequence group?
In this paper, the authors already know from molecular phylogenetic analysis of ocean water that members of the "SAR86" group within the gamma-proteobacteria are abundant worldwide, but had no idea what their role in the ecology of the ocean might be. So, again, the question is linking a particular phylogenetic group to a particular ecological process.
In attempt to find out what SAR86 is doing in the ocean, they made a cosmid bank of DNA isolated from ocean water. These cosmid clones contain DNA fragment more than 100Kbp in length. They screened these cosmids by hybridization to identify those that contained the gene for 16S rRNA, so they could identify the organism the DNA fragment came from. One such clone (EBAC31A08) proved to be a member of the SAR86 group, and the authors sequenced the entire 130Kbp DNA fragment, hoping to find genes that would provide clues about the metabolism of the organism. This isn't as hopeless as you might imagine - most gamma-proteobacteria have a genome size of about 4Mbp and 7 rRNA operons, so that about 20% of the genome is within 130Kbp of an rRNA gene. Another way top look at it is that the average size of a gene is about 1Kbp, so they expected to get about 130 genes - the chances are good that 130 genes chosen randomly from a bacterial genome will give you clues about it's phenotype.
They were successful beyond all expectation. One of the genes they found seems to be a gene encoding a rhodopsin, presumably acquired by horizontal transfer from a halophilic archaeon. Halophilic Archaea have 3 rhodopsins - the main one is bacteriorhodopsin, the light-driven proton pump they use to capture energy (via the proton motive force) from light. Halorhodopsin is a light driven chloride pump, used to bailing Cl- out of the cell (cells use organic anions like glutamate rather than harsher salt anions like chloride). They also have a sensory rhodopsin, used along with a signal transducer protein to signal the presence of sufficient light to justify turning on the genes for the other rhodopsins and retinal (the cofactor for all of these rhodopsins). Rhodopsin genes have also been transfered to eukaryotes at least once - the mold Neurospora has an archaeal-derived rhodopsin it uses for light sensing to control its diurnal cycle. It may (may!) also be that the opsins used in vision in animals originated from one of the archaeal genes. Could it be that these organisms are using this rhodopsin to grow phototrophically? Is primary production in the ocean based on two kinds of photosynthesis instead of just one?
Figure 1 is a pair of trees based on the 16S rDNA (panel A) and the rhodopsin gene (panel B). The rDNA tree is just shown to demonstrate that the genome fragment comes from an organism of the SAR86 group. The rhodopsin tree is an attempt to determine the likely function of the rhodopsin - is it related to proton-pumping, chloride-pumping, or sensory rhodopsins? The tree suggests that it might be in the sensory rhodopsin group, but this is not a strong association. We'll get back to this issue - the biochemical properties of the protein expressed in E.coli are not those of a sensory rhodopsin but those of a proton-pumping rhodopsin. Also note that the sequence is not related to the Neurospora rhodopsin gene NOP1.
In figure 2, they show that the predicted secondary structure of the "proteorhodopsin" is consistent with that of a bona fide opsin, and contains the conserved amino acids needed to bind it's cofactor, retinal.
In figure 3, they show that if they express this protein in E.coli and add retinal (E.coli doesn't make retinal, of course), the cells quickly turn a hue of red (Absorbance max of 520nm) consistent with a rhodopsin. In other words, the protein as expressed by E.coli is correctly folded and inserted into the membrane in a form that can correctly bind the cofactor.
But does it pump protens? Figure 4 shows that E.coli with both rhodopsin and retinal pumps protons from inside to outside (as measured by the change in pH of the media) when and only when provided with light. They use TTP uptake by rhodopsin/retinal containing vesicles to measure the electrical potential generated : -90mV, which is consistent with a strong proton pump.
In figure 5 they dissect the reaction cycle photometrically to show that this looks like a proton or chloride pump rather than a sensory rhodopsin. Sensory opsins have a slow reaction cycle, >300msec. The longer the recycling time, the longer the signal is sent to its associated transducer/regulator protein. This is why the rod cells in your eyes allow you to see better in the dark - their opsins have a longer reaction cycle than those of the cone cells. Proton or chloride pump rhodopsins, however, have short reaction cycles, <20msec, so they're ready to absorb another photon & pump another ion as quickly as possible. Panel A shows absorbance changes in rhodopsin-containing E.coli over time after being pulsed with a 532nm laser. Absorption increased at 400nm for a very short time, <5mec - this is the activated retinal (the "M" intermediate). The increase in absorbance at 590nm, which increases at the same time scale as the M-intermediate decays and decays with a halflife of 15msec, is another intermediate in the light cycle of retinal, the "O" intermediate. The decrease in absorbance at 520 returns to normal in the 15msec timescale too, so the O-to-groundstate transition seems to be the rate-limiting step of the photocycle, as it is in proton-pumping bacteriorhodopsins. The timescale of the cycle is clearly that of a pump rather than sensory opsin.
So, this seems to be a functional, light-driven proton-pumping rhodopsin in a group of uncultivated gamma-proteobacteria that make up as much as 10% of the biomass of the ocean surface water. In a subsequent paper, Ed Delong's lab shows that they can readily detect this rhodopsin (spectroscopically) in bacteria from ocean water, show that it actually works in cells isolated directly from the ocean, that the rhodopsin is present in large enough amounts to provide the energy a cell needs, and showed that deepwater and surface water organisms produce rhodopsins that are tuned to the differences in the wavelengths of light that are available to them.
So, the SAR86 organisms seem to be phototrophic using rhodopsin! As abundant as they are in the ocean, this represents a huge ecological impact. But are they also photosynthetic? That is, can they fix carbon, are they primary producers? So far, the answer seems to be "no". No cultivated SAR 11 organisms (there aren't many, and they're not easy to work woth) can fix carbon.
But this isn't the end of the story. It turns out a lot of marine Bacteria from many phylogenetic groups have proteorhodopsin genes. Phototrophy via rhodopsins may turn out to be an important part of primary production in the oceans.
Stable isotope probing (SIP)
This paper addresses the problem of linking microbial processes to specific organisms in isotopic "feeding" experiments to label rRNA from organisms that eat phenol, isolate the labeled rRNAs, and identify the organisms by molecular phylogenetic analysis.
Question : Who are the organisms in this environment that actually eat the phenol?
If you feed substrates enriched in specific isotopes to an organism, the biomass of that organism will in turn be enriched in those same isotopes - in other words, the molecules of that organism will be isotopically-labeled. This is how most of what we know about metabolism was determined - feed radioactive compounds to yeast, and watch the radioactivity move from one chemical compound to another as the labeled atoms pass through the metabolic pathways. If the labeled molecules in the organism are phylogenetically informative, you should be able to determine who, in a mixed culture, is eating the labeled substrate. For example, if you feed 13C-labeled acetate to an environmental sample in which you know carbon passes through acetate (for example in a wastewater cesspool), the organisms that take up and use this acetate will end up with 13C in their lipids - these can be identified using FAME, and if it is a simple population you might be able to identify the organisms by deconvoluting the FAME profile. Or if you feed some heavy-carbon labeled substrate to a population, the DNA of the organisms that gets labeled first might be the one that predominantly uses that substrate, and you might be able to identify these organisms using PCR and rRNA molecular phylogenetic analysis. In this paper, the authors describe the use of 13C-labeling of rRNA to identify the organisms that degrade phenol in a wastewater bioremediation reactor.
The environment in this case is an industrial wastewater treatment facility that uses an aerobic digestor to reduce the concentration of phenolics in the waste flow to levels that can be "released" into a public waterway. The amount of phenol is staggering - 1800 liters of wastewater per minute at more than 200mg phenol per liter is more than a ton of phenol every two days, about 200 thousand kilograms per year! The reactors contain a continuous-flow microbial sludge (10^10 cells/ml) with a volume of 1.7 million liters - the retention time is therefore about 100 minutes, during which time more than 95% of the phenol must be removed to get it down to 10ug/ml so that it can be dumped. The careful absence of information that would identify the commercial facility is conspicuous. The microbial population seems to turn over quickly - growth is continuous but controlled by grazing protists, and therefore the carbon that goes in as phenol presumably ends up as CO2 from respiration by the protists.
The basic process carried out by the authors was to add 13C (heavy) labeled phenol to a culture for 1-3 days, then extract RNA. The RNA is then fractionated by density - i.e. the amount of heavy carbon incorporated - by centrifugation in cesium TFA (tetrafluoroacetate). The more 13C incorporated, the denser the RNA, and therefore the lower in the gradient the RNA bands. The presumption here is that the organisms that actually eat phenol will incorporate the 13C from the labeled phenol into their RNA. Ribosomal RNA from gradient fractions is converted to DNA using reverse transcriptase, and PCR using rRNA-specific primers is used to amplify rDNAs from each fraction of the density gradients. The rDNAs are separated by denaturing gradient gel electrophoresis (DGGE). rDNA bands in the DGGE gels that are enriched in 13C can be reamplified and sequenced to determine their identity.
How cesium TFA density gradients work
Solutions of cesium salts, which are very dense, when subjected to very high gravitational fields in a centrifuge, will, after several hours, form stable gradients - more concentrated (and therefore more dense) at the bottom, less concentrated (less dense) at the top. In other words, although still in solution, cesium ions "settle" somewhat at very high g's. If the starting solution also contains macromolecules, these will float up or down to form bands where the densty of the macromolecule matches the density of the cesium solution. RNA binds lots of cesium, and is therefore fairly dense in cesium - about 1.7. DNA binds less cesium and so has a density of about 1.5, whereas protein doesn't bind much cesium at all and so has a density of about 1.2. Even within these bands, the molecules will separate slighty based on small differences in their density, usually because of differences in their base composition (GC pairs are denser than AU or AT pairs), or because of differences in their isotopic composition. In other words, molecules (in this case RNAs) with a higher fraction of heavy atoms will settle just a little bit lower in the gradient that those with lighter isotopes.
In their first experiment (Figure 1), the authors try to show that the labeling and fractionations actually work. They use a defined mixed culture of two Pseudomonas species - P. putida can eat phenol, P. chlorophis cannot (but it is resistant to the toxic effects of phenol). In panel A, they just show you where the bands in the DGGE that correspond to rDNAs from each organism are - P. putida is the upper band, P. chloraphis is the lower band. Panels B and C are identical, except that the cells are fed 13C phenol for either 24hr (panel B) or 72hr (panel C). In both cases lane 1 represents the bottom of the gradient (dense), lane 4 represents the top of the gradient (light). Notice than after 24hr both species are seen at in the light fraction, but only P.putida is seen in the heavy fraction - only this species seems to be able to incorporate 13C from labeled phenol and so make heavy RNA. After 3 days, P.chloraphis is also getting labeled, presumably because enough heavy carbon is making it's way into the general carbon pool.
In the second figure, the authors try to show that RNA is more efficiently labeled in 13C feeding experiments than is DNA - they only do this because others are using the same method to label DNA from cells that can use a specific substrate, and the authors show that RNA is a better marker, as long as all you want is phylogenetic information in the form of rRNA.
In the third figure, the authors get to the point. The two gels are DGGE's of rDNAs in fractions from CsTFA gradients from PCRs of the phenol-degraading community labeled for 1 hour (too soon to get significant labeling) and 8 hours. Fraction 4 is from the bottom of the gradient (most dense), fraction 13 is from the top (least dense). The authors identify 5 bands as the "major" bands in the samples, and label them A - E. As you can see, bands A, B, and D are shifted to the bottom (left, dense) of the gradient after 8 hours of growwth with 13C phenol. Each of these bands presumably represents a species that can quickly utilize phenol for growth. In figure 4, they focus on the most intense such band, band "D", and show that it specifically gets more abundant in the heavy fraction - the amount of it in the light fraction doesn't change much. Newly-made RNA, then, all goes to the heavy fraction. They also use mass spectroscopy to confirm that RNA in the denser fractions really is enriched in 13C.
All 5 of the most abundant rDNAs (bands A - E) were cut out of the DGGE gel, reamplified, and sequenced to determine the identity of the organisms they represent. The apparent phenol-degraders were an alpha-proteobacterium (band A), and two beta-proteobacteria (B and D). The single most adundant phenol degrader (band D) turns out to be a beta-proteobacterium in the genus Thauera. This was a surprise, because if you do enrichments and pure cultures, the phenol degraders you isolate from this environment are gamma-proteobacteria, members of the genus Pseudomonas. Thauera is not very well studied, but is typically a denitrifier, using nitrate as the teerminal electron acceptor for respiration. It is know to be involved in the degradation of aromatic compounds, but would generally do so anaerobically (using nitrate instead of oxygen). This would seem, then, to be a novel species of Thauera, and the authors say they're trying to isolate it for further study.
Questions for thought:
- Proteorhodopsin seems to be a major form of phototrophhy in the ocean. How could this have been missed at these years?
- Do you think you could grow E.coli photochemotrophically by expressing in it the protorhodopsin and feeding it retinal? How would you know it's using the organics only for fixed carbon rather than energy? If you did this, would you have created a new species of Esherichia?
- Do you know of any other examples where the whole ecological niche of an organism is defined by genes it acquired horizontally?
- How would you go about trying to cultivate one of the members of the SAR11 group?
- Although rRNA seems to label much better than DNA with 13C in these feeding experiments, can you think of any reasons why it might be more useful to be able to isolate the DNA (rather than rRNA) of organisms that can eat the labeled substrate?
- Do you see any other bands in the DGGE gels that might represent less abundant phenol eaters?
- How could you make adjustments in the RT-PCR to make them more quantitative?
- Can you think of systems in which adding a labeled substrate and assuming the rRNAs that get labeled represent the organisms that use the substrate directly might be mistaken?
- How would you go about trying to cultivate one of the members of this group?