We had a very nice seminar from Uwe Sauer last week, in celebration of which I thought I would write about one of his papers. Uwe would like to understand how metabolism is controlled, and as a result has done a great deal of work to develop ways to measure metabolic flux. A recent paper (Haverkorn van Rijsewijk et al. 2011, Large-scale 13C-flux analysis reveals distinct transcriptional control of respiratory and fermentative metabolism in Escherichia coli Mol. Sys. Biol. 7 477 doi:10.1038/msb.2011.9) describes the application of some of these technologies to ask to what extent metabolism of galactose and glucose by E. coli is controlled by transcription.
Chemically speaking, glucose and galactose are about as closely related as two compounds can get: they’re diastereomers, which means that they have the same numbers of C, H and O atoms, in almost the same arrangement. The only difference is in the conformation of one stereocenter. But from a bacterial point of view, they evidently look very different. If you grow two cultures of E. coli on these two different sugars, the culture that gets the glucose will grow about 3x faster than the culture that gets the galactose. Consistent with this, Haverkorn van Rijsewijk et al. (henceforward HvR et al.) find that the amount of sugar taken up by the bacteria is about 4x higher in the glucose culture than the galactose culture.
Now, the first thing that happens to galactose once it’s inside the cell is that it’s converted into glucose. You would think that this would be the end of the story of the differences between the two sugars: one is taken up less enthusiastically than the other, rather as a US shopkeeper might accept Euros more reluctantly than US dollars, but after a quick trip to an international exchange counter at a nearby bank or airport everything’s fine. But using 13C-labeled sugars and gas chromatography/mass spectroscopy to quantitatively track where the atoms in the sugar molecules go, HvR et al. show that actually the downstream fates of the two sugars are quite different from each other. In both cases, the flow of 13C atoms goes from the input sugar to glucose-6-phosphate, fructose-6-phosphate, fructose bisphosphate, and is then split into two triose-phosphate molecules, in the familiar glycolysis pathway. Up until this point, the flow diagrams for the two sugars hardly differ, once you’ve taken into account the fact that there’s much less of the galactose flow. From the trioses, the main flow goes to phosphoenol pyruvate, pyruvate and acetyl-coA. And here the pathways start to look dramatically different: instead of going into the tricarboxylic acid (TCA) cycle, as we might expect, over half the flow from glucose goes from acetyl-coA into acetate, which is secreted and lost to the cell. In contrast, almost none of the flow from galactose goes into acetate, though it doesn’t exactly follow the canonical TCA cycle either; most of it takes a short-cut through the glyoxylate shunt.
Is it surprising that acetyl-coA has different fates in these two different settings? Maybe not — although the molecule is the same, the bacterium using it is in quite a different nutritional state. To test whether transcriptional control is involved in shaping the flow of metabolite molecules, the authors took 91 E. coli mutants, each lacking a single gene that’s important in transcription (81 transcription factors, and 10 sigma factors, including all those already known to affect the central enzymes of metabolism) and grew each of them on either galactose or glucose. They then analyzed both the absolute intracellular flux of metabolites, and the normalized flux relative to sugar uptake. Normalization helps to differentiate effects that cause changes in the actual uptake of the sugar (and thus cell growth) from changes in the behavior of the downstream metabolic pathway.
The authors found that although about 2/3 of the 91 mutants showed changes in sugar uptake, none of them showed changes in the relative flux in the upper part of the pathway, from glucose-6-phosphate to the triose-phosphates. If there is transcriptional control in this part of the pathway, it’s impossible to distinguish it from the changes due to altered uptake and growth. But in the lower part of the cycle, from phosphoenolpyruvate down, the presence or absence of specific transcription factors make a big difference to the patterns of flux. The key control point is acetyl-coA, and the regulators are different depending on the sugar input. When the bacteria are grown on glucose, HvR et al. identified 9 transcription factor knockouts that showed significantly altered flux (relative to wild type) from pyruvate into either the TCA cycle or acetate. Five of these make sense given what’s known about the targets of these transcription factors (which is always reassuring), while four of them are currently unexplained and so might lead to new knowledge (although they could act indirectly). When the cells were grown on galactose, however, only 5 mutants showed significantly changed flux. In the case of galactose, there are three directions the atoms can go: into the TCA cycle, acetate, or the glyoxylate shunt. All five of the mutants that cause significant changes in flux on galactose essentially abolish the flux through the glyoxylate shunt.
This is interesting because the Sauer lab has previously shown that cyclic AMP (acting through the cAMP-receptor protein, CRP) can act to push metabolite flux through the glyoxylate shunt. But the new mutants found here aren’t in transcription factors sensitive to cyclic AMP. Two of them, however, are in transcription factors that affect the glucose transport system, the system ultimately responsible for the preferential use of glucose. This system is dependent on phosphoenol pyruvate (PEP), the transition point between the upper and lower metabolic cycles. So, is there a link between sugar uptake and the downstream behavior of the glyoxylate shunt? The authors looked — and found that all 5 mutants had much higher galactose uptake rates than wild type. Knocking out the glucose transport system led to higher galactose uptake (even though no glucose was available), and at the same time knocked out the flux through the glyoxylate shunt.
After some further detective work, HvR et al concluded that galactose uptake is actively repressed in wild-type cells. Knocking out any of the 5 transcription factors they identified relieves this repression, and this in turn causes inhibition of the glyoxylate shunt. Their evidence includes the observation that simply manipulating the rate of uptake of either glucose or galactose (using a chemostat to change the availability of the sugar) changes the flux through the glyoxylate shunt. Glucose is derepressed to start with, but if you only allow the cell to take up a small amount of glucose, the flux through the glyoxylate cycle is activated. The effect of varying sugar uptake is transduced to the glyoxylate shunt via the same cyclic AMP/CRP signal the Sauer lab identified earlier: in the mutants that showed greater galactose uptake, cyclic AMP levels were significantly decreased.
The authors rationalize all this by pointing out that when sugar uptake is low and growth is slow, the cell has a problem: it is making NADPH faster than it can use it. Slow sugar uptake is signalled to the rest of the cell by reduced cyclic AMP levels, and this has the effect of turning on the glyoxylate shunt, which produces less NADPH than the TCA cycle does and thus helps to keep the cell in balance for NADPH production and use. The cyclic AMP-mediated switch to the glyoxylate shunt therefore helps to decouple carbon catabolism from NADPH production.
The oddest thing, though (and this is something HvR et al. don’t explain) is the fact that galactose uptake is distinctly lower than optimal. The authors checked the growth rates of the mutants that take up galactose better, and find that indeed they grow faster than the wild type when growing on galactose. Clearly, there are many biological mysteries left even in the well-trodden ground of bacterial metabolism.
[Update: When Uwe mentioned this paper in his seminar (I just found my notes), he emphasized that it shows how little overall effect transcription has on steady-state flux in E. coli metabolism. Yes, there are effects — and probably when you're changing from one state to another, the effects are very important — but in this study, translational control of flux is essentially limited to one branch point (acetyl-coA to x or y), and even there the main effect is through modulating sugar intake. Yet steady-state fluxes are pretty stable, and thus are probably tightly controlled.]
Haverkorn van Rijsewijk, B., Nanchen, A., Nallet, S., Kleijn, R., & Sauer, U. (2011). Large-scale 13C-flux analysis reveals distinct transcriptional control of respiratory and fermentative metabolism in Escherichia coli Molecular Systems Biology, 7 DOI: 10.1038/msb.2011.9