These pages contain the full text of the following chapter: Jan-Hendrik S. Hofmeyr (1997) "Anaerobic energy metabolism in yeast as a supply-demand system", pp. 225-242 in New Beer in an Old Bottle: Eduard Buchner and the Growth of Biochemical Knowledge (ed. A. Cornish-Bowden), Universitat de València, Valencia, Spain
A century ago Eduard Buchner (1897) made the remarkable discovery that a cell-free soluble fraction of yeast could still convert glucose to ethanol, in other words, cell-free systems can still metabolize. Now, a hundred years later, we have a highly detailed picture of the glycolytic pathway, its enzymes and the mechanisms that regulate them. Can we claim a full understanding of what controls the carbon flux through glycolysis and the role of the mechanisms that regulate it? Reading any modern biochemistry textbook one gains the distinct impression that the answer is yes. In a nutshell, the story [as told in Stryer (1995), a popular representative of the genre] proceeds by identifying hexokinase, phosphofructokinase, and pyruvate kinase as potential sites of control, due to the fact that they catalyse "essentially irreversible reactions"; then all the "regulatory properties" of these enzymes are described in terms of allosteric effects, covalent modification by phosphorylation, and transcriptional control to cover regulation in the time-scale from seconds to hours. Phosphofructokinase, being the true committing step of glycolysis, is identified as the most important "pacemaker" and regulatory enzyme, responding both to energy charge through the ATP/AMP ratio and to biosynthetic precursors. The other two kinases, hexokinase and pyruvate kinase, however, "also set the pace of glycolysis".
This description sounds familiar and is deemed satisfactory by many because it conforms fully to what we shall call the "classical" view of metabolic regulation, where fluxes are controlled by "rate-limiting" steps which, in turn, are regulated by one or more of a number of molecular mechanisms. It usually seems taken for granted that increases in the activities of one or more of the rate-limiting steps of glycolysis would lead to an increase in flux to ethanol. Tests of this prediction have been made possible by the development of recombinant-DNA technologies that allow the quantitative manipulation of the in vivo enzyme concentration profile of cells.
Appropriately, one of the first of such tests was done on yeast glycolysis. Schaaff et al. (1989) over-expressed eight different enzymes of the Saccharomyces cerevisiae glycolytic and fermentative pathway by placing their genes on multicopy vectors; in so doing they were able to increase the specific enzyme activities between 3.7 and 13.9-fold. They over-expressed these enzymes singly or in pairs (phosphofructokinase/pyruvate kinase or pyruvate decarboxylase/ alcohol dehydrogenase) and measured the effect on the rate of ethanol production and the level of a number of metabolites in logarithmically growing cultures. In terms of the prevailing view that glycolytic flux-control resides in glycolysis itself the results were surprising: under no circumstances did increases in the activities of the different glycolytic enzymes significantly affect the rate of ethanol production ("and therefore the flux through glycolysis") or affect the concentrations of "key" metabolites in comparison to the wild type.
If flux-control does not reside in the "rate-limiting" enzymes, where does it reside? Schaaff and colleagues conclude that their data support the view of metabolic control analysis (Kacser et al., 1995; Heinrich and Rapoport, 1974) that the control of flux is shared among all enzymes of a metabolic system, a concept for which in general there is now ample evidence (Fell, 1996). What is not clear is what they regard as their "metabolic system". Those trained to take the classical view would in all probability regard only the glycolytic pathway as their system and argue that even if flux-control does not reside in one or even in a few glycolytic enzymes it must still be shared among all the glycolytic enzymes, the implication being that if one simultaneously over-expressed all the enzymes of glycolysis one would theoretically expect the glycolytic flux to increase proportionally. Transport of glucose into the cell should of course also be considered part of the glycolytic apparatus; in fact, it has been suggested that substantial control on glycolytic flux is exerted by the uptake systems (Galazzo and Bailey, 1990; Bisson et al., 1993).
In the remainder of this chapter I shall argue for another possibility, namely that steady-state glycolytic flux is controlled by reactions outside of what has traditionally been regarded as glycolysis; more specifically, it is controlled by those cellular processes that consume the key product of glycolysis, ATP. Furthermore, I shall argue that this is fully in accordance with the widely-accepted view that the properties of metabolic systems have evolved to fulfill one or more functions.
Where does a metabolic pathway begin and where does it end? We instinctively draw boundaries around pathways in order to simplify description of the metabolic tangle on the charts that adorn many a classroom and laboratory wall. The traditional textbook description of a metabolic pathway such as fermentative glycolysis in yeast is based on the net reaction
glucose + 2ADP + 2Pi –> 2 CO2 + 2ATP
There is, however, an important difference between glucose, ethanol, and CO2 on the one hand, and ADP and ATP on the other. The first are "external" substrates and products in the sense that they are ingested and excreted; they form natural metabolic boundaries. The second are "internal" substrates and products; they form artificial metabolic boundaries at which the metabolic pathway connects with other cellular processes. So, what we call an "end-product" is often the substrate for other cellular processes, e.g. ATP for biosynthesis and growth, amino acids for protein synthesis, nucleotides for nucleic acid synthesis. In principle there is of course nothing wrong with compartmenting a complex network into separate modules for purposes of analysis, whether experimental or theoretical. A problem arises, however, when the internal boundaries are carried over into an analysis of control and regulation. The reason is that while one or more steps in any module may exercise a high degree of control of some steady-state variable in the module when it is studied in isolation, it is possible that they lose that control completely when the system is expanded to include other connecting modules; this will happen when the module as a whole has little control over that variable within the expanded system (that this is a fundamental result from metabolic control analysis has not exempted control analysts from sometimes making the same mistake). The surprise caused by the failure of over-expressed glycolytic enzymes to affect flux can be largely ascribed to an unfulfilled expectation that the control profile obtained for the glycolytic module in isolation can be extrapolated to the whole system of which glycolysis is part.
That the mistake of extrapolating an intramodular control profile is often made is strange in view of the seemingly general acceptance of the functional view of metabolism (to which I fully subscribe), i.e. the view that the kinetic and regulatory properties of the enzymes of intermediary metabolism have been moulded by evolution in such a way as to allow metabolic pathways to fulfil one or more "metabolic functions". Taken seriously, this view should at least ensure that the properties of a pathway are always related to the greater cellular context. The functional view is, however, fraught with difficulties: it is one thing to accept that metabolic pathways have been "purposefully designed" by evolution; it is quite another to decide for which function(s) a particular pathway has been designed, a question that remains mostly a matter of informed guessing by reverse engineering. Nevertheless, there seems to be general consensus that the function of yeast glycolysis under anaerobic growth conditions is to maintain the cell’s energy charge at a high level (maintain the adenylate pool mostly in the form of ATP) while supplying carbon skeletons for biosynthesis mostly from pyruvate and phosphoenolpyruvate.
The analogy of the cell as a "chemical factory", developed by Cascante and Martí in this book (pp. 199–-214), leads naturally to an economic view that cellular processes consist of supply-demand or production-consumption systems. The acceptance of this analogy or terminology is not essential for the type of analysis to be described (one could just describe it as a modular analysis), but it does emphasize my own conviction that any analysis should take into account the perceived function. I shall thus analyse glycolysis as a feedback-regulated ATP-supply system for other cellular processes that consume ATP (collectively called the "demand"). Of course the introduction of such a terminology also has a persuasive element: one expects a well designed feedback-regulated supply-demand system with as main function the production of some important end-product to supply that product at a rate reflecting the demand; therefore, one expects most, probably all, of the flux-control to be associated with the collective demand for the end-product. Nevertheless, this expectation can be fully tested by experiment and need never be accepted unquestioningly.
Under anaerobic, fermentative conditions in yeast, glycolysis serves as the only net producer of ATP (it also, of course, consumes ATP in order to prime its substrates). Its key enzymes are sensitive to the composition of the adenylate pool consisting of ATP, ADP and AMP so that glycolysis is a feedback-regulated system. ATP is consumed by various demand processes: biosynthesis (if the yeast grows), active transport, maintenance of intracellular ionic composition, etc. This ATP supply-demand system is depicted in Fig. 1. The steady-state fluxes through the different glycolytic enzymes that consume or produce ATP are not necessarily strictly coupled. Each of these four enzymes occurs in a unique glycolytic sequence delimited by branchpoints. Hexokinase and phosphofructokinase are separated by the branches from glucose 6-phosphate to trehalose and to pentose phosphates, phosphofructokinase and phosphoglycerate kinase by the branch from dihydroxyacetone phosphate to glycerol, phosphoglycerate kinase and pyruvate kinase by the branch from phosphoenol pyruvate to biosynthesis. Only under non-growing conditions are the fluxes through most of these branches from glycolysis small enough for there to exist a stoichiometrically coupled "glycolytic flux" that can be measured by either the rate of glucose consumption or the rate of ethanol production (which should be 1:2 on a concentration.time–1 basis). If these conditions are not satisfied, the individual steady-state fluxes through the different sections of glycolysis should be determined. For the purpose of the following analysis we assume the experimentally obtainable situation where a single glycolytic flux can be measured. The role of adenylate kinase is considered below, but note that it is highly active and catalyses an equilibrium reaction that carries no net flux.
Fig. 1. The main reactions involved in ATP production and consumption in a fermenting yeast cell. Abbreviations: HK: hexokinase; PFK: phosphofructokinase; PGK: phosphoglycerate kinase; PK; pyruvate kinase; AK: adenylate kinase. The reaction catalysed by adenylate kinase is depicted with a dotted line to indicate that it is considered to be in equilibrium, therefore carrying no net flux. The number associated with the adenylate kinase reaction indicates reaction stoichiometry. The block designated "Demand" symbolizes the set of non-glycolytic ATP-consuming reactions.