Cooperation and Competition in the Evolution of ATP-Producing Pathways

See allHide authors and affiliations

Science  20 Apr 2001:
Vol. 292, Issue 5516, pp. 504-507
DOI: 10.1126/science.1058079

This article has a correction. Please see:


Heterotrophic organisms generally face a trade-off between rate and yield of adenosine triphosphate (ATP) production. This trade-off may result in an evolutionary dilemma, because cells with a higher rate but lower yield of ATP production may gain a selective advantage when competing for shared energy resources. Using an analysis of model simulations and biochemical observations, we show that ATP production with a low rate and high yield can be viewed as a form of cooperative resource use and may evolve in spatially structured environments. Furthermore, we argue that the high ATP yield of respiration may have facilitated the evolutionary transition from unicellular to undifferentiated multicellular organisms.

Heterotrophic organisms obtain their energy by the degradation of organic substrates into products with lower free energy. The free energy difference between substrate and product can in part be conserved by production of ATP and in part be used to drive the degradation reaction. The maximal ATP yield is obtained if the entire free energy difference is conserved as ATP. However, in this case the reaction is in thermodynamic equilibrium, and therefore the rates of substrate degradation and ATP production vanish. If some of the free energy difference is used to drive the reaction, the rate of ATP production increases with decreasing yield until a maximum is reached. Hence, for fundamental thermodynamic reasons there is always a trade-off between yield (moles of ATP per mole of substrate) and rate (moles of ATP per unit of time) of ATP production in heterotrophic organisms (1–4).

A trade-off between yield and rate of ATP production is also present in sugar degradation by fermentation and respiration. In the presence of oxygen and sugars, many organisms are in principle capable of using both pathways to produce ATP. Because the ATP production rate of respiration is rapidly saturated at high levels of resource or limited oxygen supply (5–8), these organisms can choose, at least in the evolutionary sense, to increase the rate of ATP production by using fermentation in addition to respiration. However, because the yield of fermentation is much lower than that of respiration (2 mol versus about 32 mol of ATP per mole of glucose), the use of fermentation in addition to respiration increases the rate of ATP production at the cost of a lower total yield.

If energetic limitation is an important factor for organisms in their natural environment, we then expect that the properties of ATP-producing pathways have been under strong selection pressure during evolution. The existence of a trade-off between yield and rate of ATP production leads to the following question: Under what conditions is it favorable to use a pathway with high yield but low rate, as opposed to a pathway with low yield but high rate? A cell using a pathway with high yield and low rate can produce more ATP (and thus more offspring) from a given amount of resource. However, this advantage disappears when the cell is in resource competition with cells that produce ATP at a higher rate but a lower yield. While only those cells that consume the resource more rapidly benefit from the higher rate of ATP production, all competitors exploiting the resource share the consequences of the more rapid resource exhaustion (9).

The competition between cells with different properties in ATP production can be illustrated with a simple population biological model. Assume that a resource S is produced at a constant rate v and is consumed by n different strains of cells, Ni , at a rate ofJ i S(S) per cell. We assume that the growth rates of the strains are energetically limited and proportional to the rate of ATP production, cJi ATP(S), wherec is a proportionality constant (10, 11). Finally, we assume that the death rate d of all strains is the same. Thus, the population dynamics are given byEmbedded Image Embedded Image(1)If there is only one strain (n = 1), then the steady-state level of the population is given byN 1 =y 1 vc/d, where the yield of ATP production, y 1 =J 1 ATP/J 1 S, is given by the ratio of the rates of ATP production and resource consumption. Hence, the population size N 1depends on the yield but not the rate of ATP production (12). If several strains (n > 1) compete for the resource, the outcome of competition is determined by the highest growth rate, and hence by the highest rate of ATP production. This implies that a cell population using a pathway with low yield but high rate can invade and replace a cell population using a pathway with high yield and low rate (4). However, the invading cells eventually establish a smaller population because of the lower ATP yield. The resulting evolutionary dilemma is analogous to the “tragedy of the commons,” a framework widely used in evolutionary game theory to describe evolution toward the inefficient use of a common resource (13). In energy metabolism, evolution may work to select the less efficient pathway if cells are in competition for a shared resource.

The evolutionary dilemma arises only if there is competition for shared resources. Thus, according to our hypothesis, pathways with high rate but low yield of ATP production (such as fermentation) should primarily be observed in association with the exploitation of external resources. Indeed, microorganisms growing on external sugar resources, such asSaccharomyces, Mucor, andLactobacilli, which are present in the early phases of decomposition of organic material, use fermentation for ATP production even in the presence of oxygen (7, 14, 15). In contrast, organisms metabolizing internal resources such as ingested food items are expected to use pathways with high yield but low rate of ATP production. In line with our hypothesis, higher animals indeed mostly use respiration to produce ATP from ingested sugar resources. Fermentation is observed only in situations where very high rates of ATP production are essential, as for example in muscle cells.

Further interesting differences in ATP production between microorganisms and higher organisms are manifest in the specific design of oxidative phosphorylation. The P/O stoichiometry (moles of ATP produced during the oxidation of NADH, the reduced form of nicotinamide adenine dinucleotide) in mammalian mitochondria (rat liver) is 2.5 or higher (16–18). It has been reported that this value does not maximize the rate of ATP production, but rather trades a lower rate against a higher yield, as is expected for higher organisms metabolizing internal resources (1, 3). In yeasts, such as S. cerevisiae and Candida utilis, lower stoichiometries have been reported [1.5 and 2.0, respectively (7, 19)], suggesting that in these organisms the stoichiometries of oxidative phosphorylation are adjusted to produce ATP at a higher rate. Moreover, it has been reported that fermentation is designed to maximize the rate of ATP production (2, 20). This is in agreement with our hypothesis, because fermentation should be used in addition to respiration only when high rates rather than high yields of ATP production are required. Finally, further evidence suggestive of the relevance of trade-offs between rate and yield in natural systems stems from the observation that certain organisms, such as Escherichia coli and S. cerevisiae, appear to be capable of regulating the stoichiometry of oxidative phosphorylation, and thus ATP yield and rate, physiologically (21, 22). This could enable these organisms to adjust rate and yield of ATP production to varying environmental conditions.

One way to benefit from the high ATP yield of respiration is to avoid competition by cooperating with the other consumers of a shared resource. The exclusive use of respiration in the cells of a multicellular organism can be interpreted as such cooperation. In contrast to most other cells in a multicellular organism, tumor cells often use fermentation (23, 24). Thereby they may gain a selective advantage when in competition for energy resources with other cells of the organism. Thus, tumor cells can be viewed as leaving the realm of cooperation in a multicellular organism.

Depending on the ecological properties of the habitat, cooperative resource use could also evolve in populations of unicellular organisms. The evolution of cooperation is generally facilitated in spatially structured environments (25, 26). To illustrate the effect of spatial structure on the evolution of cooperative resource use, we extend the above model to a spatial model including diffusion of cells (at a rate DN ) and of resource (at a rate DS ),Embedded Image Embedded Image(2)where ∇2 is the Laplace operator. In Fig. 1 we show a simulation of spatial resource competition between an exclusively respiring strain (respirator) and a strain that uses fermentation in addition to respiration (fermenter). The simulations are based on the following assumptions: (i) The rate of resource influx v is stochastic in space and time. (ii) The rates of resource consumptionJ i S(S) are saturating functions of the resource S. (iii) At high levels of resource, the fermenters have a considerably higher rate of resource consumption. (iv) Both fermenters and respirators have the same rate of ATP production by respiration, but the fermenter additionally produces ATP by fermentation. To account for discrete strain population size, we assigned integer values to Ni . [For details of the simulation, see (4, 27).]

Figure 1

Competition between respirators (blue) and fermenters (red) in a spatially structured environment. (A) A typical simulation of the spatial model (Eq. 2). The green line represents the resource level. In the first 50,000 time steps, only fermenters are present. Respirators are then introduced into the system at low frequency. In this simulation, respirators eventually outcompete fermenters. Parameters: cell diffusion DN = 20, resource influx R = 15. See (4, 27) for further details. (B) A snapshot of the spatial distribution of fermenters and respirators during a typical simulation. Sites where both fermenters and respirators are present are depicted in yellow. Empty sites are black. Parameters: DN = 20,R = 15. (C) The outcome of competition as a function of resource influx and cell diffusion rate. High rates of resource influx R and cell diffusionDN favor fermenters. Despite increasing resource influx, the population size decreases when fermenters outcompete respirators.

The simulations shown in Fig. 1 illustrate two points. First, as the rate of resource influx v increases, fermenters progressively outperform respirators. This is because higher resource influx rates lead to increased competition, essentially because the population density increases, which in turn makes it more likely that respirators suffer from local competition with fermenters. Second, at higher cell diffusion rates, fermenters perform better than respirators. This is because any metabolic type, whether a respirator or a fermenter, performs worse when in competition with surrounding fermenting cells, which rapidly exhaust the resource locally. At low cell diffusion rates, cells tend to be surrounded by their own metabolic type. As a consequence, fermenters perform worse. Thus, the spatial structure resulting from low diffusion rates (of cells and resource) gives fermenters a disproportionately higher selective disadvantage. As the diffusion rates of cells or resource increase, cells increasingly compete globally for the same resource, and in the limiting case, fermenters outcompete respirators as in the nonspatial model (Eq. 1).

The correlation between the exclusive use of respiration and spatial aggregation, which appears in our simulations, can be observed in dimorphic fungi. These microorganisms occur in two different phenotypes: a mycelial, multicellular form and a yeast-like, unicellular form. Interestingly, these phenotypes differ in their sugar metabolism. Mucor racemosus, for example, uses fermentation in the unicellular stage but respiration in the multicellular stage (14). In other Mucor species it was shown that fermentation is strictly associated with the yeast-like form, because inhibition or knockout of the respiration pathway results in the suppression of the mycelial form (28, 29). Furthermore, in agreement with the spatial model, fermentation is used at high glucose concentrations, whereas the respiring phenotype is observed at low glucose concentrations.

Presumably the best studied facultatively multicellular organisms are Myxococcus and Dictyostelium. Both feed almost exclusively on nonfermentable resources. Therefore, a trade-off between yield and rate resulting from the use of fermentation and respiration does not appear to exist in their natural habitat. However, it is possible that the trade-off is manifest in the specific design of the respiratory pathway. A testable prediction of our hypothesis would be that these organisms evolved a P/O stoichiometry that trades maximal rate against higher yield of ATP production in the multicellular stage.

The fact that the switch from fermentation to respiration in dimorphic fungi is tightly coupled to the transition from the yeast-like to the mycelial form suggests that the high ATP yield of respiration could have played a key role in the evolutionary transition of heterotrophic organisms from single cells to early, undifferentiated forms of multicellularity. In heterotrophic eukaryotes, aerobic respiration evolved after the rise of oxygen in the atmosphere by symbiosis with respiring bacteria, the ancestors of present-day mitochondria (30). Respiration not only allowed these organisms to produce ATP with higher yield, but also enabled them to use previously unexploitable, nonfermentable resources for ATP production. Multicellularity became abundant after the evolution of respiration and has probably evolved independently more than 10 times (31–33). The simulations of the spatial model suggest that after respiration had evolved, aggregating cells may have benefited from respiration (and, equivalently, respiring cells may have benefited from aggregation). Thus, the high ATP yield of respiration, which is a reward of cooperative resource consumption, might constitute an evolutionary advantage to respiring multicellular organisms (34, 35). Importantly, this advantage also accrues to undifferentiated multicellular organisms, whereas most other evolutionary advantages of multicellularity arise from cell differentiation and specialization, which presumably appeared after the evolution of undifferentiated multicellular organisms.

  • * Present address: Experimental Ecology and Theoretical Biology, Eidgenössische Technische Hochschule (ETH) Zürich, CH-8092 Zürich, Switzerland.

  • To whom correspondence should be addressed. E-mail: bonhoeffer{at}


Stay Connected to Science

Navigate This Article