Author response: Flagellar energy costs across the tree of life

Article Figures and data Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Flagellar-driven motility grants unicellular organisms the ability to gather more food and avoid predators, but the energetic costs of construction and operation of flagella are considerable. Paths of flagellar evolution depend on the deviations between fitness gains and energy costs. Using structural data available for all three major flagellar types (bacterial, archaeal, and eukaryotic), flagellar construction costs were determined for Escherichia coli, Pyrococcus furiosus, and Chlamydomonas reinhardtii. Estimates of cell volumes, flagella numbers, and flagellum lengths from the literature yield flagellar costs for another ~200 species. The benefits of flagellar investment were analysed in terms of swimming speed, nutrient collection, and growth rate; showing, among other things, that the cost-effectiveness of bacterial and eukaryotic flagella follows a common trend. However, a comparison of whole-cell costs and flagellum costs across the Tree of Life reveals that only cells with larger cell volumes than the typical bacterium could evolve the more expensive eukaryotic flagellum. These findings provide insight into the unsolved evolutionary question of why the three domains of life each carry their own type of flagellum. Editor's evaluation This work demonstrates convincingly that energetic considerations (building costs versus potential benefit) must be taken into account to understand the evolution of flagella. It provides compelling evidence to the long-standing question of why bacteria, archaea, and eukaryotes evolved with different types of flagella. https://doi.org/10.7554/eLife.77266.sa0 Decision letter Reviews on Sciety eLife's review process eLife digest Most creatures on Earth are single cell organisms. The tree of life comprises three domains, two of which – bacteria and archaea – are formed exclusively of creatures that spend their existence as independent cells. Yet even eukaryotes, the domain which include animals and plants, feature single cell species such as yeasts and algae. Regardless of which group they belong to, all single-celled organisms must find food in their environment. For this, many are equipped with flagella, whip-like structures that protrude from the cell and allow it to swim. In fact, archaea, bacteria and eukaryotes have all independently evolved these structures. However, flagella are also expensive for an organism to build, maintain and operate. They are only worth having if the advantages they bring to the cell compensate for their cost; many single-cell species do not carry flagella and obtain their food without having to swim. To explore this trade-off, Schavemaker and Lynch calculated the cost of building and using flagella for about 200 species across the tree of life. The analysis show that the amount of energy spent on flagella varied between 0.1% and 40% of the entire cell budget. This investment is only worthwhile if the cell is above a certain size. Smaller than this, and the organism is better off obtaining its food passively. The results also show that while eukaryotic flagella are much bigger and quite different than their bacterial counterpart, both appendages share the same patterns of cost effectiveness. However only eukaryotic cells, which are on average larger than bacteria, can afford to evolve such sizable and complex structures; making just one would cost more than the entire energy budget of a bacterial cell. Many single-cell species which are critical for the health of the planet are equipped with flagella, such as the microorganisms which recycle matter in the oceans and release carbon dioxide. Understanding the costs and benefits of flagella could explain more about this aspect of the carbon cycle, and therefore global warming. Introduction Life contains a dazzling diversity of molecular and cellular mechanisms. These are the consequence of, and are subject to, evolution. Though speculation abounds, there is yet no full integration of molecular and cellular biology with evolutionary theory (Lynch et al., 2014; Lynch and Trickovic, 2020). All features of the cell require energy for construction and operation, but they differ in their energy demand and contribution to fitness. Energy costs act as weights to evolutionary paths. If we want to know why only certain molecular and cellular features exist, we need an accounting of their baseline costs. Such an accounting has been under way for individual genes (Lynch and Marinov, 2015), membranes (Lynch and Marinov, 2017), and various other features (Lynch and Trickovic, 2020; Milo and Phillips, 2016; Raven and Richardson, 1984). Many single-celled species swim by virtue of their flagella. However, why the three domains of life – bacteria, archaea, and eukaryotes – each evolved their own type of flagellum is still a major open evolutionary question. The bacterial flagellum (Berg, 2003) consists of a helical protein filament, sometimes sheathed by a membrane (Geis et al., 1993), that is attached, via a hook region, to a basal body embedded in the cell envelope. The filament is assembled by progressive addition of monomers of the protein flagellin (FliC) to the tip, reached by diffusion through a central channel in the filament. Its rotation is driven by a proton or sodium gradient (Ito and Takahashi, 2017). The archaeal flagellum, or archaellum, is also a rotating flagellum but differs from the bacterial one in two key respects. The filament, lacking a central channel, is assembled from the base, and its rotation is driven by ATP hydrolysis (Albers and Jarrell, 2018). The eukaryotic flagellum, or cilium, is completely different from the other two. Its diameter is an order of magnitude larger (Khan and Scholey, 2018); it bends rather than rotates; and the ATP-dependent bending is caused by motor proteins arranged along the length of the flagellum. Most eukaryotic flagella have two single microtubules in the middle, surrounded by nine doublet microtubules. The microtubules are connected by protein complexes such as inner and outer dynein arms (King, 2016), nexin-DRC (Heuser et al., 2009), and radial spokes (Pigino et al., 2011). This whole protein complex, the axoneme, is always surrounded by a membrane. Like the bacterial flagellum, the eukaryotic axoneme is assembled at the tip, but with subunit delivery being facilitated by active intraflagellar transport (IFT) (Marshall and Rosenbaum, 2001). Flagellar construction and operating costs have been estimated for the bacterium Escherichia coli (Macnab, 1996) and a dinoflagellate eukaryote (Raven and Richardson, 1984). For E. coli, the construction and operating costs are 2% and 0.1%, respectively, relative to total cell energy expenditure (Macnab, 1996). Others found an operating cost of 3.4% (assuming 5 flagella, a 100 Hz rotation rate, and a 1 hr cell division time) (Ziegler and Takors, 2020). The relative construction cost for the dinoflagellate was estimated at 0.026%. The relative operating cost was 0.08–0.9%, compared to total cell maintenance cost (Lynch and Marinov, 2015; Raven and Richardson, 1984). Here, following on previous work (Lynch and Trickovic, 2020), we provide a fuller accounting of construction costs for model bacterial, archaeal, and eukaryotic flagella, and extend these results to a range of bacterial and eukaryotic species using information on flagellum number, flagellum length, and cell volume retrieved from the literature. Drawing from additional estimates of the operating costs of flagella, these results are discussed in the context of swimming speed, nutrient uptake, growth rate, scaling laws, effective population size, and the origin of the eukaryotic flagellum. Results Energy costs of flagellum construction Flagellar motility burdens the cell with costs of construction and operation. The former refers to the energy required for synthesis of the proteins and lipids that constitute the flagellum, which includes both direct and opportunity costs (Lynch and Marinov, 2015; Mahmoudabadi et al., 2019). The operating cost is the energy associated with rotating (bacteria and archaea) or beating the flagellum (eukaryotes). Starting from the framework previously outlined (Lynch and Marinov, 2015), and ignoring protein turnover (which is only a minor contribution to costs; Lynch and Marinov, 2015; Methods), we estimate the construction cost per protein component (in units of ATP hydrolyses) as: (1) Cprot=CAANpLp, where Np is the number of protein copies, Lp is the length of the protein in amino acids, and CAA is the average energy cost of an amino acid (29 ATP; direct + opportunity costs) (Lynch and Marinov, 2015). When only the volume of a protein is known, we calculate the number of amino acids using the average volume per amino acid (1.33 × 10–10 µm3; Methods). Eukaryotic flagella, and those of some bacteria, are enveloped by a membrane. Previously established approaches (Lynch and Marinov, 2017), with the inclusion of membrane proteins, leads to the membrane construction cost: (2) Cmem=2CLAtot(1−fprot)Aliphead+CAAtmprotAtotfprotVAA. For the first term, CL is the average energy cost of a single lipid molecule, Atot is the total membrane surface area, which includes both lipids and membrane proteins, Aliphead is the cross-sectional area of a lipid head-group, fprot is the fraction of membrane surface area occupied by protein (which is assumed to be 0.4) (Lindén et al., 2012), and the factor 2 accounts for the bilayer of lipids. For the second term, CAA is the energy cost per amino acid, tmprot is the thickness of the protein areas of the membrane (8 nm), and VAA is the average volume per amino acid. Equations 1 and 2 yield estimates of absolute construction costs, in number of ATPs. Relative construction costs are obtained by dividing the absolute flagellum costs by the construction cost of the entire cell obtained from Lynch and Marinov, 2015. For a discussion of assumptions and omitted costs, see Methods. The flagellar construction cost data for the three model organisms are summarised in Table 1. Table 1 Energy costs of flagella in the three domains of life. BacteriaArchaeaEukaryotesSpeciesEscherichia coliPyrococcus furiosus*Chlamydomonas reinhardtiiConstruction cost per µm (ATP)3.02 × 1071.28 × 1072.80 × 109Construction cost per flagellum (ATP)2.32 × 1082.15 × 1073.08 × 1010Number of flagella per cell3.4502Construction cost of all flagella (ATP cell–1)7.88 × 1081.07 × 1096.15 × 1010Cell volume (µm3)1.00.22122Cell division time (hr)1.01.09.15Total cost of cell (construction + operating; ATP)1.59 × 10106.50 × 1095.32 × 1012Relative construction cost, all flagella (%)5.016.51.4Operating cost per flagellum (ATP s–1)6.6 × 1042.64 × 102 †9.7 × 105Operating cost per cell cycle, all flagella (ATP)8.08 × 1084.75 × 107 ††6.39 × 1010Relative operating cost, all flagella (%)5.20.73 †1.2Relative total cost, all flagella (%)10.217.2 †2.6 * Due to gaps in knowledge of P. furiosus flagella, some data were taken from other archaea (see main text). † This estimate for the operating cost is probably too low as the flagellar rotation rate that it is based on was recorded with a bead attached, which slows down flagellar rotation. Table 1—source data 1 Breakdown of flagellar construction costs for bacteria, archaea, and eukaryotes. https://cdn.elifesciences.org/articles/77266/elife-77266-table1-data1-v1.xlsx Download elife-77266-table1-data1-v1.xlsx Bacteria – cost of the E. coli flagellum For a detailed examination of the energy costs of the bacterial flagellum, we chose that of E. coli. The protein composition of the flagellum, determined by a combination of structural and biochemical work, is summarised in Berg, 2003. We supplemented this with copy numbers for export-apparatus proteins (Fukumura et al., 2017; Minamino, 2014; Minamino, 2018) and verified the FliC copy number (Namba et al., 1989). Using protein lengths and copy numbers, we calculated the energy costs for each protein per flagellum, and by summing these, obtained the cost for the whole flagellum. The total cost of a single 7.5 µm long (Turner et al., 2000) flagellum is 2.32 × 108 ATP. A large fraction of this cost, 0.99, is in the filament (including the hook). With an average number of 3.4 flagella per E. coli cell (Harshey and Matsuyama, 1994; Turner et al., 2000), the total cost of flagella is 7.88 × 108 ATP cell–1. Because bacterial flagellum rotation is driven by the flow of protons from one side of the membrane to the other, an estimate of the operating cost requires information on the number of protons crossing the membrane, through the flagellum, per unit time. The energetic costs of this can be expressed in terms of the number of ATP hydrolyses by using the proton/ATP ratio in the ATP synthase, which is 3.33 (Jiang et al., 2001). The maximum number of stators, per flagellum, is at least 11 (Reid et al., 2006), each of which has two channels (Braun and Blair, 2001), and each channel passes 50 protons per revolution (Gabel and Berg, 2003; Samuel and Berg, 1996), leading to an estimated 1100 protons per revolution, which is similar to the value determined experimentally for Streptococcus, 1240 protons per revolution (Meister et al., 1987). The E. coli flagellum can spin at a maximal rate of 380 Hz (Chen and Berg, 2000; Gabel and Berg, 2003). The mean swimming speed for Salmonella, a close relative of E. coli, occurs at a rotation rate of 150 Hz (Magariyama et al., 2001). Taking a rotation rate of 200 Hz, a cell division time of 1 hr, 3.4 flagella per cell, and assuming continuous operation we obtain a total operating cost of 1100 × 200 × 3.4 × 3600/3.33 = 8.08 × 108 ATP cell–1. For many evolutionary considerations, the cost of a structure/function relative to the cost of an entire cell is of interest. For E. coli, the construction (or growth) and operating (or maintenance) costs of the cell are estimated to be 1.57 × 1010 and 2.13 × 108 ATP hr–1, respectively (Lynch and Marinov, 2015), and assuming a cell division time of 1 hr implies a total cell energy cost of 1.59 × 1010 ATP. Thus, the costs of constructing and operating flagella relative to the whole-cell energy budget are 5.0% and 5.2%, respectively. The operating cost is close to an earlier estimate of 3.4% but differs from the other estimate of 0.1% (see Introduction). How the 0.1% was obtained is not clear from the source. Comparing the flagellum operating cost to just the cell operating cost leads to an apparent contradiction. The flagellar operating cost exceeds the cell operating cost by 3.8-fold. This mismatch could be due to the variability of E. coli cell volume (Taheri-Araghi et al., 2015) or the fact that the flagella are not continuously rotating. The cell operating cost may also have been determined under conditions in which cells do not swim. Temperature is probably not an issue as the whole-cell operating costs are normalised to 20°C (Lynch and Marinov, 2015), and the flagellum rotation speeds are determined at 23–24°C (Chen and Berg, 2000; Gabel and Berg, 2003). Archaea – cost of the Pyrococcus furiosus flagellum For the archaeal flagellum, we use the reasonably complete structural information on the flagellum of P. furiosus, with some additional data from Sulfolobus acidocaldarius (Daum et al., 2017); see also Albers and Jarrell, 2018. The major filament component in P. furiosus is FlaB0 (Nather et al., 2014), with 1852 copies per µm of flagellum length (Daum et al., 2017) (extracted from PDBID: 5O4U). FlaB0 is chemically modified with oligosaccharides (35 sugar units per monomer) (Daum et al., 2017; Fujinami et al., 2014). We assume that each sugar unit costs as much energy as a single glucose in E. coli, which is 26 ATP (Mahmoudabadi et al., 2019). The length of the flagellar filament in P. furiosus is stated to be ‘a few 100 nm to several µm’ (Daum et al., 2017). We assume this means 0.3–3 µm, or 1.65 µm on average. Combining these data yields a construction cost of 2.15 × 107 ATP per flagellum. As in E. coli, a large fraction of this cost is in the filament, 0.98. The number of flagella per P. furiosus cell is ~50 (Daum et al., 2017), leading to a flagellar cost of 1.07 × 109 ATP cell–1. In addition to the flagella, P. furiosus has a polar cap that organises all flagella into a tuft. The polar cap is rectangular and has a thickness of 3 nm and a linear dimension of 200–600 nm (Daum et al., 2017). Assuming a square of 400 × 400 nm implies a volume of 4.8 × 10–4 µm3. We also determined the total volume of a hexagonal array of protein complexes that is attached to the polar cap. Combining these volumes with the volume per amino acid, we obtain a total polar cap construction cost of 2.43 × 108 ATP cell–1. The total cell cost is 6.50 × 109 ATP cell–1, which is obtained by applying the P. furiosus cell volume (Daum et al., 2017) to the regression in Lynch and Marinov, 2015, and assuming a 1 hr cell division time. The relative cost for flagella plus polar cap is 20.2% (16.5% for flagella alone). The rotation of the archaeal flagellum is driven directly by ATP hydrolysis, and 12 ATP are used for each rotation in Halobacterium salinarum (Iwata et al., 2019). The rotation rate is about 22 Hz (Iwata et al., 2019), but this may be an underestimate because in this measurement the flagellum was loaded with a 210 nm diameter bead, and in E. coli rotation rates decrease with an increased load (Gabel and Berg, 2003). We obtain an energy use of 264 ATP s–1 flagellum–1. Applying the H. salinarum numbers to P. furiosus, assuming a 1 hr cell division time (Nather et al., 2006), yields an operating cost of 264 × 50 × 3600=4.75 × 107 ATP cell–1. P. furiosus has a peculiar metabolism and can grow at 100°C (Kengen, 2017). We have not considered the impact of these differences on flagellar costs, but this should be investigated in the future. Eukaryotes – cost of the Chlamydomonas reinhardtii flagellum For the eukaryotic flagellum, we focus on C. reinhardtii, as its flagellum is very well characterised structurally. We include the axoneme, the membrane, and the IFT system. The axoneme contains a central pair of microtubules surrounded by nine doublet microtubules. Each microtubule doublet consists of α- and β-tubulins and about 30 additional proteins (Ma et al., 2019). Bound to the outside of each doublet are the radial spokes (Pigino et al., 2011), the radial spoke stub (Barber et al., 2012), the IC/LC complex (Heuser et al., 2012), the CSC (Dymek and Smith, 2007; Pigino et al., 2011), the Nexin-DRC (Bower et al., 2013), the MIA complex (Yamamoto et al., 2013), the tether (Heuser et al., 2012), the outer dynein arms (King, 2016; King and Patel-King, 2015; Ma et al., 2019), and finally the inner dynein arms (King, 2013; King, The central pair of microtubules of α- and and are by several other protein complexes et al., To obtain the cost of the membrane, we assume a with a diameter of µm (Khan and Scholey, 2018) and use with CL = ATP (Lynch and Marinov, 2017). we include the construction cost of IFT complexes per flagellum et al., Combining axoneme, membrane, and and assuming two flagella of 11 µm in length et al., we obtain a Chlamydomonas flagellar construction cost of × 1010 ATP cell–1. The operating cost of the Chlamydomonas flagellum determined experimentally on a axoneme without the is × ATP s–1 (Chen et al., 2015). This is similar to estimates for a eukaryotic flagellum, × ATP s–1 (Raven and Richardson, × ATP s–1 et al., assuming a total flagellum length of × µm 1—source data and × ATP s–1 (Chen et al., 2015). The Chlamydomonas operating cost for a full hr cell (Lynch and Marinov, 2015) is × 1010 ATP. The Chlamydomonas whole-cell construction and operating cost, at a hr cell division is × ATP (Lynch and Marinov, 2015), leading to the relative costs of constructing and operating both flagella of and respectively. Comparing the flagellar operating cost × 1010 to just the cell operating cost × ATP per cell reveals for E. coli, Chlamydomonas has energy in its operating budget to swim Flagellar construction costs across the tree of life Flagellar construction costs for a range of bacterial and eukaryotic species were obtained by the just determined flagellar construction costs of E. coli and Chlamydomonas with flagellum lengths and numbers, and cell volumes for species bacteria and For bacteria, we assume that all flagella are from of the same as those in E. coli amino the same for the other flagellar The absolute flagellar construction cost for bacteria of from × × ATP with a of × 108 ATP. The relative flagellar construction cost of from to with a of The absolute cost of flagella with bacterial cell but there is no relative costs and cell A to the bacterial absolute cost cell volume data yields cell volume in and absolute cost in number of ATP 1 with 2 see all Download costs of flagella in bacteria and eukaryotes as a of cell Each a species. The costs are the all flagella on a single cell. The absolute construction The are to bacteria, eukaryotic and and and The and respectively The construction costs of flagella relative to the construction costs of the entire cell. The is a to the eukaryotic The is in the In both the which in its flagellar but not belong to the Data in 1—source data 1. 1—source data 1 Table of flagellar and cellular for bacterial and eukaryotic species. Download For of the eukaryotes, we assume that flagella are of the same type as that of For three and we also include the that they carry in their flagella the axoneme and et al., which the cost per unit length by We also for all eukaryotes, that the flagellum the same cross-sectional area its length see Methods). The absolute flagellar construction cost for eukaryotes of from × × ATP with a of × ATP. The relative flagellar construction cost of from to with a of Eukaryotic and other a number of flagella, have absolute flagellum costs than the and This is even where cell volumes A to the eukaryotic absolute cost data has an of and a to the and has an of In the of relative construction cost cell eukaryotic are from the and Here, the eukaryotic a scaling with an of of eukaryotic flagellar and cellular is in 1. Discussion Flagellar costs and benefits To understand why species evolve and maintain flagella, and why different domains of life on very different of flagella, costs need to be compared to Here, benefits are analysed in three from an in swimming speed to nutrient to cell speeds were obtained from the literature for a of species from cost that swimming speed with cell volume all species are = When on just eukaryotic we find a swimming speed in more in the = in the work, using a different of an of for bacteria and eukaryotes and an of for eukaryotic (Lynch and Trickovic, 2020). with cell the swimming speed also with absolute flagellar construction cost = 2 Download The cost and benefits of flagellar Each a species. speed cell are all species from the cost for which the swimming speed is This includes bacteria, eukaryotic and The continuous is the to all The is a to the eukaryotic The for the entire speed the absolute flagellar construction The is the speed (in cell lengths per ATP of construction cost cell The is the to the bacterial The is the to the eukaryotic data and The are of the swimming or operating cost, calculated from with This an of the of energy into swimming The The relative growth rate as a of cell volume for cells in a with a of cells with flagella to cells without flagella. In all the which in its flagellar but not belong to the Data in data 1. data 1 Table with swimming swimming and cell growth Download we compared bacterial and eukaryotic flagella in their ability to swimming To allow for a direct comparison between the two swimming speed is expressed in cell lengths per second per ATP that bacterial and eukaryotic flagella a common that there is no large in cost-effectiveness between the two scaling with cell has been taken into However, data for each flagellar type or only a and volume the common be a range of cell volumes is but for the eukaryotic flagellum (see To a cell not only to a flagellum, but to it as We estimate operating costs in two The first which the to a of at a speed, through a with The second the ATP hydrolysis rate per µm of flagellum with the total flagellar length for each species only For the ATP hydrolysis rate, we used the average of the determined for and ATP s–1 see Comparing these two operating costs reveals that eukaryotes have a mean of for the of energy into swimming which is similar to earlier of swimming for single-celled organisms et al., and swimming speeds can be used to nutrient The of swimming for nutrient uptake, or the has been discussed by but without to the flagellar construction cost and However, the fitness also on it costs to and the flagella that a cell. To swimming the amount of that is obtained by swimming to be compared to