|Home | About | Journals | Submit | Contact Us | Français|
Faculty of Food Life Sciences, Toyo University, Ora-gun, Gunma 374-0193, Japan (M. Fujisawa)
This review focuses on the ATP synthases of alkaliphilic bacteria and, in particular, those that successfully overcome the bioenergetic challenges of achieving robust H+-coupled ATP synthesis at external pH values > 10. At such pH values the protonmotive force, which is posited to provide the energetic driving force for ATP synthesis, is too low to account for the ATP synthesis observed. The protonmotive force is lowered at very high pH by the need to maintain a cytoplasmic pH well below the pH outside, which results in an energetically adverse pH gradient. Several anticipated solutions to this bioenergetic conundrum have been ruled out. Although the transmembrane sodium motive force is high under alkaline conditions, respiratory alkaliphilic bacteria do not use Na+-instead of H+-coupled ATP synthases. Nor do they offset the adverse pH gradient with a compensatory increase in the transmembrane electrical potential component of the protonmotive force. Moreover, studies of ATP synthase rotors indicate that alkaliphiles cannot fully resolve the energetic problem by using an ATP synthase with a large number of c-subunits in the synthase rotor ring. Increased attention now focuses on delocalized gradients near the membrane surface and H+ transfers to ATP synthases via membrane-associated microcircuits between the H+ pumping complexes and synthases. Microcircuits likely depend upon proximity of pumps and synthases, specific membrane properties and specific adaptations of the participating enzyme complexes. ATP synthesis in alkaliphiles depends upon alkaliphile-specific adaptations of the ATP synthase and there is also evidence for alkaliphile-specific adaptations of respiratory chain components.
ATP is the dominant chemical energy currency of living cells. A minority of cells, such as mature red blood cells and some non-respiratory microbial cells, generate ATP exclusively through substrate level phosphorylation. By contrast, the majority of ATP synthesized by animal, plant and microbial cells is obtained at higher energy yield from catabolic substrates or light via use of oxidative phosphorylation (OXPHOS) or photo-phosphorylation, which depend upon the catalytic activity of F1F0-ATP synthases (which will be called ATP synthases in this article). These synthases have been part of the physiological landscape through much of the evolutionary path of contemporary organisms, perhaps as far back as LUCA, the putative “last universal common ancestor” of living organisms . ATP synthases of animals and plants are embedded by one of their domains, the F0 domain, in the inner membranes of mitochondria or in chloroplast thylakoid membranes. An interesting exception to these classic organelle localizations are recent reports of ectopic F1F0-ATP synthases in caveolae/lipid rafts on the surface of a variety of mammalian cell types, including cancer cells [2-5]. The homologous but structurally simpler bacterial ATP synthases are found in the cytoplasmic membrane or thylakoid membranes [6-8].
ATP synthesis by ATP synthases is energized by electrochemical gradients of H+ or Na+ across the membrane. ATP synthesis is coupled to downhill movement of ions fostered by those gradients across mitochondrial, thylakoid and bacterial cell membranes [6, 8]. ATP synthases are reversible, i.e., they can function as ATPases that hydrolyze ATP in concert with ion pumping (energetically uphill) in the opposite direction of ion uptake during synthesis. For many fermentative bacteria, ATPase activity is the only physiological activity of the enzyme. By contrast, the hydrolytic activity of ATP synthases in most other settings is carefully controlled, or even latent, in order to protect the cytoplasmic ATP pool [9-11]. Structure-function information that emerged over the past decade and a half supports a rotary mechanism for ATP synthesis that involves sequential conformational changes in the three catalytic sites of the synthase. Experiments on single synthase complexes have captured steps and sub-steps of the rotation. An extensive collection of high resolution structures of major portions of ATP synthases has also emerged. Although not yet representing the entire structure, the structural information has enormously fostered hypothesis-building and experimental work aimed at greater mechanistic understanding [12-24].
The importance of resolving the mechanistic details of ATP synthase function arises from the enzyme’s central role in cell physiology. The importance of further insights is underscored by the serious consequences of genetic defects in ATP synthase structural genes, regulatory genes or assembly factors [25, 26]. Dysfunctions in respiration-dependent ATP synthesis, i.e. oxidative phosphorylation (OXPHOS), are further correlated with complex disorders whose frequency increase with ageing, such as Type 2 diabetes . The mechanistic details of ATP synthase function will also enhance our ability to inactivate the ATP synthase of bacterial pathogens in which the enzyme plays an essential role in part of their pathogenic life cycle. This is the promising basis for use of diarylquinolines that target the ATP synthase of pathogenic Mycobacterium tuberculosis and its multi-drug resistant forms [28-31]. The resulting inhibition of ATP synthase reduces viability of the pathogens under hypoxic conditions in which they otherwise persist in the host . Analogously, growing insights into the mechanisms of natural ATP synthase modulators, and inhibitors of its hydrolytic mode, could lead to rational design of therapeutic strategies for human disorders that are linked to ATP synthase and to strategies for inhibiting the ATP synthases that are ectopically expressed on cancer cell surfaces [10, 33, 34].
There is another gateway to progress on the structure-function underpinnings of ATP synthase activity and its contributions to physiology. This is the growing awareness that ATP synthases from higher organisms, as well as the cytochrome oxidases that participate in ATP synthase-dependent OXPHOS, have multiple isoforms of important subunits [35-38]. Some of these have been shown to relate to tissue-specific functional difference [39-41]. Among bacteria, the pace of evolutionary adaptations and the extraordinary range of environments in which ATP synthases function have yielded a wealth of adaptations and variants in the OXPHOS machinery [42-44]. Microbial ATP synthases exhibit variations that support their adaptation to very distinct conditions of pH, temperature, oxygen levels, salt etc. [11, 31, 43, 45, 46]. These variations have structural and functional impacts on the ATP synthase that presumably contribute adaptive value under particular conditions. The recognition of these variations has challenged and expanded our understanding of how ATP synthases work. In this article, the discussion will center on the ATP synthase of alkaliphilic bacteria that are highly adapted to grow well at pH values as high as 11 [45, 47-49]. The adaptive strategies by which alkaliphiles overcome the energetic challenges to ATP synthesis at high pH started as an esoteric puzzle and grew into a model system with implications for ATP synthases in general. Before focusing on these bioenergetic challenges and adaptive strategies of alkaliphiles, we will present an overview of structural features of bacterial ATP synthases, of the operons that encode them and of the process of ATP synthase-mediated ATP synthesis. Variations that have been observed in subunit structure of diverse non-alkaliphile ATP synthases and in the organization of atp operons will be noted as part of this overview.
Bacterial ATP synthases contain eight proteins that are present in the synthase at different stoichiometric ratios per complex. Like their eukaryotic counterparts, bacterial ATP synthases have two domains, the F1 and F0 domains, as indicated in the schematic diagram in Fig. 1. The cytoplasmically located, soluble F1 domain contains three catalytic α and β subunit pairs (shades of purple in Fig. 1) and single γ, δ, and ε subunits (see reviews [6, 8, 24]). The membrane-embedded F0 domain is composed of a single a-subunit, two b-subunits and multiple c-subunits. The a, b and δ subunits (highlighted in blue in Fig. 1) comprise the stator for the rotary machine. The two b-subunits and the δ-subunit form a peripheral stalk. This stalk connects the two domains of the complex, spanning the ion-translocating a- and c-subunits of the F0 and the catalytic domain of the F1 that contains the three α-and β-subunit pairs [50-54]. The rotor element found in the F0 (highlighted in green in Fig. 1) is a homo-oligomer of hairpin-like c-subunits in most ATP synthases. The cytoplasmic loops that are exposed on the N-side of the membrane (the electrically negative side) interact with the γε sub-complex which forms the central stalk, with the γ-subunit extending asymmetrically into the F1 catalytic domain  (see section 2.3).
The number of c-subunits observed per c-rotor varies from 10-15, depending upon the organism [18, 20, 56-65]. The two c-subunit stoichiometries so far determined for eukaryotic c-rotors are at opposite ends of those ranges with the yeast stoichiometry at 10  and the spinach chloroplast stoichiometry at 14 [67-69]. By contrast to the typical homo-oligomeric rotor, the Na+-coupled ATP synthase of the anaerobic, acetogenic bacterium Acetobacterium woodii instead has a hetero-oligomeric rotor with two different c-subunits. The A. woodii c-rotor has 10 c-subunits/rotor; 9 of them are the products of two identical genes in the atp operon that have the usual 2-helix hairpin structure of other F-type ATP synthase c-subunits. The remaining rotor subunit is encoded by a third c-subunit gene and is a 4-transmembrane helix (TMH) “c”-subunit of the type found in V-type ATP synthases/ATPases . The V-type ATPase from Enterococcus hirae has 10 of the V-type subunit in its K-rotor, the V-type ATPase equivalent of the c-rotor . The larger V-type subunit has only one ion-binding site, like that of the smaller ones. Its presence in the A. woodii rotor is expected to impact the functional properties of the enzyme. Perhaps the V-type subunit enhances the propensity for ATP hydrolysis in a manner that relates to its specialized physiological circumstances [70, 72].
Bacterial ATP synthases are most often encoded in single operons that have the eight structural genes in the order atp (or unc) BEFHAGDC, respectively encoding the a, c, b, δ, α, γ, β, ε subunits (as shown for Bacillus subtilis in Fig. 2). Some bacteria, including Rhodosprillum rubrum and the cyanobacterium Synechococcus, have distinct operons for genes encoding F0 proteins and those encoding the F1 proteins, consistent with the hypothesis that the genes for the two domains evolved as separate modules [73, 74]. In these particular bacteria there are two homologous genes encoding the two b-subunits . In other bacterial operons, there are different gene orders, e.g. atpEBFHAGDC in Streptococcus mutans and Streptococcus sanguis . In most bacterial atp operons, a ninth gene precedes the eight structural genes for the ATP synthase itself. This is an atpI gene that encodes a putative hydrophobic protein predicted to have 4 TMH. The AtpI protein sequence is not highly conserved across species . In some instances atpI is under control of a separate promoter from the other atp genes , in some instances atpI is absent entirely  and in some instances atpI is absent but a different gene of unknown function takes its usual place . Gay and Walker suggested a chaperone/assembly type of function for the AtpI protein when they first described the atpI gene . Partial homology of AtpI to yeast proteins with membrane assembly functions has also been noted [80, 81]. Recently a chaperone function was clearly shown for an AtpI protein. The AtpI protein from Propionigenium modestum was shown to be required for assembly in E. coli of a hybrid ATP synthase whose F1 domain was from thermophilic Bacillus PS3 and whose F0 domain was from P. modestum (Pm) . His-tagged Pm-AtpI was found to co-purify with both the c-subunit monomer and the assembled c-ring, and a role of Pm-AtpI in assembly of the ring was suggested. Further evidence for such a role was obtained from experiments in which the P. modestum c-subunit was synthesized in vitro using purified components in the presence of liposomes . Formation of a c-ring again depended upon the presence of Pm-AtpI. The extent of the contribution of AtpI to assembly of the c-rotor in native bacterial settings is not yet clear. Deletions or disruptions atpI in E. coli resulted in a reduced growth yield but the mutants retained significant capacity for non-fermentative growth that requires ATP synthase activity [84, 85]. Deletion of atpI in alkaliphilic Bacillus pseudofirmus OF4 similarly did not affect non-fermentative growth dramatically except under specific ionic conditions (see below) . AtpI may have a less dominant role than the YidC chaperone family members that play a role in both the c- and a-subunit assembly in E. coli [87, 88]. Possibly the relative importance of the AtpI chaperone role varies among bacteria depending upon the complement of chaperones that is present and/or upon specific conditions.
The atp operon of alkaliphilic B. pseudofirmus OF4 has a small ORF upstream of the atpI gene, which was designated atpZ (Fig. 2) . The atpZ gene of B. pseudofirmus OF4 was shown to produce a hydrophobic protein product that is predicted to possess 2-TMH. Non-polar deletion of atpI, atpZ or a double atpIZ deletion resulted in a defect in non-fermentative growth at pH 7.5 that was especially pronounced at sub-optimal [Mg2+]; the relative deleterious effects of deletions was ΔatpI <ΔatpZ <ΔatpZI. In view of the demonstration of a chaperone role for P. modestum AtpI, the possibility arises that the Mg2+-related effects of AtpI status in B. pseudofirmus OF4 are an indirect effect of AtpI on assembly of the ATP synthase subunits or on assembly of AtpZ in this organism. The direct role of AtpZ also needs to be more fully explored. AtpZ has numerous homologues that are thus far restricted to the bacterial Firmicutes phylum, i.e. in low %GC Gram-positive bacteria such as those shown in Table 1. Not all members of the genera in which atpZ is found have this gene and atpZ has not yet been observed in operon sequences from several fermentative families of Firmicutes, e.g. lactobacilli, staphylococci, streptococci, mycoplasma. This suggests that the function of AtpZ is related to the physiological setting of a specific sub-set of low %GC Gram-positive bacteria. In the genomes of 5 high %GC Gram-positive bacteria, a different gene is found between atpI and atpB and is predicted to encode a hydrophobic peptide that exhibits sequence similarity to bacterial permeases .
The catalytic reaction carried out by ATP synthases/ATPases is shown below (i). The participation of a proton (H+) as a substrate in the ATP synthesis direction confers pH-dependence upon the reaction thermodynamics. When a very high external pH causes the cytoplasmic pH to rise (e.g. in B. pseudofirmus OF4 at pHout > 9.5), the lower availability of substrate H+ will therefore affect ATP synthesis. ATP synthesis at high pH also faces additional challenges described in section 3.2. The catalytic reaction that includes the coupling ions, multiple H+ or Na+ that are translocated during synthesis, is also shown below (ii); the P-side and N-side designate, respectively, the electrically positive and electrically negative side of the membrane. The majority of bacterial synthases, and the ATP synthases of animals and plants, are coupled to an electrochemical gradient of protons, the protonmotive force (pmf). The pmf is generated by proton-pumping complexes of the respiratory chain or by light-driven proteins and protein complexes that pump H+ across the membranes in which the synthases are found [89, 90]. The F1F0-ATP synthases of a smaller group of bacteria are energized by an electrochemical gradient of Na+, sodium motive force (smf), that is generated by Na+ pumping protein complexes in the membrane [6, 8]. There are motifs in the membrane-embedded c-subunits that identify Na+-coupled ATP synthase and at least two types of H+-coupled ATP synthases .
The pmf or smf respectively power downhill translocation of H+ or Na+ that energizes ATP synthesis. The translocation starts with a-subunit-mediated uptake of the ions from the P-side of the membrane and binding of the ion to successive c-subunits of the rotor at the interface of subunits a and c. After rotation of the rotor, successive c-subunits resume interaction with the a-subunit and successive release of the ions from those c-subunits occurs. The ions then complete the path across the membrane into the bacterial cytoplasm (see Fig. 1) [8, 24, 91]. No high resolution structural data are yet available for the a-subunit, but extensive biochemical and genetic evidence indicates that this ATP synthase subunit plays roles in providing the proton path from outside the membrane surface to the carboxylates of interacting c-subunits of the rotor. The ions are passed to the c-subunit carboxylates within the membrane. The c-subunit carboxylates interact with a region of the a-subunit TMH4 where there is an essential arginine residue [92, 93]. The positive charge of the conserved a-subunit arginine is proposed to facilitate the release of H+ from protonated carboxylates of c-subunits completing a rotation of the ring and to create an environment in which re-protonation occurs [91, 94-100]. The proton exit pathway to the bacterial cytoplasm has been proposed to be a discrete half channel within the a-subunit (as is shown in Fig. 1)[6, 101], but an alternative suggestion is that the exit pathway is along the a- and c-subunit interface [8, 99]. The a- and c-subunit-mediated rotation of the c-rotor causes rotation of the γ and ε subunits. These central stalk subunits interact with well-ordered cytoplasmic loops of the rotating c-ring so that torque is generated [102, 103]. The rotating γ subunit tags successive β subunits of the F1, inducing conformational changes that enable ATP synthesis to proceed [22, 104, 105].
Alkaliphilic bacteria grow optimally at pH values above 9, with extreme alkaliphiles growing optimally above pH 10. Obligate alkaliphiles are incapable of growing at near neutral pH values. Facultative alkaliphiles exhibit significant growth at near neutral pH but also exhibit deficits in growth yield or growth rates relative to neutralophilic bacteria at the same pH [45, 48, 49]. This is likely to reflect the significant number of adaptations of alkaliphile proteins, membranes and physiology that enable alkaliphiles to grow well at extremely high pH values and to tolerate sudden alkaline shifts. According to a general engineering principle referred to as either the “no free lunch” principle or the principle of “conservation of fragility” , mechanisms that increase robustness to a particular condition, e.g. growth capacity at high pH, inevitably compromise robustness (i.e., confer fragility) under another condition, e.g. near neutral pH. Obligate alkaliphiles exhibit absolute “fragility” at near neutral pH and are often the best adapted at the extreme alkaline end of the range. Facultative alkaliphiles strike a more temperate balance. Alkaliphilic bacteria can be isolated from natural environments such as highly alkaline soda lakes or alkaline hyper-saline lakes like Lake Mono, California. They can also be isolated from highly alkaline enrichments created by industrial activities, e.g. indigo dye plants, from highly alkaline segments of certain insect guts and from estuaries or soils that have periods or micro-environments of high alkalinity (e.g. periods of evaporation in estuaries or clay particles with alkaline crevices in soil) [107-111]. Alkaliphilic bacteria are from diverse taxonomic groups and have diverse physiologies but share challenges of cytoplasmic pH homeostasis and functional adaptations of proteins that are secreted or highly exposed on the outer surface. They have received attention because of their bioenergetic profiles, for the useful applications of their proteins  and their potential for bioremediation . A sampling of alkaliphiles whose genome sequences are complete or in progress is presented in Table 2.
In 1977, Garland predicted that alkaliphilic bacteria would have to maintain a cytoplasmic pH significantly below the external pH values . Experiments in many alkaliphilic strains confirm this expectation. Alkaliphiles characteristically possess an impressive capacity for alkaline pH homeostasis, maintaining cytoplasmic pH values up to 2.3 pH units lower than the pH of the bulk medium under carefully pH-controlled continuous culture conditions as well as in highly buffered batch cultures [47, 115-119]. In Fig. 3, the cytoplasmic pH of two neutralophilic bacteria, Escherichia coli and Bacillus subtilis, and four different alkaliphilic bacteria is shown at an alkaline external pH at which growth is still optimal or near-optimal for that particular organism [115, 117, 118, 120-122]. Two of the alkaliphiles, aerobic Bacillus sp. TA2.A1 (recently renamed Caldalkalibacillus thermarum TA2.A1)  and anaerobic Clostridium paradoxum , are thermophiles. Their alkaliphily is less extreme than found in the other two examples that are non-thermophilic alkaliphiles, the cyanobacterium Spirulina (also called Arthrospira) platensis  and the extreme facultative alkaliphile B. pseudofirmus OF4 . The lesser alkaliphily of the thermoalkaliphiles probably reflects trade-offs among adaptive strategies . For all the depicted bacteria, and others where it has been tested, the capacity of neutralophilic as well as alkaliphilic bacteria for maintaining a cytoplasmic pH below the external pH, depends heavily upon electrogenic Na+(Li+)(K+)/H+ antiporters. These secondary active transporters utilize the energy of the transmembrane electrical potential (Δψ) component of the pmf or smf that is generated by primary ion pumps. The antiporters take up external H+ in exchange for monovalent cations (Cat+) from the cytoplasm, with a H+/Cat+ >1. This makes it possible to acidify the cytoplasm relative to the bulk medium in a Δψ-dependent fashion [48, 49, 120, 127-129]. In alkaliphiles, antiporter-based pH homeostasis is thus far specific for Na+ as the efflux substrate (and the antiporters usually have an additional capacity for Li+/H+ antiport). This is not the case in neutralophiles, which use both Na+(Li+)/H+ and K+/H+ antiporters (Fig. 3) [48, 49]. The antiporter that has a dominant role in pH homeostasis of two extreme alkaliphiles is the hetero-oligomeric Mrp antiporter that contains 6-7 different hydrophobic proteins [48, 49, 127, 129-131]. The uniquely complex structure of Mrp antiporters is likely to be especially suitable for this role. Mrp antiporters are hypothesized to present a large surface on the outside of the inner membrane that facilitates H+ capture and funneling into the antiporter [48, 49, 131]. Mechanisms for Na+ re-entry to supply cytoplasmic Na+, the substrate for the high antiport activity, are essential to sustain antiport-dependent pH homeostasis, especially when the external [Na+] is low [48, 49, 132-135]. The major Na+ re-entry pathways are shown in Fig. 3 (and see section 3.3 below). A variety of additional mechanisms for cytoplasmic H+ retention are also present in alkaliphiles although none can thus far replace the antiport-based component [48, 49].
For two of the four alkaliphile types depicted in Fig. 3, successful pH homeostasis results in a thermodynamic problem vis a vis ATP synthesis by H+-coupled ATP synthases. Both Bacillus sp. TA2.A1 and B. pseudofirmus OF4 exhibit a reduced pmf relative to neutralophilic bacteria because the “reversed ΔpH”, i.e. acid inside relative to the outside, has an adverse effect on the total pmf. The other pmf component, the transmembrane electrical potential (Δψ, inside negative relative to outside), rises at elevated pH. However, it does not increase enough to offset the chemiosmotically counterproductive pH gradient [11, 45, 117, 118]. By contrast to the aerobic, non-fermentative alkaliphiles, C. paradoxum is not subject to this problem because in this fermentative anaerobe, the Na+-coupled ATP synthase/ATPase only functions in the hydrolytic direction, generating a Δψ and controlling cytoplasmic Na+ levels as it pumps Na+ outward [115, 136]. The ATP synthase of the fourth alkaliphile in Fig. 3, cyanobacterial Spirulina platensis, is largely and perhaps entirely sequestered in thylakoids that are not continuous with the cytoplasmic membrane [137-139]. To the extent that this sequestration is extant, the low bulk pmf across the cytoplasmic membrane does not pose a problem for ATP synthase function. The pmf across the thylakoid membrane, largely in the form of a ΔpH, is much larger than that across the cytoplasmic membrane [122, 140].
There is a further feature of alkaliphilic bacteria that was unanticipated by Garland  and is relevant to the bioenergetic challenge of carrying out OXPHOS. It was incorrectly expected that alkaliphiles would need to maintain a cytoplasmic pH close to the range maintained by nonalkaliphiles . It turns out that alkaliphiles grow well at cytoplasmic pH values that are not tolerated at all by non-alkaliphiles. Extreme alkaliphiles maintain significant growth capacities at surprisingly high cytoplasmic pH values. The optimal cytoplasmic pH for neutralophilic bacteria such as E. coli and Bacillus subtilis is 7.5-7.6. Neutralophiles that lack Na+/H+ antiporters or lack antiport function because of protonophoric uncoupling are restricted to growth in a pH range from 6.3-7.7 [141, 142]. Neutralophiles that possess active mechanisms of alkaline pH homeostasis maintain a cytoplasmic pH of 7.5-7.6 up to external pH values of 8.5 (Fig. 3). At pH values above 8.5 the growth rate of neutralophiles usually slows and there is greatly reduced capacity for growth at pH ≥ 9.0 [48, 49]. Moreover, growth arrest occurs when the cytoplasmic pH of E. coli rises to ~ pH 8 after sudden alkalinization of the medium to 8.3 and occurs in B. subtilis when this organism is shifted suddenly to medium at pH 9.0 [143, 144]. By contrast, alkaliphiles maintain a cytoplasmic pH of 7.5 and optimal growth rate when the external pH is as high as 9.5. When the external pH is even higher, the alkaliphile’s pH capacity to retain a cytoplasmic pH of 7.6 is exceeded. However, these bacteria continue to grow in the presence of cytoplasmic pH values way above those that are compatible with neutralophile survival and/or growth. For example, B. pseudofirmus OF4 grows optimally even when the cytoplasmic pH has risen to 8.2-8.3 (at pHout=10.5) and when pH 8.5-equilibrated cells are suddenly shifted to an external pH of 10.5, they maintain this cytoplasmic pH. B. pseudofirmus OF4 continues to exhibit significant non-fermentative growth at external pH values over 11, at which the cytoplasmic pH is 9.5 [45, 47, 118]. Over the range of external pH from 10.5 up past 11, the growth rate decreases in parallel to the rise in cytoplasmic pH, underscoring the centrality of pH homeostasis in spite of the extraordinary tolerance of a high cytoplasmic pH. Little is known yet about adaptations of cytoplasmic components that underpin this tolerance, but there are several “plumbing problems” that this tolerance raises with respect to synthesis of ATP. First, if an alkaliphile is to take advantage of OXPHOS as part of its metabolic strategy, it must minimize outward H+ leaks through the membrane via any of the membrane proteins involved in the process during transient reductions of pmf. Second, mechanisms of H+ capture need to address not only the thermodynamic problem of a low pmf but also the kinetic problem of H+ capture from a highly alkaline exterior.
Thus far, the problem of a low pmf has been documented primarily in Gram-positive alkaliphiles of the Bacillus genus or related genera. In these alkaliphiles, the “reversed ΔpH”, acid inside relative to outside, results from successful pH homeostasis. The reversed ΔpH detracts from the total pmf and there is incomplete compensation by a rise in Δψ. Possibly alkaliphiles will yet be found that have different proton-pumping capacities and membrane properties such that a compensatory Δψ eliminates the adverse effect of pH homeostasis on the total pmf. For the two aerobic alkaliphilic Bacillus strains shown in Fig. 3 that grow well at pH 10-10.5, the effect of pH homeostasis on the pmf is significant. In pH-controlled batch cultures, thermoalkaliphilic Bacillus sp. TA2.A1 cells exhibited a pmf of -164 mV during growth at pH 7.5, where growth is fermentative  and of -78 mV at pH 10; at pH 10, the organism can grow non-fermentatively and exhibited robust ATP synthesis . In alkaliphilic B. pseudofirmus OF4 grown under continuous culture conditions with rigorous control of the medium pH, the pmf during growth at pH 7.5 was -136 mV and at pH 10.5 was -50 mV. Typical pmf values for neutralophiles, e.g. E. coli and B. subtilis, at pH 7-7.5 are ~ -130 to -140 mV [145, 146], which is comparable to the pmf of B. pseudofirmus OF4 at pH 7.5 [116, 118]. In B. pseudofirmus OF4, however, ATP synthesis was more robust at pH 10.5 than at pH 7.5 even though the pmf was much lower at the higher pH .
Ion-coupled “bioenergetic work” such as ion-coupled solute uptake and flagellar motility in alkaliphiles is coupled exclusively to Na+ whereas it is coupled exclusively to H+ or to a mix of H+ and/or Na+-coupled transport or “Mot” systems in neutralophilic bacteria (Fig. 3) [45, 147]. The use of Na+-coupled systems for solute uptake and motility is consistent with the notion that the smf is larger than the available pmf under highly alkaline conditions. Na+ is used for motility and transport even in many of the Bacillus alkaliphiles such as B. pseudofirmus OF4 that were isolated from environments in which the Na+ concentration is not particularly high and special mechanisms to capture Na+ for antiport are important [45, 120]. It was therefore anticipated that ATP synthesis in alkaliphilic bacteria would also turn out to be Na+-coupled and was further suggested that neutralophilic organisms possessed the capacity to switch to Na+-coupling of ATP synthesis upon a drop in pmf [148-150]. ATP synthases of anaerobic alkaliphiles, most of which function as ATPases physiologically, are the only alkaliphile ATP synthases/ATPases thus far found to be Na+-coupled . This coupling choice is clearly adaptive for an alkaliphile for an ion-pumping ATPase. Use of Na+ pumping would meet the needs of Δψ and smf generation to support of Na+-coupled solute uptake and motility and would help protect against excessive Na+ accumulation under circumstances of elevated salt. Importantly, use of Na+ as the pumped ion would avoid the likely catastrophic loss of cytoplasmic H+ that could occur if the ATPase activity was coupled to H+ instead. There is one recently characterized anaerobic alkaliphile, Alkaliphilus metalliredigens , that catalyzes respiratory metal reduction and apparently possesses a Na+-coupled ATP synthase, based on sequence motifs (see Table 2). This synthase probably completes a Na+ cycle that contains putative membrane-associated, Na+-translocating organic acid decarboxylases .
By contrast, ATP synthases of aerobic or facultatively aerobic alkaliphiles whose physiological role is ATP synthesis, have thus far all been found to couple synthesis to H+ [45, 47, 153, 154]. Moreover, this synthesis is inhibited by treatments that abolish the Δψ . Coupling to H+ has similarly been found for Vibrio species. They had been expected to have Na+-coupled synthases since they have one primary respiratory pump coupled to Na+ and have a larger smf than pmf . The alkaliphile synthases exhibit no ability to use Na+ as an alternative coupling ion [47, 154, 157]. This is in contrast with the Na+-coupled synthases that are found in fermentative bacteria such as P. modestum, which can use H+ for coupling to ATP synthesis when Na+ concentrations are low .
As already noted (see equation (i) in section 2.3), the energy cost of ATP synthesis rises as the pH rises and the energy requirement for pH homeostasis at high pH further challenges alkaliphiles. Perhaps these elevated energy needs are the basis for use of pmf-coupled ATP synthases. The pmf in alkaliphiles is usually generated during electron transport events that can cover a large redox span and maximize energy conservation from growth substrates . It is notable that except for the final redox site of the terminal oxidases, the redox potentials of respiratory chain components from alkaliphilic Bacillus species are significantly lower than those of neutralophilic homologues [159-161]. Use of sodium-coupled energetics in organisms without proton-tight membranes has been proposed to have preceded proton energetic during evolution . However, among current bacterial life forms, sodium energetics may be best suited to “boutique” situations in which the niche is highly specialized. Proton-based energetics may be required in niches that require particularly high energy expenditures and/or are highly competitive niches in which fast growth rates and high growth yields are important.
A second consideration that may preclude use of the known types of Na+-coupled ATP synthases by alkaliphiles is that those synthases can also couple to H+ . In cells that have large “reversed” pH gradients, inside more acidic than the outside, the outwardly directed pH gradient might therefore inhibit synthesis of the ATP synthase even when Na+ is used as coupling ion.
Finally, there are indications that respiratory alkaliphiles and neutralophiles derive added value from using a H+-coupled ATP synthase. At acidic pH, ATP synthases/ATPases that function in the hydrolytic, H+-pumping mode are induced and have been shown to play a role in preventing adverse acidification of the cytoplasm [163-166]. Conversely, increased expression and activity of ATP synthases that function in the synthetic direction are implicated in a role in alkaline pH homeostasis by transcriptome observations and studies of ATP synthase mutants. Alkali challenge of E. coli, B. subtilis, Desulfovibrio vulgaris, and Corynebacterium glutamicum are all associated with increased expression of H+-coupled ATP synthases [76, 167-170]. A mutant of B. subtilis that has lost function of its Mrp antiporter also exhibits increased expression of its ATP synthase . The H+ taken up during ATP synthesis contribute to pH homeostasis in alkaliphiles. This was shown in studies of a panel of mutants in alkaliphile-specific motifs of the B. pseudofirmus OF4 that have defects of differing magnitude in their capacity to make ATP . The mutants and wild-type control cells were depleted of ATP and equilibrated at pH 8.5, after which they were shifted to a malate containing medium at pH of 10.5. Although antiporter-based pH homeostasis and ATP synthesis are both pmf-consuming processes, the immediate post-shift cytoplasmic pH was lowest, below 8.2, in a mutant that synthesized ATP more rapidly than wild-type. The post-shift cytoplasmic pH of the remaining mutants was higher than wild-type and correlated with the extent of their deficits in ATP synthesis. Contributions of ATP synthases to alkaline pH homeostasis would create a selective advantage for use of H+-coupled ATP synthase as long as the bioenergetic challenges for their use can be overcome or bypassed.
Before reviewing alkaliphile strategies to overcome the challenges of low H+ concentration in the bulk and sub-optimal pmf conditions, we note mechanisms that are in place in alkaliphiles that prevent use of the ATP synthase as an ATPase. Such controls are important in all ATP synthases, not just in alkaliphiles and not just in bacteria. Control of ATPase activity prevents depletion of the ATP pool under conditions in which there is a drop in pmf. For alkaliphiles, stringent controls are further needed because loss of the ATP pool is not the only major risk. The other risk is that at high pH-low pmf, the central process of pH homeostasis would be undercut by ATPase-dependent alkalinization of the cytoplasm. As noted earlier, while ATP synthases are generally described as reversible machines, this characterization does not apply to the physiological reality of many of them. For example, membrane vesicles or purified ATP synthases from aerobic alkaliphilic Bacillus species catalyze only very low rates of ATP hydrolysis, but the hydrolytic activity can be dramatically increased by solvents such as methanol or detergents such as octyl glucoside or LDAO [153, 173, 174]. The purified, reconstituted ATP synthase from thermoalkaliphile Bacillus sp. TA2.A1 was shown to catalyze ATP synthesis in response to an imposed membrane potential but it was not possible to detect proton-pumping ATPase activity even when the ATPase activity was increased to high levels by LDAO . The ATP synthases of non-thermophilic alkaliphilic Bacillus species are less extremely latent, exhibiting proton-pumping together with low, but detectable hydrolytic activities of enzymes from B. pseudofirmus OF4 and B. alcalophilus [153, 154]. It was suggested that due to the combined challenges of high temperature and high pH, the thermoalkaliphile would be especially at risk of losing protons .
It appears that most if not all F-type ATPases share one inhibitory mechanism, that of tight binding of MgADP , while other mechanisms of inhibition are more specific to the type of organelle or organism. It is thought that the MgADP is bound to a high-affinity catalytic site. Exposure of the enzyme to inorganic phosphate, or nucleotides, or a pmf can overcome the inhibition, probably by displacing the ADP [175, 176]. In addition, mitochondria and chloroplasts each have special mechanisms to inhibit the ATPase activity. Mitochondria are unique in using a naturally occurring protein, the inhibitor protein, to regulate their ATPase activity. This protein, IF1, is a small basic protein that inhibits hydrolytic activity below pH 8 but has no effect on ATP synthesis ; recently, the crystal structure of a monomeric form of the inhibitor protein bound to mitochondrial F1 was reported . The ATP synthase from chloroplasts differs from all other ATP synthases in its γ subunit, which contains a unique regulatory disulfide that is reduced physiologically by a thioredoxin system during exposure to light. This results in activation of both its hydrolytic and its synthetic activity. Chloroplast F1 can be reduced by dithiothreitol in vitro, which also stimulates ATPase activity . Structural rearrangements between γ and ε upon disulfide reduction probably account for the activation of activity . Interestingly, the cyanobacterial ATP synthase from alkaliphilic Spirulina platensis shares a number of properties with the chloroplast enzyme, including stimulation of its hydrolytic activity by methanol  and enhanced ATP synthesis at suboptimal pmf values when reduced by dithiothreitol [182, 183]. However, it lacks the regulatory disulfide found in the organelle enzyme, leading to the suggestion that other cysteines in the S. platensis γ subunit may be involved in this activation [182, 183].
Chloroplasts and many bacterial enzymes share a common regulatory feature, inhibition by the endogenous ε subunit of the F1 moiety. Either removal of the ε subunit or truncation of the C-terminal end of ε in the F1 results in activation of hydrolytic activity in E. coli, Bacillus sp. PS3, Bacillus sp. TA2.A1, or spinach chloroplasts . From structural information on the isolated ε subunit and the γε complex [185-187] of E. coli F1, and from cross-linking experiments with the E. coli  and Bacillus sp. PS3 enzymes , it was concluded that the C-terminal end of the ε subunit exists in two conformational states. When ε is in a putative “extended” state, stretched alongside the γ subunit, the enzyme’s hydrolytic activity is blocked, but not the synthetic activity. When ε is in a putative “contracted” conformation, away from γ, hydrolysis is active . Inhibition by the extended ε form is proposed to be mediated by binding of basic amino acids in the C-terminal region of ε to the acidic groups in the β subunit DELSEED region that interacts with the γ subunit. Replacement of either the ε basic residues or the β acidic residues with alanines results in activation of hydrolytic activity in Bacillus sp. PS3 . In the thermoalkaliphile, replacement of the 6 arginines in the C-terminal region increased activity 5-fold, although the enzyme could still be further stimulated 8-fold by LDAO .
Another basis for latency of the thermoalkaliphile ATP synthase, probably a dominant basis, was revealed in a structural study of the F1 domain (lacking the δ subunit) . Salt bridges were identified between two successive arginines in the N-terminus of the γ subunit and two aspartates in the helix-turn-helix C-terminal domain of the β subunit. These salt bridges result in a bent orientation of the γ subunit relative to that found in other F1 structures and appear to prevent rotation in the hydrolytic direction. The tilted γ subunit was visualized in remarkable electron micrographs of the purified ATP synthase, which showed that the γ subunit was bound to an edge of the c-ring and then angled up into the head piece . Biochemical support for the effect of the salt bridges on activity was found by mutation of the γ subunit arginines to glutamines, which created an active enzyme .
Different bacteria possess different numbers of c-subunits/ATP synthase. A complete 360° turn of the c-ring yields 3 ATP and the number of H+ or Na+ that move across to the N-side of the membrane during a full turn is equal to the number of c-subunits in the ring. Therefore the H+/ATP ratio varies among different bacterial strains. This has bioenergetic consequences because:
ΔGp is the phosphorylation potential that reflects the energy required to sustain the [ATP]/[ADP][Pi] ratio. The bulk pmf (consisting of a transmembrane difference in electrical potential, Δψ, and a transmembrane difference in pH, ΔpH) is the electrochemical gradient assessed between the aqueous phases on the two sides of the coupling membrane.
If an ATP synthase has a c-ring composed of 10 c-subunits, H+/ATP and ΔGp/pmf are both 3.3. The efficiency is greater than that of a synthase with 13 (as in thermoalkaliphile Bacillus sp. TA2.A1 [63, 65]) or 15 (as in S. platensis [20, 61]), for which the H+/ATP as well as ΔGp/pmf are respectively 4.3 and 5.0. However, since the ΔGp/pmf is highest in the synthases that have the most c-subunits/rotor ring, a high c-subunit stoichiometry would be expected to help overcome the problem of a low pmf in alkaliphiles (see discussion in ). Indeed the 13 c-subunits of the thermoalkaliphile Bacillus sp. TA2.A1 rotor is toward the high end of the 10-15 c-subunit stoichiometries reported to date [63, 65]. This can account for part of the observed capacity of this organism to synthesize ATP at pmf values that are lower than those found in other organisms. However, the c-subunit stoichiometry of 13 does not fully account for the observed ΔGp/pmf at pH 10. At pH 10, the pmf of Bacillus sp. TA2.A1 was -78 mV and the ΔGp was ~430 mV, so ~17 c-subunits/rotor would be needed instead of 13 to account for the ΔGp observed [65, 117]. For B. pseudofirmus OF4, there would have to be 29 c-subunits per ring to account for observed ΔGp/pmf ratios at pH 10.5 (ΔGp = 478 mV and pmf = -50 mV) and the stoichiometry would have to be closer to 33 c-subunits/ring at pH > 10.5 [116, 118, 172]. It currently seems unlikely that such high numbers of c-subunits will be found. Moreover, the two highest c-ring stoichiometries found to date are 14 in spinach [67-69] and 15 in S. platensis [20, 61]. In both of these organisms, the ATP synthase is sequestered in organelle membranes and energized by a substantial pmf that is largely in the form of a ΔpH. This raises the possibility that the highest c-subunit stoichiometries will turn out to be more closely associated with factors other than the magnitude of the pmf, e.g. the preponderance of ΔpH over Δψ [8, 11].
Another point about c-ring stoichiometry is relevant to the profiles of alkaliphiles. Before stable c-rings were isolated and visualized, the possibility had been raised of a variable c-subunit stoichiometry for individual bacteria. Variable c-subunit stoichiometry was proposed as a possible adaptive strategy of bacteria that faced environments in which the pmf varies, a group that would include facultative alkaliphiles [193, 194]. The hypothesis of a variable c-subunit stoichiometry in a particular organism is no longer considered tenable. Studies of intact and partially fragmented rings as well as biochemical assessments of rings isolated under different conditions support the conclusion that any given organism has a constant c-ring stoichiometry that is determined by the primary structure of the subunit [61, 195-198]. This means that a facultative alkaliphile such as B. pseudofirmus OF4 will use an ATP synthase with the same c-ring stoichiometry at the vastly different ΔGp/pmf ratios across its pH range from pH 7.5 to ≥ 10.5. Unless the c-subunit stoichiometry of extreme alkaliphiles is much higher than the current reported range, the low pmf problem that B. pseudofirmus OF4 faces at high pH will only be modestly addressed by its c-subnit stoichiometry. Moreover, an energetic price will be exacted for use of an enzyme with a high c-subunit stoichiometry at pH 7.5 even if the stoichiometry is high but within the current range. The use of a relatively inefficient enzyme for ATP synthesis at pH 7.5, under conditions in which the pmf is not low, would disadvantage the alkaliphile at pH 7.5 relative to neutralophilic organisms using enzymes with lower stoichiometries. For the alkaliphile this is an example of a trade-off in which an adaptation to the extreme environment results in poorer adaptation to the non-extreme environment. Other such trade-offs have been demonstrated [45, 135]. The affects of these cumulative trade-offs probably account for the surprising finding that the molar growth yields of B. pseudofirmus OF4 on malate are comparable at pH 7.5 and 10.5 even though the energy costs for growth at pH 10.5 are higher [116, 118].
Na+-coupling was not found in aerobic alkaliphiles and the rotors of alkaliphile ATP synthases are not expected to have c-subunit stoichiometries high enough to resolve the discrepancy of the high ΔGp achieved by alkaliphilic bacteria at low bulk pmf. Rather, this unresolved energetic conundrum became a major paradigm for a large body of recent work that revisits a tenet of the original formulation of Mitchell’s chemiosmotic hypothesis. Mitchell posited that the pmf that drives H+-coupled bioenergetic work is entirely in the form of a delocalized bulk electrochemical gradient between two aqueous phases, e.g. the bacterial cytoplasm and the external growth medium [89, 199]. Williams challenged this tenet in his own formulation that was contemporaneous with Mitchell’s, positing charge densities and proton pathways along the membrane that required proximity and specific properties of participating catalysts [200, 201]. Other bioenergeticists as well as bacterial physiologists also advanced experimental data and models for alternatives to completely delocalized energy-coupling to bulk electrochemical ion gradients [47, 202-209]. As reviewed recently by Mulkidjanian et al. , proposals of localized and/or surface associated H+ translocation gained traction after experimental evidence using new technologies showed a measurable time delay between the emergence of pumped H+ on the outer surface of a coupling membrane and their equilibration with the bulk aqueous phase [211, 212]. Evidence from experiments and modeling also emerged for a capacity for both H+ storage and H+ translocation near the surface [213-215]. Together, delayed H+ equilibration with the bulk and near surface H+ translocation create the opportunity for H+ consumers such as the antiporters and ATP synthase of alkaliphiles to couple transport and ATP synthesis to H+ that reach them at the surface before they equilibrate with the bulk. Properties of the membrane proteins and their surfaces, e.g. the large Mrp antiporter surface, and of the membrane lipids, e.g. the high cardiolipin content of B. pseudofirmus OF4 membranes [45, 47, 216], could affect the retention and/or translocation of H+ near the surface [210, 214, 217]. In this scenario, the “inadequate” bulk pmf of alkaliphiles would not be the relevant driving force because the near surface pH would replace the external bulk pH for purposes of assessment of the effective transmembrane chemical gradient of H+.
A specific model of interfacial potential barriers that impede H+ escape from the membrane surface region predicts that such a barrier is common to all membranes and that the near surface pH, referred to as pHs, could be as low as pH 6 [218, 219]. It will be important to try to assess the near surface pH experimentally in a native system such as B. pseudofirmus OF4. A surface pH in the low end of the range permitted by the computational model of pHs may clash with the observed properties of the organism. If the functional pH of the membrane surface was pH 6.0 when the bulk external pH is very high, the total pmf at pH 10.5 would be high since there is a Δψ of -180 mV at that pH. This would eliminate both the need for the special adaptations to the alkaliphile OXPHOS machinery that the alkaliphilic bacilli possess (see section 8 below) and the need for alkaliphiles to use only Na+-coupling for solute uptake and motility. A functional surface water layer at pH 6 might also be difficult to reconcile with the progressive alkalinization of the cytoplasm at external pH values above pH 9.5. Perhaps the delocalized surface pH is lower than the external pH at external pH values ≥ 10.5 because of the general property of interfacial barriers to H+ equilibration and this offsets the alkaliphile low pmf problem but does not completely resolve it. Perhaps localized H+ microcircuits are also an important part of the energy-coupling solution in alkaliphiles and elsewhere. These circuits might involve mosaics or small coupling units that involve proximity of the H+ pumps and consumers, and rely upon the special features of the protein participants and on properties of the membrane [207, 208]. The efficacy of localized H+ microcircuits has recently gained support from spectroscopic studies of model liposomal systems in which lipid headgroups behaved as H+ collecting antenna and interplay with membrane lipids increased the rate of H+ uptake by membrane-associated proteins that catalyze such uptake [220, 221].
The possibility of H+ transfers during OXPHOS during very close encounters or dynamic physical interactions between one or more respiratory chain components and the ATP synthase has also been raised. Such interactions could support effective sequestered H+ transfers in alkaliphiles and even non-alkaliphiles [159, 222, 223]. Some of the special adaptations that have been identified in alkaliphile ATP synthase (see section 8. below) may provide a gateway for genetic-biochemical tests of various sequestration and microcircuit models of H+ transfer.
There is direct experimental evidence that special adaptations of the ATP synthase play roles in OXPHOS in the native B. pseudofirmus OF4 setting. The adaptations were recognized when this alkaliphile’s atp operon was first sequenced and compared with sequence data from two other extreme alkaliphiles, Bacillus halodurans C-125 and Bacillus alcalophilus in comparison with sequences from neutralophilic bacteria. Several striking sequence deviations from the neutralophile consensus sequence were found in regions of the a- and c- subunits that had been shown to have roles in H+ translocation through the ATP synthase [45, 224, 225]. Further sequence data from a more diverse group of alkaliphilic bacteria, including alkaliphilic Bacillus species as well as some cyanobacteria, thermoalkaliphiles and Gram-negative alkaliphiles have already been instructive. The alkaliphile-specific sequence deviations from the neutralophile consensus are found among Bacillus species that possess secondary cell wall polymers such as glycoprotein S-layers or teichoic/teichuronic acids that attach to the peptidoglycan (see Fig. 3) but lack an outer membrane that is found in Gram-negative bacteria [132-135]. So far, these sequence features are not apparent in the limited number of genome atp operon sequences available for Gram-negative alkaliphiles. This raises the possibility that the outer membrane of Gram-positive bacteria acts as a barrier to the equilibration of H+ with the external medium and that such a barrier has a bioenergetic impact. Cyanobacterial alkaliphiles whose ATP synthase is sequestered in thylakoids, also lack the c- and a-subunit features possessed by the extremely alkaliphilic Bacillus species. Among the alkaliphilic Bacillus species, in which the alkaliphile-specific sequence features are found, the “fullness” of the recognized panel of sequence deviations that are found in particular alkaliphile strains correlates with how extreme the alkaliphily is in the strain. However, if the organism is a “poly-extremophile”, e.g. a thermoalkaliphile, there are a few shared alkaliphile-specific sequence features but there are even more poly-extremophile deviations from the consensus sequence that are different from those in bacteria that are only alkaliphiles. Thus far, studies testing the physiological role of alkaliphile-specific adaptations in ATP synthase in a native alkaliphile setting have all been conducted in B. pseudofirmus OF4, which is genetically tractable so that mutations can be introduced into the chromosomal atp operon [44, 172]. Use of the native setting is important since, the adaptations evolved in a context with specific homeostasis capacities, membrane lipids and respiratory chain components all in place [45, 160, 216, 226].
As shown in the alignment shown in Fig. 4, the c-subunits of extremely alkaliphilic Bacillus species such as B. pseudofirmus OF4, B. halodurans C-125 and B. alcalophilus have two major motifs that differ from the consensus motif for non-alkaliphilic Bacillus species. The first of these is the AxAxAxA (or at least three Ala residues) replacing the neutralophile consensus GxGxGxG motif in the middle of the N-terminal helix of the hairpin-like subunit. The more moderate alkaliphile Bacillus clausii has GxAxGxA, only two Ala substitutions, and the largely fermentative moderately alkaliphilic marine bacterium Oceanobacillus iheyensis has the non-alkaliphlic Bacillus consensus sequence of GxGxGxG. By contrast, thermoalkaliphile Bacillus sp. TA2.A1 has a GxSxGxS, which is a different variant than that of the alkaliphile-alone deviation from the consensus. The larger serine residues of the Bacillus sp. TA2.A1 subunit have been shown to increase the diameter of the thermoalkaliphile c-ring . This may facilitate a stoichiometry of 13 subunits in this Bacillus or in alkaliphilic Bacillus strains in general, e.g. as opposed to 10 in yeast or E. coli. However, use of an altered GxGxGxG motif that enlarges the ring diameter cannot be a general requirement for a high c-subunit stoichiometry since the largest stiochiometry to date, at 15 c-subunits/ring, is found in alkaliphilic cyanobacterium S. platensis, which has the neutralophilic consensus GxGxGxG [20, 63].
A panel of single, double, triple and quadruple Ala → Gly mutations was introduced into the AxAxAxA motif encoded by the atpE gene in the native chromosomal atp operon of alkaliphilic B. pseudofirmus OF4. Although there was no significant effect on the levels of membrane-associated β-subunit, ATPase (hydrolytic activity activated by octylglucoside) or growth on glucose, the mutations clearly affected non-fermentative growth and ATP synthesis capacity. With increasing numbers of Gly substitutions, there was progressively poorer non-fermentative growth both at pH 7.5 and 10.5, with the effect being somewhat greater at high pH. The quadruple mutant with a complete GxGxGxG motif replacing the native AxAxAxA grew normally on glucose but exhibited almost no non-fermentative growth. Almost no ATP synthesis was observed by the mutant synthase, as assayed in ADP + Pi-loaded right-side-out membrane vesicles from the quadruple mutant alkaliphile cells in comparison with wild type cells . Perhaps the AxAxAxA motif is necessary for the assembly of an optimal, somewhat high c-subunit complement in the particular rotor scaffold of alkaliphilic Bacillus species (Fig. 4).
Interestingly the “x” residues of the motif have some specificity. The earliest attempt to probe the importance of the AxAxAxA motif in B. pseudofirmus OF4 was carried out by substitution of the entire GAGIGNG found in Bacillus megaterium for the AGAIAVA found in B. pseudofirmus OF4, and there was almost no ATP synthase found in the membranes . Later mutagenesis showed that a single mutation of the first “x”, the Gly17 that follows the first Ala16 of the B. pseudofirmus OF4 c-subunit to the Ala17 found in B. megaterium caused a significant loss of membrane-associated enzyme . The importance of the “x” residues in this motif is also clear in the Na+-translocating ATP synthases/ATPases in which a Gln residue between the final Gly residues of the GxGxGQG motif in those strains is part of the Na+-binding motif (see Fig. 4).
The second major sequence deviation that was called the “alkaliphile PxxExxP motif” by Arechaga and Jones  is the presence of two Pro residues that flank the key c-subunit carboxylate of the C-terminal helix (Glu54 in B. pseudofirmus OF4). Usually only a single Pro is found three residues away on the periplasmic (C-terminal) side of the Glu. That Pro is replaced by Thr in S. platensis and by Gly in Na+-translocating enzymes (Fig. 4). It is the first Pro, Pro51, which is novel in the alkaliphile PxxExxP motif. This motif is again not found in either thermoalkaliphilic Bacillus sp. TA2.A1 or O. iheyensis. Replacement of Pro51 by Ala resulted in almost complete loss of ATP synthetic capacity at pH 10.5 and a smaller deficit at pH 7.5 in vesicle assays ; its replacement with Gly was compatible with significant synthetic capacity but there was proton leakiness that presumably accounts for significant growth deficits observed at high pH even on glucose . These results highlight the special importance of strategies that increase proton-tightness in the alkaliphile setting.
A third sequence deviation of extreme alkaliphiles is a threonine residue, Thr33, that directly precedes the conserved RQPE loop of the c-subunit on its N-terminal side next to the loop Arg34. It is found only in extremely alkaliphilic Bacillus species such as the three top species shown in Fig. 4. Ala (or Gly) is otherwise usually found at that position. Mutagenesis of the B. pseudofirmus OF4 Thr33 to Ala led to an interesting phenotype in which overall growth yield on malate at pH 10.5, but not pH 7.5, was significantly impaired. However, ATP synthesis upon reenergization of starved cells was more robust than that of wild-type at pH 10.5 and was accompanied by very strong initial pH homeostasis. In vitro, a quick peak of ATP synthesis was observed upon energization of ADP + Pi-loaded vesicles of this mutant that was followed by loss of the ATP . The presence of Thr33 near the loop edge may affect how torque is generated in some subtle way that results in strong initial torque that is quickly dissipated in the context of the alkaliphilic Bacillus scaffold.
The alignment in Fig. 5 highlights (in blue) the Lys180 that is found in the a-subunits of alkaliphilic Bacillus species that grow non-fermentatively at high pH, including thermoalkaliphilic Bacillus TA2.A1. In general the residue at this position (Gly218 in E. coli and other non-alkaliphiles) is considered to be part of a trio of residues that has critical roles in successful capture and translocation of H+ . The second one is the neighboring Glu (Glu181 in alkaliphiles and Glu219 in E. coli) that is conserved among H+-, but not Na+-translocating ATP synthases (Fig. 5). The third is highlighted (in blue) in several alkaliphiles in which it is Gly, Gly212 in B. pseudofirmus OF4. The residue at this position is different in other bacterial types, Ser in non-alkaliphilic Bacillus species, His in E. coli and Asp in Na+-translocating ATP synthases. Several lines of evidence indicate that the residue at positions of the Lys180 and Gly212 of B. pseudofirmus OF4 interact within the ion uptake pathway from the external surface. This pathway leads to the a- and c-subunit interface at which H+ are transferred to the rotor during ATP synthesis [172, 228-230]. The positive charge of the conserved Arg is critical in creating an environment that facilitates the exchange of H+ at this interface and is also highlighted in Fig. 5 (red). Consequences of mutagenesis of the Lys180 and/or Gly212 of B. pseudofirmus OF4 to the Bacillus consensus (Gly for Lys180 and Ser for Gly212) show that the presence of Lys180 in the putative H+ uptake path is critical for H+ capture and successful ATP synthesis at high pH but not as critical at near neutral pH. A further role of both Lys180 and Gly212, perhaps in concert, is indicated by the finding that they help prevent H+ loss to the outside bulk phase, e.g. during transient drops in pmf . The effect of several replacements of Lys180 in the a-subunit of thermoalkaliphlic Bacillus sp. TA2.A1 ATP synthase was also studied using an experimental system in which the enzyme was expressed in E. coli. The thermoalkaliphile synthase does not support non-fermentative growth in E. coli but can be studied in ADP + Pi-loaded vesicles with intra-and extravesicular pH values that support synthetic activity. These experiments showed that this enzyme retains significant synthetic ability with Arg and His as well as Gly replacing Lys180, with the pH profile for activity depending upon the pKa of the replacing residue . Preliminary evidence indicates that the B. pseudofirmus OF4 enzyme does not tolerate Arg replacement at this position , perhaps further reflecting the extensive differences between the adaptations of the H+-translocating subunits extremely alkaliphilic Bacillus and thermoalkaliphlic Bacillus strains (see Fig. 5).
There is another region of particular interest in the a-subunit of extremely alkaliphilic Bacillus species that merits noting. The external loop between helices II and III of the a-subunit, from a circled Glu, Glu98 to the putative border of helix III shown in Fig. 5 (and a bit further, to the circled Ser114) is generally more polar than the comparable region of non-alkaliphilic Bacillus species or E. coli . Replacement of the loop part of this region with the sequence found in B. megaterium resulted in a significant loss of capacity for non-fermentative growth and ATP synthesis at pH 10.5 but not at pH 7.5 . This loop may play a role in H+ capture on the external surface of the membrane, a possibility that will require more experimental investigation.
The extensive structure-function data already gathered for ATP synthases reveal variations in the synthases from diverse organisms that have adapted to a wide range of environmental challenges. Studies of these variations and the basis for their adaptive values will continue to enrich our understanding of the plasticity of ATP synthase function. They will also continue to shed new light on basic mechanistic principles that are common to many types of synthases and on particular mechanistic adaptations that are in place in specific groups of synthases. For example, the aerobic alkaliphilic Bacillus species have challenged and enlarged models of H+-coupling in general and will be an important paradigm in which to test these models in a native setting. The alkaliphile also appears also to have specific adaptations in connection with its unique and central challenge of alkaline pH homeostasis. Among the many future goals of interest are the following.
Experimental approaches that provide defined parameters for further models of surface-associated pmf or microcircuits need to be developed in alkaliphile settings. These parameters will facilitate evaluation of contributions of delocalized near surface H+ vs H+ whose localized transfer depends upon particular respiratory chain complexes, particular membrane lipids and/or a close proximity of the respiratory chain complex(es) to ATP synthases. One of the parameters would be experimental estimation of the pH near surface of the cytoplasmic membrane of B. pseudofirmus OF4 or a similar organism. Since extreme Gram-negative alkaliphiles appear not to require the F0 adaptations required for OXPHOS by Gram-positive alkaliphiles, it would be interesting to ascertain the extent to which their bioenergetic challenges differ from those of the well studied alkaliphilic Bacillus species and the role of the outer membrane.
Further investigations of the specific adaptations of the ATP synthase will continue to provide important information about how the alkaliphile overcomes specific aspects of its bioenergetic challenges, which will also lead to a more refined understanding of what those challenges are. As new structural data complement the work on these adaptations the insights will be much greater. Work has not yet been reported on equally interesting motifs that have been observed in respiratory chain complexes features [159, 226, 233]. Such features and motifs may facilitate specific partnering between respiratory chain complexes and alkaliphile ATP synthases either directly or in close proximity.
It remains possible that an extremely alkaliphilic bacterium or archaean will emerge from some as-yet untapped environmental niche that does compensate for the energetically adverse effect of alkaline pH homeostasis with a higher transmembrane electrical potential than heretofore observed. Or, an alkaliphilic organism may be found that resolves the energetic conundrum by using an ATP synthase rotor with an extraordinarily large number of subunits. It is also important to probe the basis for the tolerance of a high cytoplasmic pH in alkaliphiles and to probe the limits of such tolerance in diverse alkaliphiles. Perhaps it will be a major part of the resolution of the bioenergetic problem in some organisms. Nature often surprises.
The work that was conducted in the authors’ laboratory was supported by research grant GM28454 from the National Institute of General Medical Sciences of the National Institute of Health (to TAK).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.