Reply to Comment by Quistorff: ATP is not consumed solely by hydrolytic reactions

Abstract

A Comment to our recent paper that described a budget for brain metabolic water production claimed that all ATP produced by oxidation of glucose is consumed by hydrolysis, and that the net calculated production of metabolic water is equal to that obtained by combustion of glucose. However, ATP is synthesized and consumed by enzymatic reactions that do not involve water in the mechanism. Not all ATP consumed is hydrolyzed.

We welcome the opportunity to engage in scientific discussion of the points raised in the Comment by Quistorff [1] regarding our evaluation of brain metabolic water production [2]. He made two major points: (i) we “seem to have forgotten” that steady state maintains constant ATP concentration because ATP hydrolysis consumes the same amount of water as produced during ATP synthesis, and (ii) “an interesting aspect, not touched upon,….the intracellular location of consumption and production….might create a directional water flux in the cell. Albeit very small.” In fact, both of these points were addressed throughout the text and in Figs. 1 and 2 [2], yet they were overlooked by Quistorff. Our paper was entitled, “A budget for metabolic water production…” A budget is comprised of production and utilization estimates, not precisely-measured values. Quistorff ignored our ‘line-item expense’ of 28 of the 38 total water that was assigned to ATP hydrolysis for ion pumping, leaving 10 available for secretion. In contrast, he assumed all 32 ATP are hydrolyzed giving a net of 6 water from oxidation of one glucose. This discordance is not a new issue.

Response to Point 1. Not all ATP synthesis liberates water and not all ATP is hydrolyzed

A debated topic: how much metabolic water is available for secretion?

A major point of our review was that the ‘textbook version’ of metabolic water production by glucose oxidation in Eq. [1] is likely an underestimate.

[1] 1 glucose = C₆H₁₂O₆ + 6 O₂ → 6 CO₂ + 6 H₂O. Mass balance equation for glucose combustion.

As explained in detail in [2] the sequential enzymatic reactions that convert glucose to CO₂ and H₂O maximize ATP production. The enzymatic steps involving water production (+) or consumption (–) per glucose oxidized (two trioses are generated per glucose) are summarized below:

In cytosol

+ 2 Enolase: 2 2-phosphoglycerate → 2 phosphoenolpyruvate + 2 H₂O.

In mitochondria

–2 Citrate synthase: 2 Acetyl-CoA + 2 oxaloacetate + 2 H₂O → 2 citrate + 2 CoASH.

–2 Fumarase: 2 fumarate + 2 H₂O → 2 L-malate.

+ 12 Electron transport chain, Cytochrome c oxidase: 6 O₂ + 24 H⁺ + 24 e⁻ → 12 H₂O.

+ 28 Oxidative phosphorylation, ATP synthase: 28 ADP + 28 Pi → 28 ATP + 28 H₂O.

[2] + 38 Total metabolic water produced.

[3] + 10 Net (38 − 28 assigned to ATP hydrolysis) available for secretion.

[4] 1 glucose + 32 ADP + 32 Pi + 6 O₂ → 6 CO₂ + 38 H₂O + 32 ATP.

Notes:

(i) 6 O₂ are converted to 12 H₂O via the electron transport chain without ATP synthesis.

(ii) We evaluated consumption of metabolic water in the section entitled, “Influence of ATP turnover on metabolic water utilization” and in other places in the paper [2]. This concept was also presented in Figs. 1 and 2, yet this material was apparently ignored by Quistorff. ATP turns over rapidly, so the 28 water and 28 ATP produced by ATP synthase are both consumed in 1:1 stoichiometry by ATP hydrolysis. Dissipation of the proton gradient across the mitochondrial inner membrane by uncouplers separates oxidative phosphorylation from electron transport and glucose oxidation per se, and we, therefore, assigned 28 water to ATP hydrolysis. We stated that the 28 water produced in mitochondria were probably consumed during steady state by cytosolic ATP hydrolysis, mainly by ion pumps, e.g., Na⁺,K⁺-ATPase, Ca⁺⁺-ATPase, and H⁺-ATPase [2]. Estimates of the energy demands of brain Na⁺,K⁺-ATPase are on the order of ~ 65% of ATP (see p. 951–2 in [3]), or 21 ATP. Hydrolytic consumption of 65% of the total metabolic water by Na⁺,K⁺-ATPase accounts for 21 water.

(iii) We also discussed [2] the discrepancy of 10 water available for secretion in [3] vs. the 6 water produced in mass balance of combustion [1]. This issue was debated by others in earlier papers, with the possible explanation that a resonance structure of inorganic phosphate that participates in substrate-level phosphorylation reactions could mimic water by providing an oxygen to the recipient molecule, thereby consuming the equivalent of 4 of the 10 water for a net of 6 [4–6]. However, the phosphate is not water participating in the reactions.

Calculated water consumed by hydrolysis of all ATP

Quistorff [1] stated (i) ATP is synthesized with production of water and (ii) ATP is hydrolyzed consuming water. He concluded that all of the 32 ATP produced in [4] are hydrolyzed, consuming 32 H₂O, and he calculated net water produced as follows:

[5] Total H₂O minus H₂O for ATP hydrolysis = 38−32 = “properly corrected” 6 H₂O [1].

The assumptions made by Quistorff [1] are not correct, and his calculation in [5] is flawed. First, ATP synthesis can occur without producing water, as in substrate-level phosphorylation [6, 7] or phosphoryl transfer [10] reactions. Second, ATP can be utilized by phosphoryl transfer reactions that do not involve hydrolysis [8, 9, 11, 12, 13]. Quistorff provided no proof that all ATP is consumed by hydrolytic reactions.

ATP turnover reactions that do not involve water in the mechanism

(i) Substrate-level phosphorylation reactions synthesize ATP without producing water [7].

[6] Two glycolytic reactions act sequentially to transfer Pi to ADP:

Glyceraldehyde 3-phosphate (P) dehydrogenase: 2 Glyceraldehyde-3-P + 2 Pi + 2 NAD⁺ →

2 1,3-bisphosphoglycerate + 2 NADH + 2 H⁺.

Phosphoglycerate kinase: 2 1,3-Bisphosphoglycerate + 2 ADP → 2 3-phosphoglycerate + 2 ATP.

[7] Succinyl CoA synthetase: 2 Succinyl-CoA + 2 Pi + 2 GDP → 2 succinate + 2 CoASH + 2 GTP.

(ii) Examples of phosphoryl transfer reactions by kinases that can synthesize or consume ATP without water in the mechanism [7–12].

[8] Hexokinase: Glucose + 1 ATP → Glucose-6-phosphate + 1 ADP.

[9] Phosphofructokinase: Fructose-6-phosphate + 1 ATP → fructose-1,6-bisphosphate + 1 ADP.

[10] Pyruvate kinase: 2 Phosphoenolpyruvate + 2 ADP → 2 pyruvate + 2 ATP.

[11] Adenylate kinase: AMP + ATP ↔ 2 ADP.

[12] Protein kinases: γ-phosphorus of ATP is transferred a serine, threonine, or tyrosine residue in side chains of proteins.

[13] Biotin-dependent enzymes (e.g., pyruvate carboxylase, acetyl-CoA carboxylase, propionyl-CoA carboxylase, methylcrotonyl-CoA carboxylase) transfer the γ-phosphorus of ATP to a substrate.

Reevaluation of the glucose oxidation metabolic water budget

In reactions [6, 7, 10] a total of 6 ATP (ATP = GTP) are produced per glucose oxidized, and in reactions [8, 9] 2 ATP are consumed, all without involving water, for a net of 4 ATP synthesized. ATP synthase produces 28 ATP + 28 water, giving a total of 32 ATP per glucose oxidized. If these 4 ATP were hydrolyzed, then 4 water would have to come from the 10 tallied as available for secretion [3]. Then, the net water would be 6, as suggested by Quistorff [5] and [1], also resolving the 10 vs. 6 issue [4–6] discussed above.

However, the fraction of ATP hydrolyzed is not precisely known, and there are many small molecule kinases, protein kinases, and biotin-dependent carboxylases that consume ATP (e.g., [8, 9, 11, 12, 13]) without hydrolysis. The discrepant amount of 4 water is ~ 10% of the 38 total produced. It is certainly possible that various kinases and carboxylases metabolize ~ 10% of the total ATP, with ion pumps and other hydrolytic reactions consuming most, but certainly not all, ATP.

For example, brain pyruvate carboxylase is active in astrocytes: pyruvate + ATP + HCO₃ (or CO₂) → oxaloacetate + ADP + Pi. This anaplerotic reaction is involved in glutamate, glutamine, and aspartate turnover, serving to maintain glutamine levels during excitatory neurotransmission. The rate of this reaction determined in vivo in awake, non-stimulated rats was 15–20% of total glucose oxidation, depending on the metabolic model used in the calculations [13]. In this study, pyruvate carboxylase activity consumed 15–20% of the 32 ATP produced per glucose, equivalent to 5–6 water that is not used for hydrolysis. Here, the adjusted amount of water subtracted from the total in [5] would be 38−32 + 5–6 = 11–12, a net amount similar to our calculated estimate for secretion [3].

To sum up, the net 6 H₂O per glucose oxidized in [1] and [5] is probably too low. Depending on neuronal metabolite kinase, protein kinase, and biotin-dependent carboxylase activities, metabolic water available for secretion may be closer to or even slightly exceed 10 [3]. Small molecule kinases (e.g., [8, 9] and ethanolamine and choline kinases, both of which are present in synaptosomal cytosol [14]) use phosphoryl transfer mechanisms [11, 15], as do the families of protein kinases [10, 12]. In vivo fluxes mediated by these enzymes are probably small compared with pyruvate carboxylase, and the amounts of ATP consumed by these processes are not known.

Response to Point 2. Intracellular water fluxes are directional from mitochondria to cytosol

Figures 1 and 2 and discussion throughout the text in [2] noted that nearly all of the water generated by oxidation of one glucose (36 out of 38) is produced in mitochondria, whereas ATP hydrolysis mainly occurs in the cytoplasm via ion pumps. We stated that there must be a directed flow of metabolic water from mitochondria to cytosol, then to interstitial fluid, and did ‘touch on this interesting aspect but very small flux’ of metabolic water production. This flux is substantial: ~36 water flow out of mitochondria to cytosol per 2 pyruvate per glucose oxidized.

Summary

Our brain metabolic water production budget [2] is the calculated net amount of water generated by glucose oxidation, based on textbook-literature enzyme reaction mechanisms. The amounts of metabolic water assigned to ATP hydrolysis and secretion were estimates, i.e., guidelines for future studies of water fluxes in brain. We suggested that secretion of some metabolic water contributes to metabolite efflux and waste product clearance from brain via the perivascular-lymphatic drainage system. Even if the net water production were only 6 per glucose, secretion of this amount approximates the measured perivascular flow rate and would still contribute significantly to interstitial fluid flow. Furthermore, the major conclusions of our paper are still valid: continuous directed flow of metabolic water from cells to interstitial fluid, most metabolic water is produced by neurons that have higher glucose oxidation rates than other cell types, and metabolic water production is reduced during activation and when glucose oxidation rates are lower, as in deep sleep and neurological disorders, such as Alzheimer’s disease and other dementias. A caveat is that the brain volume fraction of astrocytes is small compared with neurons (about 10% and 70–75%, respectively, in cerebral cortex, with 15–20% interstitial space) [16]. Thus, adjustment of the astrocytic and neuronal total water production and secretion rates by multiplying by 1/respective volume fraction would raise the rates by 10- or 1.3-1.4-fold, respectively. This means that the astrocytes have higher rates on a per cell volume basis, but contribute a smaller fraction to the total on a per volume brain basis (equivalent to per g wet weight, as metabolic rates are generally reported).

Brain water fluxes are poorly understood and involve many processes besides fuel metabolism, including co-transport with glucose and lactate [17]. Further study is required. ATP is generated and utilized by different types of enzymatic reactions, some involve water in the reaction mechanism, whereas others do not. All ATP production routes are not equivalent and all ATP fates are not the same.

Author contributions

Both authors were equally involved in developing the concepts, analysing the data and preparing the manuscript. They both read and approved the final version of the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.