Mitochondrial ATP transporter depletion protects mice against liver steatosis and insulin resistance

Non-alcoholic fatty liver disease (NAFLD) is a common metabolic disorder in obese individuals. Adenine nucleotide translocase (ANT) exchanges ADP/ATP through the mitochondrial inner membrane, and Ant2 is the predominant isoform expressed in the liver. Here we demonstrate that targeted disruption of Ant2 in mouse liver enhances uncoupled respiration without damaging mitochondrial integrity and liver functions. Interestingly, liver specific Ant2 knockout mice are leaner and resistant to hepatic steatosis, obesity and insulin resistance under a lipogenic diet. Protection against fatty liver is partially recapitulated by the systemic administration of low-dose carboxyatractyloside, a specific inhibitor of ANT. Targeted manipulation of hepatic mitochondrial metabolism, particularly through inhibition of ANT, may represent an alternative approach in NAFLD and obesity treatment.

The lean phenotype is very clear and there is a suggestion that the liver uncoupled respiration is increased. However, I don't think that it's well supported tha t this is due to overall increases in whole-body energy expenditure and this should be tested. This is the biggest question in my mind.
I have a minor question regarding the ANT1 blot. It almost appears that this protein is also diminished.

Reviewer #2 (Remarks to the Author):
A The question raised here is to explain the findings that the ANT deletion in liver has not the expected life threatening effects given the central role of the ANT in ATP supply. Mitochondria remain intact h on deletion ANT. Interestingly, the synthesis of mitochondria is even increased . Since respiration with substrates such as pyruvate is normal, an uncoupling has to be invoked. Since the membrane potential is not decreased the uncoupling has to be subtle. Uncoupling proteins are not involved. The nature of the uncoupling remains unknown . A major disease related aspect of this paper are the fat metabolism changes associated with the ANT levels. Fatty liver disease on over-nutrition is suppressed by ANT deletion. A similar prevention of the fatty liver or even of general obesity has the injection of the inhibitor CAT in normal mice. B.This is an interesting paper where paradoxical phenomena challenge conventional wisdom. C ok D ok E The conclusions should be modified according to fortcoming results (see F) F inhibitor CAT in normal mice. The following amendments are suggested. In order to understand the status of the ANT deleted mitochondria the content of ATP synthase (AS) is needed. AS activity should be decrease in ANT depleted mitochondria. On the other hand AS might be responsible for the uncoupling. For example if ANT forms super-complexes with AS . The lack of ANT might destabilize the AS dimers which are known to form pores in the "permeability transition". Thus AS might play a key role for understanding the ANT deficient mitochondrial metabolism G. ok H. ok

Reviewer #3 (Remarks to the Author):
Cho et al analyze hepatocyte-specific ANT2 null mice, which exhibit low body weight, protection from HFD induced adiposity, and protection from fatty liver. These phenotypes are associated with hepatocyte mitochondrial proliferation, abrogated ADP-stimulated respiration, and indirect evidence of uncoupled respiration that does not require UCP2. The authors suggest ANT2 may be a target for NAFLD. The studies are interesting, but would improve from modest experimental revision. More confirmation of phosphorylationindependent proton leak, driving increased substrate oxidation, is needed.
1. Uncoupled respiration has been supported, but not demonstrated, because mitochondrial number is increased, and state 2 respiration is normalized to mitochondrial mass. A respiration experiment in liver tissue or isolated hepatocytes should be performed to demonstrate increased state 2 respiration.
2. Formal measurements of proton leak should be made, e.g. as performed in Cell Metab. 2006;3:417-27. Membrane potential appears to be normal in ANT2KO. To confirm proclivity toward uncoupling, rates of membrane potential-dependent proton leak should be measured using escalating malonate concentrations (which collapse membrane potential) 3. The hypothesis that F1F0 ATPase may favor proton leak and ATP hydrolysis is attractive, and easy to assess using oligomycin. If it is not the F1F0 ATPase, then there are other potential candidates.
4. Liver is typically 20% of resting energy expenditure. It will almost certainly by much higher in these animals. At the very least, whole animal VO2/energy expenditure must be measured.
5. The source of the Ant2fl allele is not clear. Is this new or the same as that in Nature 427, 461-465 (2004)? If it's the same allele, then Fig. 1a should move to supplemental. If it's a new construction, then more details should be given in the methods.
6. Give body weights on HFD.
7. The term pseudo-fasting should be eliminated. This is not at all the fasting state, even with weight loss and mild ketosis.

RESPONSE TO REVIEWERS
We do appreciate a thorough and positive review of the manuscript and very helpful suggestions provided by the reviewers. We have carefully considered each of the points raised by the reviewers, documented in the point-by-point responses to their concerns. We have addressed all the important criticisms and also provided the new data. Thanks to all the reviewers' helpful suggestions, we believe that the revised version of this manuscript has been significantly improved.
All changes in the text are highlighted in yellow. Figures 3d, 3e and 5

The lean phenotype is very clear and there is a suggestion that the liver uncoupled respiration is increased. However, I don't think that it's well supported that this is due to overall increases in whole-body energy expenditure and this should be tested. This is the biggest question in my mind.
We tested whole-body energy expenditure of the mice under a high fat diet using a CLAMS. Although it tended to be slightly higher in the Ant2 cKO mice, there was no statistical difference. These new data are now included in the manuscript as Fig. 5d, and discussed in the text.
I have a minor question regarding the ANT1 blot. It almost appears that this protein is also diminished.
Yes, we have repeatedly seen the tendency that expression of Ant1 protein but not mRNA was decreased in the Ant2-knockout liver. We have not seen a similar phenomenon in the liver of the systemic Ant2 hypomorphic mice (Cell Death Differ 22, 1437, 2015 or the heart of Ant2 fl/y -MHC6Cre mice (not published). We initially thought that this could be due to MIM impairment potentially caused by the complete loss of Ant2 protein in the liver mitochondria. However, as shown in the manuscript, MIM integrity has been well preserved. At this moment, we do not know the reason. It should be noted, however, that there is a residual (<4%) activity of ADP/ATP exchange in the Ant2 cKO liver mitochondria (Fig. 3a), which is likely mediated by Ant1, since it was sensitive to carboxyatractyloside.

Reviewer #2, A. The question raised here is to explain the findings that the ANT deletion in liver has not the expected life threatening effects given the central role of the ANT in ATP supply. Mitochondria remain intact h on deletion ANT.
Interestingly, the synthesis of mitochondria is even increased. Since respiration with substrates such as pyruvate is normal, an uncoupling has to be invoked. Since the membrane potential is not decreased the uncoupling has to be subtle. Uncoupling proteins are not involved. The nature of the uncoupling remains unknown. A major disease related aspect of this paper are the fat metabolism changes associated with the ANT levels. Fatty liver disease on over-nutrition is suppressed by ANT deletion. A similar prevention of the fatty liver or even of general obesity has the injection of the inhibitor CAT in normal mice. B. This is an interesting paper where paradoxical phenomena challenge conventional wisdom. We newly tested the effects of oligomycin on respiration of Ant2-depleted liver mitochondria using Seahorse XF24 (Fig. 3d). Interestingly, oligomycin did not affect their basal respiration status; thus the data implied that the F1F0 ATPase is unlikely involved in proton leak here. These data are now included and discussed in the revised manuscript.

Reviewer #3,
Cho et al analyze hepatocyte-specific ANT2 null mice, which exhibit low body weight, protection from HFD induced adiposity, and protection from fatty liver. These phenotypes are associated with hepatocyte mitochondrial proliferation, abrogated ADP-stimulated respiration, and indirect evidence of uncoupled respiration that does not require UCP2. The authors suggest ANT2 may be a target for NAFLD. The studies are interesting, but would improve from modest experimental revision. More confirmation of phosphorylation-independent proton leak, driving increased substrate oxidation, is needed.

Uncoupled respiration has been supported, but not demonstrated, because mitochondrial number is increased, and state 2 respiration is normalized to mitochondrial mass. A respiration experiment in liver tissue or isolated hepatocytes should be performed to demonstrate increased state 2 respiration.
We agree that it would be nice to demonstrate an increase in overall respiration per cell. However, unfortunately we were not able to measure respiration of perfused hepatocytes due to technical difficulties. However, to confirm the data shown Fig. 3c, we newly measured oxygen consumption rate of liver mitochondria using a Seahorse XF24 as well. As shown Fig. 3d, we demonstrated that basal respiration per mitochondria mass is similar or slightly higher in cKO, with no response to ADP at all. We hope these along with the following phosphorylation-independent respiration data would strengthen our discussion here. made, e.g. as performed in Cell Metab. 2006;3:417-27.

Formal measurements of proton leak should be
Membrane potential appears to be normal in ANT2KO. To confirm proclivity toward uncoupling, rates of membrane potential-dependent proton leak should be measured using escalating malonate concentrations (which collapse membrane potential).
We did measure phosphorylation-independent respiration and responses to malonate using an O2k respirometer. The Ant2-deleted liver mitochondria showed a significantly higher rate of phosphorylation-independent respiration and lower sensitivity to malonate. These new data are now included as Fig. 3e and Extended Data Fig. 3c. It should be noted, however, that we were not able to measure membrane potential simultaneously since we do not have a multi-sensor for the purpose.
3. The hypothesis that F1F0 ATPase may favor proton leak and ATP hydrolysis is attractive, and easy to assess using oligomycin. If it is not the F1F0 ATPase, then there are other potential candidates.
We newly tested the effects of oligomycin on respiration of Ant2-depleted liver mitochondria using Seahorse XF24 (Fig. 3d). Interestingly, oligomycin did not affect their basal respiration status; thus the data implied that the F1F0 ATPase is unlikely involved in proton leak here. These data are now included and discussed in the revised manuscript.
4. Liver is typically 20% of resting energy expenditure. It will almost certainly by much higher in these animals. At the very least, whole animal VO2/energy expenditure must be measured.
We tested whole-body energy expenditure of the mice under a high fat diet using a CLAMS. Although it tended to be slightly higher in the Ant2 cKO mice, there was no statistical difference. These new data are now included in the manuscript as Fig. 5d, and discussed in the text. Nature 427, 461-465 (2004)?

The source of the Ant2fl allele is not clear. Is this new or the same as that in
If it's the same allele, then Fig. 1a should move to supplemental. If it's a new construction, then more details should be given in the methods.
The mice have been originally generated by us and are different from those Doug Wallace's group has described (Nature 427, 461, 2004). Exons 2 & 3 are floxed in our mice, whereas Exons 3 & 4 are floxed in theirs. Since we have described our Ant2 gene targeting strategy more in details at our previous publication on systemic Ant2 hypomorphic mice (Cell Death Differ 22, 1437, 2015), we provided minimal information here. In addition, we are currently depositing our Ant2 cKO mice to the Jackson Lab to make them available to other investigators. This information is now included at the method section.

Give body weights on HFD.
We now provide such data at Fig. 5a. Furthermore, we performed more comprehensive analyses regarding the effects of Ant2 cKO on HFD-induced obesity and related metabolic changes. We included these results as an independent figure (Fig. 5).
7. The term pseudo-fasting should be eliminated. This is not at all the fasting state, even with weight loss and mild ketosis.
We removed the term from the text.
We thank all the reviewers again for their very helpful and constructive criticisms. We do believe that the revised manuscript has been significantly improved by incorporating their suggestions and advice.