A common East-Asian ALDH2 mutation causes metabolic disorders and the therapeutic effect of ALDH2 activators

Obesity and type 2 diabetes have reached pandemic proportion. ALDH2 (acetaldehyde dehydrogenase 2, mitochondrial) is the key metabolizing enzyme of acetaldehyde and other toxic aldehydes, such as 4-hydroxynonenal. A missense Glu504Lys mutation of the ALDH2 gene is prevalent in 560 million East Asians, resulting in reduced ALDH2 enzymatic activity. We find that male Aldh2 knock-in mice mimicking human Glu504Lys mutation were prone to develop diet-induced obesity, glucose intolerance, insulin resistance, and fatty liver due to reduced adaptive thermogenesis and energy expenditure. We find reduced activity of ALDH2 of the brown adipose tissue from the male Aldh2 homozygous knock-in mice. Proteomic analyses of the brown adipose tissue from the male Aldh2 knock-in mice identifies increased 4-hydroxynonenal-adducted proteins involved in mitochondrial fatty acid oxidation and electron transport chain, leading to markedly decreased fatty acid oxidation rate and mitochondrial respiration of brown adipose tissue, which is essential for adaptive thermogenesis and energy expenditure. AD-9308 is a water-soluble, potent, and highly selective ALDH2 activator. AD-9308 treatment ameliorates diet-induced obesity and fatty liver, and improves glucose homeostasis in both male Aldh2 wild-type and knock-in mice. Our data highlight the therapeutic potential of reducing toxic aldehyde levels by activating ALDH2 for metabolic diseases.

impaired fatty acid oxidation in BAT. The latter is hypothesized to account for the observed differences in energy expenditure.
The manuscript is well written, easy to understand and deals with an important and innovative research question, the role of the ALDH2 Glu504Lys mutation on energy and glucose metabolism. Using elegant studies, the authors link a well-known human mutation to a functional relevance for energy and glucose metabolism. The manuscript is considered as of importance for the field.  weight. The current gold standard is to blot body weight (g, x-axis) against total energy expenditure (kcal/h; Y-axis) followed by ANCOVA analysis using body weight as co-variate (PMID: 23520145; PMID: 23520145). On another note, the authors show in Figure 1A that body weight is significantly higher in the KI mice relative to WT. But in Fig. 2a, (despite suboptimally corrected), the energy expenditure is only slightly different at 7 (out of 24) time points. I doubt that the total mean energy expenditure will be significantly different.
Furthermore, those rather minimal differences at only a few post-hoc time points cannot explain the body weight difference shown in Fig. 1a. Does the ANCOVA give a significant genotype effect? Also, the two peaks in energy expenditure differences occur at the onset of the dark phase (I assume) and the shortly before the light goes on (I assume). Those two occasions typically coincide with increased meal intake in mice (they show typically an increase in food intake at the onset of the dark phase and shortly before the lights go on).
So I doubt that energy expenditure differences can really explain the body weight data. Are genes related to thermogenesis different in BAT of the KI mice (Ucp1, Pgc1a, prdm16, cidea9? Is isoproterenol stimulation of energy expenditure different between wt and KI mice? What is the energy expenditure during acute and chronic cold exposure? The authors say that the KI mice have no difference in basal metabolic rate. Notably, basal metabolic rate refers to the lowest energy expenditure of a post-absorptive (fasted) animal at thermoneutrality. What the authors describe here is the resting metabolic rate, not basal metabolic rate. RMR is described as the lowest energy expenditure of a resting animal at a given temperature.
Food intake should please be given as cumulative food intake measured over multiple weeks. A simple bar graph is not adequate to rule out cumulative differences over time.
The authors say to have measured cold-induced and diet-induced thermogenesis in Fig. 2f and g. But the authors show here rectal temperature. This is not cold-induced and dietinduced thermogenesis. The authors need to measure energy expenditure in the cold and after feeding a HFD. Assessment of rectal temperature cannot be declared as cold-induced thermogenesis. Also, the observation that rectal temperature is lower in the KI mice does not necessarily mean that also adaptive thermogenesis is different. The mice could also have a conductance phenotype, means they lose more energy over the skin and therefore show lower core temperature. If the mutation e.g. causes vasodilation, then the mice could lose more energy over the surface. So we can't say that lower rectal temperature means that adaptive thermogenesis is different. And the authors report in this figure no data on whether BAT function is impaired. What about expression of key thermogenic genes in BAT This manuscript suggested the activation of ALDH2 (by ALDH2 activator) reduces toxic aldehyde levels, and eventually it will be applicable for treating metabolic diseases. Based on the results obtained, the authors claimed that ALDH2 KI (human mutant gene GLu504Lys) were more susceptible to high fat diet high sucrose diet (HFHSD)-induced fatty liver and obesity, and this was due to reduced energy expenditure/impaired thermogenesis and increased aldehyde adduct formation. In addition, ALDH2 KI exhibited reduced insulin sensitivity and impaired glucose tolerance. <Major points> -The authors used homozygous ALDH2 KI mice (homozygote Aldh2*2/*2) in this study.
-The authors clearly demonstrated ALDH2 KI mice are more sensitive to obesity and insulin intolerance. Based on the results the authors presented, there was no further information what the major factor is to affect mitochondrial fatty acid oxidation, and the production of 4-HNE. One possibility is that long-term high fat diet might change intestinal microbiota changes and has a chance to produce endogenous alcohol (auto-brewery syndrome in human). In addition, high conc. of sugar might accelerate to produce endogenous alcohol, and eventually leads alcohol mediated adduct 4-HNE production. Therefore, the evaluation of blood alcohol concentration (BAC) after HFHSD in WT and ALDH2 KI might be required, and then the authors will have more ideas about the underlying mechanisms.
-In figure1d, the authors need to explain more how to evaluate fat mass, lean mass, free fluids, and total water. Especially, the measurement of free fluids, and total water. -The authors also used "#" in the figure set. What does # mean? -Line136, the authors evaluated rectal temperature 30 minutes after HFHSD feeding. Could the authors describe the way how to provide HFHSD to experimental mice at the same time for rectal temperature evaluation. Did not the authors provide HFHSD ad-libitum? -The authors evaluated mouse activity after HFHSD feeding and ALDH2 KI showed grooming behavior was slightly increased. It does not provide any further information/implication, it needs to be removed.
-After treatment of AD-9308, did the authors observe any adverse effects/side effects?
Gross/histological changes of organs or serological parameters (such as ALT, AST, BUN, creatinine, bil, amylase and so on).
<Minor points> -"Aldh2 WT and KI mice" and "Aldh2 KI and WT mice" are mixed in manuscript and figures.
It must be written with uniformity -Many spelling errors (typo) and space error are found. The authors must check them thoroughly throughout the manuscript and figures.
-fig2f, the sample information is missing -fig4b, the information of X axis is missing -fig5e,5h, what is the meaning of #?
-In materials and methods, tail blood "glucose" was collected  tail blood was collected -In materials and methods, please cross-check 58% calorie  58% kilocalorie, 12.5% calorie 

Comments from reviewer
Reviewer #1 (Remarks to the Author): The paper entitled 'A common East Asian-specific ALDH2 mutation causes obesity and insulin resistance: therapeutic effect of reducing toxic aldehydes by ALDH2 activation' by Yi-Cheng Chang and co-workers explores the behaviour of Aldh2*2/*2 homozygous knock-in (KI) mice, mimicking human Glu504Lys mutation, in developing diet-induced obesity, glucose intolerance, insulin resistance, and fatty liver on a high-fat high-sucrose diet compared with controls.
A key point of the paper is the proteomic analyses of the brown adipose tissue of the Aldh2 KI mice in which an increase of 4-HNE-adducted proteins involved in fatty acid oxidation and electron transport chain has been reported. In the present form, any MS data is consistently presented.
(1) First of all, all the raw data of the LC-MS/MS analysis has to be loaded on a public proteome identification database to give the opportunity to the reviewers to access to the results using the opportune code.

Answer:
We have uploaded to all raw mass spectrometry data to the ProteomeXchangepublic dataset base(https://massive.ucsd.edu/ProteoSAFe/static/massive.jsp). The accession code is MassIVE00091724. (2) A table reporting all the MASCOT results such as Mascot Score, empai, pep matches and sig matches has to be provided, as well.After this, I will be happy to review this part of the paper.

Answer:
We also have added the SI file in the revised Supplementary Table 2 and as follows   Table S2. SI table of identified 4-hydroxynonenal (4-HNE)-adducted protein by liquid chromatography tandem mass spectrometry (LC-MS/MS) 2. Another crucial point of the paper is the comparison of the expression levels of many different proteins or protein categories in between the KI and ctr samples. Correctly, in the immunoblotting analysis, the authors show the signal relative to GAPDH and/or HSP70 as loading control. Unfortunately, the loading is not uniform in many lanes of several experiments (such as Figure 3B, S2 and S6) and they don't comment this incongruity. Usually, when it happens, there are two ways to proceed: the researchers can make a densitometric analysis of each signal of immunoblotting and report the intensity of each band taking into account the intensity of the loading control on a histogram or they can reload the samples to obtain a more uniform loading control. The authors have absolutely to ameliorate these data and they have to provide the uncropped western blot analysis in original.

Answer:
We have added the densitometry analyses and histograms of the immunoblots in the revised manuscript. Please also see revised    In this manuscript, Chang and colleagues assessed the role of the East Asian-specific ALDH2 mutation on obesity and glucose metabolism in knock-in mice for this mutation. The authors report that mice that carry the mutation show increased body weight, glucose intolerance and hepatic steatosis. The KI mice further displayed reduced energy expenditure and impaired fatty acid oxidation in BAT. The latter is hypothesized to account for the observed differences in energy expenditure.
The manuscript is well written, easy to understand and deals with an important and innovative research question, the role of the ALDH2 Glu504Lys mutation on energy and glucose metabolism. Using elegant studies, the authors link a well-known human mutation to a functional relevance for energy and glucose metabolism. The manuscript is considered as of importance for the field. As stated in the Methods of our previous manuscript, we used mice with nearly identical body weight at the age of 10-12 weeks for every metabolic rate measurement and core temperature measurement in our study. The measured metabolic rates were further corrected by lean mass. We think the problem caused by different body weights (and different organs with different metabolic rates) will be minimized by this approach. Figure 1A that body weight is significantly higher in the KI mice relative to WT. But in Fig. 2a, (despite sub-optimally corrected), the energy expenditure is only slightly different at 7 (out of 24) time points. I doubt that the total mean energy expenditure will be significantly different. Furthermore, those rather minimal differences at only a few post-hoc time points cannot explain the body weight difference shown in Fig. 1a. Does the ANCOVA give a significant genotype effect? Also, the two peaks in energy expenditure differences occur at the onset of the dark phase (I assume) and the shortly before the light goes on (I assume). Those two occasions typically coincide with increased meal intake in mice (they show typically an increase in food intake at the onset of the dark phase and shortly before the lights go on). So I doubt that energy expenditure differences can really explain the body weight data.

On another note, the authors show in
Answer: Based on comments of reviewer 3, we performed addition experiments to see whether Aldh2 heterozygous knock-in (HE) mice are also more susceptible to diet-induced obesity. We repeated the energy expenditure measurement of Aldh2 KI, HE ,and WT mice using CLAMS system, as follows (n=19:27:13) using mice with nearly identical body weight. The energy expenditure of Aldh2 WT mice is significantly higher than KI and HE mice, which can explain the weight difference between three genotypes. Please see revised

Energy expenditure
Metabolic measurements (food and water intake, locomotor activity, VO2 consumption and VCO2 production) were obtained using the Columbus Instruments' Comprehensive Lab Animal Monitoring System (CLAMS-HC). Monitoring was performed for 3 days after mice have been acclimatized to the cages for 3 days." 4. Are genes related to thermogenesis different in BAT of the KI mice (Ucp1, Pgc1a, prdm16, cidea9? Answer: We have performed the immunological blots and RT-qPCR for these genes between BAT of WT and KI mice and found no difference in expression of all these thermogenesis genes. Please see revised

Supplementary Figure S8
Please refer to reply to Point 11.

Is isoproterenol stimulation of energy expenditure different between WT and KI mice?
Answer: We have performed isoproterenol-stimulated energy expenditure for mice of three genotypes (dose: 20 mg/kg, single subcutaneous injection). A significantly increase in energy expenditure was found in wild-type mice (WT) as compared to Aldh2 homozygous knock-in (KI) and heterozygous knick-in (HE) mice. Please

What is the energy expenditure during acute and chronic cold exposure?
Answer: Please refer to the answer for Point 9. 7. Fig. 2AThe authors say that the KI mice have no difference in basal metabolic rate. Notably, basal metabolic rate refers to the lowest energy expenditure of a post-absorptive (fasted) animal at thermoneutrality. What the authors describe here is the resting metabolic rate, not basal metabolic rate. RMR is described as the lowest energy expenditure of a resting animal at a given temperature.

Answer:
We thank the review's correction have corrected the term " basal metabolic rate" to "resting metabolic rate". 8. Fig. 2B Food intake should please be given as cumulative food intake measured over multiple weeks. A simple bar graph is not adequate to rule out cumulative differences over time.
Answer: We thank the reviewer's suggestion and measured dietary intake by weighting diet consumed by two mice of the same genotype in their home cage weekly to avoid stress cause by isolation and frequent interruption for accumulative 21 days. Please see revised

Methods
Accumulative diet intake was measured by weighting diet consumed by two mice of the same genotype in their home cage weekly for 3 weeks " Fig. 2f and g. But the authors show here rectal temperature. This is not cold-induced and diet-induced thermogenesis.

The authors say to have measured cold-induced and diet-induced thermogenesis in
The authors need to measure energy expenditure in the cold and after feeding a HFD. Assessment of rectal temperature cannot be declared as cold-induced thermogenesis. Also, the observation that rectal temperature is lower in the KI mice does not necessarily mean that also adaptive thermogenesis is different. The mice could also have a conductance phenotype, means they lose more energy over the skin and therefore show lower core temperature. If the mutation e.g. causes vasodilation, then the mice could lose more energy over the surface. So we can't say that lower rectal temperature means that adaptive thermogenesis is different.
Answer: We fully agree with the reviewer. Therefore, we measured diet-induced and cold-induced core temperature change and energy expenditure (as well as all experiments regarding core temperature and energy expenditure) using mice of the age of 10-12 weeks with nearly identical body weight to avoid the concern of body surface area and conductance according to the recommendation of an international guide for  tolerance test when mice is acutely transferred to5°C, they are forced to rely on shivering to defend its body temperature. The capacity and endurance of the shivering is unable to last for long time, and gradually nonshivering thermogenesis will take over". These statements and our finding suggested that acute cold tolerance test is not adequate measurement of non-shivering thermogenesis.  14. If fatty acid oxidation, and hence fatty acid supply to BAT, is causal for the phenotype, then supplementation of free fatty acids should reverse this effect.Hence, such a rescue experiment would be valuable to consolidate this hypothesis.
Answer: We thank the reviewer's constructive suggestion.
Consistent with the whole BAT tissue, the FAO rate of the cultured primary brown adipocytes isolated from the Aldh2 KI mice was also significantly decreased by 39.3% (P=0.02) compared with WT mice, which could be rescued by addition of 0.2 and 0.6 μCi 3 H-palmitate dose-dependently (P-for-trend=0.045) ( Fig. 3l). Please see revised Fig.3k&3l and as follows:

Serum biochemistry
Blood was collected from mice fasted for 4 hours. Fasting serum free fatty acid concentration (cat. no. LabAssay 294-63601, Wako) were measured using enzymatic assay kits according to manufacturer's instruction" 18. In Fig. 5 I miss data on food intake, energy expenditure, RER, locomotor activity, BAT function, FA oxidation, plasma level of FFA, TAG etc. The current figure is lacking quite a few metabolic data that would be helpful to better understand these data Answer: We fully agree with the reviewer's comments. However, because of the limitation of core facility during the pandemic, lack of enough mice for experiment (the phenotype of Aldh2 heterozygous knock-in (HE) mice are required for revision), and the lack of enough AD-9308 compound required for AD-9308 treatment (20 weeks), it is very difficult for us to perform similar assays for AD-9308-treated mice.

Reviewer #3 (Remarks to the Author):
This manuscript suggested the activation of ALDH2 (by ALDH2 activator) reduces toxic aldehyde levels, and eventually it will be applicable for treating metabolic diseases. Based on the results obtained, the authors claimed that ALDH2 KI (human mutant gene GLu504Lys) were more susceptible to high fat diet high sucrose diet (HFHSD)-induced fatty liver and obesity, and this was due to reduced energy expenditure/impaired thermogenesis and increased aldehyde adduct formation. In addition, ALDH2 KI exhibited reduced insulin sensitivity and impaired glucose tolerance.
<Major points> 1.The authors used homozygous ALDH2 KI mice (homozygote Aldh2*2/*2) in this study. However, in human, most people with ALDH2 variants are heterozygous ALDH2*2/*1. Therefore, heterozygous ALDH2 KI (Aldh2*1/*2 knock-in) should be tested to see whether heterozygous ALDH2*2/*1 are also more susceptible to HFHSD-induced obesity and metabolic dysfunction. Answer: We have tested whether heterozygous Aldh2 knock-in mice (HE) are susceptible to HFHSD-induced obesity and metabolic dysfunction in the Figure 1&2 and as follows.  We found elevated 4-HNE levels and increased 4-HNE adducted proteins due to reduced activity of ALDH2 of the brown adipose tissue (BAT) from the Aldh2 homozygous KI mice. Proteomic analyses of the BAT from the Aldh2 KI mice identified increased 4-HNE-adducted proteins involved in mitochondrial fatty acid oxidation (FAO) and electron transport chain (ETC), leading to markedly decreased FAO and mitochondrial respiration of BAT, which is essential for adaptive thermogenesis and energy expenditure.

Discussion
BAT is a highly specialized organ enriched in Ucp1 for adaptive thermogenesis. Although Aldh2KI mice exhibited impaired thermogenesis, unexpectedly, they did not have altered Ucp1 or associated thermogenesis gene expression. Instead, we found that the ALDH2 enzymatic activity is reduced and the 4-HNE level is increased the BAT from Aldh2KI mice. Proteomics screening found that several key mitochondrial proteins involved in mitochondrial fatty acid oxidation (FAO) and electron transfer train were modified by 4-HNE adduction, leading to markedly (~70%) reduced FAO and mitochondrial respiration. Consistently, previous studies have shown that 4-HNE is mainly generated from oxidation of mitochondrial membranes, with 30% of 4-HNE-adducted proteins located within mitochondria 35,36 . We further found the serum free fatty acid level is increased and the respiratory exchange rate (RER) measured by indirect calorimetry is increased in Aldh2KI mice, further indicating impaired fatty acid utilization.
Mitochondrial FAO and ETC are required for the maintenance of the proton gradient in the intermembranous space, which is essential for Ucp1-mediated adaptive thermogenesis. In our study, we found that the thermogenic capacity of Aldh2KI mice was reduced. Fatty acids serve as the main fuel suppliers for thermogenesis 37 . It has been estimated that fatty acids in the BAT contribute 74-84% of the fuel for thermogenesis 37 .
Cpt1 is the rate-limiting enzyme for the translocation of fatty acids into mitochondria for βoxidation.Cpt1b +/mice developed fatal hypothermia following cold challenge 38 . Adipose-specific Cpt2knockout mice presented a hypothermic phenotype when exposed to cold 39 . Mice deficient in fatty acid βoxidation enzymes, including very-long-chain acyl-CoA dehydrogenase (VLCAD), long-chain acyl CoA dehydrogenase (LCAD), and short-chain acyl CoA dehydrogenase (SCAD)also displayed cold intolerance [40][41][42] . These data indicate that mitochondrial FAO is critical for adaptive thermogenesis. Furthermore, BATspecific Lkb1-knockout mice, which have reduced expression of ETC complex proteins, also developed impaired thermogenesis 43 , indicating that the integrity of mitochondrial ETC machinery is essential for adaptive thermogenesis. These data strongly support our findings that 4-HNE adduction to mitochondrial proteins involved in mitochondrial FAO and ETC could lead to impaired adaptive thermogenesis.
Collectively, these data indicate that the significantly reduced ALDH2 activity of Aldh2KI results in elevated toxic aldehydes levels and, especially 4-HNE and increased 4-HNE adduction to proteins involved in mitochondrial reduced FAO and mitochondrial respiration of BAT, leading to markedly decreased FAO rate and mitochondrial respiration and subsequent reduced adaptive thermogenesis and energy expenditure.
The reduced thermogenesis and energy expenditure result in die-induced obesity and associated metabolic disorders including fatty liver, insulin resistance, and glucose intolerance.

Methods:
Animal behavior including awakening, drinking, feeding, grooming, hanging, resting, twisting, walking, and rearing up were recorded by the Clever HomeCage Scan system 3.0 for 24 hours after acclimation for 3 days." Answer: We thank the reviewer's comments.
(1) The gross appearance of mice was shown in revised Figure 5d and as follows. The gross appearance of perigonadal fat and liver was shown in Figure 5e&5jand the histology of perigonnadol fat and liver is shown in Figure 5i&l and as follows

Pathological examinations
Pathological examination was performed by the contract pathological core service of the Animal Centers of MedicalCollege, NationalTaiwanUniversity. Pathological changes of liver and kidney from Aldh2 KI and WT mice treated with 0, 20 or 60 mg/kg/day of AD-9308 for 20 weeks was examined and scored by H&E stain.
Serum alanine aminotransferase (ALT) and creatinine levels of Aldh2KI and WT mice treated with 0, 20 or 60 mg/kg/day of AD-9308 for 20 weeks were assayed using the FUJI DRI-CHEM clinical chemistry analyzer." and wild-type mice (WT) " throughout the manuscript.

Line88, reference number was duplicated 27,27-29
Answer: We thank the review for the careful reading and have corrected this error. Fig. 1E, put BAT, fat, and liver next to images Answer: We have added the label of tissue to the images. Please see revised Figure 1dand as follows  Table S3)."

4.
to " In vivo, AD-9308 administration showed a favorable pharmacokinetic profile in mice when administered orally or intravenously with high bioavailability in rodents and dogs (Supplementary Table S3)." 7. Fig. 2F, the sample information is missing Answer: We thank for the reviewer's correction and have added additional information in the legend of revised Fig. 2f&g as follows.

Fig. 4B, the information of X axis is missing
Answer:We thank the reviewer's correction. We have revised the legend of Figure. 2b as follows: Figure 2 (b) AD-5591(100μM) or alda-1(100μM) significantly increases the enzymatic activity of incorporation of Atp5a1 (originally assigned to wild-type-specific 4-HNE-adducted proteins) to Atp5f1a (originally assigned to 4-HNE-adducted proteins shared by both Aldh2 wild-type and knock-in mice) do not change the conclusion because the identified 4-HNE-adducted proteins specific to knock-in mice remain same. Please see the revised Results and Fig. 3i&j as follows.
"Using liquid-chromatography tandem mass spectrometry (LC MS/MS) analysis, we identified 19 4-HNE-adducted brown adipose tissue mitochondrial proteins in Aldh2 knock-in mice and 9 4-HNE adducted mitochondrial proteins in wild-type mice, with 8 proteins which are present in both Aldh2 knock-in and wild-type mice (Fig. 3i, 3j)."    Fig. 16 and Table S2 Our response: We thank the reviewer's constructive comments and the useful reference. All energy expenditure among three genotypes was re-analyzed using the generalized linear model according to international guidance (reference 60, 61). Energy expenditure (expressed as kcal/hr) of each mouse was regressed on fat and lean mass using the command "glm" implemented in STATA 14.0 and the three genotypes were coded by allelic doses as 0,1, and 2 for wild-type, heterozygous, and homozygous knock-in mice, respectively. We thank the assistant professor Pi-Hua Liu of the Clinical Informatics and Medical

Figure 3i
Statistics Research Center of Chang Gung University in Taiwan for statistical assistance.

"Statistical analyses
Energy expenditure among three genotypes was analyzed using the generalized linear model according to international guidance 60,61 . Energy expenditure (expressed as kcal/hr) was regressed on fat and lean mass using the command "glm" implemented in STATA 14.0 and the three genotypes were coded by allelic doses as 0,1, and 2 for wild-type, heterozygous, and homozygous knock-in mice, respectively 60,61 .