Write a comment

This review reflects comments and contributions from Elena Sena, Femi Arogundade & Pablo Ranea-Robles. Review synthesized by Pablo Ranea-Robles.

This study by Xiao et al. delves into an intriguing connection between metformin effects on body weight and a recently discovered exercise-induced metabolite, Lac-Phe (N-lactoyl-phenylalanine). Metformin is an anti-diabetic drug that has been used for many years, and is one of the most used anti-obesity drugs. The mechanism of action of metformin has remained elusive for many years, even though recent evidence from Hirst lab suggest that Complex I inhibition is the major mechanism of action of metformin. Authors report that metformin increases Lac-Phe levels in mice and humans, and that mice that cannot produce Lac-Phe (CDNP2 KO mice) are resistant to the anti-obesity effects of metformin, but not to the anti-diabetic effects of this drug. In vitro data in this study suggest that metformin induces Lac-Phe in macrophages and gut epithelial cells in a CDNP2-mediated manner. This study has several strengths, such as the use of CDNP2 KO mice, which impairs Lac-Phe production by deleting the required enzyme in mice, allowing mechanistic studies. The generation of CDNP2-deficient cells for in vitro experiments is another strength of this study. Another one is the use of data from two human cohorts treated with metformin, in which authors measured Lac-Phe levels and performed a mediation analysis to infer the contribution of Lac-Phe in the effect of metformin in humans. We also found several weaknesses, related to the causal role of OXPHOS inhibition and glycolysis changes in this Lac-Phe mediated phenotype, and the lack of in vivo evidence to claim a cell-autonomous role of metformin-induced Lac-Phe in mice. Below, we provide several suggestions to the authors to make their conclusions more robust.

Major comments:

• A key experiment that would support that Lac-Phe mediates anti-obesity metformin effects would be to inject Lac-Phe at the concentration reached in circulation after metformin injection in mice. This concentration is quite lower to what has been shown before to inhibit appetite and decrease body weight (Li et al 2022; DOI: 10.1038/s41586-022-04828-5), so until that experiment is performed it remains doubtful that Lac-Phe alone is mediating the observed effect of metformin. Considering the big effect observed in CDNP2 KO mice, which are resistant to metformin anti-obesity effects, the fact that other lactate-derived metabolites could mediate this effect is not even mentioned in the discussion and we suggest including it to provide a better contextualization of the observed effect in these mice.

• As we understood from this study, a limitation that is not stated anywhere in the text is that there was no body weight change in the 21 subjects from the Stanford study treated with metformin. Since a central part of this study is that Lac-Phe has an effect on body weight, this result limits the human relevance of this metabolite in the control of energy balance. At least, the reasons for this discrepancy should be discussed by the authors.

• Lac-Phe levels in metformin-treated chow-fed mice will be a relevant experiment to understand how metformin acts on Lac-Phe metabolism.

• The part in the study where authors report the differences between WT and CDNP2 KO when fed a high-fat diet are a bit confusing. First, authors report a genotype effect on body weight gain. However, these body weight data are not shown anywhere in this study, making it complicated to assess them. Moreover, when reporting the GTT data authors state that body weights between WT and CDNP2 KO mice were equal after 8 weeks of HFD. We suggest showing BW graphs for those experiments and clarifying these aspects.

• One of our major comments relates to the claim that metformin-induced Lac-Phe-mediated effects are mediated by OXPHOS inhibition. The data provided by the authors is only correlational and in vitro, and even though they mention this nuance in the text, the heading used in the results section suggest a causal role of OXPHOS inhibition in this phenotype. Moreover, it would be interesting to see the effects of OXPHOS inhibition on Lac-Phe production in cell types in which metformin does not induce Lac-Phe (hepatocytes for example). Is that inability to induce Lac-Phe production due to metformin not acting on those cells or because the proposed mechanism does not work on those cell types?  Without further evidence, we suggest toning down the way this result is reported.

• Does Lac-Phe m+3 labeling change if glycolysis is inhibited? Does induction of glycolysis alone (independent of OXPHOS inhibition) induce Lac-Phe levels? Does glycolysis inhibition abolish Lac-Phe induction by metformin? Those studies are necessary to understand the mechanisms behind Lac-Phe induction by metformin that right now remains only associative.

• The authors claim a cell autonomous effect of metformin on Lac-Phe induction. This claim is based on in vitro data using cell lines from different tissues, which lacks the more complex in vivo context. Since this is a limitation of the study, we suggest mentioning it and tone down the conclusions on that part of the study.

• Is CDNP2 expression in the different cell lines related to Lac-Phe induction by metformin?

  Minor comments:

• The target/receptor of Lac-Phe is not known yet. We think that it should be stated in the discussion/introduction for a better comprehension of the background of this study

• What do authors speculate about the source of remaining Lac-Phe in CDNP2 KO cells?

• We missed some more discussion about the connection between metformin, its mechanism of action and CDNP2 enzyme.

• Authors report the levels of appetite hormones between WT and CDNP2 KO. However, it is not clear whether these hormone levels were measured before or after these mice were fed a high-fat diet. This should be clarified. If these measurements were made in chow-fed mice, that would be a limitation for obtaining a conclusion on whether these hormones play a role or not on the phenotype.

• Reference to data shown in Figure 3D and part of Figure 5 is lacking in part of the results text (part relative to metformin effects on body weight and food intake in WT and CDNP2 KO mice)

• There is no discussion regarding the fact that some human subjects do not show an increase in Lac-Phe after metformin administration. What do the authors think about that? Is that correlated to any outcome they could measure in that population?

• Authors should add reference(s) regarding the proposed mechanisms of action of metformin, such as inhibition of endogenous glucose production in the liver (Introduction)

• There is a seminar reference missing in this article regarding metformin action of complex I by Hirst lab (Bridges et al. 2023; PMID: 36701435)

• It would be great if authors can provide full blots for CDNP2 antibody, which helps in reproducibility efforts for other researchers when they use the same antibody.

• Why do some control subjects show negative levels of Lac-Phe in Fig 1D? Graph or figure legend do not state that those levels represent post vs pre Lac-Phe levels making it a little bit confusing.

• Figure 1F y-axis label says “Lac-Phe” where it should state “Lactate”

• It would be interesting to add Lac-Phe absolute concentrations instead of relative change, so readers can better evaluate and comprehend the changes induced by Metformin.

• The connection between GDF15 and Lac-Phe is not clear in the study, and may generate a bit of confusion in the reader if that connection is not expanded a bit

• It would be interesting to show lactate m+3 levels in the glucose tracing data

• X-axis labeling in Figure 3E does not seem to be correct

• The method section lacks information on the validation of the metabolite measurements by LC-MS. Including details on the calibration curves, limits of detection, and precision and accuracy assessments would strengthen the reliability of the metabolite data.

• The section on animal experiments provides information on the treatment of mice with metformin, GDF15, and semaglutide. However, more details on the rationale behind choosing specific doses, the frequency of administration, and the methods for administering substances (e.g., oral gavage, subcutaneous injection) would be valuable. Additionally, information on randomization and blinding procedures could enhance the experimental rigor.

• It could be interesting for the reader to know more about the commonalities and differences in the mechanisms underlying the anti-obesity effects of exercise and metformin

• The mediation analysis exploring the contribution of Lac-Phe to metformin's anti-obesity effects in humans is a unique aspect of the study. However, it raises questions about the broader metabolic effects of metformin and whether Lac-Phe is a key mediator or one of several contributing factors.Comments on reporting - information on the statistical analyses or availability of data.

• The fact that data from only 21 out of 31 subjects in the Stanford study was included in the analysis should be mentioned on the results text, not only in the Methods

• Some statistical analysis of repeated measures are performed with Welch t-test (for example, Fig 5D, 5G). Why did authors choose this test instead of a two-way ANOVA?

• How are multiple comparisons accounted for in Fig. 2A? 

  Suggestions for future studies:

• Investigate other metabolites produced by CDNP2, characterizing the products of that enzyme. Analyze the relationships between Lac-Phe and other metabolites to build a comprehensive metabolic profile. This could involve metabolomic studies to uncover broader metabolic changes associated with metformin treatment.

• Extend the findings to preclinical models beyond mice to better understand the translational potential of metformin-induced changes in Lac-Phe. Consider using other animal models that may better mimic certain aspects of human metabolism.

• Investigate the factors contributing to individual variability in metformin response, particularly in terms of Lac-Phe levels and body weight changes. Genetic, environmental, and lifestyle factors may play a role in determining the magnitude of metformin's effects.

• Undertake long-term, prospective studies to assess the sustained effects of metformin on body weight and metabolism. This could involve monitoring individuals over extended periods to observe changes in Lac-Phe levels and body weight over time.

Competing interests

The author declares that they have no competing interests.