The effects of ENDOG on lipid metabolism may be tissue-dependent and may not require its translocation from mitochondria (2024)

arising from W. Wang et al. Nature Communications 10.1038/s41467-023-41757-x (2023)

In a recent article, Wang et al.¹ state that, upon lipid stimulation, the mitochondrial nuclease Endonuclease G (ENDOG) translocates to the cytoplasm where it interacts with the endoplasmic reticulum (ER) and the mTORC2/AKT/ACLY pathway. This triggers lipid synthesis, promoting non-alcoholic fatty liver disease (NAFLD) in mice. Discovering novel mechanisms involved in NAFLD is relevant because it is a prevalent condition that can cause steatohepatitis. However, some data in the article by Wang et al.¹ contradict their assertions and are open to reinterpretation.

ENDOG is a DNase/RNase in mammalian mitochondria^2,3 that translocates to the cytoplasm and nucleus during cell death to degrade nuclear DNA^4,5. ENDOG participates in mtDNA replication^2,6,7, controls cardiomyocyte growth^6,7, and regulates cell proliferation⁸. In relation to the findings presented by Wang et al.¹, we reported that Endog gene knockout promotes lipid accumulation in the heart⁶, and limits adipose enlargement in mice⁹.

Wang et al.’s results show that deletion of Endog in mice using the CRISPR/Cas9 system¹⁰ decreases lipid accumulation in the liver of fasted animals. Their Fig. ​Fig.1a1a¹ presents transmission electron microscopy (TEM) images (reproduced here in Fig.1a) and Bodipy staining (Fig.1b) showing that during fasting, a condition previously suggested by the same authors to induce ENDOG translocation to the cytoplasm¹⁰, the hepatocytes of Endog^-/- mice contain a reduced number of lipid droplets (LD) compared with those of Endog^+/- mice^1,10. Unfortunately, the experiment lacks the analysis of fed mice, limiting the conclusions that can be drawn. Nevertheless, the cancellation of ENDOG expression in HepG2 cells cultured in a standard medium also reduced lipid accumulation¹. Triglyceride deposition was diminished in ENDOG KO HepG2 cells under both normal conditions and after exposure to oleic acid (Fig.1c). These in vitro findings suggest that without exiting the mitochondria, ENDOG can regulate lipid synthesis and/or accumulation. Consequently, the necessity for ENDOG translocation to the cytoplasm and its interaction with the endoplasmic reticulum (ER) appears to be unnecessary.

Of note, we previously showed that Endog^-/- mice, which have the same C56BL/6 genetic background¹¹ as those from Wang et al. show LD accumulation in the cardiac muscle (data in ref. ⁶ and new data in Fig.1d), and in the liver (Oil Red O, Fig.1e; Bodipy, Fig.1f) compared to Endog^+/+ mice, contrasting with the findings shown by Wang et al.¹. It is possible that housing conditions may explain conflicting phenotypes at some extent. However, an independent report showed increased liver vacuolization in Endog^-/- mice¹². Considering all this evidence, and in the absence of an analysis of hepatic lipid deposition in both genotypes performed under fed conditions, it is difficult to interpret Wang et al.’s results on the role of ENDOG on hepatic LD accumulation.

The experiments presented by Wang and collaborators in Fig. ​Fig.33¹ are intended to support the main claim that HFD induces cytoplasmatic ENDOG-dependent lipogenesis and NAFLD in mice, as summarized in the diagram of Fig. 7h¹. The authors showed body weight progression and images of adipose tissue (WAT) and other organs of Endog^+/- and Endog^-/- female mice fed HFD (reproduced here in Fig.2a, b) but failed to show the results from mice fed a control diet. To our understanding, it is crucial to know the progression of body weight, WAT, and liver mass at the end of the period of Endog^+/- and Endog^-/- mice fed regular chow diet, otherwise the impact of HFD on the promotion of lipogenesis by ENDOG is uncertain. These control data are essential since previous results from our group demonstrate that Endog^-/- male mice have lower body weight gain and a reduced WAT mass compared to Endog^+/+ mice fed regular diet⁹ (Fig.2c, d). Of note, the adipose tissue was not investigated in the initial characterization of the same Endog mutant mice strain¹¹. In addition, the authors only show hematoxylin/eosin-stained liver sections of Endog^+/- and Endog^-/- mice at the end of the HFD treatment¹ but do not show TEM images as in their Fig.1a, nor do they show images from mice fed a control diet, which makes comparison with the initial state difficult.

Another aspect that does not fit with the final conclusions relates to the alterations detected in the mTORC2-AKT-ACLY lipogenic pathway. The Supplementary Figure S24 of Wang et al.¹ shows the same differences in mTOR/AKT/ACLY pathway phosphorylation due to ENDOG deficiency in both control and HFD (reproduced here in Fig.2e). These data agree with our previous results showing inhibition of the AKT pathway by ENDOG deficiency in rodent and human cells^7,8. This suggests that an HFD is not instrumental in the ENDOG-dependent effects on lipid metabolism. Therefore, Wang et al.’s in vivo results demonstrate that the relevance of ENDOG in the control of lipid metabolism is not dependent on the fat content of the diet. Thus, the title, the conclusions such as “In summary, we revealed that ENDOG is released from mitochondria under HFD (…) resulting in Acetyl-CoA production” (from their Discussion section¹), and the summary diagram, which is later used in a review by the same authors¹³ is not fully supported by the presented experimental data.

To justify ENDOG exit from mitochondria with mild stress, such as HFD/lipid treatment, in the absence of mitochondrial disruption, Wang and collaborators state that ENDOG resides in the mitochondrial intermembrane space (IMS). However, the authors do not demonstrate this assertion but refer to an article by Ohsato et al.¹⁴. In that report, the authors incubated mitochondria with a hypotonic medium followed by increasing digitonin and proteinase K treatment and observed ENDOG degradation. They suggested an IMS location of ENDOG, as the mitochondrial inner membrane (MIM) should protect matrix proteins from digestion. However, a more recent work by David and co-workers used density gradients and sequential centrifugations to show that ENDOG is located mainly in the mitochondrial matrix¹², where it can access mtDNA allowing its proven function on mtDNA replication^2,6,7. Taking this into account, ENDOG is most probably located in the mitochondrial matrix, or, in any case, its location is uncertain. Release of ENDOG from the matrix could be plausible in certain conditions as described for mtDNA in response to mitochondrial stress^15,16. However, the statement by Wang et al. in their text and their final diagram, Figure 7h¹, convey the unproven concept that ENDOG exits from the mitochondrial IMS without major alterations in mitochondrial dynamics, which is most probably not the case if residing in the mitochondrial matrix.

Lipid-induced ENDOG translocation from mitochondria to the cytoplasm is the key event from which Wang and collaborators build all the intracellular signaling proposed to be involved in ENDOG-induced lipogenesis and NAFLD. The authors insist throughout the main text and the final diagram that ENDOG must exit the mitochondria to interact with proteins in the cytosol and ER, where it displaces Rictor from 14-3-3 and triggers the ER stress response, activating TAG synthesis and NAFLD. However, the authors do not show any evidence of HFD-induced ENDOG translocation but only addressed lipid-induced ENDOG release in vitro in the hepatoblastoma HepG2 treated with oleic acid (OA)¹ (Fig.3a, b). A number of outstanding control studies would be required to reinforce the claims, such as: i) similar WB of ENDOG in cytosolic and mitochondrial fractions from ENDOG WT and KO HepG2 cells with and without OA treatment, and ideally primary hepatocytes from Endog^+/- and Endog^-/- mice under HFD and control diet to demonstrate that the faint bands are ENDOG, and ii) WB images overexposed to the same extent as ENDOG detection, of cytoplasmatic extracts blotted with antibodies against other mitochondrial proteins such as Cytochrome c (IMS), Tim23, SDHA (SDH complex, IMM), and Citrate Synthase (matrix), to know whether the procedure used can cause any mitochondrial damage. In the absence of these controls, the identity of the bands cannot be determined, and the integrity of mitochondria is unknown, hampering the drawing of conclusions. In addition to these drawbacks, the authors did not assess the specificity of ENDOG immunofluorescence (IF) detection, despite they have the appropriate mice and cellular models to do this. The IF images of ENDOG “release” still show punctuated “ENDOG” signal suggesting staining of ENDOG or other proteins still within mitochondria, despite the lack of TIM23 signal in the selected cell locations (Fig.3c). Moreover, the analysis of protein interactions and activity took place only under non physiological conditions in overexpressing settings and/or involving chaperones using total cellular extracts (i.e., including mitochondrial proteins) of HepG2 and hepatocellular carcinoma (MHCC97-H, HCCLM3) cell lines inaccurately identified as hepatocytes.

In summary, the article by Wang et al.¹ shows that the mitochondrial nuclease ENDOG is important for the regulation of lipogenesis in the liver and WAT in mice. However, we argue that key assertions about the role of ENDOG on lipid metabolism made by Wang and colleagues have not been conclusively established. The experimental conditions involving the results on the influence of ENDOG on hepatic LD accumulation in control conditions are confusing; no evidence is presented of HFD-induced ENDOG translocation to the cytoplasm in hepatocytes; experiments of OA-induced ENDOG release in the HepG2 hepatoblastoma lack essential controls; the HFD studies lack the follow-up and final analysis of Endog^+/- and Endog^-/- mice fed regular diet, and analysis of the lipogenic mTOR/AKT/ACLY signaling shows that it is altered by the lack of ENDOG but unaffected by HFD. Therefore, considering the complete set of data published and presented here, combined with previous results, the most plausible conclusion is that ENDOG influences lipogenesis due to effects that depend on its mitochondrial role in a tissue-specific manner through hitherto unidentified mechanisms.

Methods

Acknowledgements

Ministerio de Ciencia e Innovación (MICINN), Gobierno de España, grant numbers PID2019-104509RB-I00 and PID2022-139965OB-100, and AGAUR, Generalitat de Catalunya, Catalunya, grant number 2021_SGR_00758 support the research in ML and DS laboratory. MICINN grant number PID2022-137576OB-I00 to A.Z. PERMISSIONS: Oxford University Press permits reproduction of Figure 4 panels c and e from Pardo et al.⁹ in Fig.2, with License number 5676540546080.

Author contributions

L.G. performed the experiments with Oil Red O and Bodipy and contributed to writing and editing the manuscript. A.Z. made relevant conceptual contributions to mitochondrial biology and edited the manuscript. M.L. and D.S. contributed to the conception and design of the work, analysis, and interpretation of data, and wrote, and edited the manuscript. All authors approved the final version.

Peer review

Competing interests

The authors declare no competing interests.

Footnotes

References