The integrated stress response (ISR) is a pro-survival signaling pathway present in eukaryotic cells, which is activated in response to a range of physiological and pathological stressors. Such stresses commonly include cell-extrinsic factors such as hypoxia, amino acid deprivation, glucose deprivation, and viral infection. However, cell-intrinsic stresses such as endoplasmic reticulum (ER) stress (Pakos-Zebrucka et al., 2016), and mitochondrial stress can also activate the ISR (Bao et al., 2016; Quirós et al., 2017; Pereira et al., 2021). Induction of the ISR and its main effector activating transcription factor 4 (ATF4) in response to various stress conditions frequently correlates with increased levels of the endocrine factors fibroblast growth factor 21 (FGF21) and growth differentiation factor 15 (GDF15) in tissues, such as liver (Kang et al., 2021) and skeletal muscle (Ost et al., 2020). However, the upstream mechanisms mediating ISR induction in response to mitochondrial stress are unclear. Furthermore, the respective roles of FGF21 and GDF15 in the context of mitochondrial dysfunction are incompletely understood.

We recently reported that mitochondrial stress caused by deletion of the mitochondrial fusion protein optic atrophy 1 (OPA1 BKO) activated the ISR and induced its main effector ATF4 in brown adipose tissue (BAT). ATF4-mediated FGF21 induction was required to promote browning of inguinal white adipose tissue (iWAT) and improve thermoregulation in OPA1 BKO mice at baseline conditions. Although activation of the ISR in BAT led to resistance to diet-induced obesity (DIO) in an ATF4-dependent manner, this phenomenon was independent of FGF21 (Pereira et al., 2021). In the present study, we sought to investigate the upstream mechanisms mediating ISR activation in BAT in response to OPA1 deletion and the molecular mechanisms downstream of ATF4 mediating resistance to DIO.

Transcriptome analysis showed that the unfolded protein response (UPR), and specifically the protein kinase R (PKR)-like ER kinase (PERK), was induced in OPA1-deficient BAT. Given that PERK is shared by the ISR and the UPR (Pakos-Zebrucka et al., 2016), here, we tested the hypothesis that PERK is required for ATF4 induction in BAT. We also observed that in addition to FGF21, GDF15 was highly induced in BAT upon OPA1 deletion. Because GDF15 has been shown to regulate energy homeostasis in rodents (Tsai et al., 2018a; Chung et al., 2017b), here we tested the hypothesis that GDF15 is downstream of ATF4 and mediates the resistance to DIO in mice lacking OPA1 in thermogenic adipocytes. Our data reveal that PERK is dispensable for ISR and ATF4 induction in BAT in response to OPA1 deletion. Notably, GDF15 partially mediates the resistance to DIO, but is necessary for the improvement in glucose homeostasis and hepatic steatosis following high-fat feeding in OPA1 BKO mice. Furthermore, mice lacking GDF15 were unable to properly thermoregulate during cold exposure. Mechanistically, GDF15 deficiency abrogated browning of iWAT under obesogenic conditions and prevented the induction of calcium cycling in muscle, which likely contributes to increased weight gain and impaired thermogenesis in OPA1 BKO mice. These data underscore the complex regulation of systemic metabolism by various BAT-derived endocrine regulators that are released in response to mitochondrial stress and highlight novel roles for BAT-derived GDF15 in the regulation of energy homeostasis.

In our recently published study, we showed that deletion of the mitochondrial fusion protein OPA1 in brown adipocytes impaired thermogenic capacity in BAT, while paradoxically improving thermoregulation and promoting increased metabolic fitness in mice. These adaptations were mediated by the main effector of the ISR, ATF4, partially via the induction of FGF21 as a batokine. Nonetheless, the resistance to DIO and improvements in glucose homeostasis observed in OPA1-deficient mice occurred in an ATF4-dependent, but FGF21-independent manner (Pereira et al., 2021). Moreover, the molecular mechanisms leading to activation of the ISR in OPA1 BKO mice remained incompletely understood. Therefore, in the present study, we sought to investigate the molecular mechanisms mediating ISR induction and the mechanisms downstream of ATF4 promoting resistance to DIO in mice lacking OPA1 in thermogenic adipocytes.

The ISR can be induced by four different eIF2α kinases that act as early responders to disturbances in cellular homeostasis. Each kinase responds to distinct environmental and physiological stresses, which reflects their unique regulatory mechanisms (Pakos-Zebrucka et al., 2016). We and others have shown induction of the ISR downstream of mitochondrial stress, but the mechanisms mediating this induction are incompletely understood (Pereira et al., 2021; Ost et al., 2020; Forsström et al., 2019; Ost et al., 2016). RNASeq data in OPA1-deficient BAT indicated an increase in ER stress pathways and activation of the UPR in OPA1 BKO mice. Indeed, of the four different ISR kinases, the ER kinase PERK was the only one significantly enriched in our transcriptome analysis of OPA1-deficient BAT. Therefore, we tested the hypothesis that PERK, an eIF2α kinase shared by the UPR and the ISR, is required for ATF4 induction in OPA1-deficient BAT. Our data in mice demonstrated that, although ER stress is increased in response to OPA1 deletion, PERK is dispensable for eIF2α phosphorylation and ISR induction in OPA1-deficient BAT. Therefore, mice concomitantly lacking OPA1 and PERK in BAT display similar improvements in systemic metabolic health as observed in OPA1 BKO mice, including increased metabolic rates, browning of iWAT, and resistance to DIO and IR. These data suggest that an alternative ISR kinase is likely activated downstream of mitochondrial stress to induce the ISR in BAT when OPA1 is deleted. Indeed, a recent study revealed that mitochondrial dysfunction induces different paths to promote ISR activation, depending on both the nature of the mitochondrial stress and the metabolic state of the cell (Mick et al., 2020). Moreover, studies suggest that heme-regulated eIF2α kinase (HRI) is required to activate ATF4 in response to mitochondrial stress in HEK293 cells treated with oligomycin (Guo et al., 2020), and in embryonic and adult hearts of a mouse model of mitochondrial cardiomyopathy (Zhu et al., 2022). Whether the same mechanisms mediate ATF4 induction upon OPA1 deletion in BAT remain to be determined.

Transcriptome analysis also revealed that GDF15 was induced in OPA1-deficient BAT, which correlated with increased GDF15 serum levels. GDF15 is a member of the transforming growth factor-β superfamily. It acts through a recently identified receptor, an orphan member of the GFRα family called GFRAL, and signals through the Ret coreceptor (Breit et al., 2021). Several studies have demonstrated that GDF15 mediates its effects on reducing food intake, body weight, and adiposity largely by its actions on regions of the hindbrain, where GFRAL is expressed (Tsai et al., 2018a; Xiong et al., 2017; Tsai et al., 2019; Tran et al., 2018). A recent study also reported that the GDF15-GFRAL pathway is required to maintain energy expenditure during caloric restriction by increasing β-adrenergic signaling to skeletal muscle, thereby inducing fatty acid oxidation and futile calcium cycling (Wang et al., 2023). Nonetheless, several lines of evidence indicate that GDF15 may also have peripheral effects. Indeed, systemic overexpression of GDF15 prevents obesity and insulin resistance by modulating metabolic activity and enhancing the expression of thermogenic and lipolytic genes in BAT and WAT (Chrysovergis et al., 2014; Macia et al., 2012; Kim et al., 2013b). Some of these effects were independent of changes in food intake, suggesting that GDF15 might also exert its effects through yet unidentified receptors other than GFRAL (Aguilar-Recarte et al., 2022). GDF15 can be secreted by various organs such as liver (Patel et al., 2022), kidney (Baek and Eling, 2019), and skeletal muscle (Ost et al., 2020). Recently, studies demonstrated that GDF15 can also be induced in brown adipocytes in response to thermogenic stimuli (Flicker et al., 2019; Campderrós et al., 2019) and prolonged high-fat feeding (Patel et al., 2019). Like FGF21, GDF15 is induced in response to mitochondrial stress downstream of the ISR (Choi et al., 2020). Indeed, our data show that Gdf15 mRNA levels were dramatically reduced in mice lacking both OPA1 and ATF4 in BAT, suggesting that ATF4 may indirectly regulate GDF15 induction in OPA1 BKO mice, perhaps via its downstream target, CHOP, which can bind to the GDF15 promoter (Chung et al., 2017a).

Deletion of GDF15 in the OPA1 BKO background completely normalized GDF15 serum levels, confirming GDF15 secretion from BAT. Under baseline conditions, the metabolic phenotype of OPA1 BKO mice was largely unaffected by GDF15 deletion in BAT. Importantly, FGF21 levels remained elevated in DKO mice and browning of iWAT and increased resting metabolic rates under baseline conditions were preserved. These data reinforce the idea that FGF21 plays a predominant role on the baseline phenotype of OPA1 BKO mice, particularly browning of iWAT; however, FGF21 was dispensable to promote resistance to DIO in these mice (Pereira et al., 2021). Conversely, under obesogenic conditions, our data revealed that GDF15 partially mediates the resistance to DIO. These changes occurred independently of changes in food intake as DKO mice consumed similar amounts of food relative to their WT controls. Rather, our data support the idea that BAT-derived GDF15 increases metabolic rates in OPA1 BKO mice fed obesogenic diets, thereby attenuating weight gain. Furthermore, although under isocaloric conditions browning was preserved in DKO mice, GDF15 was required to maintain browning of iWAT in OPA1 BKO mice fed HFD. These data suggest that GDF15-mediated browning of iWAT during obesogenic conditions might contribute to the increased metabolic rates and reduced weight gain observed in OPA1 BKO mice, as also proposed elsewhere (Teijeiro et al., 2021; Kirschner et al., 2022; Peng et al., 2021). Furthermore, while the calcium pump Serca1 was induced in the gastrocnemius muscle of OPA1 BKO mice fed HFD, this effect was abrogated in DKO mice. This data suggests that GDF15 may also regulate futile calcium cycling in muscle to increase energy expenditure in OPA1 BKO fed an obesogenic diet. Of note, although GDF15 effects on body weight were only partial, the improvements in glucose disposal and hepatic steatosis observed in OPA1 BKO mice were completely abolished in the absence of BAT GDF15, highlighting additional mechanisms of GDF15-mediated metabolic protection that are independent of its body weight-lowering effects. Paradoxically, insulin sensitivity remained improved.

In the context of hepatic mitochondrial dysfunction, which is associated with increased hepatic FGF21 and GDF15 expression and increased liver-derived GDF15 in the circulation, GDF15 was shown to regulate changes in body and fat mass and protect against hepatic steatosis in DIO but had no effect on glucose disposal. Instead, FGF21 was required to increase insulin sensitivity, energy expenditure, and UCP1-mediated thermogenesis in iWAT of regular chow-fed mice (Kang et al., 2021), which is similar to what we previously reported for OPA1 BKO mice (Pereira et al., 2021). FGF21 and GDF15 induction and secretion was also reported in a model of mitochondrial stress in adipose tissue (both BAT and WAT). In this study, under regular chow-fed conditions, neither GDF15 nor FGF21 appeared to regulate whole-body metabolism. Conversely, as in OPA1 BKO mice, long-term induction of GDF15 attenuated progression of obesity by increasing energy expenditure in DIO, while FGF21 was dispensable for this phenotype (Choi et al., 2020). Together, our study suggests that FGF21 and GDF15 play fundamentally distinct roles as part of the mitochondrial stress response in regulating energy metabolism and glucose homeostasis in different metabolic states. Our data support a predominant role for GDF15 rather than FGF21 in promoting resistance to DIO in response to mitochondrial stress and highlights a role for GDF15 in promoting iWAT browning during DIO and potentially inducing futile calcium cycling in skeletal muscle. Some discrepancies between our model and similar models in the literature may stem from differences in the nature of the mitochondrial stress, level of GDF15 induction, and the fact that in our studies FGF21 (Pereira et al., 2021) and GDF15 were deleted in a tissue-specific manner, rather than globally. Also, additional methodological details such as age and duration of feeding might have contributed to differences in our overall conclusions.

Because OPA1 BKO mice have improved thermoregulation, we tested the role of BAT-derived GDF15 for adaptive thermogenesis. The increase in averaged baseline core body temperature observed in OPA1 BKO (Pereira et al., 2021) mice was no longer seen when GDF15 was deleted in BAT. Furthermore, in response to cold, mice lacking both OPA1 and GDF15 (DKO) became severely hypothermic as early as 4 hr into the cold exposure (4°C) protocol. Surprisingly, these mice had increased markers of browning in white adipose tissue (WAT) after 5 hr of cold exposure, despite similar degrees of sympathetic activation. Nonetheless, this increase in browning was not sufficient to maintain thermoregulation in the context of impaired BAT thermogenesis. Our data suggests increased futile calcium cycling in skeletal muscle of OPA1 BKO mice, but the induction in Serca1a levels was lost in DKO mice. Thus, in addition to severely impaired UCP1-mediated thermogenesis in BAT, inhibited compensation in muscle non-shivering thermogenesis pathways in DKO mice may contribute to overall reduced thermogenic capacity in these mice, leading to severe cold-induced hypothermia. Therefore, at least in the context of mitochondrial stress, GDF15 expression in thermogenic adipocytes plays a critical role in maintaining core body temperature in cold-exposed mice, likely by regulating UCP1-mediated and UCP1-independent thermogenesis.

In conclusion, our study has ruled out a role for PERK in inducing the ISR in response to OPA1 deletion in BAT. Furthermore, we also unveiled a role for GDF15 in attenuating DIO and improving glucose clearance and hepatic steatosis in OPA1 BKO mice. To our knowledge, this is the first demonstration that BAT-derived GDF15 may play a role in energy metabolism, glucose homeostasis, and thermoregulation in the context of mitochondrial stress. Our data also show that GDF15 may exert its effects independently of changes in food intake but predominantly by increasing energy expenditure. The specific molecular mechanisms downstream of GDF15 regulating these metabolic adaptations and the potential peripheral signaling mechanisms that mediate these effects remain to be elucidated. Furthermore, future studies exploring the role of BAT-derived GDF15 on energy metabolism and thermogenesis under physiological conditions might reveal novel GDF15-mediated effects in metabolic homeostasis.

RNASeq data (GSE218907).

1. 1. Pereira RO
   2. Weatherford ET

   (2022) NCBI Gene Expression Omnibus

   ID GSE218907. Transcriptome analysis in brown adipose tissue of mice lacking the mitochondrial protein optic atrophy 1 (OPA1) in thermogenic adipocytes.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218907

1. 1. Aguilar-Recarte D
   2. Barroso E
   3. Palomer X
   4. Wahli W
   5. Vázquez-Carrera M

   (2022) Knocking on GDF15’s door for the treatment of type 2 diabetes mellitus

   Trends in Endocrinology and Metabolism 33:741–754.

   https://doi.org/10.1016/j.tem.2022.08.004

   • PubMed
   • Google Scholar
2. 1. Baek SJ
   2. Eling T

   (2019) Growth differentiation factor 15 (GDF15): A survival protein with therapeutic potential in metabolic diseases

   Pharmacology & Therapeutics 198:46–58.

   https://doi.org/10.1016/j.pharmthera.2019.02.008

   • PubMed
   • Google Scholar
3. 1. Bao XR
   2. Ong S-E
   3. Goldberger O
   4. Peng J
   5. Sharma R
   6. Thompson DA
   7. Vafai SB
   8. Cox AG
   9. Marutani E
   10. Ichinose F
   11. Goessling W
   12. Regev A
   13. Carr SA
   14. Clish CB
   15. Mootha VK

   (2016) Mitochondrial dysfunction remodels one-carbon metabolism in human cells

   eLife 5:e10575.

   https://doi.org/10.7554/eLife.10575

   • PubMed
   • Google Scholar
4. 1. Bray NL
   2. Pimentel H
   3. Melsted P
   4. Pachter L

   (2016) Near-optimal probabilistic RNA-seq quantification

   Nature Biotechnology 34:525–527.

   https://doi.org/10.1038/nbt.3519

   • Google Scholar
5. 1. Breit SN
   2. Brown DA
   3. Tsai VW-W

   (2021) The GDF15-GFRAL Pathway in Health and Metabolic Disease: Friend or Foe?

   Annual Review of Physiology 83:127–151.

   https://doi.org/10.1146/annurev-physiol-022020-045449

   • PubMed
   • Google Scholar
6. 1. Campderrós L
   2. Moure R
   3. Cairó M
   4. Gavaldà-Navarro A
   5. Quesada-López T
   6. Cereijo R
   7. Giralt M
   8. Villarroya J
   9. Villarroya F

   (2019) Brown Adipocytes Secrete GDF15 in Response to Thermogenic Activation

   Obesity 27:1606–1616.

   https://doi.org/10.1002/oby.22584

   • PubMed
   • Google Scholar
7. 1. Choi MJ
   2. Jung S-B
   3. Lee SE
   4. Kang SG
   5. Lee JH
   6. Ryu MJ
   7. Chung HK
   8. Chang JY
   9. Kim YK
   10. Hong HJ
   11. Kim H
   12. Kim HJ
   13. Lee C-H
   14. Mardinoglu A
   15. Yi H-S
   16. Shong M

   (2020) An adipocyte-specific defect in oxidative phosphorylation increases systemic energy expenditure and protects against diet-induced obesity in mouse models

   Diabetologia 63:837–852.

   https://doi.org/10.1007/s00125-019-05082-7

   • PubMed
   • Google Scholar
8. 1. Chrysovergis K
   2. Wang X
   3. Kosak J
   4. Lee S-H
   5. Kim JS
   6. Foley JF
   7. Travlos G
   8. Singh S
   9. Baek SJ
   10. Eling TE

   (2014) NAG-1/GDF-15 prevents obesity by increasing thermogenesis, lipolysis and oxidative metabolism

   International Journal of Obesity 38:1555–1564.

   https://doi.org/10.1038/ijo.2014.27

   • PubMed
   • Google Scholar
9. 1. Chung HK
   2. Kim JT
   3. Kim H-W
   4. Kwon M
   5. Kim SY
   6. Shong M
   7. Kim KS
   8. Yi H-S

   (2017a) GDF15 deficiency exacerbates chronic alcohol- and carbon tetrachloride-induced liver injury

   Scientific Reports 7:17238.

   https://doi.org/10.1038/s41598-017-17574-w

   • PubMed
   • Google Scholar
10. 1. Chung HK
    2. Ryu D
    3. Kim KS
    4. Chang JY
    5. Kim YK
    6. Yi H-S
    7. Kang SG
    8. Choi MJ
    9. Lee SE
    10. Jung S-B
    11. Ryu MJ
    12. Kim SJ
    13. Kweon GR
    14. Kim H
    15. Hwang JH
    16. Lee C-H
    17. Lee S-J
    18. Wall CE
    19. Downes M
    20. Evans RM
    21. Auwerx J
    22. Shong M

    (2017b) Growth differentiation factor 15 is a myomitokine governing systemic energy homeostasis

    The Journal of Cell Biology 216:149–165.

    https://doi.org/10.1083/jcb.201607110

    • PubMed
    • Google Scholar
11. 1. Ebert SM
    2. Dyle MC
    3. Kunkel SD
    4. Bullard SA
    5. Bongers KS
    6. Fox DK
    7. Dierdorff JM
    8. Foster ED
    9. Adams CM

    (2012) Stress-induced skeletal muscle Gadd45a expression reprograms myonuclei and causes muscle atrophy

    The Journal of Biological Chemistry 287:27290–27301.

    https://doi.org/10.1074/jbc.M112.374777

    • PubMed
    • Google Scholar
12. 1. Fisher FM
    2. Kleiner S
    3. Douris N
    4. Fox EC
    5. Mepani RJ
    6. Verdeguer F
    7. Wu J
    8. Kharitonenkov A
    9. Flier JS
    10. Maratos-Flier E
    11. Spiegelman BM

    (2012) FGF21 regulates PGC-1α and browning of white adipose tissues in adaptive thermogenesis

    Genes & Development 26:271–281.

    https://doi.org/10.1101/gad.177857.111

    • PubMed
    • Google Scholar
13. 1. Flicker D
    2. Sancak Y
    3. Mick E
    4. Goldberger O
    5. Mootha VK

    (2019) Exploring the in vivo role of the mitochondrial calcium uniporter in brown fat bioenergetics

    Cell Reports 27:1364–1375.

    https://doi.org/10.1016/j.celrep.2019.04.013

    • PubMed
    • Google Scholar
14. 1. Forsström S
    2. Jackson CB
    3. Carroll CJ
    4. Kuronen M
    5. Pirinen E
    6. Pradhan S
    7. Marmyleva A
    8. Auranen M
    9. Kleine I-M
    10. Khan NA
    11. Roivainen A
    12. Marjamäki P
    13. Liljenbäck H
    14. Wang L
    15. Battersby BJ
    16. Richter U
    17. Velagapudi V
    18. Nikkanen J
    19. Euro L
    20. Suomalainen A

    (2019) Fibroblast Growth Factor 21 Drives Dynamics of Local and Systemic Stress Responses in Mitochondrial Myopathy with mtDNA Deletions

    Cell Metabolism 30:1040–1054.

    https://doi.org/10.1016/j.cmet.2019.08.019

    • PubMed
    • Google Scholar
15. 1. Garcia-Cazarin ML
    2. Snider NN
    3. Andrade FH

    (2011) Mitochondrial isolation from skeletal muscle

    Journal of Visualized Experiments 2011:49.

    https://doi.org/10.3791/2452

    • PubMed
    • Google Scholar
16. 1. Guo X
    2. Aviles G
    3. Liu Y
    4. Tian R
    5. Unger BA
    6. Lin Y-HT
    7. Wiita AP
    8. Xu K
    9. Correia MA
    10. Kampmann M

    (2020) Mitochondrial stress is relayed to the cytosol by an OMA1-DELE1-HRI pathway

    Nature 579:427–432.

    https://doi.org/10.1038/s41586-020-2078-2

    • PubMed
    • Google Scholar
17. 1. Kang SG
    2. Choi MJ
    3. Jung S-B
    4. Chung HK
    5. Chang JY
    6. Kim JT
    7. Kang YE
    8. Lee JH
    9. Hong HJ
    10. Jun SM
    11. Ro H-J
    12. Suh JM
    13. Kim H
    14. Auwerx J
    15. Yi H-S
    16. Shong M

    (2021) Differential roles of GDF15 and FGF21 in systemic metabolic adaptation to the mitochondrial integrated stress response

    iScience 24:102181.

    https://doi.org/10.1016/j.isci.2021.102181

    • PubMed
    • Google Scholar
18. 1. Kim KH
    2. Jeong YT
    3. Oh H
    4. Kim SH
    5. Cho JM
    6. Kim Y-N
    7. Kim SS
    8. Kim DH
    9. Hur KY
    10. Kim HK
    11. Ko T
    12. Han J
    13. Kim HL
    14. Kim J
    15. Back SH
    16. Komatsu M
    17. Chen H
    18. Chan DC
    19. Konishi M
    20. Itoh N
    21. Choi CS
    22. Lee M-S

    (2013a) Autophagy deficiency leads to protection from obesity and insulin resistance by inducing Fgf21 as a mitokine

    Nature Medicine 19:83–92.

    https://doi.org/10.1038/nm.3014

    • PubMed
    • Google Scholar
19. 1. Kim JM
    2. Kosak JP
    3. Kim JK
    4. Kissling G
    5. Germolec DR
    6. Zeldin DC
    7. Bradbury JA
    8. Baek SJ
    9. Eling TE

    (2013b) NAG-1/GDF15 transgenic mouse has less white adipose tissue and a reduced inflammatory response

    Mediators of Inflammation 2013:641851.

    https://doi.org/10.1155/2013/641851

    • PubMed
    • Google Scholar
20. 1. Kim D
    2. Paggi JM
    3. Park C
    4. Bennett C
    5. Salzberg SL

    (2019) Graph-based genome alignment and genotyping with HISAT2 and HISAT-genotype

    Nature Biotechnology 37:907–915.

    https://doi.org/10.1038/s41587-019-0201-4

    • Google Scholar
21. 1. Kirschner KM
    2. Foryst-Ludwig A
    3. Gohlke S
    4. Li C
    5. Flores RE
    6. Kintscher U
    7. Schupp M
    8. Schulz TJ
    9. Scholz H

    (2022) Wt1 haploinsufficiency induces browning of epididymal fat and alleviates metabolic dysfunction in mice on high-fat diet

    Diabetologia 65:528–540.

    https://doi.org/10.1007/s00125-021-05621-1

    • PubMed
    • Google Scholar
22. 1. Kong X
    2. Banks A
    3. Liu T
    4. Kazak L
    5. Rao RR
    6. Cohen P
    7. Wang X
    8. Yu S
    9. Lo JC
    10. Tseng Y-H
    11. Cypess AM
    12. Xue R
    13. Kleiner S
    14. Kang S
    15. Spiegelman BM
    16. Rosen ED

    (2014) IRF4 is a key thermogenic transcriptional partner of PGC-1α

    Cell 158:69–83.

    https://doi.org/10.1016/j.cell.2014.04.049

    • PubMed
    • Google Scholar
23. 1. Liao Y
    2. Smyth GK
    3. Shi W

    (2014) featureCounts: an efficient general purpose program for assigning sequence reads to genomic features

    Bioinformatics 30:923–930.

    https://doi.org/10.1093/bioinformatics/btt656

    • PubMed
    • Google Scholar
24. 1. Love MI
    2. Huber W
    3. Anders S

    (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

    Genome Biology 15:550.

    https://doi.org/10.1186/s13059-014-0550-8

    • Google Scholar
25. 1. Macia L
    2. Tsai VW-W
    3. Nguyen AD
    4. Johnen H
    5. Kuffner T
    6. Shi Y-C
    7. Lin S
    8. Herzog H
    9. Brown DA
    10. Breit SN
    11. Sainsbury A

    (2012) Macrophage inhibitory cytokine 1 (MIC-1/GDF15) decreases food intake, body weight and improves glucose tolerance in mice on normal & obesogenic diets

    PLOS ONE 7:e34868.

    https://doi.org/10.1371/journal.pone.0034868

    • PubMed
    • Google Scholar
26. 1. Mick E
    2. Titov DV
    3. Skinner OS
    4. Sharma R
    5. Jourdain AA
    6. Mootha VK

    (2020) Distinct mitochondrial defects trigger the integrated stress response depending on the metabolic state of the cell

    eLife 9:e49178.

    https://doi.org/10.7554/eLife.49178

    • PubMed
    • Google Scholar
27. 1. Mina AI
    2. LeClair RA
    3. LeClair KB
    4. Cohen DE
    5. Lantier L
    6. Banks AS

    (2018) CalR: a web-based analysis tool for indirect calorimetry experiments

    Cell Metabolism 28:656–666.

    https://doi.org/10.1016/j.cmet.2018.06.019

    • PubMed
    • Google Scholar
28. 1. Ost M
    2. Coleman V
    3. Voigt A
    4. van Schothorst EM
    5. Keipert S
    6. van der Stelt I
    7. Ringel S
    8. Graja A
    9. Ambrosi T
    10. Kipp AP
    11. Jastroch M
    12. Schulz TJ
    13. Keijer J
    14. Klaus S

    (2016) Muscle mitochondrial stress adaptation operates independently of endogenous FGF21 action

    Molecular Metabolism 5:79–90.

    https://doi.org/10.1016/j.molmet.2015.11.002

    • PubMed
    • Google Scholar
29. 1. Ost M
    2. Igual Gil C
    3. Coleman V
    4. Keipert S
    5. Efstathiou S
    6. Vidic V
    7. Weyers M
    8. Klaus S

    (2020) Muscle-derived GDF15 drives diurnal anorexia and systemic metabolic remodeling during mitochondrial stress

    EMBO Reports 21:e48804.

    https://doi.org/10.15252/embr.201948804

    • PubMed
    • Google Scholar
30. 1. Pakos-Zebrucka K
    2. Koryga I
    3. Mnich K
    4. Ljujic M
    5. Samali A
    6. Gorman AM

    (2016) The integrated stress response

    EMBO Reports 17:1374–1395.

    https://doi.org/10.15252/embr.201642195

    • PubMed
    • Google Scholar
31. 1. Patel S
    2. Alvarez-Guaita A
    3. Melvin A
    4. Rimmington D
    5. Dattilo A
    6. Miedzybrodzka EL
    7. Cimino I
    8. Maurin A-C
    9. Roberts GP
    10. Meek CL
    11. Virtue S
    12. Sparks LM
    13. Parsons SA
    14. Redman LM
    15. Bray GA
    16. Liou AP
    17. Woods RM
    18. Parry SA
    19. Jeppesen PB
    20. Kolnes AJ
    21. Harding HP
    22. Ron D
    23. Vidal-Puig A
    24. Reimann F
    25. Gribble FM
    26. Hulston CJ
    27. Farooqi IS
    28. Fafournoux P
    29. Smith SR
    30. Jensen J
    31. Breen D
    32. Wu Z
    33. Zhang BB
    34. Coll AP
    35. Savage DB
    36. O’Rahilly S

    (2019) GDF15 provides an endocrine signal of nutritional stress in mice and humans

    Cell Metabolism 29:707–718.

    https://doi.org/10.1016/j.cmet.2018.12.016

    • PubMed
    • Google Scholar
32. 1. Patel S
    2. Haider A
    3. Alvarez-Guaita A
    4. Bidault G
    5. El-Sayed Moustafa JS
    6. Guiu-Jurado E
    7. Tadross JA
    8. Warner J
    9. Harrison J
    10. Virtue S
    11. Scurria F
    12. Zvetkova I
    13. Blüher M
    14. Small KS
    15. O’Rahilly S
    16. Savage DB

    (2022) Combined genetic deletion of GDF15 and FGF21 has modest effects on body weight, hepatic steatosis and insulin resistance in high fat fed mice

    Molecular Metabolism 65:101589.

    https://doi.org/10.1016/j.molmet.2022.101589

    • PubMed
    • Google Scholar
33. 1. Peng W-Q
    2. Xiao G
    3. Li B-Y
    4. Guo Y-Y
    5. Guo L
    6. Tang Q-Q

    (2021) l-theanine activates the browning of white adipose tissue through the ampk/α-ketoglutarate/prdm16 axis and ameliorates diet-induced obesity in mice

    Diabetes 70:1458–1472.

    https://doi.org/10.2337/db20-1210

    • PubMed
    • Google Scholar
34. 1. Pereira RO
    2. Tadinada SM
    3. Zasadny FM
    4. Oliveira KJ
    5. Pires KMP
    6. Olvera A
    7. Jeffers J
    8. Souvenir R
    9. Mcglauflin R
    10. Seei A
    11. Funari T
    12. Sesaki H
    13. Potthoff MJ
    14. Adams CM
    15. Anderson EJ
    16. Abel ED

    (2017) OPA1 deficiency promotes secretion of FGF21 from muscle that prevents obesity and insulin resistance

    The EMBO Journal 36:2126–2145.

    https://doi.org/10.15252/embj.201696179

    • PubMed
    • Google Scholar
35. 1. Pereira RO
    2. McFarlane SI

    (2019) The role of brown adipose tissue in cardiovascular disease protection: current evidence and future directions

    International Journal of Clinical Research & Trials 4:136.

    https://doi.org/10.15344/2456-8007/2019/136

    • PubMed
    • Google Scholar
36. 1. Pereira RO
    2. Marti A
    3. Olvera AC
    4. Tadinada SM
    5. Bjorkman SH
    6. Weatherford ET
    7. Morgan DA
    8. Westphal M
    9. Patel PH
    10. Kirby AK
    11. Hewezi R
    12. Bùi Trân W
    13. García-Peña LM
    14. Souvenir RA
    15. Mittal M
    16. Adams CM
    17. Rahmouni K
    18. Potthoff MJ
    19. Abel ED

    (2021) OPA1 deletion in brown adipose tissue improves thermoregulation and systemic metabolism via FGF21

    eLife 10:e66519.

    https://doi.org/10.7554/eLife.66519

    • PubMed
    • Google Scholar
37. 1. Potthoff MJ
    2. Inagaki T
    3. Satapati S
    4. Ding X
    5. He T
    6. Goetz R
    7. Mohammadi M
    8. Finck BN
    9. Mangelsdorf DJ
    10. Kliewer SA
    11. Burgess SC

    (2009) FGF21 induces PGC-1alpha and regulates carbohydrate and fatty acid metabolism during the adaptive starvation response

    PNAS 106:10853–10858.

    https://doi.org/10.1073/pnas.0904187106

    • PubMed
    • Google Scholar
38. 1. Quirós PM
    2. Prado MA
    3. Zamboni N
    4. D’Amico D
    5. Williams RW
    6. Finley D
    7. Gygi SP
    8. Auwerx J

    (2017) Multi-omics analysis identifies ATF4 as a key regulator of the mitochondrial stress response in mammals

    The Journal of Cell Biology 216:2027–2045.

    https://doi.org/10.1083/jcb.201702058

    • PubMed
    • Google Scholar
39. 1. Teijeiro A
    2. Garrido A
    3. Ferre A
    4. Perna C
    5. Djouder N

    (2021) Inhibition of the IL-17A axis in adipocytes suppresses diet-induced obesity and metabolic disorders in mice

    Nature Metabolism 3:496–512.

    https://doi.org/10.1038/s42255-021-00371-1

    • PubMed
    • Google Scholar
40. 1. Tran T
    2. Yang J
    3. Gardner J
    4. Xiong Y

    (2018) GDF15 deficiency promotes high fat diet-induced obesity in mice

    PLOS ONE 13:e0201584.

    https://doi.org/10.1371/journal.pone.0201584

    • PubMed
    • Google Scholar
41. 1. Tsai VWW
    2. Husaini Y
    3. Sainsbury A
    4. Brown DA
    5. Breit SN

    (2018a) The MIC-1/GDF15-GFRAL pathway in energy homeostasis: implications for obesity, cachexia, and other associated diseases

    Cell Metabolism 28:353–368.

    https://doi.org/10.1016/j.cmet.2018.07.018

    • PubMed
    • Google Scholar
42. 1. Tsai VW
    2. Zhang HP
    3. Manandhar R
    4. Lee-Ng KKM
    5. Lebhar H
    6. Marquis CP
    7. Husaini Y
    8. Sainsbury A
    9. Brown DA
    10. Breit SN

    (2018b) Treatment with the TGF-b superfamily cytokine MIC-1/GDF15 reduces the adiposity and corrects the metabolic dysfunction of mice with diet-induced obesity

    International Journal of Obesity 42:561–571.

    https://doi.org/10.1038/ijo.2017.258

    • PubMed
    • Google Scholar
43. 1. Tsai VW-W
    2. Zhang HP
    3. Manandhar R
    4. Schofield P
    5. Christ D
    6. Lee-Ng KKM
    7. Lebhar H
    8. Marquis CP
    9. Husaini Y
    10. Brown DA
    11. Breit SN

    (2019) GDF15 mediates adiposity resistance through actions on GFRAL neurons in the hindbrain AP/NTS

    International Journal of Obesity 43:2370–2380.

    https://doi.org/10.1038/s41366-019-0365-5

    • PubMed
    • Google Scholar
44. 1. Wang D
    2. Townsend LK
    3. DesOrmeaux GJ
    4. Frangos SM
    5. Batchuluun B
    6. Dumont L
    7. Kuhre RE
    8. Ahmadi E
    9. Hu S
    10. Rebalka IA
    11. Gautam J
    12. Jabile MJT
    13. Pileggi CA
    14. Rehal S
    15. Desjardins EM
    16. Tsakiridis EE
    17. Lally JSV
    18. Juracic ES
    19. Tupling AR
    20. Gerstein HC
    21. Paré G
    22. Tsakiridis T
    23. Harper M-E
    24. Hawke TJ
    25. Speakman JR
    26. Blondin DP
    27. Holloway GP
    28. Jørgensen SB
    29. Steinberg GR

    (2023) GDF15 promotes weight loss by enhancing energy expenditure in muscle

    Nature 619:143–150.

    https://doi.org/10.1038/s41586-023-06249-4

    • PubMed
    • Google Scholar
45. 1. Xiong Y
    2. Walker K
    3. Min X
    4. Hale C
    5. Tran T
    6. Komorowski R
    7. Yang J
    8. Davda J
    9. Nuanmanee N
    10. Kemp D
    11. Wang X
    12. Liu H
    13. Miller S
    14. Lee KJ
    15. Wang Z
    16. Véniant MM

    (2017) Long-acting MIC-1/GDF15 molecules to treat obesity: Evidence from mice to monkeys

    Science Translational Medicine 9:412.

    https://doi.org/10.1126/scitranslmed.aan8732

    • PubMed
    • Google Scholar
46. 1. Zhang P
    2. McGrath B
    3. Li S
    4. Frank A
    5. Zambito F
    6. Reinert J
    7. Gannon M
    8. Ma K
    9. McNaughton K
    10. Cavener DR

    (2002) The PERK eukaryotic initiation factor 2 alpha kinase is required for the development of the skeletal system, postnatal growth, and the function and viability of the pancreas

    Molecular and Cellular Biology 22:3864–3874.

    https://doi.org/10.1128/MCB.22.11.3864-3874.2002

    • PubMed
    • Google Scholar
47. 1. Zhang Z
    2. Wakabayashi N
    3. Wakabayashi J
    4. Tamura Y
    5. Song W-J
    6. Sereda S
    7. Clerc P
    8. Polster BM
    9. Aja SM
    10. Pletnikov MV
    11. Kensler TW
    12. Shirihai OS
    13. Iijima M
    14. Hussain MA
    15. Sesaki H

    (2011) The dynamin-related GTPase Opa1 is required for glucose-stimulated ATP production in pancreatic beta cells

    Molecular Biology of the Cell 22:2235–2245.

    https://doi.org/10.1091/mbc.E10-12-0933

    • PubMed
    • Google Scholar
48. 1. Zhu S
    2. Nguyen A
    3. Pang J
    4. Zhao J
    5. Chen Z
    6. Liang Z
    7. Gu Y
    8. Huynh H
    9. Bao Y
    10. Lee S
    11. Kluger Y
    12. Ouyang K
    13. Evans SM
    14. Fang X

    (2022) Mitochondrial Stress Induces an HRI-eIF2α Pathway Protective for Cardiomyopathy

    Circulation 146:1028–1031.

    https://doi.org/10.1161/CIRCULATIONAHA.122.059594

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your work entitled "GDF15 Is Required for Cold-Induced Thermogenesis and Contributes to Improved Systemic Metabolic Health Following Loss of OPA1 in Brown Adipocytes" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by David James (Senior Editor). The reviewers opted to remain anonymous.

Essential revisions:

1) The reviewers noted a lack of inclusion of single knockout controls. If this data is available, it should be included.

2) More details regarding metabolic phenotyping should be provided.

Reviewer #1 (Recommendations for the authors):

Jena et al. follows up on a previous study on OPA1 knockout in BAT (BKO) tissue and provides mechanistic details on how certain phenotypes in OPA1 BKO are mediated. This paper tests and rules out PERK as a potential mediator of ATF4 activation in OPA1 BKO by creating a OPA1 and PERK double KO (DKO) in BAT. It also assigns the phenotypes resistance to diet induced obesity, improvements in glucose tolerance and cold induced thermogenesis to be GDF15 dependent.

The manuscript uses OPA1/PERK BAT DKO and OPA1/GDF15 BAT DKO, and compares them to wildtype (WT) without single KO comparisons. This needs to be clarified.

1. Is PERK activated in OPA1 BKO? There is no western blot showing increased PERK phosphorylation or any other means of activation.

2. Is increase in serum GDF15 levels due to OPA1 BKO in Figure 1E is it in the physiological range for the receptor GFRAL to be activated? Is giving additional GDF15 to WT or to the OPA1/GDF15 DKO to bring the concentration up to OPA1 BKO levels going to replicate these aspects of the OPA1 KO phenotype?

3. One of their conclusions states that "Our data shows that GDF15 may exert its effects independently of changes in food intake but predominantly by increasing energy expenditure." The energy expenditure data shows no difference between WT in HFD and loss of GDF15 increases energy expenditure in chow feeding. This needs clarification.

Reviewer #2 (Recommendations for the authors):

In this manuscript, the authors followup on prior findings in which they showed that mice with a brown fat specific knockout of OPA1 (OPA1 BKO) demonstrate induction of ATF4 and FGF21 secretion and resistance to dietary obesity. Since FGF21 is dispensable for these metabolic benefits, here the authors investigate what mediates the activation of the integrated stress response in OPA1 BKO mice and downstream mediators of the protection from dietary obesity. By making new brown fat-selective double mutant mice, they suggest that PERK is not required for ATF4 induction in this context and that deletion of GDF15 in brown fat partially reverses the protection from dietary obesity seen with OPA deletion.

Strengths of this manuscript include the generation of new brown fat selective double knockout models (OPA1/PERK DKO and OPA1/GDF15 DKO). Moreover, the authors performed phenotyping on both standard chow and high fat diets. Their data indicate that PERK in brown adipocytes is not required for the induction of ATF4 seen in OPA1 deletion. They further show that OPA1 BKO mice with concomitant deletion of GDF15 show attenuated protection from dietary obesity. These mice also have impaired cold tolerance. Weaknesses of the manuscript include the absence of single knockout controls in the experiments presented. In addition, the metabolic phenotyping is not comprehensive and underlying mechanisms for the phenotypes observed are not deeply explored.

1. It is very difficult to interpret the findings of the experiments on the DKO models without single knockout controls. The reader is forced to make a mental comparison to OPA1 BKO mice in a prior paper from this group, but those mice were presumably not studied contemporaneously. Because of this, it is difficult to have confidence that the data support the conclusions made. For example, in Figure 2, the authors make OPA1/PERK DKO mice, and they argue that the increase in GDF15 seen with OPA1 deletion is unchanged (Figure 2D). However, without single KO controls, how can one be sure of this? In addition, the authors argue that OPA1/GDF15 DKO mice show attenuated protection from dietary obesity (Figure 4). Again, single OPA1 KOs controls would be needed to make this conclusion more solid.

2. The metabolic phenotyping is not complete in this manuscript. Specifically, the authors should: (i) show longitudinal weight curves instead of static bar graphs; (ii) provide a more thorough analysis of browning in Figure 2, including UCP1 Westerns, qPCR of other thermogenic genes, and histology/IHC; (iii) the analysis of energy expenditure should show longitudinal data over time in the light and dark cycles, rather than bar graphs.

3. The manuscript does not contain any mechanistic exploration of the findings here. The authors begin to do this in Figure 5, but what do they think explains the fact that OPA1/GDF15 DKO mice show partial protection from DIO, yet are also more cold-sensitive? Additionally, the authors argue that the increased browning in iWAT in DKO mice is due to increased sympathetic activation, but they do not present any data in support of this claim.

4. In Figure 1, panels C and D should be in the same graph.

5. In Figure 1, GDF15 levels are in the ng/ml range, but in Figure 2, they are in the pg/ml range. This should be clarified.

6. The PERK KO in BAT was not nearly as complete as the OPA1 KO. Why do the authors think this is? And could this explain the lack of an effect on ATF4?

7. On page 10 of the text, lines 4-5 (in reference to Figure 4), shouldn't there be a NOT in this sentence?

Reviewer #3 (Recommendations for the authors):

The report by Jena et al. describes the observation that genetic inhibition of both OPA1 and GDF15 using UCP1-cre leads to reductions in circulating GDF15 and reductions in body mass. These effects are associated with increases in EE during the dark but not light cycle, lower fat mass and reduced expression of thermogenic genes in WAT. The findings are interesting but there are many incomplete findings presented which require further clarification.

1. The data around the assessment of energy expenditure is incomplete and inconsistent. For example in Figure 2 the authors present the EE relative to body but no other data from this experiment is provided. In Figure 4F they show EE is elevated in the dark cycle but not the light cycle. They then plot VO2 relative to body mass in an n=4 mice and show no difference between groups. It is unclear where this data came from and if it was the same experiment and if so whether this was collected in the dark or light cycle? In both Figure 2 and Figure 4 VO2, VCO2, and RER should be presented as it is unclear whether a shift to Fatty acid oxidation might be the primary contributor to the increase in the calculated EE. In addition the sample size for all measures is very low making it possible the results may be due to a type 1 error.

2. The authors suggest that the increase in EE may be leading to the reduced body mass of the DKO mice. To validate this hypothesis authors should complete pair feeding experiments as it is currently unclear whether these relatively small changes in EE are critical for the phenotype which is central to the primary conclusions.

3. It is now well recognized that housing mice outside of the thermoneutral zone masks changes in adaptive thermogenesis. Authors should complete experiments at thermoneutrality.

4. Throughout the paper there is paucity of information besides the assessment of a few thermogenic genes about the morphology of the brown and white adipose tissue depots. Authors should provide adipose tissue histology images and ideally electron microscopy images detailing whether there may be defects in mitochondrial function.

5. Related to the above the authors suggest reduced WAT browning is the reason for the hypothermia with cold, however, it seems highly unlikely that given the relatively low thermogenic capacity of this tissue that this is the primary driver of the effect. For example, what about the very large reduction in BAT UCP1 in the DKO mice. Alternatively, did the authors examine whether there might be differences in shivering or skeletal muscle function?

Essential revisions:

1) The reviewers noted a lack of inclusion of single knockout controls. If this data is available, it should be included.

We appreciate this recommendation. Although we have recently published our data on the OPA1 BKO mice (DOI: 10.7554/eLife.66519), we now include previously unpublished data from an independent cohort of OPA1 BKO in this revised version of our manuscript. Due to time constraints and limited resources, we were unable to conduct de novo studies in additional OPA1 BKO mice, but we have added historical data on these animals that were not previously published. The new Figure 1, Figure 1 figure supplement 1, Figure 6 —figure supplement 1 and Figure 7 —figure supplement 1 show data on body mass, energy homeostasis, indirect calorimetry and browning of WAT in OPA1 BKO mice under normal chow and HFD conditions.

2) More details regarding metabolic phenotyping should be provided.

We have expanded the data on metabolic phenotyping for the different mouse models and added more details about data collection and analysis. Although our main conclusions were unaffected by the additional data, we hope these data will be helpful, and will bring clarity thereby satisfying the reviewers’ concerns.

Reviewer #1 (Recommendations for the authors):

Jena et al. follows up on a previous study on OPA1 knockout in BAT (BKO) tissue and provides mechanistic details on how certain phenotypes in OPA1 BKO are mediated. This paper tests and rules out PERK as a potential mediator of ATF4 activation in OPA1 BKO by creating a OPA1 and PERK double KO (DKO) in BAT. It also assigns the phenotypes resistance to diet induced obesity, improvements in glucose tolerance and cold induced thermogenesis to be GDF15 dependent.

The manuscript uses OPA1/PERK BAT DKO and OPA1/GDF15 BAT DKO, and compares them to wildtype (WT) without single KO comparisons. This needs to be clarified.

1. Is PERK activated in OPA1 BKO? There is no western blot showing increased PERK phosphorylation or any other means of activation.

This is an important point, and we appreciate the opportunity to clarify this issue. Perk was initially selected from our RNASeq data in BAT of OPA1 BAT KO. Of all integrated stress response (ISR) kinases, PERK was the only one significantly enriched in BAT of OPA1 BAT KO mice. This data is now included in the revised manuscript.

2. Is increase in serum GDF15 levels due to OPA1 BKO in Figure 1E is it in the physiological range for the receptor GFRAL to be activated?

Under physiological conditions, such as after prolonged high-fat feeding, GDF15 circulating levels in mice usually double, going from ~ 100pg/mL to ~ 200 pg/mL (DOI: 10.1016/j.cmet.2018.12.016). Thus, our data suggest that the increase observed in OPA1 BKO mice (Figure 2) is within the physiological range. Importantly, administration of recombinant GDF15 within the physiological range does not lead to changes in food intake, whereas at a higher concentration (~5000 pg/mL), food intake is reduced, suggesting GDF15-mediated activation of GFRAL (DOI: 10.1016/j.cmet.2018.12.016). Although not directly tested in this study, we believe that the increase in GDF15 levels observed in OPA1 BKO mice is likely insufficient to activate GFRAL signaling in the hindbrain to reduce food intake, which might explain why we do not see suppression of food intake associated with increased GDF15 levels in our model. Indeed, it has been proposed that the anorectic effect of GDF15 involving GFRAL requires a threshold of circulating GDF15 levels that is ≥400 pg/ml in mice, with lower values not affecting appetite (10.15252/embr.201948804). Furthermore, several lines of evidence indicate that GDF15 may also have peripheral effects. Indeed, GDF15 increases thermogenesis, lipid catabolism, and mitochondrial oxidative phosphorylation independently of changes in food intake (doi.org/10.1083/jcb.201607110), suggesting that GDF15 might also exert its effects through receptors other than GFRAL and independently of appetite reduction (https://doi.org/10.1016/j.tem.2022.08.004). We have expanded our discussion to address this issue.

Is giving additional GDF15 to WT or to the OPA1/GDF15 DKO to bring the concentration up to OPA1 BKO levels going to replicate these aspects of the OPA1 KO phenotype?

These are interesting experiments to pursue. Unfortunately, due to time constraints, we were unable to conduct these studies. It is likely that restoring GDF15 circulating levels to the levels reported in OPA1 KO mice might replicate some of the OPA1 KO phenotype, particularly those shown to be mediated by GDF15. Future studies using mice lacking GFRAL should clarify which aspects of the OPA1 BAT KO phenotype are regulated by GDF15’s central effects, and which ones require GDF15’s action locally in BAT or possibly in other tissues.

Of note, the degree of GDF15 induction in OPA1 KO mice under baseline conditions is similar to that caused by cold stress or high-fat feeding in WT mice, and is not further induced in response to these stressors.

3. One of their conclusions states that "Our data shows that GDF15 may exert its effects independently of changes in food intake but predominantly by increasing energy expenditure." The energy expenditure data shows no difference between WT in HFD and loss of GDF15 increases energy expenditure in chow feeding. This needs clarification.

We appreciate the reviewer bringing this to our attention. ANCOVA analysis revealed increased energy expenditure and oxygen consumption rates as a function to body mass in OPA1 BAT KO mice under isocaloric conditions and when fed obesogenic diet (DOI: 10.7554/eLife.66519 and Figure 1). These data led us to conclude that the increase in metabolic rates resulted in the lean phenotype observed in these mice. Although OPA1/GDF15 DKO mice had increased metabolic rates under isocaloric conditions (Figure 5 and Figure 5 —figure supplement 2), which we believe is mediated by FGF21 (DOI: 10.7554/eLife.66519), upon high-fat feeding, the group effect for energy expenditure as a function of body mass was no longer statistically significant (Figure 6 and Figure 6 —figure supplement 1). This suggested to us that GDF15 might regulate changes in energy expenditure in OPA1 BKO mice fed obesogenic diet. Therefore, OPA1/GDF15 DKO mice lack this increase in resting metabolic rates, resulting in increased weight gain relative to OPA1 BKO mice when fed HFD (Figure 6A). Interestingly, OPA1 BAT KO mice fed HFD have high UCP1 expression in iWAT, while OPA1/GDF15 BAT DKO mice lack this adaptation (Figure 6 —figure supplement 1). Furthermore, we observed increased Serca1a levels in the skeletal muscle of OPA1 BKO mice, but not in OPA1/GDF15 DKO mice, suggesting GDF15 might also regulate futile calcium cycling in muscle, as recently demonstrated elsewhere (DOI: 10.1038/s41586-023-06249-4). Although not directly tested in this study, it is possible that GDF15-dependent induction of thermogenic activity in iWAT and in muscle might contribute to the increased metabolic rates observed in OPA1 BAT KO mice during obesogenic conditions contributing to leanness.

Reviewer #2 (Recommendations for the authors):

In this manuscript, the authors followup on prior findings in which they showed that mice with a brown fat specific knockout of OPA1 (OPA1 BKO) demonstrate induction of ATF4 and FGF21 secretion and resistance to dietary obesity. Since FGF21 is dispensable for these metabolic benefits, here the authors investigate what mediates the activation of the integrated stress response in OPA1 BKO mice and downstream mediators of the protection from dietary obesity. By making new brown fat-selective double mutant mice, they suggest that PERK is not required for ATF4 induction in this context and that deletion of GDF15 in brown fat partially reverses the protection from dietary obesity seen with OPA deletion.

Strengths of this manuscript include the generation of new brown fat selective double knockout models (OPA1/PERK DKO and OPA1/GDF15 DKO). Moreover, the authors performed phenotyping on both standard chow and high fat diets. Their data indicate that PERK in brown adipocytes is not required for the induction of ATF4 seen in OPA1 deletion. They further show that OPA1 BKO mice with concomitant deletion of GDF15 show attenuated protection from dietary obesity. These mice also have impaired cold tolerance. Weaknesses of the manuscript include the absence of single knockout controls in the experiments presented. In addition, the metabolic phenotyping is not comprehensive and underlying mechanisms for the phenotypes observed are not deeply explored.

1. It is very difficult to interpret the findings of the experiments on the DKO models without single knockout controls. The reader is forced to make a mental comparison to OPA1 BKO mice in a prior paper from this group, but those mice were presumably not studied contemporaneously. Because of this, it is difficult to have confidence that the data support the conclusions made. For example, in Figure 2, the authors make OPA1/PERK DKO mice, and they argue that the increase in GDF15 seen with OPA1 deletion is unchanged (Figure 2D). However, without single KO controls, how can one be sure of this? In addition, the authors argue that OPA1/GDF15 DKO mice show attenuated protection from dietary obesity (Figure 4). Again, single OPA1 KOs controls would be needed to make this conclusion more solid.

We appreciate this recommendation. Although we understand that ideally the studies in the OPA1 BKO and OPA1/GDF15 DKO mice should have been performed in parallel, our data in OPA1 BKO mice was collected several years ago, and most of it was published in 2021 (DOI: 10.7554/eLife.66519). The cost and time associated with repeating all the experiments in parallel with the OPA1/GDF15 and OPA1/PERK DKO studies would have been prohibitive. Nonetheless, we have now added additional data on OPA1 BKO mice that were not previously published. Although we were unable to conduct new studies, we have added historical data (new Figure 1) on body mass and energy homeostasis on OPA1 BAT KO under normal chow and HFD conditions. We hope that the addition of these new data will aid the reviewers in evaluating the different phenotypes described for each mouse model. We believe these data strengthen our conclusions that GDF15 regulates metabolic rates and the resistance to DIO, as well as thermoregulation in cold-exposed OPA1 BKO mice.

2. The metabolic phenotyping is not complete in this manuscript. Specifically, the authors should: (i) show longitudinal weight curves instead of static bar graphs; (ii) provide a more thorough analysis of browning in Figure 2, including UCP1 Westerns, qPCR of other thermogenic genes, and histology/IHC; (iii) the analysis of energy expenditure should show longitudinal data over time in the light and dark cycles, rather than bar graphs.

We appreciate the reviewer’s suggestions. We have now expanded the data on metabolic phenotyping for the different mouse models and added more details about data collection and analysis. Specifically, we: 1. included longitudinal weight curves; 2. provided stronger evidence for browning in both OPA1/PERK and OPA1/GDF15 DKO models; and 3. added longitudinal data over time in the light and dark cycles for energy expenditure and other metabolic parameters.

3. The manuscript does not contain any mechanistic exploration of the findings here. The authors begin to do this in Figure 5, but what do they think explains the fact that OPA1/GDF15 DKO mice show partial protection from DIO, yet are also more cold-sensitive? Additionally, the authors argue that the increased browning in iWAT in DKO mice is due to increased sympathetic activation, but they do not present any data in support of this claim.

We appreciate the opportunity to clarify mechanisms that are driving our phenotype.

In this resubmission, we include new data that we believe offers mechanistic insight to our work. For example, in DIO, although BAT thermogenesis was impaired, OPA1 BAT KO had increased browning of iWAT. This compensatory effect was lost in the absence of GDF15. Furthermore, we observed increased Serca1a levels in the skeletal muscle of OPA1 BKO mice, but not in OPA1/GDF15 DKO mice, suggesting GDF15 might also regulate futile calcium cycling in muscle, as recently demonstrated elsewhere (DOI: 10.1038/s41586-023-06249-4). Although not directly tested in this study, it is possible that GDF15-dependent induction of thermogenic activity in iWAT and in muscle might contribute to the increased metabolic rates observed in OPA1 BAT KO mice during obesogenic conditions contributing to leanness.

Therefore, in the absence of GDF15, when browning is blunted, mice gain significantly more weight (Figure 6 and Figure 6 —figure supplement 1). On the other hand, after acute cold exposure, we see preserved browning of iWAT in DKO mice, while core body temperatures were highly reduced (Figure 7). This suggests that the degree of browning was insufficient to generate enough heat for cold-induced thermogenesis, in light of greatly reduced UCP1-dependent thermogenesis in BAT and inhibited compensatory thermogenic activation in muscle (via futile calcium cycling).

4. In Figure 1, panels C and D should be in the same graph.

We have repeated this analysis in OPA1 BKO and OPA1/ATF4 DKO mice side by side. The data confirms our initial conclusion that GDF15 expression in OPA1 BKO mice is at least partially regulated by ATF4 (Figure 2).

5. In Figure 1, GDF15 levels are in the ng/ml range, but in Figure 2, they are in the pg/ml range. This should be clarified.

This was a mistake on our part and has now been corrected.

6. The PERK KO in BAT was not nearly as complete as the OPA1 KO. Why do the authors think this is? And could this explain the lack of an effect on ATF4?

We agree with the reviewers’ observation that PERK levels are not as reduced in DKO mice as OPA1. We believe this reflects PERK expression in non-adipocytes cells in BAT. However, given the greatly reduced PERK expression, we find it highly unlikely that activation of the residual PERK, could account for the increased eIF2a activation and ATF4 induction. Our interpretation is that reduced total PERK corresponds to reduced levels of phosphorylated PERK. In this scenario, if PERK was required for eIF2a phosphorylation and ISR activation in OPA1 BKO mice, OPA1/PERK DKO mice would have had reduced (or normalized) and not increased eIF2a phosphorylation and ATF4 induction.

7. On page 10 of the text, lines 4-5 (in reference to Figure 4), shouldn't there be a NOT in this sentence?

We understand this may sound confusing. We write “Indeed, the group effect from the ANCOVA analysis for oxygen consumption rates in relation to body weight was not significantly different (Figure 4P), suggesting GDF15 may mediate the increase in metabolic rates observed in OPA1 BAT KO mice in DIO”. We believe, that when examined in light of the new data added on OPA1 BKO mice that this interpretation should make more sense. In the DIO studies, OPA1 BKO mice had elevated metabolic rates, which we believe contributed to the lean phenotype in these mice. Conversely, when GDF15 is deleted in the OPA1 BKO background, mice lack this increase in metabolic rates. Therefore, we concluded that GDF15 is required for this adaptation, and, therefore, upon its deletion, mice gain significantly more weight than the OPA1 BKO.

Reviewer #3 (Recommendations for the authors):

The report by Jena et al. describes the observation that genetic inhibition of both OPA1 and GDF15 using UCP1-cre leads to reductions in circulating GDF15 and reductions in body mass. These effects are associated with increases in EE during the dark but not light cycle, lower fat mass and reduced expression of thermogenic genes in WAT. The findings are interesting but there are many incomplete findings presented which require further clarification.

1. The data around the assessment of energy expenditure is incomplete and inconsistent. For example in Figure 2 the authors present the EE relative to body but no other data from this experiment is provided. In Figure 4F they show EE is elevated in the dark cycle but not the light cycle. They then plot VO2 relative to body mass in an n=4 mice and show no difference between groups. It is unclear where this data came from and if it was the same experiment and if so whether this was collected in the dark or light cycle? In both Figure 2 and Figure 4 VO2, VCO2, and RER should be presented as it is unclear whether a shift to Fatty acid oxidation might be the primary contributor to the increase in the calculated EE. In addition the sample size for all measures is very low making it possible the results may be due to a type 1 error.

We appreciate the reviewer’s comments. Although we were unable to add more mice to these studies, in this revised manuscript we include additional metabolic data on all mouse models, including VO2, VCO2 and RER. Overall, our data reinforces the conclusion that GDF15 is likely mediating changes to resting metabolic rates in OPA1 BKO mice. Increased RER under obesogenic conditions in OPA1BKO mice, were abrogated in DKO mice, suggesting that GDF15 might also regulate fuel preference. We have included this in our discussion.

2. The authors suggest that the increase in EE may be leading to the reduced body mass of the DKO mice. To validate this hypothesis authors should complete pair feeding experiments as it is currently unclear whether these relatively small changes in EE are critical for the phenotype which is central to the primary conclusions.

Unfortunately, due to time constrains and limited number of animals, we were unable to perform pair-feeding experiments. Nonetheless, our data argues against food intake being the main driver in our phenotype. Despite having elevated GDF15 levels, OPA1 BKO mice have increased food intake under obesogenic conditions (Figure 1), which is not detected in OPA1/GDF15 BKO mice. If food intake was driving the changes in body weight, then DKO mice should weigh less than OPA1 BKO mice, but the opposite is observed. Meanwhile, resting metabolic rates are increased in OPA1 BKO mice fed HFD, but are unchanged in DKO mice. Therefore, our data points to a role for higher energy expenditure as the driver of our phenotype rather than food intake.

3. It is now well recognized that housing mice outside of the thermoneutral zone masks changes in adaptive thermogenesis. Authors should complete experiments at thermoneutrality.

This is an important point. We previously performed experiments in OPA1 BKO mice reared at thermoneutrality (DOI: 10.7554/eLife.66519). Our data showed that even under thermoneutral conditions, metabolic rates were elevated in these mice, and body mass was reduced. Based on this initial data, we decided to conduct our studies in DKO mice under room temperature conditions. Nonetheless, in this revised manuscript we show metabolic data collected in mice housed under thermoneutral conditions for 7 days (Figure 1 and Figure 5). Our data show that resting metabolic rates were increased in both OPA1 BKO and OPA1/GDF15 DKO under these conditions.

4. Throughout the paper there is paucity of information besides the assessment of a few thermogenic genes about the morphology of the brown and white adipose tissue depots. Authors should provide adipose tissue histology images and ideally electron microscopy images detailing whether there may be defects in mitochondrial function.

We appreciate the reviewer’s suggestion. We have now included morphological data (histology and TEM for OPA1/GDF15 DKO – Figure 5) and performed mitochondrial respirations in BAT (OPA1/PERK DKO – Figure 3 and OPA1/GDF15 DKO mice – Figure 5).

5. Related to the above the authors suggest reduced WAT browning is the reason for the hypothermia with cold, however, it seems highly unlikely that given the relatively low thermogenic capacity of this tissue that this is the primary driver of the effect. For example, what about the very large reduction in BAT UCP1 in the DKO mice. Alternatively, did the authors examine whether there might be differences in shivering or skeletal muscle function?

We appreciate the opportunity to expand on this point. We indeed observe reduced expression of thermogenic genes in DKO mice in DIO and in response to cold exposure. Although UCP1 protein levels were unchanged under baseline conditions, UCP1-dependent respiration was greatly inhibited. Therefore, it is likely that deletion of GDF15 further impairs BAT thermogenesis rendering DKO mice severely cold intolerant. Although we initially thought that impaired thermogenesis resulted from impaired browning of iWAT, now we show data clearly demonstrating that browning is maintained after cold exposure, and tyrosine hydroxylase levels are unchanged, showing comparable sympathetic activation of iWAT.

Regarding shivering thermogenesis, unfortunately, we were unable to measure this directly, but molecular markers of muscle function (Serca1 and GLUT1) normally affected by cold exposure, were unchanged between cold exposed WT and DKO mice. Interestingly, we now show that Serca1a is induced in gastrocnemius muscle of OPA1 BKO mice (Figure 6 —figure supplement 1), which is inhibited when GDF15 is deleted in BAT (Figure 6V). Therefore, we conclude that although it is unlikely that defects in shivering thermogenesis are contributing to our phenotype, it is possible that non-shivering thermogenesis in muscle via futile calcium cycling may be part of the compensatory mechanisms maintaining core body temperature in cold-exposed OPA1 BKO mice. Indeed, a recent study suggest that recombinant GDF15 induces muscle thermogenesis to maintain energy expenditure in mice (DOI: 10.1038/s41586-023-06249-4).

Author details

1. Jayashree Jena

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration

   Contributed equally with

   Luis Miguel García-Peña

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9929-662X

2. Luis Miguel García-Peña

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft, Project administration

   Contributed equally with

   Jayashree Jena

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-8718-6490

3. Eric T Weatherford

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft

   Competing interests

   No competing interests declared

4. Alex Marti

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Formal analysis, Validation, Investigation, Methodology

   Competing interests

   No competing interests declared

5. Sarah H Bjorkman

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Formal analysis, Investigation, Methodology

   Competing interests

   No competing interests declared

6. Kevin Kato

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8466-7149

7. Jivan Koneru

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Investigation, Methodology

   Competing interests

   No competing interests declared

8. Jason H Chen

   Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

   Contribution

   Data curation, Investigation, Methodology

   Competing interests

   No competing interests declared

9. Randy J Seeley

   Department of Internal Medicine, University of Michigan, Ann Arbor, United States

   Contribution

   Resources, Validation, Writing – review and editing

   Competing interests

   R.J.S. has received research support from Fractyl, Novo Nordisk, AstraZeneca, and Eli Lilly. R.J.S. has served on scientific advisory boards for Novo Nordisk, CinRX, Scohia, Fractyl and Structure Therapeutics. R.J.S. is a stakeholder of Calibrate and Rewind

10. E Dale Abel

    Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

    Present address

    Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, United States

    Contribution

    Resources, Visualization, Writing – review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-5290-0738

11. Renata O Pereira

    Fraternal Order of Eagles Diabetes Research Center and Division of Endocrinology and Metabolism, Roy J. and Lucille A. Carver College of Medicine, University of Iowa, Iowa City, United States

    Contribution

    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Writing – original draft, Project administration, Writing – review and editing

    For correspondence

    renata-pereira@uiowa.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0001-5809-4669

• 453

  Page views

• 73

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.