Maternal secretin ameliorates obesity by promoting white adipose tissue browning in offspring

Abstract

Our knowledge of the coordination of intergenerational inheritance and offspring metabolic reprogramming by gastrointestinal endocrine factors is largely unknown. Here, we showed that secretin (SCT), a brain‐gut peptide, is downregulated by overnutrition in pregnant mice and women. More importantly, genetic loss of SCT in the maternal gut results in undesirable phenotypes developed in offspring including enhanced high‐fat diet (HFD)‐induced obesity and attenuated browning of inguinal white adipose tissue (iWAT). Mechanistically, loss of maternal SCT represses iWAT browning in offspring by a global change in genome methylation pattern through upregulation of DNMT1. SCT functions to facilitate ubiquitination and degradation of DNMT1 by activating AMPKα, which contributes to the observed alteration of DNMT1 in progeny. Lastly, we showed that SCT treatment during pregnancy can reduce the development of obesity and improve glucose tolerance and insulin resistance in offspring of HFD‐fed females, suggesting that SCT may serve as a novel biomarker or a strategy for preventing metabolic diseases.

Maternal secretin promotes white adipose tissue browning in the offspring by inducing a global change in genome methylation pattern through the DNA methyltransferase DNMT1.

Introduction

In the past century, obesity and type 2 diabetes have become one of the major global public health concerns. It is known that an individual’s genetic and environmental factors influence obesity and/or diabetes development and progression. Besides, the genetic and environmental factors experienced by his/her parents can also contribute to one’s phenotypic plasticity regarding the development of metabolic diseases (Rando & Simmons, 2015). Recently, a growing body of evidence suggests a critical involvement of maternal diet as a metabolic modulator that dictates offspring in response to various nutritional conditions and metabolic stress (Ganu et al, 2012). It has been found that both severe under‐ or overnutrition in women during pregnancy can increase the risk of the development of metabolic dysfunction in offspring (Sullivan et al, 2011; Radford et al, 2014). Consistently, the effects of maternal obesity caused by an overnutrition diet can affect her offspring (Smith & Ryckman, 2015). Various epigenetic regulatory mechanisms, such as DNA methylation, histone modification, and noncoding RNAs, provide a plausible link between nutritional status changes early in development and the susceptibility to developing metabolic diseases later in life (Zhu et al, 2019).

Adipose tissue is highly dynamic and sensitive and can undergo adaptive changes in response to nutritional and environmental stimuli (Lee et al, 2012). The inducible browning adipocytes (also called brown‐like/beige adipocytes) offer a beneficial means to combat obesity and its metabolic consequences (Zou et al, 2017). Beige and brown adipocytes regulate whole‐body glucose and lipid homeostasis in mice and humans by dissipating energy from glucose and fatty acid oxidation (Scheja & Heeren, 2016; Park et al, 2019; Xu et al, 2021). To date, as the adult human brown adipose tissue can be identified as likely beige adipocytes by FDG‐PET, it has opened the door for taking advantage of these cells as a therapeutic strategy for metabolic disorders (Wang & Seale, 2016). White adipose tissue represents a prime target of metabolic programming induced by the maternal milieu (Moreno‐Mendez et al, 2020). Pregnancy is a critical plastic period in the biological programming of adipose tissue, as adipocyte development and physiology may be modelled by several maternal factors such as genetics, environmental influences, epigenetic changes, nutritional status, hormonal signaling, and energy balance (Borengasser et al, 2013; Lukaszewski et al, 2013; Alzamendi et al, 2021). Nevertheless, how maternal nutritional perturbations affect the programming of adipose tissue in the fetus and subsequently in the next generation remains unclear.

It has been indicated that gastrointestinal endocrine, composed of intestinal endocrine system, gut‐brain axis and nutrient‐sensing receptors, plays a key role in regulating nutrient absorption and energy metabolism, suggesting that the intestine is able to serve as an endocrine organ that modulates diabetes and obesity (Efeyan et al, 2015; Mace et al, 2015). Secretin (SCT) is a classical gastrointestinal peptide, which is required not only for digestion but also for energy metabolism (Klein, 1994; Konturek et al, 2003; Cheng et al, 2011; Li et al, 2018c). Additionally, SCT is found in extraembryonic tissues as early as embryonic day 7.5 (E7.5) and continued to be expressed in the placenta until E18.5 (Knox et al, 2011), suggesting that it may play a role in pregnancy or fetal development.

In this study, we explored the effects of maternal SCT on metabolic programming in offspring. We observed that the progeny from maternal SCT‐deficient mice exhibit altered metabolic homeostasis and display more sensitivity to high‐fat diet (HFD)‐induced obesity by decreasing the browning of inguinal white adipose tissue. Maternal SCT alters metabolic homeostasis of offspring by reprogramming white adipose tissue DNA methylation, which is dependent on the ubiquitination and proteasomal degradation of DNMT1 that in turn is regulated by AMPK signaling.

Results

SCT level is regulated by maternal nutritional status during pregnancy

First, we measured the transcript levels of SCT receptor (SCTR) and SCT in various tissues, we found that both SCTR and SCT were highly expressed in the placenta (Figs 1A and EV1A). Interestingly, the SCTR levels in the embryo were gradually increased during pregnancy (Fig EV1B), suggesting a functional involvement of SCT in fetal development. Furthermore, in a cohort of human pregnant subjects, significantly negative correlations between serum levels of SCT and TG, TC and NEFA were observed (Fig 1B), showing the interactions of SCT with maternal lipid metabolism. To assess the effects of SCT in metabolic reprogramming in response to maternal nutritional status, in addition to normal lab chow, we also used the healthy whole grain diet (WGD) and a commercialized high‐fat diet (HFD) to feed pregnant mice (Liu et al, 2020). Compared with controls, there were no changes in body weight but reductions in serum TG, TC, and NEFA in WGD mice, while HFD led to elevations in body weight and serum TG, TC, and NEFA (Fig EV1C and D). Intriguingly, WGD not only increased the SCT levels in serum and embryo in pregnant mice, but also promoted the protein expression of SCT in duodenum, while opposite effects were found in HFD‐fed pregnant mice (Fig 1C–E). As PKA contributes to SCT release (Sundaresan et al, 2012), the effects of WGD and HFD on PKA activities in pregnant mice duodenum were investigated. WGD stimulated the phosphorylation of PKA substrates, the expression of PKACα and phosphorylated PKAC, while HFD repressed the expression of phospho‐PKA substrates, phosphorylated PKAC and PKACα (Fig EV1E). Together, these results show that SCT level is regulated by maternal nutritional status during pregnancy.

Figure 1. SCT level is regulated by maternal nutritional status during pregnancy.

• A
  qPCR analysis of relative SCTR levels in different tissues [heart (He), intestine (Int), inguinal WAT (Ing), epididymal WAT (Ep), brown fat adipose (BAT), liver (Liv), hypothalamus (Hypo), placenta (Pla), skeletal muscle (Mus), kidney (Kid)] from C57BL/6J mice as indicated (n = 5 per group).
• B
  Correlation of serum SCT levels and serum triglycerides (TG), total cholesterol (TC) and nonesterified fatty acids (NEFA) levels in pregnant women in the middle trimester to the early third trimester (n = 15).
• C, D
  Plasma SCT levels of dams (C) at gestational day 18.5 (GD18.5) and embryo (D) at E18.5 (n = 8 per group).
• E
  Representative images of SCT immunohistochemical staining of duodenum from dams on standard (CON), whole grain (WG) and high‐fat (HFD) diet at GD18.5. SCT expression of WG is highest, but lower in HFD compared with CON. Scale bar, 50 μm (n = 10 per group). Right part shows quantification of SCT positive cells in duodenum from dams of three groups.

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. **P < 0.01; ***P < 0.001.

Source data are available online for this figure.

Figure EV1. SCT level is regulated by maternal nutritional status during pregnancy, related to Fig 1 .

1. qPCR analysis of relative SCTR levels in different tissues [heart (He), inguinal WAT (Ing), brown fat adipose (BAT), liver (Liv), hypothalamus (Hypo), skeletal muscle (Mus), duodenum (Duod), embryo (Embr), placenta (Pla)] from WT mice as indicated (n = 5 per group).
2. qPCR analysis of SCTR expression of placenta between embryonic days 14.5‐20.5 (E14.5‐E20.5) (n = 5 per group).
3. Body weight of CON, WG, and HFD dams at GD18.5. (n = 8 per group). Right part shows change of cumulative food intake of CON, WG and HFD dams between GD2‐GD18.
4. Serum TG, TC, and NEFA levels in dams on standard (CON), whole grain (WG) and high‐fat (HFD) diet at GD18.5. (n = 8 per group).
5. Western blot analysis detecting the phosphorylated status of protein kinase A (PKA) substrates, PKAα catalytic subunit, p‐PKAC (Thr197) and β‐actin of duodenum from dams at GD18.5.

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not statistically significant.

Source data are available online for this figure.

Maternal SCT deficiency promotes HFD‐induced obesity in offspring

To investigate the potential role of maternal SCT on fetal metabolic reprogramming, we initially generated a villus and crypt epithelial cell‐specific SCT knockout mouse model (vSCT KO) by the Cre‐lox system (SCT^f/f; villin‐Cre^+/−), and the floxed SCT mice (SCT floxed) was used as controls. Consistently, SCT mRNA and protein levels in duodenum, and serum, were greatly reduced in vSCT KO when compared with the floxed mice (Fig EV2A–C). Furthermore, we compared the male floxed SCT progeny (SCT^f/+; villin‐Cre^−/−) obtained by breeding male WT with control SCT floxed females (M‐SCT floxed, mother SCT floxed) or vSCT KO females (M‐SCT KO, mother SCT KO) (Fig 2A). There was no difference in metabolites and characteristics between SCT floxed and SCT KO dams (Table EV1). Interestingly, we found that the levels of SCT in embryos were also markedly reduced in M‐SCT KO mice (Fig EV2D); however, SCT and SCT mRNA expression levels in M‐SCT KO mice were indistinguishable from M‐SCT floxed mice in adulthood (Fig EV2E and F). Then, we measured the postnatal SCT levels in offspring to explore when SCT levels in M‐SCT KO mice were comparable to those in M‐SCT floxed mice. It was found that M‐SCT KO mice also had significantly lower SCT levels than M‐SCT floxed mice at postnatal day 1 (P1), but the SCT level of M‐SCT KO mice returned to normal after P7 (Fig EV2G). Although we did not observe any differences in the size of the fetus after birth in M‐SCT KO mice (Fig EV2H), M‐SCT KO mice on a chow diet exhibited a slightly but significantly higher body weight and impaired glucose tolerance compared with control mice (Fig EV2I and J). Interestingly, on an HFD, M‐SCT KO mice were obese and gained more body weight than controls (Fig 2B). Besides that, the increased body weight in M‐SCT KO mice is the direct result of weight gain only in the adipose tissue (Fig 2C). Consistently, HFD‐fed M‐SCT KO mice displayed elevated random plasma glucose and insulin levels, with reduced glucose tolerance and insulin sensitivity (Fig 2D–G). Moreover, the levels of TG, TC, and NEFA in HFD‐fed M‐SCT KO mice serum were higher than those in M‐SCT floxed mice (Fig 2H). However, unlike the adult phenotype, the offspring mice did not have any difference in metabolic profile at weaning (Fig EV2K–N). In addition, we found no difference in energy intake in adult offspring under chow diet, whereas the energy intake of M‐SCT KO mice was greater with a high‐fat diet (Fig EV2O). Considering that SCT KO females were SCT deficient before pregnancy, a polyclonal SCT antibody was used to neutralize the endogenous activity of SCT in wild‐type dams (Anti‐SCT). Likewise, we found no differences in metabolic phenotype between dams (Table EV2), while M‐Anti‐SCT mice were more prone to obesity after a high‐fat diet (Fig EV2P–U). Taken together, these data suggest that maternal SCT may play a role in obesity and energy homeostasis of offspring.

Figure EV2. Maternal SCT deficiency promotes HFD‐induced obesity in offspring, related to Fig 2 .

• A–D
  Target‐SCT knockout efficiency. (A) Relative level of SCT mRNA, of SCT floxed and vSCT KO mice (n = 5 per group). (B) Representative images of SCT immunohistochemical staining of duodenum, from SCT floxed and SCT KO mice. Plasma levels of SCT in SCT floxed and vSCT KO dams at GD18.5 (C), and in embryos of M‐SCT floxed and M‐SCT KO (D) (n = 8 per group).
• E
  qPCR analysis of relative SCT levels in different tissues [inguinal WAT (Ing), brown fat adipose (BAT), heart (He), liver (Liv), duodenum (Duod), skeletal muscle (Mus), hypothalamus (Hypo)] from 8‐week‐old M‐SCT floxed and M‐SCT KO mice (n = 8 per group).
• F
  Plasma levels of SCT in 8‐week‐old M‐SCT floxed and M‐SCT KO mice (n = 8 per group).
• G
  Plasma levels of SCT in M‐SCT floxed and M‐SCT KO mice at postnatal day 1 (P1), postnatal day 3 (P3) and postnatal day 7 (P7) (n = 8 per group).
• H
  Representative photograph of M‐SCT floxed and M‐SCT KO at postnatal day 1 (P1).
• I
  Body weight of 8‐week‐old M‐SCT floxed and M‐SCT KO offspring with chow diet (n = 8 per group).
• J
  Intraperitoneal glucose tolerance test (GTT) (1.0 g glucose/ kg body weight) in 8‐week‐old mice (n = 5 per group) and AUCs were calculated. Mice were fasted for 18 h before assay.
• K
  Body weight of M‐SCT floxed and M‐SCT KO offspring at weaning. (n = 8 per group).
• L
  The weight of different tissues from M‐SCT floxed and M‐SCT KO mice at weaning. (n = 8 per group).
• M
  Random blood glucose levels in M‐SCT floxed and M‐SCT KO mice at weaning. (n = 8 per group).
• N
  Serum TG, TC and NEFA levels in M‐SCT floxed and M‐SCT KO mice at weaning (n = 8 per group).
• O
  Energy intake of M‐SCT floxed and M‐SCT KO offspring with chow diet (left) and HFD (right) (n = 8 per group).
• P
  Body weight of M‐CON and M‐Anti‐SCT mice with HFD. (n = 8 per group).
• Q
  Random blood glucose levels in M‐CON and M‐Anti‐SCT mice with HFD. (n = 8 per group).
• R
  Insulin levels in M‐CON and M‐Anti‐SCT mice with HFD. (n = 8 per group).
• S
  Intraperitoneal glucose tolerance test (GTT) (1.0 g glucose/ kg body weight) in M‐CON and M‐Anti‐SCT mice with HFD (n = 6 per group) and AUCs were calculated. Mice were fasted for 18 h before assay.
• T
  Intraperitoneal insulin tolerance test (ITT) (0.75 U insulin/kg body weight) in M‐CON and M‐Anti‐SCT mice with HFD (n = 6 per group) and AUCs were calculated. mice were fasted for 6 h before assays.
• U
  Serum TG, TC and NEFA levels in M‐CON and M‐Anti‐SCT mice with HFD (n = 8 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data in panel M have been analyzed using ANCOVA with energy intake as dependent variable, group as fixed variable and body weight as covariate. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001, ns, not statistically significant.

Source data are available online for this figure.

Figure 2. Maternal SCT deficiency promotes HFD‐induced obesity in offspring.

1. Breeding scheme. M‐SCT KO offspring were derived by breeding a SCT^+/+, Cre^−/− male and SCT^f/f, Cre^+/− female; M‐SCT floxed offspring were derived by breeding a SCT^+/+, Cre^−/− male and SCT^f/f, Cre^−/− female.
2. Change of body weight of M‐SCT floxed and M‐SCT KO mice upon HFD feeding (n = 8 per group). Right part shows representative photograph of M‐SCT floxed and M‐SCT KO mice with HFD.
3. The weight of different tissues from M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
4. Random blood glucose levels in M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
5. Insulin levels in M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
6. Intraperitoneal glucose tolerance test (GTT) (1.0 g glucose/ kg body weight) in M‐SCT floxed and M‐SCT KO mice with HFD (n = 5 per group) and AUCs were calculated. Mice were fasted for 18 h before assay.
7. Intraperitoneal insulin tolerance test (ITT) (0.75 U insulin/kg body weight) in M‐SCT floxed and M‐SCT KO mice with HFD (n = 5 per group) and AUCs were calculated. mice were fasted for 6 h before assays.
8. Serum TG, TC and NEFA levels in M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. **P < 0.01; ***P < 0.001; ns, not statistically significant.

Source data are available online for this figure.

Maternal SCT promotes iWAT browning in offspring

In vivo indirect calorimetry was used to monitor the differences in energy homeostasis to explore the potential mechanisms of maternal SCT deficiency‐induced obesity. Indeed, upon HFD feeding, oxygen consumption was decreased in M‐SCT KO mice compared with M‐SCT floxed mice (Fig 3A). Interestingly, we also observed a decrease in heat production (energy expenditure) in M‐SCT KO mice (Fig 3B). Consistent with the decreased heat production, M‐SCT KO mice maintained a lower body temperature when exposed to a temperature of 4°C (Fig 3C). As interscapular brown adipose tissue (iBAT) and skeletal muscle play pivotal roles in regulating energy expenditure and body temperature, we then monitored thermogenesis in iBAT and skeletal muscle. The expression levels of thermogenic genes did not change in the iBAT and skeletal muscle, and the histological morphology of iBAT, of M‐SCT KO mice, compared with control mice (Fig EV3A–C), indicating that decreased energy expenditure in M‐SCT KO mice was most likely not due to iBAT and skeletal muscle.

Figure 3. Maternal SCT promotes iWAT browning in offspring.

• A, B
  Time‐resolved oxygen consumption (A) and heat production (B) in 48‐h light/dark cycle measured by CLAMS in M‐SCT floxed and M‐SCT KO mice with HFD (n = 6 per group), dark phase is marked as dark background. Regression‐based absolute oxygen consumption (A) or heat production (B) analysis of body weight are shown in the right part.
• C
  Body temperature of 8‐week‐old M‐SCT floxed and M‐SCT KO male offspring with chow diet during cold exposure (4°C; n = 8).
• D
  Representative HE staining and UCP1 staining of Ing from M‐SCT floxed and M‐SCT KO mice with chow diet maintained at RT or 3 days of cold exposure (4°C). Scale bar, 100 μm.
• E
  Relative mRNA levels of UCP1, CIDEA, PRDM16, PGC1α and DIO2 in iWAT from M‐SCT floxed and M‐SCT KO mice with chow diet (n = 8 per group).
• F
  Representative western blots of UCP1, PRDM16, PGC1α and HSP90 in iWAT from M‐SCT floxed and M‐SCT KO mice with chow diet after 3 days of cold exposure (4°C).
• G
  Total UCP1 protein per mg iWAT of M‐SCT floxed and M‐SCT KO mice with chow diet maintained at RT or exposed to cold for 3 days (4°C) (n = 3).
• H
  Relative mRNA levels of UCP1, PGC1α, PRDM16, CIDEA and DIO2 in SVF isolated from M‐SCT KO and M‐SCT floxed mice with HFD (n = 8 per group).
• I, J
  OCR of Oligomycin, FCCP and Antimycin A/Rotenone‐treated matured adipocytes derived from iWAT‐SVF of M‐SCT KO and M‐SCT floxed mice with HFD, and the AUC of OCR as indicated were calculated (n = 5 per group).
• K
  Glucose oxidation in adipocytes derived from iWAT‐SVF of M‐SCT KO and M‐SCT floxed mice with HFD (n = 8 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data in panel A and B have been analyzed using ANCOVA with oxygen consumption or heat production as dependent variable, group as fixed variable and body weight as covariate. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not statistically significant.

Source data are available online for this figure.

Figure EV3. Maternal SCT promotes iWAT browning of offspring, related to Fig 3 .

1. Relative mRNA levels of UCP‐1, PGC1α, PRDM16 and CIDEA in iBAT from M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
2. Representative HE staining and UCP‐1 staining of iBAT from M‐SCT floxed and M‐SCT KO mice with HFD. Scale bar, 100 μm.
3. Relative mRNA levels of UCP3, SERCA1a and SLN in skeletal muscle from M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
4. Relative mRNA levels of UCP1, PGC1α, PRDM16, ELVOL3, CIDEA, COX8B, DIO2 and COX7A1 in iWAT from M‐SCT floxed and M‐SCT KO mice with HFD (n = 8 per group).
5. Relative mRNA levels of UCP1, PRDM16, PGC1α, COX7A1, COX8B, CIDEA and DIO2 in iBAT from M‐SCT KO and M‐SCT floxed mice with chow diet after 3 days of cold exposure (4°C) (n = 8 per group).
6. Representative HE staining and UCP‐1 staining of iBAT from M‐SCT floxed and M‐SCT KO mice with chow diet after 3 days of cold exposure (4°C). Scale bar, 100 μm.
7. Relative mRNA levels of UCP3, SERCA1a and SLN in skeletal muscle from M‐SCT floxed and M‐SCT KO mice with chow diet after 3 days of cold exposure (4°C) (n = 8 per group).
8. Total UCP1 protein per mg iBAT of M‐SCT floxed and M‐SCT KO mice with chow diet maintained at RT or 3 days of cold exposure (4°C) (n = 3).
9. Relative mRNA levels of C/EBPα, C/EBPβ, C/EBPδ, PPARγ and Ap2 in SVF isolated from M‐SCT KO and M‐SCT floxed mice with HFD (n = 8 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. **P < 0.01; ***P < 0.001; ns, not statistically significant.

Source data are available online for this figure.

On the other hand, the browning of iWAT also contributes to elevated energy expenditure and adaptive thermogenesis (Ikeda et al, 2017; Wu et al, 2021). Therefore, we investigated whether the browning of iWAT was altered in M‐SCT KO mice. Interestingly, deficiency in maternal SCT was found to reduce expression of thermogenic genes in iWAT of HFD‐fed mice (Fig EV3D). Additionally, defective iWAT browning was also detected in M‐SCT KO mice at room temperature or cold exposure conditions, accompanied by enlarged adipocytes, lowered vascular density, and reduced mRNA expression of UCP1 and other thermogenic genes (Fig 3D and E). Consistently, the expressions of thermogenic genes or histological morphology in iBAT and skeletal muscle from M‐SCT KO mice appeared indistinguishable from those from M‐SCT floxed mice after cold exposure (Fig EV3E–G). Next, we compared the UCP1 content in iBAT and iWAT from M‐SCT KO mice and control mice. UCP1 protein level in iBAT was not altered in M‐SCT KO mice (Fig EV3H). In contrast, a decreased level of UCP1 protein was observed in the iWAT of M‐SCT KO mice compared with M‐SCT floxed mice under cold stress (Fig 3F and G). To further explore the role of maternal SCT in offspring iWAT, in vitro experiments were performed. Consistently, the mRNA expression levels of thermogenic and adipogenic genes were down‐regulated in matured adipocytes derived from stromal vascular fraction (SVF) isolated from iWAT of HFD‐fed M‐SCT KO mice (Figs 3H and EV3I). We also found that the basal, uncouple and maximal respiration were markedly decreased in mature adipocytes derived from iWAT‐SVF of M‐SCT KO mice (Fig 3I and J). Furthermore, adipocytes derived from iWAT‐SVF of M‐SCT KO mice displayed lower glucose oxidation (Fig 3K). Based on these data, reduced adaptive thermogenesis due to iWAT browning was one of the important reasons for the disrupted metabolic homeostasis observed in M‐SCT KO mice.

SCT participates in methylation‐dependent reprogramming of offspring

As epigenetic modifications, especially genome methylation, provide a plausible link between the intrauterine environment and phenotypic variation in offspring, we performed MeDIP‐Seq (immunoprecipitation using antibodies against 5mC followed by deep sequencing) to compare methylated DNA present in iWAT of M‐SCT floxed and M‐SCT KO mice (Weber et al, 2005; Jacinto et al, 2008). The DNA differentially methylated region (DMR) peaks revealed an increased abundance of both the total and the promoter upregulated DMR peaks (Fig 4A). In addition, the proportional distribution of hypo‐ and hypermethylated DMRs, as well as the distribution of DMR peaks within the genome, were consistent with the results of global DNA methylation patterns (Figs 4B and EV4A), indicating that DNA methylation patterns in iWAT of M‐SCT KO had undergone widespread changes. As expected, promoters and enhancers showed much greater hypermethylated DMRs, as indicated by heat maps of DMRs based on the count value or scatter plots and volcano plots with significant differences points (Fig 4C–F). Furthermore, similar results were obtained for superenhancer regions and lncRNA‐associated DMRs in promoter regions (Fig EV4B–F). We next performed KEGG pathway analyses and found that genes in hypermethylated DMRs‐associated pathways were closely related to cGMP‐PKG pathways, which are involved in numerous physiological systems, especially in enhancing differentiation and thermogenesis in adipocytes (Fig 4G). As expected, the Gene Ontology (GO) functional analysis showed that genes associated with hypermethylated DMRs were closely related to energy metabolism (Fig 4H). On the other hand, the hypermethylation of a specific locus is critical for transcriptional repression and developmental reprogramming (Dawson & Kouzarides, 2012). Thus, promoter‐specific methylation levels of vital thermogenesis‐related genes were analyzed using MeDIP‐qPCR. As aberrant DNA methylation usually occurs around the transcription start site (TSS) within a CpG island (Maunakea et al, 2010), we identified CpG islands based on the determination and the size of DNA fragments used in MeDIP. Notably, the 5mC levels in the PRDM16 and CIDEA promoters, but not in UCP1 or DIO2 promoters, were increased in the iWAT of M‐SCT KO mice (Figs 4I and J, and EV4G and H). Taken together, our results indicate that a loss of maternal SCT led to global hypermethylation of the genome in iWAT of offspring, which was to a certain extent the key reason for metabolic disorder in M‐SCT KO mice.

Figure 4. SCT participates in methylation‐dependent reprogramming of offspring.

• A
  MeDIP analyses genomic features of iWAT DNA methylation. Shown is the number of total DMR peaks (left) or total DMR peaks within promoters (right), proportions of up‐ versus down‐ regulated DMR peaks (M‐SCT KO versus M‐SCT floxed) (FDR < 0.1).
• B
  Stacked bars showing the Proportional distribution of hypo‐ and hypermethylated DMRs (M‐SCT KO versus M‐SCT floxed) in different regions.
• C
  Heat map of DMR methylation level in mRNA between M‐SCT floxed and M‐SCT KO. The color bar indicates the normalized tag counts in M‐SCT floxed or M‐SCT KO. Positive z scores, shown in the table, indicate hypermethylated regions, while negative z scores imply hypomethylated regions.
• D
  As in (C), for enhancer region.
• E
  Scatter plot representation of the most significant difference of methylation in mRNA (left) and enhancer (right) regions between M‐SCT floxed and M‐SCT KO (P < 0.005).
• F
  Volcano plot DNA methylation represents all DMRs in mRNA (left) and enhancer (right) regions. Cutoffs (vertical and horizontal dashed lines) used to identify DMRs between M‐SCT floxed and M‐SCT KO mice. The vertical lines correspond to twofold up and down, respectively, and the horizontal line represents a P value of 0.01.
• G
  Selected pathways of genes with increased DNA methylation through KEGG analysis.
• H
  Enrichment of gene ontology (GO) biological processes of DMRs showing significant different methylation levels (P < 0.01) GO terms are plotted against negative log of corrected P‐value.
• I, J
  Schematic diagram showing the CpG islands of Mus musculus PRDM16 (I) and CIDEA (J) genes. The conditions to be met by CpG islands identified were: 200–300 bp length, CG content of > 50% and observed CpG content versus expected CpG content > 0.6. Methylation levels of each CpG islands in PRDM16 (I) and CIDEA (J) are presented as the fold change relative to the average value of the M‐SCT floxed group and β‐actin promoter was used as a negative control (n = 3 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not statistically significant.

Source data are available online for this figure.

Figure EV4. SCT participates in methylation reprogramming of offspring, related to Fig 4 .

• A
  Proportional genomic distribution of hypo‐ (right) and hypermethylated (left) DMRs (M‐SCT KO versus M‐SCT floxed).
• B–D
  Heat map of DMR methylation levels in lnc RNA (B), small ncRNA (C) and superenhancer (D) regions between M‐SCT floxed and M‐SCT KO. The color bar indicates the normalized tag counts in M‐SCT floxed or M‐SCT KO. Positive z scores, shown in the table, indicate hypermethylated regions, while negative z scores imply hypomethylated regions.
• E
  Scatter plot representation of the most significant difference of methylation in lncRNA (left) and superenhancer (right) regions between M‐SCT floxed and M‐SCT KO (P < 0.005).
• F
  Volcano plot DNA methylation represents all DMRs in lncRNA (left) and superenhancer (right) regions. Cutoffs (vertical and horizontal dashed lines) used to identify DMRs between M‐SCT floxed and M‐SCT KO mice. The vertical lines correspond to twofold up and down, respectively, and the horizontal line represents a P value of 0.01.
• G, H
  Schematic diagram showing the CpG islands of Mus musculus UCP1 (G) and DIO2 (H) genes. Regions from −1,000 to +1,000 bp are depicted relative to the TSS (at 0). Vertical lines represent CpG sites and the location of CpG islands with highest density of CpG dinucleotides are detailed by red stripes (top). The conditions to be met by CpG islands identified were: 200–300 bp length, CG content of > 50% and observed CpG content versus expected CpG content > 0.6. Enrichment of 5mC at the promoter‐specific regions against the input normalized by the positive control in iWAT were analyzed by MeDIP‐qPCR (bottom). Methylation levels of each CpG islands in UCP1 (G) and DIO2 (H) are presented as the fold change relative to the average value of the M‐SCT floxed group and β‐actin promoter was used as a negative control (n = 3 per group).

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. ns, not statistically significant.

Source data are available online for this figure.

SCT represses DNMT1 expression by activating AMPKα

Mechanistically, the levels of DNA methyltransferases DNMT1, DNMT3a, and DNMT3b proteins were measured to determine the molecular mechanism underlying changes in DNA methylation in the iWAT of M‐SCT KO mice. Intriguingly, the loss of maternal SCT resulted in an increase in DNMT1 protein levels in iWAT of offspring mice, while the levels of DNMT3a and DNMT3b remained unchanged (Fig 5A). Notably, we did not observe any alteration in transcript levels of DNMT1, DNMT3a, and DNMT3b in the iWAT of M‐SCT KO mice (Fig 5B), suggesting that the regulation of DNMT1 might occur at the posttranscriptional level. As SCT is known to increase intracellular cAMP levels (Godoy et al, 2014), we speculated that AMP‐activated protein kinase α subunit (AMPKα) might participate in maternal SCT‐regulated DNA methylation reprogramming in embryos. As expected, we observed a reduced level of phospho‐AMPKα in the embryos of M‐SCT KO mice that was accompanied by a decrease in the levels of DNMT1 and phosphorylated acetyl‐CoA carboxylase (ACC), a downstream target protein of AMPKα (Fig 5C). Furthermore, we exposed mouse embryonic fibroblasts (MEFs) to various concentrations of SCT, and levels of both the phospho‐AMPKα and phospho‐ACC proteins were increased by SCT in a dose‐dependent manner, which was also accompanied by a decrease in DNMT1 expression (Fig 5D). In addition, the SCT‐induced increase in the phospho‐AMPKα level was partially reversed in MEFs transfected with an SCTR‐specific siRNA, indicating that the increase in the level of phosphorylated AMPKα induced by SCT treatment in MEFs depended on the SCT receptor (Fig EV5A). Similar results were obtained when compound C (C.C., AMPKα inhibitor) and AICAR (AMPKα activator) were used. Consistently, AMPKα inhibition by C.C. treatment significantly increased the level of the DNMT1 protein in MEFs, while AMPKα activation repressed DNMT1 expression (Fig EV5B and C). Importantly, inhibition of AMPKα by C.C. rescued the decreased DNMT1 levels in MEFs treated with SCT, while activation of AMPKα by AICAR reversed the elevated DNMT1 expression in MEFs transfected with the SCTR siRNA (Figs 5E and F, and EV5D). The level of the DNMT1 protein was upregulated in AMPKα knockout MEFs in the absence or presence of SCT (Fig 5G), further validating that the altered expression of DNMT1 is a direct effect of derepression via SCT/AMPKα.

Figure 5. SCT inhibits DNMT1 protein levels by activating AMPKα.

• A
  Western blots analysis of DNMT1, DNMT3a, DNMT3b, and HSP90 in Ing from M‐SCT KO and M‐SCT floxed mice with HFD.
• B
  Relative mRNA levels of DNMT1, DNMT3a and DNMT3b in Ing from M‐SCT KO and M‐SCT floxed mice with HFD (n = 8). Statistical significance was determined using Two‐tailed Student’s t‐test. Data are presented as mean ± SD. ns, not statistically significant.
• C
  Western blots analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in embryo of M‐SCT KO and M‐SCT floxed mice at E18.5.
• D
  Western blot analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in mouse embryo fibroblast (MEF) cells treated with 10 or 100 nM SCT.
• E
  Western blot analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in MEF cells treated with 10 nM SCT or 20 μM compound C (C.C.) for 24 h.
• F
  Western blot analysis of p‐AMPKα, AMPKα, DNMT1, SCTR and HSP90. MEF cells were transfected with indicated siRNA and then treated with 1 mM AMPK agonist (AICAR) for 24 h.
• G
  Western blot analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in AMPK^+/+ or AMPKα1/α2 double knockout (AMPK^−/−) MEF cells treated with 10 nM SCT for 24 h.
• H, I
  Western blot analysis of Flag‐DNMT1 in Flag‐DNMT1‐transfected HEK293T cells treated with AICAR with indicated dose or time periods.
• J
  HEK293T cells were transfected with indicated plasmids and then treated with 10 nM SCT for 24 h, followed by IP and IB analysis.
• K
  HEK293T cells were transfected with indicated plasmids or siRNA and then lysed, followed by IP and IB analysis.
• L
  AMPK^+/+ or AMPK^−/− MEF cells were transfected with indicated plasmids and then lysed, followed by IP and IB analysis.

Source data are available online for this figure.

Figure EV5. SCT regulates methylation by inhibiting DNMT1 via activating AMPKα, related to Fig 5 .

• A
  Western blot analysis of p‐AMPKα, AMPKα, DNMT1, SCTR and HSP90. MEF cells were transfected with indicated siRNA and then treated with 10 nM SCT for 24 h.
• B
  Western blot analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in MEF cells treated with 20 μM compound C (C.C.) for 24 h.
• C
  Western blot analysis of p‐AMPKα, AMPKα, p‐ACC, ACC, DNMT1 and HSP90 in MEF cells treated with 1 mM AICAR for 24 h.
• D
  Western blot analysis of knockout efficiency of si‐SCTR in MEF cells.
• E, F
  Western blot analysis of p‐AMPKα, AMPKα, DNMT1 and HSP90 in MEF cells treated with 10 nM SCT (E) or 1 mM AICAR (F) for 24 h, and 5 μM MG‐132 for 8 h.
• G
  Western blot analysis of knockout efficiency of si‐SCTR in HEK293T cells.
• H, I
  HEK293T cells were transfected with indicated plasmids and then treated with 20 μM C.C. (H) or 1 mM AICAR (I) for 24 h, followed by IP and IB analysis.

Source data are available online for this figure.

SCT/AMPKα activation facilitates the ubiquitination and degradation of DNMT1

The finding that DNMT1 expression decreased in MEFs upon SCT/AMPKα activation prompted us to determine the mechanism underlying the attenuation of DNMT1 expression with SCT or AMPKα activation. Interestingly, AICAR treatment reduced flag‐DNMT1 protein levels in a dose‐ and time‐dependent manner (Fig 5H and I). Moreover, the DNMT1 protein was more stable in MEFs cultured in the presence or absence of MG‐132, a protease inhibitor, suggesting that the DNMT1 protein might undergo proteasome‐dependent degradation (Fig EV5E and F). Accordingly, SCT treatment markedly increased the ubiquitination of the flag‐DNMT1 protein (Fig 5J), while the effects of SCT treatment on the ubiquitination of the flag‐DNMT1 protein were partially abolished in HEK293T cells transfected with an SCTR‐specific siRNA (Figs 5K and EV5G), further suggesting that SCT is required for the ubiquitination of DNMT1. The inhibition of AMPKα by C.C. treatment significantly repressed the ubiquitination of flag‐DNMT1 (Fig EV5H), while AMPKα activation markedly promoted the ubiquitination of flag‐DNMT1 (Fig EV5I). A similar result was obtained from AMPKα KO MEFs, and AMPKα deficiency led to a decrease in the ubiquitination of DNMT1 (Fig 5L), indicating that AMPKα is required for the ubiquitination of DNMT1 induced by SCT treatment. Based on these findings, we demonstrate that the activation of SCT/AMPKα axis facilitates the ubiquitination and degradation of DNMT1.

SCT improves metabolic homeostasis in offspring of HFD‐fed pregnant mice

Our observations in this study suggest that maternal SCT negatively regulates genome methylation and the upregulation of maternal SCT during pregnancy, which serves as a protective mechanism to improve metabolic homeostasis in offspring. We then tested whether maternal SCT supplementation during pregnancy might protect against maternal HFD‐induced metabolic disorder in offspring (Fig 6A), as evident from the reduced body weight and lean phenotype of offspring mice of HFD‐fed mice treated with SCT during pregnancy (M‐HFD + SCT mice) compared with offspring mice of female mice fed an HFD during pregnancy (M‐HFD mice) (Fig 6B). Additionally, oxygen consumption and heat production were higher in M‐HFD+SCT mice than in M‐HFD mice upon HFD feeding (Fig 6C and D). As expected, both glucose tolerance and insulin sensitivity were improved in M‐HFD + SCT mice (Fig 6E and F). Serum TG, TC and NEFA levels were attenuated in M‐HFD + SCT mice compared with M‐HFD mice (Fig 6G). Consistent with these findings, the adipocyte cell size was smaller in iWAT from M‐HFD + SCT mice than in control iWAT (Fig 6H). Moreover, SCT treatment substantially increased the mRNA expressions of thermogenic genes in iWAT of offspring (Fig 6I), further confirming that SCT treatment during pregnancy increased thermogenesis and energy expenditure in offspring by promoting the browning of iWAT. Together, our data suggest that maternal SCT supplementation during pregnancy protects offspring mice against obesity, one of the possible reasons for which was the reprogramming of inguinal adipose tissue development in the offspring.

Figure 6. SCT improves metabolic homeostasis in offspring of HFD‐fed pregnant mice.

• A
  Serum SCT levels in M‐HFD and M‐HFD+SCT mice with HFD (n = 8 per group).
• B
  Change of body weight of M‐HFD (offspring of dams with high‐fat pregnant diets) and M‐HFD+SCT (offspring of dams with high‐fat pregnant diets supplied with SCT) mice upon HFD feeding (n = 8 per group). Representative photograph of M‐HFD and M‐HFD+SCT mice after HFD feeding.
• C, D
  Time‐resolved oxygen consumption (C) and heat production (D) in 48‐h light/dark cycle measured by CLAMS in M‐HFD and M‐HFD+SCT mice with HFD (n = 5 per group), dark phase is marked as dark background. Regression‐based absolute oxygen consumption (C) or heat production (D) analysis of body weight are shown in the right part.
• E
  GTT (1 g glucose/kg body weight) in M‐HFD and M‐HFD+SCT mice with HFD (n = 5 per group) and AUCs were calculated in right part. Mice were fasted for 18 h before assays.
• F
  ITT (0.75 U insulin/kg body weight) in M‐HFD and M‐HFD+SCT mice with HFD, and AUCs were calculated in bottom (n = 5 per group). Mice were fasted for 6 h before assays.
• G
  Serum TG, TC and NEFA levels in M‐HFD and M‐HFD+SCT mice with HFD. (n = 8 per group).
• H
  Representative HE staining and UCP1 staining of iWAT from M‐HFD and M‐HFD+SCT mice with HFD. Scale bar, 100 μm (n = 5 per group).
• I
  Relative mRNA levels of UCP1, PGC1α, PRDM16, CIDEA and DIO2 in iWAT from M‐HFD and M‐HFD+SCT mice with HFD (n = 5 per group).
• J
  Schematic diagram of the working model: maternal SCT reprogrammed inguinal adipose tissue development in offspring during pregnancy.

Data information: Statistical significance was determined using Two‐tailed Student’s t‐test. Data in panel C and D have been analyzed using ANCOVA with oxygen consumption or heat production as dependent variable, group as fixed variable and body weight as covariate. Data are presented as mean ± SD. *P < 0.05; **P < 0.01; ***P < 0.001.

Source data are available online for this figure.

Discussion

An understanding of nutritional intergenerational effect is of particular importance and plays a key role in maintaining metabolic homeostasis in offspring. Maternal adaption to nutritional or environmental stimuli is related to various physiological or pathological changes in offspring (McCurdy et al, 2009; Son et al, 2020). Based on accumulating evidence, maternal HFD exposure, especially gestational exposure, is associated with perturbed metabolism in offspring, which leads to obesity and impaired glucose/lipid metabolism in adult offspring (Sasson et al, 2015; Richards et al, 2016; Ribaroff et al, 2017). Thus, a better understanding of the molecular mechanism underlying the regulation of overnutritional intergenerational effects will provide new insights into metabolic phenotypic plasticity during pregnancy and strategies to prevent metabolic diseases. In this study, we identified secretin (SCT), a gastrointestinal endocrine peptide, as a critical regulator involved in intergenerational inheritance and offspring metabolic reprogramming that might prevent maternal HFD‐induced metabolic disorder in offspring and may serve as a therapeutic target.

SCT plays a key role in energy metabolism, including fatty acid metabolism and appetite control (Chu et al, 2009; Vu et al, 2015). Recently, SCT was shown to activate thermogenesis by stimulating lipolysis in brown adipocytes and promoting satiation in the hypothalamus, suggesting that SCT might serve as a peripheral communication hub between the gut and other tissues (Li et al, 2018b). Here, we found that SCT levels were reduced in both maternal and fetal circulation following HFD feeding, while the WGD stimulated SCT release suggesting that SCT may represent a metabolic stress‐responsive modulator that participates in fetal metabolic reprogramming in a cell‐autonomous manner. Consistent with these findings, the expression of gastrointestinal hormones, such as GLP1, is repressed in L cells upon high‐fat diet feeding (Richards et al, 2016).

In this study, we used progeny mice from villus and crypt epithelial cell‐specific SCT knockout mice (vSCT KO) and mouse models with upregulated SCT levels during pregnancy to study the effects of SCT on nutritional intergenerational effects. The ablation of SCT in maternal mice markedly caused diet‐induced obesity and insulin resistance in adult offspring mice. The anti‐obesity effects of maternal SCT may result from the increased energy expenditure induced by the browning of iWAT rather than BAT and skeletal muscle. Furthermore, the SCT treatment in HFD‐fed pregnant mice exerted positive effects, including promoting thermogenesis in iWAT and improving glucose tolerance and insulin resistance in offspring mice. We also believed that the increase in energy intake was also one of the important reasons for the offspring to be prone to obesity. However, whether the change in appetite was a direct result of the deficiency of SCT in dams or involved in the hypothalamic appetite‐regulating center of the offspring had not yet been determined, and further research is needed. As maternal SCT altered the reprogramming of iWAT metabolism in offspring, we proposed that maternal SCT potentially represents an exciting biomarker or pharmacological target to combat metabolic diseases.

Adipose tissue development is particularly active during the perinatal period, and its programming occurred in gestation may lead to obesity, with further intergenerational transmission and development of chronic diseases later in life (Borengasser et al, 2013). Although a lower increase in oxidative metabolism and energy expenditure than iBAT, and several studies questioned the significant contribution of white adipose tissue to energy expenditure (Pan et al, 2020), many endogenous, pharmacological, and nutritional factors could effectively trigger WAT browning through the activation of different cellular pathways, producing a significant outcome in beige cell recruitment (Wu et al, 2012; Nedergaard & Cannon, 2014; Montanari et al, 2017).

Maternal nutritional status potentially affects the metabolic phenotype of offspring via epigenetic mechanisms, such as DNA methylation, noncoding RNAs, and histone modification (Sales et al, 2017), which provide a faithful link between changes in the intrauterine nutritional status and metabolic reprogramming in offspring. We explored epigenetic changes to verify that the effect of maternal SCT on iWAT in offspring was a direct effect. DNA methylation plays an important role in the composition and stabilization of chromosomes, the growth of embryos and the control of gene expression. By performing in‐depth research, many studies have revealed that DNA methylation is related to obesity (Meissner et al, 2008). MeDIP‐seq data indicated that iWAT of M‐SCT KO mice exhibited a higher level of hypermethylated CpGs. As DMRs were mainly enriched in cGMP‐PKG, Akt, MAPK and fructose/glucose metabolism, which are specifically related to the browning process of adipocytes (Miegueu et al, 2013), the cytosine methylation patterns of vital thermogenic genes were subsequently determined. Most notably, a transcriptional coregulator that powerfully regulates the thermogenic gene program in beige adipocytes, PRDM16, exhibited generally higher methylation levels in M‐SCT KO mice than in M‐SCT floxed mice. Only methylation within/around the promoter region is associated with gene silencing, and the PRDM16 meets the characteristic of a “key developmental gene” with high CpG density promoters (Herman & Baylin, 2003; Meissner et al, 2008). Notably, the methylation level in PRDM16 CpG‐rich regions, especially those located near the TSS, was increased, consistent with the reduced gene expression in M‐SCT KO mice, indicating that PRDM16 expression is likely to be epigenetically regulated. This finding was consistent with recent reports that the degree of methylation of the PRDM16 promoter in adipose tissue has a complex relationship with metabolic dysfunction (Kim et al, 2015). Together, these results suggest that PRDM16 may be a key nexus that integrates maternal dietary information to control offspring metabolism via maternal SCT.

Based on accumulating evidence, elevated DNMT1 expression and activities induced by a maternal HFD contribute to abnormal DNA methylation in offspring and subsequent changes in metabolic phenotypic plasticity (Grissom et al, 2015; Ramaiyan & Talahalli, 2018). Here, a maternal SCT deficiency promoted DNA hypermethylation in the iWAT of offspring mice by increasing the expression of DNMT1. Compared with control mice, we did not detect any alternation in DNMT1 mRNA levels in M‐SCT KO mice inguinal white adipose tissue, consistent with the regulation of DNMT activity by posttranslational modification (Jeltsch & Jurkowska, 2016). Meanwhile, SCT promotes AMPKα activation and subsequently suppresses DNMT1 expression, which is essential for the maternal SCT action. Additionally, AMPKα inhibits DNMT expression through direct phosphorylation of Ser730 (Marin et al, 2017). Consistent with these results, we found that SCT/AMPKα activation promotes the ubiquitination and proteasomal degradation of DNMT1, which might contribute to the alteration in genome methylation in progeny from maternal SCT‐deficient mice. Thus, we speculate that the increased DNMT1 expression induced by SCT/AMPKα downregulation might be a plausible explanation for the hypermethylation and obese phenotype observed in M‐SCT KO mice. Nevertheless, the specific mechanisms underlying SCT/AMPKα‐mediated modulation of DNMT1 ubiquitination require further study.

Taken together, our study establishes SCT as a master regulator of intergenerational effects by coordinating DNA methylation and thermogenesis in iWAT of offspring (Fig 6J). Genetic downregulation of maternal SCT results in metabolic disorder, and maternal SCT treatment prevents the maternal HFD‐induced obesity phenotype in offspring. Mechanistically, SCT promotes the ubiquitination and proteasomal degradation of DNMT1, thus subsequently resulting in a decrease in DNMT1 expression and activity. Based on this discovered nutrition intergenerational regulatory potential of SCT, we propose that SCT may represent a potential therapeutic target to treat metabolic diseases involving maternal overnutrition. Through our study, we highlight the role of white adipose tissue browning in the intergenerational genetic regulation of SCT. However, changes in other tissues and key metabolic regulatory pathways cannot be ruled out as important causes for the phenotypic changes in offspring of SCT‐deficient dams.

Materials and Methods

In vivo study

Serum secretin, TG, TC, and NEFA levels were measured in fifteen pregnant volunteers in the middle second trimester to the early trimester (24–31 weeks) (30 ± 5 years, BMI 24 ± 5). The ethical committee evaluated and approved the studies, and all participants gave their written informed consent before any study procedures (clinical trial number: (GKLW) 2018‐19). C57BL/6 mice were maintained on a 12‐h light/dark cycle in a temperature‐ and humidity‐controlled facility with ad libitum access to food and water. Female mice (age, 8–10 weeks; weight, 23–25 g) were mated with wild‐type (WT) males fed on a chow diet. Pregnancy was dated by the presence of a vaginal plug (day 0.5) and males mice were removed immediately. Pregnant mice were divided into different groups according to their diet throughout gestation: standard (CON) diet (AIN‐93G‐based diet), whole grain (WG) diet or high‐fat (HFD) diet (D12492). All dams were started on their respective diets on gestational day 0.5 (GD0.5), body weight and food intake of dams were recorded daily during the 21‐day pregnancy. Plasma and tissue samples were obtained beginning at 8 a.m., and all overnight fasted mice were sampled within 2 h from mice fasted for 10–12 h.

Villus and crypt epithelial cells‐specific SCT knockout mice (vSCT KO, SCT^f/f; villin‐Cre^+/−) were generated using the Cre‐lox system by crossing SCT^f/f mice (Zhang et al, 2014) to villin‐Cre^+/− mice. Female vSCT KO and floxed mice (SCT^f/+; villin‐Cre^−/−) mice were mated with WT male fed on a chow diet. Offspring from vSCT KO females were designed as M‐SCT KO, and those from floxed mice mothers were designed as M‐SCT floxed. Both M‐SCT floxed and M‐SCT KO dams were maintained on the same AIN‐93G‐based diet. At birth, the litter size was aligned to 8, and different group of offspring mice did not from the same litter. More than three groups of littermates from each dam were analyzed in individual experiments. All dams were maintained on AIN‐93G‐based standard diet during lactation. At postnatal day 21, all pups were weaned onto normal chow diet for 8 weeks and recorded the changes in body weight and food intake. For cold stimulation, mice were maintained in a cold environment (4°C) for 3 days, and the rectal temperature was performed with a digital thermometer. Another batch of 8‐week‐old mice were fed an HFD for 8 weeks, then mice were individually placed into Comprehensive Laboratory Animal Monitoring System (CLAMS) cages to measure VO₂ and heat production. Serum and tissues were obtained for further analyses.

In the SCT supplementation study, dams fed an HFD were i.p. injected with modified SCT (HSDGTFTSELSRLREEARLKRLLQGLV), two injections per day from GD1.5 to GD18.5 (Wang et al, 2013), every 12 h, and the dose of each injection was 10 nmol/day. For endogenous SCT blocking experiment, pregnant WT mice fed a chow diet were s.c. injected with anti‐SCT antibody (Phoenix Pharmaceuticals, G‐067‐04) (500 pmol/mouse) (Anti‐SCT). All animal studies were conducted in accordance with guidelines and regulations of the Guide for the Care and Use of Laboratory Animals and approved by the Laboratory Animal Ethics Committee of Jiangnan University (permission no. JN. No 20180930c0231231[212]). All efforts were made to minimize animal suffering.

Cell culture, transfection and immunoprecipitation

HEK293T and MEF cells were cultured in DMEM (Gbico) containing 10% fetal bovine serum (Gbico), 100 units/ml penicillin and 100 μg/ml streptomycin in a humidified atmosphere of 5% CO₂ at 37°C. Cells were transfected with plasmids (Flag‐DNMT1, HA‐Ub) or siRNA oligos targeting SCTR (target sequence: mouse‐‐‐AAC ATG AAT GGC TCC TTT AAT GA, human‐‐‐CAG ACA CTT TCT GGA AGA TGT TG) at 70% confluence using Lipofectamine 3000 (Invitrogen) and Opti‐MEM (Invitrogen) according to the manufacturer protocols for 24 h, followed by treatment with 10 nM SCT (ab123762, abcam), 100 nM SCT, 1 mM AICAR, 20 μM compound C (C.C.) for 24 h, or 5 μM MG‐132 for 8 h. Cells were harvested 48 h after transfection.

For stromal vascular fraction (SVF) isolation, inguinal adipose tissues (iWAT) were dissected from 5‐week‐old vSCT KO and WT mice. Briefly, iWAT were minced and digested with 2% collagenase type I in DMEM for 30 min at 37°C, followed by quenching with complete medium. The cell suspension was centrifuged, filtered through a 70 μm filter (BD Biosciences) after washing, and then inoculated into a 10 cm dishes in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin (Invitrogen). The differentiation of SVF‐derived preadipocytes was induced in growth medium with 5 μg/ml insulin, 0.5 mM IBMX (Sigma), 1 μM DEX (Sigma), 1 nM T3 and 5 μM Rosiglitazone (Sigma) for 48 h, then replaced with growth medium supplemented with insulin, T3 and Rosiglitazone for 4 days. For glucose oxidation assay, differentiated adipocytes in a six‐well plates were incubated with DMEM containing 2% FBS for 2 h. Cells were incubated with KRB–HEPES buffer containing 2% BSA and 5 mM glucose in the presence of 0.5 µCi/ml [1‐¹⁴C] glucose at 37°C for 1 h. Subsequently, 30% hydrogen peroxide (350 µl) was added into each well, and [¹⁴C] CO₂ was trapped in the smears supplemented with 300 µl 1 M benzethonium hydroxide solution at room temperature for 20 min. Glucose oxidation was determined by counting radioactivity trapped in the wipe smears using a scintillation counter as previous described (Zhang et al, 2016).

Cells growing in 10‐cm dishes were rinsed once with cold PBS and lysed on ice in CHAPS buffer for 15 min (40 mM HEPES [pH 7.5], 120 mM NaCl, 1 mM EDTA, 10 mM pyrophosphate, 10 mM glycerophosphate, 50 mM NaF, EDTA‐free protease inhibitors [Roche], 0.3% CHAPS). Cell lysates were centrifuged at 13,000 g for 10 min, and the supernatant was used for immunoblotting analysis. For immunoblotting, protein was separated by sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE), heated at 100°C for 5 min, then electrophoresed and immunoblotted. For immunoprecipitation (IP), incubated the cell lysates containing overexpressed protein with specific antibodies and A/G Sepharose protein beads at 4°C overnight.

Biochemical analysis of serum

Maternal blood and fetal blood were collected after decapitation following laparotomy then centrifuged at 4°C at 3,000 r.p.m. for 5 min. Triglycerides (TG) and total cholesterol (TC) concentrations of serum were measured using commercial kits (TG, LabAssay, Cat#290‐63701; TC, LabAssay, Cat#294‐65801). Levels of plasma nonesterified fatty acids (NEFA) (Millipore), secretin (Cloud‐clone) and insulin (Millipore) were determined using ELISA kits.

Histology

Duodenum, iWAT, and iBAT were fixed in 4% paraformaldehyde for 24 h and then embedded in paraffin. Sections was used for immunohistochemical staining or hematoxylin and eosin (H&E) staining according to the standard protocols. Immuno‐stained with secretin (bs‐0088R, bioss), UCP‐1 (ab10983, abcam). The electron microscopic observations were conducted through scanning electron microscope (Axio Vert A1, Carl Zeiss AG).

RT‐PCR

RNA was extracted from different tissues from mouse using Trizol reagent (Invitrogen) as described before (Li et al, 2018a). One microgram of RNA was transcribed to cDNA using oligo (dT) and Superscript (Takara). Real‐time PCR was carried out by Applied Biosystems Stepone. The following primers were used as follows: SCTR F: 5′‐GTGGGCTGTCACCAGACAC‐3′, R: 5′‐ATCAGCAGGAGGGTGGACTT‐3′; UCP‐1 F: 5′‐AGGCTTCCAGTACCATTAGGT‐3′, R: 5′‐CTGAGTGAGGCAAAGCTGATTT‐3′; PGC1α F: 5′‐TTCATCTGAGTATGGAGTCGCT‐3′, R: 5′‐GGGGGTGAAACCACTTTTGTAA‐3′; PRDM16 F: 5′‐CCAAGGCAAGGGCGAAGAA‐3′, R: 5′‐AGTCTGGTGGGATTGGAATGT‐3′; ELVOL3 F: 5′‐GATGGTTCTGGGCACCATCTT‐3′, R: 5′‐CGTTGTTGTGTGGCATCCTT‐3′; CIDEA F: 5′‐TGACATTCATGGGATTGCAGAC‐3′, R: 5′‐GGCCAGTTGTGATGACTAAGAC‐3′; COX8B F: 5′‐TGTGGGGATCTCAGCCATAGT‐3′, R: 5′‐AGTGGGCTAAGACCCATCCTG‐3′; DIO2 F: 5′‐CAGTGTGGTGCACGTCTCCAATC‐3′, R: 5′‐TGAACCAAAGTTGACCACCAG‐3′; COX7A1 F: 5′‐GCTCTGGTCCGGTCTTTTAGC‐3′, R: 5′‐GTACTGGGAGGTCATTGTCGG‐3′; UCP3 F: 5′‐CTGCACCGCCAGATGAGTTT‐3′, R: 5′‐ATCATGGCTTGAAATCGGACC‐3′; SLN F: 5’‐ATGGAGAGGTCTACTCAGGAGCTGT‐3′, R: 5′‐CTCACGAGGAGCCACATAAGGA‐3′; SERCA1a F: 5′‐GCGCACTCCAAGTCCACAGA‐3′, R: 5′‐GCTCATTGGGGCCGTATTTT‐3′; C/EBPα F: 5′‐CAAGAACAGCAACGAGTACCG‐3′, R: 5′‐GTCACTGGTCAACTCCAGCAC‐3′; C/EBPβ F: 5′‐AGTGGCCAACTTCTACTACG‐3′, R: 5′‐AGAGGTCGGAGAGGAAGTC‐3′; C/EBPδ F: 5′‐CGACTTCAGCGCCTACATTGA‐3′, R: 5′‐CTAGCGACAGACCCCACAC‐3′; PPARγ F: 5’‐TCGCTGATGCACTGCCTATG‐3′, R: 5′‐GAGAGGTCCACAGAGCTGATT‐3′; Ap2 F: 5′‐AAGGTGAAGAGCATCATAACCCT‐3′, R: 5′‐TCACGCCTTTCATAACACATTCC‐3′; DNMT1 F: 5′‐ATCCTGTGAAAGAGAACCCTGT‐3′, R: 5′‐CCGATGCGATAGGGCTCTG‐3′; DNMT3a F: 5′‐CTGTCAGTCTGTCAACCTCAC‐3′, R: 5′‐GTGGAAACCACCGAGAACAC‐3′; DNMT3b F: 5′‐AGCGGGTATGAGGAGTGCAT‐3′, R: 5′‐GGGAGCATCCTTCGTGTCTG‐3′.

Western blotting

Samples were homogenized in RIPA buffer with protease inhibitor cocktail and phosphatase inhibitor cocktail. The gray intensity of blotting bands was evaluated through Image J. The following primary antibody used in this study: p‐PKA substrate, 9624, cell signaling technology; PKA Cα, 4782, cell signaling technology; p‐PKA C, 4781, cell signaling technology; UCP1, ab10983, abcam; PRDM16, PA5‐20872, Thermo Fisher Scientific; PGC1α, sc518025, santa cruz biotechnology; DNMT1, AF5150, beyotime biotechnology; DNMT3a, AF1732, beyotime biotechnology; DNMT3b, AF1384, beyotime biotechnology; HA‐tag, 3724, cell signaling technology; Flag‐tag, 14793, cell signaling technology; β‐actin, 4967, cell signaling technology; HSP90, 4874, cell signaling technology; SCTR, bs‐0089R, Bioss. Secondary antibodies were incubated at a dilution of 1:2,000.

MeDIP

These experiments were conducted on the basis of a previous report (Papageorgiou et al, 2011). Male offspring fed the chow diet at 3 weeks age (the end of lactation period) were chosen for methylated DNA immunoprecipitation sequencing (MeDIP‐seq) analysis. Genomic DNA samples of inguinal white adipose tissue were sonicated to a size 200–800 bp fragments, and about 1 μg of fragmented sample was ligated to Illumina’s genomic adapters with Genomic DNA Sample Kit (FC‐102‐1002, Illumina). Then, 300–900 bp ligated DNA fragments were immunoprecipitated by 5mC antibody (Diagenode). The enriched DNA was purified by AMPure XP beads followed amplified by PCR, and quality control (QC) was performed to assess the quality of MeDIP experiment and sequencing library before sequencing. Sequencing was carried out using the Illumina NovaSeq 6000 following the NovaSeq 6000 S4 Reagent Kit (300 cycles) protocol.

The mapped reads were used for statistically significant Methylation regions and Differentially Methylated regions (DMRs) detection. LncRNA, mRNA, Small ncRNA and Enhancer associated MeDIP enriched regions with statistically significant and DMRs within promoter between two groups were identified using a q‐value threshold of 10⁻⁴ by MACS2 software or by diffReps.

After the successful process of MeDIP, qPCR amplification was carried out for quantification of methylated DNA targeted parts in UCP1, PRDM16, CIDEA and DIO2 CpG islands. Immunoprecipitated DNA (IP) and input control fractions as templates were subjected to SYBR Green qPCR SYBR Premix Ex Taq II Master Mix, and the real‐time PCR reactions were carried out. The following primers were used as follows: UCP1(A) F: 5′‐GTA GAT GTA TGG GAA GGG GGT T‐3′, R: 5′‐CAA AAC CCT AAC CAC ATC ACC T‐3′; UCP1(B) F: 5′‐GTA GAT GTA TGG GAA GGG GGT T‐3′, R: 5′‐CAA AAC CCT AAC CAC ATC ACC T‐3′; PRDM16(A) F: 5′‐AGG ATT TTA GTT TTT GAA AGG G‐3′, R: 5′‐CCA ACT CCA AAC TC TAC CAC‐3′; PRDM16(B) F: 5′‐TTT YGG TTG TTT GTT TTT G‐3′, R: 5′‐AAA CAC RAA TCC AAA CTA CTT T‐3′; PRDM16(C) F: 5′‐GTT GYG GGT TAG TTT TTT TTA‐3′, R: 5′‐AAC TCT CCR AAA CCT TCTC‐3′; PRDM16(C) F: 5′‐GTT GYG GGT TAG TTT TTT TTA‐3′, R: 5′‐AAC TCT CCR AAA CCT TCT C‐3′; PRDM16(D) F: 5′‐GAG GAT GAT GAA TAT ATT GTA AA‐3′, R: 5′‐AAA AAC TTA CTT TTA ACT AAC TTC C‐3′; PRDM16(E) F: 5′‐GTA TTY GGG TTA GGG GTG‐3′, R: 5′‐ACC CRA ACR CCT AAC ACT ATA‐3′; PRDM16(F) F: 5′‐AAA GGG GAG AGT AAT TTT TTT TTT‐3′, R: 5′‐AAA TCT CCA TAT TAC TTC CCC A‐3′; PRDM16(G) F: 5′‐TAT GTG YGA AGG TGT TTA AAT TGA‐3′, R: 5′‐TCR ACC RAA AAC CRA AAC‐3′; CIDEA(A) F: 5′‐ATT TTG TAG TTT GYG AAG AAA GT‐3′, R: 5′‐CTC AAA AAT ACC AAA ACT TCC C‐3′; CIDEA(B) F: 5′‐TTT TGG AGA TTT TAG TAT GTA G‐3′, R: 5′‐CRA ATA ATT CTA ACT CCC TA‐3′; CIDEA(C) F: 5′‐GTT TTG GGA TAG GAG GAT‐3′, R: 5′‐CCA AAA AAC AAA AAC CC‐3′; CIDEA(D) F: 5′‐GGT TTT GGG ATA GGA GGA T‐3′, R: 5′‐ACA CRA AAA ACT ATA ACC CA‐3′; DIO2(A) F: 5′‐AGG TAG AGA AGA TGG GAT TTT TT‐3′, R: 5′‐AAA CTA TTC CAA ACA CAA CRC‐3′; β‐actin F: 5′‐GGT GAA GGT GAC AGC AGT‐3′, R: 5′‐TGG GGT GGC TTT TAG GAT‐3′. The relative methylation fold change was calculated after normalization to input and β‐actin was used as a positive or negative control.

Measurement of mitochondrial respiration

Differentiated primary adipocytes were inoculated into the XF24 V7 cell culture plate according to the test requirements of the program, and then induced the differentiation of adipocytes normally. Assay Media (pH 7.4, 0.22 μm filter sterilization) was used to wash the cells, then changed the fluid and the Oxygen consumption rate (OCR) was determined by Seahorse Bioscience XF24 Extracellular Flux Analyzer (Seahorse Bioscience). And the machine automatically added Oligomycin 2 μM, FCCP 1.5 μm and 1 μM Antimycin A/Rotenone. Basal respiration, uncoupled respiration, and maximal respiration were calculated according to the mitochondrial oxygen consumption value and the instructions.

Statistical analysis

All values were reported as mean ± SD. Two‐tailed Student’s t‐test was applied for comparison. Statistical analysis was performed with GraphPad Prism 8.0 software. P values < 0.05 were considered statistically significant. All experiments were performed at least three times, and representative data were shown.

Author contributions

Lamei Xue: Conceptualization; Data curation; Formal analysis; Writing—original draft. Juan Sun: Data curation. Jinxin Liu: Data curation. Chaoping Hu: Resources. Dandan Wu: Formal analysis; Funding acquisition. Chenzhipeng Nie: Data curation. Kuiliang Zhang: Data curation. Yu Wang: Data curation. Lei Zhao: Resources. Xihua Li: Resources. Yan Lu: Formal analysis; Validation. Li Zhang: Resources; Investigation. Duo Zhang: Data curation; Supervision. Mingcong Fan: Data curation. Haifeng Qian: Data curation. Haowen Jiang: Data curation; Formal analysis. Jiemin Wong: Resources; Supervision. Yuyin Li: Data curation; Formal analysis. Hao Ying: Resources; Funding acquisition; Investigation. Billy KC Chow: Resources; Investigation; Writing—review & editing. Li Wang: Conceptualization; Formal analysis; Writing—original draft. Yan Li: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Writing—original draft; Writing—review & editing.

In addition to the CRediT author contributions listed above, the contributions in detail are:

LX, BC, LW, and YLi designed research; LX, JS, JL, CN, KZ, YW, MF, and HQ performed research; CH, DW, LZ, XL, YLu, LZ, DZ, HJ, JW, YLu, and HY contributed new reagents/analytic tools; LX, BC, LW, and YLi analyzed data; and LX, BC, LW, and YLi wrote the manuscript.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Source Data for Expanded View

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

EMBO reports (2022) 23: e54132.

Data availability

The MeDIP‐seq of genome‐wide methylation state in M‐SCT floxed and M‐SCT KO inguinal white adipose tissue is available in National Center for Biotechnology Information (NCBI) under the accession code (GSE162406): https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE162406.