We are in the midst of a worldwide obesity epidemic. The global prevalence of obesity has doubled since 1980 (1) and is now an international public health issue (2). Obesity is frequently associated with impaired glucose tolerance, dyslipidemia, and hypertension, and the combination of these metabolic disorders is associated with atherosclerosis-related diseases. Therefore, the concept of metabolic syndrome has been proposed (3). Increasing evidence has demonstrated that environmental disruptions in the early susceptible period affect health and promote a predisposition to non-communicable diseases (NCDs) in later life, based upon which the concept of the Developmental Origins of Health and Disease (DOHaD) was established (4, 5). Eriksson et al. (2001) identified a high body mass index (BMI) at 7 years old as a risk factor for adult obesity with low-birth weights in 3,659 individuals born at Helsinki University Hospital between 1924 and 1933 (6). Ong et al. conducted systematic reviews and demonstrated that rapid infantile weight gain was consistently associated with an increased risk of obesity later in life (7). Ekelund et al. (2007) showed that rapid infantile weight gain was associated with clustered metabolic disruptions in 128 affected individuals at the age of 17 years (8). Martin et al. (2017) performed a systemic review and found that longer-term health outcomes found catch-up growth were associated with a higher body mass, BMI, or cholesterol in later life; however, catch-up growth following low birth weight may have short-term benefits for individuals with low birth weight (9). Singhal et al. (2017) reported that growth acceleration in healthy infants born at term (either normal or small for gestation) may be associated with an increased long-term risk of obesity and NCDs; however, the benefits of rapid infant weight gain for later neurodevelopment supports its encouragement in infants born preterm (10). These findings led us to the concept that rapid infantile growth (RG) in term newborns, particularly with low-birth weights, is causatively associated with a predisposition to obesity and metabolic syndrome in later life. However, to the best of our knowledge, the underlying molecular mechanisms by which RG in small newborns programs a predisposition to the deterioration of fat deposition have not yet been elucidated. Indeed, risk of NCDs resulted from RG in small newborns applies well to ‘Mismatch hypothesis’ (11) (12), one of central hypotheses of DOHaD theory. In the present study, we hypothesized the presence of specific pattens in gene expression profiles of the adipose tissue of pups with RG and undernutrition in utero (UN).

We established a mouse model of UN using maternal energy restriction, which develops the phenotypes of various NCDs (13–15), and subsequently demonstrated that diet-induced fat deposition in adipose tissue and the liver was significantly greater in UN pups than in normal nutrition (NN) pups (13, 16–18). Oral treatments with the hydrophilic secondary bile acid tauroursodeoxycholic acid (TUDCA), an endoplasmic reticulum (ER) stress alleviator, markedly ameliorated developmentally-induced hepatic steatosis in UN pups, but not in NN pups (17, 18). Therefore, modifications to this animal model of UN are promising for investigating the effects of RG on developmentally-programed fat deposition and its amelioration by TUDCA (TU).

The overall aim of the present study was to examine the characteristics of gene expression profiles associated with the changes in the amount of adult adipose tissue deposition, i.e. deteriorations induced by RG after UN and improvements by TU. The specific objectives of the present study were 1) to develop a mouse model of RG soon after birth by reducing the number of pups per litter to 4 pups (UN-RG) compared to 8 pups per litter for NN, 2) to investigate the effects of TU on UN-RG pups and NN pups fed a high-fat diet (HFD), 3) to perform a microarray analysis followed by a gene enrichment analysis of epididymal fat tissues (NN-Veh vs UN-RG-Veh and UN-RG-Veh vs UN-RG-TU) in order to identify common gene ontologies (GO) terms, 4) to conduct heatmap and principal component analyses of the common GO identified, 5) to identify genes of interest (GOI) from these GO terms and subject them to quantitative RT-PCR, and 6) to immunohistochemically assess the infiltration of macrophages in epididymal adipose tissue in consideration of the characteristics of the GO terms and GOI identified.

Increasing evidence supports RG being causatively associated with obesity and/or metabolic syndrome in later life, particularly in neonates born small (6–10); however, the benefits of RG for later neurodevelopment supports its encouragement in infants born preterm (10). In the present study, we newly established a mouse model of RG during the lactation period following UN (UN-RG) ( Figures 1 , 2 ), by modifying our previous model of UN (13–18). UN-RG pups developed diet-induced obesity ( Figure 3 ) as well as metabolic disorders ( Table 2 ), mimicking the phenotype of diet-induced obesity and associated metabolic syndrome causatively connected with rapid infantile growth (6–10).

As shown Figures 2C, D , mean body weight from 8.5 d to 16.5 d was significantly higher in UN-RG pups than in NN pups, but similar at weaning 18.5 d. Our previous study showed that mean body weight of simple UN pups from fetal period (18.5 dpc) to early infantile period (3.5 d) was significantly lower than in NN pups (13). Therefore, it is noted that the present UN-RG pups experienced drastic changes from relatively small fetal body weight to large neonatal body weight (5.5 d-16.5 d), which we believe is an appropriate animal model of the small human babes who experienced rapid growth after birth and suffered from NCDs in later life, as reported by epidemiological studies (6, 7), corresponding to ‘Mismatch’ hypothesis in DOHaD theory (11, 12).

In the present study, we focused to design the animal model for mimicking the epidemiological outcomes of infantile rapid growth of small neonate (6, 7) and did not simultaneously carry out only RG pups experiment due mainly to animal facility limitation. However, we developed UN-RG animal model, considering that risk of NCDs resulted from RG in small human newborns applies well to ‘Mismatch hypothesis’ (11, 12), one of central hypotheses of DOHaD theory. ‘Mismatch hypothesis’ is based on the concept that small fetuses developmentally adapted to the low energy supply and caused ‘Mismatch’ to the abundant energy supply after birth, resulted in predisposition to obesity and associated metabolic disorders (11, 12). It is undeniable that the adult phenotype of UN-RG pups might be similar to that of RG pups. It is also unclear if an experience of the UN affected the degree of RG, although it was speculated that main contributor of RG was increased breast feeding in UN-RG pups. Nevertheless, we would like to note that simple RG pups model is a clearly different research target from UN-RG pups, because simple RG pups mimic babies with normal birth weight who experienced rapid infantile growth, which is not directly associated with ‘Mismatch hypothesis’ (11, 12).

We performed a comprehensive analysis of gene expression profiles in epididymal adipose tissue as a visceral fat pad. We initially investigated the effects of UN-RG on gene expression profiles in epididymal adipose tissue using a microarray analysis (NN-Veh vs UN-RG-Veh), displayed the results in volcano plots ( Figure 4A ), and then conducted a gene enrichment analysis, which identified the top 20 GO terms in ascending order of P-values ( Figure 4B ). Since TU significantly attenuated diet-induced obesity ( Figures 3B, C ) and epididymal fat deposition ( Figure 3D ), we examined its effects on gene expression profiles in epididymal adipose tissue using a microarray analysis (UN-RD-Veh vs UN-RG-TU). The results obtained were displayed in volcano plots ( Figure 5A ). A subsequent gene enrichment analysis identified the top 20 GO terms in ascending order of P-values ( Figure 5B ).

In 4 inflammatory GO terms, heatmap analysis of the mRNA expression of representative genes showed the genes of increased expression by UN showed a decrease by TU; while the genes of decreased expression by UN showed an increase by TU ( Figures 6A–D ), which suggests their coordinated contribution to the inflammatory changes observed in fat pads in response to UN and TU. However, we were unable to obtain a clear pattern for heatmaps by clustering related genes and, thus, simply listed gene names on the vertical axis ( Figures 6A–D ). We then performed a principal component analysis, which revealed similar changes in gene expression patterns between NN and TU in comparison with UN ( Figure 6E ).

We subsequently acknowledged the following 4 GO terms: 1) Inflammatory response, 2) Response to interferon-γ, 3) Regulation of cytokine production, and 4) Cellular response to interferon-β, all of which were commonly detected in the two different GO terms groups ( Figures 4B , 5B ), for the purpose of identifying potential critical pathways contributing to the plasticity observed in the regulation of fat deposition, which responded not only to UN, but also to TU ( Figures 3B–D ). All 4 GO terms were directly related to inflammatory reactions, suggesting that inflammation is a key phenomenon in the plasticity of the regulation of fat deposition in this animal model. Top GO term of natural killer cell-mediated cytotoxicity (UN-RG-Veh vs NN-Veh) also supports a possible involvement of inflammation ( Figure 4B ) (22, 23). It is noted that GO terms of carnitine metabolism and regulation of apoptotic cell clearance, direct pathways of lipid metabolism, were demonstrated by comparison between NN-Veh and UN-RG-Veh ( Figure 4B ), but not between UN-RG-Veh and UN-RG-TU ( Figure 5B ). We speculate that UN-RG might induce lipid metabolism as well as inflammation and that TU might affect mainly inflammatory regulation. More investigation is necessary to prove this speculation. TUDCA is an alleviator of ER stress, and it is noted that changes of Kdelr 3, ER protein retention receptor 3, gene expression was observed by comparison between NN-Veh and UN-RG-Veh (original microarray data is available by referring GSE18830 of the GEO repository).

Recent studies reported that obesity and its metabolic disruption are closely associated with low-level systemic inflammation, based upon which the concept of ‘metaflammation’ was proposed (24–27). Hotamisligil et al. (2017) proposed adipose tissue, the liver, pancreas, and brain as representative organs of ‘metaflammation’ (27). Lauterbach et al. (2017) suggested the principal involvement of tissue-resident macrophages in ‘metaflammation’ (28). Therefore, we performed an immunohistochemical analysis of macrophages in epididymal adipose tissue ( Figure 8A ) and found that UN-RG induced significant increases in the numbers of macrophages, while TU reversed this effect ( Figure 8B , Supplementary Figure 2 ). These results support the critical contribution of ‘metaflammation’ to the adjustment of fat deposition in this animal model. There was a tendency of increase in the gene expression ratio of CD11c/CD163 in epididymal adipose tissue of UN-RG pups at 16.5 wks, but statistically not significant ( Supplementary Figure 1 ). It is a future aim of the study to investigate the characteristics of these macrophages.

Previous studies proposed the potentially long-lasting embryonal memory of immune cells, including tissue-resident macrophages. Mass et al. (2021) suggested that fetal-derived immune cells are the prime transmitters of the long-term consequences of prenatal adversities, causing inflammatory, degenerative, and metabolic disorders (29). The potential commitment of erythro-myeloid progenitors produced in the extra-embryonic yolk sac to the ability to establish long-lasting immunological memory has been proposed (29–31); however, the exact mechanisms by which the memory of tissue resident macrophages is transferred to older generation macrophages remain unclear. Troger et al. (2013) demonstrated that the ability of the innate immune system of the growth-restricted fetus to mount an immune response was weakened after birth (32). Gotz et al. (2007) reported that prenatal stress may alter the number and proliferation of lymphocytes as well as the cytotoxicity of natural killer cells in the blood of male adult rat offspring (33). The gene expression profiles obtained in the present study also support the causative relationship of an environmental imbalance during the early critical period with the predisposition of adults to metaflammation; however, we did not detect immune memory in the macrophages of adipose tissue in this animal model.

In the search for the main molecular mechanisms in this animal model, we identified 7 GOI: Saa3, Ubd, S100a8, Hpx, Casp1, Agt, and Ptgs2, which were observed at least twice in the 4 GO terms that we selected in the present study, and also meet the conditions of a ≤−2 or ≥2 linear fold change and P-value <0.05 ( Table 3 ). Quantitative RT-PCR reconfirmed the changes in their gene expression profiles in response to UN-RG and TU, except for Casp1 ( Figure 7 ). We currently have no clear explanation for the discrepancy regarding Casp1. Although limited information is currently available on their contribution to metaflammation in the DOHaD scheme, these genes play the following critical roles in the regulation of general inflammation. Saa3 (34), Agt (35), and Ptgs2 (36) are all involved in the inflammation of obesity. Ubd has been implicated in inflammation in nonalcoholic fatty liver disease (37). Hpx functions as an antioxidant and regulates inflammation (38). Casp1 is an effector of NLRC4 inflammasomes (39) and converts pro-IL-1β and pro-IL-18 into their biologically active forms (40). Regarding the activation of macrophages, Saa3 has been shown to promote the infiltration of macrophages in adipose tissue (41), while S100a8 plays a central role in obesity-promoting macrophage-based inflammation (42) and induces the secretion of cytokines from mononuclear cells (43). We would like to speculate possible a series of reactions by these genes in macrophage-based metaflammation, 1) induction of macrophage migration by Saa3, 2) induction of the secretion of cytokines from macrophage by S100a8, 3) activation of cytokines secreted from macrophage by Casp1, and 4) possible connection to insulin sensitivity by Ubd nd/or Hpx. More intensive studies are needed to clarify the direct involvement of these genes in the developmentally-induced augmentation of macrophage infiltration and metabolic disruption in the adipose tissue of this animal model. Epigenetic changes in these seven GOI are now under investigation by our research group. It is our future aim to investigate the involvement of these GOI in the developmentally-induced regulation of fat deposition and macrophage infiltration in this animal model. Our previous study showed that gene expression of MCP-1 and TNF-α correlated with subcutaneous adipose tissue weight at 17 wks of simple UN pups, but not in NN pups (16), indirectly suggested a possible association between fat deposit and expression of inflammatory cytokines. We have no clear explanation why these two genes were not identified as GOI in UN-RG pups, because we did not carry out microarray analysis at that time and MCP-1 and TNF-α were selected as representative inflammatory cytokines in UN pups, and also because the present UN-RG pups was designed to represent ‘Mismatch hypothesis’ in DOHaD theory (11, 12), quite different from simple UN pups at that time.

There are some limitations in the present study. We investigated 4 groups of pups, i.e. the NN-Veh, NN-TU, UN-RG-Veh, and UN-RG-TU groups, but not a NN-RG-Veh or NN-RG-TU group because of the restrictions of our animal facility accommodation. It is noted that Lizarraga-Mollinedo reported that different growth speed in the pups with undernourishment in utero showed conflicting levels of gene expression patterns in adipose tissue compared to normally nourished pups by using rat animal model (44). We did not investigate a condition without HFD due to the same animal facility limitation. We did not investigate time course of gene expression profiles. Furthermore, we were unable to obtain clear patterns for heatmaps by clustering related genes and, thus, simply listed gene names on the vertical axis of heatmaps ( Figures 6A-D ). We identified 7 GOI from 4 GO terms based on the results of microarray and gene enrichment analyses; however, we did not demonstrate their direct contribution to the programing of metaflammation in this animal model. We did not measure the plasma levels of inflammatory cytokines due to lack of the serum. Further study is needed to prove the involvement of 7 GOI in the pathophysiology of metaflammation. As for quantitative RT-PCR, 18S ribosomal mRNA expression was used as an internal control; however, we did not assess the effect of RG or TU on its expression. The main contributor of RG was supposed to be breastfeeding in this animal model; however, we could not assess the changes in breastfeeding of dams due to technical limitation. In the preset study, we focused on the assessment of epididymal adipose tissue, as a visceral fat, and did not investigate subcutaneous and retroperitoneal adipose tissues. It is a future aim of the study to investigate the changes in gene expression in subcutaneous and retroperitoneal adipose tissues.

In summary, we newly established a mouse animal model of RG during the lactation period following UN ( Figures 1 , 2 , Table 2 ), the offspring of which developed diet-induced obesity and metabolic disorders, which were significantly improved by TU ( Figure 3 ), although we did not assess UN pups without GR. The microarray analysis and subsequent gene enrichment analysis ( Figure 4 , Figure 5 ) identified 4 GO terms of inflammatory pathways, indicating the critical contribution of chronic inflammation to developmentally-programed fat deposition and its amelioration by TU. The results of the heatmap and principal component analyses of the representative genes of the 4 GO terms ( Figure 6 ), changes in the gene expression profiles of 7 GOI ( Table 3 , Figure 7 ) selected from 4 GO terms, and immunohistochemistry on macrophages ( Figure 8 ) collectively support the proposal of the ‘Developmental Origins of Metaflammation’. Future epigenetic and immunological investigations, with a focus on 7 GOI, are needed to further establish this concept.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

The animal study was reviewed and approved by the Animal Research Committee, Hamamatsu University School of Medicine (H20-014). Written informed consent was obtained from the owners for the participation of their animals in this study.

MS and YK-K performed and analyzed all experiments. MS and HI wrote the manuscript. MU, KM, and HI designed the study. NF-I, TO, and MasM performed histological examinations. MS, TO, KK, TI, and MadM supported data collection and animal experiments, sample preparation, and data analysis. NT, TU, MN, and YK-K supported data collection, discussions, and statistical analyses. MS, TO, KK, and TI supported data analyses and preparation of the manuscript. KM supported the gene enrichment analysis. All authors contributed to the article and approved the submitted version.

This work was supported by JSPS KAKENHI Grant Numbers, JP20H03823, JP20K09666, and JP20K16886, and AMED under Grant Number JP20gm1310009.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.