(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



LXR-dependent enhancer activation regulates the temporal organization of the liver’s response to refeeding leading to lipogenic gene overshoot [1]

['Noga Korenfeld', 'Institute Of Biochemistry', 'Food Science', 'Nutrition', 'The Robert H. Smith Faculty Of Agriculture', 'Food', 'Environment', 'The Hebrew University Of Jerusalem', 'Rehovot', 'Tali Gorbonos']

Date: 2024-09

Transitions between the fed and fasted state are common in mammals. The liver orchestrates adaptive responses to feeding/fasting by transcriptionally regulating metabolic pathways of energy usage and storage. Transcriptional and enhancer dynamics following cessation of fasting (refeeding) have not been explored. We examined the transcriptional and chromatin events occurring upon refeeding in mice, including kinetic behavior and molecular drivers. We found that the refeeding response is temporally organized with the early response focused on ramping up protein translation while the later stages of refeeding drive a bifurcated lipid synthesis program. While both the cholesterol biosynthesis and lipogenesis pathways were inhibited during fasting, most cholesterol biosynthesis genes returned to their basal levels upon refeeding while most lipogenesis genes markedly overshoot above pre-fasting levels. Gene knockout, enhancer dynamics, and ChIP-seq analyses revealed that lipogenic gene overshoot is dictated by LXRα. These findings from unbiased analyses unravel the mechanism behind the long-known phenomenon of refeeding fat overshoot.

Funding: This work was supported by the Canada-Israel Health Research Initiative, jointly funded by the Canadian Institutes of Health Research (cihr-irsc.gc.ca), the Israel Science Foundation ( www.isf.org.il ), the International Development Research Centre, Canada ( https://idrc-crdi.ca/en ) and the Azrieli Foundation ( https://azrielifoundation.org/ ) to IG (#1469/19, 3533/19) and CLC (#109155-001) as well as the European Research Council ( https://erc.europa.eu/homepage ) to IG (ERC-StG #947907). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The transcriptional regulation and underlying enhancer dynamics of the fasted state has previously been studied, leading to meaningful insights into liver biology and regulation of metabolic pathways [ 38 ]. In stark contrast, the response to refeeding has received considerably less attention and the refeeding enhancer landscape, its dynamics and kinetics after cessation of fasting have not been explored. Moreover, the ad libitum fed state and refed state are considered synonymous and most studies (apart from a few exceptions [ 39 , 40 ]) use either ad libitum fed or refed states to represent a “fed” state. Herein, we aimed to inspect the transcriptional and chromatin events occurring upon refeeding, their kinetic behavior and their molecular drivers. We found distinct, temporally organized transcriptional programs occurring upon refeeding with an early wave of transcription followed by a later wave. These programs were driven by enhancer activation that also showed kinetic behavior. Using unbiased genome-wide approaches and gene knockout models, we show that a lipogenic gene program is part of the second wave of transcription and is directly regulated by LXRα.

TFs regulate gene expression by binding to DNA regulatory elements (enhancers and promoter-proximal regions) and initiate a series of events resulting in enhancer activation followed by increased rate of gene transcription by RNA polymerase II (i.e., gene induction). These events include TF-dependent recruitment of co-activators, histone modifying enzymes, and chromatin remodelers [ 32 ]. It has become clear from many studies that enhancer activity in fully differentiated cells is altered by various hormonal and metabolic cues. This is beautifully exemplified in hepatocytes whose enhancers are dynamically activated by a myriad of cues in a TF-dependent manner [ 17 , 33 – 37 ].

The TF termed carbohydrate response element-binding protein (ChREBP) also supports lipogenesis, partly due to activation of glycolysis which supplies lipogenic precursors [ 23 ]. Similar to SREBP1c, ChREBP mRNA levels are also induced by LXR [ 24 , 25 ]. In addition to those mentioned above, other TFs were reported to regulate certain aspects of lipogenesis and cholesterol biosynthesis: thyroid hormone receptor (ThR) [ 26 ], upstream stimulatory factor 1 (USF-1) [ 27 , 28 ], X-box binding protein 1 (XBP-1) [ 29 ], and liver receptor homolog 1 (LRH-1) [ 30 ]. In addition to TFs activating lipogenesis, the BCL6 TF was shown to repress lipid catabolism in the fed state [ 31 ].

Another critical TF family regulating lipogenesis is the liver X receptor (LXR) family, composed of 2 members: LXRα (encoded from Nr1h3) and LXRβ (encoded from Nr1h2). LXRα is considered the principal LXR in hepatocytes and the major member responsible for lipogenesis (although in the absence of LXRα, LXRβ partially compensates for it [ 14 ]). LXRs induce many lipogenic genes and their deletion severely impairs lipid homeostasis [ 15 – 18 ]. Part of the positive effect of LXRs on lipogenesis is indirect, through induction of Srebf1 and the resulting increase in SREBP1c levels [ 19 – 22 ].

Both hepatic lipogenesis and cholesterol synthesis are heavily regulated transcriptionally with dedicated transcriptional programs activating them in the fed state [ 5 , 9 – 12 ]. These programs include induction of genes encoding enzymes, transporters and carriers participating in these 2 anabolic pathways (for brevity, we term these genes and their encoded products “lipogenic genes” or “cholesterol biosynthesis genes”). Several transcription factors (TFs) were reported to govern hepatic induction of these genes under fed conditions. Two central TFs regulating lipid synthesis are members of the sterol regulatory element-binding protein (SREBP) family: SREBP1c (encoded from Srebf1) and SREBP2 (encoded from Srebf2). The activity of these TFs is regulated by proteolytic cleavage occurring in the Golgi. Following cleavage, SREBPs enter the nucleus, bind their DNA recognition motif, and induce gene transcription. The activation of SREBPs is controlled by cholesterol whereby cholesterol inhibits SREBP2 activation. SREBP1c is also inhibited by cholesterol but it is commonly accepted that the major regulation on SREBP1c activity is feeding-dependent whereby insulin levels rising in the fed state robustly activate SREBP1c [ 13 ]. Several studies using gene knockout techniques showed that SREBP1c mostly induces lipogenesis genes while SREBP2 induces cholesterol biosynthesis genes [ 11 , 12 ].

The liver plays a central role in maintaining homeostatic metabolic pathways important in both the fed and fasted states. During fasting, glycogen is broken down to supply glucose, and gluconeogenesis is enhanced to produce glucose from non-carbohydrate precursors. Additionally, fatty acid oxidation is augmented to produce ATP and supply precursors for the production of ketone bodies which are then used as an energy source for the brain and other tissues [ 2 ]. In the fed state, fatty acid oxidation is dampened and instead fatty acid synthesis is active. Synthesized fatty acids (together with glycerol) are used to produce triglycerides, the principal energy storage molecule in mammals. The pathways of fatty acid and triglyceride synthesis in the liver are termed “de novo lipogenesis” or simply “lipogenesis” [ 3 , 4 ]. Another major biosynthetic pathway active during the fed state is cholesterol biosynthesis [ 5 ]. Cholesterol serves as a constituent of membranes, lipoproteins and is the initial substrate for the synthesis of bile acids, steroids, and certain vitamins. Both lipogenesis and cholesterol biosynthesis require acetyl CoA as a precursor. Several metabolic pathways converge to acetyl CoA and two were shown to be important in supporting lipogenesis and cholesterol biosynthesis: the glycolytic production of pyruvate (which is eventually converted to citrate and then to acetyl CoA) and the production of acetyl CoA from acetate (acetyl CoA production from acetate is considered to be minor in physiological conditions) [ 6 , 7 ]. In addition to precursors, lipogenesis and cholesterol biosynthesis require NADPH which is supplied by malic enzyme activity as well as by enzymes in the pentose phosphate pathway [ 4 , 8 ].

When food is readily accessible and its consumption is possible at will (ad libitum), most mammals will eat several meals during their wake time and fast for a few hours during their inactive phase. However, mammals are often faced with longer periods of fasting for reasons such as inaccessibility of food, illness, or voluntary fasting [ 1 ]. In both the fed and fasted states, bodily homeostasis is maintained due to metabolic adjustments aimed at preserving energy supply to cells and storage of excess energy. When food is consumed again after a period of fasting (i.e., refeeding), a metabolic switch occurs and tissues transition from frugal energy usage and the internal production of fuel to using energy available from food constituents and storage of excess energy in specialized molecules.

Results

Refeeding gene regulation distinctly diverges from the basal ad libitum state We aimed to gain a better understanding of the hepatic transcriptional response to refeeding and its kinetic progression from early to later time points. Thus, we designed an experiment in which control mice had ad libitum access to food while all other groups were fasted for a period of 24 h. We chose to fast mice for 24 h because this length of fasting was repeatedly shown to lead to maximal fasting-dependent gene regulation and to optimally manifest the metabolic attributes of fasting (gluconeogenesis, glycogen breakdown, fatty acid oxidation, ketogenesis, etc.) [33,41,42]. One group of mice was euthanized at the end of the fasting period while the other groups had food reintroduced into the cage. Then, mice were euthanized at 3 different time points (3 h, 10 h, and 24 h) following refeeding. The groups were termed Adlib, Fasted, Refed_3h, Refed_10h, and Refed_24h, respectively (Fig 1A). The hepatic transcriptome of all groups was profiled via RNA-seq. To broadly assess the difference between the transcriptomes of different groups, we performed a t-distributed stochastic neighbor embedding (tSNE) analysis. This showed a gradual departure from the basal gene expression pattern (Adlib) first to the Fasted state and then to the different Refed stages. Surprisingly, the Refed states were noticeably distinct from the basal Adlib state and the liver gene expression pattern did not return to the basal state even in animals who had unlimited access to food for 24 h following fasting (Fig 1B). To further show the difference between the Adlib and Refed states, we performed pairwise differential gene expression analysis and defined genes whose expression is higher in the Adlib condition compared to the Fasted condition. We then compared them to the genes whose expression is higher in the Refed_24h condition compared to the Fasted condition. Adlib and Refed states are considered similar conditions because they are both “fed” states in which energy from food is readily available. If indeed the Adlib and Refed_24h states are very similar to each other, we would expect the transcriptional program of these 2 conditions in comparison to Fasted to largely overlap. We also note that this is the only refeeding time point that matches in circadian time that of the fasted samples. However, we found only partial overlap between the 2 gene groups where only 35% of genes induced in the Refed_24h/Fasted comparison were also induced in the Adlib/Fasted comparison (Fig 1C and S1 Table; in all analyses throughout the study, a gene was considered differentially regulated if it passed 2 cutoffs: fold change (FC) ≥1.5 and adj. p-value ≤0.05). We then sought to directly measure differential gene expression between the Adlib and Refed conditions; 2 biological conditions commonly perceived as interchangeable. We found that 2,209 genes were induced in at least 1 refeeding time point as compared to the Adlib state (S1 Table). While the early refed time point led to a higher number of induced genes (Fig 1D), the potency of gene induction, as measured by FC, was higher in later refeeding time points (Fig 1E). Collectively, these data show that refed mice have a fundamentally different milieu of expressed genes at all measured refed time points compared to ad libitum fed mice. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Gene expression upon refeeding distinctly diverges from the basal ad libitum state. (A) Experimental design—the different groups were collected at different food availability states; animals from the Adlib group did not experience fasting prior to collection. The Fasted group was collected at the end of a 24 h fasting bout. The Refed groups were collected 3, 10, and 24 h following reintroduction of food. (B) The global transcriptomic similarity between replicates and experimental groups was measured by a t-SNE analysis. Biological replicates cluster closely to each other, showing high transcriptomic similarity and attesting to the technical quality of the experiment. The Refed groups notably diverge from the Adlib group as well as from each other, pointing to marked differences in gene expression in the Refed groups compared to Adlib and to kinetic progression of the transcriptomic response to refeeding. (C) Evaluation of Refed-induced genes vs. Adlib-induced genes (both compared to the Fasted group) shows a distinct and nonoverlapping set of induced genes in the 2 groups. A gene was considered differentially regulated compared to Fasted if it passed 2 cutoffs determined by DESeq2: FC ≥1.5 and adj. p-value ≤0.05. These cutoffs are consistent throughout the study. (D) A direct pairwise comparison between the Adlib and each Refed time point uncovers thousands of genes differentially regulated upon refeeding, showing that the transcriptomes of the Refed and Adlib conditions are far from identical. (E) The FC values of refeeding-induced genes in each Refed time point were plotted (compared to Adlib; i.e., the genes marked in red in panel D). The highest fold induction is observed in later Refed time points (10 and 24 h). FC, fold change; t-SNE, t-distributed stochastic neighbor embedding. https://doi.org/10.1371/journal.pbio.3002735.g001

Refeeding gene regulation is temporally organized While pairwise comparisons are useful in obtaining discrete gene lists with distinct FC values, they fail to encompass dataset-wide gene expression patterns and trends. In order to expose these complex patterns, we created 2 gene lists for downstream analyses which represent 2 opposite biological responses: fasting-induced genes and fasting-repressed genes. A list of fasting-induced genes was created by extracting all genes showing lower expression in either of the fed states (Adlib, Refed_3h, Refed_10h, or Refed_24h) as compared to the Fasted state (n = 2,312; S2 Table). To generate the reciprocal type of regulation—fasting-repressed genes, we collected all genes showing higher expression in either of the fed states as compared to the Fasted state (n = 2,825; S2 Table). Then, each gene list was put through a clustering analysis to reveal major gene expression patterns and trends. Clustering of the fasting-induced genes revealed 3 clusters consisting of 3 gene expression patterns (Fig 2A). The first cluster showed fasting-induced genes whose levels remain high after 3 h of refeeding and only return to basal levels after 10 or 24 h of refeeding (Pattern A). Pathway enrichment analysis showed many genes from this cluster participate in lipid catabolism, fatty acid oxidation, and ketogenesis. This aligns with the well-known induction of these pathways during fasting [43]. In the second cluster, fasting-induced genes quickly reversed to their Adlib levels with repression evident as early as 3 h following refeeding (Pattern B). Enriched pathways in this cluster were related to signal transduction and phosphorylation. Interestingly, in the third cluster we found a group of genes whose levels do not overtly change between the Adlib and Fasted states but is markedly reduced following 3 h of refeeding and going back to basal levels at later refeeding time points (Pattern C). Genes from this cluster belonged to various signaling pathways. A representative gene from each cluster is depicted (Fig 2B and S1 Data) and a schematic illustration of each pattern is shown in Fig 2C. The full gene lists and enriched pathways of each pattern are detailed in S2 Table. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Induction of several distinct gene expression programs following refeeding. (A) k-means clustering of fasting-induced genes (n = 2,312; k = 3) shows 3 major gene expression patterns. Blue: minimum expression value of the gene. Red: maximum expression value of each gene (minimum and maximum values of each gene are set independently to other genes). (B) The normalized read values are shown for a representative gene from each cluster. Conditions in which gene expression was different from Adlib in a statistically significant manner are marked with asterisks. Numerical values for this panel are detailed in S1 Data. (C) Schematic illustration of patterns recovered in clustering analysis (Fig 2A): Pattern A: Fasting-induced genes whose expression wanes slowly upon refeeding. Pattern B: Fasting-induced genes going back to their basal levels quickly upon refeeding. Pattern C: Genes with similar expression between Adlib and Fasted but quickly repressed upon refeeding and then go back to basal levels at later refeeding time points. (D) k-means clustering of fasting-repressed genes (n = 2,825; k = 3). Blue: minimum expression value of the gene. Red: maximum expression value of each gene (minimum and maximum values of each gene are set independently to other genes). (E) Careful inspection of the second cluster revealed it represents 2 gene expression patterns. The normalized read values are shown for a representative gene from each pattern. Conditions in which gene expression was different from Adlib in a statistically significant manner are marked with asterisks. Numerical values for this panel are detailed in S1 Data. (F) Schematic illustration of patterns recovered in clustering analysis (Fig 2D): Pattern D: Fasting-repressed genes recovering to their Adlib basal levels upon refeeding. Pattern E: Genes repressed by fasting and upon refeeding overshoot above their basal levels. Pattern F: Genes with similar expression between Adlib and Fasted but potently induced in later refeeding time points. Pattern G: Genes quickly and transiently induced in Refed_3h. (G) Definitions of the genes belonging to each pattern based on the detailed cutoffs. These cutoffs are schematically represented by arrows in panel F: Gray arrow indicates no statistically significant change between 2 conditions, red arrow indicates gene induction and blue arrow indicates gene repression.—All individual data points are presented ± SD; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by ordinary one-way ANOVA followed by Dunnett’s post hoc analysis. RPKM, reads per kilobase per million reads. https://doi.org/10.1371/journal.pbio.3002735.g002 Clustering of fasting-repressed genes revealed 3 major clusters (Fig 2D). Upon careful inspection, the 3 clusters represented 4 gene expression patterns (see gene examples and schematic illustrations in Fig 2E and 2F and S1 Data). The first cluster showed gene repression upon fasting which did not resolve 3 h after refeeding. Only 24 h after refeeding did gene expression recover and return to its basal Adlib levels (Pattern D). This intuitive pattern was expected and is the one most chiefly considered in the literature [44]. The second cluster revealed 2 intriguing patterns where following fasting-mediated gene repression, genes were strongly induced in later refeeding time points to a level higher than the basal state (Pattern E). Within the same cluster, it appears that the expression of some genes does not differ between the Adlib and Fasted conditions, but does show overt induction upon refeeding (Pattern F). The third cluster was also unexpected with a clear pattern of immediate gene induction following 3 h of refeeding which wanes almost completely after 10 h of refeeding (Pattern G). Taken together, these findings reveal a dynamic transcriptional response to refeeding with clear kinetics whereby gene induction following refeeding is partitioned to an early response and a late response. Also, refeeding often leads to strong gene induction higher than the basal ad libitum state. Because the Refed_3h and Refed_10h conditions were collected at different time points during the day, it is conceivable that some genes induced in these conditions are induced not due to refeeding but rather due the effect of the circadian clock which significantly alters liver gene expression throughout the day [45]. The Adlib, Fasted, and Refed_24h groups were collected at zeitgeber time 1 (ZT1) while the Refed_3h and Refed_10h groups were collected at ZT4 and ZT11, respectively. To test the contribution of rhythmic gene expression on these groups, we compared the genes induced in the Refed_3h or the Refed_10h condition (as compared to Adlib) to genes induced in ad libitum-fed mice whose transcriptome was profiled in ZT points similar to Refed_3h and Refed_10h [46]. There was very little overlap between rhythmic clock-controlled genes in the relevant time points and genes induced by refeeding in Refed_3h and Refed_10h (S1 Fig), suggesting that most of the genes revealed in our clustering analysis are genes responding to refeeding per se and not to the circadian clock. To further explore the transcriptional events occurring upon refeeding, we considered the 4 refeeding patterns from Fig 2F (Patterns D, E, F, and G) and defined distinct inclusion criteria for each pattern. The criteria are based on fold change and statistical significance cutoffs designed for each pattern (Fig 2G). In Patterns F and G, the prominent change in gene expression is observed when comparing the Refed states to Adlib while the difference between the Adlib and Fasted states is mild or even nonexistent. Pattern G, termed “early refeeding induction,” was defined by cutoffs to represent significant induction in the Refed_3h as compared to Adlib but no induction compared to the other conditions. This resulted in 1,047 early-refed-induced genes and pathway enrichment analysis revealed that many of these genes participate in protein synthesis, ribosomal biogenesis, rRNA processing, etc. (S2 Table). This induction of genes related to protein synthesis presumably serves to support the previously reported increase in liver cell mass and hepatocyte proliferation following refeeding [47–50]. For Pattern F (late refeeding-induced genes), we used a similar approach in which genes were included if they were induced in Refed_10h or Refed_24h as compared to Adlib but were not induced in Fasted or Refed_3h compared to Adlib. This resulted in 747 refeeding-induced genes enriched with similar pathways as early refeeding-induced genes (S2 Table). The 2 other gene expression patterns (Patterns D and E) were characterized by repression during fasting, compared to the Adlib state. In Pattern D, fasting-repressed genes recover and go back to their basal, pre-fasting expression levels. In contrast, genes in Pattern E showed an overshoot pattern where expression is reduced during fasting and upon refeeding, it markedly exceeds basal levels. Again, to strictly differentiate between “recovered” genes and “overshoot” genes, we used distinct cutoffs. Because the Refed_10h and the Refed_24h gene expression trends were similar, we focused only on the Refed_24h time point. Genes repressed by fasting compared to both Adlib and Refed_24h that also show no higher expression in Refed_24h compared to Adlib were defined as recovered (n = 419; S2 Table). Genes both repressed by fasting compared to Adlib and induced by refeeding (again, compared to Adlib) were determined as overshoot genes (n = 74; S2 Table).

Lipogenic genes and cholesterol biosynthesis genes are differentially regulated during refeeding Given their different expression patterns, we hypothesized the 2 gene groups—“recovered” and “overshoot”—may have different functions and therefore we performed pathway enrichment analysis for each group. A prominently enriched pathway in the recovered group was cholesterol biosynthesis, in line with the known repression of cholesterol biosynthesis genes during fasting [5]. In the overshoot group, cholesterol biosynthesis was also enriched but in addition, the pathways for lipogenesis, fatty acid synthesis, triglyceride synthesis, glycolysis, and the pentose phosphate pathway were enriched as well (S2 Table). These results suggest that while most cholesterol synthesis genes only go back to their basal level after refeeding, genes related to other lipid metabolic pathways overshoot following refeeding. To examine this possibility, we manually collected and curated from the literature all genes related to cholesterol biosynthesis as well as lipogenesis and divided them into groups: The CHOL group consists of all genes previously shown to participate in the cholesterol biosynthesis pathway. The LIPO group consists of all genes shown to participate in lipogenesis (fatty acid synthesis, fatty acid elongation, and triglyceride synthesis). Many genes are intimately related to both cholesterol biosynthesis and lipogenesis because they aid and support both pathways in various ways. For example, genes contributing to the formation of acetyl-CoA (the precursor for both biosynthetic pathways) and genes replenishing NADPH, a cofactor needed for both cholesterol biosynthesis and lipogenesis pathways. We collected all these genes in a group termed AID. After obtaining the 3 lists, we excluded all genes not expressed in liver as well as genes not repressed by fasting as compared to any of the fed conditions (S2 Table). After these stringent filtering steps, we were left with 61 genes belonging to either the LIPO, CHOL, or AID groups (Fig 3A and S3 Table). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Upon refeeding most cholesterol biosynthesis genes recover to pre-fasting levels while lipogenesis genes overshoot. (A) To faithfully define genes whose encoded proteins participate in lipid synthesis or aiding pathways, we first collected all relevant genes from the literature. We then applied 2 filtering steps in which all genes not expressed in liver (RPKM below 1) and were not repressed by fasting were excluded. The 3 groups were abbreviated as follows: LIPO–lipogenesis genes; CHOL–cholesterol biosynthesis genes; AID–genes from pathways needed to aid both lipogenesis and cholesterol biosynthesis (e.g., to produce acetyl-CoA or replenish NADPH levels). For further details and full gene lists, see S3 Table. (B) The extent to which LIPO, CHOL, and AID genes are repressed by fasting (Adlib/Fasted) or induced by refeeding (Refed_24h/Adlib) was measured. While average fasting repression FC was similar between LIPO and CHOL genes, refeeding induction FC was higher in LIPO genes. Each point represents the FC of a single gene. Numerical values for this panel are detailed in S1 Data. (C) The expression level of all CHOL, LIPO, and AID genes is presented, showing robust overshoot induction of many LIPO and AID genes following refeeding with most CHOL genes showing a recovered pattern. All genes from panel A are shown and were sorted based on refeeding-induction FC. Genes induced in Refed_24h compared to Adlib in a statistically significant manner are marked with a black asterisk while those that did not pass the adj. p-value cutoff are marked by “ns.” Genes significantly repressed in Refed_24h compared to Adlib are marked with a gray asterisk (adj. p-values were determined by DESeq2). (D) Liver triglycerides and total cholesterol were quantified, showing increased liver triglycerides following refeeding. All individual data points are presented ± SD; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001 by two-tailed unpaired t test. One significant outlier was removed based on the ROUT method, Q = 1%. Numerical values for this panel are detailed in S1 Data. FC, fold change; RPKM, reads per kilobase per million reads. https://doi.org/10.1371/journal.pbio.3002735.g003 To test for different regulatory modes, we first assessed whether CHOL and LIPO genes are repressed to different strengths during fasting. Comparison of FC values (Fasted compared to Adlib) showed no difference in repression potency between LIPO and CHOL genes. Then, we aimed to assess if refeeding induction values differ between groups. Here, and throughout the rest of the text, the term “refeeding induction” refers to an increase in the Refed_24h condition as compared to Adlib. We compared the FC values of Refed_24h to Adlib and found a significant difference between groups: most CHOL genes showed little-to-no refeeding induction while many LIPO and AID genes were robustly induced by refeeding as compared to the basal, Adlib condition (Fig 3B and S1 Data). To represent this visually, we plotted the expression values of all 61 genes across all conditions. This shows that most (although not all) CHOL genes show a recovered pattern while LIPO genes mostly overshoot (Fig 3C). AID genes largely followed the LIPO pattern of expression with high refeeding induction above Adlib levels (Fig 3C). The common transcriptional pattern between LIPO and AID genes suggests a mutual transcriptional regulator and implies that the induction of AID genes serves to support lipogenesis. Indeed, many AID genes are commonly considered to facilitate lipogenesis much more prominently than cholesterol biosynthesis (e.g., Acly, Me1). To test whether these gene expression changes alter hepatic lipid levels, we quantified liver triglyceride and cholesterol levels. In accordance with LIPO gene overshoot, triglyceride levels were increased following refeeding as compared to Adlib. In contrast, there was no change in liver cholesterol levels, aligning with the lack of cholesterol gene overshoot (Fig 3D and S1 Data). Plasma cholesterol and triglyceride levels were unchanged between Adlib and Refed_24h (S2A Fig and S1 Data). Collectively, these findings show that while both cholesterol biosynthesis and lipogenesis pathways are similarly repressed by fasting, they are regulated in a starkly different manner following refeeding with many LIPO and AID genes overshooting above pre-fasting levels. Similarly to LIPO and AID gene expression, liver triglycerides levels also overshoot above basal levels following refeeding. The overshoot phenomenon is evident 24 h following refeeding. To test if it persists even longer, we modified the fasting-refeeding experiment and examined longer refeeding periods—in addition to the Refed_24h time point, we collected livers 72 h and 1 week after the reintroduction of food. LIPO and AID genes showed overshoot expression upon 24 h of refeeding, as expected. Interestingly, for some genes the overshoot phenomenon lingered also in the Refed_72h group where gene expression remained higher than Adlib. By 1 week after refeeding, all genes returned to their basal expression (S2B Fig and S1 Data). Accordingly, hepatic triglyceride content tended to be higher 24 and 72 h after refeeding (although it did not reach statistical significance; S2C Fig and S1 Data). Therefore, we found that gene and fat overshoot upon refeeding lasts for 3 days after reintroduction of food. Our experiments included 24 h of fasting prior to refeeding. This period of fasting extends beyond the mice’s inactive phase through which mice fast voluntarily for several hours. Shorter periods of fasting of around 8 h during the inactive phase are considered mild and do not lead to maximal glycogen depletion, ketonemia, weight loss, and other parameters of the fasting response [33,41]. We aimed to test if periods of fasting closer to voluntarily overnight fasting lead to overshoot upon refeeding. Therefore, we performed a fasting-refeeding experiment where mice fasted for only 8 h followed by 24 or 72 h of refeeding. Under these conditions, genes did not overshoot (S2D Fig and S1 Data). Thus, short-term fasting periods during the inactive phase are not followed by gene overshoot.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002735

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.