A potential probiotic- Lachnospiraceae NK4A136 group: Evidence from the restoration of the dietary pattern from a high-fat diet


 Background: High-fat diet (HFD) that contributes to obesity is one of the pivotal risk factors for metabolic syndrome and cancers. The dietary pattern can shape the intestinal bacterial community and influence the physiological parameters. This study aimed to investigate whether the short-term dietary pattern shift from HFD to a balanced chow diet (CD) could correct HFD-induced colonic dysbiosis and reverse adverse health effects and identify the specific bacteria that changed by dietary patterns. Results: C57BL/6 mice fed with an HFD for 10 months, followed by a CD for 3 months, served as the dietary shift model. Stool samples were collected for bacterial analysis. Physiological parameters, such as serum adipokines, blood lipid levels, and hepatic function, were monitored in control and dietary shift groups. HFD-induced weight gain was mitigated by the dietary shift. A highly similar structure at the phylum, genus, and species levels was observed in the beta diversity of colonic bacteria in mice that underwent the dietary shift as compared to those in the control group. Notably, the abundance of Peptococcaceae and Akkermansiaceae in HFD-fed mice reduced after the dietary shift; whereas the diminished amount of probiotic Lachnospiraceae NK4A136 group in HFD-fed mice was restored to the level comparable to those in controls after the dietary shift. Conclusions: Our finding suggests that a dietary switch from a long-term HFD to a short-term balanced diet has the potential to correct colonic dysbiosis and restore physiological homeostasis. The Lachnospiraceae NK4A136 group has the potential to be a probiotic.


Background
In developed countries, the overconsumption of dietary fat is the most common cause of obesity, metabolic syndrome, and several chronic diseases. Studies have revealed that the high-fat diet (HFD)induced intestinal dysbiosis is associated with a decrease in mucus thickness and low-grade in ammation. The consumption of HFD lowers the expression of tight junctions in the intestinal epithelium, which has been associated with subsequent obesity, insulin resistance, and hyperglycemia in both human and animal models [1,2]. HFD could also adversely affect the brain by impairing cognitive functions or promoting depression-like symptoms [3][4][5]. Several animal studies have demonstrated that dietary shifts from HFD to a balanced diet could correct adverse effects resulted from HFD with permanent or transgenerational bene ts [6,7]. and human studies have revealed a higher Firmicutes to Bacteroidetes (F/B) ratio in the gut of obese individuals [17,19,20]. Furthermore, a lowered abundance of Proteobacteria has been found after HFD exposure [20][21][22]24]. However, HFD-induced alteration of gut microbiota could be modi ed by body weight control [19], beta-glucan [25] or pectin -containing diet dietary [26], metformin [20], or probiotics interventions such as Bacteroides uniformis CEST 7711 [23] or B. pseudocatenulatum [21]. Whether the short-term dietary pattern shift from HFD to a balanced diet could correct HFD-induced colonic dysbiosis and reverse adverse physiological alterations still requires further investigation. The lack of ber intakes is also associated with the HFD-induced gut dysbiosis [27,28]. The short-chain fatty acid (SCFA)producing bacteria such as Lachnospira, Akkermannsia, Bi dobacterium, Lactobacillus, Ruminococcus, Roseburia, Clostridium, Faecalibacterium, and Dorea was a signi cant decrease in human receiving low ber diet [29,30].. In addition, intestinal Bacteroides acidifaciens were increases in mice fed a high ber diet [31] and played a role in obesity prevention and improves insulin sensitivity in mice [32]. Notably, HFD has also been linked to enhanced tumorigenicity of colonocytes [33,34]. A possible mechanism is the disruption of the colonic mucosal barrier, which increases the adherence of pathogenic bacteria mediated by HFD-induced dysbiosis [35]. The increase of ber mucus-degrading bacteria, Akkermansia muciniphila, was found in mice fed with a ber-de cient [36] or red meat-mimic heme diet [37], and resulted in the pathogenic bacterial adhesion and invasion to the colonic mucosa [36] or caused epithelial hyperproliferation [37]. However, whether the mucus-degrading bacteria also contribute to HFD and dietary pattern shift is still unknown.
Lachnospiraceae NK4A136 group, which is belonged to the family of Lachnospiraceae, was noticed recently. Lachnospiraceae is characterized by the anaerobic and spore-forming features with the ability to ferment the plant polysaccharides into SCFAs and ethanol [38]. The role of Lachnospiraceae NK4A136 group in diet-associated metabolism is still controversial. In HFD mice, the numbers of Lachnospiraceae NK4A136 group were reduced and negatively correlated to triglyceride levels [39]. The elevation of Lachnospiraceae NK4A136 group was found in mice fed with Gracilaria Lemaneiformis-derived sulfated polysaccharide and positively correlated to the secretion of bile acids [40]. Moreover, obese mice with fucosylated chondroitin sulphate feeding showed the anti-in ammation effects as well as the elevation of Lachnospiraceae NK4A136 group [41]. However, Liu et al. have reported that the abundance of Lachnospiraceae NK4A136 group did not affect in high-fat diet-fed mice [40]; Zhang et al. demonstrated that the increase of Lachnospiraceae NK4A136 group in type 2 diabetes mellitus rats and the bacterial level returns back to control after fed with anthraquinone-glycoside [42]. Whether the Lachnospiraceae NK4A136 group level could be modi ed by dietary fat content remains elusive.
In the current study, we aimed to (1) evaluated the effects of both short-term (3 months) and long-term (10 months) HFDs on biological changes and colonic bacterial alteration, (2) investigated the therapeutic potential of a short-term (3 months) dietary pattern shift from an HFD to a balanced CD in terms of modulation of the intestinal bacterial community, and (3) identi ed speci c bacteria responsible for distinct dietary effects.

Results
HFD contributed to the elevation of body weight and alteration in the bacterial community in mice Mice were fed with either a control balanced CD or an HFD for the indicated period (Fig. 1a). The percentage of calories from fat was 12% in the CD and 62% in the HFD. Compared with controls, the average body weight of mice fed with the HFD for 3 and 10 months increased signi cantly by 1.30-and 1.29-fold, respectively ( Fig. 3a and 3d). Colonic stool samples were collected at the indicated time points, and the diversity of the bacterial community was investigated. The taxonomic pro les of bacterial betadiversity were analyzed using PCA. Using weighted UniFrac to analyze the beta diversity, the distribution of points related to both short-term HFD (SHFD) and long-term HFD (LHFD) samples were compared with those of the control group. HFD-related clusters were close to PC1, which accounted for 73.2% of the total variation, indicating that the HFD had a considerable effect on the colonic bacterial community in mice (Fig. 4a). A similar cluster distribution was also found in the lower-level classi cation of bacterial taxonomy in terms of genus and species, suggesting an alteration of bacterial diversity in mice fed with either an SHFD or an LHFD ( Fig. 4b and 4c).

Effects of HFD and dietary shift on gut microbiota alteration
Obesity-associated gut bacterial changes were found in mice with both SHFD and LHFD consumption The relative abundance of several bacteria taxa was further analyzed to investigate the contribution of diet in modulating the bacterial community of the colon. An increased F/B ratio is known to be associated with obesity. Stool samples derived from both SHFD and LHFD mice revealed that the composition of bacteria at the phylum level was altered in the DS group; major changes were observed in Bacteroidetes, Firmicutes, Verrucomicrobia, and Patescibacteria (Fig. 5a). A signi cant decrease in the abundance of Bacteroidetes was noted (Fig. 5b), whereas no statistically signi cant difference was observed in the number of Firmicutes and the F/B ratio in mice fed with the different diet (SHFD vs. C-SHFD, p = 0.13; LHFD vs. C-LHFD, p = 0.10; Fig. 5c-d). In addition, the phyla Verrucomicrobia and Patescibacteria were relatively abundant in the stool sample from mice fed with LHFD.
At the lower taxonomic level, further analysis was conducted to elucidate which bacterial families were affected after HFD consumption. Compared with controls, Peptococcaceae (Fig. 8a) and Akkermansiaceae (Fig. 8b) were signi cantly increased in the stool samples of mice fed with either an SHFD or an LHFD. However, the elevation of Tannerellaceae only appeared in mice fed with an SHFD, not in mice fed with an LHFD (Fig. 8c). Our nding suggests a correlation between HFD-induced colonic dysbiosis and obesity in mice.
Moreover, we further differentiated the composition of the characteristic microbes in species by linear discriminant analysis effect size (LEfSe). Bacteriodes acidifaciens and peptococcaceae were abundant in SHFD and LHFD groups ( Fig. 6a-d). On the other hand, Lachnospiraceae NK4A136 group was abundant in C-SHFD and C-LHFD groups ( Fig. 6a-d). After fed with CD, Bacteriodes acidifaciens was still abundant in SHIFT group ( Fig. 7 and 8d). The proportion of Lachnospiraceae NK4A136 group was reduced while fed with HFD (p=0.056 at 10th month) and gradually increased to the level as NDC group (Fig. 8e).
The short-term dietary shift corrected HFD-induced weight gains and colonic dysbiosis to regular balance chow diet (CD) The composition of gut bacteria could be modulated by dietary shifts, such as the alteration of macronutrient composition, total calories, or speci c supplements. To evaluate the effect of a short-term dietary shift from an HFD to a regular CD on the recovery of colonic bacteria composition and obesityrelated parameters, mice fed with an LHFD were shifted to a CD for 3 months. During the 3-month period, the weight gain of the DS mice was comparable to that of the NDC mice (Fig. 3c). The circulating levels of adipokine, leptin, and adiponectin were not signi cantly different between groups, suggesting the restoration of adipocyte homeostasis after the dietary shift ( Fig. 3d and 3e).
PCA revealed that stool samples collected from the mice that shifted to a CD for 3 months after HFD feeding for 10 months (SHIFT mice) displayed a smaller difference along PC2, which accounted for 10.9% of total variations at the phylum level (Fig. 4a). A similar pattern of cluster distribution was also found in the genus and species levels. These results suggest that the short-tern dietary shift corrected the HFD-induced alteration in bacterial diversity. The effect of dietary shift on bacterial taxa was further investigated. Compared with controls, the abundance of Firmicutes (Fig. 5b) and Bacteroidetes (Fig. 5c) as well as the F/B ratio (Fig. 5d) were not signi cantly different in mice after the short-term dietary shift, indicating that obesity-associated dysbiosis was corrected by the shift to a normal diet. In addition, an HFD led to elevated Verrucomicrobia (Fig. 5e) and Patescibacteria ( Fig. 5f) abundance, which decreased after the dietary shift; no signi cant difference was noted between groups. Similar results in the family level revealed no signi cant difference in Peptococcaceae (Fig. 8a), Akkermansiaceae (Fig. 8c), and Tannerellaceae (Fig. 8b) between the DS and NDC groups. Moreover, the colon morphology of the SHIFT group was not different to that of the parallel controls of the LHFD mice (C-SHIFT group) ( Fig. 8d and 8e) No difference in blood lipid pro les and liver function parameters between groups after short-term dietary shift HFD-induced abnormal blood lipid pro les and liver dysfunction are known risk factors for obesityassociated diseases. No signi cant difference in serum total cholesterol (TCHO), high-density lipoprotein (HDL), and triglyceride (TG) levels were noted between DS and NDC mice (Fig. 9a). In terms of liver function, serum glutamic oxaloacetic transaminase, glutamic pyruvic transaminase, and total bilirubin were not notable different between groups after the short-term dietary shift (Fig. 9b). The hepatic tissues of NDC and DS mice were processed for H&E staining. Neither group exhibited signs of pathological defects (Fig. 9c).

Discussion
Our results provided evidence that (1) HFD-induced colonic dysbiosis is associated with weight gain, which was signi cantly reversed by the short-term dietary switch from an HFD to a normal balanced CD; (2) both an SHFD or an LHFD could lead to colonic dysbiosis with an increase in the abundances of Peptococcaceae and Akkermansiaceae families; (3) the increased in the abundance of Bacteriodes acidifaciens and decreased of Lachnospiraceae NK4A136 group in the DS. (Fig. 3a-c); (4) No changes in blood lipid pro les and liver function were found in DS mice in comparison with the control group. These results suggest that the correction of HFD-induced dysbiosis through dietary modi cation may be a pivotal factor for the maintenance of physiological homeostasis, and the potential health problem caused by obesity could be partially resolved by dietary changes.
The formula of HFDs was diverse and affected microbiota differentially [43]. In general, obesity was linked to a 50% reduction in the abundance of Bacteroidetes in mice and humans [17,19,20]. Furthermore, this change was reversed after weight loss [19]. A HFD can increase Firmicutes and decrease Bacteroidetes in the human gut [44]. Tang et al have been reported that the abundances of Erysipelotrichaceae, Family_XIII, Ruminococcaceae, ratAN060301C, Clostridiales Coprococcus, Intestinimonas, Parabacteroides, Pseudobutyrivibrio, and Roseburia are increased after HFD feeding. Some microbes such as De uviitaleaceae, De uviitaleaceae, Lachnospiraceae, Peptococcaceae, vadinBB60, Christensenellaceae, Coriobacteriaceae, Peptostreptococcaceae, Prevotellaceae, RF9, Ruminococcaceae, and S24-7 (Muribaculaceae) are decreased after a HFD feeding [45]. Ruminococcus, Akkermannsia, Bacteroidetes, Faecalibacterium, Bi dobaterium, Lactobacillus and Blautia are positively corelated to type 2 diabetes. Furthermore, Rumonococcus and Fusobacterium are negatively correlated to type 2 diabetes [46]. These microbes involves in the gut permeability, metabolism, and in ammation in directly or in directly way [46]. In our study, we also found that the decreased proportions of Bacteroidetes, and Muribaculaceae and the increased proportions of Akkermansia, Parabacteroides, Intestinimonas, and Roseburia (Fig. 6a-d and Additional le 1) indicating our HFD feeding was workable. However, the abundance of Peptococcaceae was increased in SHFD and LHFD which might because of the differences of HFD formula (Fig. 6a-d, 7a) [43]. In the current study, we found that the diet shifting could restore an imbalanced microbiota. However, some genera were still affected even after the dietary shift, such as Acetatifactor, Ruminiclostridium, GCA−900066575, Ruminiclostridium 5, Ruminococcaceae UCG−004, Lachnospiraceae UCG−006, Intestinimonas, and Lachnospiraceae AC2044 group labeled as C2 (Additional le 1). Some genera might also be affected by the dietary shift, such as Roseburia, Anaerotruncus, Oscillibacter, Marvinbryantia, and Lachnospiraceae UCG−001 labeled as C4 (Additional le 1). The functions of these genera remain unclear at present. They might play roles in the homeostasis of nutrition, metabolism, and even the immune system. Further study on their functions is warranted.
The function of Lachnospiraceae NK4A136 group is not clear. In HFD mice, Lachnospiraceae NK4A136 group is reduced and negatively correlates to triglyceride [39]. Instead, Liu et al have been reported that the abundance of Lachnospiraceae NK4A136 group has no effect at high-fat diet-fed mice; however, it elevates after fed with a sulfated polysaccharide from Gracilaria Lemaneiformis [40]. The diverse data might be due to the ingredients of HFD and the duration for HFD feeding. Lachnospiraceae NK4A136 group positively correlates to bile acids indicating it involving in cholesterol homeostasis [40]. Zhang et al have found that the abundance of Lachnospiraceae NK4A136 group increases in type 2 diabetes mellitus rat and returns to the level as the control after fed with anthraquinone-glycoside [42]. Moreover, obese mice with fucosylated chondroitin sulphate feeding showed the anti-in ammation effects as well as the elevation of Lachnospiraceae NK4A136 group [41]. In our study, the abundance of Lachnospiraceae NK4A136 group decreased during HFD feeding and returned to the level as NDC mice after fed with CD ( Fig. 6 and 7e) suggesting that it might be a potential probiotic. Lachnospiraceae NK4A136 group needs further investigation in the following study.
High ber dietary is a factor to in uence gut microbiota and produce SCFAs to regulating metabolism by gut microbiota [47]. High ber diet signi cantly elevated microbes, involving SCFA production, such as Lachnospira, Akkermannsia, Bi dobacterium, Lactobacillus, Ruminococcus, Roseburia, Clostridium, Faecalibacterium, and Dorea [30]. In our study, the proportions of Bacteriodes acidifaciens and Akkermansiaceae family were increased signi cantly in mice fed with either an SHFD or an LHFD (Fig.6ad, 8c, and 8d). Moreover, a 3-month CD shift could not restore the level of Bacteriodes acidifaciens ( Fig. 7  and 8d). Bacteriodes acidifaciens has been reported that it prevents obesity and improves insulin sensitivity [32] and its abundance elevates in the high ber dietary [31,48]. In our study, CD dietary contained 5.3 % crude ber, 15.4 % neutral detergent ber, and 6.3 % acid detergent ber. The ber in HFD was 6.5 % and derived from powdered cellulose. The composition of ber was different between CD and HFD. The abundance of Bacteriodes acidifaciens was increased in the DS group might be affected by the bers or HFD ( Fig. 6 and 8d). After the dietary shift, abundance of Bacteriodes acidifaciens was still signi cantly higher in the DS group ( Fig. 7 and 8d). It indicates that Bacteriodes acidifaciens was not easy to be affected by CD. he mucus-degrading A. muciniphila, which belongs to the Akkermansiaceae family, utilizes colonic mucus as a carbohydrate source and is associated with increased pathogen susceptibility through enhanced bacterial colonization in the epithelium when mice were fed with a berfree diet [36]. A. muciniphila also increased in mice receiving a red meat-mimic heme diet, leading to colonic mucolysis [37]. An HFD as well as red meat-free and ber-free diets are risk factors for colorectal cancer; however, the mechanism is under-investigated. Although reports have indicated that increased A. muciniphila may have bene cial effects on obese mice, [49] our study provides new insights into the role of Akkermansiaceae bacterial in mice receiving an HFD. At the endpoint, the lipids level and adipokines were similar between the NDC and DS groups (Fig. 3d, 3d, 9a-c). Bacteriodes acidifaciens and Akkermansiaceae family might help the homeostasis of lipids level and adipokines [32].
An HFD has several effects on animal anatomy and physiology and results in pathological changes. Although HFD-fed mice exhibited increased blood lipids, their liver function was not affected [50]. In the current study, histology of the liver revealed similar morphologies in the DS and NDC mice (Fig. 9g). The DS and NDC mice also exhibited similar blood lipid levels. Moreover, although TCHO, HDL, and TG levels decreased in the DS mice, this decreased level was not signi cant (Fig. 9a-c). We further examined adipokines, leptin, and adiponectin ( Fig. 3d and 3e). Studies have reported elevated leptin and unchanged adiponectin in the serum of HFD-fed mice [51,52]. The results of the present study indicated that gut length and blood lipid levels were restored in the DS mice.
Despite these ndings, our study has some limitations. Sample sizes for determining the microbiota pattern after LHFD and dietary shifts were small (N = 3 for NDC and N = 5 for DS). The mice were maintained in a relatively simple environment, which enabled their physiology and microbiota to be restored more through the dietary shift to a regular balanced diet. In this study, we only investigated the dominant bacteria. Some minor species that play critical roles may have been overlooked. However, we attempted to provide an overall observation of the condition of the DS group. Hopefully, this information can inspire researchers to investigate therapies to counteract the effects of HFD.

Conclusion
Our results revealed that a short-term dietary shift could modulate colonic dysbiosis caused by an HFD and maintain physiological homeostasis. Therefore, nutritional intervention is an effective tool to prevent the adverse health problem associated with an HFD

Animals and experimental design
Male C57BL/6 mice aged 6-8 weeks were purchased from the National Laboratory Animal Center and were raised at the animal center in Taipei Tzu Chi Hospital. All experimental procedures were approved by the Institutional Animal Use and Care Committee of Taipei Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation (105-IACUC-007). Mice were maintained according to the recommendations of the Guide for the Care and Use of Laboratory Animals (National Institutes of Health). At age 10 weeks, the mice were randomly assigned into 2 groups. Group 1, the normal balanced diet control (NDC; N = 3) group was fed with a CD for 13 months. The formula for the CD (Prolab RMH2500 5PI4, LabDiet, St. Louis, MO, USA) was 12.1% fat, 28.8% protein, and 59.1% carbohydrates to produce calories ( Fig. 1a and 1b). Group 2, the dietary shifting (DS) group (N = 5), was fed with an HFD for 10 months, followed by a balanced CD for 3 months. The diet was acquired by itself. The formula for the HFD (58Y1, TestDiet) was 61.6% fat, 18.1% protein, and 20.3% carbohydrates to produce calories ( Fig. 1a and 1b). CD dietary contained 5.3 % crude ber, 15.4 % neutral detergent ber, and 6.3 % acid detergent ber. The ber in HFD dietary was 6.5 % and derived from powdered cellulose. Each group was further divided into 3 subgroups at 3-3.5 (SHFD), 10 (LHFD), and 13 (SHIFT) months. The parallel controls were named C-SHFD, C-LHFD, and C-SHIFT at 3-3.5, 10, and 13 months, respectively. The bodyweight of each mouse was recorded once weekly, and stool samples were collected at six time points (3-3.5, 7, 10, 11, 12, and 13 months) during the distinct diet exposure (Fig. 2). Mice were euthanized at 13 months by using iso urane (Rhodia Organique Fine, Bristol, UK). All animal experiments were performed in the animal room.

Blood Sampling and Stool collection
Blood samples collected from the hearts at the endpoint of the exam (at 13 months) were centrifuged at 3000 rpm for 15 min at 4 °C for serum collection. Stool samples were collected in the morning; the stool in each cage had been removed the previous evening. The stool samples were stored at 4 °C before sequencing.

Cytokine analysis and blood chemistry analysis
Serum samples were used for cytokine analysis by using a Proteome Pro ler Array Mouse XL Cytokine Array Kit (ARY028; R&D Systems, Minneapolis, MN, USA). In general, 200 μl serum samples were added into the antibody-coating membrane provided by the kit. The signals were detected by the BioSpectrum 810 Imaging System (UVP, Upland, CA, USA) and quanti ed by Image J software [53]. Remaining serum was analyzed with Fuji Dri-Chem 4000i in Taiwan Mouse Clinic for blood chemistry analysis.

Sequencing of 16S rRNA genes and data analysis
Bacterial DNA was extracted from fecal samples by using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany). After DNA extraction, we delivered the samples to Genomics (Taipei, Taiwan) for 16S rRNA gene sequencing on an ABI 3730XL DNA Analyzer (Applied Biosystems, Foster City, CA, USA). Data were analyzed using ClustVis software [54]. Beta diversity was analyzed using weighted UniFrac analysis principal component analysis (PCA) with all operational taxonomic units (OTUs). Diversity at the genus and species level were analyzed using principal component analysis (PCA) with OTUs excluding OTUs below 0.05 % after normalization. Heatmap was generated using the data at the genus level excluding OTUs below 0.05 % after normalization. The parameters used for PCA and heatmap were row centering, unit variance, SVD with imputation. Moreover, the heatmap was represented with collapsed columnsmedian. The signi cant biomarkers were analyze by linear discriminant analysis effect size (LEfSe) [55]. The parameters for LEfSe: alpha value for the factorial Kruskal-Wallis test among classes was 0.05 and the LDA scores were 4 at C-SHFD, SHFD, C-LHFD, and LHFD groups and 2 at C-SHIFT and SHIFT groups.

Histological analysis
Liver and colon tissues were collected and xed in 10% formalin and embedded in para n wax. Five-µmthick sample sections were depara nized and stained with hematoxylin and eosin (H&E). Histological slides were observed using an ECLIPSE TE2000-U microscope (Nikon, New York, NY, USA).

Statistical analysis
All data are expressed as the mean ± standard error and analyzed with GraphPad Prism 5 software. The NDC and DS groups were compared using Student's t test. Signi cance was established at p < 0.05.      HFD feeding for 3 months; C-SHFD: the parallel controls of the SHFD group; LHFD: HFD feeding for 10 months; C-LHFD: the parallel controls of the LHFD group; SHIFT; dietary shift to a CD for 3 months after HFD feeding for 10 months; C-SHIFT: the parallel controls of the SHIFT group. *: p < 0.05 versus the individual control group. N = 3 in the NDC group and N = 5 in the DS group. HFD, high-fat diet; CD, chow diet; SHFD, short-term HFD; LHFD, long-term HFD; NDC, normal balanced diet control; DS, dietary shifting. Figure 6 LEfSe analysis. (a, c) LDA score at the SHFD and LHFD, , respectively. (b, d) Cladogram at the SHFD and LHFD, respectively. The LDA signi cant thresholds were 4.0 at SHFD and LHFD stages. SHFD: HFD feeding for 3 months; C-SHFD: the parallel controls of the SHFD group; LHFD: HFD feeding for 10 months; C-LHFD: the parallel controls of the LHFD group. N = 3 in normal balanced diet control group and N = 5 in the dietary shift group. HFD, high-fat diet; CD, chow diet; SHFD, short-term HFD; LHFD, long-term HFD.