Age-dependent sex difference of non-alcoholic fatty liver disease in TSOD and db/db mice

According to previous clinical studies, the prevalence of non-alcoholic fatty liver disease (NAFLD) is higher in men than women only during the reproductive age. Animal models of NAFLD that reflect sex differences in humans have not been established. In this study, we examined sex differences in the hepatic lesions of Tsumura Suzuki obese diabetes (TSOD) and db/db mice, which are representative genetic models of NAFLD. Male and female TSOD and db/db mice were fed with a normal diet and tap water ad libitum. Six male and female mice of each strain were sacrificed at the ages of 3 and 9 months, respectively, and serum biochemical, pathological, and molecular analyses were performed. Serum aspartate aminotransferase (AST) levels were significantly higher in male than female mice of both strains at the age of 3 months; however, at 9 months, significant sex differences were not observed. Similarly, alanine aminotransferase (ALT) levels were significantly higher in male mice than in female TSOD mice at the age of 3 months; however, at 9 months, significant sex differences were not observed. Image analysis of histological slides revealed that the frequency of the steatotic area was significantly higher in male than female db/db mice at the age of 3 months; however, significant sex differences were not observed at 9 months. The frequency of Sirius red-positive fibrotic area was significantly higher in male than female mice in both strains at the age of 3 months; however, significant sex differences were not observed at 9 months. Serum AST and ALT levels and hepatic steatosis and fibrosis in TSOD and db/db mice showed age-dependent sex differences consistent with those observed in human NAFLD. These mice may be suitable for studying sex differences of the disease.


Introduction
The accumulation of fat in hepatocytes (steatosis) in the absence of significant alcohol intake is called non-alcoholic fatty liver disease (NAFLD). Non-alcoholic fatty liver disease includes a spectrum of liver diseases ranging from simple steatosis (non-alcoholic fatty liver) to non-alcoholic steatohepatitis (NASH), which can progress to liver cirrhosis and hepatocellular carcinomas (HCC) [1][2][3]. It is a hepatic manifestation of metabolic syndrome and constitutes a major health concern worldwide, with an estimated prevalence of 25% in the general population [4] and 43-70% in patients with type 2 diabetes [5].
Previous clinical and epidemiological studies have suggested that there are sex differences in NAFLD in humans. During the reproductive age, the prevalence of NAFLD is higher in men than in women; however, following menopause, the prevalence in women becomes similar to or surpasses that in men [6][7][8][9][10]. Changes in estrogen levels are thought to be an important cause, but the detailed molecular mechanisms underlying age-dependent sex differences in human NAFLD have not yet been elucidated.
Animal models are useful for elucidating pathogenic mechanisms and developing new therapies for NAFLD/NASH. Ideal animal models should recapitulate all features of human diseases. In particular, animal models that accurately reflect sex differences in human NAFLD must be established to elucidate the mechanisms regulating sex differences and develop new therapies based on them. To date, several studies have examined sex differences in animal models of NAFLD/NASH; however, the results have been conflicting. In some studies, hepatic lesions were more severe in males than in females [11][12][13][14]; however, in other studies, the opposite was found [15,16] or there was no obvious sex difference [17]. The major problem with most of these studies is that they did not examine age-dependent changes that are characteristic of sex differences in NAFLD in humans. To the best of our knowledge, there has been only one study in which age-dependent changes were examined, but age-dependent changes in humans were not evident in the study [18]. Therefore, animal models of NAFLD that reflect sex differences in humans have not yet been established.
Tsumura Suzuki obese diabetes (TSOD) mice are a polygenic model of metabolic syndrome and spontaneously develop diabetes mellitus, obesity, glucosuria, hyperglycemia, and hyperinsulinemia. They exhibit a liver histopathology that mimics human NASH and eventually develop liver tumors, including HCC [19,20]. Db/db mice possess a natural mutation in the leptin receptor gene and exhibit obesity, insulin resistance, diabetes, and hepatic steatosis [21,22]. Currently, both TSOD and db/db mice are regarded as representative animal models of NAFLD. As these mice reflect many pathophysiological aspects of human NAFLD, we hypothesized that they might also reflect sex differences in human NAFLD. Therefore, in this study, we examined age-dependent sex differences in hepatic lesions of TSOD and db/db mice. In addition, significant molecules associated with sex differences in the livers of mice with NAFLD were investigated using real-time reverse transcription polymerase chain reaction (RT-PCR).

Animals and experimental protocol
Twelve male and 12 female TSOD and db/db mice were each prepared at the Institute for Animal Reproduction (IAR) (Kasumigaura, Japan). They were kept in an animal experiment laboratory at IAR with free access to a normal diet (MF; Oriental Yeast, Tokyo, Japan) and tap water ad libitum. The conditions were as follows: 25˚C, 45% humidity, and a 12 h light/12 h dark cycle. Six male and six female mice of each strain were sacrificed at the ages of 3 and 9 months, respectively, by collecting blood samples from the inferior vena cava (IVC) under deep anesthesia by isoflurane. Serum was separated by centrifugation (3000 rpm/10 min). The mice were starved for 17 h before sacrifice, and blood glucose levels, using blood from the tail vein, were measured by glucose analysis equipment (StatStrip Xpress 900; Nova Biomedical, Boston, MA, USA) immediately before sacrifice. Body weight was measured simultaneously. After blood was collected from the IVC, the liver of each mouse was extirpated and weighed, and samples for histological analysis, RNA purification, and snap-freezing were collected. This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The experimental protocol was approved by the Ethics Committee of IAR (on Mar. 1, 2018; Permit No. N562) and International University of Health and Welfare (on Dec. 16, 2019;Permit No. 19009NA).

Histopathological analysis
For histological analysis, the central parts of the two large liver lobes of each mouse were cut along the long axis, and 3 mm-thick liver tissue was taken from each lobe. Therefore, two liver tissue samples were collected from each mouse. The tissues were fixed in 10% neutral-buffered formalin and embedded in paraffin. Sections (3-4 μm-thick) were made from each liver tissue, and hematoxylin and eosin (H&E) and Sirius red staining were performed to evaluate hepatic histopathology, including fibrosis. Therefore, the largest cut surfaces of the two large liver lobes were used to generate histological slides for each mouse (two sections per mouse). Both sections of all mice were used for staining and evaluated in all histological examinations (including image analysis). Histopathology was evaluated semi-quantitatively using the scoring validated by Kleiner et al. [23]. Detailed scoring criteria for steatosis, lobular and portal inflammation, and hepatocellular ballooning have been described in our previous studies [24,25]. The method for calculating the NAFLD activity score (NAS) and staging of fibrosis was also described in our previous papers [24,25]. The histological slides were observed by three hepatic pathologists (E.D., T.F., and Y.T.). Often, each pathologist's evaluation showed good concordance. When there were minor differences in the evaluation, the three pathologists discussed and made the final decision.

Whole-slide quantitative image analysis
We performed whole-slide quantitative image analysis to evaluate hepatic steatosis and fibrosis using previously validated methods [26,27]. All H&E-and Sirius red-stained slides were scanned at 40× resolution using a digital slide scanner NanoZoomer S210 (Hamamatsu Photonics, Hamamatsu, Japan). Digital images of the entire liver sections were captured using NDP.scan v3.3 software (Hamamatsu Photonics). The prevalence of macro/micro lipid vesicles in hepatocytes and Sirius red-positive fibrotic areas in these whole-slide images were analyzed using the open-source pathology image analysis software QuPath v0.2.0 (University of Edinburgh, Edinburgh, UK). The representative tissue areas were annotated using manual tools, and the pixel classifier (artificial neural network MLP) was trained on annotations. Once the tissue classifier was trained, the algorithms were applied to all the images, and the steatosis and fibrosis areas (%) were computed as the proportion to the classified tissue.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining
To evaluate apoptotic hepatocytes, we performed TUNEL staining using the DeadEnd colorimetric TUNEL system (Promega, Madison, WI, USA) according to the manufacturer's instructions. Entire histological slides were observed, and the total number of positively stained hepatocytes was counted for each mouse.

Hepatic levels of TG and T-Cho
We measured TG and T-Cho levels in the liver biochemically using approximately 50 mg of frozen tissue from each mouse. The detailed methods were described in our previous paper [25].

Real-time RT-PCR assays
Total RNA (50 ng/sample) was isolated from mouse liver specimens using ISOGEN (Nippon Gene, Tokyo, Japan) or RNAzol RT Reagent (Molecular Research Center, Cincinnati, OH, USA) and then reverse transcribed using Oligo (dT) primers and the Transcriptor First Strand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's recommendations. Quantitative real-time PCR was performed using 5 μL of 20-fold diluted cDNA in a final volume of 20 μL using the FastStart Universal Probe Master Mix (Roche Diagnostics). The sequences of primers and probes are listed in Table 1. Data were analyzed using Light Cycler 96 Software Ver.1.1 (Roche Diagnostics GmbH, Basel, Switzerland). The comparative Ct method was used for the relative quantitation of samples.

Statistical analysis
Continuous variables are presented as the mean ± standard deviation (SD). A one-tailed or two-tailed t-test was performed to evaluate the significance of the differences since the comparison was made between two groups (male and female) and normal distribution could be assumed for these continuous data. A one-tailed or two-tailed test was appropriately selected based on the null hypothesis that male predominance of NAFLD and its associated metabolic disorders is observed only in young mice. For semi-quantitative data in the histological evaluation, data are presented as median (min. to max.). The Mann-Whitney U test was used to determine statistical significance since the comparison was made between two groups and normal distribution could not be assumed for these categorical data. Outliers confirmed by the Smirnov-Grubbs's test or Thompson's test were removed since these are generally the most accepted outlier tests. Statistical analyses were performed using BellCurve for Excel, Version 3.21 (Social Survey Research Information, Tokyo, Japan). Statistical significance was set at a p value < 0.05. Table 2 provides data regarding body weight, liver weight, and the liver-to-body weight ratio. The body and liver weights of male TSOD mice at 3 months were significantly higher than those of age-matched females (p < 0.001 for both); however, no significant difference in the liver-to-body weight ratio was observed. There was no significant sex difference in 3-monthold db/db mice with respect to body and liver weights or liver-to-body weight ratio. The liver weight and liver-to-body weight ratio of 9-month-old TSOD mice were significantly higher in males than in females (p = 0.012 and 0.031, respectively). The body and liver weights of male db/db mice at the age of 9 months were significantly higher than those of age-matched females (p = 0.001 for both); however, no significant difference in the liver-to-body weight ratio was observed. Table 3 shows the serum data corresponding to each mouse strain at 3 and 9 months of age. Serum AST levels were significantly higher in 3-month-old male TSOD and db/db mice than in their female counterparts (p = 0.009 and 0.022, respectively). However, at 9 months, no significant sex difference was observed in either strain. ALT levels were significantly higher in 3-month-old male TSOD mice than their female counterparts (p = 0.017); however, at 9 months, significant sex differences were not observed. T-Cho levels were significantly higher in male TSOD mice than in female TSOD mice at 3 months (p < 0.001), and they were significantly higher in 9-month-old male TSOD and db/db mice than in their female counterparts (p = 0.003 and 0.041, respectively). At 3 and 9 months, TG levels were significantly higher in

PLOS ONE
Age-dependent sex difference of NAFLD in TSOD and db/db mice male TSOD mice than in females (p = 0.031 and 0.036, respectively). Glucose levels in the tail vein blood were significantly higher in 3-month-old male db/db mice than their female counterparts (p = 0.001), but they were significantly higher in 9-month-old female TSOD and db/ db mice than in males (p = 0.014 and 0.025, respectively). Insulin levels were significantly higher in 3-month-old male TSOD mice than in females (p = 0.014) and 9-month-old male TSOD and db/db mice than their female counterparts (p < 0.001 and p = 0.010, respectively). Adiponectin levels were significantly higher in female TSOD mice than in males at the ages of 3 and 9 months (p < 0.001 for both) but were significantly higher in male db/db mice than in females at the age of 3 months (p = 0.044). HOMA-R levels were significantly higher in male TSOD mice than female mice at 3 and 9 months of age (p = 0.039 and p < 0.001, respectively). CRP levels were significantly higher in 3-month-old male TSOD and db/db mice than in their female counterparts (p < 0.001 and p = 0.001, respectively) and significantly higher in 9-month-old male TSOD mice than in their female counterparts (p = 0.022). FGF21 levels were higher in female mice than in male mice in both age groups and strains, and a significant sex difference was observed in 9-month-old TSOD mice (p = 0.001). TNF-α was undetectable in all samples suggesting very low serum levels.

Histopathological findings
With respect to liver histopathology, intralobular and portal inflammation was observed in male and female TSOD mice, although hepatic steatosis and ballooning were not conspicuous. Intralobular inflammation tended to be more severe in male mice than female mice at the age of 3 months (Fig 1). In db/db mice, steatosis was conspicuous in both the 3-and 9-month-old mice. Intralobular and portal inflammation was milder than in TSOD mice, and ballooning was not conspicuous at either age. Steatosis was more severe in male mice than female mice at 3 months (Fig 2). In the semi-quantitative analysis, steatosis grade was significantly higher in male mice than female db/db mice at 3 months (p = 0.027); however, significant sex differences were not observed at 9 months. Intralobular inflammation in 3-month-old TSOD mice tended to be more severe in male mice than females (p = 0.058) ( Table 4).

Whole-slide quantitative image analysis of histological slides
To evaluate steatosis and fibrosis accurately and objectively, we performed whole-slide quantitative image analysis using H&E-and Sirius red-stained slides. The frequency of steatotic area was significantly higher in 3-month-old male db/db mice than in their female counterparts (p < 0.001); however, significant sex differences were not observed at 9 months (Fig 3). The frequency of Sirius red-positive fibrotic area was significantly higher in male TSOD and db/db mice than females at 3 months (p < 0.001 and p = 0.005, respectively); however, no significant sex difference was observed in either strain at 9 months (Fig 4). Fig 5 shows total number of TUNEL-positive apoptotic hepatocytes in each mouse. The number of positively stained hepatocytes was significantly higher in 3-month-old male TSOD mice than in females (p = 0.027), but there was no significant sex difference in 9-month-old TSOD mice. For db/db mice, there was no significant sex difference in both age groups.

Fig 2. Photomicrographs of the liver of db/db mice at 3 and 9 months. A-D)
In db/db mice, steatosis is conspicuous in both 3-and 9-month-old mice. Intralobular and portal inflammation is milder compared to that in TSOD mice, and ballooning is not conspicuous at both ages. Steatosis is more severe in male mice than females at 3 months. (H&E staining, original magnification ×200). https://doi.org/10.1371/journal.pone.0278580.g002

PLOS ONE
Age-dependent sex difference of NAFLD in TSOD and db/db mice

TG and T-Cho levels in the liver
At 3 months of age, TG levels in the livers of TSOD mice showed no significant sex difference; however, at 9 months, they were significantly higher in females than in males (p = 0.025). For db/db mice, hepatic TG levels were significantly higher in male mice than in females at 3 and 9 months of age (p = 0.011 and 0.023, respectively). T-Cho levels in the livers of male db/db mice

Fig 3. Image analysis for quantification of steatosis. A-E)
There is no significant sex difference in the frequency of steatotic area in 3-and 9-month-old TSOD mice (one-tailed t-test for young mice and two-tailed t-test for old mice).

PLOS ONE
Age-dependent sex difference of NAFLD in TSOD and db/db mice were significantly higher than those in females at 3 and 9 months of age (p = 0.037 and 0.021, respectively) (Fig 6).

Gene expression in the liver
The hepatic expression levels of FGF21 mRNA, determined by real-time RT-PCR, were significantly higher in female TSOD mice than in males at 3 months (p = 0.003) and significantly

PLOS ONE
Age-dependent sex difference of NAFLD in TSOD and db/db mice higher in female TSOD and db/db mice than in males at 9 months (p = 0.003 and 0.021, respectively) (Fig 7A). For the expression levels of myeloid differentiation primary response (MYD)88, peroxisome proliferator-activated receptor (PPAR)α, and PPARγ, significant sex differences were not observed in either strain or age group (Fig 7B-7D). The expression of TNF-α and IL-6 was not detectable, although two sets of primers were tried. In the additional real-time RT-PCR analysis, the expression of genes examined was not detectable for many samples probably owing to deterioration of the sample quality after longterm preservation. Actually, the expression of SREBP-1c, CD11b, CD11c, and IL-8 was not detectable for all samples. However, the expression of ACC, CD36, FASN, CPA 1A, F4/80, and IL-1b was detectable at least partially (S1 Fig). At 3 months of age, the expression levels of FASN were significantly higher in male TSOD mice than in females (p = 0.037), but they were significantly higher in female db/db mice than in males (p = 0.005). At 9 months of age, FASN expression showed no significant sex difference in both strains. Expression levels of CPT 1A were significantly higher in 9-month-old female TSOD mice than in their male counterparts (p = 0.009). For the expression levels of ACC, CD36, F4/80, and IL-1b, no significant sex difference was detected.

Discussion
In this study, serum AST and/or ALT levels were significantly higher in 3-month-old male TSOD and db/db mice than in females; however, at the age of 9 months, there was no significant sex difference. Image analysis of liver histopathology revealed that steatosis and/or fibrosis were significantly more severe in 3-month-old male TSOD and db/db mice than their female counterparts; however, at the age of 9 months, there was no significant sex difference. These findings are consistent with sex differences observed in human NAFLD, with male predominance at a young age and equal prevalence or female predominance at an old age (especially after menopause). Serum ALT is considered important for determining the severity of NAFLD no significant sex difference; however, at 9 months, they are significantly higher in females than in males (two-tailed t-test). For db/db mice, hepatic TG levels are significantly higher in male mice than in females at 3 and 9 months of age (one-tailed t-test for young mice and two-tailed t-test for old mice). B) T-Cho levels in the livers of male db/db mice are significantly higher than those in females at 3 and 9 months of age (one-tailed t-test for young mice and two-tailed t-test for old mice).
https://doi.org/10.1371/journal.pone.0278580.g006 since it is a highly specific marker of liver cell damage. The correlation coefficients between serum ALT levels and body weight, between serum ALT and AST levels, and between serum ALT levels and NAS were 0.569, 0.690, and 0.353, respectively, suggesting weak-to-moderate and positive correlations.
Conflicting results have been reported in previous studies on sex differences in NAFLD animal models [11][12][13][14][15][16][17]. In studies reporting male predominance, methionine-and choline-deficient (MCD) diet [11,14], high fat diet [13], or hepatocyte-specific Pten-deficient mice models [12] were used. In contrast, in studies reporting female predominance, high-fructose [15] and cafeteria [16] diet models were used. The limitation of these studies is that they did not examine age-dependent changes. The sacrificial age was 9-40 weeks in studies reporting male predominance and 17-24 weeks in studies reporting female predominance; the animals were not necessarily young in studies reporting male predominance or old in studies reporting female predominance. To the best of our knowledge, only one study has examined age-dependent sex differences in a NAFLD animal model [18]. In the study, hepatic lesions of p62/Sqstm1 and Nrf2 double-knockout mice were examined at 8, 30, and 50 wk of age. In the study, inflammatory activity, extent of fibrosis, and serum AST levels were consistently higher in male mice than female mice. Sex differences in inflammatory activity were not statistically significant at 8 wk, but significant male predominance was observed at 30 and 50 wk. Thus, the model did not A) The expression levels of fibroblast growth factor (FGF)21 mRNA are significantly higher in female TSOD mice than males at 3 months and significantly higher in female TSOD and db/db mice than males at 9 months (two-tailed t-test). B-D) For the expression levels of myeloid differentiation primary response (MYD)88, peroxisome proliferator-activated receptor (PPAR)α, and PPARγ, significant sex differences are not observed in both strains and age groups (two-tailed ttest).
https://doi.org/10.1371/journal.pone.0278580.g007 correctly reflect sex differences in human NAFLD. In the present study, serum AST and ALT levels and the extent of hepatic steatosis and fibrosis in TSOD and db/db mice basically reflected sex differences in human NAFLD; therefore, these mice models are thought to be more appropriate than p62/Sqstm1 and Nrf2 double-knockout mice for sex difference studies.
Here, we set the sacrificial age of young mice at 3 months and that of old mice at 9 months. We set the sacrificing age of young mice at 3 months because the sacrificing age is 3 months or older in most animal experiments on NAFLD. In contrast, we set the sacrificing age of old mice at 9 months because after this age: 1) liver tumors develop in TSOD mice, which may influence morphological and chemical findings in the liver [19]; and 2) db/db mice begin to die [29]. There are no detailed data on the reproductive ages of TSOD and db/db mice. However, in the IAR, male TSOD mice over 28 weeks of age, female TSOD mice over 27 weeks of age, male db/db mice over 26 weeks of age, and female db/db mice over 32 weeks of age are not used for breeding because pregnancy rates and litter sizes evidently decrease. Compared to young mice, 9-month-old TSOD and db/db mice have an evidently lower reproductive capacity. In the future, examinations with longer observational periods using animal models with more detailed data on the reproductive age should be performed. To the best of our knowledge, there has been no report on serum estrogen levels in mice of approximately 30 weeks of age. However, Kubota et al. [30] reported that serum estradiol levels of 50-week-old female 129/Sv mice were lower than those of their 10-week-old counterparts. A large amount of serum is necessary to measure levels of estrogen. In the present study, we were not able to measure serum levels of estrogen since only small amounts of serum were available from mice and we measured the levels of many other serum markers.
Furthermore, we found glucose levels were significantly higher in 3-month-old male db/db mice than their female counterparts but significantly higher in 9-month-old female TSOD and db/db mice than their male counterparts. As NAFLD is closely associated with diabetes mellitus, this finding may be associated with age-dependent sex differences in NAFLD in these mice. In humans, it has been reported that the prevalence of diabetes is higher in men aged < 60 years and in women at older ages [31]. Therefore, the blood glucose levels in TSOD and db/db mice reflect sex-dependent findings in humans. In this study, serum T-Cho, TG, insulin, and adiponectin levels showed significant sex differences in 3-and 9-month-old TSOD and db/db mice. However, they did not show age-dependent changes that could explain the severity of NAFLD. Androgens are known to decrease plasma adiponectin [32], and thus, it is reasonable that adiponectin levels were significantly higher in female TSOD mice than in males. However, unexpectedly, adiponectin levels were significantly higher in male db/db mice than in females at the age of 3 months. It is difficult to determine the cause of this; however, it has been reported that male, but not female, db/db mice show resistance to the effects of adiponectin [33], and this sex-related difference in the adiponectin sensitivity of db/db mice might be associated with this finding. CRP is a marker of inflammation, and its levels were remarkably higher in male TSOD and db/db mice than in females at 3 months of age. However, at 9 months of age, the sex difference became less remarkable in TSOD mice, and there was no significant sex difference in db/db mice. This finding is compatible with the age-dependent sex difference in NAFLD.
In the histopathological analysis, steatosis decreased with age in male db/db mice. This finding is consistent with the report by Trayhurn et al. showing that hepatic lipogenesis in db/db mice increases progressively from weaning until 8 weeks of age and then decreases [34]. However, in the present study, hepatic steatosis in 3-month-old female db/db mice was much milder than that in their male counterparts, and there was no remarkable difference in the extent of steatosis between 3-month-old and 9-month-old female db/db mice. However, the extent of fibrosis in db/db mice was slightly increased with age for both sexes.
Hepatocyte apoptosis plays a role in the activation of NAFLD/NASH, and apoptotic hepatocytes stimulate immune cells and hepatic stellate cells to initiate the progression of fibrosis in the liver through the production of inflammasomes and cytokines [35]. Hepatocyte apoptosis in TSOD mice was significantly more prevalent in males only at 3 months of age, and this might comprise one of the mechanisms of age-dependent sex difference in NAFLD in this strain. However, positively stained hepatocytes in db/db mice showed no significant sex difference in both age groups.
Regarding the biochemical quantification of hepatic TG and T-Cho, both 3-month-old and 9-month-old male db/db mice showed significantly higher levels than their female counterparts. However, in the semi-quantitative analysis and quantitative image analysis of histopathology, only 3-month-old male db/db mice showed significantly more severe steatosis than females. Thus, there was a discrepancy between biochemical and histopathological data. One of the causes of this might be sampling bias associated with the biochemical analysis since only a small amount of liver tissue (approximately 50 mg for each mouse) was analyzed. In contrast, the largest cut surfaces of the two large liver lobes were analyzed in the histopathological examinations.
Moreover, the hepatic expression levels of FGF21 mRNA, determined by real-time RT-PCR, were significantly higher in 3-month-old female TSOD mice than their male counterparts and significantly higher in 9-month-old female TSOD and db/db mice than males. Serum levels of FGF21 basically reflected the expression levels of FGF21 mRNA in the liver. FGF21 is known to increase energy expenditure, fat utilization, and lipid excretion, causing weight loss, increased insulin sensitivity, decreased blood glucose and lipid levels, and the amelioration of hepatic steatosis [36][37][38][39]. The higher expression of this molecule in female mice than in males may explain male predominance of NAFLD at a young age. However, it was difficult to attribute the age-dependent sex difference in NAFLD to this molecule, since its expression level was significantly higher in old females than males. Lee et al. [14] reported that FGF21 expression was increased in the liver tissue by the MCD diet, and the degree of upregulation was significantly higher in the livers of female mice. On the other hand, Gasparin et al. [16] reported that FGF21 mRNA expression was exclusively induced in male mice when mice were fed a cafeteria diet. Therefore, it seems that sex differences in the expression levels of FGF21 depend on the NAFLD model used. Thus, the role of FGF21 in sex differences in NAFLD should be examined further in the future. MYD88 is an adaptor protein that plays a pivotal role in innate and adaptive immunity. PPARs are nuclear receptors that regulate glucose and lipid metabolism, as well as inflammation [40]. In this study, expression levels of these genes did not show significant sex differences. Expression of TNF-α and IL-6, representative pro-inflammatory cytokines, was not detectable. In the additional real-time RT-PCR assays, expression of the genes examined was not detectable for many samples even though we tried two different methods. Frozen liver samples were preserved for more than 4 years before the additional real-time RT-PCR experiments, and sample deterioration was probably the cause of this. The samples from 3-month-old mice were preserved longer than those from 9-month-old mice, and the detection of gene expression was especially difficult when using the samples from 3-month-old mice. Although significant sex differences were observed in the expression levels of FASN and CPT 1A, the interpretation of this result requires caution considering the decreased quality of samples.
Our study has several limitations. Hepatic steatosis in TSOD mice was very mild, and it may be questionable whether TSOD mice fed the MF diet are appropriate as a model of NAFLD. Although remarkable steatosis was observed in the livers of db/db mice, necroinflammatory changes were mild, and ballooning degeneration was not conspicuous. This finding conforms to previous observations that db/db mice do not spontaneously develop NASH [20,22]. In the future, animal models that better reflect both sex differences and the histopathology of NASH should be developed. In addition, the detailed molecular mechanisms underlying age-dependent sex differences in NAFLD were not elucidated in this study. In the future, extensive molecular studies should be performed using TSOD or db/db mice, or further improved animal models. In particular, it is known that clear sex differences exist in immune regulation and response [41], and it would be interesting to examine the associated molecules. Since the present study was performed using mice, only a small amount of serum was available and only a limited number of serum markers could be measured. Especially, serum levels of IL-6, IL-1, and lipopolysaccharide should be examined in the future.

Conclusions
In this study, serum AST and ALT levels and hepatic steatosis and fibrosis in TSOD and db/db mice showed age-dependent sex differences, consistent with those in human NAFLD. These mice may be suitable for studying sex differences of the disease. Elucidation of the detailed mechanisms and development of animal models with more severe histopathological changes should be performed in the future.
Supporting information S1 Fig. Expression of genes that are associated with lipid metabolism and inflammation in the liver, as determined by additional real-time reverse transcription polymerase chain reaction. A) At 3 months, expression levels of fatty acid synthase (FASN) are significantly higher in male TSOD mice than in females but are significantly higher in female db/db mice than in males (one-or two-tailed t-test). B) Expression levels of carnitine palmitoyltransferase (CPT) 1A are significantly higher in 9-month-old female TSOD mice than in their male counterparts (two-tailed t-test). C-F) For expression levels of acetyl-CoA carboxylase (ACC), CD36, F4/80, and interleukin (IL)-1b, no significant sex difference is detected (one-or two-tailed ttest). (TIF) S1