Abstract

   Caveolin-1 (CAV1) is a crucial regulator of lipid accumulation and
   metabolism. Previous studies have shown that global Cav1 deficiency
   affects lipid metabolism and hepatic steatosis. We aimed to analyze the
   consequences of hepatocyte-specific Cav1 knockout under healthy
   conditions and upon non-alcoholic fatty liver disease (NAFLD)
   development. Male and female hepatocyte-specific Cav1 knockout
   (HepCAV1ko) mice were fed a methionine/choline (MCD) deficient diet for
   4 weeks. MCD feeding caused severe hepatic steatosis and slight
   fibrosis. In addition, liver function parameters, i.e., ALT, AST, and
   GLDH, were elevated, while cholesterol and glucose level were reduced
   upon MCD feeding. These differences were not affected by
   hepatocyte-specific Cav1 knockout. Microarray analysis showed strong
   differences in gene expression profiles of livers from HepCAV1ko mice
   compared those of global Cav1 knockout animals. Pathway enrichment
   analysis identified that metabolic alterations were sex-dimorphically
   regulated by hepatocyte-specific CAV1. In male HepCAV1ko mice,
   metabolic pathways were suppressed in NAFLD, whereas in female knockout
   mice induced. Moreover, gender-specific transcription profiles were
   modulated in healthy animals. In conclusion, our results demonstrate
   that hepatocyte-specific Cav1 knockout significantly altered gene
   profiles, did not affect liver steatosis and fibrosis in NAFLD and that
   gender had severe impact on gene expression patterns in healthy and
   diseased hepatocyte-specific Cav1 knockout mice.

   Subject terms: Mechanisms of disease, Mechanisms of disease, Mechanisms
   of disease, Transcriptomics, Transcriptomics

Introduction

   Non-alcoholic fatty liver disease (NAFLD) is a common metabolic liver
   disorder, characterized by excessive deposition of intrahepatic
   fat^[42]1. The incidence of NAFLD has increased in recent years and
   thus, NAFLD is considered a new societal challenge, requiring novel
   strategies for disease prevention and treatment^[43]2. NAFLD
   encompasses two histological subtypes: non-alcoholic fatty liver
   characterized by simple liver steatosis and non-alcoholic
   steatohepatitis (NASH), which describes hepatic steatosis and
   inflammation with fibrosis. NASH patients have a high risk of
   developing cirrhosis^[44]3,[45]4. For a deeper understanding of the
   disease and for developing therapy, sophisticated NAFLD animal models
   have been developed. The commonly used models are diverse high caloric
   diets (e.g., Western diet) and the methionine/choline deficient (MCD)
   diet. The MCD diet leads to impairment of the secretion of hepatic very
   low density lipoproteins (VLDL), due to the absence of
   phosphatidylcholine precursors^[46]5.

   Caveolin-1 (CAV1) is the main structural protein of
   caveolae—omega-shaped invaginations of plasma membrane defining a
   specific endocytic route. CAV1 regulates cholesterol transport and
   membrane distribution, as well as diverse signaling pathways^[47]6.
   Moreover, CAV1 is a regulator of metabolism, effecting lipid and
   glucose metabolism, insulin resistance-, and diet-induced
   obesity^[48]7–[49]10. These phenomena are supported by findings in
   global Cav1 null mice, that present with strongly elevated triglyceride
   and free fatty acid levels, and which are resistant to high fat
   diet-induced obesity^[50]11. It can also be inferred that CAV1 is an
   essential factor in lipid regulation during liver regeneration after
   partial hepatectomy, given that lipid droplet accumulation was reduced
   and cell cycle was impaired in hepatocytes of global Cav1 null
   mice^[51]12. In another study, Cav1 null mice displayed decreased
   adiponectin abundance and reduced metabolic flexibility under fasting
   conditions. This influenced hepatic steatosis, thus arguing for a
   non-hepatic CAV1 control of liver metabolic alterations^[52]13. This
   prompted us to perform an in-depth study to clarify hepatocyte-specific
   functions in healthy liver and NAFLD and focused on whether
   hepatocyte-specific CAV1 deficiency alters metabolic processes in
   healthy and diseased livers. We additionally aimed to determine gender
   influence on CAV1 functions, given that males have a higher
   susceptibility for developing NAFLD in mice and human^[53]14,[54]15.
   Interestingly, CAV1 and sex hormones are interacting to interfere with
   metabolic processes by regulating hormone signaling and altering
   distinct hormones or hormone receptors^[55]16.

   We report that hepatocyte-specific CAV1 does not affect liver steatosis
   and fibrosis in the MCD induced NAFLD model, but impacts severely on
   gene expression profiles, especially in diseased livers of males and
   females.

Results

Hepatocyte-specific Cav1 knockout in mice

   We first confirmed hepatocyte-specific CAV1 deletion by genotyping for
   Cre recombinase and Cav1 loxp sites (Fig. [56]1a). In our studies,
   homozygous HepCAV1^−/− mice, which were positive for Cre-recombinase
   (heterozygous), were considered as hepatocyte-specific CAV1 knockouts
   (HepCAV1ko), whereas those with HepCAV1^−/− and negative
   Cre-recombinase were used as hepatocyte-specific CAV1 wild-type
   (HepCAV1wt, i.e., as reference control group).

Fig. 1. Knockout of hepatocyte-specific Cav1 in mice.

   [57]Fig. 1
   [58]Open in a new tab

   a PCR based genotyping of mice including Cre-recombinase test and
   Cav-flx test. HepCAV1ko mice showed one Alb-cre positive band.
   Wild-type mice showed one band at 491 bp, and HepCAV1^+/− mice showed
   two bands of sizes at 444 bp and 491 bp, and HepCAV1^−/− showed one
   band at 444 bp. b mRNA expression level of Cav1 in isolated primary
   hepatocytes from males showed a significant difference between
   HepCAV1ko and HepCAV1wt (N = 7), p = 0.0402 using t-test. c Time course
   of mRNA expression of Cav1 in primary hepatocytes isolated from male
   and female wild-type and HepCAV1^−/− mice. In the wild-type mice, Cav1
   is strongly induced over time. d Protein expression of CAV1 in isolated
   primary hepatocytes (males) after 48 h and 72 h cell culture and
   densitometric analysis of expression intensity. e mRNA levels of Cav1
   in liver tissue of males (N = 6) and females (N = 6) showed decreased
   expression in HepCAV1ko mice, with p values almost reaching
   significance (data of males were analyzed by Mann-Whitney, and data of
   females by t-test). f Cav1 expression alteration upon MCD diet (4
   weeks; N = 6) in male and female mice. Total liver Cav1 expression of
   male was higher than female, while no significant change was detected.
   Cav1 expression showed a reduction tendency in male wild-type mice,
   while an increasing tendency was found in females upon MCD diet. Data
   were analyzed by Two-way ANOVA. WT: HepCAV1wt mice; KO: HepCAV1ko mice;
   Con: control diet; MCD: MCD diet.

   To test CAV1 expression in hepatocytes, mRNA level of Cav1 in freshly
   isolated hepatocytes (N = 7) was measured. As expected, Cav1 was
   significantly reduced (p = 0.0402) in HepCAV1ko compared to HepCAV1wt
   mice (Fig. [59]1b). However, due to the low basal expression of CAV1 in
   primary hepatocytes, CAV1 was not detected by Western blot. Hence, an
   alternative system to confirm knockout efficiency at protein level was
   used. Specifically, CAV1 expression of primary hepatocytes is induced
   over time when cultured in vitro^[60]17. Hence, we cultured primary
   cells up to 4 days and harvested RNA and protein lysates at 48 h
   (N = 4) and 72 h (N = 2). Cav1 increased over time in wild-type males
   and females (Fig. [61]1c), but only very weak in HepCAV1ko hepatocytes.
   CAV1 protein expression was significantly higher in HepCAV1wt at both
   time points, as assessed upon quantification (Fig. [62]1d). In
   HepCAV1wt mice, CAV1 abundance increased over time, as expected^[63]17,
   whereas no change in HepCAV1ko mice was observed.

   With respect to gender, mRNA expression of Cav1 was obviously reduced
   in males and females, although p values did not reach significance
   (males: p = 0.0649, N = 6; females: p = 0.0771, N = 6, Fig. [64]1e). It
   has been reported that high fat diet reduces Cav1 expression in murine
   livers^[65]10. Thus, we checked whether we could recapitulate this
   finding. The MCD diet led to a reduction of Cav1 expression in male
   wild-type mice by tendency, similar as for high fat diet, while an
   increasing trend was found in females (Fig. [66]1f). Therefore, a
   gender-specific regulation seems to occur. Underlining this fact, total
   liver Cav1 expression in males was higher than in females, while not
   being significant (p = 0.45, Fig. [67]1f).

Lack of hepatocyte-specific CAV1 did not alter body weight and liver/body
weight ratio in healthy and NAFLD mice

   Prior start of experiments, at the age of 8 weeks, no significant
   weight difference was observed between HepCAV1wt and HepCAV1ko mice in
   both sexes (males: HepCAV1wt: 24.92 g ± 2.92 vs HepCAV1ko:
   24.09 g ± 1.85; females: HepCAV1wt: 19.88 g ± 1.86 vs HepCAV1ko:
   19.88 g ± 1.61; N = 12 for each sex, Fig. [68]2a). Multiple regression
   analysis showed that gender and diet were significant in body weight
   change, whereas the genotype and its interactions between gender and
   diet were not relevant (Fig. [69]2b). The liver/body weight ratio did
   not significantly change upon MCD diet in any genotype or any sex (Fig.
   [70]2c).

Fig. 2. Alteration of body weight and liver/body weight ratio.

   [71]Fig. 2
   [72]Open in a new tab

   a Body weight comparison between HepCAV1wt and HepCAV1ko mice (N = 12).
   Data of both males and females were analyzed by Mann-Whitney test. b
   Body weight change of males and females during 8th–12th week of
   experiment. Body weight was significantly reduced upon MCD diet
   compared with control diet over time, regardless of genotype. c
   Liver/body weight ratio of mice in both genders. No significant
   differences were determined after MCD feeding by t-test. WT: HepCAV1wt
   mice with control diet; KO: HepCAV1ko mice with control diet; WT MCD:
   HepCAV1wt mice fed MCD diet; KO MCD: HepCAV1ko mice fed MCD diet.

Hepatocyte-specific Cav1 knockout does not affect serum parameters

   To evaluate liver function and integrity, liver injury parameters, such
   as ALT, AST, GLDH and ALP, were measured (Fig. [73]3a, see Table
   [74]1). Plasma ALT and AST were elevated by MCD diet, but gender and
   genotype did not show any influence. Alkaline phosphatase did not
   change significantly in HepCAV1wt mice, while it showed significant
   difference in HepCAV1ko male mice upon MCD diet. GLDH increased
   significantly after MCD diet in HepCAV1wt males and females, while
   considering the knockouts, only in males. Significant differences were
   additionally detected between HepCAV1ko male and female mice fed MCD
   diet, i.e., reduced GLDH release in females, which indicated a gender
   difference in connection with hepatocyte CAV1.

Fig. 3. Plasma indices and pathological changes of HepCAV1ko mice upon
feeding MCD diet for 4 weeks.

   [75]Fig. 3
   [76]Open in a new tab

   a Plasma parameters indicating liver injury, i.e., AST, ALT, GLDH, ALP.
   b Lipid metabolism-related parameters, including cholesterol and
   triglycerides in serum. c Carbohydrate metabolism-related parameters,
   including glucose and urea in serum. d H&E staining. e Oil Red O
   staining for testing liver steatosis. f Sirius Red staining for
   detecting fibrosis status. g Quantification of Oil Red O staining.
   N = 6 for each group and data was analyzed by ANOVA with multiple
   comparisons and Tukey’s posthoc test.

Table 1.

   Plasma parameters in mice. Data was shown as mean ± SD.
   Males Females
   WT con KO con WT MCD KO MCD WT con KO con WT MCD KO MCD
   ALT 23 ± 13.78 30 ± 24.07 293.8 ± 186.4 401.5 ± 156.3 18.17 ± 5.46
   14.67 ± 3.93 364.5 ± 180.5 257.8 ± 177.4
   AST 48.9 ± 15.47 73.17 ± 50.49 262.8 ± 136.8 266.3 ± 98.35
   70.83 ± 17.63 53 ± 5.76 251.7 ± 102.7 188.3 ± 84.53
   ALP 107.8 ± 33.43 98.17 ± 35.44 139.2 ± 26.45 151.2 ± 18.19 150.3 ± 28
   146 ± 24.13 156.2 ± 38.82 126.7 ± 15.82
   GLDH 15.83 ± 6.97 23.75 ± 21.68 228.2 ± 105.8 320.3 ± 96.61
   7.552 ± 2.05 6.29 ± 2.46 251.2 ± 182 138.4 ± 77.43
   Cholesterol 148 ± 57.78 141.2 ± 52.49 51.67 ± 10.07 48.17 ± 17.96
   126.2 ± 34.39 138.2 ± 31.68 51.67 ± 22.6 52.67 ± 12.16
   Triglycerides 84.33 ± 28.32 72.17 ± 36.14 57.67 ± 14.22 53.17 ± 6.31
   82.83 ± 12.91 74.33 ± 22.39 44.17 ± 11.62 43.67 ± 5.05
   Urea 38.12 ± 4.054 37.8 ± 4.56 53.78 ± 5.44 48.77 ± 5.94 39.13 ± 6.99
   40.95 ± 8.75 46.32 ± 11.03 53.92 ± 9.96
   Glucose 273.8 ± 27.01 228.7 ± 44.8 155 ± 29.26 123 ± 16.69 246.3 ± 44.2
   256.3 ± 41.23 146.2 ± 24.85 135.5 ± 24.44
   [77]Open in a new tab

   Regarding metabolic activity, serum cholesterol, and triglyceride
   levels were measured (Fig. [78]3b). Serum cholesterol showed a
   significant reduction in both genders and genotypes upon MCD feeding.
   Serum triglyceride level only significantly reduced in HepCAV1wt female
   mice fed MCD compared to controls, but not observed in the other groups
   (HepCAV1wt male, HepCAV1ko female and HepCAV1ko male mice). Urea and
   glucose, important indicators of detoxification and glucose metabolism,
   respectively, were also tested (Fig. [79]3c). Urea increased
   significantly in HepCAV1wt male mice after feeding MCD diet, while in
   all other groups upon MCD feeding, urea was only slightly increased
   only. Next, glucose itself was significantly reduced upon MCD diet in
   both genders and genotypes. However, these differences were independent
   of CAV1 abundance.

Histology of HepCAV1wt and HepCAV1ko mice fed MCD diet

   Complementing findings on CAV1 functions on serum parameters and
   liver/body weight, architectural changes of NAFLD mouse livers were to
   be analyzed. Thus, HepCAV1wt and HepCAV1ko mice were initially tested
   for hepatic steatosis. Notably, H&E staining and Oil Red O staining
   revealed severe steatosis with various sizes of lipid deposition in the
   livers of MCD diet mice (Fig. [80]3d, e). To assess fibrotic
   remodelling after NAFLD development, Sirius Red O staining was
   performed. Four weeks MCD feeding yielded only slight fibrosis (Fig.
   [81]3f). Quantification of Oil Red O staining positive area showed no
   significant difference in HepCAV1ko mice compared to control livers of
   males and females (Fig. [82]3g). However, all these alterations were
   independent of an hepatocyte-specific Cav1 knockout. Ki67 staining was
   performed to determine the degree of cell proliferation. No significant
   differences in Ki67 staining were identified between HepCAV1ko and
   HepCAV1wt with control or MCD diet (Fig. [83]S1). Hence, the lack of
   hepatocyte CAV1 is not a critical factor for the development of liver
   steatosis and fibrosis in the MCD model.

Minimal overlap between HepCAV1ko and global Cav1 knockout datasets

   To assess the impact of hepatocyte-specific Cav1 knockout on liver gene
   expression profiles, we analyzed microarray data of male HepCAV1ko and
   HepCAV1wt mice liver. We found 1211 genes regulated (p < 0.05), among
   them 623 upregulated and 588 downregulated genes (Fig. [84]S2A, B,
   Table [85]S1). KEGG pathway analysis showed that several signaling
   pathways were deregulated (upregulated genes, connected to e.g.,
   calcium-, adipocytokine-, glucagon-, and oxytocin signaling pathways).
   Among the downregulated genes, we identified cancer-related signaling
   pathways, the peroxisome proliferator-activated receptor (PPAR)
   signaling pathway, the TGF-β signaling pathway and steroid hormone
   biosynthesis (Fig. [86]4a, Table [87]S2). Genes in fatty acid
   β-oxidation and peroxisomal function were also suppressed in gene
   ontology analysis (Fig. [88]S2C). To compare differences between
   HepCAV1ko and global Cav1 knockout (GloCAV1ko), we accessed a published
   dataset from GloCAV1ko livers ([89]GSE35431, NCBI). Here, we initially
   analyzed the KEGG pathways and found that both, up- and downregulated
   genes in [90]GSE35431 were mainly linked with metabolic processes and
   pathways. Downregulated genes were also related to nervous system
   diseases, such as Parkinson’s, Huntington’s, and Alzheimer’s disease.
   Relevant genes are also listed in Table [91]S3. For further comparison
   of HepCAV1ko and GloCAV1ko datasets in context of metabolic processes,
   we screened for metabolic genes in each dataset. One hundred and
   fourteen metabolic genes were deregulated in HepCAV1ko compared to
   HepCAV1wt mice, while 426 metabolic genes were deregulated in GloCAV1ko
   compared to GloCAV1wt mice, demonstrating that the global knockout has
   tremendous more impact on metabolic processes, outperforming HepCAV1ko
   mice almost 4-fold (Fig. [92]4b). To summarize, the global absence of
   CAV1 has stronger impact on metabolic processes than a CAV1 knockout
   specifically in hepatocytes, indicating that although hepatocytes are
   the main metabolic cells in the liver, the environment owns potent
   influence which could, to a significant degree, be CAV1 dependent.

Fig. 4. Gene expression profiling of HepCAV1ko and overlap with GloCAV1ko
expression data.

   [93]Fig. 4
   [94]Open in a new tab

   a KEGG pathway annotation of up- or downregulated genes from male
   HepCAV1ko vs HepCAV1wt mice. b Metabolic gene numbers and
   classification in livers of GloCAV1ko (426 genes) and HepCAV1ko (114
   genes) mice. c Overlap of upregulated genes in [95]GSE35431 and
   HepCAV1ko datasets. d Distribution of downregulated genes in
   [96]GSE35431 and HepCAV1ko datasets. e Genes upregulated in
   [97]GSE35431, while inversely expressed in HepCAV1ko dataset. f Genes
   downregulated in [98]GSE35431, while upregulated in HepCAV1ko dataset.

   Next, we compared the GloCAV1ko with the HepCAV1ko to assess the degree
   of overlap. Indeed, only 29 upregulated and 25 downregulated genes
   overlapped in the two datasets (Fig. [99]4c, d). Interestingly, there
   were also some genes showing opposite regulation, as 25 genes were
   upregulated in GloCAV1ko dataset while downregulated in the HepCAV1ko
   dataset, and 33 genes were downregulated in GloCAV1ko and upregulated
   in the HepCAV1ko livers (Fig. [100]4e, f). Among the overlapping genes,
   only 13 genes were connected to metabolism (Cyp2b9, Ces1b, Ugt1a9,
   Galc, Mat2b, Acadl, Ip6k2, Cox5b, Kcnk5, Tpo, Pomt2, Dhcr7, Pck1). In
   summary, there were less than 2% of total genes overlapping and
   contained only a limited number of metabolic genes. This indicated that
   the type (hepatocyte-specific vs global) of Cav1 knockout has distinct
   consequences on gene expression patterns in mouse livers.

Gene expression changes upon MCD diet

   We considered three main features as independent variables in this
   study, namely: genotype (HepCAV1ko vs HepCAV1wt), diet (MCD vs control
   diet) and gender (males vs females). Principal component analysis
   showed that the diet factor accounted for 26.6% of effects, gender for
   16.5%, and the genotype solely for 2.3%, whereas residual effects were
   also strong (54.6%, Fig. [101]5a). Thus, the impact of the
   hepatocyte-specific Cav1 knockout on disease outcome was indeed minor.
   In contrast, MCD diet, which is a well described NAFLD model strongly
   altered the gene expression profiles. Initially, expression data from
   HepCAV1wt male mice that were fed the MCD diet for 4 weeks was analyzed
   and compared with data from HepCAV1wt male mice fed control diet. 3861
   genes were regulated (p < 0.05, 1973 upregulated and 1888
   downregulated, Fig. [102]5b, c, Table [103]S4). KEGG pathway annotation
   indicated that the MCD diet caused overall upregulation of genes
   related to metabolic pathways and several metabolic processes,
   including retinol metabolism (e.g., Cyp2c55, Cyp2b9, Cyp3a11) and
   glutathione metabolism (e.g., Gsta4, Gclc, Ggct), but also genes linked
   to DNA replication (e.g., Lig1, Pole, Pola1) and cell cycle (e.g.,
   E2f1, Dbf4, Tgfb3, Fig. [104]5d, Table [105]S5). These expression
   changes were confirmed using gene ontology analysis, which identified
   several metabolism-related alterations, including oxidation-reduction
   processes and lipid metabolism and cell cycle (Fig. [106]5e). However,
   downregulated genes were also mainly classified to metabolic pathways
   as analyzed by KEGG, e.g., drug metabolism (e.g., Upb1, Nat1, Ugt2b1)
   and amino acid degradation (e.g., valine, leucine and isoleucine;
   Aldh6a1, Acadsb, Acsf3, Fig. [107]5d, Table [108]S5). Similar changes
   were found in gene ontology analysis with respect to metabolic
   processes, mitochondrial, complement activation and endoplasmic
   reticulum (Fig. [109]5e).

Fig. 5. Microarray analysis for MCD diet-induced changes in male HepCAV1wt
mice.

   [110]Fig. 5
   [111]Open in a new tab

   a Principal component analysis revealed the proportion of factors
   including diet, gender, genotype and residual. b Number of
   differentially regulated genes (N = 3861). c List of top and bottom 20
   significantly regulated genes. d KEGG pathway annotation of up- or
   downregulated genes proves that MCD diet has most impact on metabolic
   pathways. e Molecular functions, biological processes and cellular
   components analyses of up- or downregulated genes confirmed the results
   from KEGG pathway annotation. red: upregulated, green: downregulated.

Gender differences in gene expression profiles

   From phenotypic and microarray principal component analyses, gender was
   determined as another crucial factor, that exerted strong effects on
   gene expression changes. Hence, gender differences in HepCAV1wt and
   HepCAV1ko mice under healthy conditions and NAFLD were investigated.
   First, we compared gene patterns of HepCAV1wt in healthy males and
   females to evaluate basal gender differences. Surprisingly, 2618 genes
   were significantly deregulated between males and females (p < 0.05,
   1311 higher in males and 1307 higher in females, Tables [112]2,
   [113]S6). KEGG pathway analysis showed that genes significantly higher
   expressed in male mice were related to PPAR signaling, ribosome and
   various metabolic processes including fatty acid metabolism and steroid
   hormone biosynthesis. Genes which were significantly more expressed in
   female mice were also closely related with metabolic processes, such as
   drug-, retinol-, and glutathione metabolism (Fig. [114]6a, Table
   [115]S7). Interestingly, fatty acid metabolism was detected in both
   upregulated and downregulated gene sets, hence, a reliable conclusion
   in which direction gender effects are to be interpreted is impossible.

Table 2.

   Number of total regulated genes and metabolic genes in comparison of
   males and females.
   WT con (M vs F) KO con (M vs F) WT MCD (M vs F) KO MCD (M vs F)
   Total deregulated genes 2618 2667 1777 1402
   Upregulated genes 1311 1107 951 756
   Downregulated genes 1307 1560 826 646
   Total deregulated metabolic genes 382 433 204 239
   Upregulated metabolic genes 146 136 117 98
   Downregulated metabolic genes 236 297 87 141
   Metabolic /total deregulated genes (%) 14.59% 16.24% 11.48% 17.05%
   [116]Open in a new tab

Fig. 6. Gender differences of deregulated genes in males compared to females.

   [117]Fig. 6
   [118]Open in a new tab

   a KEGG pathway annotation of highly or lowly regulated genes comparing
   males to females in HepCAV1wt mice with control diet. b KEGG pathway
   annotation of up- or downregulated revealed that downregulated genes
   were related to lipid metabolism after MCD diet in HepCAV1ko compared
   to HepCAV1wt male mice. c KEGG pathway annotation of deregulated genes
   showed that HepCAV1ko in female induced metabolic processes when
   compared to HepCAV1wt female mice upon MCD diet. M males, F females.

   Furthermore, gender analyses of healthy and NAFLD HepCAV1ko mice showed
   that the number of deregulated gene was less upon MCD diet compared to
   control diet groups of both genotypes (wild-type: 2618 genes vs 1777
   genes upon MCD diet, knockout: 2667 genes vs 1402 genes upon MCD diet),
   which indicated that the diet dominated gender effects (Table [119]2).
   From the KEGG pathway analysis for up- or downregulated genes in each
   group, metabolic reprogramming was shown as a common phenomenon (Fig.
   [120]S3).

   Ratios of metabolic/total deregulated genes of male versus female mice
   in each group (WT control diet—14.59%; KO control diet—16.24%; WT MCD
   diet—11.48%; KO MCD diet—17.05%) were between 11–18%. To further
   evaluate the gender difference caused by Cav1 knockout mice between
   wild-type and knockouts, we compared overlapping genes after normal
   diet. Here, 126 genes were downregulated (20.3%) and 60 upregulated
   (9.7%) in both dataset (Fig. [121]S4A, Table [122]S8). After MCD diet,
   only 16 genes showed a downregulation (4%) and 20 were upregulated (5%,
   Fig. [123]S4B, Table [124]S8). Thus, the gender-dependent differences
   caused by Cav1 knockout in metabolic gene expression patterns were
   enhanced after MCD diet compared with control fed mice.

Hepatocyte-specific Cav1 knockout in NAFLD development

   Although no phenotypic differences between HepCAV1ko and HepCAV1wt were
   observed under healthy conditions, transcriptomic alterations were
   determined. Thus, it was of interest to see whether the absence of
   hepatocytic Cav1 was reflected on the gene expression level in diseased
   mice. Gene expression profiles of HepCAV1wt and HepCAV1ko upon feeding
   MCD diet were analyzed in context of gender.

   In male MCD fed mice, 1000 genes were significantly regulated in
   HepCAV1ko compared to HepCAV1wt mice (in total 471 upregulated,
   including Dmbt1, Fbl, Egr1, and 529 downregulated genes, including
   Evol3, Cyp2c40, Cyp2c38; top 20 genes are listed in Fig. [125]S5A, B).
   Upregulated genes were related to SNARE interactions in vesicular
   transport, amino sugar and nucleotide sugar metabolism. Downregulated
   genes were mainly involved in distinct metabolic processes (including
   arachidonic acid, tryptophan, and fatty acid metabolism, Fig. [126]6b).
   Biological process analysis confirmed the finding shown in pathway
   annotation with presenting lipid metabolic processes and
   oxidation-reduction processes (Fig. [127]S5C). As conclusion, we report
   that male HepCAV1ko mice mainly responded with a downregulation of
   lipid metabolism-related genes upon MCD diet. However, on the phenotype
   level, this functional outcome did not manifest.

   In females, 961 genes were deregulated (476 upregulated genes,
   including Cyp3a16, A1bg, Trav9-2, and 485 downregulated genes,
   including Slco1a1, Gm6614, Ugt1a7c in the knockouts, Fig. [128]S6A, B).
   Deriving from KEGG pathway annotation, upregulated genes were related
   to metabolic pathways, e.g., carbon and amino acid metabolism, while
   downregulated genes were less clusterable in KEGG pathway, only showing
   cancer-relation, including microRNAs in cancer, viral carcinogenesis
   and prostate cancer (Fig. [129]6c). Biological process analysis and
   molecular function nevertheless supported the results from KEGG pathway
   annotation, showing metabolic processes and oxidoreductase activity as
   upregulated (Fig. [130]S6C). Therefore, in females, HepCAV1ko induced
   metabolic processes while in males those were reduced. A few strongly
   altered metabolic genes from array analyses in males and females were
   tested by PCR. As in the array, Elov3 and Csad were significantly
   changed, and for Dpys and Got1 we found a clear trend (Fig. [131]S7),
   supporting our array findings.

   Deregulated metabolic genes in the HepCAV1ko were analyzed in four data
   subsets (male HepCAV1ko mice control diet vs wild-type, male HepCAV1ko
   mice MCD diet vs wild-type, female HepCAV1ko mice control diet vs
   wild-type, female HepCAV1ko mice MCD diet vs wild-type). As shown in
   Fig. [132]4b and Fig. [133]S8, metabolic events were classified in
   diverse programs. Among them, small molecular transport, ion transport,
   fatty acid, glycan, and amino acid are most relevant metabolic
   processes. In context of the glucose metabolism relevance in liver
   regeneration^[134]9, we more closely analyzed whether genes related to
   glycometabolism were affected by diet (sex dependent). Functional
   involvement of altered genes in gluconeogenesis and glycolysis was
   rather low (~1–3% of metabolic altered genes). However, genes related
   to glycan metabolism were altered more frequent, with a share of 7–11%
   of metabolic altered genes (Table [135]S9).

   When overlapping all deregulated genes from males and females, only a
   small number of genes were found regulated in both sexes (Fig. [136]S9,
   Table [137]S10), indicating that CAV1 exerts gender-dependent effects
   in NAFLD disease.

   To conclude, these results provide evidence that CAV1 prominently
   influences metabolic processes during disease development, but also
   determines gender effects.

Discussion

   In this study, we investigated the effects of hepatocyte-specific CAV1
   in healthy and in NAFLD mice phenotype evaluation of tissue and gene
   expression profiling as well as plasma analysis.

   The hepatocyte-specific knockout using the Cre/loxP system was
   successfully established, although mRNA expression of Cav1 in liver
   tissue (p = 0.0649 in males, 0.0771 in females) did not reach
   significance, which is due to normal Cav1 expression in non-parenchymal
   cells (and especially LSEC express higher levels compared to
   hepatocytes). In context of distinct cell types expressing higher CAV1
   levels, this protein will exert cell specific functions and
   interestingly, in LSEC, CAV1 impacts on eNOS regulation, autophagy and
   defenestration—crucial events during liver diseases^[138]18,[139]19.
   Therewith, applying the cell type specific knockout in context of LSEC
   and other liver cells will be relevant to broaden our understanding on
   CAV1 in liver disease contexts. Regarding parenchymal cells, the
   knockout in isolated hepatocytes was evident—as a significant reduction
   of Cav1 was shown on RNA level in both sexes. On protein level, some
   CAV1 expression was still detectable in knockout hepatocyte lysates,
   likely due to the conditional knockout system that could be leaky.
   However, compelling evidence proves CAV1 reduction in the HepCAV1ko and
   thus, we moved onto study its functions in vivo.

   CAV1 regulation in NAFLD showed conflicting results in the literature.
   CAV1 expression was repressed in L02 and AML12 cells cultured with free
   fatty acids to trigger steatosis in vitro. In vivo using C57BL/6 J mice
   fed a high fat diet (HFD) to induce NAFLD, CAV1 expression was reduced.
   Similar findings were obtained in obese mice, and even in NAFLD
   patients^[140]10,[141]20. These findings are in line, showing a
   reduction of CAV1 expression upon feeding MCD in male mice, but not in
   females. However, in another study, C57BL/6 male mice fed a high fat
   and cholesterol enriched diet showed increased CAV1 expression in
   liver^[142]21. Regarding our study, CAV1 expression showed a tendency
   towards upregulation in MCD fed females. Hence, gender plays a crucial
   role in controlling CAV1 expression in NAFLD. Li et al. may provide a
   key explanation on the mechanism of CAV1 regulation in NAFLD: CAV1
   expression was proved to be reduced by upregulated miR199a-5p in
   hepatocytes^[143]20. Indeed, this microRNA is regulated during liver
   diseases^[144]22,[145]23. Thus, it is worth investigating whether
   miR199a-5p expression is gender-dependent.

   The liver is pivotal in regulating systemic homeostasis and balancing
   metabolism, in concert with other tissue, e.g., adipose tissue and the
   gastrointestinal tract. In recent studies, CAV1 was proven as an
   important membrane protein for regulating metabolism and different
   signalling pathways, but most in vivo studies were done with global
   Cav1 knockout animals, reporting some contradictory outcomes. For
   example, CAV1 was proven to protect from lipid accumulation in L02 and
   AML12 hepatocyte cells upon FFA treatment and in mice after 4 weeks
   HFD^[146]10. While in the Asterholm study, opposite results were shown,
   i.e., that Cav1 knockout mice had reduced hepatic steatosis after 8
   weeks HFD in response to 24 h fasting^[147]13. These conflicting
   results implicated that CAV1 abundance related steatosis may be
   influenced by diet/feeding conditions, for example whether animals were
   fasting prior analysis. They also assumed that the liver may play a
   secondary compensatory role as the metabolic dysfunctions should be
   mediated by adipose tissue^[148]13. These results were consistent with
   our findings using hepatocyte-specific Cav1 knockout mice, pointing to
   the possibility that CAV1 in other tissues/cells affects susceptibility
   for steatosis. However, when studying liver regeneration upon partial
   hepatectomy, Fernández-Rojo and colleagues determined hepatocyte
   autonomous effects of CAV1^[149]9. Cav1 knockout animals shifted
   towards carbohydrate-dependent adaptation. In AML12 cells, lack of CAV1
   under metabolic stress led to alterations in the extracellular
   acidification rate, indicating alterations in lactate efflux and thus
   of glycolysis. The MCD NAFLD/NASH model is also triggering massive
   metabolic stress and hence, a similar adaptation might take place. With
   absent CAV1, glucose availability thus may influence hepatocyte
   integrity and proliferation capacity and therewith directly influence
   disease progression. Other animal models without lack of essential
   nutritional compounds, such as Western diet, might be more appropriate
   to study those mechanisms due to the mode of action. On gene expression
   level, data did not indicate major alterations of glycolysis or
   gluconeogenesis in dependency of CAV1. Comparing our analyses with
   GloCAV1ko mice data that were showing phenotypes like decreased body
   mass, increased liver/body ratio, increased triglyceride and
   cholesterol levels^[150]13,[151]24, this was not observed in HepCAV1ko
   mice under control conditions. Also, comparing KEGG pathways and
   metabolic events of GloCAV1ko and HepCAV1ko mice, strong differences
   and little overlap were detected, thus consistent with the phenotypes.

   Sexual dimorphism was discussed as second important factor in weight
   average proportion analysis. In this study, strong metabolic
   differences between the sexes were determined, regardless of genotype
   and diet. In previous studies, gender-specific metabolic differences
   caused by CAV1 were reported. In the CAV1 intestinal epithelial cell
   knockout mouse model, plasma low density lipoprotein (LDL) cholesterol
   increase by HFD was prevented in males, but not in females^[152]25.
   Also, LDL cholesterol was only detected in male knockout mice^[153]11.
   Here, we observed a GLDH gender difference in the HepCAV1ko mice upon
   MCD diet. More interestingly, male HepCAV1ko mice demonstrated
   suppressed, while females had induced metabolic gene expression
   patterns upon MCD feeding. However, strong basal differences (healthy
   conditions) between male and female FVB mice had been monitored (Table
   [154]2, Fig. [155]6a). Based on this, CAV1 seems to play sex-dependent
   metabolic functions in hepatocytes, although more studies are needed to
   determine in which contexts phenotypes become relevant. Regarding the
   mechanisms of CAV1 gender effects, some studies indicated that it may
   due to the interaction between CAV1 and sex hormone signaling to
   regulate metabolic processes, like androgen receptor^[156]26 or steroid
   signaling, such as E2^[157]27. Whether this regulation applies also in
   this setting has yet to be proven.

   We identified significant alterations of gene expression profiles in
   HepCAV1ko mice in NAFLD, while not on the disease level. A possible
   reason for this is the short time feeding (4 weeks) which only presents
   the early phase of disease (with slight fibrosis). The HFD model could
   be applied to investigate later phases of NASH-disease for further
   studying effects of CAV1.

   In summary, our findings provide evidence that hepatocyte-specific Cav1
   knockout does not affect MCD mediated NAFLD development, but alters
   expression level of a plethora of genes. Cav1 deletion led to strong
   differences in gene expression profiles in male and female livers, in
   both healthy and NAFLD mice. Future studies will shed light on whether
   gender differences with respect to CAV1 function are of
   pathophysiological relevance in the development of metabolic liver
   diseases.

Materials and methods

Mice

   Floxed CAV1^−/− mice were obtained from Prof. Philip Scherer
   (University of Texas) and details about the mice are depicted in the
   study of Asterholm et al.^[158]13. In this project, floxed CAV1^−/−
   mice were crossed with Albumin promoter Cre-recombinase (Alb-Cre)
   heterozygous mice (BL/6-TgN(Alb-Cre−/ + )) to obtain HepCAV1ko mice and
   wild-type littermates (HepCAV1wt). The mice genotype was confirmed by
   PCR of ear tissue using the following primers: floxed Cav1 (forward:
   5′-3′: GTG CAT CAG CCG CGT CTA CTC and reverse: 5′-3′: GGC CGT AAC CTG
   AAT CTC TTC CCT TTG) and Alb-Cre PCR (forward: 5′-3′: AAG TTG AAT AAC
   CGG AAA TGG and reverse: 5′-3′: AGC TAC ACC AGA GAC GGA AAT). Mice were
   housed in a pathogen-free standard environment. Group allocation for
   the experiments was randomized and not blinded. Sample analyses were
   not blinded. All procedures with mice were performed according to
   national and international guidelines. Prior ethics approval was
   obtained from the local ethics committee of the state
   Baden-Württemberg.

Experimental diet

   Mice were kept with a normal diet ad libitum until 8 weeks of age.
   Prior start of the experiments, both HepCAV1ko and HepCAV1wt mice were
   randomly divided into two groups for either feeding the MCD diet
   (N = 12) or control diet with methionine and choline (N = 12) for 4
   weeks. MCD and control diets were purchased from Ssniff ([159]E15652
   and [160]E15654, Germany). Each diet group included 6 male and 6 female
   mice. Mice were weighted twice per week during the 4 weeks of
   experimentation.

Serum analyses

   At the end of the experiment, blood from retrobulbar plexus was
   preserved for further serum analyses. Blood biochemical indices, among
   alanine aminotransferase, aspartate aminotransferase (AST), alkaline
   phosphatase, glutamatdehydrogenase (GLDH), triglyceride, cholesterol,
   urea, and glucose were measured using a Hitachi automatic analyzer.

Histological analysis

   For histopathologic evaluation, livers were fixed with 4%
   paraformaldehyde, embedded in paraffin, and cut into 4 μm thick slides.
   Hematoxylin and eosin (H&E, Darmstadt, Germany) and Sirius red
   stainings (PSR, Missouri, USA) were done to assess liver inflammation
   and fibrosis. For detection of lipid droplets in livers, cryostat
   sections of 8 μm were stained with Oil Red O according to standard
   procedures. Sections were imaged at ×20 magnification. Quantification
   of Oil Red staining was done with ImageJ software (Version1.51j8).
   Immunohistochemistry for Ki67 (12202 S, Cell Signaling) was performed
   to determine cell proliferation activity in tissue of HepCAV1wt and
   HepCAV1ko mice fed normal or MCD diet. Quantification was done by
   counting positive nuclei upon ×20 magnification in 20 random fields per
   mouse. For each mouse, values were averaged and shown in the graph
   (Fig. [161]S1).

Hepatocyte isolation and culture

   Primary hepatocytes were isolated with a two-step collagenase (C2-22,
   Merk Biochrom) perfusion method as described previously^[162]28.
   Freshly isolated hepatocytes were then seeded on 6-well plates
   (3 × 10^5 per well) and cultured in Williams E medium supplemented with
   1% l-glutamine, 1% penicillin/streptomycin, 0.1% dexamethasone, and 10%
   fetal bovine serum^[163]28. Upon attachment (approximately after 4 h),
   cells were cultured in starvation medium without dexamethasone and FBS.
   RNA and protein lysates of primary hepatocytes were harvested after 48
   and 72 h.

RNA isolation and quantitative PCR

   Total RNA was isolated from whole liver tissue or primary hepatocytes
   using InviTrap Spin Universal RNA Mini Kit (Stratec Biomedical AG,
   Germany), according to the manufacturer’s instructions. RNA
   concentration was determined using the Tecan Infinite M200 microplate
   reader. Subsequently, RNA (1 µg) was reverse transcribed to cDNA using
   random primers and Revert Aid H Minus Reverse Transcriptase (Thermo
   Scientific). Quantitative polymerase chain reaction (qPCR) analysis was
   performed in a 10 µl total reaction volume using EvaGreen qPCR Mix Plus
   (Solis BioDyne, Tartu, Estonia) on a StepOne (Applied Biosystems). The
   qPCR reaction mix was composed of 2 µl EvaGreen, 12.5 ng cDNA, 5 pmol
   of gene primer mix, and then filled to 10 µl with nuclease-free water.
   The experiments were performed in triplicates (or as stated) and
   peptidylprolyl isomerase A (Ppia) was used for normalization. The gene
   primer sequences are as follows (5′-3′): Cav1: forward: GAA GGG ACA CAC
   AGT TTC GAC; reverse: GGA TGC CGA AGA TCG TAG ACA. Ppia: forward: GAG
   CTG TTT GCA GAC AAA GTT; reverse: CCC TGG CAC ATG AAT CCT GG. Elovl3:
   forward: TTC TCA CGC GGG TTA AAA ATG G; reverse: GAG CAA CAG ATA GAC
   GAC CAC. Got1: forward: GCG CCT CCA TCA GTC TTT G; reverse: ATT CAT CTG
   TGC GGT ACG CTC. Dpys: forward: GGT GGA GCG ACA AGG TAA AAG; reverse:
   CTG CTC GTC TTG CAC CAT GT. Csad: forward: GAA GAA GGG GAC CAT GAT GA;
   reverse: CAC CAC CAT TCG GAA GAA GT.

Western blot analysis

   CAV1 protein abundance was tested with tissue lysates from whole liver
   and isolated primary hepatocytes as previously described^[164]29.
   Briefly, RIPA buffer with freshly added phosphatase and protease
   inhibitors was used to extract protein from tissue and cells. Protein
   concentration was assessed with the Bio-Rad protein assay kit according
   to the manufacturer’s instructions and quantified with the Tecan
   Infinite M200 via absorbance measurement at 690 nm. Equal amounts of
   protein (30 µg) were separated by SDS-PAGE (10–12% gels) and blotted
   onto a PVDF membrane (Millipore, Billerica, MA, USA). The membrane was
   blocked with 5% nonfat dry milk in TBST at room temperature for 1 h,
   and incubated overnight at 4 °C with primary antibodies. Primary
   antibodies against CAV1 (T3267, Cell Signaling) and GAPDH (sc25778,
   Santa Cruz Biotechnology) were used. HRP-linked anti-mouse (sc-2005,
   Santa Cruz) and anti-rabbit antibodies (sc-2357, Santa Cruz) were used
   as secondary antibodies. Signals were visualized by incubating the
   blots in Supersignal Ultra solution (Pierce, Hamburg, Germany) and
   recorded with imaging system Fusion SL4 (PEQLAB, Germany). The target
   protein bands were quantified by using Fluorchem Q system
   (ProteinSimple, California, USA).

Microarray analysis

   Prior microarray analysis, RNA isolated from liver tissue was tested by
   capillary electrophoresis on an Agilent 2100 bioanalyzer (Agilent) for
   confirming quality. Gene expression profiling was subsequently
   performed using mouse Affymetrix MoGene 2.0 ST Array Chips (Life
   Technologies, Germany). Biotinylated antisense cRNA was then prepared
   according to the Affymetrix standard labelling protocol with the
   GeneChip WT Plus Reagent Kit and the GeneChip Hybridization, Wash and
   Stain Kit (both from Affymetrix, Santa Clara, USA). Afterwards, the
   hybridization on the chip was performed on a GeneChip Hybridization
   oven 640, then dyed in the GeneChip Fluidics Station 450 and thereafter
   scanned with a GeneChip Scanner 3000. All of the equipment used was
   from the Affymetrix Company (Affymetrix, High Wycombe, UK).

Data analyses

   In addition to our own microarray data, the dataset [165]GSE35431 was
   downloaded from the Gene Expression Omnibus website
   ([166]https://www.ncbi.nlm.nih.gov/geo/). This dataset included gene
   profiles of livers from global Cav1 knockout (GloCAV1ko) and wild-type
   littermates (GloCAV1wt). A p value < 0.05 was defined as statistical
   significant. The significantly regulated genes were followed by Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway functional annotation
   and gene ontology analysis with DAVID Bioinformatics Resources 6.8
   ([167]https://david.ncifcrf.gov/). Venny tool was used to identify
   overlapping genes in two or more datasets
   ([168]http://bioinfogp.cnb.csic.es/tools/venny/). Metabolic genes were
   defined as published in another gene list^[169]30.

Statistics

   Statistical analyses were performed with GraphPad Prism version 6.0
   software. Statistical testing was done with two sided t-test or
   analysis of variation (ANOVA) where applicable. Normally distributed
   data was analyzed with t-test and non-normally distributed data was
   analyzed with Mann-Whitney test. We used ANOVA with multiple
   comparisons and Tukey’s posthoc test. Quantification of staining was
   done using ImageJ. Data is presented as mean ± SD. Each experiment was
   performed at least three times (or as indicated). Statistical
   significance is indicated as follows: *p < 0.05; **p < 0.01;
   ***p < 0.001. ****p < 0.0001

Supplementary information

   [170]Supplementary figure and table legends^ (17.2KB, docx)
   [171]Figure S1^ (3.1MB, png)
   [172]Figure S2^ (120.4KB, png)
   [173]Figure S3^ (113.4KB, png)
   [174]Figure S4^ (173.1KB, png)
   [175]Figure S5^ (119.4KB, png)
   [176]Figure S6^ (119.5KB, png)
   [177]Figure S7^ (72.2KB, png)
   [178]Figure S8^ (121.9KB, png)
   [179]Figure S9^ (511.8KB, png)
   [180]Table S1^ (23.1KB, xlsx)
   [181]Table S2^ (13.3KB, xlsx)
   [182]Table S3^ (14.1KB, xlsx)
   [183]Table S4^ (24KB, xlsx)
   [184]Table S5^ (13.1KB, xlsx)
   [185]Table S6^ (26.2KB, xlsx)
   [186]Table S7^ (12.3KB, xlsx)
   [187]Table S8^ (15.1KB, xlsx)
   [188]Table S9^ (10.8KB, xlsx)
   [189]Table S10^ (10.7KB, xlsx)

Acknowledgements