Impact of liver-specific GLUT8 silencing on fructose-induced inflammation and omega oxidation.

Excessive consumption of high-fructose diets is associated with insulin resistance, obesity, and non-alcoholic fatty liver disease (NAFLD). However, fructose differentially affects hepatic regulation of lipogenesis in males and females. Hence, additional studies are necessary in order to find strategies taking gender disparities in fructose-induced liver damage into consideration. Although the eighth member of facilitated glucose transporters (GLUT8) has been linked to fructose-induced macrosteatosis in female mice, its contribution to the inflammatory state of NAFLD remains to be elucidated. Combining pharmacological, biochemical, and proteomic approaches, we evaluated the preventive effect of targeted liver GLUT8 silencing on liver injury in a mice female fructose-induced non-alcoholic steatohepatitis female mouse model. Liver GLUT8-knockdown attenuated fructose-induced ER stress, recovered liver inflammation, and dramatically reduced fatty acid content, in part, via the omega oxidation. Therefore, this study links GLUT8 with liver inflammatory response and suggests GLUT8 as a potential target for the prevention of NAFLD.

Introduction

Non-alcoholic fatty liver disease (NAFLD) constitutes a global public health concern since it is the fastest-growing cause of end-stage liver disease and hepatocellular carcinoma (HCC). The prevalence of NAFLD parallels that of obesity and metabolic syndrome and is expected to continue rising over the next few years. One of the reasons for the increase in liver disease is related to a significant augment in the consumption of diets enriched in sugars, such as fructose which is a major component of the most widely used sweeteners in the western world, especially in beverages (Jensen et al., 2018; Vos and Lavine, 2013). Data gleaned over the last few years have allowed uncovering some of the mechanisms related to NAFLD development and associated comorbidities to be discovered. In this context, animals fed with high fructose diets have emerged as an important dietary model alongside high-fat/cholesterol or the methionine and choline-deficient diets (Lau et al., 2017; Van Herck et al., 2017; Zhong et al., 2020). Indeed, fructose intake has been proposed as a key player in the development of fatty liver disease (Basaranoglu et al., 2013; Hannou et al., 2018; Jegatheesan and De Bandt, 2017). Hence, a better understanding of liver fructose metabolism will facilitate the design of novel pharmacological therapies for NAFLD.

On the other hand, it is generally accepted that mechanisms involved in the development of liver disease are determined by the existence of sex differences. Both basic animal models and clinical studies have reported that males present higher severity and risk of NAFLD than female, at least during the reproductive stage (DiStefano, 2020b; Du et al., 2017; Lonardo et al., 2019). This sexual dimorphism has a crucial role in the onset, progression and treatment response of NAFLD. In fact, male mice fed long-term on high-fat diet (HFD) display steatohepatitis and inflammasome activation, whereas female mice have steatosis without inflammation (Ganz et al., 2014). In general, in animal NALFD models, male subjects present more severe steatosis and steatohepatitis, more pro-inflammatory/profibrotic cytokines, and a higher incidence of hepatic tumors than females. Contrastingly, it has also been reported that female mice develop more severe steatosis than males when fed a cafeteria diet (Gasparin et al., 2018). These apparent discrepancies could be explained by the fact that the liver proteins involved in glucose and lipid metabolism, inflammation or oxidative stress are differentially regulated between sexes. In this context, several studies indicate that this disparity is due to the preventive role of estrogen in hepatic steatosis (Chukijrungroat et al., 2017) among other mechanisms. Accordingly, recent studies have showed that higher hepatic estrogen-related receptor α expression in female mice contributes to the sex disparity in the assembly and secretion of hepatic triglyceride (TG)-rich very low-density lipoproteins (Yang et al., 2020).

Interestingly, despite females being at lower risk of NAFLD development when comparing with males (Balakrishnan et al., 2020; Lonardo et al., 2019), rodent data support the idea that this risk can increase when liver damage is mediated by fructose intake (Choi et al., 2017; Hyer et al., 2019; Spruss et al., 2012). Accordingly, human studies also revealed that women could be more affected than men by the ingestion of higher amounts of fructose than men and a fructose-rich diet sustained over time may lead to changes in hepatic fatty acid (FA) portioning and eventually to increased liver fat content (DiStefano, 2020a; Kang and Kim, 2017; Low et al., 2018; Rodgers et al., 2019). In line with this, studies in rats indicate that these sex differences were related to the fact that the liver enzyme fructokinase, which controlls fructose metabolism in the liver, was markedly induced by fructose liquid ingestion in females but not in males (Vila et al., 2011). In contrast to the large amount of data reporting the effect of high-fructose diet in male animals, the study of the impact of elevated consumption of fructose t in females has been neglected.

GLUT8 (Slc2A8) is the eighth member of facilitated hexose transporters superfamily. Despite knowledge about the tissue distribution and subcellular localization of GLUT8, its complete physiological function remains obscure. It has been shown that global GLUT8-deficient female mice present impaired hepatic first-pass fructose metabolism and, therefore, are protected from fructose-induced macrosteatosis (Debosch et al., 2014); however, its contribution to the physiopathology of NAFLD development is poorly understood. In this study, we examine in-depth two preclinical models of diet-induced NAFLD-non-alcoholic steatohepatitis (NASH) in female: one based on high-fructose and the other based on high-fat, to allow direct comparisons. Specifically, we addressed in female mice the following questions: (1) how detrimental is a fructose-enriched-diet and its impact in terms of deregulation of lipid metabolism, ER stress, oxidative stress and inflammation; (2) the impact of specific GLUT8 silencing on these parameters; and (3) the potential role of GLUT8 in the response to fructose-intake in hepatocytes (HCs) and liver stellate cells.

Our data show that long-term high-fructose intake is associated with NAFLD-NASH development in the absence of increased adiposity and/or obesity. This includes classical features of liver disease, such as liver fat deposition, lipotoxicity, and inflammation. Whole quantitative proteomic analysis and pathway enrichment showed a similarity with some of the known features of human liver disease. This adds further support to the translational value of our data, showing the prevention of NASH in animals with genetic silencing of liver GLUT8.

Results

High fructose and high fat promote different patterns of fatty liver

In contrast to HFD, animals exposed to prolonged exposure to a high-fructose diet showed similar body weight and adiposity to control animals under standard chow (Figures S1 and S2). Liver weight was increased in the high-fructose diet at all time points in comparison to control animals (Figure 1A). Following 22 weeks of exposure to either, HFD or high-fructose diet, there was a marked increase in liver fat accumulation (Figures 1B and 1C), and TG liver content (Figure 1D) as well as in cholesterol serum levels (Figure 1E) while levels of non-esterified fatty acids (NEFAs) levels were unaffected with the exception of elevated NEFAs in 12 week-HFD females (Figure 1D). Assessment of glucose homeostasis following GTT showed, as expected, insulin resistance in HFD animals and normal responses to insulin in high-fructose diets (Figure S3).

Figure 1

High-fructose and high-fat diets promote fatty liver accumulation in a different pattern

C57BL/6 female mice were fed with Standard Diet (SD), high-fructose (60% Fruct) or high-fat diet (60% HFD). Twelve days (n = 8 per group), 12 weeks (n = 4 per group), and 22 weeks (n = 3–4 per group) post-feeding, livers were isolated, (A) weighed, and subjected to (B) hematoxylin and eosin (H&E) (Scale bar: 50 μm). Total liver FA was measured by (C) Oil Red dyeing (Scale bar: 50 μm), and (D) the liver content of triglycerides (TG) and non-esterified FAs (NEFAs).

(E) Serum cholesterol levels were determined by colorimetric assay.

(F) DNL process was evaluated by measuring FAS and ACC protein expression. Representative blots from indicated time points are shown. Cropped blots are used in the figure. Samples derived from the same experiment and blots were processed in parallel. Results are expressed as mean ± SEM, ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001; One-Way ANOVA or Kruskall-Wallis test following by a post-hoc test (∗ compared to SD, # compared to 60% Fruct).

Animals exposed to high-fructose diet exhibited increased protein levels of lipogenic enzymes such as fatty acid synthase (FAS) and acetyl-CoA carboxylase (ACC) (Figure 1F), while protein levels of enzymes involved in lipid mobilization like adipose triglyceride lipase (ATGL) and carnitine palmitoyltransferase 1A (CPT1A) were reduced after 22 weeks fructose ingestion (Figure S4C). Gene expression of elongation of very-long-chain fatty acids (VLCFAs) protein 6 (Elovl6) and the endoplasmic reticulum enzyme stearoyl-CoA desaturase (Scd1) were differently regulated after 22 weeks of fructose consumption (Figure S4B). Overall, these data suggest that the nature of liver fat depots differ under high-fructose and HFD conditions. Both preclinical models exhibited deterioration of liver function, as shown by increased levels of alpha-fetoprotein (Afp) (Figure 2A), a marker of steatosis and HCC; a potential reduction in bile acid (BA) metabolism, indicated by the significantly reduced levels of sulfotransferase family 2A member 1 (Sult2A1) which mediates the sulfate conjugation of BAs, and hepatocyte nuclear factor 4 alpha (Hnf4a), a master regulator of hepatic function, specifically in animals exposed to high-fructose diet (Figure 2A).

Figure 2

High-fructose diet impairs liver function and triggers inflammation

C57BL/6 female mice fed on SD, 60% Fructose or 60% HFD for 22 weeks (n = 8 per group).

(A) Liver function was evaluated by gene expression of alpha-fetoprotein (Afp), sulfotransferase 2A1 (Sult2A1), and hepatocyte nuclear factor 4 Alpha (Hnf4a).

(B and C) (B) Liver ER stress markers were determined by gene expression of Bip, Atf4, Chop, and total Xbp1(discerning between unspliced usXbp1 and spliced sXbp1) (n = 8 per group), and (C) proteins levels of ATF6 and phosphorylated forms of the ER stress sensor proteins (pPERK and pIRE) (n = 3–4 per group).

(D and E) Liver inflammatory response was analyzed by (D) protein expression of anti-inflammatory related proteins (FGF21 and pCREB) (n = 3–4 per group); and (E) gene expression of pro-inflammatory cytokines IL1b and IL6 (n = 8 per group).

(F) Liver fibrosis was evaluated by gene expression of the fibrotic markers (Col1a2, Col3, Timp1 and aSma) (n = 8 per group).

(G and H) Red positive area was quantified from Red Sirius staining images (Scale bar: 20 μm) to estimate the levels of collagens in livers from different diets in (G) the plate of hepatocytes and (H) nearby liver portal field. Results are expressed as mean ± SEM, ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001, ∗∗∗∗p ≤ 0.0001; One-Way ANOVA or Kruskall-Wallis test following by a post-hoc test (∗ compared to SD, # compared to 60% Fruct).

Fructose triggers liver inflammation

As a vital organ for protein synthesis and detoxification, the liver is especially susceptible to ER stress. ER stress takes place when unfolded or misfolded proteins accumulate in the ER lumen. Thus, we analyzed the unfolded protein response (UPR) sensors: Inositol-requiring enzyme 1 (IRE1), protein kinase R-like endoplasmic reticulum kinase (PERK), and activating transcription factor-6 (ATF6), as well as downstream genes, such as binding immunoglobulin protein (Bip), Activating transcription factor-4 (Atf4), CCAAT-enhancer-binding protein homologous protein (Chop) and X-box binding protein (Xbp1) (Figures 2B and 2C). Interestingly, only the IRE sensor was significantly activated by fructose diet (Figure 2C), whereas Bip and Atf4 remained downregulated (Figure 2B). The activation of IRE1 induces splicing of Xbp1 mRNA, which is involved in responding to ER stress. Spliced XBP1 (sXbp1) binds to the endoplasmic reticulum stress elements, which promotes fibroblast growth factor 21 (FGF21) expression. Therefore, ER stress increases FGF21 synthesis as a protective mechanism since this helps to diminish importantly the oxidative stress via activation of ERK and cAMP-responsive element-binding protein (CREB). Our data indicate that this antioxidant mechanism is suppressed by fructose, because Xbp1s, FGF21, and phosphorylated levels of CREB (pCREB) were downregulated (Figures 2B and 2D). It is known that prolonged IRE activation or unresolved ER stress leads to ER stress-induced inflammasome activation and Interleukin 1 beta (IL1b) production, as well as other inflammatory mechanisms mediated by cytokines such as Interleukin 6 (IL6). Thus, we examined IL-6 and IL1b expression and found a significant increase in fructose-fed animals (Figure 2E). These results indicate that high levels of pIRE in fructose diet are accompanied by activation of inflammation-induced pathogenic ER stress pathways (Figure 5A). Moreover, we also analyzed markers of liver fibrosis such as collagen type 1 alpha 2 (Col1a2), collagen type III (Col3), tissue inhibitor of metalloproteinase-1 (Timp1), and alpha-smooth muscle actin (aSma) in our experimental model (Figure 2F). The results obtained suggested that, unlike HFD, fructose diet had an insignificant effect on the progression of liver fibrosis (Figures 2F–2H).

Figure 5

Liver GLUT8 knockdown improves ER stress and decrease inflammation/inflammasome induced by fructose diet

C57BL/6 female mice received a tail vein injection of 1x106 TU/mL lentiviral particles to inhibit Slc2A8 (shG8-Fruct) or to use as control (Cont-Fruct) and fed with 60% Fructose diet for 22 weeks (n = 7–8 per group).

(A) Schematic representation of the switched ER stress mechanism observed after GLUT8 knockdown.

(B and C) (B) Expression of the main genes involved in ER stress signaling including usXbp1 and uXbp1, and (C) protein levels of ATF6 and phospho-IRE ER stress sensors, chaperones (GPR78/BiP), and the C/EBP Homologous Protein (CHOP) Transcription Factor.

(D and E) Inflammation measured by gene expression of (D) IL6 and IL15 receptor produced during the immune response, a major macrophage marker (F4-80), tumor necrosis factor-alpha (Tnfa), and transforming growth factor-beta (Tgfb); and (E) gene expression of inflammasome markers (IL1b, Nlrp3, Asc, IL18).

(F) protein expression of Interleukin-1 converting enzyme (Caspase 1), that proteolytically cleaves precursors of the inflammatory cytokines IL1b and IL18.

(G) Protein levels of the phosphorylated form of cAMP response element-binding protein (CREB). Results are expressed as mean ± SEM. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001P; unpaired t test or Mann-Whitney test.

Specific liver GLUT8 silencing reduces fructose-induced NAFLD

It has been previously described that global GLUT8KO female mice exhibit attenuated fructose-induced hepatic TG and cholesterol accumulation (Debosch et al., 2014). To test whether this was an intrinsic liver effect, we assessed if targeted depletion of liver GLUT8 was sufficient to prevent the fructose-induced steatosis phenotype in C57BL/6 females. First, we validated the zone-specific GLUT8 knockdown after administration of Cont-Fruct or shG8-Fruct lentiviruses, using a specific antibody against GLUT8 (Figure S5 and Table S3). Immunohistochemistry analysis revealed a 79% reduction of liver immuno-positive signal in the shG8 animals (Figure 3A) and this reduction in Glut8 expression was specifically within the liver (Figures S6A and S6B). After this Glut8 silencing, there was an increased expression of both Glut2 and Glut5 (Figure S7). This upregulation in fructose transporters could be explained as a regulatory mechanism to compensate for the silencing of Glut8 and/or may be an insight that the GLUT8-knockdown recapitulates a more normo-physiological liver condition, gaining prominence the two well-established glucose and fructose transporters.

Figure 3

Liver GLUT8 silencing reduces steatosis and fibrosis

C57BL/6 female mice received a tail vein injection of 1x106 TU/mL lentiviral particles to inhibit Slc2A8 (shG8-Fruct) or to use as control (Cont-Fruct), and fed with 60% Fructose diet for 22 weeks (n = 7–8 per group).

(A–E) (A) Specific silencing of liver GLUT8 was tested by immunohistochemistry GLUT8 protein expression (scale bar: 20 μm). Liver FA accumulation was evaluated by (B) H&E staining (scale bar: 50 μm), (C) liver weight (G), (D) TG content in liver and (E) Oil Red staining (scale bar: 100 μm).

(F and G) Circulating levels of (F) TG and (G) cholesterol (mg/dL).

(H and I) Liver fibrosis was evaluated in terms of (H) genetic expression of key biomarkers (Timp1, Col1a1, Col1a2, and Col3) and (I) Sirius red staining taking the portal vein into account. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001P, ∗∗∗∗p ≤ 0.0001; unpaired t test or Mann-Whitney test.

Under fructose diet, the hepatic GLUT8 knockdown reduced lipid content, as shown by the lack of white lipid droplets in liver slides stained with H&E (Figure 3B), lower liver weight (Figure 3B), decreased liver TG (Figure 3D) and significant reduction of Oil Red staining (Figure 3E); and further supported by serum TG (Figure 3F) and reduced cholesterol levels (Figure 3G). These results demonstrated a dramatic reduction in FA stores. Unexpectedly, we also observed a decreased in genes involved in fibrosis (Figure 3H), supported by Sirius staining (Figure 3I).

Omega oxidation is upregulated by the inhibition of liver GLUT8

To establish a metabolic signature and to understand the mechanisms underlying the role of GLUT8 in NAFLD development, we performed a quantitative proteomic analysis using SWATH technology. Moreover, this analysis took account of an enriched-membrane and cytosolic protein fractions. Venn diagrams represent common and unique proteins in Cont-Fruct and shG8-Fruct livers (Figure 4A). We focused our analysis on proteins upregulated in the following processes: (1) lipid metabolism, (2) endoplasmic reticulum-associated protein degradation (ERAD), (3) cytochrome P450 superfamily, (4) oxidative stress, and (5) fibrosis (Table 1, and Figure 4B).

Figure 4

Under the high-fructose diet, liver GLUT8 knockdown promotes the omega oxidation and lipolysis

C57BL/6 female mice received a tail vein injection of 1x106 TU/mL lentiviral particles to inhibit Slc2A8 (shG8-Fruct) or to use as control (Cont-Fruct), and fed with 60% Fructose diet for 22 weeks (n = 7–8 per group).

(A) The Venn diagram illustrates common and unique proteins in control (Cont-Fruct) and liver GLUT8 knockdown (shG8-Fruct) in both cytosol and membrane enriched fractions.

(B) Volcano plot of quantitative proteomic data from cytosol and membrane fractions. Volcano plots are depicted with the fold change and p value calculated by t test. The averages of Cont-Fruct group (n = 4) were compared with the average of the data from shG8 group (n = 4).

(C–F) (C) Lipid metabolism was evaluated by protein expression of FAS, ACC, and CPT1A, and by (D) gene expression of beta-oxidation (Cpt1a, Cpt2) markers. Gene expression of essential components of (E) peroxisomal and (F) omega oxidation.

(G) Regulation of the epoxygenase P450 pathway in shG8-Fruct group compared to Cont-Fruct.

(H) The liver lipolysis was evaluated by ATGL protein expression.

(I) The oxidative stress was assayed by the gene expression of Alcat1 and Nrf2. Results are expressed as mean ± SEM. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001; unpaired t test or Mann-Whitney test. Black line represents cropped blots.

Table 1

Proteins upregulated in control-fructose (fold change >1) and upregulated in shG8-Fructose (fold-change < 1) in the cytosol and membrane isolated fractions

Class		Protein symbol	Uniprot ID	Protein name	Fold change	p value
Cytosol	Lipid metabolism	APOA1	Q00623	Apolipoprotein A-I	1.41	1.95E-02
		HMCS2	P54869	Hydroxymethylglutaryl-CoA synthase	0.73	8.93E-03
	ERAD	PSA5	Q9Z2U1	Proteasome subunit alpha type-5	2.83	1.58E-05
		PSMD1	Q3TXS7	26S proteasome non-ATPase regulatory subunit 1	1.56	2.24E-02
		PSME2	P97372	Proteasome activator complex subunit 2	1.49	7.61E-03
		PSB8	P28063	Proteasome subunit beta type-8	1.47	1.67E-02
		PSA7	Q9Z2U0	Proteasome subunit alpha type-7	1.34	4.73E-02
		PSA1	Q9R1P4	Proteasome subunit alpha type-1	1.24	4.41E-02
		HS90B	P11499	Heat shock protein HSP 90-beta	1.18	2.06E-03
		HS90A	P07901	Heat shock protein HSP 90-alpha	1.13	7.46E-03
		BAP31	Q61335	B-cell receptor-associated protein 31	0.00	4.26E-02
	Cytochrome P450	CP2E1	Q05421	Cytochrome P450 2E1	0.86	8.14E-03
		CP4AE	O35728	Cytochrome P450 4A14	0.74	3.55E-02
	Fibosis	FIBG	Q8VCM7	Fibrinogen gamma chain	1.20	3.27E-02
Membrane	Lipid metabolism	CACP	P47934	Carnitine O-acetyltransferase	3.98	1.30E-02
		ACOT3	Q9QYR7	Acyl-coenzyme A thioesterase 3	1.59	1.63E-02
		CYB5	P56395	Cytochrome b5	1.32	1.19E-02
		DHB12	O70503	Very-long-chain 3-oxoacyl-CoA reductase	1.16	9.45E-03
		FABPL	P12710	Fatty acid-binding protein, liver	0.85	2.25E-02
		SGPL1	Q8R0X7	Sphingosine-1-phosphate lyase	0.40	4.93E-02
		AAKG1	O54950	5'-AMP-activated protein kinase subunit gamma-1	0.00	4.64E-02
	Cytochrome P450	CP3AP	O09158	Cytochrome P450 3A25	0.89	2.84E-02
		CP2J5	O54749	Cytochrome P450 2J5	0.88	1.47E-02
	Oxidative stress	PRDX3	P20108	Thioredoxin-dependent peroxide reductase	1.91	1.28E-02
		NU3M	P03899	NADH-ubiquinone oxidoreductase chain 3	0.80	3.81E-02
	Fibosis	FIBB	Q8K0E8	Fibrinogen beta chain	1.62	2.03E-04

Significantly upregulated proteins in Cont-Fruct mice (fold-change > 1) included: carnitine O-acetyltransferase and acyl-coenzyme A thioesterase 3, which regulates intracellular levels of acyl-CoA; cytochrome b5, which in turn contributes to the sterol biosynthetic pathway; and very-long-chain 3-oxoacyl-CoA reductase, which participates in the production of VLCFAs. These findings, alongside supplementary data concerning biological processes involved in lipid metabolism (Figure S8), support the idea that de novo lipogenesis (DNL) was upregulated in Cont-Fruct mice compared to shG8-Fruct mice. Proteasome subunits also prevailed in the Cont-Fruct group (Table1), suggesting an enhanced activation of the ERAD system. Fibrinogen gamma and beta were also upregulated in this group (Table1), in agreement with Sirius staining (Figure 3I). In contrast, significantly upregulated proteins in shG8-Fruct mice (fold-change < 1) comprised cytochromes P450 monooxygenases involved in the metabolism of FAs, including cytochrome P450 2E1 and 4A14, which hydroxylate FAs specifically at the omega-1-position and display the highest catalytic activity for saturated FAs; or cytochrome P450 3A25 that oxidizes a variety of unrelated compounds, such as steroids and FAs; and cytochrome P450 2J5 with oxidoreductase activity (Table 1, and Figure 4B). While the main mechanism linked to beta-oxidation such as CPT1A was downregulated (Figures 4C and 4D), other members of the cytochrome P450 superfamily, such as Cyp4a10 and Cyp4a14, were also enhanced according to gene expression analysis (Figure 4F). In concordance, a comparative study using a FunRich program denoted that proteins involved in the epoxygenase P450 pathway were highly upregulated in shG8-Fruct mice (Figure 4G). Interestingly, when we checked genes involved in the mitochondrial beta-oxidation, such as Cpt1a and carnitine palmitoyltransferase 2 (Cpt2), or peroxisomal oxidation of VLCFAs mediated by Acyl-coenzyme A oxidase-1(Acox1) and ATP binding cassette subfamily D member 1 (Abcd1), we did not observe any significant differences after liver GLUT8 knockdown (Figures 4D and 4E). We also analyzed lipolysis and observed that ATGL protein levels were significantly increased under the inhibition of liver GLUT8 (Figure 4H). These results indicate major lipid oxidation and mobilization in the absence of liver GLUT8, where omega oxidation seems to plays an important role. In addition, when GLUT8 was silenced in the liver, we observed a general attenuation of glucose metabolism, although there was still gluconeogenesis activation in the membrane fraction (Figure S9).

Hepatic GLUT8 knockdown reduces oxidative stress and decreases the inflammation/inflammasome linked to pIRE over activation

Analysis of the oxidative stress process using SWATH analysis revealed that the NADH-ubiquinone oxidoreductase chain 3 protein (NAD3), which mediates the transfer of electrons from NADH to the respiratory chain was upregulated in shG8-Fruct animals, whereas, in Cont-Fruct mice upregulation of the mitochondrial thioredoxin-dependent peroxide reductase (PRDX3), which is involved in cell protection against oxidative stress (Figures 4B and Table 1), confirmed the presence of fructose-induced oxidative stress and mitochondrial dysfunction in a Glut8-dependent manner. Further evidence of fructose-induced oxidative stress is shown in the biological processes affected (Figure S10A) and gene expression of oxidative stress markers such as lysocardiolipin acyltransferase 1 (Alcat1) and nuclear factor erythroid 2-related factor 2 (Nrf2) in the livers of Cont-Fruct vs shG8-Fruct mice. Alcat1(usually upregulated in mouse models of NAFLD) and Nrf2 (that orchestrates an antioxidant response) were downregulated in shG8-Fruct mice (Figure 4I).

Both, oxidative stress and ER stress contribute to the process of liver inflammation. In fact, the data obtained indicate that fructose pIRE-induced pathological ER stress switches to an adaptative/recovery ER stress-mediated by ATF6 (Figure 5A). Interestingly, the examination of the ER stress markers after liver GLUT8 silencing show that, unlike pIRE and CHOP, proteins such as ATF6 and BIP were markedly upregulated (Figures 5B and 5C). These results were supported by FunRich comparative analysis indicating the contribution of the main biological processes in Cont-Fruct and shG8-Fruct involved in ER stress (Figure S10B).

Next, we evaluated the liver inflammatory response by mRNA expression of IL6, EGF-like module-containing mucin-like hormone receptor-like-1 (F4/80), Interleukin 15 receptor, tumor necrosis factor-alpha, and tumor growth factor-beta. We observed that fructose-induced inflammation was significantly reduced under the lack of GLUT8 (Figure 5D). Similarly, ablation of liver GLUT8 reduced the expression of inflammasome markers such as IL1b, NOD-, LRR- and pyrin domain-containing protein 3 (Nlrp3), the adapter protein Asc (apoptosis-associated speck-like protein containing a CARD), and Interleukin 18 (IL18) (Figure 5E). In addition, protein levels of cleaved Caspase1, an enzyme that proteolytically cleaves precursors of the inflammatory cytokines IL1b and IL18 were significantly reduced (Figure 5F). Accordingly, we also observed that pCREB, which has anti-inflammatory functions (Wen et al., 2010), was enhanced (Figure 5G).

The inhibition of GLUT8 affects human LX2 and THLE2 cells differently

There are four major liver cell types which spatiotemporally cooperate to shape and maintain liver functions (HCs, hepatic stellate cells (HSCs), Kupffer cells, and liver sinusoidal endothelial cells. Since HCs comprise 55-65% of the liver's mass, and HSCs have a remarkable range of functions in normal and injured liver, so we explored the cell type-specific roles of GLUT8 in the liver by examining GLUT8 expression and its impact on cellular function impact on HCs and HSCs using established hepatic cells lines cultures (AML12 mouse HCs, human LX2 HSCs, and human THLE2 HCs). Immunohistochemistry confirmed the presence of GLUT8 in AML12 and LX2 cells (Figure 6A), so then we analyzed GLUT8 mRNA and protein expression in LX2 and AML12 cells under different media mimicking the SD (physiologycal glucose concentration), fructose-enriched or HFD-like conditions (Figure 6B). Fructose-enriched medium upregulated GLUT8 expression in LX2 cells, whereas in AML12 cells GLUT8 was downregulated (Figures 6C and 6D). The inflammation markers induced by fructose in the in vivo studies are also significantly upregulated in LX2 cells (Figure 6E). Moreover, the expression of the genes of the intercellular adhesion molecule 1 (ICAM1) and the disintegrin and metalloproteinase 17 (ADAM17) involved in inflammation and well-known markers of HSCs activation was also elevated (Figure 6F). LX2 cells showed an unexpected intracellular GLUT8 localization pattern, so an as-yet-unknown intracellular role of GLUT8 in hepatic cells cannot be ruled out. Indeed, parallel studies point to the presence of GLUT8 in the late endosomal/lysosomal compartments (Figure S11) as has been demonstrated in other tissues, such as the testis (Diril et al., 2009). Next, we used human siRNA to silence GLUT8 in LX2 and human HCs THLE2 cells and looked at gene expression of DNL, lipolysis, oxidation, and inflammation markers after 24hr fructose exposure (Figure 7A). The efficiency of GLUT8 silencing was around 50% in LX2 (Figures 7B) and 75% in THLE2 (Figure 7E) cells. GLUT8 knockdown in LX2 cells significantly decreased IL6 gene expression (Figure 7B), indicating a reduction of inflammatory signals (Figure 7C). Interestingly, whilst the human cytochrome P450 monooxygenase CYP4F2, which predominantly catalyzes the omega-oxidation of long Fas and VLCFAs, was significantly reduced in GLUT8 silenced LX2 cells; CYPA11 and CYP4F3, which are involved in the metabolism of various endogenous substrates, including FAs and their oxygenated derivatives (oxylipins), were upregulated (Figure 7D). In THLE2 cells, GLUT8 knockdown cells significantly decreased IL1b and IL18 gene expression (Figure 7F), indicating that the activation of the inflammasome response was reduced. Importantly, the gene expression of the three cytochrome P450 monooxygenases analyzed was downregulated (Figure 7G). GLUT8 silencing mainly reduced fructose-induced DNL in THLE2 cells, since the fatty acid synthase (FASN) gene expression and FAS and ACC protein levels were downregulated (Figures 7H and 7I). Enhanced beta-oxidation mediated by CPT1A and lipolysis by ATGL was only observed in LX2 cells (Figure 7J). Altogether, these results indicate that the inhibition of GLUT8 differently reduces the FA content in both LX2 and THLE2, as confirmed by Oil Red staining (Figure 7K), probably due to reduced lipogenesis in HCs and enhanced lipid oxidation and lipolysis in stellate cells.

Figure 6

GLUT8 is differentially regulated by fructose in LX2 and AML12 cells

Fructose induces inflammation in both cell types.

(A) GLUT8 expression assayed by immunohistochemistry in human stellate cells (LX2) and mouse hepatocytes (AML12) (n = 3) and their negative controls (neg cont.) in basal media (scale bar: 20 μm).

(B) Schematic depiction of 24hr cells treatments with different nutrients.

(C and D) Regulation of Glut8 mRNA under basal (SD), Fructose (Fruct.), and high-fat (HFD) media; (C) quantification of mRNA and (D) representative images of immunohistochemistry.

(E) Genetic expression in hepatic stellate cells (LX2) and hepatocytes (AML12) of the main cytokines involved in the inflammatory response (IL6) and inflammasome (IL1b) activation.

(F) Genetic expression in LX2 of other inflammatory markers, ICAM1 and ADAM7. Results are expressed as mean ± SEM, ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001; One-Way ANOVA or Kruskall-Wallis test following by a post-hoc test (∗ compared to SD, # compared to 60% Fructose).

Figure 7

GLUT8 silencing in hepatic human cells in fructose media reduces DNL in THLE2 hepatocytes, increases oxidation/lipolysis in LX2 stellated cells and ameliorates inflammation in both

(A) Schematic representation of the experimental process.

(B–-D) Impact of GLUT8 silencing in LX2 cells and its effects on (C) mRNA expression of inflammatory-related genes, and (D) human cytochrome P450-related omega-oxidation, after 24hr incubation in fructose media (n = 3 independent experiments conducted by triplicate).

(E–G) Impact of GLUT8 silencing in THLE2 cells and its effects on (F) inflammation markers, and (G) expression of cytochrome P450 monooxygenases involved in the metabolism of FAs after 24hr fructose enriched media (n = 3 independent experiments conducted by triplicate).

(H–J) Lipid metabolism in LX2 and THLE2 cells after GLUT8 knockdown assayed by (H) gene expression of FASN and protein expression of (I) lipogenic (FAS, ACC) and (J) lipolytic (CPT1A, ATGL) factors. Representative blots are shown (bellow J).

(K) Measure of lipid content in LX2 and THLE2 cells by Oil Red staining (scale bar: 20 μm). Results are expressed as mean ± SEM. ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001P; unpaired t test or Mann-Whitney test.

Discussion

Within the heterogeneity in NAFLD risk profiles, the diet consumed and differences in how it is metabolized constitute a key factor for the accurate identification of high-risk individuals and personalized preventive/therapeutic strategies. Female subjects seem to be more sensitive to fructose metabolism, hence one of these strategies may be the reduction of liver fructose intake achieved by sugar facilitated transporters (GLUT family) across the cell membranes. Fructose metabolism involves its conversion into glucose and studies have showed that elevated fructose consumption is more cytotoxic than glucose (Ter Horst and Serlie, 2017). This conversion is only partly carried out in intestinal cells, thus, at higher luminal concentrations, fructose is metabolized in the liver. There, fructolysis bypasses the step using glucokinase and is much faster than glycolysis, thus providing increased substrate for all central carbon metabolic pathways, including glycolysis, glycogenesis, gluconeogenesis, lipogenesis, pentose phosphate pathway, and oxidative phosphorylation (Hannou et al., 2018). On the other hand, fructose rather than glucose induces barrier deterioration of the intestinal epithelial cells (IECs). Moreover, fructose metabolites trigger ER stress and inflammation in both in IECs and in HCs. Noteworthy ER-stress-activated IRE1 and TNF signaling stimulate hepatosteatosis (Todoric et al., 2020).

Besides glucose transporter 2 (GLUT2) and glucose transporter 5 (GLUT5), a previous study based on a germ-line GLUT8 knock-out mouse model recognized GLUT8 as a liver fructose transporter that mediates fructose-induced DNL and macrosteatosis in female mice (Debosch et al., 2014). In our study, we achieved a liver-targeted GLUT8 silencing in order to understand better the role of this new player in NAFLD progression. Our results, in a fructose-induced female NASH model, demonstrated a surprising recovery of the inflammatory/ER stress status in liver GLUT8-knockdown animals. We observed a reduced liver weight, decreased hepatic FA, plasma cholesterol levels, and, curiously, improvement of the fibrotic state. These effects could be explained by (1) the reduction of DNL, (2) the increased FA oxidation, or (3) maybe both. In our experimental model, we have not observed significant changes in FAS and ACC expression despite the reduction in liver steatosis. Nevertheless, after GLUT8 silencing, the inflammation process decreased markedly. In agreement with recent data, this decrease could explain the reduction in liver steatosis present in GLUT8-knockdown mice (Todoric et al., 2020). Additionally, an upregulation of the cytochrome P450 epoxygenase pathway and increased ATGL activity indicate major oxidation and lipolysis of accumulated FAs.

It is known that malonyl-CoA generated via DNL limits FA oxidation by inhibiting CPT1A, the enzyme required for translocation of long-chain fatty acids into the mitochondria (Hannou et al., 2018). However, when beta-oxidation is impaired due to fructose-induced defects in CPT1A mechanisms, the silencing of liver GLUT8 promotes a subsidiary pathway for maximal fat oxidation: omega oxidation (Miura, 2013). This unusual FA oxidation is carried out by the liver, kidney and lung, as demonstrated in microsomal preparations from mice and other species (Ichiha et al., 1969). The proposed biological significance of omega-oxidation is the generation of more soluble acids that are easily secreted into the blood and finally excreted in the urine (Miura, 2013). Mortensen et al. (Mortensen, 1980) demonstrated that omega oxidation of FAs might have important metabolic influence in situations where living organisms lacked carbohydrates and largely have to utilize fats for energy demand. Despite the idea that omega oxidation catalyzed by Cyp2E1, Cyp 4A10, and Cyp 4A14 constitutes a mechanism of lipid-induced cellular injury in NAFLD due to formation of ROS species (Browning and Horton, 2004), many other studies consider FA omega oxidation a rescue pathway for FA oxidation disorders in humans (Wanders et al., 2011). Moreover, the cytochrome P450 epoxygenase pathway regulates hepatic inflammatory response in fatty liver disease (Schuck et al., 2014). At least, in our female fructose model, the upregulation of the murine cytochrome P450 epoxygenase pathway seems to have a protective effect against liver steatosis.

Importantly, steps in DNL and very-low-density lipoprotein synthesis occur at the ER membrane; thus, fructose-induced lipogenesis may elicit an ER stress response that contributes to NAFLD pathogenesis and progression (Malhi and Kaufman, 2011). As a vital organ for protein synthesis and detoxification, the liver is especially susceptible to ER stress. The signaling initiated by IRE and other UPR pathways can restore homeostasis, however, prolonged or unresolved ER stress leads to inflammation via the sustained interaction of IRE and TNF receptor-associated factor 2 (TRAF2), leading to NFkB activation (Guo and Li, 2014). As a consequence, IRE is a key partner in the ER stress-induced inflammasome and inflammasome-independent inflammation mechanisms. Indeed, in our study, the exacerbated fructose-induced activation of IRE was accompanied by increased IL1b and IL6 gene expression. Unexpectedly, after inhibition of GLUT8 in the liver we observed ATF6 activation and enhanced BIP protein levels. Although ER stress pathways are usually associated with their pathogenic effects, this result is not unusual taking the protective role of UPR activation to maintain homeostasis into account (Cinaroglu et al., 2011). In fact, in murine models, it has been shown that the activation of the ATF6 pathway protects against hepatic steatosis, via activation of PPAR-alpha stimulating FA oxidation under high-fat, high-sucrose diet (Chen et al., 2016). Moreover, glucose deprivation activates ATF6 but suppresses the SREBP-2 regulated transcription and, consequently, its lipogenic effect (Zeng et al., 2004).

One of the unresolved issues from our in vivo study relates to the liver cell types involved in the fructose-induced liver injury and its prevention by silencing GLUT8. GLUT8 is expressed in both, HCs and HSCs, and their intracellular localization is not surprising since it coincides with previous findings in other cell types (Schmidt et al., 2009). Our in vitro studies support that the absence of GLUT8 impacts lipid metabolism in human HCs and stellate cells differentially: reducing lipogenesis in THLE2 cells while promoting FA oxidation in LX2 cells, respectively. This dual effect may explain the potent diminution of fat content observed in the liver.

In summary and based on our data we hypothesize that the lack of GLUT8 in liver mimics a cellular glucose/fructose deprivation state, which in turn modulates lipogenesis and promotes FA oxidation as a source of energy. Limited carbohydrates availability in liver cells may be achieved by 1) reduced fructose uptake in the plasma membrane of HCs as previously described, but importantly also by (2) the impaired intracellular import/export via facilitative transporters such as GLUT8. Since beta-oxidation mediated by CPT1A mechanisms is compromised under the fructose diet, and alternative pathways, omega-oxidation and lipolysis, are used to breaking fat. Beyond lipid metabolism, fructose-induced pathogenic ER stress, and inflammation/inflammasome are attenuated in the absence of GLUT8, preventing the first steps of NAFLD development. Whilst HCs are mainly involved in relation to lipogenesis, stellate cells GLUT8 appears mainly to regulate lipid oxidation; however, our data indicate that both cell types are involved in terms of the inflammatory response. Interestingly, while rich fructose medium induces a high HSCs activation that over time results in the formation of liver fibrosis (Koyama and Brenner, 2017), GLUT8 silencing reduces ACC expression in LX2 cells that could suppress the activation of these cells and thus the fibrosis process (Bates et al., 2020).

This study may serve as a warning against excessive consumption of sugar-sweetened beverages and manufactured food products containing high-fructose corn syrup; as this may lead to potential pathophysiological consequences, especially in females due to greater susceptibility to fructose uptake and metabolism by the liver. In addition, this study provides a drug target with the potential to prevent fructose-induced liver injury and raises the question whether, given the high energy demand for the proliferation of liver tumor cells, restriction of carbohydrate supply mediated by GLUT8 and, in turn, diminution of fat sources, may slow down HCC progression.

Limitations of the study

Our study contributes to a better understanding of the role of GLUT8 in the development of NAFLD in the context of high amounts of fructose diet intake. Despite liver-specific knockdown of GLUT8 improves liver fat deposition, lipotoxicity, and inflammation in female mice, we have not provided potential mechanisms to explain why female mice seem to be more responsive to high fructose diet than male mice. Future studies should also include a model in male mice to address possible discrepancies between fructose consumption and sex-specific effects. Moreover, the logical approach will be to compare male and females with/without gonadectomy and with/without replacement therapy in order to uncover the molecular underpinning of these sex differences. Similarly, a high fructose diet is only one of the diet-induced NAFLD animal models and does not recapitulate all NAFLD human features. It would be interesting to elucidate in further studies using an American lifestyle-induced obesity syndrome or cafeteria diets, which are more representative of human habits, whether the beneficial effects of hepatic GLUT8 silencing can be replicated. Of note, consideration of the deleterious effects of fructose intake, including the gender-related one, also needs to be refined taking into account that the end-alterations in liver function are multifactorial and the impact of fructose may be related to other factors, such as age, previous or concurrent diet, timing, dosage, and also dependent of other co-morbidities. In order to extrapolate the conclusions of this study, all these limitations should be considered.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact Amparo Romero-Picó (amparo.romero@usc.es).

Materials availability

This study did not generate new unique reagents.

Data and code availability

All data supporting the current study are available from the corresponding author on request.

Methods

All methods can be found in the accompanying Transparent methods supplemental file.