Biliary elimination of cholesterol can be modulated by hepatocyte mitochondrial Aquaporin-8 in mice

TL;DR

This study shows that a protein called AQP8 in liver cells can influence how cholesterol is removed from the body through bile in mice.

Contribution

The study reveals a new role for mitochondrial AQP8 in modulating biliary cholesterol excretion via regulation of ABCG5 and SREBP-2.

Findings

Knockdown of mtAQP8 reduces SREBP-2 and ABCG5 expression, decreasing biliary cholesterol excretion.

Expression of hAQP8 increases SREBP-2, LXR, and ABCG5, enhancing biliary cholesterol excretion.

Mitochondrial antioxidants prevent upregulation of SREBP-2 and ABCG5, blocking increased cholesterol excretion.

Abstract

Sterol regulatory element-binding protein (SREBP) transcription factors directly or indirectly regulate key genes involved in hepatic cholesterol homeostasis, including biliary elimination. The ATP-binding cassette transporter G5 (ABCG5), located in hepatocyte canalicular plasma membranes, strongly controls the excretion of unesterified cholesterol into bile. Recently, we demonstrated in cultured hepatocytes that mitochondrial aquaporin-8 (mtAQP8), a channel protein capable of conducting H2O2, is involved in SREBP-controlled cholesterol synthesis. In this study, we evaluated whether hepatic mtAQP8 participates in modulating the biliary elimination of cholesterol. Using C57BL/6 mice, we found that adenovirus-induced mtAQP8 knockdown significantly downregulated the expression of sterol regulatory element-binding protein-2 (SREBP-2) and, through liver X receptor (LXR), that of ABCG5, which in turn decreased biliary cholesterol excretion. In contrast, mice with adenovirus-mediated mitochondrial human AQP8 (hAQP8) expression significantly increased the expressions of SREBP-2, LXR, and ABCG5 and, consequently, the biliary excretion of cholesterol. A mitochondrial-targeted antioxidant, which has been shown to quench mitochondrial H2O2 release, did not affect mitochondrial hAQP8 expression, but prevented upregulation of SREBP-2, ABCG5, and biliary cholesterol excretion. Our data further support the involvement of mtAQP8 in hepatic cholesterol metabolism, suggesting that it modulates biliary cholesterol elimination through ABCG5 gene expression.

The online version contains supplementary material available at 10.1038/s41598-026-39058-6.

Figures

Knockdown of hepatocyte mtAQP8 in mice transduced with AdAQP8sh. The delivery of AdAQP8sh and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for AQP8 in liver mitochondrial fraction, with corresponding densitometric analysis. A 28 kDa immunoreactive band corresponding to endogenous mitochondrial mouse AQP8 is observed. Each lane was loaded with 25 µg of protein. Prohibitin, an inner mitochondrial membrane marker, is shown as the control for equal protein loading. Lanes were cropped from non-ad

Hepatic expression of SREBP-2 and LXR in mtAQP8-knockdown mice. The delivery of AdAQP8sh and control vectors was detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. Total lysates were prepared and subjected to immunoblotting as described in Materials and Methods. (A) Representative immunoblot for SREBP-2 in total lysates (90 µg protein/lane) and densitometric analysis. Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. (B) Representative immunoblo

Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in mtAQP8-knockdown mice. The delivery of AdAQP8sh and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for ABCG5 in total lysates (90 µg protein/lane) and canalicular plasma membranes (35 µg proteins/lane). Anti-β-actin antibody was used as a control for equal protein loading. Membranes were cut prior to antibody hybridisation based on molecular weight ranges. Lanes were cropped from non-adjacent regions of the

Hepatic expression of human mtAQP8 in mice transduced with AdhAQP8. The delivery of AdhAQP8 and control vectors was detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for hAQP8 in liver mitochondrial fraction. A 28 kDa immunoreactive band corresponding to endogenous mitochondrial mice AQP8 (weak staining) and human AQP8 is observed. The AQP8 antibody used is capable of detecting both mice and human isoforms but with higher affinity for the latter^[11](#CR11),[17](#CR17)^, accordingly, densitometric analysis was not pe

Hepatic expression of SREBP-2 and LXR in AdhAQP8-transduced mice. The delivery of AdhAQP8 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. Total lysates were prepared and subjected to immunoblotting as described in Materials and Methods. (A) Representative immunoblot for SREBP-2 in total lysates (90 µg protein/lane) and densitometric analysis. Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. (B) Representative immunob

Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in AdhAQP8-transduced mice. The delivery of AdhAQP8 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for ABCG5 in total lysates (90 µg protein/lane) and canalicular plasma membranes (35 µg protein/lane). Anti-β-actin antibody was used as a control for equal protein loading. For total lysate, the blot was reprobed by using anti-β-actin antibody and for canalicular plasma membrane, membrane was cut prior to an

Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in AdhAQP1-transduced mice. The delivery of AdhAQP1 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for hAQP1 in liver mitochondrial fraction. hAQP1-expressing hepatic plasma membranes from AdhAQP1 transduced-mice were used as a positive control. As expected, mouse mitochondria did not express AQP1, nor did the AdhAQP1 gene transfer induce mitochondrial expression of this aquaporin. The very week band obser

Effect of the mitochondria-targeted antioxidant MitoTempo. AdhAQP8-transduced mice were treated or not with MitoTempo, and the expression of hAQP8, SREBP-2 and ABCG5, and biliary cholesterol excretion were evaluated. (A) Representative immunoblot for hAQP8 in liver mitochondrial fraction, with corresponding densitometric analysis (*n* = 4). A 28 kDa immunoreactive band corresponding to human AQP8 is observed. Each lane was loaded with 25 µg of protein. Prohibitin, an inner mitochondrial membrane marker, is shown as the control for equal protein loading. (B) Representative immunoblot for SREBP-

Full text

Introduction

Cholesterol is a fundamental component of the cell membrane and plays a role in the production of bile acids, vitamin D and steroid hormones. To maintain cholesterol homeostasis, the amount of cholesterol acquired by cells through de novo synthesis or lipoprotein uptake must be tightly coupled to the amount that is lost through degradation or excretion^1^.

Because cholesterol is virtually insoluble in aqueous solutions, its solubilization in plasma and bile requires specialized transport mechanisms. In plasma, cholesterol is mainly carried by lipoproteins, whereas in bile it is solubilized within mixed micelles and vesicles. Excess cholesterol accumulation in the arterial wall promotes atherosclerosis and cardiovascular disease, while impaired biliary solubilization by bile salts and/or phospholipids results in cholesterol monohydrate crystal precipitation and gallstone formation^2^.

In the liver, biliary cholesterol secretion is the major route for cholesterol elimination and a key determinant of bile composition. Its dysregulation plays a central role in cholesterol gallstone disease, a highly prevalent disorder driven by alterations in hepatic cholesterol metabolism and biliary lipid transport, highlighting biliary cholesterol excretion as a potential therapeutic target beyond surgery^2,3^.

The ATP-binding cassette transporter G5 (ABCG5), as part of the ABCG5/ ATP-binding cassette transporter G8 (ABCG8) heterodimer, tightly controls hepatobiliary cholesterol secretion in mice^4^. Thus, canalicular ABCG5 seems to be responsible for around 82% of hepatic cholesterol excretion into bile^5^.

Alterations in normal hepatic ABCG5 expression and function can cause cholesterol related disorders. Thus, loss-of-function mutations in ABCG5 cause sitosterolemia, i.e., decreased biliary secretion of plant and animal sterols and sitosterol accumulation in the blood, as well as hypercholesterolemia^1,2^. Conversely, ABCG5 overexpression promotes the formation of cholesterol gallstones^2,5^. This underscores the need for strict control of ABCG5 expression and function in the liver.

The sterol regulatory element-binding protein-2 (SREBP-2), a transcription factor that regulates key genes involved in hepatic cholesterol homeostasis, has been shown to induce hepatic Abcg5 expression in transgenic mice, suggesting a role in modulating sterol excretion pathways^6^. In parallel, Liver X Receptors (LXRs) are nuclear receptors that play a central role in sensing intracellular cholesterol levels and activating the transcription of genes involved in its efflux and metabolism^7^, including Abcg5 and Abcg8, both in vitro and in vivo^8^. The expression of these transporters can be upregulated by sterol metabolites such as oxysterols, which serve as endogenous LXRs ligands^9^. In hepatocytes, oxysterol-dependent regulation of ABCG5 is primarily mediated by LXRα, the predominant hepatic LXR isoform^8^. Importantly, the increased cholesterol biosynthesis resulting from SREBP-2 overexpression can generate oxysterols, which can serve as endogenous LXR ligands thereby providing endogenous activators of the LXR pathway^6^. This suggests a mechanistic link between SREBP-2-mediated cholesterol synthesis and LXR-driven cholesterol excretion pathways^10^.

Despite these advances, the contribution of mitochondrial signaling mechanisms to the regulation of biliary cholesterol secretion remains poorly understood.

Aquaporin-8 (AQP8) is a multifunctional channel protein expressed in various cellular membranes, including the inner mitochondrial membrane of hepatocytes. Specifically, the mitochondrial isoform of AQP8 functions as a peroxiporin, facilitating the efflux of hydrogen peroxide (H_2_O_2_) from mitochondria^11–14^. We recently provided experimental evidence, using cultured hepatocytes, suggesting that AQP8 is a cholesterol-responsive SREBP-2 target gene^15^, and that mitochondrial AQP8 (mtAQP8), via H_2_O_2_, is involved in SREBP-2-controlled cholesterol synthesis^16^. In this study, using adenovirus-transduced mice, we evaluated whether hepatic mtAQP8 participates in modulating the biliary elimination of cholesterol.

Results

Biliary cholesterol excretion is decreased in mtAQP8 knockdown mice

Mice were transduced with an adenovector encoding a short hairpin RNA targeting murine Aqp8 mRNA (AdAQP8sh), or control adenovector (see Materials and Methods for details). After 72 h, the downregulation of hepatocyte mtAQP8 expression was confirmed by liver subcellular fractionation and immunoblotting. Figure 1A shows a band of expected size at around 28 kDa, corresponding to endogenous mouse mtAQP8, which is decreased by about 60% in AdhQP8sh-transduced mice. As shown in Fig. 1B, AQP8 mRNA expression was also significantly downregulated by around 60% in AdhQP8sh-transduced mice.

Fig. 1. Knockdown of hepatocyte mtAQP8 in mice transduced with AdAQP8sh. The delivery of AdAQP8sh and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for AQP8 in liver mitochondrial fraction, with corresponding densitometric analysis. A 28 kDa immunoreactive band corresponding to endogenous mitochondrial mouse AQP8 is observed. Each lane was loaded with 25 µg of protein. Prohibitin, an inner mitochondrial membrane marker, is shown as the control for equal protein loading. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Data (means ± SEM) are expressed as percentage of controls, n = 3. *P < 0.05 from controls. (B) Messenger RNA expression of mouse AQP8 assessed by real-time PCR in AdAQP8sh transduced mice. Data (means ± SEM) were normalized to housekeeping gene, Hypoxanthine Phosphoribosyltransferase1 (HPRT1), n = 3–7. *P < 0.05 from controls.

SREBP-2 transcription factor regulates key genes involved in hepatic cholesterol homeostasis^6^. Another key cholesterol-related transcription factor, the LXR, is involved in the regulation of cholesterol homeostasis at various levels, including biliary elimination^8^. As shown in Fig. 2, the hepatic expression of SREBP-2 in mtAQP8 knockdown mice was significantly decreased. LXR protein abundance was also significantly decreased, in parallel with reduced expression of the LXR target gene ABCG5. The canalicular cholesterol transporter ABCG5 was markedly downregulated by 70–90% in mtAQP8 knockdown mice, both in total lysates and in canalicular plasma membranes (Fig. 3A). Consistently, ABCG5 mRNA levels were reduced by approximately 50% in AdAQP8sh-transduced mice (Fig. 3B). In line with the decreased expression of ABCG5, biliary cholesterol excretion was reduced by 45% in AdAQP8sh-transduced mice compared with control adenovirus–treated mice (Fig. 3C).

Fig. 2. Hepatic expression of SREBP-2 and LXR in mtAQP8-knockdown mice. The delivery of AdAQP8sh and control vectors was detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. Total lysates were prepared and subjected to immunoblotting as described in Materials and Methods. (A) Representative immunoblot for SREBP-2 in total lysates (90 µg protein/lane) and densitometric analysis. Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. (B) Representative immunoblot for LXR in total lysates (100 µg protein/lane) and densitometric analysis. The blots were reprobed by using anti-β-actin antibody as a control for equal protein loading. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Data (means ± SEM) are expressed as percentage of controls, n = 3–4. *P < 0.05 from controls.

Fig. 3. Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in mtAQP8-knockdown mice. The delivery of AdAQP8sh and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for ABCG5 in total lysates (90 µg protein/lane) and canalicular plasma membranes (35 µg proteins/lane). Anti-β-actin antibody was used as a control for equal protein loading. Membranes were cut prior to antibody hybridisation based on molecular weight ranges. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Densitometric analysis of three/four separate experiments in each group. Data (means ± SEM) are expressed as percentage of controls, n = 3–4. *P < 0.05 from controls. (B) Messenger RNA expression of ABCG5 assessed by real-time PCR in AdAQP8sh transduced mice. Data (means ± SEM) were normalized to housekeeping gene, Hypoxanthine Phosphoribosyltransferase1 (HPRT1), n = 4–5. *P < 0.05 from controls. (C) Biliary cholesterol excretion. Bile was collected in 30-minute period, and cholesterol were determined by thin layer chromatography, as detailed in Material and Methods. Data are given as mean ± standard error of four/five independent experiments per group. *P < 0.05 from controls.

Taken together, our data suggest that mtAQP8 knockdown in mice reduces biliary cholesterol excretion by decreasing ABCG5 expression, likely through the mediation of SREBP-2 and LXR.

Biliary cholesterol excretion is induced in mitochondrial hAQP8 expressed mice

Mice were transduced with an adenovector encoding hAQP8, AdhAQP8, or a control adenovector (see Materials and Methods for details). Mitochondrial hAQP8 expression was confirmed by liver subcellular fractionation and immunoblotting. A 28 kDa immunoreactive band corresponding to endogenous mouse mtAQP8 (weak staining) and hAQP8 is observed (Fig. 4A). The antibody used is capable of detecting both mice and human isoforms of AQP8 but with higher affinity for the latter^11,17^, therefore, densitometric analysis was not performed. Figure 4B shows that hAQP8 mRNA was strongly expressed in AdhQP8-transduced mice and was undetectable in control adenovirus–treated mice.

Fig. 4. Hepatic expression of human mtAQP8 in mice transduced with AdhAQP8. The delivery of AdhAQP8 and control vectors was detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for hAQP8 in liver mitochondrial fraction. A 28 kDa immunoreactive band corresponding to endogenous mitochondrial mice AQP8 (weak staining) and human AQP8 is observed. The AQP8 antibody used is capable of detecting both mice and human isoforms but with higher affinity for the latter^11,17^, accordingly, densitometric analysis was not performed. Each lane was loaded with 25 µg of protein. Prohibitin, an inner mitochondrial membrane marker, is shown as the control for equal protein loading. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. (B) Messenger RNA expression of human AQP8 assessed by real-time PCR in AdhAQP8 transduced mice. Data (means ± SEM) were normalized to housekeeping gene, Hypoxanthine Phosphoribosyltransferase1 (HPRT1). Statistical analysis was not performed because, as expected, hAQP8 expression was undetectable in control group.

Protein expression of SREBP-2 (Fig. 5) was markedly upregulated in mitochondrial hAQP8-expressed mice. LXR protein levels were significantly increased, in parallel with changes in the expression of the LXR target gene, ABCG5, in both total lysates and canalicular plasma membranes (Fig. 6A), which was found to be markedly upregulated in AdhAQP8-transduced mice.

Fig. 5. Hepatic expression of SREBP-2 and LXR in AdhAQP8-transduced mice. The delivery of AdhAQP8 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. Total lysates were prepared and subjected to immunoblotting as described in Materials and Methods. (A) Representative immunoblot for SREBP-2 in total lysates (90 µg protein/lane) and densitometric analysis. Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. (B) Representative immunoblot for LXR in total lysates (100 µg protein/lane) and densitometric analysis. The blots were reprobed by using anti-β-actin antibody as a control for equal protein loading. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Data (means ± SEM) are expressed as percentage of controls, n = 3–4. *P < 0.05 from controls.

Fig. 6. Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in AdhAQP8-transduced mice. The delivery of AdhAQP8 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for ABCG5 in total lysates (90 µg protein/lane) and canalicular plasma membranes (35 µg protein/lane). Anti-β-actin antibody was used as a control for equal protein loading. For total lysate, the blot was reprobed by using anti-β-actin antibody and for canalicular plasma membrane, membrane was cut prior to antibody hybridisation based on molecular weight ranges. Densitometric analysis of three/ four separate experiments in each group. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Data (means ± SEM) are expressed as percentage of controls, n = 3–4. *P < 0.05 from controls. (B) Messenger RNA expression of ABCG5 assessed by real-time PCR in AdhAQP8 transduced mice. Data (means ± SEM) were normalized to housekeeping gene, Hypoxanthine Phosphoribosyltransferase1 (HPRT1), n = 4. *P < 0.05 from controls. (C) Biliary cholesterol excretion. Bile was collected in a 30-minute period, and cholesterol were determined by thin layer chromatography, as detailed in Material and Methods. Data are given as mean ± standard error of four independent experiments per group. *P < 0.05 from controls.

Figure 6B shows that ABCG5 mRNA expression was also significantly upregulated by about 100% in AdhAQP8-transduced mice. According to the ABCG5 upregulation, biliary cholesterol excretion was found to be significantly increased by about 100% in mice with mitochondrial hAQP8 expression compared with control adenovirus–treated mice (Fig. 6C).

To test whether expression of a different aquaporin (AQP) can also induce biliary cholesterol excretion, we used the adenovector AdhAQP1 encoding human AQP1 (hAQP1)^18,19^. AQP1 is an archetypal aquaporin with endogenous hepatic expression in peribiliary vascular endothelia and cholangiocytes, but absent in hepatocytes^20,21^. As we previously reported in rats^19^, after AdhAQP1 administration in mice, hAQP1 was mainly expressed in hepatocyte plasma membranes (data not shown) but not in mitochondria (Fig. 7A). Expression of hAQP1 in hepatocytes did not alter either ABCG5 expression (Fig. 7B) or biliary cholesterol excretion (Fig. 7C).

Fig. 7. Hepatic canalicular cholesterol transporter, ABCG5 and biliary cholesterol excretion in AdhAQP1-transduced mice. The delivery of AdhAQP1 and control vectors were detailed in Materials and Methods. The experiments were performed 72 h after adenoviral infusion. (A) Representative immunoblot for hAQP1 in liver mitochondrial fraction. hAQP1-expressing hepatic plasma membranes from AdhAQP1 transduced-mice were used as a positive control. As expected, mouse mitochondria did not express AQP1, nor did the AdhAQP1 gene transfer induce mitochondrial expression of this aquaporin. The very week band observed is due to cross-contamination with membranes expressing hAQP1. Each lane was loaded with 25 µg of protein. Prohibitin was used as a mitochondrial protein loading control. (B) Representative immunoblot for AQP1 and ABCG5 in isolated hepatocyte total lysates (50 µg protein/lane). The blots were reprobed by using anti-β-actin antibody as a control for equal protein loading. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Densitometric analysis of three separate experiments in each group. Data (means ± SEM) are expressed as percentage of controls, n = 3. No significant differences were observed for ABCG5 from controls. (C) Biliary cholesterol excretion. Bile was collected in a 30-minute period, and cholesterol was determined by thin layer chromatography, as detailed in Material and Methods. Data are given as mean ± standard error of six independent experiments per group.

Taken together, our data suggest that mitochondrial hAQP8 expression in mice induces biliary cholesterol excretion by increasing ABCG5 expression, likely via SREBP-2 and LXR.

The mitochondria-targeted antioxidant MitoTempo prevented mtAQP8-induced biliary cholesterol excretion

Since AQP8 can facilitate mitochondrial H_2_O_2_ release and H_2_O_2_ has been described to activate SREBP-2^16,22–24^, we assessed the effect of MitoTempo, a mitochondria-targeted antioxidant, which has been shown to quench mitochondrial H_2_O_2_ release^25^ on mtAQP8-induced biliary cholesterol excretion. In previous observations, we found that in normal cells, MitoTEMPO administration does not alter endogenous mtAQP8 expression, but significantly reduces mitochondrial H_2_O_2_ release^16^. AdhAQP8-transduced mice were treated or not with MitoTempo. As shown in Fig. 8, MitoTempo did not affect mitochondrial hAQP8 expression in AdhAQP8-transduced mice, but significantly prevented the upregulation of SREBP-2 and ABCG5, as well as the increased of biliary cholesterol excretion.

Fig. 8. Effect of the mitochondria-targeted antioxidant MitoTempo. AdhAQP8-transduced mice were treated or not with MitoTempo, and the expression of hAQP8, SREBP-2 and ABCG5, and biliary cholesterol excretion were evaluated. (A) Representative immunoblot for hAQP8 in liver mitochondrial fraction, with corresponding densitometric analysis (n = 4). A 28 kDa immunoreactive band corresponding to human AQP8 is observed. Each lane was loaded with 25 µg of protein. Prohibitin, an inner mitochondrial membrane marker, is shown as the control for equal protein loading. (B) Representative immunoblot for SREBP-2 in total lysates (90 µg protein/lane). Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. Densitometric analysis of four separate experiments in each group. (C) Representative immunoblot for ABCG5 in total lysates (90 µg protein/lane). Anti-β-actin antibody was used as a control for equal protein loading. Membrane was cut prior to antibody hybridisation based on molecular weight ranges. Lanes were cropped from non-adjacent regions of the same gel; grouping is indicated by dotted lines. Full-length, uncropped blots are provided in the Supplementary Information. Densitometric analysis of three separate experiments in each group. Data (means ± SEM) are expressed as percentage of controls, n = 3–4. *P < 0.05 from AdhAQP8 without MitoTempo. (D) Effect of MitoTempo on biliary cholesterol excretion in human mtAQP8 mice. Bile was collected in 30-minute period, and cholesterol were determined by thin layer chromatography. Data are given as mean ± standard error (n = 3–5). Statistical significance was assessed by Student’s t-test comparing AdhAQP8 vs. AdhAQP8 + MitoTempo. *P < 0.05 from AdhAQP8 without MitoTempo.

These data suggest the involvement of mitochondria-derived H_2_O_2_ in the upregulation of biliary cholesterol excretion in mitochondrial hAQP8 expressed mice.

Discussion

The main finding in this study relates to the role of mtAQP8 in the modulation of biliary cholesterol excretion. We provide experimental evidence that (i) mtAQP8 knockdown in mice reduced biliary cholesterol excretion by decreasing ABCG5 expression, through downregulation of SREBP-2 and LXR; (ii) mitochondrial hAQP8 expression in mice induced biliary cholesterol excretion by increasing ABCG5 expression, via upregulation of SREBP-2 and LXR; and (iii) the mitochondria-targeted antioxidant MitoTempo prevented mitochondrial hAQP8-induced increase in biliary cholesterol excretion.

We previously demonstrated, using human and rat hepatocytes, that mtAQP8 is involved in the SREBP-2-controlled cholesterol synthesis^16^. Our present data suggest that mtAQP8 may also modulate biliary cholesterol elimination, further supporting its key role in hepatic cholesterol metabolism.

Hepatic cholesterol homeostasis is tightly regulated by the interplay between SREBP-2 and LXRs^26^. Although hepatic oxysterol levels were not directly assessed in this study, the coordinated changes observed in SREBP-2 expression, LXR protein levels, ABCG5 expression, and biliary cholesterol excretion are consistent with a functional interaction between these pathways, as previously described in the literature^9,10^. Future studies will be required to directly determine whether mtAQP8-dependent modulation of cholesterol metabolism alters hepatic oxysterol production.

LXR activation by oxysterols induces ABCG5 and ABCG8 expressions, promoting biliary cholesterol secretion. These effects are abolished in mice deficient in Abcg5 and Abcg8^8,27^, underscoring the essential role of the pathway in hepatic lipid balance. While LXR transcriptional activity was not directly assessed, the coordinated regulation of ABCG5 mRNA and protein expression, a well-established direct transcriptional target, and biliary cholesterol excretion is consistent with functionally relevant modulation of LXR-dependent signaling under conditions of mtAQP8 modulation.

mtAQP8 functions as a peroxiporin, facilitating the diffusion of H₂O₂ across the mitochondrial membrane^13,16^. Mitochondria-derived H₂O₂ is increasingly recognized as a redox signaling molecule involved in the regulation of hepatic lipid metabolism. Low levels of H₂O₂ have been shown to activate SREBPs in hepatocytes, thereby promoting lipogenesis^22,24^. HepG2 cells and primary hepatocytes treated with H_2_O_2_ significantly increased lipid accumulation compared to untreated HepG2 cells and significantly increased the expression of mRNA for genes related to cholesterol synthesis and uptake^23^. We recently provided experimental evidence that overexpression of mtAQP8 in human hepatocyte-derived cell lines leads to increased mitochondrial H₂O₂ release, elevated SREBP-2 protein expression, and enhanced cholesterol biosynthesis. Furthermore, treatment with a mitochondria-targeted antioxidant in these cells prevented the stimulation of hepatic cholesterol synthesis^16^. These findings suggest that the peroxiporin mtAQP8 via H_2_O_2_, is involved in SREBP-2-controlled cholesterol synthesis. SREBP-2 activation via H_2_O_2_ may occur indirectly through signaling pathways involving mitogen-activated protein kinases (MAPKs)^28^ and further supported by studies on mitochondrial ROS signaling^24,29,30^. In addition, under oxidative stress conditions, H₂O₂ has been implicated in the modulation of LXR activity via AMPK-dependent pathways^31^, although no evidence currently supports a direct activation of LXR by H₂O₂ in normal hepatocytes or in vivo models. Further studies are needed to elucidate the molecular mechanisms linking mitochondrial H₂O₂ to LXR activation, including whether H₂O₂ acts through indirect signaling cascades or directly modifies LXR via oxidative post-translational mechanisms. It is worth noting that AQP8 has also been implicated in the regulation of hepatic transcription factors, in addition to LXR, including farnesoid X receptor, suggesting a broader role in metabolic homeostasis^32^.

Our previous findings demonstrated that a mitochondria-targeted antioxidant prevents the upregulation of SREBP-driven cholesterogenesis in mtAQP8-overexpressing hepatocytes^16^, supporting the involvement of mitochondrial H₂O₂ signaling in this process. Consistent with these findings, our present data suggest that mitochondrial hAQP8 promotes biliary cholesterol excretion by modulating mitochondrial H₂O₂ signaling, which in turn affects SREBP-2 and LXR pathways, leading to increased ABCG5 expression (Fig. 8). The inhibitory effect of MitoTempo on mitochondrial hAQP8-induced cholesterol excretion further underscores the critical role of mitochondrial H_2_O_2_ in regulating hepatic cholesterol elimination.

To explore whether another aquaporin could similarly induce biliary cholesterol excretion, we investigated the effects of hAQP1^19^. AQP1 is the archetypal water channel protein^33^, nevertheless there is experimental evidence indicating that it is also capable of conducting H₂O₂^34,35^. However, unlike AQP8, AQP1 is not localized to mitochondria. Despite the successful hepatocyte-specific expression of hAQP1 via adenoviral delivery, our findings indicated that hAQP1 did not affect ABCG5 expression or biliary cholesterol output (Fig. 7). The divergent outcomes observed with AQP1 and AQP8 expression reinforce the functional specificity of mitochondria-located AQP8 in modulating cholesterol handling and provide additional evidence to support the role of mitochondrial H₂O₂ as a signaling intermediate in cholesterol metabolic pathways.

In hepatocytes, AQP8 is also expressed on the canalicular membrane, where it facilitates the osmotic transport of water during bile formation^36^. It was recently demonstrated that canalicular AQP8 facilitates bile dilution and prevents gallstone formation in mice^37^. While mtAQP8 primarily transports H_2_O_2_ rather than water^38^, its role in modulating redox signaling may influence biliary cholesterol excretion and hepatic cholesterol homeostasis. Our findings align with this, suggesting that targeting mtAQP8 could have therapeutic potential for gallstone disease.

In conclusion, our results suggest that AQP8, via mitochondria-derived H_2_O_2_, plays a role in SREBP-2-LXR-controlled biliary cholesterol excretion by modulating the canalicular expression of ABCG5. Based on our previous^15,16^ and current observations, mtAQP8 appears to play an important role in hepatic cholesterol homeostasis. Given the central role of cholesterol imbalance in metabolic liver diseases, mtAQP8 could represent a promising target for therapeutic intervention.

Materials and methods

Materials and reagents

Isoflurane was from Baxter Healthcare Corporation. Sucrose was obtained from MP Biomedicals. Protease inhibitors Phenylmethylsulfonyl fluoride (PMSF) was purchased from Sigma-Aldrich (St. Louis, MO, USA) and leupeptin was obtained from Chemicon Millipore (St. Louis, MO, USA). For the immunoblotting materials, Polyscreen PVDF transfer membrane was from Perkin Elmer Life and Analytical Sciences (Waltham, MA, USA) and nitrocellulose membranes was from Cytiva Life Sciences Amersham (Germany); Pierce ECL Western blot analysis substrate was from Thermo Fisher Scientific (Waltham, MA USA). TRIzol reagent were all from Invitrogen (Carlsbad, CA). MitoTempo was purchased from Santa Cruz Biotechnology (Dallas, TX). Solvents for thin layer chromatography were heptane and acetic acid from Cicarelli (Santa Fe, Argentine) and diisopropylether from Sigma-Aldrich (Darmstadt, Germany). Methanol was obtained from Cicarelli (Santa Fe, Argentine).

Animals

Male C57BL/6 mice, 6–8 weeks of age, were maintained on a standard laboratory diet and water ad libitum and housed in a temperature- and humidity-controlled environment under a constant 12-h light-dark cycle. Animals received human care, according to criteria outlined in the “Guide for the Care and Use of Laboratory Animals” (National Research Council, Washington D.C.: National Academy Press, 1996). All the experimental protocols were approved by the Institutional Animal Use Committee of the National University of Rosario, Argentina (Resolution n° 201/2021). All methods were performed in accordance with the relevant guidelines and regulations, as well as in accordance with the ARRIVE guidelines.

Adenoviral vectors

Adenoviral vectors were employed to achieve expression of human AQP8 or suppression of endogenous hepatic AQP8 expression in vivo.

Adenoviral-mediated human AQP8 expression

achieved by administration of AdhAQP8, a serotype 5 replication-deficient bicistronic recombinant adenoviral vector encoding hAQP8 and enhanced green fluorescent protein (EGFP)^16,17^. As control, it was used an adenovector which encodes only for EGFP.

Adenoviral-mediated knockdown of AQP8

given the lack of available specific pharmacological inhibitors for AQP8, gene silencing approaches was employed. Hepatic knockdown of AQP8 in vivo was achieved via administration of AdAQP8sh, a recombinant adenoviral vector encoding a short hairpin RNA (shRNA) targeting murine Aqp8 mRNA. ShRNA-expressing adenoviral construct, along with a scrambled-sequence control vector, were designed and validated in-house. All vectors were produced by the Vector Core Facility of Instituto Leloir, Buenos Aires, Argentina. Preliminary experiments confirmed their efficacy in downregulating mtAQP8 expression in murine liver tissue. Since none of the control adenovectors (EGFP and scrambled) exhibited significant differences compared to sterile saline solution, we opted to use a single adenovector control group.

Adenoviral-mediated human AQP1 expression

achieved by administration of AdhAQP1, a serotype 5 replication-deficient recombinant adenoviral vector encoding hAQP1^18,19^.

Adenoviral administration

The mice were anesthetized with isoflurane; the induction was conducted in a saturated chamber with isoflurane 4% under oxygen at a flow rate of 0.8 L/min until the loss of the righting reflex. After induction, the mice were moved to a thermostated platform, and anesthesia was maintained with isoflurane 2% under oxygen at the same flow rate through a facemask during the procedure. A 2 cm midline laparotomy was performed just below the xyphoid through the alba line to expose the duodenum. The cystic duct was clamped and, then, a bile duct catheterization was carried out by placing a PE-10 polyethylene catheter (Intramedic, Clay Adams, Parsippany, NJ) through the Vater papilla^17,18^. A virus dose of 3.10^9^ PFU/liver (AdhAQP8 or AdhAQP1), 5.10^9^ PFU/liver (AdAQP8sh) or control adenovector was suspended in 100 µl of sterile saline was retrogradely infused into the biliary tract during a period of 3 min. The catheter was maintained in place for 15 min to prevent backflow. After the procedure was completed, the catheter was gently removed, the duodenal puncture and abdominal wall were sutured and the animal was allowed to recover. The experiments were performed 72 h after adenoviral infusion. We performed preliminary experiments to determine the optimal adenovirus administration dose in downregulating mtAQP8 expression (see Supplementary Fig. S1 online). The efficacy of the adenoviral vector in modulating hepatic AQP8 expression and AQP1 has been previously validated and reported^16–19^. The mitochondria-targeted antioxidant MitoTempo (5 mg/kg body weight i.p.) was administered 24 and 48 h after adenoviral infusion. Under these/similar conditions, MitoTempo was found to block mitochondrial H_2_O_2_^25^.

Isolation of mouse hepatocytes

Given the high endogenous expression of AQP1 that we observed in hepatic endothelial cells, isolated hepatocytes were used to verify the success of AQP1 gene transfer specifically into hepatocytes. First, the C57BL/6 male mice were anesthetized with isoflurane, and the surgical procedure was performed as described above, using the adenoviral vector AdhAQP1 or control^19,39^. After 72 h after adenoviral infusion, mice were anesthetized with an intraperitoneal mixture of ketamine (100 mg/kg bw) and xylazine (10 mg/kg bw)^40^. Anesthesia was confirmed by the absence of pedal reflex and the animals were maintained on a heated pad. Then, hepatocytes were isolated from control or AdhAQP1 C57BL/6 male mice by collagenase perfusion and mechanical disruption as previously described^40^. Total hepatocyte lysates were prepared using ice-cold RIPA buffer supplemented with protease inhibitors.

Bile secretion studies and cholesterol excretion

After 72 h of the adenoviral infusion, mice were anesthetized as mentioned above and maintained under this condition during bile collection. For this purpose, a middle abdominal incision was made, the cystic duct was clamped and, then, a bile duct catheterization was carried out by placing a PE-10 polyethylene catheter (Intramedic, Clay Adams, Parsippany, NJ) through the Vater papilla. Bile was collected for two 30-min periods. Bile flow was determined by gravimetry, assuming a density of the bile of 1.0 g/ml. At the end of bile collection, blood samples were taken by cardiac puncture, animals were euthanized by exsanguination, and the livers harvested for evaluation. To determine the cholesterol biliary excretion, total lipids were extracted using the method developed by Folch et al.^41^, the organic phase was separated and finally the solvent was evaporated under N_2_ stream. Subsequently, lipids were resuspended in 50 µL of chloroform and resolved by thin layer chromatography. For this, a mixture of heptane: diisopropylether: acetic acid in a 15:10:1 ratio was used. Finally, the band corresponding to cholesterol was detected using a cholesterol-specific staining method^42,43^, and subsequently quantified by densitometric analysis. Experiments were performed in duplicate using independent extractions. For quantification, a pure cholesterol standard (Sigma-Aldrich, St Louis, MO) of known concentration was included. Biliary cholesterol excretion was calculated as the product of bile flow and cholesterol concentration.

Preparation of cellular fractions and hepatic canalicular plasma membrane

Livers were homogenized by 15 up-down strokes with a loose fitting Dounce homogenizer in four volumes of 0.3 M sucrose (Merck Chemicals, Darmstadt, Germany), containing 0.1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich, St Louis, MO) and 0.1 mM leupeptin (Chemicon, Millipore). Liver homogenates were subjected to low-speed centrifugation to obtain postnuclear supernatants. A fraction of the cell lysates was first subjected to low-speed centrifugation, 500 × g for 10 min at 4 °C, to remove nuclei, resulting in postnuclear supernatants. These supernatants were then centrifuged, 6000 × g for 10 min at 4 °C, to isolate the mitochondrial fraction as a pellet. The mitochondrial pellet was washed twice and subsequently resuspended in 0.3 M sucrose supplemented with protease inhibitors. The remaining post-mitochondrial supernatant was further centrifuged at 200,000 × g for 60 min using a discontinuous sucrose gradient (1.3 M) to obtain plasma membrane-enriched fractions, following established protocols^19^. Afterwards, the plasma membrane band was removed, diluted to 0.3 M, and centrifuged at 200,000 g for 60 min at 4 °C resulting in the plasma membrane fraction, which was subsequently purified by centrifugation at 100,000 × g for 90 min on a continuous (9–60%) sucrose gradient. Purified plasma membranes were loaded onto a discontinuous sucrose gradient consisting of 31%, 34%, and 38% sucrose. Following centrifugation at 270,000 × g for 3 h, the band above the 31% layer corresponding to the canalicular plasma membrane was collected, diluted with washing buffer, and further centrifuged at 200,000 × g for 1 h^18,19^. To prepare total whole lysates, 25 mg livers were washed with cold PBS and lysed in RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with protease and phosphatase inhibitors. Lysates were incubated on ice for 30 min and clarified by centrifugation at 13,000 × g for 15 min at 4 °C. Supernatants were collected. Proteins were determined according to Lowry et al.^44^. by using bovine serum albumin as standard.

Immunoblotting

Western blottings of membrane fractions were performed as previously described by our laboratory^16,18^, using rabbit anti- hAQP8 [EPR8397] (Abcam), mouse anti- mouse AQP8 (14-Z, sc-81870, Santa Cruz Biotechnology, USA), rabbit anti- mouse SREBP-2 (sc-5603, Santa Cruz Biotechnology), rabbit anti- mouse ABCG5 (Invitrogen, Thermo Fisher), mouse anti- mouse LXRα/β (Santa Cruz Biotechnology), mouse anti-β-actin (Sigma-Aldrich, St Louis, MO), and rabbit anti-prohibitin (Abcam) antibodies. After that, the blots were repeatedly washed and incubated with horseradish peroxidase-conjugated corresponding secondary antibodies. In some experiments, membranes were cut prior to antibody hybridisation based on molecular weight ranges. At the end, the proteins were detected by enhanced chemiluminescence detection system (ECL) using the Amersham ImageQuant 500 system. Unprocessed blot images are provided in the Supplementary Information. Densitometric analysis of the developed bands was performed using Image J Software^45^. The rabbit anti- hAQP8 [EPR8397] used is capable of detecting both mice and human isoforms but with higher affinity for the latter^11,17^.

Real-Time RT-PCR

Total RNA from a liver sample was isolated by using TRIzol reagent (Invitrogen) and cDNA was produced by using MMLV RNase H reverse transcriptase (Promega), according to manufacturer’s instructions. HOT FIREPol EvaGreen qPCR Mix Plus was employed to assess transcript levels and amplification reactions were performed with a StepOne Real-Time PCR System (Applied Biosystems). Primers used for mouse AQP8 detection were 5ʹ-TGTGTAGTATGGACCTACCTGAG-3ʹ and 5ʹ-ACCGATAGACATCCGATGAAGAT-3ʹ, for human AQP8 5ʹ-CCACGCTGGGGAATATCA-3ʹ and 5ʹ-GAGGAGCATCACCAGGTTG-3ʹ. For detection of ABCG5 were 5ʹ-AGCTCTTCCAACACTTCGAC-3ʹ and 5ʹ-TACGTTTCTATTTCCCGCTC-3ʹ and those for housekeeping gene Hypoxanthine Phosphoribosyltransferase1 (HPRT1) were 5ʹ-TCAGTCAACGGGGGACATAAA-3ʹ and 5ʹ-GGGGCTGTACTGCTTAACCAG-3ʹ. The size of the products amplified by each pair of primers was 145 and 92 bp for mouse and human AQP8, respectively; 193 bp for ABCG5 and 142 pb for HPRT1. Relative levels of the AQP8/ABCG5 mRNA normalized to HPRT1 were calculated based on the 2‐ΔΔCt method^16^.

Statistical analysis

Data are expressed as means ± S.E.M. The samples passed the normality test. Significance was determined by Student’s t-test for pairwise comparisons between two experimental conditions; P < 0.05 was considered statistically significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2