Vps33b is crucial for structural and functional hepatocyte polarity

Abstract

Background & aims: In the normal liver, hepatocytes form a uniquely polarised cell layer that enables movement of solutes from sinusoidal blood to canalicular bile. Whilst several cholestatic liver diseases with defects of hepatocyte polarity have been identified, the molecular mechanisms of pathogenesis are not well defined. One example is arthrogryposis, renal dysfunction and cholestasis syndrome, which in most patients is caused by VPS33B mutations. VPS33B is a protein involved in membrane trafficking that interacts with RAB11A at recycling endosomes. To understand the pathways that regulate hepatocyte polarity better, we investigated VPS33B deficiency using a novel mouse model with a liver-specific Vps33b deletion.

Methods: To assess functional polarity, plasma and bile samples were collected from Vps33b liver knockout (Vps33b^fl/fl-AlfpCre) and control (Vps33b^fl/fl) mice; bile components or injected substrates were quantitated by mass spectrometry or fluorometry. For structural analysis, livers underwent light and transmission electron microscopy. Apical membrane and tight junction protein localisation was assessed by immunostaining. Adeno-associated virus vectors were used for in vivo gene rescue experiments.

Results: Like patients, Vps33b^fl/fl-AlfpCre mice showed mislocalisation of ATP-binding cassette proteins that are specifically trafficked to the apical membrane via Rab11a-positive recycling endosomes. This was associated with retention of bile components in blood. Loss of functional tight junction integrity and depletion of apical microvilli were seen in knockout animals. Gene transfer partially rescued these defects.

Conclusions: Vps33b has a key role in establishing structural and functional aspects of hepatocyte polarity and may be a target for gene replacement therapy.

Lay summary: Hepatocytes are liver cells with tops and bottoms; that is, they are polarised. At their bottoms they absorb substances from blood. They then, at their tops, secrete these substances and their metabolites into bile. When polarity is lost, this directional flow of substances from blood to bile is disrupted and liver disease follows. In this study, using a new mouse model with a liver-specific mutation of Vps33b, the mouse version of a gene that is mutated in most patients with arthrogryposis, renal dysfunction and cholestasis (ARC) syndrome, we investigated how the Vps33b gene product contributes to establishing hepatocyte polarity. We identified in these mice abnormalities similar to those in children with ARC syndrome. Gene transfer could partly reverse the mouse abnormalities. Our work contributes to the understanding of VPS33B disease and hepatocyte polarity in general, and may point towards gene transfer mediated treatment of ARC liver disease.

Keywords: Bile; Bile canaliculi; Canalicular membrane; Cholestasis; Gene transfer; Hepatocyte polarity; Liver diseases; Protein trafficking; Rab11a; Vps33b.

Graphical abstract

Fig. 1

ARC liver disease in Vps33b^fl/fl-AlfpCre mice. (A) Plasma ALP activity, (B) plasma ALT activity (C) liver mass and (D) total plasma bile acid levels assessed in 14-week-old normal chow fed and cholic acid (CA) fed Vps33b^fl/fl-AlfpCre and Vps33b^fl/fl mice (n = 4–6). (E) Trichrome staining of liver sections from 14-week-old CA fed mice and 12-month-old mice fed with normal diet (n = 3). Graphs are presented with medians and IQR. p values (Mann Whitney U test): ALP activity, p = 0.0317 and p = 0.0022; ALT activity p = 0.0591 and p = 0.1775; plasma bile acids, p = 0.0043 and p = 0.0022; liver mass, p = 0.0117 and p = 0.0050 (normal chow fed and CA fed comparisons respectively).

Fig. 2

Functional polarity defects in Vps33b^fl/fl-AlfpCre mice. (A) Immunostaining for detection of BSEP and (B) MRP2 localisation in Vps33b^fl/fl-AlfpCre and Vps33b^fl/fl liver sections. (C) Plasma and bile levels of cholesterol and (D) phospholipid in Vps33b^fl/fl-AlfpCre and Vps33b^fl/fl mice assessed by fluorometric analysis (n = 5, p = 0.0079 in all tests, Mann Whitney U test). (E) Cumulative biliary levels of CLF and (F) blood levels of CLF measured in Vps33b^fl/fl-AlfpCre and Vps33b^fl/fl (n = 4). All mice were fed a cholic acid supplemented diet and sacrificed at 14–15 weeks of age. Graphs are presented with medians and IQR.

Fig. 3

Gene expression analysis. Heat map showing gene expression of the top 50 genes identified as differentially expressed by PLS loading in the first component. Analysis was performed on liver RNA samples from 10–14-week-old, cholic acid fed Vps33b^fl/fl and Vps33b^fl/fl-AlfpCre mice. Color Key: 4 (red) = lower expression level; 12 (yellow) = higher expression level. Left hand trees link genes with similar expression patterns but do not necessarily infer a functional relationship.

Fig. 4

In vivo gene transfer of human VPS33B. (A) Schematic of ssAAV2/8 vectors containing wild-type (hVPS33B) and codon optimised (hVPS33Bco) human VPS33B cDNA. (B) Plasma alkaline phosphatase activity levels (p = 0.0016, Kruskal-Wallis test), (C) total plasma bile acid levels (p = 0.0015, Kruskal-Wallis test) (D) Plasma cholesterol levels (p = 0.0023, Kruskal-Wallis test) (E) plasma phospholipid levels (p = 0.0047, Kruskal-Wallis test) and (F) CEA protein localisation assessed 9 weeks after injection of ssAAV2/8-EFS-hVPS33B-WPRE and ssAAV2/8-EFS-hVPS33Bco-WPRE (1 × 10¹² vg/mouse) in 5-week-old Vps33b^fl/fl-AlfpCre mice (n = 4–6). All mice were fed cholic acid supplemented diet and sacrificed at 14 weeks of age. Graphs are presented with medians and IQR.

Fig. 5

Tight junction defects in Vps33b^fl/fl-AlfpCre mice. Appearance of hepatocellular tight junctions assessed by immunostaining for Claudin1 in (A) Vps33b^fl/fl and (B) Vps33b^fl/fl-AlfpCre mouse liver sections and in (C) sections of Vps33b^fl/fl-AlfpCre mouse liver 9 weeks after injection of ssAAV2/8-EFS-hVPS33Bco-WPRE vector (1 × 10¹² vg/mouse injected at 5 weeks of age) (n = 3). (D) Cumulative FD-40 levels measured in bile fractions from Vps33b^fl/fl-AlfpCre and Vps33b^fl/fl mice collected after intravenous FD-40 injection (n = 4–5, p = 0.0134). All mice were fed a cholic acid supplemented diet and sacrificed at 14–15 weeks of age.

Fig. 6

Bile canaliculus ultrastructure. Transmission electron micrographs of bile canaliculi in (A) Vps33b^fl/fl and (B) Vps33b^fl/fl-AlfpCre mouse liver and in (C) liver of Vps33b^fl/fl-AlfpCre mice injected with 1 × 10¹² vg/mouse of ssAAV2/8-EFS-hVPS33Bco-WPRE vector. (D) Measurements of canalicular membrane length (± SEM) as a factor of canalicular perimeter assessed using ITEM software on mouse hepatocyte electron micrographs (n = 62–90, p <0.0001, One way ANOVA, Bonferroni’s test). All mice were fed a normal chow diet and sacrificed at 14–15 weeks of age.