MYO5B and bile salt export pump contribute to cholestatic liver disorder in microvillous inclusion disease

Authors

  • Muriel Girard,

    Corresponding author
    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    2. INSERM, UMR 781, Université Paris Descartes-Sorbonne Cité, Institut Imagine, Paris, France
    • Address reprint requests to: Muriel Girard, M.D., Ph.D., Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, 12 School Street Medicine, 75270 Paris Cedex 06, France. E-mail: muriel.girard@nck.aphp.fr; fax: +33-1-44-49-25-01.

    Search for more papers by this author
  • Florence Lacaille,

    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    Search for more papers by this author
  • Virginie Verkarre,

    1. Department of Pathology, Necker-Enfants Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    Search for more papers by this author
  • Raphael Mategot,

    1. INSERM, UMR 781, Université Paris Descartes-Sorbonne Cité, Institut Imagine, Paris, France
    Search for more papers by this author
  • Gerard Feldmann,

    1. INSERM, UMR 773, Centre de Recherche Biomédicale Bichat Beaujon CRB3, Paris, France
    Search for more papers by this author
  • Alain Grodet,

    1. INSERM, UMR 773, Centre de Recherche Biomédicale Bichat Beaujon CRB3, Paris, France
    Search for more papers by this author
  • Frédérique Sauvat,

    1. Department of Pediatric Surgery, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Sabine Irtan,

    1. Department of Pediatric Surgery, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Anne Davit-Spraul,

    1. Department of Pediatric Hepatology and Biochemistry, Bicêtre Hospital, Assistance Publique-Hôpitaux de Paris, University Paris-Sud 11, Paris, France
    Search for more papers by this author
  • Emmanuel Jacquemin,

    1. Department of Pediatric Hepatology and Biochemistry, Bicêtre Hospital, Assistance Publique-Hôpitaux de Paris, University Paris-Sud 11, Paris, France
    Search for more papers by this author
  • Frank Ruemmele,

    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    Search for more papers by this author
  • Dominique Rainteau,

    1. INSERM, UMR S938, Centre de Recherche Saint-Antoine, and Sorbonne Universités, UPMC Université Paris 06, Paris, France
    Search for more papers by this author
  • Olivier Goulet,

    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    Search for more papers by this author
  • Virginie Colomb,

    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    Search for more papers by this author
  • Christophe Chardot,

    1. Department of Pediatric Surgery, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France
    Search for more papers by this author
  • Alexandra Henrion-Caude,

    1. INSERM, UMR 781, Université Paris Descartes-Sorbonne Cité, Institut Imagine, Paris, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Dominique Debray

    1. Department of Pediatric Gastroenterology and Hepatology, Necker Enfants-Malades Hospital, Assistance Publique-Hôpitaux de Paris, Université Paris Descartes-Sorbonne Cité, Paris, France
    2. INSERM, UMR S938, Centre de Recherche Saint-Antoine, and Sorbonne Universités, UPMC Université Paris 06, Paris, France
    Search for more papers by this author
    • These authors contributed equally to this work.


  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 34

Abstract

Microvillous inclusion disease (MVID) is a congenital disorder of the enterocyte related to mutations in the MYO5B gene, leading to intractable diarrhea often necessitating intestinal transplantation (ITx). Among our cohort of 28 MVID patients, 8 developed a cholestatic liver disease akin to progressive familial intrahepatic cholestasis (PFIC). Our aim was to investigate the mechanisms by which MYO5B mutations affect hepatic biliary function and lead to cholestasis in MVID patients. Clinical and biological features and outcome were reviewed. Pretransplant liver biopsies were analyzed by immunostaining and electron microscopy. Cholestasis occurred before (n = 5) or after (n = 3) ITx and was characterized by intermittent jaundice, intractable pruritus, increased serum bile acid (BA) levels, and normal gamma-glutamyl transpeptidase activity. Liver histology showed canalicular cholestasis, mild-to-moderate fibrosis, and ultrastructural abnormalities of bile canaliculi. Portal fibrosis progressed in 5 patients. No mutation in ABCB11/BSEP or ATP8B1/FIC1 genes were identified. Immunohistochemical studies demonstrated abnormal cytoplasmic distribution of MYO5B, RAB11A, and BSEP in hepatocytes. Interruption of enterohepatic BA cycling after partial external biliary diversion or graft removal proved the most effective to ensure long-term remission. Conclusion: MVID patients are at risk of developing a PFIC-like liver disease that may hamper outcome after ITx. Our results suggest that cholestasis in MVID patients results from (1) impairment of the MYO5B/RAB11A apical recycling endosome pathway in hepatocytes, (2) altered targeting of BSEP to the canalicular membrane, and (3) increased ileal BA absorption. Because cholestasis worsens after ITx, indication of a combined liver ITx should be discussed in MVID patients with severe cholestasis. Future studies will need to address more specifically the effect of MYO5B dysfunction in BA homeostasis. (Hepatology 2014;60:301–310)

Abbreviations
Ab

antibody

ABC

ATP-binding-cassette

ATP

adenosine triphosphate

BA

bile acid

BCs

bile canaliculi

BSEP

bile salt export pump

CLD

cholestatic liver disease

CK7

cytokeratin 7

GGT

gamma-glutamyl transpeptidase

H&E

hematoxylin and eosin

IHC

immunohistochemical

ITx

intestinal transplantation

IV

intravenous

mRNA

messenger RNA

MRP2

multidrug resistance-associated protein 2

MVID

microvillous inclusion disease

PEBD

partial external biliary diversion

PFIC

progressive familial intrahepatic cholestasis

PN

parenteral nutrition

qRT-PCR

quantitative reverse-transcriptase polymerase chain reaction

SEM

standard error of the mean

TEM

transmission electron microscopy

Tx

transplantation

UDCA

ursodeoxycholic acid

Microvillous inclusion disease (MVID; MIM 251850) is a congenital disorder of enterocytes that is responsible for severe diarrhea of neonatal onset. Affected children present with total and definitive intestinal failure, are dependent on parenteral nutrition (PN),[1] and are candidates for intestinal transplantation (ITx). Intestinal biopsies show typical images of “double rail” resulting from the disappearance of microvilli and inclusion bodies in the cytoplasm at transmission electron microscopy (TEM).[1, 2] The molecular defect has recently been discovered through mutations of the MYO5B gene, which encodes a motor protein of the cytoskeleton involved in regulation of membrane trafficking in polarized epithelial cells.[3, 4] MYO5B is required for canalicular formation and targeting of endosomes that contain apical adenosine triphosphate (ATP)-binding-cassette (ABC) transporters to the canalicular membrane by binding to small GTPase Rab proteins, specifically to Rab11a, in liver cell cultures.[5-7] Little is known about regulation of MYO5B in humans, but divergent effects of MYO5B mutations in kidney and intestinal epithelial cells have been reported, suggesting a tissue-specific regulation of the expression of the MYO5B protein.[8, 9]

We previously reported on a large series of children with MVID followed in our center since 1995.[10] Among them, 8 developed a cholestatic liver disease (CLD) akin to progressive familial intrahepatic cholestasis (PFIC). Thus far, expression of MYO5B in hepatocytes and the mechanisms by which MYO5B mutations may affect hepatic biliary function in MVID patients have never been investigated. In this study, we describe the specificities of MVID-associated CLD. To gain insights into liver disease pathogenesis, we used liver biopsy specimens obtained before transplantation (Tx) and investigated the expression of MYO5B, RAB11A, and bile acid (BA) apical transporters, such as the bile salt export pump (BSEP) and multidrug resistance-associated protein 2 (MRP2), that use different apical sorting pathways.[9] We provide evidence for altered expression of MYO5B, RAB11A, and BSEP, but not MRP2, in hepatocytes from MVID patients. In addition, the critical role of active ileal BA reabsorption function is discussed.

Patients and Methods

Patients

Diagnosis of MVID was based on clinical and histological features, as previously described, and confirmed by genetic study.[1] Of 28 children with MVID followed in our center, 8 developed intrahepatic cholestasis with episodes of jaundice and pruritus. Clinical features and liver function tests were reviewed. Total BA concentration in serum was measured by an enzymatic technique using 3 α-hydroxysteroid dehydrogenase. BA composition of serum and bile was analyzed in 1 cholestatic patient by high-performance liquid chromatography, as previously described.[11] Children with CLD (referred to as MVID-CLD) were divided into two groups according to time of onset of cholestasis: Group 1 included 5 children who developed CLD early in life, before ITx, and group 2 included 3 who developed CLD subsequent to ITx. Written informed consent was obtained from parents for genetic studies carried out in all children.

Histopathology

In MVID-CLD patients, liver biopsies (n = 25) were performed in the routine checkup before Tx and repeated once a year during the waiting time and also after Tx, if clinically indicated. Histological features of biopsies obtained before ITx were compared to those of 15 MVID children without CLD (referred to as MVID-no-CLD) and 6 on PN for short bowel syndrome (PN-controls). Liver tissues were fixed in 4% acetic formalin, embedded in paraffin, cut into 4-µm sections, and stained with hematoxylin and eosin (H&E). Perls' reaction, reticulin, Sirius Red, and cytokeratin 7 (CK7) staining were also performed. The METAVIR score was used for grading portal fibrosis from F0 (no fibrosis) to F4 (cirrhosis).[12] Lobular fibrosis was graded from 0 (absent) to 2 (prominent).[13] Portal inflammation, lobular inflammation, and ductular reaction were graded from 0 (absent) to 3 (severe). Immunostaining with anti-MYO5B antibodies (Abs; C-term human MYO5B; LS-B3118-50 against AA: 1093-1112; LifeSpan Biosciences, Inc., Seattle, WA), anti-RAB11A Abs (C-terminus of human RAB11A; catalog no. 610656; BD Biosciences, San Jose, CA), anti-BSEP Abs (a generous gift from Prof. Stieger, Zurich, Switzerland), and anti-MRP2 Abs (clone M2 III-6; Chemicon International) was performed in pretransplant liver biopsies obtained from 5 MVID-CLD children and 7 MVID-no-CLD children, in 5 PN-controls matched for age and grade of liver fibrosis, and 3 healthy controls. We ensured that anti-MYO5B Abs recognized the mutant protein in MVID patients, even in case of nonsense mutation.

TEM

Liver biopsies were performed before ITx in 4 children with MVID-CLD (2 from groups 1 and 2, respectively), in 2 with MVID-no-CLD, and in 2 PN-controls. All were analyzed by TEM. Liver samples were cut into 1-mm3 blocks, fixed in a 2% paraformaldehyde/glutaraldehyde solution and buffered in 0.1 M of phosphate buffer (pH 7.2) for 1 hour at 4°C. After washing with the same buffer, blocks were postfixed in a 1% tetroxide solution for 1 hour at room temperature, then dehydrated in graded alcohols and embedded in epoxy resins. Ultra-thin sections were cut, stained with uranyl acetate and lead citrate, and examined on a JEOL 1010 electron microscope (JEOL, Ltd., Tokyo, Japan). A total of 124 bile canaliculi (BCs) were analyzed in MVID-CLD patients (groups 1 and 2 combined), 30 in MVID-no-CLD children, and 28 in PN-controls. Morphologic changes, such as BC dilatation, intracanalicular cholestasis, absence of microvilli in BCs, and thickening of BC membrane, were expressed for each group of patients as frequencies. In addition, an ultrastructural morphometric study was performed. Ten to twelve electron microgaphs centered on a BC were randomly taken for each patient at an original magnification of 12,000×. The surface area of the lumen of each BC and the perimeter of each canalicular membrane were measured using the software program provided with the electron microscope (Imaging System Olympus Analysis V0.3 software; Olympus Corporation, Tokyo, Japan). A total of 46, 18, and 12 BCs were respectively studied in MVID-CLD patients, MVID-no-CLD patients, and PN-controls. Results were expressed for each group as mean ± standard error of the mean (SEM).

Genomic DNA Sequence Analysis

DNA extraction from peripheral blood samples, polymerase chain reaction (PCR), and direct sequencing of coding regions and splice sites of MYO5B, ABCB11/BSEP, ATP8B1/FIC1 genes were carried out as previously described.[3, 14]

Statistical Analysis

Quantitative variables were expressed as mean + standard error of the mean (SEM) and qualitative variables as frequencies. Main features were compared between groups by using the Mann-Whitney test for quantitative variables and the Fisher exact test for qualitative variables, with statistical significance set at a P value of less than 0.05.

Results

Liver disease of 8 MVID-CLD patients (5 boys and 3 girls) was studied. Six children received an isolated small bowel and colon Tx at a median age of 3 years, and 1 received a combined liver and ITx at 1.3 years of age (Tables 1 and 2). All transplanted patients received a tacrolimus-based immunosuppressive regimen. Ages at last follow-up ranged from 2.8 to 14 years. The consanguinity rate was similar in the MVID-CLD and MVID-no-CLD groups.

Table 1. Biological Features of MVID-CLD Patients During an Acute Episode of Cholestasis
 PatientsSexMYO5B MutationAge at Onset of Cholestasis (Months)AST/ALT (UI/L)GGT (UI/L)Total Bilirubin/Conjugated Bilirubin (µmol/L)Serum BA (µmol/L)
  1. Peak values are recorded during the most severe acute episode of cholestasis after ITx, except for patients 4 (combined liver plus ITx) and 5 (not transplanted).

  2. Abbreviations: AST, aspartate aminotransferase; ALT, alanine aminotransferase; BA, bile acids; NA, nonavailable.

Group 11FCompound heterozygous393/521488/79122
2FHeterozygous377/818204/136154
3MHomozygous3103/623093/6878
4MHomozygous389/582881/5976
5MCompound heterozygous696/10240170/7085
Group 26MHomozygous4451/5620211/169222
7FCompound heterozygous6075/5227111/9189
8MHomozygous43124/1214283/69NA
Table 2. Surgical Treatment and Outcome of Cholestasis in MVID-CLD Patients
 PatientsAge at TxType of TxSurgical Procedure After Tx (Time After Tx)Outcome of CholestasisAge at Last Follow-up
  1. a

    Died 2 months after procedure.

  2. Abbreviations: m, months; PEBD, partial external biliary diversion; y, year.

Group 1117 mBowel + colonIleal exclusion (3 m), PEBD (2 y)Partial remission6 y 3 m
212 mBowel + colonRemoval of graft (4 y)Remission8 y 7 m
360 mBowel + colonRemoval of graft (2 y)Partial remission12 y 6 m
414 mLiver + bowel + colonRemission2 y 11 m
5NoneNoneRemission2 y 8 m
Group 2641 mBowel + colonNasobiliary drainage (3 m), ileal exclusion (1 y)Remission9 y 6 m
741 mBowel + colonGallbladder drainage (2 y)Remission*5 y 4 m
835 mBowel + colonRemoval of graft (5 y)Remission14 y

Main Features of MVID-CLD

The main clinical features of the 8 patients who developed cholestasis (MVID-CLD) are shown in Tables 1 and 2 and compared to the clinical features of MVID-no-CLD patients in Supporting Table 1. Sequencing of MYO5B was performed in the whole series of MVID children followed in our center.[3, 15] Of the 8 MVID-CLD patients, 4 carried homozygous mutations, 3 were compound heterozygous, and 1 was heterozygous, whereas the second mutation was not identified (Table 1). There was no clustering of mutations (Supporting Fig. 1).

All MVID-CLD children presented with jaundice, pruritus, and hepatomegaly. In the 5 children of group 1, jaundice occurred between 3 and 6 months of age while on PN and before transplantation. In the 3 children of group 2, jaundice occurred within 3 and 24 months after ITx once they were off PN. For 2 children, jaundice occurred before the ileostomy was pulled down. None of these 3 children was jaundiced or had pruritus before ITx. In both groups, cholestasis had initially a similar course, characterized by intermittent jaundice and occurrence of severe and debilitating pruritus. The pruritus was severe enough to cause excoriations and skin thickening of hands and feet. Biochemical tests showed mild elevation of transaminases activity (1.5-5.0 times the upper normal value), increased serum direct bilirubin (1-20 times the upper normal value), and increased serum total BA levels (6-40 times the upper normal value), despite normal gamma-glutamyl transpeptidase (GGT) activity at the time of cholestasis (Table 1). Serum albumin levels and prothrombin time were normal. Ultrasonography of the liver ruled out biliary obstruction.

In bile, total BA concentration studied in 1 MVID-CLD patient was low (3.3 mmol/L). Cholic acid was the major BA in bile, whereas chenodeoxycholic acid was detected in low amounts (Supporting Table 2).

Treatment and Outcome of Cholestasis

Follow-up ranged from 2.5 to 12.5 years from onset of cholestasis. Outcome of cholestasis and pruritus (shown in Table 2) varied according to therapeutic interventions.

Children of group 1 (i.e., pre-Tx onset of cholestasis) were treated with ursodeoxycholic acid (UDCA), cholestyramine, and either oral or intravenous (IV) rifampicin without efficacy in 3 patients (patients 1, 2, and 3), with partial efficacy (patient 4) or full efficacy (patient 5) in 1, respectively. In patients 1, 2, and 3, pruritus worsened after ITx, despite rifampicin, naltrexone, or sertralin therapy. Patient 1 underwent a 40-cm ileal exclusion of the transplanted bowel together with pull down of the stoma at 3 months post-ITx, as proposed in children with PFIC,[16, 17] resulting in a transient remission of pruritus. Partial external biliary diversion (PEBD), as described previously,[17] was achieved 2 years later by performing a cholecystostomy, resulting in partial remission of pruritus. Because of intractable acute rejection, the small bowel graft was removed 4 and 2 years after ITx, respectively, in patients 2 and 3. Unexpectedly, pruritus resolved subsequently in patient 2 without recurrence during 4 years of follow-up after graft removal, whereas patient 3 experienced intermittent and less-severe episodes of pruritus than before graft removal (Table 2). Patient 4 received a primary combined liver and ITx and had no relapse of pruritus. Patient 5 did not undergo Tx: jaundice and pruritus improved after a few weeks of IV rifampicin treatment. This patient had no relapse during 1.5 years of follow-up from withdrawal of therapy.

Children of group 2 (i.e., post-ITx onset of cholestasis) were treated with UDCA, cholestyramine, and IV rifampicin without any effect on pruritus. Patient 6 underwent nasobiliary drainage, which was immediately successful based on both clinical and biochemical markers of cholestasis, but relapsed after removal; ileal exclusion was then performed 1 year post-ITx, resulting in complete remission of cholestasis and pruritus during 4 years of follow-up. Patient 7 underwent gallbladder drainage 3 months after onset of cholestasis, resulting in immediate resolution of pruritus; the child died of septic shock a few weeks after the procedure. In patient 8, the bowel graft was removed 5 years post-ITx because of intractable acute rejection, resulting in complete remission of pruritus during the 9 years of follow-up despite resuming PN.

Altogether, these observations suggest that the onset or worsening of cholestasis is linked to active ileal BA reabsorption function.

Histological and Ultrastructural Characterization of Hepatic Lesions

Main histological features are shown in Table 3. The first and last biopsies were performed at a mean age of 19.8 months and 5.5 years, respectively. Histological features were quite similar in both groups. In group 1, the main features at first biopsy, before ITx, were mild portal fibrosis and portal tract inflammation (Fig. 1A1,B1). Canalicular cholestasis was present in 4 patients. Only 1 patient had ductular bile plugs (Fig. 1B2). Ductopenia was suspected on H&E staining in patients 1 and 4 (Fig. 1B3), but was not confirmed by cytokeratin 7 (CK7) staining that revealed dystrophic ducts in the portal tracts. CK7 staining also showed a ductular reaction at the periphery of the portal tracts in these patients (Supporting Fig. 2). There was no sign of acute cholangitis or periductal fibrosis. No lobular fibrosis, but mild lobular inflammation, was noted and steatosis was uncommon. Portal fibrosis worsened in 3 patients (patients 1, 3, and 4; Fig. 1A2), whereas lobular fibrosis developed in 2 (patients 1 and 3). In group 2, liver biopsies performed before Tx (i.e., before onset of cholestasis) revealed moderate (F2) portal, but no lobular, fibrosis associated with mild portal tract inflammation. More specifically, 3 patients displayed canalicular cholestasis and 2 moderate ductular reaction. Mild steatosis was noted in 2 children. Portal fibrosis progressed in patients 6 and 8, and lobular fibrosis developed in patient 7. Histological features that were compared between MVID-CLD and MVID-no-CLD children (shown in Supporting Table 3) were similar.

Table 3. Liver Histological Features in MVID-CLD Patients
 PatientsAge (Months) at CLD/TxAge (Months) at First/Last BiopsyFeatures at First/Last Biospy
Portal FibrosisPortal InflammationDuctular ReactionHepatocellular Cholestasis/Canalicular CholestasisLobular Inflammation/Lobular FibrosisSteatosis(%)
  1. Dash (—) indicates “not done.” Fibrosis was graded from 0 to 4 according to the METAVIR score. Portal and lobular inflammation and ductular reaction were graded from 0 (absent), 1 (mild), 2 (moderate), to 3 (severe). Lobular fibrosis was graded from 0 (absent), 1 (present), to 2 (prominent).

Group 113/1714/28F1/F21/21/1(0/+)/(+/+)2-0/2-20/0
23/129/36F1/F11/10/0(0/+)/(+/+)0-0/1-00/0
33/6043/147F1/F41/20/2(0/+)/(+/+)0-0/1-130/10
43/146/14F1/F31/10/1(0/+)/(+/+)1-0/1-00/0
56/—19/—F0/—1/—0/—(0/0)/—1-0/—0/—
Group 2644/4121/77F2/F31/12/1(0/+)/(0/0)0-0/0-010/30
760/4113/63F2/F21/10/1(0/+)/(0/0)0-0/0-10/0
843/3534/163F2/F41/12/0(+/+)/(+/+)0-0/0-010/30
Figure 1.

Histopathology of the liver in MVID-CLD. (A) Progression of liver fibrosis in MVID-CLD patients after ITx. (A1) Patient 1: 14 months old, before Tx. Portal fibrosis is scored F1 (picrosirius 25×). (A2) Patient 1: 28 months old, after ITx. Portal tract fibrosis is scored F2 and is associated with centrilobular fibrosis (picrosirius 25×). (B; B1) Portal tract inflammation (H&E 200×). (B2) Canalicular bile plugs and pseudorosette formation with hepatocyte clarification (H&E 400×). (B3) Portal tract inflammation with no bile duct visible (H&E 200×).

In addition, we studied the ultrastructural aspect of BCs in liver specimens obtained from MVID patients dependent on PN before ITx and in PN-controls (Fig. 2). Results of the morphometric study of BCs are summarized in Supporting Table 4. In PN-controls, all BCs displayed a normal ultrastructural aspect (Fig. 2A). The mean surface area of canaliculi lumens was 0.5281 ± 0.03438 µm2 and the mean perimeter of canalicular membranes was 3.291 ± 0.1766 µm. In MVID-no-CLD, TEM revealed dilatation of the lumen (66%), canalicular cholestasis (53%), thickening of the canalicular membrane (3%), and disappearance of microvilli within canaliculi (50%), but no inclusion bodies containing microvilli in hepatocytes. The mean surface area of canaliculi lumens (1.832 ± 0.3076 µm2) and the mean perimeter of canalicular membranes (6.090 ± 0.5688 µm) were significantly increased, compared to PN-controls (P < 0.0001 and P < 0.0001, respectively; Fig. 2C,D). In MVID-CLD patients (groups 1 and 2 combined), the following features were observed: dilatation of the lumen (78%); canalicular cholestasis (70%); pericanalicular thickening (63%); and disappearance of microvilli within canaliculi (88%; Fig. 2B). Bile plugs were noted in only a few canaliculi. No inclusion bodies containing microvilli were observed in hepatocytes. The mean surface area of the canaliculi lumens (2.170 ± 0.3856 µm2) and the mean perimeter of canalicular membranes (6.829 ± 0.8923 µm) were significantly increased, compared to PN-controls (P < 0.005 and P < 0.005, respectively; Fig. 2C,D). Morphological and ultrastructural abnormalities of canaliculi were specific features in MVID patients. The only feature that differed significantly between MVID-CLD and MVID-no-CLD was the absence of microvilli in MVID-CLD patients (P < 0.005; Fig. 2C).

Figure 2.

Ultrastructural abnormalities of the liver. (A) Normal BC (6,000×) in a PN-control (fibrosis and age matched). Normal canaliculus with presence of microvilli (black arrow) and no thickening of the canaliculi membrane. (B) BC (5,000×) in a MVID-CLD patient. Dilatation of the BC, absence of microvilli, intracanalicular cholestasis, and thickening of the canalicular membrane (white arrow) are shown. (C) Morphologic changes of BCs in MVID-CLD, MVID-no-CLD, and PN-controls. BC dilatation and absence of microvilli in BCs are expressed for each group of patients as frequencies (% of total number of BCs analyzed) and compared between groups. (D) Ultrastructural morphometric study of BCs in MVID-CLD, MVID-no-CLD, and PN-controls. The surface area of the BC lumens and the perimeter of canamicular membranes are expressed for each group as mean ± SEM and compared between groups. A P value <0.05 is considered significant. NS, not significant.

Molecular Characterization of MVID-CLD

The clinical phenotype of liver disease in MVID patients was similar to PFIC and “benign recurrent cholestasis,” which are, respectively, caused by mutations in ABCB11 (encoding BSEP protein) and ATP8B1 (encoding FIC1 protein) genes. However, none of the patients carried disease-causing mutations in either gene. A decreased function of BSEP related to the V444A polymorphism (rs2287622),[18] which has been associated with drug-induced cholestasis[19] and intrahepatic cholestasis of pregnancy,[19] was also questioned. No significant increase in the C-allele frequency of the ABCB11 polymorphism, V444A, was observed among MVID-CLD patients (64% vs. 50% in MVID-no-CLD). Genotypes are presented in Supporting Table 5.

To gain insight into the pathogenesis of liver dysfunction in MVID patients, we investigated whether MYO5B messenger RNA (mRNA) was expressed in normal and MVID livers by quantitative reverse-transcriptase (qRT)-PCR. No significant difference in expression level of MYO5B mRNA between controls (healthy and PN-controls) and MVID (CLD or no-CLD) patients (Supporting Fig. 3) was noted, even when taking into account the type of MYO5B mutation. However, an abnormal subcellular distribution of MYO5B in hepatocytes from MVID patients was demonstrated by immunohistochemical (IHC) analysis (Fig. 3). In PN-controls as well as in healthy controls, MYO5B was mainly localized to the cell periphery of hepatocytes, delineating a frame to each cell. In contrast, in MVID patients, MYO5B immunoreactivity was intense and granular in the cytoplasm of hepatocytes. Because binding of MYO5B to RAB11A is required for subcellular positioning of recycling endosomes[6] and BC formation,[5] RAB11A immunostaining was also performed. Whereas RAB11A immunoreactivity was detected exclusively at the canalicular membrane of hepatocytes in healthy and PN-controls, abnormal subcellular cytoplasmic staining was demonstrated in MVID patients (Fig. 3).

Figure 3.

MYO5B and RAB11A immunostaining in liver (400×). MYO5B immunostaining in controls (healthy and PN-controls) is localized to the cell periphery delineating a frame to each cell, with very weak granular cytoplasmic staining. In contrast, in MVID patients, MYO5B is not localized at the cell periphery, but immunoreactivity is intense and granular in the cytoplasm of hepatocytes. RAB11A immunoreactivity is detected exclusively at the canalicular membrane of hepatocytes in healthy and PN-controls. In MVID patients, no canalicular staining is shown, but RAB11A immunoreactivity is weak and granular in the cytoplasm of hepatocytes. Each photograph is representative of 5-7 patients.

IHC analysis of BSEP and MRP2 was then performed in MVID patients and controls (Fig. 4). In PN-controls, BSEP staining was mainly localized to the canalicular membrane, showing a linear, thin, and strong staining. The cytoplasm was also faintly stained. In MVID-no-CLD as in MVID-CLD patients, we observed an increased granular cytoplasmic staining with an intense and thickened canalicular staining that overflowed the canalicular membrane. In contrast, MRP2 immunostaining was alike in all groups (Fig. 4).

Figure 4.

BSEP and MRP2 immunostaining in liver (200×). In PN-controls, BSEP staining was mainly localized to the canalicular membrane, showing a linear, thin, and strong staining. The cytoplasm was also faintly stained. In MVID-no-CLD as in MVID-CLD patients, we observed an increased granular cytoplasmic staining with an intense, thickened canalicular staining that overflowed the canalicular membrane. In contrast, MRP2 immunostaining was alike in all groups. Each photograph is representative of 5-7 patients.

Altogether, these results suggest that impaired MYO5B/RAB11A apical recycling endosomes in hepatocytes and abnormal sorting of BSEP to the canalicular membrane predispose to development of liver disease in MVID patients.

Discussion

Despite the wide tissue expression of MYO5B,[20] MVID symptoms were thought to be restricted to the intestine. Our findings, together with the recent description of transient Fanconi syndrome in MVID patients,[9] expand to other cell types than enterocytes the alteration of apical membrane trafficking in polarized cells. The large cohort of MVID patients followed in our center gave us a unique chance to characterize this peculiar low GGT intrahepatic cholestasis and investigate the mechanisms by which MYO5B mutations may affect hepatic biliary function. Our results provide evidence for (1) disorganization of the canalicular pole of hepatocytes, as assessed by electron microscopy, (2) altered expression of MYO5B, RAB11A, and BSEP in hepatocytes, and (3) the critical role of intestinal BA reabsorption in MVID patients.

Prevalence of CLD in MVID patients was approximately 30% in our series, including 3 children who develop cholestasis after ITx. Among the 20 MVID-no-CLD patients, only 4 had undergone ITx alone (Supporting Table 1). It cannot be excluded that further patients will develop cholestasis once they undergo ITx. MVID children presented with severe pruritus, intermittent jaundice, and unexpected low-serum GGT activity either before or after ITx. Ultrastructural analyses of livers in MVID patients on PN before ITx revealed abnormal canalicular features, such as dilatation and absence of microvilli, that were not observed in PN-controls. However, these canalicular changes have been reported to a milder degree in patients with PFIC1,[21] PFIC2,[22] biliary atresia, and neonatal hepatitis[23] and may occur more or less as a result from severe cholestasis. Unlike the description in enterocytes,[1] no inclusion bodies were observed in hepatocytes, such as in renal tubular cells.[9] Whether cholestasis in these patients could have been related to PN or anticalcineurin inhibitors (i.e., tacrolimus) used after ITx appears unlikely. Pruritus with low GGT activity is definitely not a feature of intestinal failure-associated liver disease and PN.[24] Likewise, tacrolimus-induced cholestasis,[25] which is a very rare side effect, is questionable. Indeed, cholestasis appeared before ITx for the majority of patients and resolved after biliary or gallbladder drainage in 2 of 3 post-ITx recipients, despite continuation of tacrolimus immunnosuppression (Table 2).

To a large extent, liver disease in MVID patients showed clinical and histological similarities with PFIC, though no mutation was found in the ATP8B1 and ABCB11 gene. Moreover, the BA profile in bile, which was studied in 1 child was similar to that reported in PFIC1 patients with down-regulation of BSEP.[14, 26] BSEP, the rate-limiting step for BA efflux, is an apical membrane ABC transporter protein, sorted to the apical membrane of hepatocytes from the trans-Golgi network by RAB11A-positive apical recycling endosomes.[27] Therefore, we hypothesized that mistargeting of BSEP could contribute to pathogenesis of MVID-associated liver disease.

We demonstrated, by IHC, that MYO5B and RAB11A were expressed at the canalicular pole of hepatocytes in normal liver, consistently with previous data obtained in the Wif-B9 polarized cell line.[5] In contrast, MYO5B and RAB11A displayed an abnormal subcellular distribution in hepatocytes from MVID patients. Moreover, we demonstrated an abnormal BSEP canalicular staining pattern in liver cells of MVID patients (Fig. 4). However, the expression of MRP2, known to be sorted directly to the apical membrane and to bypass apical recycling endosomes,[28] was unchanged (Fig. 4). These IHC studies suggest that CLD in MVID is related to impairment of the MYO5B/RAB11A recycling endosome pathway, thus altering BSEP apical targeting.

The reason why not all MVID patients develop CLD remains unknown. No MYO5B genotype-phenotype correlation could be demonstrated. Using qRT-PCR, expression level of MYO5B mRNA did not significantly differ between MVID-CLD and MVID-no-CLD patients. Little is known about regulation of MYO5B in humans, but divergent effects of MYO5B mutations in kidney and intestinal epithelial cells have been reported, suggesting a tissue-specific regulation of the expression of the MYO5B protein.[8, 9] A decrease in BSEP function by the V444A polymorphism was excluded because a similar frequency of this polymorphism was found among cholestatic and noncholestatic patients. BSEP targeting to the apical membrane involves different cellular proteins.[27, 29] Indeed, consequences of mutations in VPS33B or in VIPAR (also called C14ORF133) in arthrogryposis-renal dysfunction-cholestasis (MIM208085) syndrome include mis-sorting of BSEP to the basolateral membrane, resulting in low-GGT intrahepatic cholestasis.[27] Our results suggest that candidate modifier genes of cholestasis may be related to BSEP endocytosis recycling pathways and/or BA intestinal absorption.

Ileal BA uptake may be impaired to various degrees among MVID patients. In response to decreased canalicular BA secretion and decreased intraintestinal BA concentration, up-regulation of the apical sodium-dependent BA transporter (ASBT/SLC10A2) normally occurs and may lead to a harmful increase in intestinal reabsorption of BA, which could exacerbate the cytotoxic effects of BA in the liver and accelerate progression of liver disease.[30] Therefore, it can be suggested that symptomatic cholestasis occurs in MVID patients if BA ileal absorption is sufficiently efficient. On the contrary, disruption of intestinal BA absorption may, in some way, have a beneficial effect. It is noteworthy that after ITx, the superior mesenteric vein of the intestinal graft is reimplanted in the native inferior vena cava. Therefore, BA escape first-pass hepatic extraction from the portal circulation leads to an increase in systemic BA levels that may result in worsening or onset of pruritus. In addition, the portal vein bypass may alter, to some degree, the coordinated intestinal regulation of BA synthesis in the hepatocyte, which could also contribute to the worsening of liver damage.[31] In line with the critical role of intestinal BA absorption in MVID patients, cholestasis subsided once the enterohepatic cycling of BA was interrupted (i.e., after removal of the intestinal graft, ileal exclusion, nasobiliary drainage, or PEBD).

Therefore, occurrence of cholestasis in MVID patients raises questions about the appropriate therapeutic strategy. Because ongoing cholestasis worsens after ITx, leading to liver fibrosis, indication for a combined liver and ITx needs to be considered in children with severe cholestasis. In children undergoing isolated ITx, we recommend the conservation of the gallbladder that could be used for PEBD if cholestasis develops, which prevents a fraction of secreted BA from reaching the ileum.[32, 33] Nasobiliary drainage may help to select potential responders to biliary diversion. Ileal bypass should be avoided because it removes a part of the transplanted bowel and appears not to ensure good long-term remission of cholestasis.[16]

MVID patients are predisposed to the development of a PFIC-like cholestatic liver disease. Our results suggest that cholestasis results from (1) impairment of the MYO5B/RAB11A apical recycling endosome pathway in hepatocytes, (2) altered expression of BSEP at the canalicular membrane, and (3) increased ileal BA absorption and hepatic BA uptake. This study expands the repertoire of BSEP-related liver disease. Future studies will need to address more specifically the consequences of MYO5B mutations in ileal BA absorption. Together with the recent observation of kidney involvement in MVID, our work supports the need to increase our awareness of other phenotypes associated with MYO5B mutations.

Acknowledgment

The authors thank Marie-Elisabeth Samson-Bouma (Necker-Enfants Malades Hospital, Paris, France) for technical assistance in electron microscopy, Caroline Werl of the Tumorothèque and Annie Postel and Catherine Gandon of the Pathology Unit in Necker Hospital for their technical assistance in immunohistochemistry, and Thomas Müller (Innsbruck Medical University, Innsbruck, Austria) for performing MYO5B sequencing. The authors are grateful to Prof. Bruno Stieger (Zurich, Switzerland) for providing us with BSEP antibody and the National French Reference Center for Rare Digestive Diseases (Paris, France) for providing MRP2 antibody. The authors also thank Chris Gordon for writing assistance and the AMFE for their financial support to perform sequencing analysis.

Ancillary