Myo5b knockout mice as a model of microvillus inclusion disease

Inherited MYO5B mutations have recently been associated with microvillus inclusion disease (MVID), an autosomal recessive syndrome characterized by intractable, life-threatening, watery diarrhea appearing shortly after birth. Characterization of the molecular mechanisms underlying this disease and development of novel therapeutic approaches is hampered by the lack of animal models. In this study we describe the phenotype of a novel mouse model with targeted inactivation of Myo5b. Myo5b knockout mice show perinatal mortality, diarrhea and the characteristic mislocalization of apical and basolateral plasma membrane markers in enterocytes. Moreover, in transmission electron preparations, we observed microvillus atrophy and the presence of microvillus inclusion bodies. Importantly, Myo5b knockout embryos at day 20 of gestation already display all these structural defects, indicating that they are tissue autonomous rather than secondary to environmental cues, such as the long-term absence of nutrients in the intestine. Myo5b knockout mice closely resemble the phenotype of MVID patients and constitute a useful model to further investigate the underlying molecular mechanism of this disease and to preclinically assess the efficacy of novel therapeutic approaches.

Microvillus inclusion disease (MVID) is an autosomal recessive syndrome affecting the intestinal epithelium 1,2 . It was first described in 1978 and it is characterized by the onset of abundant neonatal watery diarrhea that most commonly starts within the first days of life, and can cause the loss of up to 30% of body weight within 24 h 1,2 . In some cases (<20%), MVID manifests at later times, within the first 3-4 months of life.
The diagnosis of MVID is based on the detection of morphological abnormalities in the intestinal epithelium using a combination of light and electron microscopy. Histological examination of the small intestinal mucosa reveals a variable degree of villus atrophy. In addition, there is a characteristic accumulation of periodic acid-Schiff (PAS)-positive cytoplasmic granules in intestinal epithelial cells 3,4 . Transmission electron microscopy (TEM) of intestinal epithelial biopsies is used to confirm the diagnosis. Ultrastructural defects in small intestinal enterocytes include the shortening of microvilli and the presence of distinctive cytoplasmic vacuoles lined by microvilli, known as microvillus inclusion bodies 5 .
Inactivating mutations in MYO5B have recently been associated with the majority of cases of MVID 6,7 . MYO5B codes for the unconventional myosin Vb, an actin-based motor involved in plasma membrane recycling through its interactions with RAB GTPases 8 . The loss of a functional Myosin Vb protein results in profound protein trafficking defects in enterocytes leading to the mistargeting of apical and basolateral proteins 9,10 . These abnormalities in the structure of the apical brush border and the mislocalization of membrane proteins are likely responsible for the absorption defects and the watery diarrhea observed in Scientific RepoRts | 5:12312 | DOi: 10.1038/srep12312 these patients, but the detailed molecular mechanisms remain to be fully elucidated. Currently, the only treatment options available for this uniformly fatal disease are total parenteral nutrition and intestinal transplant 11 .
Here we describe the phenotype of the first animal model with targeted inactivation of Myo5b. Mice deficient for this myosin show perinatal mortality, watery diarrhea and the characteristic structural defects of patients with MVID. This study provides formal demonstration of MYO5B mutations as the cause of microvillus inclusion disease. Moreover, the availability of this mouse model will decisively contribute to shed new light on the underlying molecular mechanisms of this disease and the development and testing of new therapeutic approaches for MVID patients.

Results
Survival of Myo5b knockout mice. Homologous recombination was used to introduce a targeting cassette including the mouse En2 splice acceptor and the SV40 polyadenylation sequences after exon 4 of Myo5b, which is predicted to generate a null allele through splicing to a lacZ "gene trap" element (Supplementary Figure 1) 12 . As expected, mice homozygous for the trapped allele showed no Myosin Vb expression in their intestine (henceforth referred to as Myo5b knockout mice; Fig. 1). Wild type, heterozygous or Myo5b knockout embryos at day 20 of gestation (E20) showed no difference in their size or weight ( Fig. 2A,B). Animals were born at Mendelian ratios (n = 99; Chi-square test, p = 0.14; Fig. 2C), but Myo5b knockout mice invariably died within the first 12 h after being born (Fig. 2D). No differences were observed between wild type and knockout newborn mice in their body size (Fig. 2E) or the gross histology of the gastrointestinal tract (Supplementary Figure 2) or other organs studied, including the lungs, liver, central nervous system, heart, pancreas and spleen. No cyanotic episodes or respiratory distress was observed in Myo5b knockout mice. However, newborn Myo5b knockout mice showed reduced bodyweight compared to wild type and heterozygous mice (Fig. 2F). Moreover, knockout mice showed signs of diarrhea and wrinkled skin, possibly due to dehydration (Fig. 2G,H). Although newborn Myo5b knockout mice showed no suckling defects (presence of a milk spot; Fig. 2G,H), they had significantly reduced blood glucose levels compared to wild type and heterozygous littermates (Fig. 2I). This is consistent with the watery diarrhea and absorption defects observed in patients with MVID 1,2 and likely contributed to the death of the Myo5b knockout mice within hours of birth 13 .

Mislocalization of apical brush border proteins in the enterocytes of Myo5b knockout mice.
At the ultrastructural level, the intestinal enterocytes of Myo5b knockout newborn mice showed the characteristic cytoplasmic accumulation of periodic-acid Schiff (PAS) staining observed in MVID patients 3

Ultrastructural defects in the brush border of the enterocytes of Myo5b knockout mice.
Transmission electron microscopy (TEM) analysis of the intestinal epithelium revealed the presence of microvillus inclusion bodies in the cytoplasm of absorptive cells from Myo5b knockout E20 embryos (Fig. 4A). In addition, the apical surface of the intestinal enterocytes showed widespread microvilli atrophy and reduced packing with areas with few/absent microvilli (

Discussion
We describe here the phenotype of the first mouse model with targeted inactivation of Myo5b. Germline mutations in this gene are associated with microvillus inclusion disease (MVID) 6 , a congenital disorder of the intestinal epithelium causing persistent life-threatening watery diarrhea 1,2 . Myosin Vb is an actin-based molecular motor with a key role in vesicle trafficking and plasma membrane recycling through its interaction with the small GTPases RAB11 and RAB8 8,9 . It is not surprising, therefore, that inactivation of either Rab11a or Rab8a in the mouse intestine resulted in nutrient malabsorption, intracellular accumulation of apical proteins in intestinal epithelial cells, shortening of microvilli and microvillus inclusion bodies 16,17 . Interestingly, inactivation of the small GTPase Cdc42 also caused microvilli shortening and microvillus inclusions in intestinal epithelial cells 18 . However, diarrhea, one of the hallmarks of MVID patients, was not observed in Rab11 or Cdc42 knockout mice 17,18 , and Rab8 knock out mice survived for approximately 5 weeks after birth, more closely resembling the phenotype of late Scientific RepoRts | 5:12312 | DOi: 10.1038/srep12312 onset MVID. Importantly, no mutations in RAB8 or RAB11 GTPases have been identified in MYO5B mutation negative MVID patients 19 , suggesting that Myo5b deficient mice represent the optimal animal model for human microvillus inclusion disease.
Indeed, Myo5b knockout mice showed all the typical features observed in patients with early onset MVID, the most common form of this disease accounting for >80% of the cases 4 . Myo5b deficient mice showed no overt defects during embryonic development, having normal size and weight. However, newborn Myo5b knockout mice showed watery diarrhea and died during the first 12 h of life, likely due to dehydration and/or reduced nutrient availability secondary to absorption defects, as exemplified by the low blood glucose levels observed, although the contribution of each of these symptoms to the death of Myo5b deficient mice cannot be conclusively determined. The body weight reduction observed in Myo5b knockout mice (8% in approximately 6 h) is consistent with the fluid loss reported in early onset MVID patients (>30% of body weight in 24 h; i.e., 7.5% in 6 h) 3 . In previous studies, the perinatal mortality of Klf4 or Scd2 newborn knockout mice was attributed to a 5-10% reduction of body weight due to transepidermal water loss 20,21 . In humans, this rapid rate of dehydration would result in hypovolemic shock leading to death, as observed in Myo5b knockout mice. On the other hand, newborn mice have previously been shown to go through a transitory phase of severe hypoglycemia (about 10 mg/dL within 2 h of birth) until glucose levels are restored due to gluconeogenesis and eventually nutrient absorption of maternal milk 13,22,23 . Consistent with the phenotype observed in Myo5b knockout animals, the incapacity   Consistent with the changes observed in the intestinal epithelium of MVID patients 1,2 , important structural defects were observed in the enterocytes of Myo5b knockout newborn mice, including the mislocalization of apical and basolateral markers, microvillus atrophy and the presence of microvillus inclusion bodies. Collectively, this study provides formal demonstration of the inactivation of Myo5b as the cause of microvillus inclusion disease. Moreover, these results indicate that the absence of a functional Myosin Vb protein, rather than the presence of pathogenic Myosin Vb mutations 24 , is responsible for the intestinal defects observed in MVID patients. In addition, the presence of ultrastructural defects in the enterocytes of Myo5b-deficient E20 embryos indicates that this phenotype is tissue-autonomous. However, the characteristic villus atrophy observed in patients with MVID was not observed in Myo5b knockout E20 embryos or newborn mice (Supplementary Figure 2E), suggesting that this phenotype is secondary to environmental cues, such as the prolonged absence of nutrients in their gastrointestinal tract 25 . The Myo5b knockout model described here will be instrumental for the characterization of the molecular mechanisms downstream of Myosin Vb responsible for the phenotype observed in patients with MVID, and should significantly contribute to the identification of novel therapeutic approaches for these patients.
It has been reported that up to 75% of MVID patients die before 9 months of age 4 . Different pharmacological approaches have been used in an attempt to stop/reduce the severe diarrhea in these patients, but none of them has proven effective 4 . Patients are dependent on total parenteral nutrition, which over time often causes liver damage and sepsis. Small-bowel transplantation is the only option available to avoid parenteral nutrition and improve the quality of life and the long-term prognosis of these children 11 . However, intestinal transplantation is associated with high rates of rejection and/or mortality 11 , and additional therapeutic options are urgently needed for these patients. The Myo5b knockout model described here will constitute an ideal system to preclinically test the efficiency of possible new treatment options, including pharmacological or gene therapy using for example autologous reimplantation of intestinal epithelium grown ex vivo following restoration of functional Myosin Vb 26 .
In conclusion, we describe here the phenotype of Myo5b knockout mice, which closely phenocopies human early-onset microvillus inclusion disease. These experiments confirm the important role of Myosin Vb in the formation of the apical brush border and the sorting of apical and basolateral proteins in intestinal absorptive cells, and formally demonstrate that the loss of a functional Myosin Vb protein is responsible for the phenotype observed in MVID patients. The availability of this mouse model of MVID will not only contribute to the characterization of the molecular pathological mechanisms downstream of Myosin Vb leading to novel therapeutic approaches, but also provides an ideal system to preclinically test different treatment options.

Transmission electron microscopy. Duodenal samples were collected from Myo5b wild type and
Myo5b knockout E20 embryos (at least 3 animals per genotype). Samples were fixed with 2.5% glutaraldehyde and 2% paraformaldehyde and processed following standard procedures. Ultra-thin sections were mounted on copper grids, contrasted with uranyle acetate/lead citrate double-staining, and observed in a Jeol JEM-1400 (Jeol LTD, Tokyo, Japan) transmission electron microscope equipped with a Gatan Ultrascan ES1000 CCD camera. The brush border architecture was evaluated on a minimum of 12 enterocytes per animal. Microvilli length (actin rootlet and actin core bundles) and microvilli density (microvilli/μ m) were measured using ImageJ software. Duodenum biopsy sample from a MVID patient carrying a homozygous MYO5B nonsense mutation (c.4366C > T, p.1456X) was obtained after removal of the diseased intestine during the transplantation procedure 10 . The sample was fixed in 2% glutaraldehyde in phosphate buffer, rinsed in 6.8% sucrose in phosphate buffer, and postfixed in a solution of 1% osmium tetroxide in 0.1 mol/L sodium cacodylate buffer containing 11.2% potassium ferrocyanide. Samples were dehydrated with ethanol and processed according to standard procedures upon embedding. Ultra-thin sections were mounted on copper grids and contrasted with uranyle acetate and lead citrate double-staining.
Histology and immunohistochemistry. Myo5b wild type and knockout embryos were obtained at day 20 of gestation (E20) and sacrificed by decapitation on ice-cold PBS. Newborns were collected within 6 hours of birth, and sacrificed by decapitation. Both embryos and newborn mice were weighted and measured using a caliper. Blood samples were obtained form tail clips of newborn mice. Glucose levels were measured with a Glucocard G+ meter (Menarini diagnostics, Barcelona). The small and large intestine were dissected from embryos or newborn mice, their length measured and then fixed overnight with 4% formalin, dehydrated by serial immersion in 50%, 70%, 96%, 100% ethanol and xylene and embedded in paraffin. In parallel, formalin-fixed duodenal samples obtained from newborns were cryoprotected in 15% sucrose in PBS overnight, then 30% sucrose overnight. Samples were then immersed in OCT (VWR) and frozen on dry-ice for cryosectioning.