Disruption of Slc52a3 gene causes neonatal lethality with riboflavin deficiency in mice

Abstract

Homeostasis of riboflavin should be maintained by transporters. Previous in vitro studies have elucidated basic information about riboflavin transporter RFVT3 encoded by SLC52A3 gene. However, the contribution of RFVT3 to the maintenance of riboflavin homeostasis and the significance in vivo remain unclear. Here, we investigated the physiological role of RFVT3 using Slc52a3 knockout (Slc52a3-/-) mice. Most Slc52a3-/- mice died with hyperlipidemia and hypoglycemia within 48 hr after birth. The plasma and tissue riboflavin concentrations in Slc52a3-/- mice at postnatal day 0 were dramatically lower than those in wild-type (WT) littermates. Slc52a3-/- fetuses showed a lower capacity of placental riboflavin transport compared with WT fetuses. Riboflavin supplement during pregnancy and after birth reduced neonatal death and metabolic disorders. To our knowledge, this is the first report to indicate that Rfvt3 contributes to placental riboflavin transport, and that disruption of Slc52a3 gene caused neonatal mortality with hyperlipidemia and hypoglycemia owing to riboflavin deficiency.

Disruption of gene causes

R A

P

Hiroki, AtsushiYonezawa, Kaori, YoshiakiYao Sugano, Shunsaku Nakagawa, Satoshi , Takayuki Nakagawa, IkukoYano, Satohiro Masuda, Ken-ichi Inui & Kazuo Matsubara

in vitro in vivo knockout ( / /

/ / fetuses showed a lower of

The water-soluble vitamin B2 (riboavin) is involved in various metabolic reactions. It is converted to active coenzyme forms, avin mononucleotide (FMN) and avin adenine dinucleotide (FAD), which are required as cofactors for redox reactions, including carbohydrate, lipid, and amino acid metabolism. Riboavin is essential for cellular function; deciency of riboavin leads to various clinical abnormalities (e.g. anemia, neurodegeneration and growth retardation)1,2. The low maternal riboavin intake contributes to the risk of some disorders in ospring, like congenital heart defects and transverse limb deciency24, suggesting the necessity of riboavin for healthy fetal growth.

Transporters play an important role in riboavin homeostasis5. Until now, three riboavin transporters have been identied, and designated as the RFVT family (RFVT1-3)69. Riboavin transporter 3 (RFVT3) encoded by the SLC52A3 gene was identied based on the sequence of bacterial riboavin transporter impX in 20097.

Previous in vitro studies have extensively documented basic information about RFVT3713. This transporter consists of 469 amino acids with 11 putative membrane-spanning domains, and is strongly expressed in the small intestine, kidney, testis and placenta7,8. The results of previous studies indicate that RFVT3 can transport riboavin8; however, riboavin analogs, FMN and FAD, are poor substrates for RFVT313. In addition, recent in vitro studies suggest that RFVT3 is functionally involved in intestinal riboavin absorption in the apical membranes of intestinal epithelial cells12,13.

Recently, several clinical reports have mentioned that SLC52A3 mutations have been detected in patients with a neurological disorder (Brown-Vialetto-Van Laere syndrome; BVVLS), wherein fatty acid metabolism is impaired1421. BVVLS is an orphan disease as only 92 patients have been documented prior to 201422. In patients with SLC52A3 mutation, plasma riboavin levels tend to be lower than those in healthy subjects despite normal dietary intake16. In several case reports, the neurological symptoms and abnormal fatty acid metabolism of patients improved aer supplementation with riboavin14,1618. Based on these limited clinical reports, disruption of SLC52A3 might disturb riboavin homeostasis and trigger BVVLS. However, the contribution of RFVT3 to the maintenance of riboavin homeostasis and its physiological signicance in vivo remain unclear.

Japan. Present materials should be addressed to A.Y. (email: [email protected])

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

In this in vivo study, we investigated the impact of Rfvt3 on riboavin homeostasis using Slc52a3 knockout (Slc52a3/) mice. Slc52a3/ mice exhibited several phenotypes including riboavin deciency and neonatal sudden death. Importantly, their conditions were improved by riboavin supplementation. We demonstrated for the rst time that Rfvt3 contributes to placental riboavin transport, and disruption of Slc52a3 caused neonatal lethality with hyperlipidemia and hypoglycemia owing to riboavin deciency.

Targeted disruption of the Slc52a3/ mice in C57BL/6 background were obtained from the Knockout Mouse Project (KOMP) Repository. The mouse Slc52a3 gene was integrated with the SA- Geo-pA trapping cassette between exons 1 and 2 (Supplementary Fig. S1). In genomic long-range PCR analysis, the integrated Slc52a3 allele was detected in the heterozygous mice, indicating that the SA-Geo-pA trapping cassette was correctly targeted to the Slc52a3 locus. Heterozygous male and female mice were mated, and the targeted Slc52a3 allele was detected in these ospring genomic DNA isolated from tail biopsies by PCR analysis. To conrm the disruption of the Slc52a3 gene, Slc52a3 mRNA expression was examined. Real-time PCR analysis demonstrated that Slc52a3 mRNA in the kidney was hardly detected in Slc52a3/ mice, and was signicantly lower in Slc52a3+/ mice than in WT mice.

/ Genotyping of newborn pups from Slc52a3+/ parents (n = 8) at birth revealed no fetal wastage; the mutant allele segregated in Mendelian fashion (P = 0.85, Fig.1a). Most of the newborn Slc52a3/ mice were alive for more than 8 hr, but died within 48 hr aer birth. Survival times of Slc52a3/ mice were signicantly shorter than those of WT mice (P < 0.001, Fig.1b). Macroscopically, although their appearance was not visibly dierent from those of WT mice (Fig.1c), the body weights of Slc52a3/ mice were signicantly lower than those of WT mice at P0 (Fig.1d). Organogenesis appeared to be macroscopically normal (Supplementary Fig. S2). The amount of maternal milk ingestion seemed to be relatively lower in Slc52a3/ pups compared with that in WT pups. The lungs of Slc52a3/ mice oated, indicating that the initial postnatal expansion occurred normally. Slc52a3/ pups as well as WT pups could exercise their arms and legs. Histological analysis of femoral muscle tissues demonstrated that no obvious abnormality was found in Slc52a3/ pups (Supplementary Fig. S3). In addition, apoptosis-inducing factor, mitochondrion-associated 1 (AIFM1) expression levels were also similar in femoral muscles in WT and Slc52a3/ pups. Urinary glutaric acid, a marker for fatty acid metabolism disorder, was signicantly higher in Slc52a3/ mice (P<0.05, Fig.1e).

Plasma riboavin levels in Slc52a3/ mice at P0 were dramatically lower than those in WT mice (Fig.2a). Additionally, plasma concentrations of FAD, but not FMN, were also decreased in Slc52a3/ mice compared with WT mice (Fig.2c,e). In almost all tissues examined, the disruption of Slc52a3 led to not only signicantly lower riboavin but also FMN and FAD levels (Fig.2b,d,f).

The mRNA distribution of Slc52a3 in the mouse placenta was examined by in situ hybridization (Fig.3a). Positive staining for Slc52a3 was detected in the junctional and labyrinth zones in the placenta (Fig.3c,d). Slc52a3 mRNA was not detected in the deciduum (Fig.3b). Hybridization with a sense-probe for Slc52a3 showed an absence of nonspecic binding to the junctional and labyrinth zones (Fig.3e,f). Real-time PCR analysis conrmed the disruption of Slc52a3 in the placenta of Slc52a3/ fetuses (Fig.3g). We next investigated the functional involvement of Rfvt3 in placental riboavin transport. In Slc52a3/ fetuses, the body and placenta weights were signicantly lower than in the WT mice (Table1). Fetal and placental weight ratios did not dier between the WT and Slc52a3/ fetuses. Upon intravenous administration of [3H]riboavin to pregnant dams, Slc52a3/ fetuses exhibited signicantly lower [3H]riboavin levels (Fig.3h). Placental [3H]riboavin transport capacities (adjusted by placental weight) in Slc52a3/ fetuses were also signicantly lower than in WT fetuses (Fig.3i). We conrmed that placental transport of D-[U-14C]glucose was not signicantly aected by knockout of Slc52a3 (Fig.3j,k).

/ We then examined the eect of riboavin supplementation on the survival of Slc52a3/ mice (n= 8) compared with WT littermates (n=12). Briey, Slc52a3+/ mother mice were supplemented with riboavin (50 mg/L) in drinking water and newborn pups were given a daily subcutaneous injection of 0.75 mg/kg riboavin until weaning at the P21 (Fig.4a). Under the riboavin treatment, the survival of Slc52a3/ pups (n= 7) was comparable to that of WT littermates (n= 11) (Fig.4b). In terms of body weights, the values in Slc52a3/ mice tended to be lower than those in WT littermates (Fig.4c).

We evaluated the plasma acylcarnitine and blood glucose levels in Slc52a3/ pups with or without riboavin supplementation. Slc52a3+/ dams were supplemented with riboavin during pregnancy, and newborn pups were sacriced 59 hr aer birth (Fig.5a). Plasma riboavin levels were signicantly (P < 0.05) increased in Slc52a3/ pups when their dams were given riboavin via drinking water (Supplementary Table S1). In almost all tissues examined, the concentrations of FMN and FAD were also increased by riboavin supplementation. Weight loss at birth was signicantly improved in supplemented Slc52a3/ pups (Fig.5b). Analysis of plasma acylcarnitine levels in untreated Slc52a3/ neonatal mice demonstrated elevated levels of several acylcarnitines (C4, C5, and C14) compared with those in WT mice (Table2). These abnormalities of acylcarnitine in Slc52a3/ mice were signicantly improved by riboavin supplementation (Table2). Furthermore, in most of Slc52a3/ pups, blood glucose levels were undetectable (<20 mg/dL) (Fig.5c). Glucose levels in Slc52a3/ pups were signicantly increased by riboavin supplementation (Fig.5c). To conrm the inuence of riboavin supplementation on blood glucose levels, we conducted an experiment using adult Slc52a3/ mice as shown in Supplementary Fig. S4. At 3-week old (riboavin treated condition), the blood glucose levels in Slc52a3/ mice were comparable to those in WT mice. However, two weeks aer terminating riboavin

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 1. Phenotypic analyses of Slc52a3/ mice. (a) Genotypic distribution immediately aer birth (n=63 from 8 litters). (b) Kaplan-Meier survival curves for WT (n= 14; open circles, black line), Slc52a3+/ (n=34; closed circles, green line) and Slc52a3/ (n= 13; closed triangles, blue line) littermates. (c) Gross appearance of whole body of WT, Slc52a3+/ and Slc52a3/ newborn pups at postnatal day 0 (P0). (d) Body weightof WT, Slc52a3+/ and Slc52a3/ littermates at P0. (e) Urinary glutaric acid levels in WT and Slc52a3/ littermates at P0. Each scale bar represents 1cm length. Each column or point represents the meanSD. Values where *P< 0.05, or ***P< 0.001, signicantly dierent from WT.

supplementation (at 5-week old), the blood glucose levels in Slc52a3/ mice were signicantly lower than those in WT mice. As for the blood lactate levels at P0, there was no signicant dierence between WT and Slc52a3/ pups (Fig.5d).

To identify the biochemical mechanisms underlying the hypoglycemia, targeted quantitative metabolomics, focusing on glycolysis, tricarboxylic acid cycle and amino acid metabolites, was performed using liver samples obtained from WT, Slc52a3/ and riboavin-supplemented Slc52a3/ pups at P0. The data from the metabolomics analysis and acylcarnitine levels were simultaneously analyzed by hierarchical cluster analysis (Fig.6). The metabolites of upstream glycolysis (glucose-6-phosphate, and fructose-6-phosphate) were placed in the same classication as glucose (lower in Slc52a3/ than WT pups). Most amino acids, including a gluconeogenic precursor alanine, belong to the same group as C4-acylcarnitine, showing that these levels tended to be higher, but not lower, in Slc52a3/ pups than in WT pups or riboavin-supplemented Slc52a3/ pups.

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 2. Riboavin, FMN, and FAD levels in plasma and tissues from WT and Slc52a3/ mice. Plasma and tissues were obtained from WT (n=14) and Slc52a3/ (n= 13) littermates at postnatal day 0 (P0) immediately aer birth. Plasma concentrations of riboavin (a), FMN (c) and FAD (e) were measured by HPLC. Tissue levels of riboavin (b), FMN (d) and FAD (f) were also determined. Each column represents the mean SD. Values where **P<0.01, ***P< 0.001, signicantly dierent from WT.

Discussion

In this study, we investigated the physiological role of Rfvt3 in vivo using Slc52a3/ mice. Pups from Slc52a3+/ parents showed a Mendelian distribution of the genotypes, while Slc52a3/ pups exhibited the severe phenotype of sudden death almost within 48 hr aer birth. The appearance of tissues obtained from these pups was macroscopically normal, and lung tissues oated in water. These data suggest that Slc52a3 null embryos were not lethal in utero, and that Slc52a3/ pups seemed not to have died from respiratory compromise or developmental defects in the tissues. In Slc52a3/ newborn pups, plasma and tissue riboavin concentrations were prominently lower compared with those in WT littermates. As for FMN, tissue level in Slc52a3/ pups was signicantly lower than that in WT pups, although the plasma level was comparable. It was reported that FMN and FAD are poor substrates for RFVTs, compared with riboavin8,13. FMN and FAD are generated from riboavin by riboavin kinase and FAD synthetase in the cells. Therefore, it was suggested that the tissue levels of FMN and FAD in Slc52a3/ pups were lower than those in WT pups, owing to a decrease in the tissue riboavin concentration. Most importantly, riboavin level was signicantly recovered and neonatal death was apparently reduced in Slc52a3/ pups by riboavin supplementation. Taken together, Rfvt3 is indispensable for the maintenance of riboavin homeostasis in pups and for neonatal survival.

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 3. Involvement of Rfvt3 in the placental riboavin transport. (a) mRNA expression of Slc52a3 was examined by in situ hybridization. The insets indicate the magnication region. (bd) Higher magnication images corresponding the solid ames in (a): deciduum (b), junctional zone (c) and labyrinth zone (d). The blue signals were positive staining for Slc52a3 (arrows). Hybridization with a sense probe for Slc52a3 (negative control) in the junctional zone (e) and labyrinth zone (f). Scale bar: 100m. (g) Slc52a3 mRNA expression in the placentas from WT and Slc52a3/ fetuses derived from Slc52a3+/ dams. Total RNA isolated from the placentas from WT and Slc52a3/ fetuses was reverse-transcribed and mRNA level of Rfvt3 was determined by real-time PCR. Each column represents the mean SD (WT, n=6; Slc52a3/, n= 6 from 2 litters).(h,i) Placental transport of [3H]riboavin in WT and Slc52a3/ fetuses. Slc52a3+/ mice were mated,and [3H]riboavin was administered intravenously to pregnant dams at gestation day 16.5. Five minutes aer administration, fetuses and placentas were collected. Placental transport of [3H]riboavin was expressed per gram of placenta (h) or per gram of fetus (i). Placental transport of [14C]glucose was also examined (j,k). Each column represents the mean SD (WT, n=7; Slc52a3/, n= 8 from 3 litters). Values where ***P<0.001 were signicantly dierent from WT.

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

WT Slc52a3/ Fetal weight (mg) 72222 59422**

Placental weight (mg) 1145 1022* Fetal to placental weight ratio 6.370.26 5.820.22

Table 1. Fetal and placental wet weights of WT, Slc52a3+/, and Slc52a3/ fetuses at E16.5. All data are expressed as the mean SD. Values where *P<0.05 or **P< 0.01 indicate signicant dierence from WT.

Figure 4. Survival of Slc52a3/ mice with riboavin supplementation. (a) Protocol for the riboavin supplementation. Slc52a3+/ mice were mated, and pregnant dams were supplemented with 50mg/L of riboavin in drinking water ad libitum from gestational day 0 to 3 weeks postpartum. Newborn pups were also administered 0.75mg/kg riboavin subcutaneously once a day until weaning at the postnatal day 21. Kaplan-Meier survival (b) and body weights (c) for WT and Slc52a3/ mice with riboavin administration. The body weight values are exressed as the mean SD for mice that were alive at each time point (WT, n=12; Slc52a3/, n= 8 from 4 litters).

It has been suggested that placental riboavin transport is mediated by transporters. Studies with perfused human placental tissues have identied a saturable riboavin uptake process23,24. This active riboavin transport has also been observed in human trophoblast-derived BeWo cell monolayers25. However, the molecular entity of the placental riboavin transporter and its physiological importance are poorly understood as yet. We have previously reported that RFVT3 is expressed in the human placenta8, and the present study demonstrated the signicant role of Rfvt3 in the placental riboavin transport in vivo. In situ hybridization revealed that Rfvt3 was expressed in the labyrinth zone, which is the feto-maternal exchange site in the placenta. Furthermore, the placental riboavin transport capacity in Slc52a3/ fetuses was signicantly lower than that in WT fetuses. In Slc52a3/ newborn pups immediately aer birth, plasma riboavin concentrations were dramatically lower than in WT pups. These results indicate that Rfvt3-mediated placental riboavin transport should play a key role in the maintenance of riboavin homeostasis in fetuses.

Riboavin is converted to the active coenzyme forms (FMN and FAD), which are indispensable for various metabolic reactions. The initial step in fatty acid -oxidation is catalyzed by acyl-CoA dehydrogenase, which requires FAD as a co-factor in the reaction. The present study indicates that disruption of Slc52a3 induced a disorder in fatty acid metabolism, including abnormal levels of plasma acylcarnitine and urinary glutaric acid. Plasma and tissue levels of FAD in Slc52a3/ pups were lower than in WT pups, indicating that riboavin deciency signicantly aected FAD homeostasis in these mice. Remarkably, this abnormality of fatty acid metabolism in Slc52a3/ pups was clearly improved by riboavin supplementation of their dams during pregnancy. These results suggest that disruption of Slc52a3 induced the deciency of riboavin in the fetuses and led to the failure in fatty acid -oxidation. The relevant features of this phenotype in Slc52a3/ mice were reproduced in a clinical report by Bosch et al.16. Accordingly, three patients with SLC52A3 mutation exhibited increased plasma

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 5. Blood glucose and lactate levels in Slc52a3/ neonatal mice with or without riboavin supplementation. (a) Protocol for the riboavin supplementation. In the riboavin-supplemented group(spl (+)), the pregnant dams were given 50mg/L of riboavin in drinking water ad libitum from gestational day 0. Postnatal day 0 (P0) pups were collected at 59 hr aer birth. (b) Body weights of WT and Slc52a3/ pups. (c) Blood glucose levels in WT and Slc52a3/ pups. Closed circles represent undetected data (<20mg/ dL). (d) Blood lactate levels in WT and Slc52a3/ pups. Each column represents the mean SD (WT spl(), n= 7; WT spl(+), n=5; Slc52a3/ spl(), n=6; Slc52a3/ spl(+), n= 8). Values where *P<0.05,**P< 0.01, or ***P< 0.001 indicates signicant dierence.

WT Slc52a3/

RS () RS (+) RS () RS (+) C0 48.914.6 47.93.3 55.514.1 43.87.2C2 7.703.35 8.631.17 5.671.18 4.142.96C3 0.161.28 0.110.24 0.250.39 0.150.21C4 0.200.34 1.250.99 14.737.54*** 3.884.89C5 0.090.09 0.200.11 2.811.23*** 0.310.29C8 0.040.07 0.030.07 0.140.21 0.060.09C14 0.050.07 0.030.07 1.870.69*** 0.660.61C16 0.130.15 0.050.12 0.570.74 0.310.41

Table 2. Plasma acylcarnitine levels in WT and Slc52a3/ mice with (+) and without () riboavin supplementation (RS). (unit: M). In the riboavin-supplemented group, the pregnant dams were given 50mg/L of riboavin in drinking water from gestational day 0. Postnatal day 0 (P0) pups were collected aer 59hr of birth. All data are expressed as the mean SD. Values where ***P< 0.001 indicate signicant dierence from WT (RS ()). Values where P<0.01, P< 0.001 indicate signicant dierence from Slc52a3/ (RS ()).

acylcarnitines and urine organic acids, and mimicked an inherited disorder aecting mitochondrial fatty acid -oxidation: i.e. multiple acyl-CoA dehydrogenase deciency (MADD). All three patients were found to be decient in plasma avin in spite of normal dietary intake, and the biochemical abnormalities were clearly improved

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

Figure 6. Metabolome analysis in Slc52a3/ neonatal mice. Hierarchical clustering of targeted metabolomics data. Amino acids, metabolites of glycolysis and TCA cycle in liver, plasma acylcarnitine, and blood glucose in WT, Slc52a3/ pups (Slc52a3/ spl()) and riboavin-supplemented Slc52a2/ pups (Slc52a3/ spl(+)) are listed on the abscissa. The color in the heat map reects the relative metabolite abundance level according to the z-score.

by riboavin administration16. Taken together, it is strongly suggested that disruption of Slc52a3 induced the deciency in riboavin and led to the abnormality in fatty acid -oxidation.

Newborn Slc52a3/ mice exhibited hypoglycemia, although transplacental transport of glucose in Slc52a3/ pups was not changed. In addition, amino acids including the gluconeogenic precursor alanine were not decreased in Slc52a3/ pups. It has been suggested that the hypoglycemia was not due to insufficient glucose supplement from the maternal or insufficient gluconeogenic precursors. Hypoglycemia is one of the major clinical signs of fatty acid oxidation defects both in patients and mouse models26. Lack of mitochondrial trifunctional protein (MTP), which catalyzes mitochondrial long-chain fatty acid -oxidation, leads to neonatal hypoglycemia and sudden death aer birth27. In addition, deciency in very long-chain acyl-CoA dehydrogenase (VLCAD) or medium-chain acyl-CoA dehydrogenase (MCAD) causes hypoglycemia under fasting conditions28,29. During

periods of high metabolic demand, mitochondrial fatty acid -oxidation is stimulated to preserve glucose, which is an indispensable energy source for the brain30. In LCAD knockout mice, peripheral tissues utilizes more glucose to compensate for impaired fatty acid oxidation under fasting condition31. Enhanced peripheral glycogenolysis has also been observed in patients with carnitine palmitoyltransferase II deciency or VLCAD deciency during prolonged low-intensity exercise32,33. In several -oxidation-decient mice models, hypoglycemia has also been observed. Thus, it is suggested that Slc52a3/ pups might have increased glucose demand due to impaired fatty acid oxidation, resulting in hypoglycemia. We speculated that, in utero, mother helps Slc52a3/ embryos to obtain nutrients and metabolize fatty acid.

Neonatal death is one of the most striking phenotypes in Slc52a3/ mice. Similarly, several mouse models of defects in mitochondrial -oxidation of fatty acids have been demonstrated to result in neonatal mortality26.

Disruption of the MTP gene results in fatty acid metabolism abnormality and death 636 hr aer birth27. In addition, MCAD knockout mice also manifest signicant neonatal mortality29. Although the precise mechanisms of neonatal loss in these mice remain undetermined, impaired fatty acid oxidation might be lethal in newborn pups.

Recent clinical reports have identied SLC52A3 mutations in patients with BVVLS1421. Although the patho-physiology of BVVLS is not well known, bulbar palsy, sudden hearing loss, facial weakness, respiratory compromise and muscle weakness have been frequently observed in these patients22,34. Because of early death in Slc52a3/ mouse pups, the several symptoms of BVVLS including hearing loss and muscle weakness could

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

not be clearly evaluated in this study. However, it has also been reported that some patients die in infancy or youth22,34, and fatty acid metabolism disorders have been observed in some patients14,16. The neurological symptoms and increased levels of plasma fatty acids have been improved by the administration of a high dose of riboavin in some cases14,16,18. Thus, the phenotypes of Slc52a3/ mice were similar to those in the patients. While there is no information on the status of BVVLS patients immediately aer birth, we assume that some patients with SLC52A3 mutations might have died from some unknown cause. Although further studies are needed, Slc52a3/ mice could serve as an experimental BVVLS model, and may help clarify its underlying mechanism and lead to development of new drugs for treatment of this disease.

In conclusion, Rfvt3 is involved in placental riboavin transport and regulates riboavin homeostasis in neonatal pups in mice. Disruption of Slc52a3 caused neonatal lethality with hyperlipidemia and hypoglycemia attributable to riboavin deciency. These data support that riboavin supplementation can prevent neonatal death in BVVLS patients with SLC52A3 mutations, and may help to clarify the mechanisms involved in their neurological symptoms.

Methods

Animal experiments were conducted in accordance with the Guidelines for Animal Experiments of Kyoto University. All protocols were approved by the Animal Research Committee, Graduate School of Medicine, Kyoto University. Slc52a3/ embryos (C57BL/6N-2310046K01Riktm1a(KOMP)Wtsi) were obtained from Knockout Mouse Project repository (2310046K01Rik; synonym for Slc52a3)35. The targeting vector is described in Supplementary Fig. 1a. To conrm the homologous recombination, mouse genomic DNA was isolated from tail biopsies of offspring (F1) using DNeasy Blood and Tissue Kits (Qiagen), and used for the long-range PCR. Long-range PCR analysis was performed using TaKaRa Ex Taq Hot Start Version (TAKARA BIO) according to the following profile immediately

aer an initial 3-min denaturation step at 94 C: 94 C for 30 sec, 55 C for 30 sec, 72 C for 9 min, 32 cycles. The primer sets were as follows: forward primer 5-GAGTGGCACTTGACATTCGAGGATAAGC-3 and reverse primer 5-CACAACGGGTTCTTCTGTTAGTCC-3 for the 5 homology arm, and forward primer 5-TCTATAGTCGCAGTAGGCGG-3 and reverse primer 5-GCGGCTTTTACCAAGAATATCAAG-3 for the 3 homology arm based on the sequence. Then, F1 Slc52a3 heterozygous knockout (Slc52a3+/) animals were inter-crossed to generate WT, Slc52a3+/ and Slc52a3/ mice. To determine the genotype, total DNA was isolated as described above and PCR analysis was performed with a forward primer 5-TCCTGAGGCAAAAGTGGAAA-3 and a reverse primer 5-GGCTGAAATAGCCCACCTTT-3 for detecting wild-type alleles, and a forward primer 5-GAGATGGCGCAACGCAATTAAT-3 and a reverse primer 5-GGCTGAAATAGCCCACCTTT-3 for detecting mutant alleles. PCR conditions were as follows: 94 C for 30 sec, 55 C for 30 sec, 72 C for 3 min, 32 cycles. The mice were housed in a temperature-controlled environment with an alternating 12-hr light/dark cycle, and were given a standard chow diet (F-2; Funabashi Farm) and water at libitum before experiments.

Total RNA was isolated from the kidney and placenta using RNeasy Mini Kit (Qiagen), and then was reverse-transcribed. To determine the mRNA expression level of Slc52a3, real-time PCR was performed as described previously8. TaqMan Gene Expression assays were purchased from Life Technologies: Slc52a3, Mm00510189_m1.

In situ In situ hybridization was performed by the method of Genosta, Inc. as previously described13. Hybridization was performed with Slc52a3-specic probes (sequence position; 10451645 in the open reading frame) at concentrations of 300 ng/ml in the Prove Diluent-1 (Genosta) at 60 C for 16 hr. The sections were counterstained with Kernechtrot stain solution (Muto Pure Chemicals).

+/ Transport activity for [3H] riboavin across the placenta of pregnant Slc52a3+/ mice, which were mated with Slc52a3+/ male mice, at gestation date 16.5 was determined as previously reported36. Pregnant dams were anesthetized with an intraperitoneal administration of sodium pentobarbital. Surgery was performed on each animal lying on a heating pad to maintain constant body temperature. Then, radioisotope-labeled riboavin (500nM of [3H]riboavin, 0.903 TBq/mmol, Moravek Biochemicals) or glucose (40 M of D-[U-14C]glucose, 11.5 GBq/mmol, GE Healthcare) was administrated as a bolus via the femoral vein (10mL/kg). Five minutes aer administration, fetuses and placentas were removed and weighted. Fetal tails were collected, and used for genotyping as described above. Fetuses were homogenized in saline, and the samples were placed into Ultima Gold scintillation cocktail (PerkinElmer) followed by solubilization in SOLVABLE (Packard Instrument Co.). The radioactivity was measured by liquid scintillation counting, and the levels of [3H]riboavin and D-[U-14C]glucose in each fetus were determined and their transporter activities were calculated.

/ Male and female mice, of both whose genotypes were Slc52a3+/, were intercrossed overnight, and then the successful copulation was veried by observation of a vaginal plug. Slc52a3+/ dams were supplemented with 50mg/L of riboavin in drinking water ad libitum from gestational day 0 to the day of parturition, and the P0 pups were used for the studies. P0 pups were collected immediately aer birth or 59hr aer birth. The pups were sacriced by decapitation, and the blood and tissues were collected. For the survival analysis, Slc52a3+/ dams were supplemented as described above from gestational day 0 to 3 weeks just before delivery. Aer birth, the newborn pups were also administered daily with 0.75mg/kg riboavin subcutaneously up to P21. Blood was collected from weaned mice by tail snip.

SCIENTIFIC REPORTS

9

www.nature.com/scientificreports/

Blood and urine samples were collected from P0 pups. The glutaric acid levels in urine samples were determined using a gas chromatography mass spectrometer (GC/MS) at Shimadzu Techno-Research Inc. The concentrations of riboavin, FMN, and FAD in plasma and tissue samples were measured by HPLC as described previously37.

Plasma acylcarnitine levels were measured by Electrospray tandem mass spectrometry (ESI-MS/MS) at Shimadzu Techno-Research Inc.38 with some modications. Briey, a total volume of 10L methanol/acetone/water (7:7:2) was added to 2.5 L plasma, and evaporated to dryness for 20 min at 50 C aer standing for 10 min at room temperature. The dried sample was redissolved in an aliquot of 100L internal standard solution (LABELED CARNITINE STANDARDS SET B, Cambridge Isotope Laboratories, Inc.). The sample was agitated vigorously for 1hr and centrifuged. The supernatant was evaporated for 1hr at 50C, and dissolved in 100L mobile phase (water/methanol/acetonitrile (20:40:40) containing 0.05% formic acid) for ESI-MS/MS analysis, using LCMS-8040 triple quadrupole mass spectrometer equipped with two LC-30AD HPLC pumps and SIL-30AC autosampler (Shimadzu). Blood glucose and lactate levels were determined using the NIPRO FreeStyle Blood Glucose Monitoring System (NIPRO) and Lactate Pro LT-1710 (Arkray Inc.), respectively.

Liver samples were removed from the P0 pups. Frozen tissue was plunged into 50% acetonitrile solution containing internal standards (H3304-1002, Human Metabolome Technologies) at 0 C in order to inactivate the enzymes. The tissue was then homogenized and centrifuged (2300 g, 5 min). Subsequently, the upper aqueous layer was centrifugally ltered through a Millipore 5-kDa cuto lter to remove proteins. The ltrate was centrifugally concentrated and resuspended in water for capillary electrophoresis coupled to mass spectrometry (CE-MS) analysis. Cationic compounds were measured in the positive mode of CE-TOFMS, and anionic compounds were measured in the positive and negative modes of CE-MS/ MS as previously reported3941. Metabolome measurements and hierarchical cluster analysis were performed at a service facility of Human Metabolome Technologies, Inc.

Data are expressed as means SD, and dierences were statistically analyzed by the unpaired Students t-test. Multiple comparisons were performed by one-way ANOVA with either the Bonferroni test (parametric) or Kruskal-Wallis with the Dunns test (nonparametric) using GraphPad Prism 5.0 (GraphPad Soware Inc.). For Mendelian randomization analysis, the chi-square test with 2-degree freedom was used. For the survival analysis, the dierences in survival rates were examined with the log-rank test. Dierences where P< 0.05 were considered as statistically signicant.

1. Depeint, F., Bruce, W. R., Shangari, N., Mehta, R. & OBrien, P. J. Mitochondrial function and toxicity: role of the B vitamin family on mitochondrial energy metabolism. Chem Biol Interact 163, 94112 (2006).

2. Powers, H. J. Riboavin (vitamin B-2) and health. Am J Clin Nutr 77, 13521360 (2003).3. Robitaille, J., Carmichael, S. L., Shaw, G. M. & Olney, R. S. Maternal nutrient intake and risks for transverse and longitudinal limb deciencies: data from the National Birth Defects Prevention Study, 19972003. Birth Defects Res A Clin Mol Teratol 85, 773779 (2009).

4. Smedts, H. P. et al. Maternal intake of fat, riboavin and nicotinamide and the risk of having ospring with congenital heart defects. Eur J Nutr 47, 357365 (2008).

5. Foraker, A. B., Khantwal, C. M. & Swaan, P. W. Current perspectives on the cellular uptake and trafficking of riboavin. Adv Drug Deliv Rev 55, 14671483 (2003).

6. Yonezawa, A., Masuda, S., Katsura, T. & Inui, K. Identication and functional characterization of a novel human and rat riboavin transporter, RFT1. Am J Physiol Cell Physiol 295, C632641 (2008).

7. Yamamoto, S. et al. Identication and functional characterization of rat riboavin transporter 2. J Biochem 145, 437443 (2009).8. Yao, Y. et al. Identication and comparative functional characterization of a new human riboavin transporter hRFT3 expressed in the brain. J Nutr 140, 12201226 (2010).

9. Yonezawa, A. & Inui, K. Novel riboavin transporter family RFVT/SLC52: identication, nomenclature, functional characterization and genetic diseases of RFVT/SLC52. Mol Aspects Med 34, 693701 (2013).

10. Ghosal, A., Sabui, S. & Said, H. M. Identication and characterization of the minimal 5-regulatory region of the human riboavin transporter-3 (SLC52A3) in intestinal epithelial cells. Am J Physiol Cell Physiol 308, C189196 (2015).

11. Jiang, X. R. et al. RFT2 is overexpressed in esophageal squamous cell carcinoma and promotes tumorigenesis by sustaining cell proliferation and protecting against cell death. Cancer Lett 353, 7886 (2014).

12. Subramanian, V. S. et al. Dierential expression of human riboavin transporters 1, 2, and 3 in polarized epithelia: a key role for hRFT-2 in intestinal riboavin uptake. Biochim Biophys Acta 1808, 30163021 (2011).

13. Yoshimatsu, H. et al. Functional involvement of RFVT3/SLC52A3 in intestinal riboavin absorption. Am J Physiol Gastrointest Liver Physiol 306, G102110 (2014).

14. Anand, G. et al. Early use of high-dose riboavin in a case of Brown-Vialetto-Van Laere syndrome. Dev Med Child Neurol 54, 187189 (2012).

15. Bandettini di Poggio, M. et al. A novel compound heterozygous mutation of C20orf54 gene associated with Brown-Vialetto-Van Laere syndrome in an Italian family. Eur J Neurol 20, e9495 (2013).

16. Bosch, A. M. et al. Brown-Vialetto-Van Laere and Fazio Londe syndrome is associated with a riboflavin transporter defect mimicking mild MADD: a new inborn error of metabolism with potential treatment. J Inherit Metab Dis 34, 159164 (2011).

17. Bosch, A. M. et al. The Brown-Vialetto-Van Laere and Fazio Londe syndrome revisited: natural history, genetics, treatment and future perspectives. Orphanet J Rare Dis 7, 83 (2012).

18. Ciccolella, M. et al. Brown-Vialetto-van Laere and Fazio-Londe overlap syndromes: a clinical, biochemical and genetic study. Neuromuscul Disord 22, 10751082 (2012).

19. Dezfouli, M. A., Yadegari, S., Nassi, S. & Elahi, E. Four novel C20orf54 mutations identied in Brown-Vialetto-Van Laere syndrome patients. J Hum Genet 57, 613617 (2012).

20. Green, P. et al. Brown-Vialetto-Van Laere syndrome, a ponto-bulbar palsy with deafness, is caused by mutations in c20orf54. Am J Hum Genet 86, 485489 (2010).

21. Koy, A. et al. Brown-Vialetto-Van Laere syndrome: a riboavin-unresponsive patient with a novel mutation in the C20orf54 gene. Pediatr Neurol 46, 407409 (2012).

22. Manole, A., Fratta, P. & Houlden, H. Recent advances in bulbar syndromes: genetic causes and disease mechanisms. Curr Opin Neurol 27, 506514 (2014).

SCIENTIFIC REPORTS

10

www.nature.com/scientificreports/

23. Dancis, J., Lehanka, J. & Levitz, M. Transfer of riboavin by the perfused human placenta. Pediatr Res 19, 11431146 (1985).24. Dancis, J., Lehanka, J. & Levitz, M. Placental transport of riboavin: dierential rates of uptake at the maternal and fetal surfaces of the perfused human placenta. Am J Obstet Gynecol 158, 204210 (1988).

25. Huang, S. N. & Swaan, P. W. Riboavin uptake in human trophoblast-derived BeWo cell monolayers: cellular translocation and regulatory mechanisms. J Pharmacol Exp Ther 298, 264271 (2001).

26. Spiekerkoetter, U. & Wood, P. A. Mitochondrial fatty acid oxidation disorders: pathophysiological studies in mouse models. J Inherit Metab Dis 33, 539546 (2010).

27. Ibdah, J. A. et al. Lack of mitochondrial trifunctional protein in mice causes neonatal hypoglycemia and sudden death. J Clin Invest 107, 14031409 (2001).

28. Spiekerkoetter, U. et al. Changes in blood carnitine and acylcarnitine proles of very long-chain acyl-CoA dehydrogenase-decient mice subjected to stress. Eur J Clin Invest 34, 191196 (2004).

29. Tolwani, R. J. et al. Medium-chain acyl-CoA dehydrogenase deciency in gene-targeted mice. PLoS Genet 1, e23 (2005).30. Hue, L. & Taegtmeyer, H. The Randle cycle revisited: a new head for an old hat. Am J Physiol Endocrinol Metab 297, E578591 (2009).

31. Houten, S. M. et al. Impaired amino acid metabolism contributes to fasting-induced hypoglycemia in fatty acid oxidation defects. Hum Mol Genet 22, 52495261 (2013).

32. OR, M. C., Norgaard, M. G., Sacchetti, M., van Engelen, B. G. & Vissing, J. Fuel utilization in patients with very long-chain acyl-coa dehydrogenase deciency. Ann Neurol 56, 279283 (2004).

33. Orngreen, M. C. et al. Fuel utilization in subjects with carnitine palmitoyltransferase 2 gene mutations. Ann Neurol 57, 6066 (2005).

34. Sathasivam, S. Brown-Vialetto-Van Laere syndrome. Orphanet J Rare Dis 3, 9 (2008).35. Collins, F. S., Rossant, J. & Wurst, W. A mouse for all reasons. Cell 128, 913 (2007).36. Wyrwoll, C. S., Seckl, J. R. & Holmes, M. C. Altered placental function of 11beta-hydroxysteroid dehydrogenase 2 knockout mice. Endocrinology 150, 12871293 (2009).

37. Yao, Y. et al. Involvement of riboavin transporter RFVT2/Slc52a2 in hepatic homeostasis of riboavin in mice. Eur J Pharmacol 714, 281287 (2013).

38. Shigematsu, Y. et al. Newborn mass screening and selective screening using electrospray tandem mass spectrometry in Japan. J Chromatogr B Analyt Technol Biomed Life Sci 776, 3948 (2002).

39. Soga, T. & Heiger, D. N. Amino acid analysis by capillary electrophoresis electrospray ionization mass spectrometry. Anal Chem 72, 12361241 (2000).

40. Soga, T. et al. Quantitative metabolome analysis using capillary electrophoresis mass spectrometry. J Proteome Res 2, 488494 (2003).

41. Soga, T. et al. Simultaneous determination of anionic intermediates for Bacillus subtilis metabolic pathways by capillary electrophoresis electrospray ionization mass spectrometry. Anal Chem 74, 22332239 (2002).

This study was supported in part by a grand-in-aid for Scientic Research (KAKENHI) from the Ministry of Education, Science, Culture and Sports of Japan.

H.Y., A.Y., S.M. and K.-i.I. conceived and designed the experiments. H.Y., A.Y., K.Y. and K.M. mainly contributed to writing the manuscript. H.Y., K.Y., Y.Y., K.S., S.N., S.I., T.O., T.N. and I.Y. carried out and analyzed all experiments. All the authors participated in the discussion of the results, and all had reviewed the manuscript.

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Yoshimatsu, H. et al. Disruption of Slc52a3 gene causes neonatal lethality with riboavin deciency in mice. Sci. Rep. 6, 27557; doi: 10.1038/srep27557 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

11