Effect of riboflavin deficiency on development of the cerebral cortex in Slc52a3 knockout mice

Riboflavin transporter 3 (RFVT3), encoded by the SLC52A3 gene, is important for riboflavin homeostasis in the small intestine, kidney, and placenta. Our previous study demonstrated that Slc52a3 knockout (Slc52a3−/−) mice exhibited neonatal lethality and metabolic disorder due to riboflavin deficiency. Here, we investigated the influence of Slc52a3 gene disruption on brain development using Slc52a3−/− embryos. Slc52a3−/− mice at postnatal day 0 showed hypoplasia of the brain and reduced thickness of cortical layers. At embryonic day 13.5, the formation of Tuj1+ neurons and Tbr2+ intermediate neural progenitors was significantly decreased; no significant difference was observed in the total number and proliferative rate of Pax6+ radial glia. Importantly, the hypoplastic phenotype was rescued upon riboflavin supplementation. Thus, it can be concluded that RFVT3 contributes to riboflavin homeostasis in embryos and that riboflavin itself is required during embryonic development of the cerebral cortex in mice.

Cell proliferation of radial glia in E13.5 Slc52a3−/− embryos. Radial glia in the ventricular zone were stained with antibodies against Pax6 and cell proliferation was determined by bromodeoxyuridine (BrdU) and Ki67 detection (Fig. 4a). No difference in thickness of the ventricular zone ( Fig. 4b) or number of Pax6 + radial glia ( Supplementary Fig. 4d) was observed between WT and Slc52a3−/− embryos, which was supported by RT-PCR analyses ( Supplementary Fig. 4e). Similarly, no difference between WT and Slc52a3−/− embryos was noted regarding the extent of the BrdU + area in the ventricular zone (Fig. 4c). The number of Ki67 + cells was similar in WT and Slc52a3−/− embryos (Fig. 4d). Pax6/Tbr2 double immunofluorescence revealed that the ratio of Pax6 + cells expressing Tbr2 was significantly lower in Slc52a3−/− embryos (Fig. 4e). In contrast, no statistically significant difference was found in the ratios of Tbr2 + BrdU + to Tbr2 + cells (Fig. 4f) and Ki67 + BrdU + to Ki67 + cells (Fig. 4g) between WT and Slc52a3−/− embryos.

Discussion
In the present study, we investigated the role of riboflavin in cerebral cortex development in mice. Slc52a3−/− mice exhibited abnormal brain morphology as a consequence of lower riboflavin levels. Histological analysis of the cerebral cortex showed decreased thickness of cortical layers at P0 and E13.5. Slc52a3−/− embryos displayed fewer immature neurons and intermediate neural progenitors in the cerebral cortex, and none of them showed signs of cell death. Importantly, brain hypoplasia was rescued by riboflavin supplementation. Taken together, these findings indicate that RFVT3 contributes to riboflavin homeostasis in embryos and plays an important role during embryonic development of the cerebral cortex in mice. The six-layered mammalian cerebral cortex is formed through the coordinated processes of neurogenesis and migration (Fig. 6a) 29 . During cortical neurogenesis, radial glia undergo proliferative divisions aimed at selfrenewal and neurogenic divisions to generate either neurons or intermediate neural progenitors. Progenitors in the ventricular and subventricular zones divide and give rise to neurons specific to each layer. Early-born neurons populate layers VI and V, whereas late-born neurons migrate past them to progressively populate layers IV and II/IV. Immunostaining of P0 Slc52a3−/− brain sections showed that the thickness of each cortical layer, including the intermediate zone, and the number of layer neurons were significantly decreased in absolute but not in relative terms. Disruption of the Slc52a3 gene was associated with a reduction in the number of neurons in every cortical layer, instead of just a specific one. At E13.5, Tbr2 + intermediate neural progenitors and Tuj1 + neurons were significantly fewer but without signs of cell death. In contrast, there was no change in the population of Pax6 + radial glia and BrdU + cell proliferation. Pax6/Tbr2 double immunofluorescence revealed that the proportion of Pax6 + cells expressing Tbr2 was significantly lower in Slc52a3−/− embryos. However, the proportion of Tbr2 + BrdU + and Tbr2 + Tuj1 + cells were similar in WT and Slc52a3−/− embryos. These results suggest that the transition from intermediate neural progenitors to neurons and the proliferation of intermediate neural progenitors were similar in WT and Slc52a3−/− embryos; however, differentiation from radial glia to intermediate neural progenitors decreased in Slc52a3−/− embryos. The sequential expression of Pax6 and Tbr2 is linked to differentiation from radial glia to intermediate neural progenitors 30 . Intermediate neural progenitors produce glutamatergic projection neurons for all cortical layers 31 . Therefore, riboflavin deficiency during cortical neurogenesis influences the neuronal differentiation of radial glia, but not cell proliferation, causing atrophy of cortical neurons (Fig. 6b). The molecular mechanisms underlying riboflavin-dependent neuronal differentiation will be elucidated in future studies.
Disruption of the Slc52a3 gene affects the development of the cerebral cortex even though RFVT3 is hardly expressed in the brain. Our previous study showed that RFVT3 was highly expressed in the placenta and the placental [ 3 H]riboflavin transport capacity in Slc52a3−/− fetuses was significantly decreased at E16.5 12 . Reduction of placental transport to Slc52a3−/− fetuses decreased riboflavin levels in fetal plasma and, consequently, in their tissues. In this study, excess riboflavin supplementation by subcutaneous tablet implantation and oral administration to the mother rescued the hypoplasia phenotype in developing embryos owing to passive diffusion of riboflavin through the placenta. The SLC52A2 gene, which encodes RFVT2, is expressed in several tissues including the brain 9 , where it affects tissue distribution of riboflavin. Hence, disruption of Slc52a3 could decrease the transport of riboflavin via the placenta to the fetus without affecting its tissue distribution. www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ BVVL syndrome patients with mutations in SLC52A3 commonly present with bulbar palsy, hearing loss, muscle weakness, and respiratory symptoms in infancy or later in childhood 26,27 . Neurological diagnostic tests have shown peripheral neuropathy, anterior horn dysfunction, and chronic denervation in most cases. Although brain magnetic resonance imaging (MRI) revealed no remarkable changes, abnormal T2-weighted intensities have been sometimes described in cerebellar, cortical, subcortical (basal ganglia and external capsule), and brainstem regions of some patients with SLC52A3 mutations 27 . Brain malformations, especially cerebral cortical hypoplasia found in Slc52a3−/− mice, are similar to those in the above patients. Recent neuropathological features of the central nervous system in two patients with different SLC52A3 mutation types reflected the above clinical phenotype, although the degree of atrophy of cerebellar Purkinje cells was somewhat different between the two patients. It has been suggested that different mutations might influence the length of the disease and degree of atrophy of specific brain structures 15 . In future studies, model mice with knock-in patient mutations will clarify the relationship between the pathology and symptoms of each BVVL syndrome patient.
In conclusion, RFVT3 contributes to riboflavin homeostasis in embryos via placental transport. The Slc52a3 gene disruption resulted in atrophy of the cerebral cortex, and this was associated with abnormal differentiation of radial glia during the embryonic period. Hence, riboflavin appears to play an important role in the embryonic development of the cerebral cortex.

Methods
Mice. Slc52a3−/− mice in a C57BL/6 background (C57BL/6N-2310046K01Rik tm1a(KOMP)Wtsi ) were obtained from the Knockout Mouse Project Repository and were generated as described previously 12 . Mice were maintained under a 12-h light/dark cycle and provided a standard chow diet (F-2; Funabashi farm) and water ad libitum before experiments. To determine the genotype of the offspring, genomic DNA was isolated from tail biopsies, and PCR analysis was performed as previously reported 12 . Heterozygous male and female mice were mated www.nature.com/scientificreports/ overnight, and vaginal plugs were checked the following morning. Plug detection was considered to correspond to day 0.5 of pregnancy (E0.5). Brains were dissected at E13.5 and P0. 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 (permission number: Med-Kyo20121).
The TUNEL assay was carried out on brain sections from P0 pups and E13.5 embryos using a MEBSTAIN Apoptosis TUNEL Kit (MBL) in accordance with the manufacturer's instructions.
In situ hybridization was performed as previously described 12  Image analysis and quantification. Coronal sections containing the dorsal telencephalon close to the diencephalon were used for quantification. The thickness of the total cortex was defined as the distance between the apical surface and basal lamina. The thickness of the intermediate zone was defined as the distance between the cortical plate and the ventricular zone. The thickness of the Cux1 + , Ctip2 + , Tuj1 + , and Pax6 + layers and the cell numbers were measured based on confocal micrographs using NIS-Elements AR analysis software. Quantification of Tbr2 + and BrdU + areas was performed based on confocal micrographs using ImageJ software. For each marker, the thickness of a particular area relative to the total cortex area or ventricular zone was calculated and expressed as a percentage.
Measurement of riboflavin, FMN, and FAD. Brain samples were collected from P0 pups. The concentrations of riboflavin, FMN, and FAD in brain samples were measured by high-performance liquid chromatography as described previously 12 .
Real-time PCR. Real-time PCR was performed as described previously 12 . For Slc52a3 and Slc52a2 expression analysis, total RNA was isolated from WT mouse brains at P0 and E13.5 using the RNeasy Mini Kit (Qiagen) and then reverse transcribed. TaqMan gene expression assays for Slc52a3 (Mm00510189_m1) and Slc52a2 (Mm01205717_g1) were from Life Technologies. For Tbr2 and Pax6 analysis, total RNA was isolated from the cerebral cortex using the RNeasy Mini Kit and reverse transcribed. TaqMan gene expression assays for Tbr2 (Mm01351985_m1) and Pax6 (Mm00443081_m1) were from Thermo Fisher Scientific. Relative expression levels were normalized to glyceraldehyde 3-phosphate-dehydrogenase (GAPDH) (Mm99999915_g1).
Statistics. Statistical analyses were performed using GraphPad Prism software (version 8.3.1). All values are expressed as the mean ± S.E.M. Differences were analyzed for significance using an unpaired Student's t-test. Significance between variables is shown based on the P-value obtained (*P < 0.05, **P < 0.01, ***P < 0.001).