MFSD7c functions as a transporter of choline at the blood–brain barrier

Nguyen, Xuan Thi Anh; Le, Thanh Nha Uyen; Nguyen, Toan Q.; Thi Thuy Ha, Hoa; Artati, Anna; Leong, Nancy C. P.; Nguyen, Dat T.; Lim, Pei Yen; Susanto, Adelia Vicanatalita; Huang, Qianhui; Fam, Ling; Leong, Lo Ngah; Bonne, Isabelle; Lee, Angela; Granadillo, Jorge L.; Gooch, Catherine; Yu, Dejie; Huang, Hua; Soong, Tuck Wah; Chang, Matthew Wook; Wenk, Markus R.; Adamski, Jerzy; Cazenave-Gassiot, Amaury; Nguyen, Long N.

doi:10.1038/s41422-023-00923-y

Download PDF

Article
Open access
Published: 02 February 2024

MFSD7c functions as a transporter of choline at the blood–brain barrier

Xuan Thi Anh Nguyen¹^na1,
Thanh Nha Uyen Le¹^na1^nAff17,
Toan Q. Nguyen¹,
Hoa Thi Thuy Ha¹,
Anna Artati²,
Nancy C. P. Leong¹,
Dat T. Nguyen¹,
Pei Yen Lim^1,3,
Adelia Vicanatalita Susanto^1,4,5,
Qianhui Huang ORCID: orcid.org/0009-0008-1390-6713¹,
Ling Fam¹,
Lo Ngah Leong⁶,
Isabelle Bonne^6,7,8,
Angela Lee⁹,
Jorge L. Granadillo⁹,
Catherine Gooch⁹,
Dejie Yu¹⁰,
Hua Huang^10,11,12,13,
Tuck Wah Soong^10,11,12,13,
Matthew Wook Chang^1,4,5,
Markus R. Wenk^1,3,
Jerzy Adamski^1,14,15,
Amaury Cazenave-Gassiot^1,3 &
…
Long N. Nguyen ORCID: orcid.org/0000-0002-9857-2239^1,3,8,13,16

Cell Research volume 34, pages 245–257 (2024)Cite this article

2509 Accesses
50 Altmetric
Metrics details

Subjects

Abstract

Mutations in the orphan transporter MFSD7c (also known as Flvcr2), are linked to Fowler syndrome. Here, we used Mfsd7c knockout (Mfsd7c^–/–) mice and cell-based assays to reveal that MFSD7c is a choline transporter at the blood–brain barrier (BBB). We performed comprehensive metabolomics analysis and detected differential changes of metabolites in the brains and livers of Mfsd7c^–/–embryos. Particularly, we found that choline-related metabolites were altered in the brains but not in the livers of Mfsd7c^–/– embryos. Thus, we hypothesized that MFSD7c regulates the level of choline in the brain. Indeed, expression of human MFSD7c in cells significantly increased choline uptake. Interestingly, we showed that choline uptake by MFSD7c is greatly increased by choline-metabolizing enzymes, leading us to demonstrate that MFSD7c is a facilitative transporter of choline. Furthermore, single-cell patch clamp analysis showed that the import of choline by MFSD7c is electrogenic. Choline transport function of MFSD7c was shown to be conserved in vertebrates, but not in yeasts. We demonstrated that human MFSD7c is a functional ortholog of HNM1, the yeast choline importer. We also showed that several missense mutations identified in patients exhibiting Fowler syndrome had abolished or reduced choline transport activity. Mice lacking Mfsd7c in endothelial cells of the central nervous system suppressed the import of exogenous choline from blood but unexpectedly had increased choline levels in the brain. Stable-isotope tracing study revealed that MFSD7c was required for exporting choline derived from lysophosphatidylcholine in the brain. Collectively, our work identifies MFSD7c as a choline exporter at the BBB and provides a foundation for future work to reveal the disease mechanisms of Fowler syndrome.

Brain endothelial GSDMD activation mediates inflammatory BBB breakdown

Article 17 April 2024

Small extracellular vesicles from young plasma reverse age-related functional declines by improving mitochondrial energy metabolism

Article Open access 16 April 2024

Antisense oligonucleotide therapeutic approach for Timothy syndrome

Article Open access 24 April 2024

Introduction

Missense mutations in MFSD7c (also known as FLVCR2) have been associated with Fowler syndrome.¹ The major clinical symptoms reported in these patients are the severe dilation of cerebral blood vessels and mild microcephaly. Most severe cases result in embryonic death, but some patients have survived after birth.^1,2,3,4,5,6 Although the cause of Fowler syndrome is unclear, it is linked to the loss of functions of MFSD7c, an orphan transporter. To gain a better understanding of the disease mechanisms, the molecular role of MFSD7c must be elucidated. We recently reported that mice with a global deletion of Mfsd7c had late gestation lethality with severe dilation of central nervous system (CNS) vasculature.² Specific deletion of Mfsd7c in endothelial cells also resulted in similar phenotypes, indicative of its critical roles in the blood vessels.⁷ Mfsd7c knockout (KO) embryos exhibited multiple neurological phenotypes that are probably linked to primary defects of the CNS vasculature. These phenotypes resemble the glomeruloid structures in the CNS blood vessels of patients with Fowler syndrome.^2,7

MFSD7c belongs to the Major Facilitator Superfamily (MFS) of transporters which facilitate the movement of small molecules through cell membranes.⁸ MFSD7c was reported to exert heme import activity.⁹ However, this observation has not been confirmed in the literature. Recently, it has been shown that MFSD7c is involved in thermogenesis in response to heme. Besides its localization in the mitochondria, this study and our results also showed that MFSD7c was expressed in the plasma membrane.^2,10 Furthermore, we and others showed that MFSD7c was expressed in endothelial cells of the CNS vasculature.^2,7 These results suggest that MFSD7c is a membrane transporter for small molecules at the blood–brain barrier (BBB). Utilizing mouse KO models coupled with comprehensive metabolomics analyses, we revealed the specific changes in metabolites due to the loss of Mfsd7c. These results guide us to demonstrate that MFSD7c is a transporter for choline. We show that MFSD7c facilitates choline transport via the plasma membrane in a concentration-dependent manner. Unexpectedly, we find that lack of MFSD7c at the BBB causes choline accumulation in the brain. A stable isotopic tracing study reveals that choline from plasma lysophosphatidylcholine (LPC) is delivered to the brain. The excessive amount of choline is exported out of the brain via MFSD7c. The current work identifies MFSD7c as a choline transporter and provides a foundation for future studies to reveal the disease mechanisms of Fowler syndrome. Additionally, the identification of MFSD7c as a choline transporter at the BBB lays the groundwork for future research to reveal the metabolic fates of choline in the brain.

Results

Lack of Mfsd7c affects choline levels in the brain

Deletion of Mfsd7c resulted in late gestation lethality in mice.^2,7 Because MFSD7c is predicted to be a membrane transporter, its deletion is anticipated to cause insufficient transport of essential nutrients in and/or out of the brain. To deorphanize the physiological ligands for MFSD7c, we performed a comprehensive analysis of metabolites in the brains of wild-type (WT) and global Mfsd7c KO (Mfsd7c^–/–) embryos at gestation day 14.5 (E14.5), a time point when KO embryos already exhibited vascular growth defects.² This analysis encompassed more than 520 metabolites from metabolic pathways for carbohydrates, lipids, amino acids, and cofactors (Supplementary information, Tables S1–S10). First, we found that the levels of glycolytic metabolites such as lactate were significantly increased, whereas the levels of 3-phosphoglycerate and phosphoenolpyruvate were reduced (Fig. 1a). These metabolic changes are congruent with the increased glycolysis and possibly reduced mitochondrial activity as a consequence of hypoxia that occurs in the brain of Mfsd7c^–/– embryos.^2,7 MFSD7c was shown to regulate heme transport.¹⁰ However, in our data, heme and bilirubin levels in fetal brains were comparable between Mfsd7c^–/– and WT embryos (Supplementary information, Fig. S1a, b). Second, we detected increased levels of several long-chain acylcarnitines (Fig. 1b, c; black arrowheads), whereas the levels of L-carnitine were significantly reduced in the brain of Mfsd7c^–/– embryos (Fig. 1d). Nevertheless, we ruled out that MFSD7c transports these acylcarnitines by performing the import assay with palmitoyl-carnitine (C16:0 carnitine) and octanoyl-carnitine (C8:0 carnitine) (Supplementary information, Fig. S1c, d). Again, these changes might reflect the defects of mitochondrial activity due to hypoxia in the brain of KO embryos. Indeed, mitochondrial morphology and the expression level of several mitochondrial markers were reduced in the brain of Mfsd7c^–/– embryos (Supplementary information, Fig. S2). Interestingly, we noted increased levels of choline, palmitoylcholine, and oleoylcholine in the brains of Mfsd7c^–/– embryos relative to WT embryos (Fig. 1b; red arrowheads). In contrast, the levels of CDP-choline were reduced in Mfsd7c^–/– embryos (Fig. 1d). These changes in choline metabolites were specific to the brain as their levels were unaltered in the livers from Mfsd7c^–/– embryos (Fig. 1e). Instead, we detected a striking increase in the levels of several neutral lipids such as monoacylglycerols and diacylglycerols in the livers of Mfsd7c^–/– embryos (Fig. 1f, arrowheads). Deficiency of choline has been linked to mitochondrial defects.^11,12,13,14 These differential changes in the levels of choline and choline metabolites in the brain and the accumulation of neutral lipids in the livers of Mfsd7c^–/– embryos suggest that MFSD7c may be involved in choline metabolism.

**Fig. 1: Comprehensive metabolomics analysis identifies changes in the levels of choline and choline metabolites in the brain of *Mfsd7c* KO embryos.**

MFSD7c mediates the transport of choline

The accumulation of choline and decrease in CDP-choline levels in the brains of Mfsd7c^–/– embryos suggest that MFSD7c is important for the regulation of choline levels. Thus, we tested whether MFSD7c regulates choline levels by a transport mechanism. We overexpressed human (hMFSD7c) or mouse Mfsd7c (mMfsd7c) in HEK293 cells and utilized radioactive [³H]-choline for transport assays. Interestingly, overexpression of hMFSD7c or mMFSD7c led to 1.5–2-fold increase of intracellular levels of radioactive signals, implicating that MFSD7c mediates the import of choline (Fig. 2a). We validated our choline transport assay by including neuronal choline transporter CHT1 (also known as SLC5a7) as a positive control (Fig. 2a). The import of choline mediated by hMFSD7c was increased with increasing levels of choline in medium and with increased time (Fig. 2b, c). To reaffirm that expression of MFSD7c is necessary for choline import to the cells, we included the missense mutation S203Y in our transport assays. In comparison to WT MFSD7c, the S203Y mutant exhibited abolished choline transport activity when incubated with the indicated concentrations of choline (Fig. 2b, c). The reduced import of choline of the mutant was not due to reduced protein expression levels nor defective localization in the plasma membrane (Supplementary information, Fig. S3a, b). MFSD7c is conserved from fish to humans, but not in yeasts (Supplementary information, Fig. S3c). We tested the choline import function for several MFSD7c orthologs including the ones from zebrafish (DaMfsd7c_a and DaMfsd7c_b isoforms), medaka fish (MeMfsd7c), and frog (XeMfsd7c). Consistently, these MFSD7c orthologs also exhibited choline transport activity (Fig. 2d). In the yeast S. cerevisiae, HNM1 was shown to import choline as well as L-carnitine.¹⁵ We performed complementation assays of hMFSD7c in Hnm1 KO S. cerevisiae. First, we showed that Hnm1 KO cells exhibited reduced choline uptake compared to WT yeast cells (Fig. 2e). Second, when WT yeast cells and Hnm1 mutant yeast cells were overexpressed with hMFSD7c, they both exhibited significantly increased choline uptake compared to parental cells (Fig. 2e; Supplementary information, Fig. S4a, b). These results indicate that hMFSD7c can rescue the loss of yeast Hnm1 and reaffirm that MFSD7c is a choline transporter. Choline transport by neuronal choline transporter CHT1 is inhibited by hemicholinium-3 (HC-3), a competitive inhibitor of CHT1.^16,17 HC-3 inhibited CHT1, but not MFSD7c, suggesting that these two choline transporters exert different transport properties (Fig. 2f). SLC44a1 and SLC44a2 have been reported to be choline transporters. However, we were unable to show choline import activity for these transporters under the tested conditions (Supplementary information, Fig. S4c, d). MFSD7c preferred choline as a substrate as it did not transport betaine, a choline metabolite that is also found in the brain (Fig. 2g, h). Similarly, Mfsd7c did not transport L-carnitine, acetyl-carnitine, ethanolamine, and acetylcholine under the tested conditions (Fig. 2i–k; Supplementary information, Fig. S4e). Together, our results reveal that MFSD7c is a conserved transporter for choline in vertebrates.

**Fig. 2: Expression of MFSD7c increases choline uptake.**

MFSD7c functions as a facilitative transporter for choline

We noted that the transport activity of MFSD7c was slightly increasing over time (Fig. 2c), suggesting that intracellular levels of choline might affect the influx of choline. Thus, we characterized its transport properties by testing whether MFSD7c utilizes cations for transport activity. Replacement of sodium with lithium did not affect choline import by MFSD7c (Fig. 3a; Supplementary information, Fig. S5a). Next, we tested whether lowering intracellular choline levels could increase the influx of choline via MFSD7c. To test this hypothesis, we co-expressed hMfsd7c with human choline kinase A (hChKA), which phosphorylates choline to phosphorylcholine, thus lowering the intracellular concentration of choline. Remarkably, co-expression of hMfsd7c with hChKA increased choline uptake by approximately 9-fold (Fig. 3b), whereas sole expression of hMfsd7c increased choline uptake by 1.5–2-fold (Fig. 2a). In the presence of hChKA, the import of choline was significantly increased over time (Fig. 3c). Co-expression of hChKA with hMfsd7c also greatly enhanced choline uptake in a dose-dependent manner (Fig. 3d; Supplementary information, Fig. S5b). The kinetics (Km) and Vmax for choline import by human Mfsd7c under this condition were 100 µM and 0.117 µmol/well/60 min, respectively. Choline uptake by the S203Y mutant in the absence of hChKA was completely inhibited (Fig. 2b, c). Interestingly, we noted that there was a slight increase in choline import in the S203Y mutant when co-expressed with hChKA, indicating that the transport activity of S203Y is severely delayed, but not abolished (Fig. 3d; Supplementary information, Fig. S5c). The Km and Vmax for importing choline by S203Y mutant under this condition were 1197 µM and 0.039 µmol/well/60 min, respectively. We also examined whether the expression of choline acetyltransferase (CHAT), a neuronal enzyme for acetylcholine synthesis, would increase choline uptake by converting choline into acetylcholine. Indeed, co-expression of CHAT and hMfsd7c greatly increased choline uptake (Fig. 3e). In the absence of ETNK1, an ethanolamine kinase, MFSD7c did not increase uptake of ethanolamine (Fig. 2k). However, transport of ethanolamine by MFSD7c was increased to a significant level when ETNK1 was co-expressed (Fig. 3f). Interestingly, co-expression of hMfsd7c with hChKA also slightly increased L-carnitine uptake (Fig. 3g). However, L-carnitine was unable to compete for choline import, suggesting that L-carnitine is a weak ligand (Supplementary information, Fig. S5d). Furthermore, we tested whether the transport of choline by MFSD7c is bidirectional. Indeed, our results showed that expression of hMfsd7c was required for the release of intracellular choline (Fig. 3h). Choline is a positively charged molecule. The import of choline would increase the membrane potential. To this end, we performed a whole-cell patch clamp of HEK293 cells co-expressing hMfsd7c or S203Y mutant with hChKA. Cells with overexpression of hChKA alone were used as a control (Supplementary information, Fig. S5e). Incubation of control cells or cells expressing S203Y mutant with 100 or 200 µM choline did not significantly increase membrane potential (Fig. 3i). In contrast, expression of hMfsd7c significantly increased membrane potential when 100 or 200 µM choline was included in the bath solution (Fig. 3i, j). Interestingly, the removal of choline reversed the increased membrane potential, suggesting that the influx of choline generates the membrane potential. These results show that the import of choline by MFSD7c is electrogenic. Collectively, our results show that MFSD7c facilitates the transport of choline and it behaves like a channel for the movement of choline via the plasma membrane.

**Fig. 3: MFSD7c exhibits properties of a facilitative transporter for choline.**

Mfsd7c mediates choline export at the BBB

Choline uptake at the BBB has been documented in previous studies.^18,19,20 However, the identity of the long-sought choline transporter was unknown. To test whether MFSD7c plays a role in choline transport in vivo, we generated conditional KO mice for Mfsd7c in the endothelial cells using Cdh5-CreERT² (Fig. 4a). Next, we intravenously injected control (Mfsd7c^f/f) and Mfsd7c^f/fCdh5-CreERT² (hereafter, EcMfsd7c-KO) mice with a dose (2 mM in blood) of radioactive choline and measured the distribution of radioactive signals in brains, lungs, livers, hearts, and kidneys after 30 min. We chose this supraphysiological dose as the half-life of choline in plasma is within 3 min.²¹ The radioactive choline levels in blood were not different between the two genotypes after 5 min choline injection, indicating that choline was equally delivered to the blood of the mice (Fig. 4b). We found that radioactive signals were significantly reduced in the brains of EcMfsd7c-KO mice compared to control mice (Fig. 4c). However, we noted that the majority of injected choline was present in peripheral organs such as livers and kidneys; and the radioactivity in the organs was comparable between the genotypes, indicating that reduced choline import to the brain was specific in EcMfsd7c-KO mice (Fig. 4d–g). These results indicate that Mfsd7c can mediate the uptake of exogenous choline at the BBB, a result which is consistent with the findings in the literature.^18,19,20 Nevertheless, the injected amount of choline was well above physiological concentrations of choline in blood.^22,23 Perhaps, Mfsd7c mediates to import a small amount of choline to the brain at this supraphysiological dose. Of note, the radioactive signal in the brain accounted for 2%–3% of total radioactive signals.

**Fig. 4: MFSD7c regulates choline levels in the brain.**

Our metabolomic analysis showed that the levels of choline in the brain of Mfsd7c^–/– embryos were elevated (Fig. 1d). Mfsd7c is expressed in both sides of the CNS endothelial cells, which might suggest that Mfsd7c is also required for choline import from the brain parenchyma.² To gain insights into the mechanism by which choline level is elevated in the brain of Mfsd7c-KO embryos, we injected deuterated choline (choline-d9) into the circulation of Mfsd7c pregnant dams and harvested Mfsd7c-KO embryos for lipidomic analysis of choline-containing lipids and choline (Fig. 4h). We found that the total levels of endogenous phospholipids such as LPC, phosphatidylcholine (PC), and sphingomyelin, as well as the deuterated phospholipids containing choline-d9 in the brains and livers of KO and control littermates, were comparable (Supplementary information, Table S11). Interestingly, we found that the levels of endogenous choline and choline-d9 were significantly elevated in the brains but not livers of KO embryos (Fig. 4i; Supplementary information, Table S11). Free choline imported to the brain is expected to be suppressed in the KO embryos. However, these results suggest that choline imported to the brains of Mfsd7c-KO embryos must be via other routes. We hypothesize that LPC delivers choline to the brain of Mfsd7c-KO embryos (Fig. 4h). The above data suggest a possibility that MFSD7c is also necessary for the export of choline from the brain parenchyma.

LPC is one of the most abundant lipids in the plasma^24,25 and is imported to the brain by Mfsd2a via the BBB for phospholipid synthesis.^26,27 To test our hypothesis that choline from LPC is accumulated in the brain of EcMfsd7c-KO mice, we injected a dose (300 µM in blood) of deuterated LPC-d49 which consists of choline-d13, palmitate-d31, and glycerol-d5 into the circulation of EcMfsd7c-KO and control mice. We then employed mass spectrometry (MS) for detection of choline-d13 and deuterated LPC and PC (Fig. 4j). In the plasma, a part of LPC-d49 was remodeled to produce LPC-d31 and LPC-d13. However, LPC-d49 was still the major deuterated LPC in the blood of the mice at 2 h post-injection (Fig. 4k). In the brain, we were able to detect deuterated PC species, which were present with similar amounts in the phospholipid pool of EcMfsd7c-KO and control mice. These results indicated that plasma LPC-d49 was equally imported to the brains for PC synthesis (Fig. 4l; Supplementary information, Table S12). Remarkably, we found that most of the deuterated PC species in the brains of the mice contained palmitate-d31 (PC-d31), instead of the intact form of LPC-d49 (PC-d49) (Fig. 4l; Supplementary information, Table S12). This indicated that there was a dramatic remodeling of plasma-derived LPC in the brain. As a result, the level of choline-d13 in the brains of EcMfsd7c-KO mice was significantly increased compared to that of control mice (Fig. 4m). Consistently, the levels of endogenous choline were confirmed to be elevated in EcMfsd7c-KO mice in the same analysis (Fig. 4m; Supplementary information, Fig. S6a, b), whereas the levels of choline-d13 in the plasma and livers of EcMfsd7c-KO and control mice were comparable, highlighting a specific elevation of choline in the brain of EcMfsd7c-KO mice (Supplementary information, Table S12). Furthermore, the levels of acetylcholine were increased in the brains of EcMfsd7c-KO adult mice, especially from mice fed with a choline-deficient diet (CDD) (Supplementary information, Fig. S6a, b). Since free choline import to the brain is likely suppressed or reduced in EcMfsd7c-KO mice (Fig. 4c), these results reveal an unexpected finding that excessive choline from LPC/PC metabolism is to be transported out of the brain and this mechanism is mediated by the activity of MFSD7c at the BBB.

Missense mutations of MFSD7c result in reduced choline transport activity

Missense mutations of MFSD7c have been reported in patients with Fowler syndrome. However, it remains unknown whether these mutations cause a loss or gain of choline function of MFSD7c. Thus, we performed mutagenesis to generate several homozygous and compound heterozygous missense mutations including S203Y, T430R, and T430M, and tested their transport activity (Fig. 5). These homozygous missense mutations S203Y, T430R, and T430M were chosen because the first two former mutations caused lethal Fowler syndrome, whereas the latter mutation caused non-lethal Fowler syndrome.^1,2,28,29 We also examined the activity of missense mutations P340L and T430A that were reported in a non-lethal Fowler patient.² First, we confirmed that these missense mutations did not affect the expression and localization of MFSD7c, except for L483R (Fig. 5a, b). Second, we measured choline transport activity and found that most of these missense mutations of MFSD7c exhibited significantly reduced or abolished choline transport activity (Fig. 5c). A patient homozygous for S203Y was aborted at 20 weeks during gestation.² Choline transport activity of this missense mutation was abolished (Fig. 5c). The T430M mutation was associated with non-lethal Fowler syndrome, whereas homozygous T430R mutation was reported to be lethal.^1,4 We showed that T430M had ∼76.9% activity compared to that of WT protein, whereas T430R is an inactive mutant (Fig. 5c). Previously, compound heterozygous mutants (P340L and T430A) were found in a 2-year-old patient, who is still living.² His brain Magnetic Resonance Imaging (MRI) showed mild microcephaly. We found that both alleles had partially decreased choline transport activity (Fig. 5c). The current study also identified a patient with phenotypes associated with Fowler syndrome. Whole exome sequencing identified two missense mutations of the MFSD7c coding sequence that led to a substitution of proline at position 276 to serine (P276S) inherited from the mother and leucine at position 483 to arginine (L483R) from the father (Fig. 5d–f). The patient exhibited severe microcephaly with neurological disorders (Fig. 5f). Transport activity of these missense mutants showed that P276S retained 58.2% choline transport activity, whereas L483R is likely unstable, leading to a severe reduction of activity in MFSD7c (Fig. 5c). Expending these results, we also tested the missense mutations that have been reported in the literature. We found that most lethal mutations caused a significant reduction of choline transport activity (Supplementary information, Fig. S7 and Table S13). Nevertheless, several missense mutations led to a partial loss of choline transport activity, but the patients carrying these mutations exhibited severe clinical features. It is unclear whether other factors also contribute to the pathological conditions in these cases. Together, these results suggest that reduced choline transport activity of MFSD7c may be associated with the pathogenesis of Fowler syndrome. The discovery of the molecular function of MFSD7c as a choline transporter is an important step to facilitate the understanding of the disease mechanisms in patients with Fowler syndrome.

**Fig. 5: Mutations of *Mfsd7c* affect the choline transport function.**

Although our results show that accumulation of choline might be a causal factor, we took a systemic approach to address whether deficiency of choline or choline accumulation in the brain is linked to the phenotypes. First, if choline is imported into the brain for phospholipid synthesis, it is expected that the brain phospholipid profile of Mfsd7c KO mice must be reduced. We performed a lipidomic analysis of the brains from Mfsd7c KOs and controls. However, there were no significant changes in the levels of major phospholipids containing choline such as LPC, PC, and SM in Mfsd7c KO embryos and adult mice (Supplementary information, Fig. S8a, b and Tables S14–S17). Additionally, we compared the brain transcriptomes of KO embryos to those of WT embryos obtained from dietary choline-deficient mice as we argued that deficiency of choline would recapitulate the deletion of Mfsd7c (Supplementary information, Fig. S9 and Table S18). Nevertheless, we did not observe any significant change in gene expression from WT embryos obtained from dams fed with a CDD compared to that of Mfsd7c KO embryos. These results indicate that deficiency of choline is unlikely to be the cause for the phenotypes in the brain of Mfsd7c^–/– embryos. Second, we tested whether the depletion of choline from diets would improve the phenotypes in Mfsd7c^–/– embryos. We depleted choline from the diets of pregnant Mfsd7c^–/– mice and examined the CNS blood vessels of Mfsd7c^–/– embryos. We found that the depletion of choline slightly, but significantly reduced the dilation of blood vessels in the cortices of E14.5 Mfsd7c^–/– embryos (Supplementary information, Fig. S10). Nonetheless, maternal choline deficiency did not prevent the migratory defects of the blood vessels to the ganglionic eminence regions of these Mfsd7c^–/– embryos (Supplementary information, Fig. S10d), probably due to the dominant source of choline from LPC. Thus, a dietary intervention approach to reduce brain choline levels might not be an effective approach to reverse the phenotypes of Mfsd7c^–/– embryos.

Discussion

The import of choline to the brain via the BBB was evidenced several decades ago.^18,19,20,30 Most of these prior studies used in situ rat models and reported that exogenous choline is rapidly taken up to the brain. These experiments were often stopped within 1–3 min and the whole brain was collected without perfusion.^18,19,30 Although these studies inferred that plasma choline was imported to the brain via the BBB, it was rather elusive whether choline is taken up to the brain at physiological conditions. To our knowledge, the molecular machinery for choline import at the BBB as well as the source of choline anticipated by these prior studies remains uncharacterized. MFSD7c is specifically expressed in endothelial cells of the CNS blood vessels.^20,31,32,33 Deletion of Mfsd7c in endothelial cells recapitulates the phenotypes obtained from the global KO, implicating its major role at the BBB.⁷ The current work identifies MFSD7c as a choline transporter, which facilitates choline transport bidirectionally. Employing Mfsd7c KO models, we observe that injection of a high dose (2 mM) of radioactive choline to the blood increases choline import to the brain via MFSD7c. This result recapitulates the prior reports in the literature and suggests that if a high level of choline is delivered to the blood, a small amount of choline might be imported to the brain by MFSD7c. Nevertheless, the import function of choline by MFSD7c might not be relevant in physiological conditions because a higher level of choline present in the blood might force the transport direction toward the brain side. In our in vivo transport experiments, we detected approximately 2%–3% injected choline in the brain amongst these analyzed organs (not all organs were collected). These results are well in line with numerous reports using radioactive choline (11C or 18F-choline) for cancer diagnosis. Most of radioactive choline was accumulated in the peripheral organs such as liver, kidney, lung, and prostate.^34,35,36 Consistently, the amount of injected choline present in the brain accounts for approximately 1%–3% of total injected choline in humans.³⁴ Thus, it seems that the brain is not equipped with an effective machinery to actively import blood choline in comparison with peripheral tissues. Furthermore, the half-life of choline in the circulation is 1–3 min. It is believed that the liver sequesters blood choline for the biosynthesis of PC and LPC in which the latter is then released into the circulation. Thus, it might be that the detected amount of radiolabelled choline imported to the brain of mice and humans contains a portion of LPC and/or other choline metabolites such as glycerolphosphorylcholines (GPC). These findings argue against the previously established notion that blood choline is imported into the brain.

Employing Mfsd7c KO mice, we unexpectedly discover that the expression of MFSD7c at the CNS blood vessels is required for the export of choline from the brain. We demonstrate that plasma LPC is a carrier of choline to the brain. Within the brain, the remodeling of phospholipids derived from LPC liberates choline. Our data show that LPC-derived choline is accumulated in the brain of Mfsd7c KO mice, implicating that an excessive amount of choline is exported out of the brain. The levels of acetylcholine were slightly increased in Mfsd7c KO mice, suggesting that the increased choline is used for acetylcholine synthesis. We argue that LPC-Mfsd2a pathway is an active pathway that likely provides a sufficient amount of choline in the brain. The levels of plasma LPC can reach 400–500 µM,²⁵ whereas plasma choline level is ∼10–15 µM.^22,23 Plasma LPC is transported to the brain by Mfsd2a which is one of the most abundant genes expressed in the BBB.^26,32,37,38 These data point to the conclusion that LPC provides a major source of choline to the brain, not free choline from the plasma. Thus, it is likely that the major physiological function of MFSD7c at the BBB is to reduce excess choline in the brain.

At this point, it is still unclear whether MFSD7c exports brain choline to the plasma or choline is accumulated in the endothelial cells. The increased levels of acetylcholine in the brain of Mfsd7c KO mice might hint that at least a portion of choline is indeed accumulated in the brain parenchyma since acetylcholine is mainly generated by cholinergic neurons.³⁹ MFSD7c is also expressed at the abluminal side of the endothelial cells. Thus, the excessive amount of choline may be delivered to the endothelial cells by MFSD7c. Based on our results, it is suggested that the accumulation of choline in the brain due to Mfsd7c mutations might be a causal factor for the phenotypes. The dietary choline depletion approach has been applied to test whether the accumulation of choline would be a detrimental factor. However, this approach could not eliminate the contributions of brain choline from other pathways such as LPC-Mfsd2a or GPC, resulting in a modest improvement of the phenotypes. Thus, the pathological consequences of choline accumulation to the defective BBB of Mfsd7c KO mice remain to be investigated.

In summary, our work identifies Mfsd7c as a choline transporter at the BBB. This study provides a foundation for future investigation on a potential mechanism linking the defective choline export function of MFSD7c to the pathogenesis of Fowler syndrome. The discovery of MFSD7c as a choline exporter at the BBB reveals an unexpected route of choline transport between blood and brain and opens up new avenues delving into the metabolic fates of choline in the brain.

Materials and methods

Mice

Global deletion of Mfsd7c has been described previously.² To generate the postnatal deletion of Mfsd7c in the BBB, Mfsd7c^f/f was crossed with Cdh5-Cre^ERT2 mice⁴⁰ to generate Mfsd7c^f/fCdh5-Cre^ERT2 mice (EcMfsd7c-KO). At about 2 months of age, mice were injected daily with 5 doses of 200 µg/g bodyweight tamoxifen prepared in corn oil via oral gavage.⁴¹ Male and female mice with age more than 8 weeks old were used for experiments. Mice were maintained at a constant temperature of 20 °C with 12-h light/12-h dark cycle on normal chow diets. All experimental protocols and procedures in protocol R19-0567 and R23-0973 were approved by IACUC committees under National University of Singapore.

Plasmids

All plasmids were constructed in pcDNA3.1 vector for overexpression unless stated otherwise. hChKA (D10704.1), ETNK1, CHAT (NM_020986.4) genes were synthesized from GenScript and inserted into pcDNA3.1, respectively. pcDNA3.1-hMfsd7c, pcDNA3.1-hMfsd7cT430M and S203Y were generated previously.² The other missense mutations including hMfsd7cT430A, P276S, L483R, P340L, and T430R were generated by mutagenesis and confirmed by Sanger sequencing. Zebrafish (DaMfsd7c_a, XM_688497.7; DaMfsd7c_b: XM_003200755.5), medaka fish (MeMfsd7c, XM_004082328.4), and frog Mfsd7c (XeMfsd7c, NM_001016982.2) coding sequences were synthesized by GenScript and cloned into pcDNA3.1 plasmid for overexpression, respectively. The pESC-HIS-hMfsd7c plasmid (GenScript) was used for overexpression in S. cerevisiae.

Untargeted metabolite analysis

Untargeted metabolomics analysis was performed at the Genome Analysis Center, Research Unit Molecular Endocrinology and Metabolism, Helmholtz Center Munich. Frozen WT and Mfsd7c KO embryonic brain and liver samples at E14.5 were weighed and placed into pre-cooled (dry ice) 2 mL homogenization tubes containing ceramic beads with a diameter of 1.4 mm (Precellys® Keramik-Kit 1.4 mm). Pre-cooled water with a ratio of 5 µL/mg tissue was added into the tubes. The samples were then homogenized in Precellys 24 homogenizer (PEQLAB Biotechnology GmbH, Germany) equipped with an integrated cooling unit 3 times for 20 s at 5500 rpm each, with 30 s intervals (to ensure freezing temperatures in sample vials) between the homogenization steps. 100 µL of the brain and liver homogenates were loaded onto two separate 2 mL 96-deep well plates, one plate for the brain sample set and the other for the liver samples. Three types of quality control samples were analyzed in concert with the experimental samples in each sample set: samples generated from a pool of human plasma; samples generated from a small portion of each experimental sample served as technical replicates throughout the dataset; and extracted water samples served as process blanks. Experimental samples and controls were randomized across the metabolomics analysis.

The homogenates in each well of the 2 mL 96-deep well plate were extracted with 500 µL methanol, containing four recovery standards to monitor the extraction efficiency. After centrifugation, the supernatant was split into aliquots in two 96-well microplates. The first 2 aliquots of each sample set were used for ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) analysis in positive electrospray ionization modes (i.e., early and late eluting compounds). 1 aliquot of each sample set was used for UPLC-MS/MS analysis in negative ionization mode, and the rest were kept as reserve for backup. The extract aliquots were dried on a TurboVap 96 (Zymark).

Prior to UPLC-MS/MS analysis, the dried extract samples were reconstituted in acidic or basic LC-compatible solvents, each of which contained eight or more standard compounds at fixed concentrations to ensure injection and chromatographic consistency. The UPLC-MS/MS platform utilized a Waters Acquity UPLC with Waters UPLC BEH C18-2.1 × 100 mm, 1.7 μm columns and Q-Exactive high resolution/accurate mass spectrometer (ThermoFisher Scientific) interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. Extracts reconstituted in acidic conditions were gradient eluted using water and methanol containing 0.1% formic acid, while the basic extracts, which also used water/methanol, contained 6.5 mM ammonium bicarbonate. The MS analysis alternated between MS and data-dependent MS2 scans using dynamic exclusion, and the scan range was from 80 to 1000 m/z.

Metabolites were identified by automated comparison of the ion features in the experimental samples to a reference library of chemical standard entries that included retention time, molecular weight (m/z), preferred adducts, and in-source fragments as well as associated MS spectra and curated by visual inspection for quality control using software developed at Metabolon. Chromatographic peaks were quantified using area-under-the-curve.

In vivo transport of radiolabelled choline

To study the uptake of choline into the brain, control (Mfsd7c^f/f) and EcMfsd7c-KO mice aged 4–6 months were intravenously injected with 2 mM radiolabelled choline (stock: 333 mM choline with 0.05 µCi/µL). After 5 min of injection, 50 µL blood was collected and the radioactive levels were quantified to monitor the initial radiolabelled levels in the circulation. After 30 min of injection, mice were perfused with 20 mL of PBS. Then, brains, lungs, hearts, kidneys, and livers were collected for quantification of radioactive signals. A similar amount of tissues was homogenized in RIPA buffer and the radioactive signals were quantified by a liquid scintillation counter. The radioactive signals from tissue from each mouse were normalized to radioactive signals from blood collected at 5 min post-injection.

Choline-d9 and LPC-d49 experiments

For stable isotopic tracing with choline-d9 (containing 9 deuterium atoms) injection, pregnant Mfsd7c^+/– mice at 12.5 days of gestation were injected daily by intravenous route with an amount of 2 mM choline-d9/day until gestation day 14.5 (total of 3 doses of choline-d9 from E12.5–14.5). The embryos were collected at E15.5 after the last dose at E14.5 and genotyped. The brains and livers of embryos were harvested for lipid extraction for lipidomic analysis. Briefly, an equal amount of brain and liver lysates from WT, heterozygous and KO embryos were extracted with 1-butanol:methanol (ratio 1:1) method for choline and lipid analyses by LC-MS/MS.

For stable isotopic LPC-d49 (870308P-5MG, Sigma-Aldrich, containing choline-d13, glycerol-d5, and palmitate-d31) injection, EcMfsd7c-KO and control mice aged 7–10 months were injected with a dose of 300 µM LPC-d49 in 12% BSA. After 2 h of injection, plasma was collected from the blood; then the mice were perfused with cold PBS to remove blood. After that the brains and livers were collected. Brains and livers were homogenized in PBS and the same amount of tissue lysates were used for lipid and choline extraction with chloroform:methanol (1:2 v/v) method.⁴² Organic phase was used for lipid analysis and the aqueous phase (upper phase) was used for choline analysis by LC-MS/MS.

Lipids and choline quantification by LC-MS/MS

Samples were randomized before extraction and analysis. Blank samples (10 µL MilliQ water), matrix blanks (samples extracted without spiking internal standards, see below), and pooled QC samples were used to assess method performance. In addition, diluted pooled QC samples were used to assess response linearity. Blanks, blank extracts, QC, and diluted QC were interspersed with study samples throughout the analytical run. Phospholipids and choline were separated using HILIC column (Kinetex 2.6 µm HILIC 100 Å, 150 × 2.1 mm, Phenomenex, Torrance, CA, USA) on Agilent 6495A and 6495C triple quadrupole mass spectrometers (Agilent Technologies). Mobile phases A (50% acetonitrile (LC-MS grade, ThermoFisher Scientific) and 50% 25 mM ammonium formate (Sigma-Aldrich), pH 3.5) and B (95% acetonitrile and 5% of 25 mM ammonium formate, pH 3.5) were mixed at the following gradient: 0–6 min, 99%–75% B; 6–7 min, 75%–10% B; 7–7.1 min, 10%–99% B; 7.1–10.1 min, 99% B. The flow rate was 0.6 mL/min, and the sample injection volume was 1 µL. MS parameters were as follows: electrospray ionization, gas temperature 200 °C, gas flow 12 L/min, sheath gas flow 12 L/min, and capillary voltage 3500 V. Phospholipids were quantified at the sum composition level using multiple reaction monitoring (MRM) using transition to phosphocholine headgroup (m/z 184 for endogenous lipids and adjusted to appropriate m/z for deuterated lipids). Choline was quantified using MRM using a transition from 104 to 60 (adjusted to appropriate m/z for deuterated choline). These internal standards (ISs) were used for lipidomics and choline analysis by LC-MS/MS. IS solutions were prepared in butanol/methanol (BuMe) (1:1, v/v) or chloroform/methanol (1:2) containing phospholipid IS SPLASH Mix (Avanti, 330707) (containing 5.3 µM 15:0–18:1(d7) PC, 0.19 µM 15:0–18:1(d7) PE, 0.13 µM 15:0–18:1(d7) PS (Na Salt), 0.93 µM 15:0–18:1(d7) PG (Na Salt), 0.27 µM 15:0–18:1(d7) PI (NH₄ Salt), 0.27 µM 15:0–18:1(d7) PA (Na Salt), 1.20 µM 18:1(d7) Lyso PC, 0.27 µM 18:1(d7) Lyso PE, 13.32 µM 18:1(d7) d18:1–18:1(d9) SM). Choline-d9 was used as IS in choline analysis.

Transport assays in HEK293 cells

hMfsd7c or mouse mMfsd7c cDNA constructed in pcDNA3.1 was transfected to HEK293 cells using lipofectamine 2000 (ThermoFisher Scientific). Mock was treated with an empty plasmid. For experiments in which hChKA was used, 1–1.5 µg pcDNA3.1-hChKA plasmid was co-transfected with 2–2.5 µg plasmid of hMfsd7c or mutant plasmids. Cells transfected with 1–1.5 µg pcDNA3.1-hChKA plasmid were used as control. After 18–30 h post-transfection, cells were used for transport assays with 100 µM [³H] choline (ARC: ART 0197) in DMEM containing 10% FBS. The transport assays were stopped after 1 h incubation at 37 °C by washing once with cold plain DMEM medium. Cell pellets were lyzed in RIPA buffer and transferred to scintillation vials for quantification of the radioactive signal using Tricarb liquid scintillation counter. The transport activity of MFSD7c was expressed as DMP for [³H] isotopes and CPM [¹⁴C] isotopes. For dose-dependent transport activity, HEK293 cells were similarly transfected with hMfsd7c or hMfsd7cS203Y mutant plasmid alone or co-transfected with hChKA. Transport activity was conducted with 20, 45, 90, and 250 µM [³H] choline in DMEM containing 10% FBS for 1 h at 37 °C. For the time-dependent assay, hMfsd7c was transfected or co-transfected with hChKA. Transfected HEK293 cells were incubated with 100 µM [³H] choline in DMEM containing 10% FBS for 15, 30, 60, and 120 min at 37 °C. For transport assay with [¹⁴C] ethanolamine, [³H] L-carnitine, and [³H] acetyl-carnitine, [³H] palmitoyl-carnitine, [¹⁴C] octanoyl-carnitine, HEK293 cells were co-transfected with MFSD7c and hChKA as described above. Transport assays were performed with 100 µM [¹⁴C] ethanolamine, 1 mM [³H] L-carnitine, 1 mM [³H] acetyl-carnitine, 10 µM [³H] palmitoyl-carnitine, or 100 µM [¹⁴C] octanoyl-carnitine in DMEM with 10% FBS for 1 h. Cells were washed once with cold plain DMEM and lyzed with RIPA buffer for radioactive quantification. For import assays with the missense mutants, these plasmids were co-transfected with hChKA to HEK293 cells. Import assays were performed with 100 µM [³H] choline in DMEM containing 10% FBS for 60 min. Radioactive signals from cells were quantified and converted to a percentage of WT activity.

For export assays, HEK293 cells were transfected with empty, hMfsd7c, or mMfsd7c plasmids. After 24 h of transfection, cells were loaded with 500 µM [³H] choline for 1 h. The cells were then washed to remove the remaining [³H] choline in the medium and incubated with plain DMEM medium to stimulate the release. The cells were harvested at 0 (before release) and 1 h (after release) after incubation with DMEM for radioactive quantification as described above.

Sodium and pH-dependent transport assays

For sodium-dependent assays, sodium is replaced with lithium in transport buffer (NaCl buffer: 150 mM NaCl, 5 mM KCl, 10 mM HEPES-Na, pH 7.4; LiCl buffer: 150 mM LiCl, 5 mM KCl, 10 mM HEPES, pH 7.4 with HCl). For pH-dependent assays, NaCl buffer was adjusted to pH 6.5 and pH 8.5, respectively. For transport assay conditions, overnight hMfsd7c and hChKA co-transfected HEK293 cells were washed once with the same buffer before being incubated with 100 µM [³H] choline for 15 min at 37 °C. Cells were lyzed in RIPA buffer and mixed with scintillation fluid for radioactive quantification.

Electrophysiology

For single-cell recording of membrane potential alteration during transport of choline, HEK293FT cells were cultured in DMEM medium supplemented with 10% FBS and 1% penicillin and streptomycin and maintained in 5% CO₂ incubator at 37 °C. For transfection, cells were seeded on the Petri dishes and grown overnight. Subsequently, 2 µg of empty pIRES2-EGFP (empty plasmid), pIRES2-EGFP-hMfsd7c or mutant pIRES2-EGFP-S203Y plasmid was co-transfected with 2 µg of pIRES2-RFP-hChKA plasmid using lipofectamine 2000. The transfected cells were incubated for 24 h in 5% CO₂ incubator at 37 °C. At 48 h post-transfection, the cells were split and seeded on poly-_D-lysine-coated coverslips for one more day before recording.

Whole-cell patch clamp recordings and data analysis

To record the changes in resting membrane potential, the internal solution (pipette solution) contained 130 mM K-gluconate, 10 nM KCl, 5 mM EGTA, 10 mM HEPES, 1 mM MgCl₂, 0.5 mM Na₃GTP, 4 mM Mg-ATP, 10 mM Na-phoshocreatine, pH 7.4 (adjusted with KOH) was filled in the pipette tip. The external solution contained: 10 mM Glucose, 125 mM NaCl, 25 mM NaHCO₃, 1.25 mM NaH₂PO₄·2H₂O, 2.5 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, pH 7.4 (300–310 mOsm). Single cells with both EGFP (empty, hMfsd7c or S203Y plasmid) and RFP (hChKA) fluorescence were visualized under a fluorescent microscope for patching. Resting membrane potential was recorded under the current clamp with an Axopatch200B or multiclamp 200B amplifier (Molecular Device) with 0 pA current injection. Subsequently, 200 µM choline was perfused to the cells after 2 min of stable baseline recording. After 5 min of recording, the ligand was washed away. Approximately, 10 individual cells were recorded for each condition. The data were analyzed using GraphPad Prism V software (San Diego, CA) and Microsoft (Seattle, WA) Excel. Data are expressed as mean ± SEM.

Expression of hMfsd7c in S. cerevisiae Hnm1 mutants

S. cerevisiae acquires choline via HNM1. The S. cerevisiae Hnm1 mutants were acquired from Dharmacons. WT and Hnm1 mutant yeasts were grown in Yeast extract-Peptone-Dextrose (YPD) medium at 30 °C overnight before transformation. The pESC-HIS-hMfsd7c plasmid was transformed into WT and Hnm1 mutant yeasts following the previously described protocol.⁴³ After heat shock at 42 °C for 30 min and recovery in YPD for 2 h, the yeasts were spread on 2% agar plates prepared in yeast nitrogen base (YNB) medium (Sigma-Aldrich) supplemented with 2% glucose and yeast synthetic drop-out medium supplements without histidine (Sigma-Aldrich). Positive colonies were picked up randomly and further confirmed by transport assay and western blot.

Transport assay for yeast cells

WT, Hnm1 mutant and transformed yeast cells were grown in YNB medium with 2% glucose and drop-out medium supplements without histidine at 30 °C overnight followed by induction of hMfsd7c expression in YNB medium with 2% galactose for at least 5 h at 30 °C. An 800 µL of yeast cells at OD₆₀₀ = 1 was centrifuged, washed twice with distilled water, then incubated in PBS containing 100 µM [³H] choline for 10 min. The transport assay was stopped by washing twice with PBS containing 100 µM choline. Yeast cells were lyzed with lysis buffer containing 2% Triton-X, 1% SDS, 100 mM NaCl, 10 mM Tris-HCl (pH 8.0), 1 mM EDTA (pH 8.0) at 95 °C for 5 min for radioactive quantification by liquid scintillation counter.

Western blot analysis

For protein extraction, transfected HEK293 cells with hMfsd7c, mMfsd7c, hMfsd7c mutants were lysed with RIPA buffer, respectively. For validation of deletion of Mfsd7c in mice, micro-vessels from PBS-perfused brains of WT and EcMfsd7c-KO mice were isolated, homogenized and lyzed with RIPA buffer. For Western blot analysis of hMfsd7c expression in yeasts, 1 mL of yeast culture at OD₆₀₀ = 1 from WT, Hnm1 mutant yeasts or transformed yeasts was used for total protein extraction. BCA assay was used for total protein quantification. These antibodies were used: VDAC (Cell Signaling, CS4866), CoxIV (Invitrogen, MA5-15686), OPA1 (BD biosciences, 612602), MRPS35 (Protein biotech, 16457), CHKA (Cell Signaling, CS13422S), HMOX1 (Proteintech, 10701-1-AP), NDUFS1 (Proteintech, 12444-1-AP). Western blot analysis was conducted as previously described.²

Fowler patient recruitment

A 5-month-old infant born at 39w1d with multifocal epilepsy, ventriculomegaly, developmental delay, and lissencephaly was identified via GeneMatcher collaboration.⁴⁴ She was born to a 38-year-old mother. Pregnancy was complicated by gestational diabetes and maternal hypertension. There was concern for lissencephaly on prenatal ultrasound and MRI. Prenatal CMA and karyotype were normal. She was delivered vaginally complicated by shoulder dystocia. She required brief positive pressure ventilation and was admitted to the NICU on NIPPV. Apgars at 1 and 5 min were 4 and 9, respectively. Birth parameters include length of 49 cm (47th percentile WHO), weight of 2.86 kg (20th percentile WHO), and head circumference of 32 cm (6th percentile WHO). Her whole exome sequencing at birth showed two variants of uncertain significance in FLVCR2, c.826C>T (Maternal) and c.1448T>G (Paternal). The missense mutations were confirmed by Sanger sequencing. There was no family history of a similar presentation. This was the first pregnancy between this patient’s parents. Mother had a previous miscarriage and a healthy older daughter. Parents report no consanguinity. The IRB approval for this case study was exempted by the Human Research Protection Office (HRPO) of Washington University in St. Louis.

RNA sequencing and analysis

For bulk RNA sequencing of whole brains from E14.5 WT and KO embryos from normal chow diet, this dataset has been deposited in GEO database with the accession number: GSE148854. For RNA sequencing of isolated endothelial cells from heterozygous and KO embryos at E14.5, we re-analyzed the dataset GSE129838.⁷ For RNA sequencing of whole brains from E14.5 WT embryos from CDD, pregnant WT mice were fed with CDD at the mating day until embryo collection. The whole brain of E14.5 embryos was collected for bulk RNA sequencing. This dataset has been deposited in the GEO database with the following accession number: GSE239589.

Immunofluorescence staining for neocortical vessels

Mfsd7c^+/– female mice were fed on CDD for 2 weeks and then time-mated with the heterozygous male. E14.5 embryos were dissected from the pregnant female for collection of brains for immunostaining of CNS vessel morphology. Brain section was permeabilized in 0.5% triton X-100 (PBST), blocked in 5% normal goat serum and then incubated with anti-mouse GLUT1 (Abcam, ab40084,1:200) at 4 °C overnight, followed by 3 washes in PBS at 10-min intervals. Then, the goat anti-mouse antibody (Alexa Fluor 488, ThermoFisher Scientific, A11034,1:500) was added to the slides for 1-h incubation at room temperature to visualize the GLUT1 signal. Images were taken by the confocal microscope Zeiss LSM710. Vessel phenotypes were analyzed using measurement tools in ImageJ.

Choline and acetylcholine assays

Choline and acetylcholine from plasma, liver, and brain samples were measured by fluorometric assay (Abcam: ab65345). Tissues were homogenized in choline buffer. Choline and acetylcholine were assayed according to the manufacturer’s instructions.

Transmission electron microscope (TEM)

WT (n = 3) and Mfsd7c^–/– (n = 3) embryos at E14.5 were collected from the same pregnant mice. Samples were fixed with a solution containing 2.5% glutaraldehyde in 0.1 M PBS buffer (pH 7.3). Samples were then washed and post-fixed with 1% buffered osmium. The samples were dehydrated in increasing concentrations of ethanol and then infiltrated and embedded in Araldite medium. The samples were then polymerized in a 60 °C oven for approximately 2 days. Ultrathin sections were cut using a Leica Ultracut microtome and then stained with lead citrate. The stained samples were examined in a JEM 1400 transmission electron microscope (JEOL USA, Inc., Peabody, MA) using an accelerating voltage of 100 kV. Digital images were obtained using a CMOS Matataki Flash 2K CCD camera.

Statistical analysis

Data were analyzed using GraphPrism9 software. Statistical significance was calculated using parametric and non-parametric t-tests, one- or two-way ANOVA as indicated in the figure legends. P < 0.05 was considered as statistically significant.

Data availability

All data have been included in the manuscript.

References

Meyer, E. et al. Mutations in FLVCR2 are associated with proliferative vasculopathy and hydranencephaly-hydrocephaly syndrome (Fowler syndrome). Am. J. Hum. Genet. 86, 471–478 (2010).
Article CAS PubMed PubMed Central Google Scholar
Kalailingam, P. et al. Deficiency of MFSD7c results in microcephaly-associated vasculopathy in Fowler syndrome. J. Clin. Invest. 130, 4081–4093 (2020).
CAS PubMed PubMed Central Google Scholar
Thomas, S. et al. High-throughput sequencing of a 4.1 Mb linkage interval reveals FLVCR2 deletions and mutations in lethal cerebral vasculopathy. Hum. Mutat. 31, 1134–1141 (2010).
Article CAS PubMed Google Scholar
Kvarnung, M. et al. Mutations in FLVCR2 associated with Fowler syndrome and survival beyond infancy. Clin. Genet. 89, 99–103 (2016).
Article CAS PubMed Google Scholar
Al-Murshedi, F. et al. Variability of non-lethal Fowler syndrome phenotype associated with FLVCR2 variants. Clin. Genet. 98, 520–521 (2020).
Article CAS PubMed Google Scholar
De Luca, C. et al. Expanding the clinical spectrum of Fowler syndrome: three siblings with survival into adulthood and systematic review of the literature. Clin. Genet. 98, 423–432 (2020).
Article PubMed Google Scholar
Santander, N. et al. Lack of Flvcr2 impairs brain angiogenesis without affecting the blood-brain barrier. J. Clin. Invest. 130, 4055–4068 (2020).
CAS PubMed PubMed Central Google Scholar
Yan, N. Structural biology of the major facilitator superfamily transporters. Annu. Rev. Biophys. 44, 257–283 (2015).
Article CAS PubMed Google Scholar
Duffy, S. P. et al. The Fowler syndrome-associated protein FLVCR2 is an importer of heme. Mol. Cell. Biol. 30, 5318–5324 (2010).
Article CAS PubMed PubMed Central Google Scholar
Li, Y. et al. MFSD7C switches mitochondrial ATP synthesis to thermogenesis in response to heme. Nat. Commun. 11, 4837 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Sha, W. et al. Metabolomic profiling can predict which humans will develop liver dysfunction when deprived of dietary choline. FASEB J. 24, 2962–2975 (2010).
Article CAS PubMed PubMed Central Google Scholar
Craciunescu, C. N., Johnson, A. R. & Zeisel, S. H. Dietary choline reverses some, but not all, effects of folate deficiency on neurogenesis and apoptosis in fetal mouse brain. J. Nutr. 140, 1162–1166 (2010).
Article CAS PubMed PubMed Central Google Scholar
Corbin, K. D. & Zeisel, S. H. Choline metabolism provides novel insights into nonalcoholic fatty liver disease and its progression. Curr. Opin. Gastroenterol. 28, 159–165 (2012).
Article CAS PubMed PubMed Central Google Scholar
Johnson, A. R. et al. Deletion of murine choline dehydrogenase results in diminished sperm motility. FASEB J. 24, 2752–2761 (2010).
Article CAS PubMed PubMed Central Google Scholar
Fernandez-Murray, J. P., Ngo, M. H. & McMaster, C. R. Choline transport activity regulates phosphatidylcholine synthesis through choline transporter Hnm1 stability. J. Biol. Chem. 288, 36106–36115 (2013).
Article CAS PubMed PubMed Central Google Scholar
Ribeiro, F. M. et al. The "ins" and "outs" of the high-affinity choline transporter CHT1. J. Neurochem. 97, 1–12 (2006).
Article CAS PubMed Google Scholar
Haga, T. Molecular properties of the high-affinity choline transporter CHT1. J. Biochem. 156, 181–194 (2014).
Article CAS PubMed Google Scholar
Allen, D. D. & Smith, Q. R. Characterization of the blood-brain barrier choline transporter using the in situ rat brain perfusion technique. J. Neurochem. 76, 1032–1041 (2001).
Article CAS PubMed Google Scholar
Murakami, H., Sawada, N., Koyabu, N., Ohtani, H. & Sawada, Y. Characteristics of choline transport across the blood-brain barrier in mice: correlation with in vitro data. Pharm. Res. 17, 1526–1530 (2000).
Article CAS PubMed Google Scholar
Lockman, P. R. & Allen, D. D. The transport of choline. Drug. Dev. Ind. Pharm. 28, 749–771 (2002).
Article CAS PubMed Google Scholar
Haubrich, D. R., Wang, P. F. & Wedeking, P. W. Distribution and metabolism of intravenously administered choline[methyl- 3-H] and synthesis in vivo of acetylcholine in various tissues of guinea pigs. J. Pharmacol. Exp. Ther. 193, 246–255 (1975).
CAS PubMed Google Scholar
Jacob, R. A., Jenden, D. J., Allman-Farinelli, M. A. & Swendseid, M. E. Folate nutriture alters choline status of women and men fed low choline diets. J. Nutr. 129, 712–717 (1999).
Article CAS PubMed Google Scholar
Psychogios, N. et al. The human serum metabolome. PLoS One 6, e16957 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Quehenberger, O. et al. Lipidomics reveals a remarkable diversity of lipids in human plasma. J. Lipid Res. 51, 3299–3305 (2010).
Article CAS PubMed PubMed Central Google Scholar
Tan, S. T., Ramesh, T., Toh, X. R. & Nguyen, L. N. Emerging roles of lysophospholipids in health and disease. Prog. Lipid Res. 80, 101068 (2020).
Article CAS PubMed Google Scholar
Nguyen, L. N. et al. Mfsd2a is a transporter for the essential omega-3 fatty acid docosahexaenoic acid. Nature 509, 503–506 (2014).
Article ADS CAS PubMed Google Scholar
Chan, J. P. et al. The lysolipid transporter Mfsd2a regulates lipogenesis in the developing brain. PLoS Biol. 16, e2006443 (2018).
Article PubMed PubMed Central Google Scholar
Radio, F. C. et al. Proliferative vasculopathy and hydranencephaly-hydrocephaly syndrome or Fowler syndrome: report of a family and insight into the disease’s mechanism. Mol. Genet. Genomic. Med. 6, 446–451 (2018).
Article CAS PubMed PubMed Central Google Scholar
Kvarnung, M. et al. Genomic screening in rare disorders: new mutations and phenotypes, highlighting ALG14 as a novel cause of severe intellectual disability. Clin. Genet. 94, 528–537 (2018).
Article CAS PubMed Google Scholar
Cornford, E. M., Braun, L. D. & Oldendorf, W. H. Carrier mediated blood-brain barrier transport of choline and certain choline analogs. J. Neurochem. 30, 299–308 (1978).
Article CAS PubMed Google Scholar
Zhang, Y. et al. An RNA-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cortex. J. Neurosci. 34, 11929–11947 (2014).
Article CAS PubMed PubMed Central Google Scholar
Vanlandewijck, M. et al. A molecular atlas of cell types and zonation in the brain vasculature. Nature 554, 475–480 (2018).
Article ADS CAS PubMed Google Scholar
Sabbagh, M. F. et al. Transcriptional and epigenomic landscapes of CNS and non-CNS vascular endothelial cells. Elife 7, e36187 (2018).
Article PubMed PubMed Central Google Scholar
Challapalli, A. et al. Biodistribution and radiation dosimetry of deuterium-substituted 18F-fluoromethyl-[1, 2-2H4]choline in healthy volunteers. J. Nucl. Med. 55, 256–263 (2014).
Article CAS PubMed Google Scholar
Calabria, F. et al. PET/CT with (18)F-choline: physiological whole bio-distribution in male and female subjects and diagnostic pitfalls on 1000 prostate cancer patients: (18)F-choline PET/CT bio-distribution and pitfalls. A southern Italian experience. Nucl. Med. Biol. 51, 40–54 (2017).
Article CAS PubMed Google Scholar
Kitajima, K. et al. ¹¹C-Choline-Avid but ¹⁸F-FDG-nonavid prostate cancer with lymph node metastases on positron emission tomography. Case Rep. Oncol. 9, 685–690 (2016).
Article PubMed PubMed Central Google Scholar
Daneman, R. et al. The mouse blood-brain barrier transcriptome: a new resource for understanding the development and function of brain endothelial cells. PLoS One 5, e13741 (2010).
Article ADS PubMed PubMed Central Google Scholar
Ben-Zvi, A. et al. Mfsd2a is critical for the formation and function of the blood-brain barrier. Nature 509, 507–511 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Picciotto, M. R., Higley, M. J. & Mineur, Y. S. Acetylcholine as a neuromodulator: cholinergic signaling shapes nervous system function and behavior. Neuron 76, 116–129 (2012).
Article CAS PubMed PubMed Central Google Scholar
Sorensen, I., Adams, R. H. & Gossler, A. DLL1-mediated Notch activation regulates endothelial identity in mouse fetal arteries. Blood 113, 5680–5688 (2009).
Article PubMed Google Scholar
Pitulescu, M. E., Schmidt, I., Benedito, R. & Adams, R. H. Inducible gene targeting in the neonatal vasculature and analysis of retinal angiogenesis in mice. Nat. Protoc. 5, 1518–1534 (2010).
Article CAS PubMed Google Scholar
Vu, T. M. et al. Mfsd2b is essential for the sphingosine-1-phosphate export in erythrocytes and platelets. Nature 550, 524–528 (2017).
Article ADS CAS PubMed Google Scholar
Gietz, R. D. & Woods, R. A. Transformation of yeast by lithium acetate/single-stranded carrier DNA/polyethylene glycol method. Methods Enzymol. 350, 87–96 (2002).
Article CAS PubMed Google Scholar
Sobreira, N., Schiettecatte, F., Valle, D. & Hamosh, A. GeneMatcher: a matching tool for connecting investigators with an interest in the same gene. Hum. Mutat. 36, 928–930 (2015).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This study was supported in part by Singapore Ministry of Health’s National Research Council NMRC/OFIRG/0066/20, Singapore Ministry of Education grants (T2EP30221-0012, T2EP30123-0014, and NUHSRO/2022/067/T1 (to L.N.N.)), the Synthetic Biology Translational Research Programme of the Yong Loo Lin School of Medicine of the National University of Singapore (NUHSRO/2020/077/MSC/02/SB to M.W.C.). We thank the SG Academies South-East Asia Fellowship (SASEAF) Programme funded by National Research Foundation (to X.T.A.N). We also thank L.K.V. for providing technical assistance with SLC44a1/2.

Author information

Thanh Nha Uyen Le
Present address: Hudson Institute of Medical Research, Clayton, VIC, Australia
These authors contributed equally: Xuan Thi Anh Nguyen, Thanh Nha Uyen Le.

Authors and Affiliations

Department of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Xuan Thi Anh Nguyen, Thanh Nha Uyen Le, Toan Q. Nguyen, Hoa Thi Thuy Ha, Nancy C. P. Leong, Dat T. Nguyen, Pei Yen Lim, Adelia Vicanatalita Susanto, Qianhui Huang, Ling Fam, Matthew Wook Chang, Markus R. Wenk, Jerzy Adamski, Amaury Cazenave-Gassiot & Long N. Nguyen
Metabolomics and Proteomics Core, Helmholtz Center Munich, German Research Center for Environmental Health, Neuherberg, Germany
Anna Artati
Singapore Lipidomics Incubator (SLING), Life Sciences Institute, National University of Singapore, Singapore, Singapore
Pei Yen Lim, Markus R. Wenk, Amaury Cazenave-Gassiot & Long N. Nguyen
NUS Synthetic Biology for Clinical and Technological Innovation (SynCTI), National University of Singapore, Singapore, Singapore
Adelia Vicanatalita Susanto & Matthew Wook Chang
Synthetic Biology Translational Research Programme, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Adelia Vicanatalita Susanto & Matthew Wook Chang
Electron Microscopy Unit, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Lo Ngah Leong & Isabelle Bonne
Department of Microbiology and Immunology, Immunology Translational Research Programme, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Isabelle Bonne
Life Sciences Institute, Immunology Programme, National University of Singapore, Singapore, Singapore
Isabelle Bonne & Long N. Nguyen
Division of Genetics and Genomic Medicine, Department of Pediatrics, Washington University in St Louis, Saint Louis, MO, USA
Angela Lee, Jorge L. Granadillo & Catherine Gooch
Electrophysiology Core Facility, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Dejie Yu, Hua Huang & Tuck Wah Soong
Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Hua Huang & Tuck Wah Soong
Healthy Longevity Translational Research Program, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Hua Huang & Tuck Wah Soong
Cardiovascular Diseases Program, National University of Singapore, Singapore, Singapore
Hua Huang, Tuck Wah Soong & Long N. Nguyen
Institute of Experimental Genetics, Helmholtz Zentrum München, German Research Center for Environmental Health, Neuherberg, Germany
Jerzy Adamski
Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia
Jerzy Adamski
Immunology Translational Research Program, Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore
Long N. Nguyen

Authors

Xuan Thi Anh Nguyen
View author publications
You can also search for this author in PubMed Google Scholar
Thanh Nha Uyen Le
View author publications
You can also search for this author in PubMed Google Scholar
Toan Q. Nguyen
View author publications
You can also search for this author in PubMed Google Scholar
Hoa Thi Thuy Ha
View author publications
You can also search for this author in PubMed Google Scholar
Anna Artati
View author publications
You can also search for this author in PubMed Google Scholar
Nancy C. P. Leong
View author publications
You can also search for this author in PubMed Google Scholar
Dat T. Nguyen
View author publications
You can also search for this author in PubMed Google Scholar
Pei Yen Lim
View author publications
You can also search for this author in PubMed Google Scholar
Adelia Vicanatalita Susanto
View author publications
You can also search for this author in PubMed Google Scholar
Qianhui Huang
View author publications
You can also search for this author in PubMed Google Scholar
Ling Fam
View author publications
You can also search for this author in PubMed Google Scholar
Lo Ngah Leong
View author publications
You can also search for this author in PubMed Google Scholar
Isabelle Bonne
View author publications
You can also search for this author in PubMed Google Scholar
Angela Lee
View author publications
You can also search for this author in PubMed Google Scholar
Jorge L. Granadillo
View author publications
You can also search for this author in PubMed Google Scholar
Catherine Gooch
View author publications
You can also search for this author in PubMed Google Scholar
Dejie Yu
View author publications
You can also search for this author in PubMed Google Scholar
Hua Huang
View author publications
You can also search for this author in PubMed Google Scholar
Tuck Wah Soong
View author publications
You can also search for this author in PubMed Google Scholar
Matthew Wook Chang
View author publications
You can also search for this author in PubMed Google Scholar
Markus R. Wenk
View author publications
You can also search for this author in PubMed Google Scholar
Jerzy Adamski
View author publications
You can also search for this author in PubMed Google Scholar
Amaury Cazenave-Gassiot
View author publications
You can also search for this author in PubMed Google Scholar
Long N. Nguyen
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

L.N.N., T.N.U.L., and T.Q.N. performed in vitro transport assays. T.Q.N. and L.N.N. performed in vivo transport assays. A.A. and J.A. performed metabolite analysis. D.Y., H.H., and T.W.S. performed electrophysiological assays. Q.H. and L.F. helped with cell cultures. T.N.U.L., A.V.S., and M.W.C. performed yeast experiments. H.T.T.H., X.T.A.N., T.Q.N., A.C.-G., P.Y.L., and M.R.W. performed lipidomics assays. A.L., J.L.G., and C.G. provided clinical data. D.T.N and N.C.P.L. performed transport assays and western blot for mutants. L.N.L. and I.B. performed electron microscopy. X.T.A.N. performed in vivo experiments. L.N.N. conceived and designed the study and experiments, analyzed data, prepared the figures, and wrote the manuscript.

Corresponding author

Correspondence to Long N. Nguyen.

Ethics declarations

Competing interests

The authors declare no competing interests.

Supplementary information

Supplementary information Fig S1

Supplementary information Fig S2

Supplementary information Fig S3

Supplementary information Fig S4

Supplementary information Fig S5

Supplementary information Fig S6

Supplementary information Fig S7

Supplementary information Fig S8

Supplementary information Fig S9

Supplementary information Fig S10

Supplementary Tables 1–5

Supplementary Tables 6–10

Supplementary Table 11

Supplementary Table 12

Supplementary Table 13

Supplementary Table 14

Supplementary Table 15

Supplementary Table 16

Supplementary Table 17

Supplementary Table 18

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Nguyen, X.T.A., Le, T.N.U., Nguyen, T.Q. et al. MFSD7c functions as a transporter of choline at the blood–brain barrier. Cell Res 34, 245–257 (2024). https://doi.org/10.1038/s41422-023-00923-y

Download citation

Received: 18 December 2023
Accepted: 26 December 2023
Published: 02 February 2024
Issue Date: March 2024
DOI: https://doi.org/10.1038/s41422-023-00923-y