Nel L et al. (2026), Structural and biochemical comparison of the FL...

XB-ART-61840

Life Sci Alliance 2026 Jul 11;97:. doi: 10.26508/lsa.202503583.

Show Gene links Show Anatomy links

Structural and biochemical comparison of the FLVCR and CTL membrane protein families in eukaryotes.

Nel L, Driller JH, Driller R, Frain KM, Pedersen BP.

???displayArticle.abstract???
The organic cation choline is essential for eukaryotic metabolism. Recently, the feline leukemia virus subgroup C receptor-related (FLVCR, SLC49) family was demonstrated as central for basal choline transport, questioning the role of the choline transporter-like (CTL, SLC44) family in this capacity. Here, we use Xenopus laevis oocytes to confirm that FLVCR1 (SLC49A1) and FLVCR2 (SLC49A2) proteins are choline transporters. CTL1 (SLC44A1) does not transport choline under the same conditions, supported by other CTL proteins, Arabidopsis thaliana CherI and Saccharomyces cerevisiae PNS1, which also display no choline transport activity. We present the atomic structures of FLVCR2, CTL1, and PNS1. The 3.4 Å cryo-EM structure of FLVCR2 has choline in the binding pocket. The 3.3 Å cryo-EM structure of CTL1 and the 2.7 Å crystal structure of PNS1 reveal an unusual protein fold, weakly related to the mitochondrial carrier family (SLC25). The unusual fold appears incompatible with transmembrane transport and implies a different and, so far, unknown function for CTL proteins. Our results support FLVCR proteins as choline transporters and suggest a nontransport role for CTL proteins.

???displayArticle.pubmedLink??? 42114998
???displayArticle.pmcLink??? PMC13160678
???displayArticle.link??? Life Sci Alliance

???attribute.lit??? ???displayArticles.show???

	Figure 1. Overview of the metabolism and transporters of choline and ethanolamine in humans.(A) Choline and ethanolamine need to be transported across biological membranes to be metabolized further by multiple enzymes. For instance, choline can be acetylated to produce the neurotransmitter acetylcholine. Both choline and ethanolamine can be phosphorylated in the Kennedy pathway to form phospholipids, which are essential components of all membranes. Within mitochondria, the oxidation of choline forms betaine, a precursor to amino acids like methionine and cysteine. Abbreviations: CHAT, choline acetyltransferase; EK, ethanolamine kinase; ECT, phosphoethanolamine cytidylyltransferase; EPT, ethanolaminephosphotransferase; CK, choline kinase; CCT, phosphocholine cytidylyltransferase; CPT, cholinephosphotransferase; PEMT, phosphatidylethanolamine N-methyltransferase; PSS1, phosphatidylserine synthase 1; PSS2, phosphatidylserine synthase 2; PCP, phosphatidylcholine:ceramide choline phosphotransferase; LPCAT, acyl-CoA:lysophosphatidylcholine acyltransferase; PSD, phosphatidylserine decarboxylase; PLA, platelet-activating factor; CHDH, choline dehydrogenase; BMHT, betaine:homocysteine methyltransferase; MTHF, 5,10-methylene-tetrahydrofolate; THF, tetrahydrofolate; MTHFD, methylenetetrahydrofolate dehydrogenase; MTHFR, methylenetetrahydrofolate reductase; MAT, methionine adenosyltransferase; SAH, S-adenosylhomocysteine hydrolase; and MS, methionine synthase. (B) Experimentally solved structures representing the human solute carrier (SLC) families implicated in choline transport. Models and topology diagrams are colored from N terminus (blue) to C terminus (red). The high-affinity choline transporter family (CHT, SLC5) has a LeuT-like fold and is exemplified by CHT1 (PDB: 8J77). Both the organic cation transporter family (OCT, SLC22) and the feline leukemia virus subgroup C receptor–related (FLVCR, SLC49) family have a major facilitator superfamily (MFS) fold. However, OCT1 (PDB: 8ET6) has an extracellular domain between TM1 and TM2 that FLVCR1 (PDB: 8QCT) does not have. Two unique but similar folds are typified by the mitochondrial carrier family (MCF, SLC25), represented by uncoupling protein 1 (UCP1, PDB: 8HBV), whereas the choline transporter-like (CTL, SLC44) family is characterized by CTL1 (PDB: 7WWB). Because no experimentally solved structure of SLC25A48 is available, we use the structure of uncoupling protein 1 (SLC25A7) to represent the fold.
	Figure S1. Overview of Xenopus laevis oocyte assays.(A) Schematic depicting the principle of using frog oocytes to assay protein activity with radioactive substrates. (B) Representative confocal microscopy images of GFP signal from recombinantly expressed CTL1, CherI, PNS1, FLVCR1, and FLVCR2 on the plasma membrane of oocytes (scale bar: 200 μm). (C) Time-dependent, linear uptake of choline by FLVCR1 and FLVCR2 (n = 3–5). A simple linear regression was applied to fit the data. (D) Titration curve of FLVCR1 and FLVCR2 with H3-ethanolamine (n = 4–6). Saturation was not achieved, and a Km value could not be determined when applying a Michaelis–Menten least squares fit. The plots appear to be fitted with a simple linear regression fit but is in fact a Michaelis–Menten least squares fit. (E) Uptake of 0.5 μM C14-choline by CTL1, PNS1, and CherI under different pH conditions (n = 3–6). Each data point represents an average of two to six oocytes and the SD as error bars. No fit was applied to the data. (F) Uptake of 2 μM C14-choline by CTL1, PNS1, and CherI under different pH conditions (n = 4–5). Each data point represents an average of four to five oocytes and the SD as error bars. No fit was applied to the data.
	Figure 2. Biochemical characterization of proposed choline transporters in Xenopus laevis oocytes.Statistical significance is denoted by asterisks that correspond to certain probabilities (P-values). Not significant (ns; P > 0.05), significant (P < 0.05), highly significant (P < 0.01), very high significance (P < 0.001), and extremely high significance (**P < 0.0001). (A) Uptake of 100 μM H3-choline by CTL1, PNS1, CherI, FLVCR1, and FLVCR2 (n = 4–6). Compared with the negative control of water, choline uptake is not observed for CTL1 (P = 0.7376), PNS1 (P = 0.8621), or CherI (P = 0.9875). Choline uptake is observed for FLVCR1 (P = 0.0208) and FLVCR2 (P < 0.0001). Statistics were calculated from a one-way ANOVA with Dunnett’s multiple comparisons test. (B) Uptake of 100 μM H3-ethanolamine by CTL1, PNS1, CherI, FLVCR1, and FLVCR2 (n = 6). Compared with the negative control of water, ethanolamine uptake is not observed for CTL1 (P > 0.9999), PNS1 (P = 0.9977), or CherI (P > 0.9999). Ethanolamine uptake is observed for FLVCR1 (P < 0.0001) and FLVCR2 (P < 0.0001). Statistics were calculated from a one-way ANOVA with Dunnett’s multiple comparisons test. (C) Titration curve of FLVCR1 with H3-choline (n = 4–6). Measurements were corrected by subtracting the average of counts obtained from water for each concentration. A Michaelis–Menten fit was applied to obtain the apparent affinity (Km) for FLVCR1 as 26 ± 8 μM. (D) Titration curve of FLVCR2 with H3-choline (n = 3–6). Measurements were corrected by subtracting the average of counts obtained from water for each concentration. A Michaelis–Menten fit was applied to obtain the apparent affinity (Km) for FLVCR2 as 164 ± 21 μM. (E) Evaluating the substrate scope of FLVCR1 and FLVCR2 with choline, ethanolamine, carnitine, and histamine (n = 2–6). When comparing water with H3-choline, FLVCR1 (P < 0.0001) and FLVCR2 (P < 0.0001) exhibited transport. For water compared with H3-ethanolamine, FLVCR1 (P < 0.0001) and FLVCR2 (P < 0.0001) showed transport. When comparing water with H3-carnitine, FLVCR1 (P = 0.9090) and FLVCR2 (P = 0.7424) did not exhibit transport. For water compared with H3-histamine, FLVCR1 (P > 0.9999) and FLVCR2 (P > 0.9999) did not show transport. When comparing water to C14-choline, FLVCR1 (P < 0.0001) and FLVCR2 (P < 0.0001) exhibited transport. Statistics were calculated from a one-way ANOVA with Sidak’s multiple comparisons test. (F) Transport of H3-choline by FLVCR1 and FLVCR2 under different pH conditions (n = 3–4). Measurements were corrected by subtracting the average of counts obtained from water for each pH condition. (G) Influence of protons and the membrane potential on the transport of choline by FLVCR1 and FLVCR2 (n = 2–4). When comparing FLVCR1 without and with CCCP (P = 0.2283), no significant dependency on protons was observed. For FLVCR2 with and without CCCP (P = 0.0011), a highly significant dependency on protons was observed. Statistics were calculated from a one-way ANOVA with Sidak’s multiple comparisons test. (H) Transport of H3-choline (C) and H3-ethanolamine (E) by FLVCR1 and FLVCR2 in the presence (Na) of 96 mM sodium or absence (K) of sodium (using 96 mM potassium; n = 3–5). Neither FLVCR1 nor FLVCR2 seems to depend on sodium to transport choline or ethanolamine. For negative controls, water and choline with sodium and with potassium, P = 0.9801, as well as water and ethanolamine with sodium or potassium, P = 0.9996 were obtained. For FLVCR1 with choline and sodium or potassium, P = 0.9999. For FLVCR1 and ethanolamine and sodium or potassium, P = 0.9152. For FLVCR2 with choline and sodium or potassium, P = 0.1761. For FLVCR1 and ethanolamine and sodium or potassium, P = 0.9795. Statistics were calculated from a one-way ANOVA with Sidak’s multiple comparisons test.
	Figure 3. Cryo-EM structure of FLVCR2.(A) Diagram of the major facilitator topology exhibited by FLVCR2, divided into an N domain (rosy brown) and C domain (sand). The twofold pseudosymmetry point is indicated by the black oval. (B) Model and map of FLVCR2 (PDB: 9QU4) in a schematic of its localization in the plasma membrane. The map shows where the Protein A repeats (gray) were linked to the N terminus, but this fiducial was not built in the model. (C) Map of the DDM:CHS micelle, shown as a gray surface, with our model of FLVCR2 inside in black. The AlphaFold2 prediction of Protein A fused to FLVCR2, colored according to the predicted local distance difference test (pLDDT) scale, is superposed onto our model. (D) Unmodeled, elongated map regions, possibly of sterols and retained lipids, form belts around FLVCR2, particularly on the face where the membrane would be in contact with the extracellular environment. (E) Model of FLVCR2 at 3.4 Å adopts the inward-facing conformation. Choline (brown) is bound within a pocket located at a central twofold pseudosymmetry point. Residues involved in coordinating choline are shown as sticks and annotated in the map (gray). (F) Hydrophobicity of the binding pocket with choline inside. The slices were made through the front and the top of FLVCR2. (G) Electrostatics of the binding pocket with choline inside. The slices were made through the front and the top of FLVCR2. (H) Hydrophobicity coloring of FLVCR2 from the front and the back. (I) Electrostatic coloring of FLVCR2 from the front and the back.
	Figure S2. Purification and cryo-EM data processing of FLVCR2 in DDM.(A) Schematic of the construct of FLVCR2 used for structure determination. The three repeats of Protein A served as a fiducial to facilitate particle alignment during data processing. The thrombin site was used after Ni-NTA purification to remove the 10× His-Tag. (B) SDS–PAGE gel of fractions taken during the purification of FLVCR2. Lanes 1–7: Ni-NTA elution fractions; lane 8: molecular weight marker; lanes 9–15: fractions of SEC. The asterisk () denotes the sample used to prepare grids for cryo-EM. (C) Size-exclusion chromatography trace of FLVCR2 from an S200 Superdex Increase 10/300 column. The asterisk () shows the fraction used to prepare grids for cryo-EM. (D) Representative micrograph from the data collected on the copper-support grid with R1.2/1.3 300 mesh. Particles were evenly distributed within the ice. (E) Representative 2D classes obtained from picked particles. Multiple orientations were sampled of FLVCR2, with the Protein A fiducial protruding from the micelle. (F) Summary of the data processing steps performed in cryoSPARC.
	Figure S3. Purification and cryo-EM data processing of CTL1 in LMNG.(A) Schematic of the construct of CTL1 used for structure determination. The thrombin site was used after affinity purification to remove the 3× Strep-tag. (B) SDS–PAGE gel of fractions taken during the purification of CTL1. Lane 1: lysate flow-through; lane 2: beads after elution; lane 3: elution from beads; lane 4: molecular weight marker; lane 5: concentrator flow-through; lane 6: concentrated sample injected for SEC; lane 7: molecular weight marker; lanes 8-15: fractions of SEC. The asterisk () denotes the sample used to prepare grids for cryo-EM. (C) Size-exclusion chromatography trace of CTL1 from an S200 Superdex Increase 10/300 column. The asterisk () shows the fraction used to prepare grids for cryo-EM. (D) Representative micrograph from the data collected on the copper-support grid with R1.2/1.3 300 mesh. Particles were evenly distributed within the ice. (E) Representative 2D classes obtained from picked particles. A variety of orientations were sampled of CTL1. (F) Summary of the data processing steps performed in cryoSPARC. (G) Maps of CTL1 in LMNG and peptidisc are very similar (map correlation = 0.86), with the largest differences observed in the position of M3 and M4. In the peptidisc map (beige), M3/M4 are pulled toward the core of the protein, whereas M3/M4 are further away from the rest of the protein in the LMNG map (teal).
	Figure 4. Cryo-EM structure of CTL1.(A) Diagram of the novel topology of CTL1. The threefold pseudosymmetry point is indicated by the black triangle. (B) Model and map of CTL1 (PDB: 9QU3) in a schematic of its localization in the plasma membrane. The model of CTL1 shows a unique fold with 10 transmembrane helices (M1–10 in shades of blue and green) arranged in a rotary fashion around three amphipathic helices (AH1–3 in yellow) laying parallel to the membrane. The extracellular domain and M3 and M4 were not built and are shown in gray in the map. (C) Map of the LMNG:CHS micelle, shown as a gray surface, with our model of CTL1 inside in black. The AlphaFold2 prediction of CTL1, colored according to the predicted local distance difference test (pLDDT) scale, is superposed onto our model. (D) Unmodeled map regions of putative sterols and lipids form a belt around the transmembrane helices that are near the extracellular domain. Lipid- and sterol-like densities are also observed in a defined groove at the back of CTL1. (E) Model of CTL1 adopts a bowl-like conformation that is open toward the extracellular environment, with a possible binding pocket at the central pseudosymmetry point. Residues of the putative binding pocket are annotated and shown in the map (gray), along with an unidentified ligand (yellow outline) at a lower contour level of the map. (F) Hydrophobicity coloring of CTL1 shown from the front, back, and top. The outer surface of CTL1 is mostly hydrophobic, with a groove at the back where lipids or sterols likely bind, whereas the putative binding pocket mostly contains hydrophilic residues. (G) Electrostatic coloring of CTL1 shown from the front, back, and top. The outer surface of CTL1 is mostly neutral to positively charged, whereas distinct positive (Lys544, Lys454, Arg457, His584, and Lys491) and negative (Asp537, Glu591, and Asp595) patches are observed near the opening of the proposed binding pocket.
	Figure S4. Purification and X-ray crystallography of PNS1.(A) Schematic of the construct of PNS1 used for structure determination. Limited proteolysis with trypsin was performed on full-length PNS1. (B) SDS–PAGE gel of fractions taken during the purification of PNS1. Lane 1: PNS1 before trypsin digest; lanes 2–3: PNS1 after trypsin digest; lane 4: molecular weight marker; lanes 6–14: fractions of SEC. The asterisk () denotes the sample used for crystallography. (C) Size-exclusion chromatography trace of PNS1 from an S200 Superdex Increase 10/300 column. The asterisk () shows which fractions were used in crystallography experiments. (D) Crystals of PNS1 obtained in vapor diffusion and lipidic cubic phase droplets.
	Figure 5. X-ray crystallography structure of PNS1.(A) Diagram of the topology of PNS1. The threefold pseudosymmetry point is indicated by the black triangle. (B) Model of PNS1 in a schematic of its localization in the plasma membrane. PNS1 at 2.7 Å (PDB: 9F63) shows a novel fold with 10 transmembrane helices (M1-10) arranged in a rotary fashion (coral, yellow, pink, purple, and orange) and three amphipathic helices (AH1-3 in green) that lay parallel to the membrane. (C) Densities of lipid- or sterol-like molecules (purple mesh) are present in a defined groove at the back of PNS1. Within the putative binding pocket, densities for water molecules and an unknown ligand are also observed (purple mesh). (D) Model of PNS1 adopts a bowl-like conformation that is open toward the extracellular environment, with a possible binding pocket at the central pseudosymmetry point. Residues of the putative binding pocket are annotated and shown in the map (gray mesh), along with nonprotein densities attributed to an unidentified ligand and possible water molecules (purple mesh). Distances (green) are shown between residues that could form salt bridges or interactions with water molecules. (E) Hydrophobicity coloring of PNS1 shown from the front, back, and top. The outer surface of PNS1 is mostly hydrophobic, with a groove at the back where lipids or sterols likely bind, whereas the putative binding pocket mostly contains hydrophilic residues. (F) Electrostatic coloring of PNS1 shown from the front, back, and top. The outer surface of PNS1 is mostly neutral to positively charged, whereas distinct positive (Arg296, Arg354, Lys389, and Arg495) and negative (Glu293, Glu393, and Glu492) patches are observed within the proposed binding pocket.
	Figure 6. Comparison of SLC44 structures determined in the current study.(A) Depth and width of the bowl-like cavity of CTL1 and PNS1 shown from the top, front, and back. (B) Superposition of CTL1 and PNS1 reveal that transmembrane M3/M4 of CTL1 is arranged in a manner that makes its structure more open compared with that of PNS1. Helices M1, M5, and M7 are also more open in CTL1 compared with PNS1. (C) Putative binding pockets of CTL1 and PNS1 are very different because only Y495CTL1 and Y394PNS1 are conserved. (D) Superposition of CTL1 and PNS1 with cartoons in gray and conserved residues, based on the sequence alignment of Fig S6, shown as sticks (teal for CTL1 and coral for PNS1). The models are superposed and shown from the top, front, and back. (E) AlphaFold-predicted models of full-length CTL1 (left) and PNS1 (right) colored according to evolutionary conservation of residues by ConSurf.
	Figure S5. Comparing our FLVCR2 and CTL1 structures with those in literature.(A) Superposition of FLVCR2 solved in the current study (PDB: 9QU4 in sand) to other structures in the inward-facing conformation and a close-up view of the substrate binding pocket. All the structures, except the model and labels in pink, are of human FLVCR2. When superposing the three models to PDB: 9QU4, the RMSDCa is 0.7 Å for all the structures. (B) Overlay of CTL1 determined in the current study (teal color) and in previous work (gray). Regions that were modeled to various degrees are indicated in colors corresponding to the models.
	Figure S6. Sequence alignment and secondary structure annotation of CTL1 and PNS1.The secondary structure annotation of CTL1 (top) and PNS1 (bottom) is colored according to their respective topology diagrams (Figs 4A and 5A). Residues in black are conserved and correspond to the amino acids shown in the models of Fig 6D.
	Figure S7. Consulting structures for clues to SLC44 members’ function and their comparison with predicted choline transporter folds.(A) DALI output CysZ, a sulfate permease in prokaryotes, with a Z score of 9.7 in relation to PNS1. CysZ has three pairs of helices (red, orange, and purple) that span the membrane to various extents. Three short helices (teal) are arranged in a tripod configuration, surrounding the transmembrane helices. One monomer is shown from the top and front and then within the hexameric assembly adopted in the crystal structure (PDB: 6D9Z). (B) In eukaryotes, DALI output SLC25A48 (Z score = 8.8) for PNS1 and SLC25A45 (Z score = 8.6) for CTL1. Because a structure of only SLC25A7 is available, we use this as a representative of the fold (PDB: 8HBV). SLC25s commonly have a core formed by six transmembrane helices, arranged in groups of 2 (red, orange, and purple) and are interconnected by three helices (teal) that are parallel to the membrane. PNS1 and CTL1 have M1 and M2 (light gray) as additions at the periphery of this core, as well as M3 and M4 (dark gray) that form another helix–hairpin pair. (C) AlphaFold predictions obtained via UniProt of SLC25A48 (the choline transporter in the inner mitochondrial membrane, UniProt: Q6ZT89), CherI (proposed to transport choline in Arabidopsis thaliana, UniProt: Q94AN2), and HNM1 (the choline transporter in yeast, UniProt: P19807). The models are colored according to the predicted local distance difference test (pLDDT) scale.