In vivo evidence for GDP-fucose transport in the absence of transporter SLC35C1 and putative transporter SLC35C2

Abstract

Slc35c1 encodes an antiporter that transports GDP-fucose into the Golgi and returns GMP to the cytoplasm. The closely related gene Slc35c2 encodes a putative GDP-fucose transporter and promotes Notch fucosylation and Notch signaling in cultured cells. Here, we show that HEK293T cells lacking SLC35C1 transferred reduced amounts of O-fucose to secreted epidermal growth factor-like repeats from NOTCH1 or secreted thrombospondin type I repeats from thrombospondin 1. However, cells lacking SLC35C2 did not exhibit reduced fucosylation of these epidermal growth factor-like repeats or thrombospondin type I repeats. To investigate SLC35C2 functions in vivo, WW6 embryonic stem cells were targeted for Slc35c2. Slc35c2[-/-] mice were viable and fertile and exhibited no evidence of defective Notch signaling during skeletal or T cell development. By contrast, mice with inactivated Slc35c1 exhibited perinatal lethality and marked skeletal defects in late embryogenesis, typical of defective Notch signaling. Compound Slc35c1[-/-]Slc35c2[-/-] mutants were indistinguishable in skeletal phenotype from Slc35c1[-/-] embryos and neonates. Double mutants did not exhibit the exacerbated skeletal defects predicted if SLC35C2 was functionally important for Notch signaling in vivo. In addition, NOTCH1 immunoprecipitated from Slc35c1[-/-]Slc35c2[-/-] neonatal lung carried fucose detected by binding of Aleuria aurantia lectin. Given that the absence of both SLC35C1, a known GDP-fucose transporter, and SLC35C2, a putative GDP-fucose transporter, did not lead to afucosylated NOTCH1 nor to the severe Notch signaling defects and embryonic lethality expected if all GDP-fucose transport were abrogated, at least one more mechanism of GDP-fucose transport into the secretory pathway must exist in mammals.

Keywords: GDP-fucose transporter; Notch signaling; O-fucose glycans; gene deletion; skeletal development.

Figure 1

GDP-fucose synthesis, transport, and utilization. The diagram summarizes known mechanisms of GDP-fucose synthesis in the cytosol from glucose or mannose via a de novo pathway from GDP-Man or from fucose via a salvage pathway (63). From the cytosol, GDP-fucose is transported into the Golgi by SLC35C1 (blue oval). The related protein SLC35C2 (green oval) also resides in the Golgi (of rat liver at least) but has not been shown directly to transport GDP-fucose. In this work, we show that elimination of both SLC35C1 and SLC35C2 does not prevent the transfer of fucose to NOTCH1 providing strong evidence for the existence of either an ER GDP-fucose transporter (gray oval) like Efr in Drosophila (see text) or some novel means of GDP-fucose transfer (?) to the ER. POFUT2 also resides in the ER and transfers GDP-fucose to thrombospondin type 1 repeats (TSRs). Transfer of fucose to the core of complex or hybrid N-glycans (by FUT8) or the GlcNAc in lactosamine units in glycans of glycoproteins and glycolipids (by different FUTs) generating LeX or sialyl-LeX epitopes occurs in the Golgi. All fucose residues that are terminal (i.e. not further substituted) are recognized by the fucose-binding lectin AAL. See text for further details and references. AAL, Aleuria aurantia lectin; ER, endoplasmic reticulum; FUT, fucosyltransferase; LeX, Lewis X.

Figure 2

Deletion of SLC35C1 in HEK293T cells reduced O-fucosylation of EGF repeats and TSRs but deletion of SLC35C2 did not.A, flow cytometric analysis of fluorescent AAL binding to WT, SLC35C1-KO (clone 25.3) and SLC35C2-KO (clone 59) HEK293T cells, or WT control (Blank-AAL incubated with 5 mM L-Fucose). The results are representative of three replicates, with mean fluorescence intensity (MFI) normalized to WT (100%). SLC35C1-KO was 3.8 ± 0.24% and SLC35C2-KO was 94 ± 2.7% of WT. Similar results were obtained with a second clone for each KO (Fig. S1E). B, AAL binding to SLC35C1-KO (clone 25.3) cells expressing a cDNA encoding SLC35C1 or SLC35C2. The results are representative of four replicates, with MFI normalized to WT (100%). SLC35C1-KO was 4.8 ± 1.1%, SLC35C1-KO+C1cDNA was 90.7 ± 11.6%, and SLC35C1-KO+C2cDNA was 6.3 ± 1.4% of WT. Similar results were obtained with a second SLC35C1-KO clone (Fig. S1F). C, NOTCH1 EGF1-5 and TSP1 TSR1-3 were purified from the medium of WT, SLC35C1-KO (clone 25.3), or SLC35C2-KO (clone 59) cells expressing mouse NOTCH1 EGF1-5-Myc-His₆, or TSP1 TSR1-3-Myc-His₆, digested with proteases, and analyzed by nano-LC-MS/MS. The relative abundance of O-fucosylated peptides (percent of total modified and unmodified peptides) from NOTCH1 EGF2, 3, and 5, and TSP1 TSR1, 2, and 3 are plotted. Note that the fucosylated forms of TSRs include both the monosaccharide and disaccharide forms of O-fucose. Relevant MS/MS spectra are in Figs. S2 and S3. D, LFNG was purified from the medium of WT, SLC35C1-KO (clone 25.3), or SLC35C2-KO (clone 59) cells expressing mouse Lfng-Myc-His₆, digested with protease, and analyzed by nano-LC-MS/MS. The relative abundance of N-glycan fucosylated peptide (percent of total modified and unmodified peptides) bearing the only N-glycan on LFNG is plotted. An MS/MS spectrum for this peptide is shown in Fig. S4. A two-tailed assuming unequal variance t test was used to determine the statistical significance of fucosylated glycoforms. Error bars are SD. AAL, Aleuria aurantia lectin; cDNA, complementary DNA; EGF, epidermal growth factor; LFNG, Lunatic Fringe; MS/MS, tandem mass spectrometry; TSP, thrombospondin; TSR, thrombospondin type 1 repeat.

Figure 3

Targeted deletion of Slc35c2 in the mouse.A, the Slc35c2 genomic locus of 11 exons was targeted at exon 4 with the pFlox vector that includes a neomycin (neo) and thymidine kinase (tk) cassette and two loxP sites (black triangles) that flank ∼800 bp of genomic DNA. The deleted allele was obtained by crossing mice with a targeted Slc35c2 allele with mice carrying a ZP3-Cre recombinase transgene (64, 65). Thus, exon 4 was removed, leaving one loxP site. B, BamHI; H, HindIII; Xb, XbaI; S, SalI. B, genomic DNA was extracted from WW6 WT and targeted WW6 ES cells C2PC70-5 and C2P30-2, digested with BamHI, and analyzed by Southern analysis using probe P1 to detect 8.7 kb (WT) and 11.2 kb (Targeted) fragments. C, characterization of C2PC70-5 targeted ES cells. PCR genotyping of Slc35c2 in four pups from a cross between C2PC70-5 heterozygotes. Bands derived from a 390 bp WT allele (+) and a 230 bp deleted allele (Del). D, RT-PCR of complementary DNA prepared from mouse liver RNA from 3-week pups using primers within the Slc35c2 ORF. RT-PCR of actin transcripts served as control. Note the reduced level of truncated mutant transcript (Del) compared with WT transcript (+). E, Western blot analysis of mouse testis lysate from Slc35c2[+/+] and Slc35c2[−/−] males (left) and mouse embryo fibroblasts (MEF) derived from Slc35c2[+/+] and Slc35c2[−/−] embryos. The anti-SLC35C2 antibody was previously described (28). ES, embryonic stem.

Figure 4

Skeletal development in Slc35c2, Slc35c1, and Slc35c1:Slc35c2 null embryos.A–H, skeletal preparations from ∼E18.5 embryos of various Slc35c1 and Slc35c2 compound genotypes in the C57BL/6J background are shown. Genotypes are indicated by C1 for Slc35c1 and C2 for Slc35c2. Skeletons were stained with Alcian blue (cartilage) and Alizarin red (bone). T1 is first thoracic vertebra, L1 is first lumbar vertebra, S1 is first vertebra in the sternum. Skeletal abnormalities were similar in mice lacking SLC35C1 and mice null for both SLC35C1 and SLC35C2. White asterisks placed near disorganized cartilage reflect bone defects in thoracic and lumbar regions. Black asterisks are placed near rib and tail defects giving examples of bifurcated, broken, bent, or missing ribs or tail. Detailed skeletal abnormalities are described for individual mice in Tables S2–S4.

Figure 5

Fucosylation of NOTCH1 in Slc35c1:Slc35c2 null lung.A and B, portions of lung or liver from nine 28 day pups of the indicated genotypes were solubilized as described in Experimental procedures and proteins analyzed by lectin blot using biotinylated AAL alone or biotinylated AAL in the presence of 0.5 M fucose. Antibody to ACTB detected relative loading of each sample. The dotted line shows where the blot was cut prior to incubation with respective antibodies. C, NOTCH1 was immunoprecipitated from lung lysate of 28 day pups of the indicated genotype as described in Experimental procedures and Fig. S6. Antibody to NOTCH1 ECD detected immunoprecipitated NOTCH1 (NOTCH1 IP+Ab) and NOTCH1+AAL detected immunoprecipitated, fucosylated NOTCH1. AAL, Aleuria aurantia lectin; ECD, extracellular domain.

Figure 6

Expression of related Slc35 genes in Slc35c1:Slc35c2 null lung and liver.A, body, lung, and liver weights of 28-day pups of the indicated genotype. B and C, relative expression was determined by comparing Gapdh and Actb expression with candidate transporter genes by real time quantitative reverse transcription polymerase chain reaction as described in Experimental procedures. Control pups were Slc35c1[+/−]Slc35c2[+/−] or Slc35c1[+/−]Slc35c2[−/−] (Control) compared to Slc35c1[−/−]Slc35c2[−/−] double null pups. Error bars are SEM. p < 0.05 (∗), p < 0.01 (∗∗), p < 0.005 (∗∗∗), and p < 0.0001 (∗∗∗∗) determined by two-tailed Student’s t test with Welch’s correction.