Site-specific O-glucosylation of the epidermal growth factor-like (EGF) repeats of notch: efficiency of glycosylation is affected by proper folding and amino acid sequence of individual EGF repeats

Abstract

O-Glucosylation of epidermal growth factor-like (EGF) repeats in the extracellular domain of Notch is essential for Notch function. O-Glucose can be elongated by xylose to the trisaccharide, Xylα1-3Xylα1-3Glcβ1-O-Ser, whose synthesis is catalyzed by the consecutive action of three glycosyltransferases. A UDP-glucose:protein O-glucosyltransferase (Poglut/Rumi) transfers O-glucose to serine within the O-glucose consensus. Subsequently, either of two UDP-xylose:glucoside xylosyltransferases (Gxylt1 or Gxylt2) transfers xylose to O-glucose. Finally, a UDP-xylose:xyloside xylosyltransferase (Xxylt1) transfers xylose to Xylα1-3Glcβ1-O-EGF. Our prior site-mapping studies demonstrated that O-glucose consensus sites are modified at high but variable stoichiometries in mouse Notch1 and identified a novel glycosylation site with alanine in place of proline, suggesting a revised, broader consensus sequence (CXSX(P/A)C). Here we examined the molecular basis for this site specificity. A panel of EGF repeats from human coagulation factor 9 (FA9), mouse Notch1, and Notch2 were bacterially expressed and purified by reverse phase HPLC for use in in vitro enzyme assays. We demonstrate that proper folding of EGF repeats is essential for glycosylation by Poglut/Rumi, that alanine can substitute for proline in the context of coagulation factor 9 EGF repeat for O-glucose transfer, confirming the new consensus sequence, and that positively charged residues within the O-glucose consensus sequence reduce efficiency of glycosylation by Poglut/Rumi. Moreover, proper folding of EGF repeats is also important for the activities of Gxylt1, Gxylt2, and Xxylt1. These results indicate that protein folding and amino acid sequences of individual EGF repeats fundamentally affect both attachment and elongation of O-glucose glycans.

FIGURE 1.

Poglut/Rumi differentiates folding isomers of EGF repeats. A, RP-HPLC analysis of untreated control or reduced/alkylated FA9 EGF repeat is shown. The masses of the species in each peak, which were determined by mass spectrometry, are shown. Note that the mass was increased by 347.3 after reduction and alkylation, which corresponded to modification with carbamidomethyl groups at all six cysteine residues. mAU, milli-absorbance units. B, Poglut activity of Poglut/Rumi using native (filled circles) or reduced and alkylated (open circles) FA9 EGF repeat is shown. The values indicate the mean ± S.E. C, elution profiles of mN1 EGF27 folding isomers (top, RT23.1 and RT26.1). RT23.1 (middle) and RT26.1 (bottom) eluted at the same elution time after reduction to break the disulfide bonds. Note that there was the same, six-mass increase for both species, which corresponded to reduction of all three disulfide bonds. D, after the folding isomers of mN1 EGF27 were incubated overnight with Pofut1 and LFng in the absence (top) or the presence (bottom) of GDP-fucose (Fuc) and UDP-GlcNAc, the reaction mixtures were applied to RP-HPLC. The elution profiles are shown, and masses of each species were determined. E, after the folding isomers of mN1 EGF27 were incubated with Poglut/Rumi in the absence (top) or the presence of UDP-glucose (Glc) (bottom), the reaction mixtures were applied to RP-HPLC. The elution profiles are shown, and determined masses are indicated. The theoretical masses of EGF repeats are summarized in supplemental Table 2.

FIGURE 2.

Amino acid sequence within the O-glucose consensus affects Poglut and Poxylt activities of Poglut/Rumi. A, RP-HPLC elution profiles are shown of the reaction mixtures of FA9 EGF repeat wild type (WT), P55A, P55R, or N54S incubated with Pofut1 and LFng in the absence (top) or the presence of GDP-Fuc and UDP-GlcNAc (bottom) for 10 h. mAU, milli-absorbance units. B, RP-HPLC elution profiles are shown of the reaction mixtures of FA9 EGF repeat WT, P55A, P55R, or N54S incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-xylose (Xyl) (bottom) for 10 h. Masses of each species were determined. When FA9 (P55A) was incubated with Poglut/Rumi and UDP-Glc, there was a peak containing unmodified substrate whose mass was detected as 5670.7 (indicated by *). C, Rumi Poglut (left) or Poxylt activity (right) toward FA9 WT, P55A, P55R, or N54S in kinetic assays is shown. mN2 EGF16 was also used as acceptor substrate. The values indicate mean ± S.E. D, elution profiles of the reaction mixtures of mN2 EGF16 incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-Xyl (bottom) for 10 h on RP-HPLC are shown, and masses of each species were determined.

FIGURE 3.

Poglut and Poxylt activities of Poglut/Rumi vary among EGF repeats. A, RP-HPLC elution profiles are shown of the reaction mixtures of mN1 EGF12 or EGF27 incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-Xyl (bottom) for 10 h. mAU, milli-absorbance units. B, shown is Rumi Poglut (left) or Poxylt activity (right) toward mN1 EGF12, EGF27, or mN2 EGF16 in kinetic assays. The values indicate the mean ± S.E.

FIGURE 4.

Positively charged amino acids in the O-glucose consensus sequence reduce Poglut activity of Poglut/Rumi. A, shown is the amino acid sequence of EGF repeats 12, 16, 20, 27, 28, and 33 from mN1. The O-glucose consensus site is underlined in both. B, variability of amino acid sequences of EGF repeats is shown by WebLogo (31). Alignment is shown of the primary sequence of 10 amino acids containing the O-glucose consensus sequence of 17 EGF repeats from mN1 (top), EGF repeats (only wild type) tested in the present study (second), EGF repeats that were better substrates for Poglut/Rumi (Rumi high, third), and EGF repeats that were poorer substrates for Poglut/Rumi (Rumi low, bottom). EGF repeats that were modified more efficiently (over 5 nmol/min/mg) at 10 μm in the Poglut assays were defined as Rumi high. This group contained EGF12, EGF20, and EGF33 from mN1, EGF16 from mN2, FA9 EGF repeat, and FA7 EGF repeat (15). EGF repeats, which showed lower efficiency, were defined as Rumi low. This group contained EGF16, EGF27, and EGF28 from mN1. C, shown are RP-HPLC elution profiles of the reaction mixtures of mN1 EGF27 (R1028N) incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-Xyl (bottom) for 10 h. Masses of each species were determined. mAU, milli-absorbance units. D, shown is Rumi Poglut activity toward mN1 EGF27 (WT), R1028N mutant, or mN2 EGF16 in kinetic assays. The values indicate the mean ± S.E. E, RP-HPLC elution profiles of the reaction mixtures of mN1 EGF16, EGF20, EGF28, or EGF33 were incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-Xyl (bottom) for 10 h. Masses of each species were determined. The asterisk indicates O-xylosylated EGF16, whose mass was 7186.5. F, Rumi Poglut activity toward mN1 EGF16, EGF20, EGF28, or EGF33 in kinetic assays is shown. The values indicate the mean ± S.E. G, shown are RP-HPLC elution profiles of the reaction mixtures of the H608A and H608D mutants of mN1 EGF16 incubated with Poglut/Rumi in the absence of donor (top) or in the presence of UDP-Glc (middle) or UDP-Xyl (bottom) for 10 h. Asterisks indicate O-xylosylated H608A or H608D of EGF16 whose mass was 7120.6 or 7165.6, respectively. H, shown is Rumi Poglut (left) or Poxylt activity (right) toward the wild type, H608A, or H608D mutant of mN1 EGF16 and mN2 EGF16 in kinetic assays.

FIGURE 5.

Gxylt1 or Gxylt2 and Xxylt1 can synthesize O-xylose trisaccharide on mN2 EGF16. A, mN2 EGF16 with O-xylose was incubated without enzyme or with Gxylt1, Gxylt1 and Xxylt1, Gxylt2, or Gxylt2 and Xxylt1 for 10 h in the presence of UDP-Xyl. Products were analyzed by RP-HPLC. mAU, milli-absorbance units. B, using mN2 EGF16 unglycosylated, modified with O-Glc, O-Xyl, or with Xyl-Glc disaccharide at 10 μm and 30 ng of each enzyme, the xylosyltransferase activities of Gxylt1 (B) or Gxylt2 (C) were analyzed. The values indicate the mean ± S.E.

FIGURE 6.

Gxylt1, Gxylt2, and Xxtlt1 prefer properly folded FA9 EGF repeat as substrate. A, RP-HPLC analysis of untreated or reduced and alkylated FA9 EGF repeat modified with O-Glc (left) or a Xyl-Glc disaccharide (right). The mass increase of Glc-FA9 or Xyl-Glc-FA9 after denaturing was 348.5 or 348.1, respectively, corresponding to carbamidomethylation of all 6 cysteines of the EGF repeats. mAU, milli-absorbance units. B and C, shown is the activity of Gxylt1 (B) and Gxylt2 (C) toward untreated or reduced and alkylated FA9 EGF repeat modified with O-Glc in kinetic assays. D, shown is activity of Xxylt1 toward untreated or reduced and alkylated FA9 EGF repeat modified with a Xyl-Glc disaccharide in kinetic assays. The values indicate the mean ± S.E.