The EGF repeat-specific O-GlcNAc-transferase Eogt interacts with notch signaling and pyrimidine metabolism pathways in Drosophila

Abstract

The O-GlcNAc transferase Eogt modifies EGF repeats in proteins that transit the secretory pathway, including Dumpy and Notch. In this paper, we show that the Notch ligands Delta and Serrate are also substrates of Eogt, that mutation of a putative UDP-GlcNAc binding DXD motif greatly reduces enzyme activity, and that Eogt and the cytoplasmic O-GlcNAc transferase Ogt have distinct substrates in Drosophila larvae. Loss of Eogt is larval lethal and disrupts Dumpy functions, but does not obviously perturb Notch signaling. To identify novel genetic interactions with eogt, we investigated dominant modification of wing blister formation caused by knock-down of eogt. Unexpectedly, heterozygosity for several members of the canonical Notch signaling pathway suppressed wing blister formation. And importantly, extensive genetic interactions with mutants in pyrimidine metabolism were identified. Removal of pyrimidine synthesis alleles suppressed wing blister formation, while removal of uracil catabolism alleles was synthetic lethal with eogt knock-down. Therefore, Eogt may regulate protein functions by O-GlcNAc modification of their EGF repeats, and cellular metabolism by affecting pyrimidine synthesis and catabolism. We propose that eogt knock-down in the wing leads to metabolic and signaling perturbations that increase cytosolic uracil levels, thereby causing wing blister formation.

Figure 1. Human EOGT requires a DXD motif for optimal activity.

(A) Human EOGT is active in S2 cells. Western blots of lysates (top) and immunoprecipitates from conditioned medium (bottom) from S2 cells treated with (+) or without (−) dsRNA against eogt, and transfected with N(EGF1-20)-AP and GFP, Ago61 or EOGT, as noted. The target of each antibody is shown on the right. Arrow identifies Ago61 band; *identifies non-specific bands. (B) Alignment of putative catalytic regions of Eogt homologs. % identity with the full-length human protein is shown on the right. C. gri: Cricetulus griseus, T. adh.: Trichoplax adhaerens, C. int.: Ciona intestinalis. (C) Western analysis of lysates from S2 cells treated with eogt dsRNA as noted, and transfected with GFP, EOGT wild-type (DYD), or EOGT mutant (AYA) cDNA. Mutant EOGT(AYA) was less active even though consistently expressed at much higher levels. (D, E) The proposed Eogt consensus site is present in several EGF repeats of Drosophila Dl (D) and Ser (E). (F) Western analysis of lysates from S2 cells transfected with soluble extracellular domain of His-tagged Dl (Dl-His) or Ser (Ser-His) and either EOGT or Ago61 cDNA as indicated.

Figure 2. Human EOGT can substitute for Drosophila eogt in vivo.

(A) Schematic of the eogt locus. The region deleted in eogt ^ex10 by imprecise excision of P-element BG00673 is indicated by a dotted line and flanking sequences. Coordinates of the deletion are given relative to the start codon. The deletion removed the start codon and 224 aa of the coding region. Locations of the dsRNA Shigen/R-3 and VDRC/44572 are shown as grey bars on top. (B) Homozygous eogt^ex10 mutants die in L2. Number of dead offspring at indicated stages of an eogt ^ex10/CyO, twi-GFP strain. Dead embryos expressed GFP and were probably homozygous for the balancer. Animals dying in L1-L3 did not express GFP and were thus homozygous for eogt ^ex10. All animals that survived to adulthood were heterozygotes carrying CyO (n = 89). (C) Drosophila eogt (tub>eogt) and human EOGT (tub-EOGT), but not mouse Ago61 (tub-Ago61), rescued eogt ^ex10 animals that obtained a transgene. (D) Western blots of adult fly lysates confirm that human EOGT and Ago61 were expressed in the stocks assessed for rescue in (B). *non-specific band.

Figure 3. Loss of eogt phenocopies loss of dp in vivo.

(A–C) Compared to control clones (A), Ubx-Flp-induced eogt^ex10 clones (B) caused vortex phenotypes on the thorax (arrow), similar to dp^lv clones (arrow in C). (D–F) ap-Gal4-mediated RNAi knock-down of eogt phenocopied the vortex phenotype of eogt mutant clones (arrow in E), similar to dp knock-down (arrow in F). Note that ap-Gal4 alone had a different bristle phenotype (D). (G–I) Compared to control clones (G), eogt^ex10 (H) and dp^lv clones (I) caused deformed wings with blisters.

Figure 4. Molecular targets of Eogt.

(A) Western analysis using mAb CTD110.6 of early pupal lysates of tub>dp ^IR and control siblings. An ∼75 kDa band served as loading control. The GFP signal in the lower panel confirmed the respective genotypes (i.e. all dp knock-down pupae contained tub-Gal4 and not the TM3, Ser, act-GFP balancer over which it was kept). (B) Western analysis of control w¹¹¹⁸ and ogt mutant sxc⁶, eogt^ex10 and sxc⁶, eogt^ex10 double mutant L2 larval extracts to detect O-GlcNAcylated proteins using mAb CTD110.6 (upper panel). α−Tubulin was used as loading control (lower panel).

Figure 5. Temperature-sensitive wing blister assay for eogt interactors.

Wings of flies raised at 22.5°C (A, C, E) or 27°C (B, D, F). en>eogt^IR wings were normal at 22.5°C (A) but blistered in the posterior compartment at 27°C (B). At the low temperature, blistering was induced when one gene dose of eogt (eogt^ex10/+) or dp (dp^olvR) was removed (C and E, respectively). While blistering at the higher temperature was not affected by co-expression of Ago61 (D), it was suppressed by co-expression of human EOGT (F), even when one gene dose of eogt had been removed.

Figure 6. N gene dose influences wing blister frequency in en>eogt^IR wings.

The histogram depicts % animals with blisters of the genotype indicated below expressed in en>eogt^IR (Baseline). N dose is indicated on top. Total number of animals of each genotype is shown at the base of each column. P-values were calculated using the Two-proportion Z-test. ***p<0.001; *p<0.05; ns, not significant.

Figure 7. Wing blister phenotypes of en>eogt^IR interactions with N and pyrimidine metabolism mutant alleles.

Adult wings from flies with the eogt^IR chromosome and the alleles shown developed at the indicated temperature. Several alleles of N suppressed the wing blister phenotype due to eogt knock-down (A, C, D). (B) N^55E11 suppression was reverted by a genomic N transgene. (E, H) Interactions of en>eogt^IR with pyrimidine metabolism mutants. (E) Dhod⁸ suppressed the wing blister phenotype of en>eogt^IR. (F) The suppression of en>eogt^IR by r^70b was reverted by a transgene encoding constitutively active R^Su(b). (G) Example of a wing with blister from a fly with one null allele of pyd3 (pyd3^Lb5). (H) Example of a blistered wing of a fly overexpressing pyd3 in an en>eogt^IR; pyd3^Lb10/+ background.

Figure 8. Interactions between Eogt, pyrimidine metabolism and Notch signaling in the posterior wing.

(A) The diagram shows the Eogt-catalyzed addition of O-GlcNAc from UDP-GlcNAc (UDP-blue square) to EGF-containing proteins Dp and N, and key steps of the pyrimidine synthesis and catabolism pathways in a wild-type wing cell. Repression of pyrimidine neo-synthesis by Dp was shown biochemically in dp mutant larvae , . Protein products of genes tested for interaction with en>eogt^IR flies are indicated. (B) In en>eogt^IR wings, reduced Eogt leads to loss of O-GlcNAc on Dp, N, Dl, Ser and other EGF-containing substrates. Genetic interactions with mutant alleles that resulted in suppression of wing blisters at 27°C are in green, while those that caused enhancement of wing blisters at 22.5°C are in magenta. Enhanced activity of initial enzymes in pyrimidine synthesis due to reduced Dp function is indicated by gray lines. The combined data suggest the unifying model that an increase in cytoplasmic uracil concentration is a likely cause of wing blisters when Eogt levels are reduced. The loss of O-GlcNAc from Dp and N may also contribute to the wing blister phenotype by reducing signals that influence pyrimidine biosynthesis.