Structure of the polyisoprenyl-phosphate glycosyltransferase GtrB and insights into the mechanism of catalysis

Abstract

The attachment of a sugar to a hydrophobic polyisoprenyl carrier is the first step for all extracellular glycosylation processes. The enzymes that perform these reactions, polyisoprenyl-glycosyltransferases (PI-GTs) include dolichol phosphate mannose synthase (DPMS), which generates the mannose donor for glycosylation in the endoplasmic reticulum. Here we report the 3.0 Å resolution crystal structure of GtrB, a glucose-specific PI-GT from Synechocystis, showing a tetramer in which each protomer contributes two helices to a membrane-spanning bundle. The active site is 15 Å from the membrane, raising the question of how water-soluble and membrane-embedded substrates are brought into apposition for catalysis. A conserved juxtamembrane domain harbours disease mutations, which compromised activity in GtrB in vitro and in human DPM1 tested in zebrafish. We hypothesize a role of this domain in shielding the polyisoprenyl-phosphate for transport to the active site. Our results reveal the basis of PI-GT function, and provide a potential molecular explanation for DPM1-related disease.

Figure 1. GtrB catalyses the formation of UndP-glucose in vivo and in vitro.

(a) GtrB (TM domain in blue, GT domain in orange) catalyses the transfer of a glucose from UDP-glucose to UndP, producing UndP-glucose. (b) Cells overexpressing GtrB accumulate a species detected by negative ion ESI/MS at m/z of 1,007.75, corresponding to the [M-H]-ion of UndP-glucose. Its identity was further confirmed by tandem mass spectrometry (MS/MS). (c) When incubated with varying concentrations of UndP for 1 h, membranes prepared from GtrB-overexpressing cells accumulate UndP-glucose (UndP-Glc) in a concentration-dependent manner (circles), whereas those prepared from control cells overexpressing a membrane protein of unrelated function do not (squares). Error bars are provided as ±s.e.m., n=3.

Figure 2. Architecture of GtrB, an integral membrane PI-GT.

The tetrameric assembly of GtrB, in which each of the four polypeptide chains is represented as a ribbon drawn in a different colour, is shown here (a) from a position orthogonal to the TM helices and (b) looking down the fourfold symmetry axis of the tetramer from the extracellular side. (c) A single protomer is shown. The protomer is comprised of a GT-A fold GT domain (orange), two amphipathic juxtamembrane helices (grey; juxtamembrane (JM)), two TM helices (blue) and a C-terminal β-hairpin (red). (d,e) The GtrB tetramer is shown in the same two views and orientations as in a,b, with three subunits represented in grey ribbons and the fourth in rainbow colouring from the N-terminus (blue) to the C-terminus (red). (f) The GtrB monomer shown in the same orientation as in c, represented in rainbow colouring from the N-terminus (N; blue) to the C-terminus (C; red).

Figure 3. The catalytic mechanism of GtrB.

(a) The disposition of key residues in the active site and interactions with the acceptor and donor substrates. An anomalous difference Fourier map (contoured at 5 σ above the mean) calculated from a tungstate-soaked wild-type (WT) crystal is represented as purple mesh. Residues D94, D96, R122, D157, R200 and the UDP are shown in stick representation. Mn²⁺ is represented as a purple sphere. (b) Mutation of key residues in the acceptor and donor sites abolishes GtrB activity. (c) The catalytic mechanism of GtrB does not require a catalytic base. Error bars are provided as s.e.m., n=3.

Figure 4. A putative mechanism for substrate translocation in GtrB and DPM1 function in a zebrafish model.

(a) A speculative hypothesis is that the phosphate headgroup of UndP (red) could first bind at R290, near the cytoplasmic face of the inner membrane. (b) The substrate could then diffuse along a pathway lined with conserved hydrophobic (teal) and positively charged (blue) residues. (c) Finally, the phosphate headgroup is coordinated by R122 and R200 at the acceptor site, where catalysis occurs. (d) Mutation of hydrophobic and positively charged conserved residues lining the region between JM1 and JM2 abrogates GtrB activity. Error bars are provided as s.e.m., n=3. (e) The architecture of GtrB may be shared by the DPMS complex. CDG1e mutations G152V and S248P in the conserved JM1 and JM2 linkers are represented in purple and red, respectively. (f) Phenotypes typical of dpm1 loss-of-function mutations are fully evident at 2 days post fertilization (d.p.f.) in the zebrafish embryo. Morphants show smaller head (microcephaly) and smaller eyes, kinked tail and occasional vascular defects in the tail vein (arrowhead). Scale bar, 500 μm (g) Functional analysis of DPM1 mutants was performed by injecting human DPM1 mRNA following dpm1 morpholino injection and embryos were scored as normal or affected. Human DPM1 mRNA was able to improve the morphant phenotypes (see Supplementary Fig. 8f for details) and the effect of different mutations was normalized to rescue levels. mRNA carrying the R147A or R234A mutations (equivalent to R122 and R200 in GtrB, respectively) showed almost complete loss of function. Known human missense changes (R92S, G152V and S248P) were also tested. R92S and G152V greatly abolished protein function, whereas S248P, which leads to milder human phenotypes, showed substantial residual activity. The presumed location of G152V and S248P on the juxtamembrane region is shown. R92S is expected to reside close to the active site in the GT domain. Error bars: s.e.m., **P<0.01, ***P<0.001, unpaired Student’s t-test for each mutant, n=6 (R147A), n=6 (R234A), n=3 (R92S), n=5 (G152V), n=4 (S248P). NS, not significant.