Fig. 8
Increased mannose trimming to M7-5 by ERManI and especially by EDEM1 and EDEM2 on a denatured glycoprotein. a The addition of changes relative to control for M7, M6, and M5, for experiments in vitro using free N-glycans, native vitellogenin, and denatured vitellogenin. The graph shows data for ERManI, EDEM1, and EDEM2 (three independent experiments for ERManI and EDEM1 and two for EDEM2). (*P values ERManI native glycop. vs. free N-glycans 0.03, denat. vs. native glycop. 0.02, EDEM1 denat. glycop. vs. free N-glycans 0.04, denat. vs. native glycop. 0.02, Student’s t test (unpaired, two-tailed). b Working model illustrating the possible mechanism of increased mannosidase activity on partially unfolded or misfolded substrate glycoproteins. For ERManI, the folded protein moiety sterically hinders ERManI activity, leading to removal of only the terminal branch B mannose from most molecules of the native glycoprotein (upper panel), whereas the unfolded state allows engagement of the mannosidase (lower panel) and increased mannose trimming (red arrow). For EDEM1 and EDEM2 there is little or no activity on the folded molecules, whereas the unfolded state exposes determinants (possibly hydrophobic domains) on the substrate glycoprotein that allow protein–protein binding and engagement of the mannosidase with the glycans. For partially folded or misfolded glycoprotein molecules, containing disulfide bridges (lower panel), TXNDC11 and PDI, which are in complexes with EDEM1 and EDEM2, reduce the disulfides, leading to unfolding. The mannosidase activity produces M5-6, which bind to OS-9, which in turn targets the glycoprotein to ERAD. M7A (with branch A intact) could eventually bind to OS-9, but proteasomal inhibition causes accumulation of M5-6 on ERAD substrates and not M71,2; M7A can still be engaged in the calnexin folding cycle
