N‐glycosylation mediated folding and quality control in serine proteases of the hepsin family

N‐linked glycans are specifically attached to asparagine residues in a N‐X‐S/T motif of secretory pathway glycoproteins. N‐glycosylation of newly synthesized glycoproteins directs their folding via the lectin chaperones calnexin and calreticulin that are associated with protein‐folding enzymes and glycosidases of the endoplasmic reticulum (ER). Misfolded glycoproteins are retained in the ER by the same lectin chaperones. The work by Sun et al. (FEBS J 2023, 10.1111/febs.16757 ) in this issue focusses on hepsin, a serine protease on the surface of liver and other organs. The authors deduce that spatial positioning of N‐glycans on one side of a conserved domain of hepsin, known as the scavenger receptor‐rich cysteine domain, regulates calnexin selection for hepsin maturation and transport through the secretory pathway. If N‐glycosylation is elsewhere on hepsin, then it is misfolded and has a prolonged accumulation with calnexin and BiP. This association coincides with the engagement of stress response pathways that sense glycoprotein misfolding. The topological considerations of N‐glycosylation dissected by Sun et al. may help unravel how key sites of N‐glycosylation sites required for protein folding and transport have evolved to select the lectin chaperone calnexin pathway for folding and quality control.


Introduction
Calnexin mediates the folding of newly synthesized Nlinked glycoproteins through a transiently exposed mono-glucosylated (Glc1 Man9 GlcNac2) residue that is now well understood with the roles of several of the associated protein folding enzymes having been resolved [1]. Central to the quality control of secretory N-linked glycoproteins is the glucosyl transferasemisfolded glycoprotein sensor UGGT [2]. UDPglucose:glycoprotein glucosyl transferase (UGGT) readds a single glucose to an incomplete or misfolded glycoprotein for repeated founds of lectin chaperonemediated folding or prolonged retention of misfolded glycoproteins.
Sun et al. [3] in this issue, add to our knowledge of the important role of N-glycosylation spatial positioning in secretory pathway glycoproteins for folding and quality control [4][5][6]. The new study considers the scavenger receptor cysteine-rich (SRCR) domain of hepsin a type II transmembrane serine protease on the plasma membrane of hepatocytes. Here, Sun et al. consider the single site of N-glycosylation in hepsin previously shown as necessary for association with calnexin, secretory transport and the acquisition of catalytic activity on the cell surface. The latter is a consequence of a cleavage of cell surface located hepsin to generate the catalytically active serine protease C-terminal domain itself linked by disulfide bonds to the SRCR domain and the membrane-spanning Nterminal domain of hepsin.
The authors were puzzled as to why other similar type II transmembrane serine proteases had been selected for alternative sites of N-glycosylation that were known to be required for their transport, processing and acquisition of catalytic activity. These include other type II transmembrane serine proteases, human enteropeptidase and corin whose key N-gycosylation sites required for maturation are not in the SRCR domain but rather in the C-terminal protease domains of the proteins. Accordingly, Sun et al. expressed tagged wild-type hepsin and mutants that removed the N-glycosylation site in the SRCR domain and mutants that added new sites in the protease domain corresponding to that in enteropeptidase or that in corin.
The authors assessed the generation of the Cterminal disulfide-bonded protease domain of hepsin and its catalytic activity. Both were diminished when the single site at N112 in hepsin was removed; the addition of new sites in the protease domain was unable to rescue cleavage to generate the C-terminal protease domain or catalytic activity.
The authors imaged tagged hepsin and the Nglycosylation mutants by immunofluorescence confocal microscopy with endoplasmic reticulum (ER) colocalization concluded with anti-KDEL for all mutants but not for wild-type hepsin.
Corin, enteropeptidase and the type II transmembrane serine protease known as TMPRSS4 are normally expressed with N-glycosylation also in their respective SRCR domains. Again, the authors tested by hepsin mutation if these new sites could rescue the loss of the key N112 site. All new sites were Nglycosylated, but again, such hepsin mutants showed diminished cleavage to generate the disulfide-bonded C-terminal domain. Hence, the wild-type Nglycosylation site at N112 in the SRCR domain of hepsin could not be replaced by those used in three other related type II transmembrane serine protease proteins. Sun et al. [3] now considered replacing the N112 site in hepsin with those at positions 67, 100, 113, 126, which are all in the SRCR domain. Although all sites could be N-glycosylated, only N100 or N113 was functional for hepsin maturation as assessed by cleavage of the C-terminal domain. Imaging also indicated that like wild-type hepsin, the mutants at sites N100 or N113 did not colocalize with anti-KDEL while the other mutants did.
The authors modelled the positions of the Nglycosylation sites on the known structure of hepsin. They mapped sites N100, N112 (wild-type) and N113 to one side of the SRCR domain. The N-glycosylation sites that did not lead to maturation, N67, N80, N126, N135 mapped to the opposite side. The authors extended the analysis to remove the wild-type N112 site and generate a new N59 N-glycosylation site of hepsin since it should map to the same side of the SRCR domain as the wild-type N112. The N59 site was N-glycosylated and hepsin maturation was comparable with wild-type hepsin. The authors concluded that N-glycosylation sites that enabled hepsin maturation required a single N-glycosylation site in the SRCR domain provided it was exposed and positioned to one side of the domain.
Sun et al. [3] next considered the roles of the chaperones of the ER. Both calnexin and BiP were increased in coIPs of tagged hepsin deleted for N112 or with the N-glycosylation site at N126 as compared with wildtype hepsin, or the sites at N59, N100 or N113, which promoted hepsin maturation. When the authors immunoprecipitated calnexin or BiP at different times of the expression of tagged wild-type hepsin or selected mutants, they observed at 12 h of expression increased association with the mutant deleted for Nglycosylation or at position N126, neither of which had led to hepsin correct folding. By 18 and 24 h of expression, wild-type and mutant hepsins competent or not for maturation were observed in calnexin or BiP co-immunoprecipitates.
Next, Sun et al. considered the consequences of stress sensors of protein misfolding in the ER. They observed elevated levels of Erp72, ATF4 and CHOP for hepsin N-glycosylation mutants that were diminished in maturation. XBP1 levels did not change after the expression of wild-type or any of the hepsin mutants competent or not for maturation. Sun et al. also assessed the soluble secretory serine protease with an SRCR domain, complement factor I, to conclude that the N-glycosylation site at position 177 in the SRCR domain as required for secretion. Modelling suggested an exposed N-glycosylation site similar in topology to that of the N112 site in wild-type hepsin.

Conclusion
Taken together the data suggest that it is the spatial position of the N-glycan in the SRCR domain of hepsin as determining correct hepsin folding. They hypothesize the relevance of the spatial accessibility of calnexin as related to the position for N-glycosylation required for hepsin maturation. An additional path of enquiry may consider the glycoprotein misfolding sensor, UGGT, that monitors the topology of Nglycosylation sites with respect to misfolded regions on glycoproteins [6]. It is repeated rounds of glucosylation of misfolded glycoproteins by UGGT that leads to their apparent retention on calnexin with misfolded glycoproteins also extended to BiP and activation of ER stress sensors. The stress response pathways deduced by Sun et al. are also an avenue for further study for hepsin and other type II transmembrane serine proteases. Progress in understanding the mechanisms of Nglycan-mediated pathways of folding and quality control will continue to provide insight into the evolution of N-glycosylation and the secretory pathway.