Fig 2: N272 glycosylation site variants.(A) WB analysis of supernatants from transfected WT and N272 variant constructs under reducing conditions. Recombinant expression of N271del (lane 3) and N272del (lane 5) was decreased. N271A, N272A, N272D, N271-N272del, K273del, and S275del (lanes 2, 4, 6, 7, 8, and 9) had normal secretion comparable to WT (lane 1). See Table 1. N272-linked glycan site has an atypical N-glycosylation consensus sequence, NNKIS, which is highlighted in blue, purple, orange, and red. (B) WB analysis of variants N272del and K273del before and after treatment with glycosidases. Before treatment, K273del (lane 3) had a slightly higher Mr compared with N272del (lane 2). After treatment, WB demonstrates that N272del (lane 5) and K273del (lane 6) had the same Mr. Human purified C1-INH was used as a positive control (lane 7, before treatment; lane 8, after treatment). Adopted from Ren et al. (6). Δ, post-degylcosylation. (C–F) Functional analysis of the N272 glycosylation site variants. (C and E) Absorbance is plotted against protein concentration. (D and F) Comparison of PKa and FXIIa binding between WT and N272 glycosylation site variants. The binding affinity of N271A, N272A, and N272D for PKa and FXIIa was comparable to WT, whereas N271del, N272del, K273del, and N271-N272del exhibited impaired binding activity to both substrates. Interestingly, S275del exhibited a markedly decreased binding to PKa, but not to FXIIa. Data represent mean ± SEM of 3 separate experiments. ***P < 0.001, ****P < 0.0001 by 1-way ANOVA with Dunnett’s multiple-comparison test. (G) Structural analysis of N272del and K273del. K273 is located in the loop immediately after hF. The deletion of K273 does affect the conformation of hF. K273del, previously reported, results in a new N-glycosylation site in C1-INH (37). The N271 residue is shown in pink, N272 in blue sphere, and K273 in purple.

Fig 3: Illustrations of N-glycans on the SERPIN domain in C1-INH and a C1s̄/C1-INH complex.(A) Molecular model of N-glycans on the SERPIN domain in C1-INH (PDB: 5DU3). Molecular modeling was performed on the GlyCAM-Web tool, Glycoprotein Builder (https://glycam.org). The N-linked glycan structure was generated based on mass spectrometry results (14). The N-GlcNAc linkage conformation was based on the simulation generated from Glycoprotein Builder. (B) Structure of a C1s̄/C1-INH complex (PDB: 8W18). The structure of active C1s̄ is shown in a cartoon representation in rainbow. The RCL is in violet. The P1 R466 and P1′ T467 residues are displayed as rainbow sticks and are responsible for trapping C1s̄. The disulfide bridges are labeled and colored in red. Recombinant C1s̄ harbors an S632A mutation, making it catalytically inert (PDB: 8W18) (9). Figures were produced using PyMOL (3.0).

Fig 4: N238 glycosylation site variants.(A) Western blot (WB) analysis of supernatants from transfected WT C1-INH and variant constructs of N238 under reducing conditions. The recombinant expression of N238A and A239D was comparable to WT (see Table 1). The secretion of N238del was markedly decreased compared with that of WT. The consensus sequence of N238 glycosylation NXS/T is highlighted in blue, orange, and red. (B–E) Functional analysis of N238 variants. (B and D) Absorbance is plotted against protein concentration. (C and E) Relative absorbance (RA) was computed as the absorbance of the variant divided by the absorbance of the WT at concentrations of 1 μg/mL, 500 ng/mL, 250 ng/mL, and 125 ng/mL. Data represent 3 separate experiments, with bars corresponding to SEM. *P < 0.05 for the percentage difference in FXIIa binding between N238A and WT by 1-way ANOVA with Dunnett’s multiple-comparison test. The P value for the percentage difference in FXIIa binding between A239D and WT is 0.358. For PKa binding, the P values for the percentage differences of N238A and A239D compared with WT are 0.918 and 0.254, respectively. (F) Structural analysis of N238 (PDB: 5DU3). N238 is located on the surface of helix E (hE). N238del results in the disruption of the consensus sequence NXS, which is required for the attachment of N-glycan. In the absence of glycosylation, N238del likely leads to protein misfolding.

Fig 5: Illustration of key domains in C1-INH.The reactive center loop (RCL) is in pink. The regions involved in SERPIN function are labeled. The P15-P9 portion of the RCL, the hinge domain, is highly conserved and facilitates the insertion of RCL into β-sheet A (SA). The breach region lies on the top of SA, the initial insertion site of RCL. The shutter domain, composed of S3A and S5A, is in the center of SA and facilitates the RCL insertion. The gate region consists of strands 3 and 4 from β-sheet C (S3C and S4C). N-glycosylation sites are shown as blue spheres. The P1 and P′ are displayed as rainbow sticks and are responsible for trapping the target protease. Strands of central SA are in cyan. The 2 disulfide bridges are labeled and colored in red. Protein structures used for modeling were obtained from the PDB database (PDB: 5DU3). The figure was generated using Pymol (3.0) and serves as a model for structure analyses in Figures 4–7. NTD, amino-terminal domain; CTD, carboxyl-terminal domain.