Fig 2: Purified recombinant HAI-2 proteins repressed cellular matriptase activation and prostate cancer cell motility. (A,B) Effects of purified recombinant HAI-2 proteins on matriptase in N2 (A) and PC-3 cells (B). Cells were treated with the indicated concentrations of rHAI-2 (Cat.1106-PI-010, R&D system, MN, USA, which was produced from mouse myeloma cells) and incubated for 20 hours. The conditioned media and cells lysates were collected for immunoblotting analysis using anti-HAI-2 pAb, M32 and M69 mAbs to detect the levels of rHAI-2 in the conditioned media and matriptase (total and activated levels) in cells, respectively. ß-actin was used as control. (C,D) Effects of purified recombinant HAI-2 (rHAI-2) proteins on N2 (C) and PC-3 (D) cell migration and invasion. Cells were seeded at a density of 2 × 105 (N2) or 5 × 104 (PC-3) cells per well of Boyden chambers and treated with the indicated concentrations of rHAI-2 proteins. Boyden chambers were coated with or without Matrigel before cell seeding for cell invasion and migration assays. Migration assays were performed for 24 hours (N2) and 16 hours (PC-3) and invasion assays were carried out for 48 hours (N2) and 24 hours (PC-3). The results of cell migration and invasion were statistically calculated from three independent experiments and represented as mean ± SD. (*p < 0.05; **p < 0.01, one-way ANOVA).

Fig 3: Dynamic bindings and computational molecular docking between HAI-2’s KDs and matriptase’s protease domain. (A,B) Examination of the dynamic bindings of GST-KD1 (A) and GST-KD2 (B) to matriptase’s protease domain using BIAcore assay. The purified recombinant HAI-2’s KD1 and KD2 GST-fusion proteins were dialyzed and diluted in running buffer to the indicated concentrations. One hundred nanomolar of GST fusion proteins were used as a negative control. To measure the dynamic bindings between HAI-2’s KDs and matriptase’s protease domain, the indicated concentrations of the analytes (HAI-2’s KD GST-fusion proteins) were injected to Trx-MTX PD-immobilized CM5 chip at 30 µl/min for 120 sec. The dissociation was carried out by passing running buffer 30 µl/min for 180 sec. The estimated dissociation constants (KD) were statistically calculated by the affinity analysis of BIAevaluation software v1.0 (GE Lifescience, CT, USA). (C) The amino acid sequence alignment of the HAI-1’s and HAI-2’s KD1 and KD2 functional regions. The amino acid sequence alignment was performed by the MUltiple Sequence Comparison by Log-Expectation (MUSCLE, EMBL-EBI). ? indicated the important residues which potentially interacted with matriptase’s protease domain40. (D) The neighbor-joining tree of phylogenetic analysis was performed by MUSCLE (EMBL-EBI). The values mean the genetic distances for each KD. (E,F) The computational docking models of HAI-2 KD1 (E) and KD2 (F) with matriptase’s protease domain. The structural models of matriptase’s protease domain (Matriptase PD, PDB ID: 4ISO) and HAI-2 KD1 (PDB ID: 4U32) were downloaded from Protein Data Bank. HAI-2’s KD2 was constructed by the computational docking according to HAI-2’s KD1 (PDB ID: 4U32). The best-fit docking models of HAI-2’s KD1-matriptase PD and HAI-2’s KD2-matriptase PD were generated by ClusPro 2.0 (https://cluspro.bu.edu/)77–81 and drew by CCP4mg software (MRC Laboratory of Molecular Biology, Cambridge, UK). (G) The superposition of the docking models of KD1 and KD2 with matritpase’s protease domain. (H/I) The interaction residues between the active site in HAI-2’s KD1 (R48) (H) and KD2 (R143) (I) and matriptase’s protease domain (catalytic site S190) in the docking models. (J) The superposition of the residue R48 in HAI-2’s KD1 and the residue R143 in HAI-2’s KD2 with the residue S190 in matriptase’s protease domain in the docking models.

Fig 4: Characterization of HAI-2’s KDs in inhibiting matriptase activation and prostate cancer cell motility. (A) Schematic structures of wild-type HAI-2 and its variant mutants with domain deletions or active site mutations. (B) N2 cells were stably transfected with the plasmids encoding the wild type and mutants of HAI-2. Western blot revealed the levels of HAI-2 protein, total matriptase and activated matriptase using anti-c-Myc, M32 and M69 mAbs, respectively. ß-actin was shown as control. (C/D) Examination of the effects of HAI-2 mutant proteins on N2 cell invasion (C) and migration (D) using transwell assays. N2 cells which stably expressed HAI-2 mutant proteins were seeded at a density of 3 × 105 cells per upper chamber of Boyden chambers and incubated for 16 hours. Data were statistically calculated from three independent experiments and shown as mean ± SD. (*p < 0.5; **p < 0.01; ***p < 0.001, one-way ANOVA).

Fig 5: Role of HAI-2 in repressing matriptase activation and prostate cancer cell motility. (A) Examination of CWR22Rv1, 103E and N2 cell migration and invasion using transwell assays. Cells were seeded at a density of 2 × 105 cells per well into a Boyden chamber coated with or without Matrigel. Regular culture medium containing 10% FBS was used as a chemoattractant in lower wells. The periods of incubation time were 24 hours for migration assays and 48 hours for invasion assays. The migrating and invasive cells on the bottoms of Boyden chamber were stained with 1% crystal violet and imaged by a microscopy (magnification, ×100). Three independent experiments were performed and a set of representative images were shown. (B) Quantification of the migrating and invasive cells in Fig. 1A by ImageJ software. The data were statistically calculated from of three independent experiments and presented as mean ± SD. (**p < 0.01; ***p < 0.001, one-way ANOVA). (C) The activation process of matriptase and the schematic structure of matriptase and HAI-1 on plasma membrane. Upon activation, latent matriptase (zymogen, 70 kDa) can undergo a cleavage process at the residue of R614 by autoactivation, androgen-induced TMPRSS247, COX-299 or ErbB-2 signaling45, to become active and immediately form a complex with HAI-1 (120 kDa). The activated matriptase-HAI-1 complexes were used to generate monoclonal antibodies and two monoclonal antibodies (M32 and M69) were used in this study. M32 can recognize the third LDLR domain of matriptase, and M69 can specifically interact with the activated protease domain of matriptase100, as shown in the figure. Matriptase is composed of an intracellular region in the amino terminus, followed by a transmembrane domain, one SEA, two CUB, four LDLR domains and a protease domain in the carboxyl terminus. HAI-1 has two Kunitz domains (KD1 and KD2) in the amino terminus, one LDLR domain between these two KDs, a transmembrane domain and a short intracellular region in the carboxyl terminus. Matriptase’s and HAI-1’s domain abbreviations: SEA, Sperm protein, Enterokinase and Agrin; CUB, C1r/C1s, Uegf and Bmp1; LDLR, LDL receptor100. (D) Immunoblot analysis of total and activated matriptase as well as HAI-2 in CWR22Rv1, 103E and N2 cells. Forty micrograms of cell lysate per sample were separated by SDS-PAGE and immunoblotted using monoclonal M32, monoclonal M69 and polyclonal anti-HAI-2 antibodies to detect total matriptase (activated and latent matriptase), activated matriptase, and HAI-2 proteins, respectively. ß-actin was used as a control.