SPINT2 (HAI-2) missense variants identified in congenital sodium diarrhea/tufting enteropathy affect the ability of HAI-2 to inhibit prostasin but not matriptase

Abstract

The syndromic form of congenital sodium diarrhea (SCSD) is caused by bi-allelic mutations in SPINT2, which encodes a Kunitz-type serine protease inhibitor (HAI-2). We report three novel SCSD patients, two novel SPINT2 mutations and review published cases. The most common findings in SCSD patients were choanal atresia (20/34) and keratitis of infantile onset (26/34). Characteristic epithelial tufts on intestinal histology were reported in 13/34 patients. Of 13 different SPINT2 variants identified in SCSD, 4 are missense variants and localize to the second Kunitz domain (KD2) of HAI-2. HAI-2 has been implicated in the regulation of the activities of several serine proteases including prostasin and matriptase, which are both important for epithelial barrier formation. No patient with bi-allelic stop mutations was identified, suggesting that at least one SPINT2 allele encoding a protein with residual HAI-2 function is necessary for survival. We show that the SCSD-associated HAI-2 variants p.Phe161Val, p.Tyr163Cys and p.Gly168Ser all display decreased ability to inhibit prostasin-catalyzed cleavage. However, the SCSD-associated HAI-2 variants inhibited matriptase as efficiently as the wild-type HAI-2. Homology modeling indicated limited solvent exposure of the mutated amino acids, suggesting that they induce misfolding of KD2. This suggests that prostasin needs to engage with an exosite motif located on KD2 in addition to the binding loop (Cys47/Arg48) located on the first Kunitz domain in order to inhibit prostasin. In conclusion our data suggests that SCSD is caused by lack of inhibition of prostasin or a similar protease in the secretory pathway or on the plasma membrane.

Introduction

Congenital diarrheal disorders (CDD) represent a group of clinical conditions that present with life-threatening enteral loss of water and electrolytes in newborns, with a high proportion of prematurely born infants. CDD arise from alterations in the transport of nutrients and electrolytes across the intestinal mucosa, from enterocyte and enteroendocrine cell differentiation and polarization defects and from the modulation of the intestinal immune response. The number of CDD forms, as well as the number of underlying disease genes, are gradually increasing. Knowledge of these clinical characteristics and CDD genes enables non-invasive, next-generation sequencing-based testing to facilitate an early diagnosis (1). One of these disorders, a syndromic form of congenital sodium diarrhea (SCSD; Diarrhea 3, secretory sodium, congenital, syndromic, OMIM #270420) has been reported in less than 40 patients in the literature (2–8). SCSD is also referred to as a syndromic form of tufting enteropathy (OMIM #613217), on grounds of finding characteristic ‘tufts’ on intestinal histology, consisting of focal crowds of enterocytes within the epithelium. In addition, intestinal crypts might appear dilated with features of pseudo-cysts and abnormal regeneration with branching (9). Epithelial tufts are not encountered in every diagnostic intestinal biopsy from patients with syndromic or non-syndromic tufting enteropathy (2,3,6). The diarrhea is severe, most often requiring life-long total or partial parenteral nutrition (PN) or small bowel transplantation (2–6,10–13). By using the positional candidate approach, mutations causing syndromic and non-syndromic tufting enteropathy (SCSD) were identified in SPINT2, encoding Kunitz-type serine protease inhibitor 2 (SPINT2, also referred to as hepatocyte growth factor activator inhibitor 2 (HAI-2) and placental bikunin; MIM #605124) (3) and in EPCAM (OMIM #185535), encoding epithelial cell adhesion molecule (11). How HAI-2 variants identified in patients with SCSD cause diarrhea, sodium loss and malformations is incompletely understood. However, a mechanism for diarrhea development was recently suggested involving HAI-2, EpCAM and claudin-7 (14). It is well established that EpCAM stabilizes claudin-7 and claudin-1 (15), and loss of EpCAM function results in a redistribution of claudin-7 and modification of tight junctions in intestinal epithelial cells. It was recently shown that EpCAM is a substrate for the serine protease, matriptase, and it was suggested that HAI-2 carrying SCSD missense mutations are less efficient in inhibiting matriptase than wild-type (WT) HAI-2, resulting in decreased cellular levels of full-length EpCAM (14).

The human genome contains 115 annotated protease inhibitors responsible for regulating the activity of the 612 known human proteases (16). Thus, on average each inhibitor regulates several proteases. In line with this, HAI-2 has been shown to inhibit a number of proteases in vitro, including matriptase (17), hepatocyte growth factor activator (18), trypsin, plasma kallikrein, tissue kallikrein and plasmin (19), supporting its role as a protease inhibitor. Recently, HAI-2 has been purified from human milk in complexes with the activated forms of serine proteases matriptase [encoded by Suppression of Tumorigenicity-14 (PRSS14) gene] or prostasin (PRSS8), but not as free HAI-2 (20) supporting previous implications of these two serine proteases as HAI-2 targets. HAI-2 is a membrane-bound serine protease inhibitor that contains two Kunitz domains (KDs) [Kunitz Domain 1 (KD1) and Kunitz Domain 2 (KD2)] and is expressed in most epithelial cells. Each KD displays a typical pear-shaped fold with the binding loop at the top (21). Kunitz-type inhibitors are known to competitively prevent access of physiologically relevant substrates to the serine proteases by inserting their P1-P4 residue, located in the binding loop, into the active site cleft of the serine protease. The binding loop mimics the substrate and forms an interaction resembling an enzyme-substrate Michaelis complex (22). In HAI-2, Arg48-Cys47 and Arg143 serve as P1-P2 residues in KD1 and as P1 in KD2, respectively. HAI-2 Cys47Phe was designed based on the SNP rs1804770. rs1804770 has not been associated with a phenotype nor observed in any patient with SCSD. HAI-2 Arg48Leu and HAI-2 Cys47Phe/Arg48Leu are laboratory-generated mutations. We have recently shown that mutations in the binding loop of HAI-2 KD1 lead to a reduced ability of HAI-2 to inhibit matriptase in the secretory pathway (23). Matriptase is known to play a crucial role in maintaining functional epithelial barriers and has been linked to cancers originating from epithelial cells and to lethal barrier defects in the intestine in mice (24,25). Depletion of either matriptase or prostasin from Caco-2 cells causes defects in intestinal epithelial barrier formation and tight junction assembly (24,26,27). It has been shown in a Xenopus Laevis oocyte system that one of the HAI-2 mutants found in SCSD patients, Tyr163Cys, is unable to inhibit the serine proteases prostasin and tmprss13, in contrast to WT HAI-2 (28).

Here, we report the clinical course of three additional patients with SCSD in the context of previously published cases. We identify two novel SPINT2 mutations, expanding the mutational spectrum to 13 variants. We further show that the SCSD-associated HAI-2 variants Phe161Val, Tyr163Cys and Gly168Ser are fully able to inhibit matriptase but display reduced ability to inhibit prostasin. We show that the HAI-2-mediated inhibition of prostasin depends on both the binding loop of KD1 and the interaction of an exosite located on KD2, which is disturbed by the SCSD-associated mutations, Phe161Val, Tyr163Cys and Gly168Ser.

Results

Clinical presentation of SCSD

Patient 1 is a 7-year-old girl, born to healthy, unrelated Dutch parents. Mild intra-uterine growth retardation was seen on routine ultrasound in the third trimester, and she was born after 38 weeks of gestation by cesarean section weighing 2455 g [percentile 1 (p1)], length, 44 cm (p1), head circumference, 32 cm (p3) and Apgar scores of 1/6/9. She had bilateral choanal atresia, uvula bifida, pre-auricular pits and single horizontal palmar creases. She passed meconium during delivery. The first 3 weeks of her life she passed stool of appropriate consistency 2–5 times a day, there was no vomiting and she was fed 60 cc seven times a day. However, she did not grow well, so extra kilocalories were added. After 2–3 weeks she acutely developed severe diarrhea and metabolic acidosis. The enteral nutrition was stopped for 3 days and the diarrhea ceased as well. Duodenal biopsies showed variable partial villous atrophy with normal brush border staining on PAS and CD10. There was no evidence of an inflammatory reaction or enterocyte vacuolation. In retrospect, there were a few areas display-ing tufts or dilated crypts as described in tufting enteropathy. Other biopsies (antrum, esophagus, colon) were found to be normal. At 10 weeks of age, her fecal [Na⁺] was 76 mmol/L (ref.: 20–50 mmol/L) and pH was 6.5 (ref.: 5.3–7.5), while her serum [Na⁺] was normal (ref.: 136–146 mmol/L), and urine [Na⁺] and [Cl^-] were 149 and 214 mmol/L (ref.: 10–226 and 10–210 mmol/L), respectively. At this point, the diarrhea deteriorated because of an enteral Klebsiella infection. Since that age, she is on 100% total PN (TPN), with additional enteral tube feeding of 60 ml vaminolact and 40 ml water/day and she drinks additional 10–15 cc water six times a day. Oral Na⁺, K⁺ and citrate⁻ supplementation was started at that time and she currently receives 100 ml NaK citrate solution (360 mmol Na/Liter and 240 mmol K/liter) per day and two times a day 15 ml Nabic 8.4% intravenously. Based on the pattern of a severe congenital diarrhea and choanal atresia, targeted SPINT2 gene mutation analysis was initiated and confirmed the clinical suspicion of SCSD. Endocrine causes of secretory diarrhea were not formally excluded. Intermittently, the liver was enlarged and ultrasound investigations showed an inhomogeneous liver parenchyma, indicative of cirrhosis, but at follow-up consistent with steatosis. Neither cholestasis nor abnormal liver synthesis parameters were seen and transaminases were mildly elevated. At the age of 4 years and 9 months, her weight was 16.7 kg (p39), length, 104 cm (p36) and BMI, 15.4 kg/m2 (p58). At the age of 7 years and 3 months she weighs 21.5 kg (p25), length is 113 cm (<p1) and BMI is 16.8 kg/m² (p75). Urine [Na⁺] is 109 mmol/L and urine osmolarity 516 mmol/L. For TPN, she has received several replacements of central venous catheter. Neither oral feeding could be introduced step by step nor a duodenal tube as she became very discomforted with increased diarrhea and vomiting. She only accepts vaminolact up to 60 ml/day by percutaneous endoscopic gastrostomy tube. She is thriving well. Her defecation is five times a day, half of the time watery, half-loose stool (Bristol scale types 5–7). She vomited regularly 0–3 times in the morning until the age of 7 years. She has urinary and anal continence. She is having frequent ear, nose, throat (ENT) infections. She attends regular school. She has had all her regular vaccinations. Fecal calprotectin concentrations were moderately elevated during her first years of life with 500–3000 mg/L (ref.<50 mg/L), suggesting chronic intestinal inflammation and are normal now.

Patient 2 is a 9-year-old boy, born to healthy unrelated German parents. Polyhydramnios was noted during pregnancy. He was born after 39 weeks of gestation weighing 3750 g (p79), length, 54 cm (p98) and head circumference, 36 cm (p89). Right-sided choanal atresia and left-sided cleft lip-palate, as well as an anal atresia and an enterocutaneous fistula, and abnormally positioned toes were noted at birth. He underwent surgery for his facial malformations, for recto-anal plastic reconstruction and removal of urethral valves. He has chronic diarrhea since the 12th day of life, with 10 stools per day (Bristol scale type 7) and requires TPN since then. He had had highly elevated fecal lactoferrin and calprotectin levels and bloating and benefits from cyclic oral antibiotic therapy. At the age of 9 years and 4 months, he is short with a length of 115.0 cm (<p1) and weighs 26.1 kg (p22). He receives 50% of fluids and 90% of calorie intake parenterally. In total, his Na+ intake amounts to 20 mmol/kg/day. Endoscopies revealed absence of the ileocecal valve, absence of villi in the terminal ileum and a short coecum, abnormally localizing to the left side. Of note, haustra appeared to be absent from the large bowel, and there were increased mucosal granularity and lymphofollicular hyperplasia suggesting chronic inflammation. Histology revealed signs of mild chronic inflammation in all parts of the bowel from the duodenum to the colon. He has had episodes of elevated transaminases without clinical signs of liver dysfunction and with normal liver synthesis results, as often observed with long-term TPN in young children. He experiences recurrent ENT infections and has never had keratitis. His psychomotor development is mildly delayed, which is considered due to his hospitalization and severe disease course. He attends a regular school and is 1 year late for his age. He had had all regular vaccinations. He had a normal karyotype (46,XY) and normal chromosomal microarray result. He had been tested negative for SKIV2L (Ski2-like RNA helicase gene) variants excluding trichohepatoenteric syndrome type 2 (OMIM #614602).

Table 1

Molecular and relevant clinical findings in SCSD patients

Parameter 1)	Patient 1	Patient 2	Patient 3
SPINT2 mutation (exon; protein alteration) allele 1	c.447G>A (5; p.Trp149*)	c.488A>G (5; p.Tyr163Cys)	c.481T>G (5; p.Phe161Val)
SPINT2 mutation (exon; protein alteration) allele 2	c.488A>G (5; p.Tyr163Cys)	c.488A>G (5; p.Tyr163Cys)	c.481T>G (5; p.Phe161Val)
Consanguinity	No	No	Yes
Ethnicity	Dutch	German	Arab
Current age (month)	7 years	9 years and 4 months	Died at age 21 months
Sex	Female	Male	Female
Gestational age at birth (weeks)	38	39	39
Birth weight (g)	2455	3750	2920
Present weight (kg/SD score weight for age)	21.5 (−1)	26.1 (−0.78)	8.35 (−1.9)
Present length (cm/SD score length for age)	113 (−2.5)	115.0 (−3.4)	75 (−2.3)
Polyhydramnios	No	Yes	No
Associated findings	Choanal atresia, uvula bifida, preauricular pits, single transverse palmar creases	Choanal atresia (right sided), ventrally displaced anus and enterocutaneous fistula, absence of the ileocecal valve, urethral valve, atrial septal defect, cleft lip-palate, toe abnormalities	Meckel’s diverticulum, gastroduodenal intussusception, keratits
Meconium	+	No data	No data
Bowel movements/d	5–10	10	Continuously
Histology	Villous atrophy, tufting enteropathy	Mild chronic inflammation in all parts of the bowel from the duodenum to the colon	Unspecific chronic inflammation in duodenum
PN	Since 3 weeks of age	Since 12th day of life	Since birth
% Lifetime on PN	100	100	100
Current treatment	Hardly tolerates enteral fluids/food	He receives 50% of fluids and 90% of calorie intake parenterally.	n.a.
Average Na intake (mmol/kg/d)	10	9	10
Complications	Chronic intestinal inflammation in first years of life Osteoporosis	Chronic intestinal inflammation Intestinal bacterial overgrowth	Gastrointestinal-transit disorder

Patient 3, a girl, is the fifth child born to healthy Arab parents who are first cousins. Four sibs are healthy. She was born at term by cesarean section due to lost cardiotocogram after an otherwise uneventful pregnancy. She showed good adaption postnatally (Apgar 9/10/10), pH 7.32, birth weight, 2880 g (P8), length, 49 cm (P10) and head circumference, 34 cm (P16). There were no overt congenital malformations. She did not tolerate any oral feeding after birth and presented with recurrent vomiting even when getting null per mouth. At the age of 5 weeks, she underwent surgery where a duodenal obstruction caused by herniation of part of an unusually mobile stomach into the duodenum, and a Meckel’s diverticulum were removed; in addition, the hypermobile gastrum was fixated to the ventral abdominal wall. Ileus recurred 1 week later, and surgery for removal of adhesions relating to the first surgery was performed. At that time her weight was 2920 g (−4.0 SD, p0.1) and her length was 54 cm (−1.4 SD, p7). Subsequently, chronic diarrhea developed, with 10–15 watery stools per day. She vomited several times a day, did not tolerate any oral feeding until she died at the age of 21 months due to multi-organ failure following dehydration due to acute gastrointestinal infection. She had been on stable TPN at home, receiving 10 mmol Na⁺/kg/day. At the age of 5 months, bilateral keratitis was noted and SCSD was suspected prompting SPINT2 mutation analysis. She had been growing parallel but below the third percentile and her psychomotor development was mildly delayed, which was considered as secondary to hospitalizations and the initially complicated course of the disease. At the age of 19 months, she was short with a length of 75 cm (−2.3 SD, p1), and weighed 8.35 kg (−1.9 SD, p3). She received 100% of fluids and caloric intake parenterally.

The most important clinical features and pathological laboratory values seen in the presented SCSD patients are compiled in Table 1. Table 2 shows a compilation of clinical symptoms in all 34 SCSD patients from 26 families with identified SPINT2 variants, reported to date.

Table 2

Clinical symptoms in 34 SCSD patients from 26 families with identified SPINT2 variants, reported in this study and in the literature

Clinical features	Present in
Premature birth	2/7
Polyhydramnios	3/13
Chronic diarrhea of congenital or infantile onset
Outcome
weaned from PN (two patients weaned of PN:	2/34
3-year-old girl, with genotype c.593-1G>A/c.593-1G>A;
18-year-old girl with genotype Tyr163Cys/?+)
requires partial or total PN	11/34
received small bowel transplantation	8/34
deceased, death from treatment-associated complications (line sepsis, acute liver failure, transplant issues)	14/34
Choanal atresia (uni- or bilateral, bony or membraneous) Associated with distinct SPINT2 genotypes; mostly, but not always associated with the most frequent genotype, Tyr163Cys/Tyr163Cys	20/34
Intestinal atresia (includes anal duodenal, jejunal, ileal, anal and multiple atresia)	6/34
Pilar dysplasia/brittle hair	9/34
Keratitis/corneal erosions (with mild to severe photophobia and chronic conjunctival discharge; spontaneous corneal perforation reported in one case)	26/34
Associated with distinct SPINT2 genotypes. Mostly, but not always Associated with the most frequent genotype, Tyr163Cys/Tyr163Cys
Cleft palate and cleft lip-palate	3/34
Hexadactyly, abnormally formed or positioned fingers or toes	4/34
Mild psychomotor delay considered secondary to hospitalization	2/34
Hypertelorism	5/34
Preauricular pits (uni- or bilateral)	3/34
Broad and high nasal root and long philtrum	7/34
Chronic ear infections	7/34
Single transverse palmar crease (uni- or bilateral)	2/34
Optic nerve coloboma	3/34

The following symptoms were reported in single patients: double kidney, aortic-valve hamartoma, intestinal malrotation, gastroduodenal intussusception, absence of the ileocecal valve, ureteral duplicity, olfactive bulb agenesis, fistula of the philtrum, macrocephaly and rectovaginal fistula.

SPINT2 variant detection in SCSD

Bi-allelic SPINT2 mutations were identified in all three patients in this series, in whom clinical presentations had suggested SCSD as the likely diagnosis (Table 1). Patient 1 was compound heterozygous for SPINT2 mutations c.447G>A (p.Trp149^*) and c.488A>G (p.Tyr163Cys). c.447G>A is a novel variant and predicts nonsense-mediated mRNA-decay and potentially represents a null mutation. Patient 2 is homozygous for SPINT2 mutation c.488A>G (p.Tyr163Cys). Patient 3 is homozygous for the novel SPINT2 mutation c.481T>G (p.Phe161Val). So far, 13 different SPINT2 variants have been identified in SCSD patients. All these variants were private or very rare with respect to public databases with ~70 000 exomes from individuals without severe childhood disabilities, and in silico analyses predicted that they were all compromising the function of the gene product (PHRED Combined Annotation Dependent Depletion scores; http://cadd.gs.washington.edu/home) (Table 3). Collectively, the nature of these variants, their low allele frequencies among control populations, the evolutionary conservation of the residues affected by missense changes and in silico predictions suggested they were disease causing. Of note, these 13 variants included 3 premature termination codons, 2 start codon removals and 3 canonical splice-site variants. Therefore, loss of function of the mutant proteins is indicated, in line with the observed recessive inheritance of SCSD. The c.488A>G (p.Tyr163Cys) variant represents the most common SCSD-associated variant and was present on 40/68 (59%) of disease alleles. This variant represents a founder mutation, as demonstrated by a shared haplotype around the c.488A>G variant in seven unrelated SCSD patients using high-density SNP array genotyping (data not shown). All four human HAI-2 missense variants affect highly conserved residues of KD2 but localize outside the canonical binding loop (Figs 1A and B and 2).

Table 3

Compilation of SPINT2 variants identified in patients with diagnoses of SCSD and (syndromic) tufting enteropathy

cDNA variant	Allele frequency Exome aggregation consortium ExAC	Geographic origin	Protein variant	Domain	CADD score	Reference
c.1A>T	0	France	Start codon removal	Signal peptide	22.2	(4)
c.2T>C	0	The Netherlands	Start codon removal	Signal peptide	23.5	(3)
c.166_167dupTA	0	USA	p. (Asn57Thrfs*24)	KD1	34	(8)
c.172dupG	0	Germany, Switzerland	p.Val58Glyfs*3a	KD1b	28.9	(4)
c.247G>T	0	Italy	p.Glu83*	KD1	36	(4)
c.337+2T>C	0.00001647 2/121406	Sweden	Splicing defect (predicted)	EC	24.3	(3)
c.442C>T	0	France	p.Arg148Cys	KD2	35	(4)
c.447G>A	0	The Netherlands	p.Trp149*	KD2	40	This report
c.481T>G	0	Turkey	p.Phe161Val	KD2	26.7	This report
c.488A>G	0.0003067 37/120652	Europe, Canada, USA	p.Tyr163Cys	KD2	25.9	(3)
c.502G>A	0	Algeria	p.Gly168Ser	KD2	29.6	(4)
c.553+2T>A	0	Sweden	Splicing defect (predicted)	Adjacent to TM	22.6	(3)
c.593-1G>A	0	Austria	Decreased expression, 24-aa inframe insertion	Adjacent to TM	24.5	(3)

^aPresent on one allele, with no variant identified on the second allele.

^bKD1 and KD2, Kunitz domains 1 and 2; EC, extracellular domain (KD1-KD2 linker domain); TM, transmembrane domain.

Figure 1

Human HAI-2 missense mutations affect highly conserved residues. (A) Conservation analysis of a variety of human KDs by multiple sequence alignment (ClustalW). Kunitz domain amino acids were aligned from; HAI-2 (O43291), APLP2 (Q06481), APP (P05067), TFPI-2 (P48307), TFPI-1 (P10646) and HAI-1 (O43278). Invariant amino acids are shown in a consensus line at the bottom. The Kunitz motif comprises a chain of around 60 amino acid residues stabilized by three disulphide bonds, the six conserved cysteine residues are each boxed in green. All three KD2 SCSD missense mutations affect invariantly conserved residues (marked in blue); the specificity determining P1 amino acid is marked red. (B) Amino acid multiple sequence alignment (ClustalOmega) of human HAI-2 KD2 (H. Sapiens, O43291:133-183), encoded by SPINT2, to a variety of Kunitz domain-containing proteins from different organisms. The aligned sequences were choosen amongst the 20 organisms with the lowest identity score (<60%) to HAI-2 KD2 from a UniProt BLAST. Organisms aligned; Merops nubicus (A0A091RA22), Acanthisitta chloris (A0A091MSB9), Astyanax mexicanus (W5LK57), Pelodiscus sinensis (K7GB12), Scleropages formosus (A0A0P7WW45), Hydatigera taeniaeformis (A0A0R3X6J5), Tyto alba (A0A093H0H8), Bambusicola thoracicus (A0A2P4TEU0), Meleagris gallopavo (G1MQW1), Gallus gallus (F1P0A7), Colinus virginianus (A0A226PMC9), Neovison vison (U6DW97), Chelonia mydas (M7AU62), Ailuropoda melanoleuca (G1L6K6), Propithecus coquereli (A0A2K6F518), Ictidomys tridecemlineatus (I3MER0), Melanochromis Auratus (A0A1U7R2P8), Chamaeleo calyptratus (M9T1N5), Pan troglodytes (A0A2J8N121) and Pongo abelii (A0A2J8WZI3).

Figure 2

HAI-2 SCSD-associated variants Phe161Val, Tyr163Cys and Gly168Ser stabilize matriptase and display unaffected inhibitory effect toward matriptase. (A) Each column represents the rate of turnover of the chromogenic substrate S-2288 without antibody (PBS, light grey column), with antibody aZ-mAb-6 inhibiting matriptase activity (aZ-mAb-6, black column) or with a control antibody (control, dark grey column) in extracts obtained by lysis of HEK293 cells transiently transfected as indicated. Each column represents the mean value for the rate of turnover from at least three technical replicas with standard deviations indicated. The experiment was repeated at least three times with essentially the same result. (B) Extracts from (A) were subjected to SDS-PAGE and WB using antibodies against matriptase (AF3946), HAI-2 (HPA) and GAPDH under boiled and reducing conditions. Molecular mass markers (kDa) are indicated to the left. The results are representative of three independent experiments.

Human SCSD-associated HAI-2 variants inhibit matriptase

Matriptase is a modular, ~95 kDa, protease consisting of a short cytoplasmic N-terminal peptide, a signal anchor that functions as a single-pass transmembrane domain (TM), a sea urchin sperm protein, enteropeptidase and agrin (SEA) domain (residues 86–201), two complement C1r/s urchin embryonic growth factor and bone morphogenetic protein-1 (CUB) domains (residues 214–334), four low-density lipoprotein receptor class A domains (residues 452–604) and a trypsin-like serine protease domain (SPD) (residues 614–855). The Gly149-Ser150 peptide bond of newly synthesized matriptase is hydrolysed in the secretory pathway, resulting in the SEA domain-cleaved form of matriptase, which subsequently can undergo cleavage after Arg614 in the SPD domain, generating the activated form. The Arg614-cleaved matriptase rapidly complexes with HAI-1 or HAI-2, whereby it becomes proteolytically silent (29). In order to investigate the inhibitory effects of the SCSD-associated HAI-2 variants Phe161Val, Tyr163Cys and Gly168Ser toward matriptase, we determine the peptidolytic activity by measuring the color development from degradation of a general substrate for serine proteases (S-2288) in vitro. Peptidolytic activity derived from matriptase was determined by comparing the rate of peptidolytic activity in the presence of a monoclonal antibody that works as a specific competitive inhibitor of matriptase activity (aZ-mAb-6) to the rate of peptidolytic activity in the absence of aZ-mAb-6 (30). The aZ-mAb-6 antibody inhibits both the zymogen form and the activated form of matriptase (30) and inhibits >90% of matriptase activity under the conditions used.

HEK293 cells were transiently transfected with constructs encoding matriptase alone or matriptase with either WT HAI-2 or a mutated form of HAI-2. The cells were lysed after 2 days and the peptidolytic activity was determined in the presence of anti-matriptase antibody aZ-mAb-6, a control antibody or phosphate buffered saline (PBS) (Fig. 2A). The laboratory-generated mutation not associated with SCSD, HAI-2 Cys47Phe/Arg48Leu, was included as a positive control, as we have previously shown that it has a reduced ability to inhibit matriptase, and WT HAI-2 was included as a negative control as it efficiently inhibits matriptase (23). Furthermore, the mutation HAI-2 Arg143Leu was included in order to investigate whether the binding loop of KD2 plays a role in the inhibition of matriptase. Extracts obtained by lysis of cells co-expressing matriptase and WT HAI-2 or a HAI-2 variant (Arg143Leu, Phe161Val, Tyr163Cys and Gly168Ser) all exhibited a low peptidolytic activity that was unaffected or close to unaffected by the presence of anti-matriptase aZ-mAb-6 antibody (Fig. 2A). In extracts of cells co-expressing matriptase and HAI-2 Cys47Phe/Arg48Leu a significant proteolytical activity was detected and this activity was significantly diminished by the presence of anti-matriptase aZ-mAb-6 antibody but not by the control antibody, signifying that unopposed matriptase activity is present. Extracts of cells expressing matriptase alone also depicted low levels of peptidolytic activity that was unaffected by the presence of aZ-mAb-6. We have previously shown that WT matriptase can only be detected in cellular lysates if co-expressed with HAI-1 or HAI-2 (31,32), explaining the low levels of matriptase activity present when matriptase is expressed alone. We previously reported that the HAI-2 mutant Cys47Phe/Arg48Leu affects the cleavage of matriptase SEA-domain by showing a single band of 95 kDa on western blots as opposed to the WT matriptase double band at 95 and 80 kDa (23) as seen when comparing lanes 2 and 3 in Figure 2B. Western blot analysis of the cell extracts showed that similar amounts of matriptase, HAI-2 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were present in all samples, except when matriptase is expressed alone (Fig. 2B, lane 1), since the matriptase protein is undetectable in the absence of HAI-2 (31,32). Our result suggests that the SCSD-associated HAI-2 variants Phe161Val, Tyr163Cys and Gly168Ser do not affect the ability of HAI-2 to inhibit matriptase.

HAI-2 SCSD-associated variants Phe161Val, Tyr163Cys and Gly168Ser display reduced ability to inhibit prostasin

We investigated whether the SCSD-associated variants Phe161Val, Tyr163Cys and Gly168Ser affect the ability of HAI-2 to inhibit the proteolytic activity of prostasin. No specific peptide-based substrates or inhibitors are available for prostasin. However, it was recently shown that matriptase and matriptase Ser805Ala are substrates for prostasin (33). WT matriptase can only be detected in cellular lysates if co-expressed with HAI-2 and/or HAI-1 (31) and is therefore unsuited as substrate in a biochemical assay. The catalytically inactive matriptase Ser805Ala mutant does not depend on co-expression with HAI-1 or HAI-2 to be stable (23) and was chosen as a substrate for prostasin. Prostasin catalyzed cleavage of matriptase Ser805Ala generates a 30 kDa matriptase fragment. HEK293 cells were transiently transfected with matriptase Ser805Ala, prostasin and HAI-2 WT or mutated HAI-2 in various combinations and after 2 days, cell extracts were obtained by lysis and analyzed by SDS-PAGE and western blotting (WB) (Fig. 3). No 30 kDa matriptase fragment was detected when matriptase Ser805Ala was expressed alone, whereas prostasin co-expression with matriptase Ser805Ala produced a prominent band of 30 kDa, confirming that prostasin is capable of cleaving matriptase Ser805Ala. It has previously been shown that HAI-2 is an inhibitor of prostasin (34). Co-expression of matriptase Ser805Ala, prostasin and WT HAI-2 displayed no 30 kDa fragment, confirming that WT HAI-2 is an inhibitor of prostasin (34). When co-expressed with HAI-2 mutated in the KD1, P2 and P1 position (HAI-2 Cys47Phe/Arg48Leu) a prominent 30 kDa fragment was detected suggesting that HAI-2 Cys47Phe/Arg48Leu is unable to inhibit prostasin-catalyzed cleavage. When co-expressed with HAI-2 mutated in the P1 position of KD2 (HAI-2 Arg143Leu) no 30 kDa fragment was observed, suggesting that the binding loop of the KD2 is unimportant for inhibition of prostasin-catalyzed cleavage. Co-expression with the SCSD-associated HAI-2 variants Phe161Val, Tyr163Cys and Gly168Ser all displayed a reduced ability to inhibit prostasin-catalyzed cleavage, as visualized by the 30 kDa band. This suggests that the SCSD-associated HAI-2 variants have reduced the ability to inhibit prostasin. However, the 30 kDa matriptase fragment is less prominent for the SCSD-associated HAI-2 variants, as compared with HAI-2 Cys47Phe/Arg48Leu, suggesting that the SCSD-associated HAI-2 variants have a reduced but not an abolished inhibitory ability toward prostasin. All samples were analyzed by SDS-PAGE and WB for the expression of HAI-2 and prostasin, as controls.

Figure 3

HAI-2 SCSD-associated variants Phe161Val, Tyr163Cys and Gly168Ser display reduced ability to inhibit prostasin-catalyzed cleavage. Extracts obtained by lysis of HEK293 cells transiently transfected with matriptase Ser805Ala (mat Ser805Ala) alone, matriptase Ser805Ala with prostasin (pro), matriptase Ser805Ala and prostasin together with HAI-2 WT or mutated HAI-2 variants (Cys47Phe/Arg48Leu, Arg143Leu, Phe161Val, Tyr163Cys or Gly168Ser), as indicated above the figure, were analyzed by SDS-PAGE and WB using antibodies against matriptase (AF3946) under boiled and reducing conditions. AF3946 recognizes the 30 kDa fragment resulting from prostasin-catalyzed cleavage of matriptase Ser805Ala as is indicated to the right. The HAI-2 Cys47Phe/Arg48Leu mutant, known to significantly reduce the inhibitory abilities of the HAI-2 KD1 against matriptase, and the HAI-2 Arg143Leu mutant located in the equivalent mutation site of the KD2, were included as controls. Molecular mass markers (kDa) are indicated to the left. The results are representative of at least three independent experiments.

Figure 4

Predicted structure of human HAI-2 KD2. The structural model of human HAI-2 KD2 was constructed using the protein structure prediction software I-TASSER. The residues affected by SCSD-associated missense variants are shown in blue and the P1 Arg143 in red. The three disulfide bonds stabilizing the Kunitz domain structure are shown in green.

Evidence for the presence of an exosite for prostasin on KD2 of HAI-2

We generated a structural model of human HAI-2 KD2 using the protein structure prediction software I-TASSER (35–37) (Fig. 4). The submitted sequence was residue 133–183 from GenBank accession no. O43291-1. The first model chosen by I-TASSER is presented using PyMol software. The model has a C-score of 1.20 (which typically ranges from −5–2) signifying a model with a high confidence.

The residues affected by SCSD-associated missense variants, Phe161, Tyr163 and Gly168 (shown in blue) are predicted by I-TASSER to have no or very limited solvent exposure, scoring 0, 1 and 2, respectively, on a scale from 0 (buried) to 9 (highly exposed). We therefore suggest that the SCSD-variants affect the folding of KD2 and thereby the interaction with prostasin at an exosite on KD2, since it is unlikely that they interact directly.

Discussion

Including the three patients presented here, 34 SCSD patients with SPINT2 mutations from 26 families have been reported in the literature (2–8,12). All these patients presented with intractable diarrhea, which most often had its onset within the first 2 weeks of life. Episodes of intestinal pseudo-obstruction can precede the onset of diarrhea for several weeks, and prompt intestinal surgery, as seen in patient 3 reported here. All but two reported patients had at least one additional clinical symptom of the typical spectrum, and the associations of a congenital diarrhea with either congenial choanal atresia or keratitis of infantile-onset appear to be pathognomonic for underlying SPINT2 variants. The most common symptoms observed in this disease, choanal atresia, keratitis, intestinal tufting and atresias suggest an epithelial defect that might be caused by abnormal activation of signal molecules or enzyme cascades in the secretory pathway due to unopposed activity of one or more proteases. Thirteen patients were reported with intractable diarrhea and SPINT2 variants, who all had typical histologic findings of epithelial tufting enteropathy (4,6), whereas such findings were not seen in 15 patients in another study (3). Diagnosing the typical histologic findings of epithelial tufting enteropathy (9) might be difficult and requires sufficient amounts of intestinal biopsy material.

The risk of death from acute dehydration, from PN-related infections and liver failure, as well as in the context of intestinal transplantation is high (14/34 patients). Even successful treatment with PN and successful intestinal transplantation requiring strict immunosuppression greatly affect the patients’ quality of life. However, when successfully managing the first years of life with PN the results can be rewarding: of six patients with SCSD and the same homozygous c.593-1G>A variant, five died before 13 months of age from treatment-related complications, whereas one patient, now 12 years of age, could be weaned from PN, has full continence, attends regular school and participates in all activities of life under oral sodium supplementation; this is a follow-up on our patient A2HE (1) reported in 2009 (3). Similarly, of 16 patients reported with the most common genotype, the homozygous p.Tyr163Cys mutation, 5 died early from therapy-related complications, and 3 patients were able to manage with partial PN at the age of 12 years.

These observations might also suggest a genotype–phenotype correlation, with the p.Tyr163Cys variant being less functional than the variant containing 24 additional amino acids inserted between the second Kunitz and the TM encoded by the c.593-1G>A variant (3). Similar to p.Tyr163Cys, other missense variants affecting conserved residues in the KD2 appear to confer a severe disease course: the single patient with the homozygous p.Phe161Val mutation died from acute dehydration, and one of two patients with a homozygous p.Gly168Ser variant requires TPN at age 12 years, his sister had been provided with a bowel transplantation (4). It is noteworthy, that so far no persons have been described who harbored two SPINT2 null alleles, a situation that might be associated with more severe developmental defects or embryonically lethal as seen in the mouse (38–40).

The most frequent genotype among SCSD patients is a homozygous p.Tyr163Cys mutation; this genotype is most often associated with both choanal atresia and corneal erosions, but the presence of only one of these symptoms or their absence has been observed.

Prostasin, matriptase, HAI-1 and HAI-2 constitute a system of proteases and protease inhibitors that regulates multiple aspects of development, postnatal homeostasis and tissue remodeling, although HAI-2 has other targets as well. Matriptase, prostasin, HAI-1 and HAI-2 are co-expressed in most developing and adult mammalian epithelia (29,32,33,39–44).

Individuals homozygous for one of the SCSD-causing HAI-2 missense mutations Phe161Val, Tyr163Cys or Gly168Ser, all display a similar phenotype, and we speculated that the three HAI-2 mutants cause SCSD by a similar mechanism. It has previously been shown in an oocyte system, that WT HAI-2 inhibits prostasin, whereas HAI-2 Tyr163Cys does not (28). We therefore investigated the ability of the three SCSD-associated HAI-2 mutants Phe161Val, Tyr163Cys or Gly168Ser to inhibit prostasin using the matriptase Ser805Ala mutant as a substrate for prostasin. Matriptase is one of several natural protein substrates for prostasin. Unlike WT matriptase, matriptase Ser805Ala does not depend on HAI-1 or HAI-2 for protein stability (23) and it has no enzymatic activity, as the serine residue in the catalytic triad is mutated (23). Our findings suggest that WT HAI-2 efficiently inhibited prostasin-catalyzed cleavage, as did HAI-2 Arg143Leu, whereas HAI-2 Cys47Phe/Arg48Leu and the three SCSD-causing HAI-2 mutants appeared to be less efficient inhibitors of prostasin-catalyzed cleavage (Fig. 3).

Our data suggests that the binding loop of HAI-2 KD1 (including Cys47/Arg48) is involved in binding to the active site of prostasin, whereas the binding loop of the HAI-2 KD2 (Arg143) seems to have no importance in the inhibition of prostasin. The three SCSD-causing HAI-2 mutations are all located in KD2 outside of the binding loop, and they all display diminished ability to inhibit prostasin, suggesting that HAI-2, when inhibiting prostasin, interacts with the binding loop in the KD1 of HAI-2 (Cys47/Arg48) along with a surface exposed exosite in KD2 of HAI-2 located away from the binding loop. We find it unlikely that the SCSD-causing mutations induce misfolding of the entire HAI-2 molecule since they are fully able to inhibit matriptase. Homology modeling of HAI-2 KD2 using I-TASSER showed that the residues Phe161, Tyr163 and Gly168 that are mutated in SCSD, have no or limited solvent exposure (Fig. 4), suggesting that these amino acids do not directly form part of the exosite, but induce misfolding of KD2 disturbing the exosite. Analysis of nine single amino acids mutations of solvent exposed amino acids on KD2 of HAI-2, did not identify any positions that affect the ability of HAI-2 to inhibit prostasin proteolytic activity (data not shown). We have previously shown that both prostasin and HAI-2 are located in the secretory pathway and on the apical plasma membrane where interaction could take place (29,31). It is difficult to assess the physiological importance of prostasin enzymatic activity, since critical in vivo functions of prostasin can be performed by proteolytically-inactive or zymogen-locked variants (Prss8^R44Q) of the protease and are compatible with normal or near-normal murine development (42,44,45). HAI-2 seems to be a physiologically important inhibitor for prostasin, since prostasin-HAI-2 complexes could be purified from human enterocytes, whereas prostasin-HAI-1 complexes were not detected (34), even though both HAI-1 and HAI-2 are able to inhibit prostasin in vitro (17). Thus, although HAI-1 is also expressed in epithelial cells of the intestinal tract, it is unable to compensate for mutated HAI-2.

In conclusion, our data suggests that SCSD patients have unopposed prostasin activity located in the secretory pathway and/or on the (apical) plasma membrane. However, it remains unclear whether the unopposed prostasin activity is causing SCSD, since HAI-2 is likely to have many targets.

The identity of the protease or proteases responsible for the intestinal abnormalities observed in SCSD patients has thus not been unequivocally determined. Contrasting with our results, it was recently suggested that the SCSD-causing HAI-2 mutations affect the ability of HAI-2 to inhibit matriptase, resulting in increased matriptase-catalyzed EPCAM degradation and affecting epithelial tightness through the level of claudin-7. This suggestion was based on lower EPCAM levels in CACO-2 cells stably expressing HAI-2 Tyr163Cys as compared with cells stably expressing HAI-2 WT siRNA, both with down-regulation of endogenous HAI-2 through siRNA treatment (14). However, this experiment is not easy to interpret, since serine proteases have overlapping substrate specificities and HAI-2 might target multiple proteases. An alternative interpretation of this experiment may be that prostasin, endogenously expressed in Caco-2 cells, normally inhibited by WT HAI-2, is now only partly inhibited by HAI-2 Tyr163Cys and therefore degrades EPCAM. As the experiment does not include a reagent that specifically targets matriptase, EPCAM may be degraded by any protease targeted by HAI-2. Matriptase is one of many serine proteases with no known specific peptide substrate or inhibitor and it has therefore been difficult to determine its proteolytic activity in complex biological samples. In the present study, we employ a newly described monoclonal antibody, aZ-mAb-6, that has been shown to be a potent fast-acting specific competitive inhibitor of human matriptase. aZ-mAb-6 blocks the activity of both zymogen and activated human matriptase (30) used in this study in combination with a broad-spectrum serine protease substrate (S-2288), to determine matriptase activity. We have previously shown that WT HAI-2 efficiently inhibits matriptase and that this inhibition is compromised by mutations in the HAI-2 KD1, including mutations in the binding loop of KD1 (HAI-2 Cys47Phe/Arg48Leu) (23). However, introducing a corresponding mutation in the binding loop of HAI-2 KD2 (HAI-2 Arg143Leu) did not have any effects on HAI-2s inhibitory abilities toward matriptase (Fig. 2). Likewise, we found that the three SCSD-causing HAI-2 mutations Phe161Val, Tyr163Cys or Gly168Ser, all located in KD2 outside of the binding loop, had no effect on the ability of HAI-2 to inhibit matriptase (Fig. 2). In a recent study it was shown that both recombinantly expressed KD1 and recombinantly expressed KD2 interact with and inhibit matriptase, and that KD1 is a more potent inhibitor of matriptase than KD2 (46). A model of KD2 based on structure alignment/comparison, showed that the position of the SCSD-associated mutations in HAI-2 have limited solvent exposure and therefore, are likely to induce misfolding of KD2. We have earlier shown that HAI-2 Tyr163Cys has maintained its ability to support stabilization and SEA-domain cleavage of co-expressed matriptase (23), suggesting that at least part of the molecule, most likely KD1, is correctly folded. The SCSD-associated mutations in KD2 thus may exhibit WT inhibitory ability because the mutations do not cause misfolding of KD1, which consequently remains fully functional and inhibits matriptase despite a misfolded KD2.

Figure 5

Proposed schematic structure of inhibitory complexes between prostasin or matriptase and HAI-2 WT or SCSD-mutated HAI-2. A schematic overview of suggested HAI-2 inhibitory complexes between (A) prostasin (green) and WT HAI-2, (B) prostasin and SCSD-associated HAI-2 mutants, (C) matriptase (red) and WT HAI-2 and (D) matriptase and SCSD-associated HAI-2 mutants. Prostasin binds KD1 of HAI-2 along with a yet unknown exosite located on or near KD2 affected by the three SCSD-associated HAI-2 mutations (Phe161Val, Tyr163Cys and Gly168Ser, marked in blue). Matriptase only needs to bind KD1 of HAI-2 and inhibition is thus unaffected by SCSD-associated HAI-2 mutations. The inhibitory complexes may form early in the secretory pathway, as matriptase and HAI-2 KD1 mutants have been observed co-localizing with an endoplasmic reticulum (ER) marker (23), and since HAI-2 mainly resides in the ER (31). Each circle represents an amino acid, and the P1 residue of each KD is marked in red. (E) A schematic overview of HAI-2 and the HAI-2 KD sequences marking the P1 residues (red) and the SCSD-causing mutation sites (blue).

Our results thus suggest that HAI-2 interacts with prostasin and matriptase in different ways. When HAI-2 interacts with prostasin, both the inhibitory loop of KD1 and an unknown exosite on KD2 are involved (Fig. 5A). We propose, that the SCSD-associated HAI-2 mutations induce misfolding of the HAI-2 KD2, compromising the inhibitory complex, leading to a less efficient inhibition of prostasin (Fig. 5B). When HAI-2 inhibits matriptase KD1 interacts with the inhibitory loop of matriptase (Fig. 5C), this inhibition is not affected by mutations in KD2 of matriptase (Fig. 5D). Note that matriptase can also be inhibited by WT HAI-2 KD2 albeit less efficiently than by HAI-2 KD1.

In conclusion, our results suggest that the SCSD-associated mutations in HAI-2 may cause the disease due to reduced ability to inhibit prostasin or a prostasin-like protease.

Materials and Methods

Patients

Three unrelated patients with watery diarrhea of congenital or infantile onset were included in this study (Table 1). SCSD was suspected in these patients based on the presence of at least one clinical finding present in addition to chronic diarrhea: hyponatremia, intestinal atresia, choanal atresia, keratitis or skeletal abnormalities of hands or feet.

SPINT2 mutation detection

Written informed consent for genetic testing was obtained from the patients’ parents. Genomic DNA was prepared from peripheral leukocytes using standard protocols. Targeted PCR amplification and direct sequencing was performed for the coding region (7 exons) and all exon–intron boundaries of the SPINT2 gene (serine peptidase inhibitor, Kunitz type, 2; NCBI Ref. seq. NM_021102.3) in genomic DNA from the patients. Parental samples were tested for the presence of mutations detected in their children to confirm bi-allelic inheritance of mutations in the index patients. PCR primers and conditions are available from the authors upon request.

Cell culturing, transfections and DNA constructs

The human embryonic kidney cell line HEK293 was grown in Dulbecco’s Modified Eagle’s Medium supplemented with 2 mm L-glutamine, 10% fetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C in an atmosphere of 5% CO₂. For experiments, cells were seeded 1 day prior to transfection into either 6- or 12-well Corning® Costar® cell culture plates (cat. no. CLS3512 or CLS3516, Sigma-Aldrich, Copenhagen, Denmark) and grown to 80% confluence. For transient expression, adherent HEK293 cells were transfected using Lipofectamine™ 2000 (Invitrogen, Hvidovre, Denmark), according to the protocol supplied by the manufacturer. For co-transfections, the same overall quantity of plasmid was used. The cDNA encoding full-length human WT matriptase and HAI-2 were incorporated into pcDNA3.1 plasmid vectors. The HAI-2 cDNA used in the present study contains a naturally occurring SNP resulting in the amino acid substitution Val200Leu (rs11548457, allele frequency of 2.8% in 61.000 individuals). Full-length human prostasin cDNA was incorporated into the pIRES-EGFP vector (32). Empty expression plasmids were used for mock transfections. Mutations in the cDNA encoding matriptase and HAI-2 were introduced by site-directed mutagenesis using the GeneArt Site-Directed Mutagenesis System (cat. no. A13282, Life Technologies, Roskilde, Denmark) according to the manufacturer’s recommendations and verified by sequencing. The mutations matriptase Ser805Ala, HAI-2 Cys47Phe/Arg48Leu (23), HAI-2 Arg143Leu (31), HAI-2 Tyr163Cys (3) and HAI-2 Gly168Ser (4) have previously been described. Cell extracts were obtained by lysis in PBS, containing 1% Triton X-100, 0.5% deoxycholate (lysis buffer) 48 h post-transfection, spun for 20 min at 20.000 × g and the supernatant was stored at −20°C.

SDS-PAGE and WB

Samples were prepared by addition of 2 × SDS sample buffer (1:1) and for reducing conditions 0.2 m dithiothreitol was added. Proteins were separated on either 7 or 10% SDS polyacrylamide gels and transferred to Immobilon-P PVDF membranes (Merck, Søborg, Denmark). The membranes were blocked with 10% non-fat dry milk in PBS containing 0.1% Tween-20 (MPBS-T) for 1 h. The individual PVDF membranes were probed with primary antibodies diluted in 1% MPBS-T at 4°C overnight, followed by 3 × 5 min wash in PBS-T and 1 h incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies diluted in 1% MPBS-T. After 3 × 5 min wash in PBS-T the signal was developed using ECL® (enhanced chemiluminescence) reagent Super Signal West Femto Maximum Sensitivity Substrate (cat. no. 34095, Thermo Fischer Scientific, Hvidovre, Denmark) or Pierce ECL Western Blotting Substrate (cat. no. 32106, Thermo Fischer Scientific, Hvidovre, Denmark) according to the protocol supplied by the manufacturer and visualized with a Fuji LAS-1000 camera and Intelligent DarkBox II (FujiFilm Sweden AB, Malmö, Sweden), using the program LAS1000 Lite v1.5.

Antibodies

All samples for WB were boiled under reducing conditions and blots were probed with primary polyclonal sheep anti-human matriptase antibody (cat. no. AF3946, R&D systems, Abington, United Kingdom) (1:1000), polyclonal rabbit anti-human HAI-2 antibody (cat. no. HPA011101, Sigma) (1:1000), monoclonal mouse anti-human prostasin (cat. no. 612173, BD Transduction Laboratories™, Albertslund, Denmark) recognizing prostasin and with GAPDH antibody (cat. no. ZG003, Thermo Fischer Scientific, Hvidovre, Denmark) (1:2000) as control. Blots were probed with secondary HRP-conjugated antibodies (1:5000), Donkey anti-sheep IgG-HRP antibody (cat. no. HAF016, Thermo Fischer Scientific, Hvidovre, Denmark) Goat anti-rabbit IgG antibody (cat. no. AP307P, Merck, Søborg, Denmark) or with Goat anti-mouse (cat. no. T30953, Invitrogen). The monoclonal aZ-mAb-6 (30) was used to inhibit matriptase activity.

Peptidolytic activity assay

Extracts obtained by lysis of transiently transfected HEK293 cells, prepared as previously described, were used for an activity assay using chromogenic substrate H-d-Isoleucyl-l-prolyl-l-arginine-p-nitroaniline (cat. no. S-2288, Chromogenix, Frederiksberg, Denmark), substrate for a broad spectrum of serine proteases, in combination with antibody aZ-mAb-6 inhibiting matriptase activity or a control antibody (30). All samples were diluted to ensure that the maximal signal activity did not exceed the signal obtained for samples containing 100 pM matriptase SPD (47). Under these conditions the aZ-mAb-6 antibody (at a final concentration of 10 × K_i) is able to inhibit >90% of the activity. Cell extracts were combined with either PBS, control antibody or aZ-mAb-6 antibody to a final concentration of 200 nm (corresponding to 10 × Ki of the aZ-mAb-6 antibody according to (30) in a similar volume and the samples were diluted in 20 mm HEPES pH 7.4, 140 mm NaCl supplemented with 0.1% BSA (Sigma-Aldrich, Copenhagen, Denmark) (HBS buffer) to a final volume of 72,75 μl in a Corning 96-well half-area plate and heated to 37°C for 1 h. Then 300 μm chromogenic substrate S-2288 was added and color development was measured at 405 nm every 5 min for 5 h in a standard plate reader (BioTek Synergy HT) at 37°C. The rate of substrate turnover was determined from color development resulting from the pseudo-first-order reaction due to a substrate concentration far greater than the expected pM range of protease content. All measurements were adjusted for the optical path length and a background sample prepared with lysis buffer and reaction velocities calculated and indicated in mAU/min.

Conflict of Interest statement. None declared.

Funding

Jubiläumsfonds der Österreichischen Nationalbank [grant no. 16678] and Faculty of Health and Medical Sciences Foundation for Employees and Students at the University of Copenhagen [A5296].

Author notes

The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint first authors and the last two authors as joint last authors.