The Channel-Activating Protease CAP1/Prss8 Is Required for Placental Labyrinth Maturation

The serine protease CAP1/Prss8 is crucial for skin barrier function, lung alveolar fluid clearance and has been unveiled as diagnostic marker for specific cancer types. Here, we show that a constitutive knockout of CAP1/Prss8 leads to embryonic lethality. These embryos presented no specific defects, but it is during this period, and in particular at E13.5, that wildtype placentas show an increased expression of CAP1/Prss8, thus suggesting a placental defect in the knockout situation. The placentas of knockout embryos exhibited significantly reduced vascular development and incomplete cellular maturation. In contrary, epiblast-specific deletion of CAP1/Prss8 allowed development until birth. These CAP1/Prss8-deficient newborns presented abnormal epidermis, and died soon after birth due to impaired skin function. We thus conclude that a late placental insufficiency might be the primary cause of embryonic lethality in CAP1/Prss8 knockouts. This study highlights a novel and crucial role for CAP1/Prss8 in placental development and function.


Introduction
Proteases or peptidases are enzymes able to catalyze the hydrolysis of peptide bonds representing about 2% of all proteins in human [1] or mouse [2]. Proteases are often involved in signaling cascades, where a zymogen is cleaved and converted into its active form, allowing amplification but also fine-tune regulation of a signal. Moreover, endogenous protease inhibitors are present within a cell, giving a high level of complexity for these proteins in vivo [3]. They are involved in many processes, from the classical digestive function, to cellular signaling and specific protein cleavage. Caspases are important in apoptosis, matrix metalloproteinases (MMPs) are involved in tissue remodeling, thrombin plays a key role during blood clotting, while deubiquitinating enzymes (DUBs) are required for sub-cellular localization of proteins. Serine proteases are implicated in cell growth, survival, invasion, angiogenesis and inflammation and appear to be involved in human disease, including cancer. Therefore, there is a growing interest in developing therapies to control the activity and expression of proteases [4], e.g., inhibitors of the angiotensinconverting enzyme, in the treatment of hypertension.
Mouse channel activating protease 1, also termed CAP1/Prss8 for protease serine S1 family member 8, has been named for its ability to increase ENaC-mediated sodium current in vitro [5,6] and in vivo [7,8]. According to the peptidase classification [2], it belongs to the subfamily A of the S1 chymotrypsin family as trypsin and chymotrypsin. The S1 family is characterized by a catalytic triad, histidin, aspartate and serine (HDS), that are gathered to form the catalytic site by folding of the protein [9]. CAP1/Prss8 belongs to a subgroup of S1 serine proteases that are anchored to plasma membranes via a glycosyl-phophatidylinositol linkage (GPI-anchored) located at the lipid rafts that are important for cell-cell communication [10][11][12]. Human CAP1/Prss8 has been previously isolated from seminal fluids of prostate cancer patients and named prostasin [13]. First described as a cancer marker, it has later been linked to reduced tumor growth as seen by reduced invasiveness of breast [14], gastric [15] and prostate cancer [16,17].
Using mice carrying a conditional knockout allele of CAP1/ Prss8 [28], we have previously demonstrated that CAP1/Prss8 is required for normal skin barrier function resulting in postnatal death due to severe dehydration when CAP1/Prss8 is deleted in skin [29]. In addition, increased expression of CAP1/Prss8 induces severe skin defects [30]. Moreover, CAP1/Prss8 is involved in water and salt balance in lung alveolar type II cells [7]. Furthermore, we have recently described two spontaneous animal models carrying mutations in the CAP1/Prss8 gene, presenting skin defects and constitutively reduced basal activity of ENaC [8].
In the present study, we unveil a novel and important role of CAP1/Prss8 during embryonic development. Our data clearly illustrate that proper expression of CAP1/Prss8 in the placenta triggers late placental labyrinth maturation, thereby maintaining late gestation embryonic development and allowing the embryo to develop to term.

A Constitutive Knockout of CAP1/Prss8 Results in Embryonic Lethality
Serine proteases are implicated in many physiological and pathological processes. In order to uncover new roles of the membrane-bound serine protease CAP1/Prss8, we generated constitutive knockouts of CAP1/Prss8 (D/D) by intercrossing heterozygous mutant mice (D/+). CAP1/Prss8 knockout mice were missing at weaning (Fig. 1A), and at birth (data not shown), suggesting embryonic lethality. To identify the cause we dissected embryos from E8.5 to E16.5 and determined the representation of each genotype (Fig.1B). CAP1/Prss8 knockout mice were found underrepresented, and none of the knockout embryos was found alive from E14.5 onwards as determined by lack of heart beating and visible pulsatile blood vessels. Examination of the knockout embryos did not reveal any obvious developmental defects, like neural tube closure, or heart malformations (Fig. 1C). Most often, only the placenta and/or the degraded dead CAP1/Prss8 knoxkout embryo were found (Fig. 1C).

Late Differentiation Defect is Detected in CAP1/Prss8 Knockout Placentas
While wildtypes and heterozygotes showed an organized arborization of the vessels starting from the umbilical cord, placentas of knockout embryos exhibited only a couple of large vessels and marginal ones ( Fig. 2A). Histopathological analysis of E12.5 to E14.5 embryos revealed that knockout placentas are less organized with less marked placental cotyledons, and fewer blood vessels at E14.5. At E12.5 and E14.5, the labyrinth size did not differ amongst the experimental and control groups (E12.5; wt: 8756120 pixel; het: 8906267 pixel; ko: 7326149 pixel versus E14.5; wt: 9456259 pixel; het: 1346665 and ko: 8876275 pixel) (see also Fig. 2B). At the cellular level, a regular transition of the placenta histology can be observed in wildtypes. At E12.5, the placenta is mainly composed of cuboidal cytotrophoblasts that are fusing to form a syncytium from E10.5 to E16.5. This results in a thinner separation between maternal blood (small anucleate erythrocytes) and the fetal blood (large nucleated erythrocytes). In addition, the surface of exchange progressively increases. This process is observed in heterozygotes as well. In contrast, knockout placentas showed a majority of cuboidal cytotrophoblasts at E12.5 and this was maintained at all later time points, indicating incapability to form a proper syncytium (Fig. 2C).
All together, these data highly suggest an altered development of the placenta of knockout embryos from E12.5 onwards.

Increased Transient CAP1/Prss8 Expression is Required for Normal Placenta Maturation
To evaluate the importance of CAP1/Prss8 in the placenta, we followed its expression level by quantitative RT-PCR. In wildtype, CAP1/Prss8 mRNA transcript expression in the placenta increases from E11.5 onwards, reaching a peak around E13.5 before declining to basal levels ( Fig. 3A). No CAP1/Prss8 expression was detected in placenta from knockout embryos, while heterozygotes showed an intermediate expression level (Fig. 3A). In the embryo itself, the expression level of Prss8 was increasing 2-fold between E12.5 and 14.5 but overall remained 10-to 100-fold lower than in the placenta. At time points considered in this study, no transient peak of Prss8 expression was detected in the knockout embryo (data not shown; see also  To study further the consequences of CAP1/Prss8-deficiency, we analyzed the mRNA transcript expression of the markers pRb, Gcm1, EOMES and PAR2. pRb, EOMES and Gcm1 are transcription factors involved in placental differentiation and/or development. Absence of retinoblastoma protein (pRb) leads to a placental defect and embryonic death at E14.5 [29]. In our study, pRb expression in wildtype placentas increased from E11.5 to E14.5, and this was not affected by the absence of CAP1/Prss8 (Fig. 3C). The T-box transcription factor Eomesodermin (Eomes) is involved in the formation of the extra-embryonic trophoblast lineage [30]. Eomes is necessary for gastrulation and early events of placenta formation. We found that Eomes expression in the placenta decreased with time showing no difference between wildtype and knockout (Fig. 3C). The Protease-Activated Receptor-2, PAR2, is a G-coupled receptor found to be activated by CAP1/Prss8 in mouse embryo and in the skin [30]. Our results show that PAR2 is equally expressed in the placenta of both wildtype and knockout embryos (Fig. 3C). The knockout of the glial cell missing 1 (Gcm1) transcription factor leads to a failure of branchomorphogenesis and labyrinth formation from E8.5 onwards. Interestingly, Gcm1 promotes the fusion of the cytotrophoblast and syncytiotrophoblast [31]. In wildtype placenta, Gcm1 mRNA expression increased till E13.5, and did not significantly differ from the knockout placenta (Fig. 3C).
To assess the consequences of the lack of CAP1/Prss8 in the embryos during the perinatal period, two litters were monitored until birth. Conditional knockouts were born according to the expected Mendelian distribution (Fig. 5A). Histological observation of the skin of the CAP1/Prss8 lox/D ;Sox2::Cre Tg/+ pups right after birth showed hyperkeratosis (Fig. 5C). This is a characteristic phenotype of skin-specific CAP1/Prss8 knockout animals as has been previously described and leads to early postnatal death due to severe dehydration [29]. Four more litters from the same breeding (CAP1/Prss8 D/+ ; Sox2::Cre Tg/+ x CAP1/Prss8 lox/lox ) were kept until weaning, and no knockout (CAP1/Prss8 lox/D ; Sox2:: Cre Tg/+ ) mice could be detected at that time point (Fig. 5B).
We concluded that the lack of CAP1/Prss8 in the embryo is compatible with normal placenta function and embryo development demonstrating that the embryonic death of constitutive CAP1/Prss8 knockout animals is primarily caused by placenta failure.

Discussion
Placenta and embryo are highly dependent on each other and form a tightly regulated system with a placenta progressively adapting to the increasing requirements of the embryo. To determine the consequences of CAP1/Prss8 deficiency during embryonic development, we analyzed in detail a complete CAP1/ Prss8 knockout mouse model, and found that the absence of the protein leads to embryonic lethality. The present study shows a novel important role of the membrane-bound serine protease CAP1/Prss8 in placental development during late gestation.
Looking at different time points during embryonic development, we found that 50% of the knockout embryos were dead at day E12.5 or E13.5, and 100% from day E14.5 onwards (Fig. 1). We therefore concluded that CAP1/Prss8 deletion resulted in increased embryonic lethality. No obvious developmental delay and/or defect was observed in E12.5 embryos. Cell proliferation (Ki67) and TUNEL staining, and vascularization of the placenta, as visualized by the staining for the endothelial cell marker CD31, was not different from the wildtype (data not shown), thus rather suggesting a functional defect. From that time point onwards, we found the knockout embryos at varying developmental stages, and they often presented signs of advanced post-mortem degradation as illustrated in Fig. 1C, and thus were not informative for the cause of lethality (Fig.1).
The placenta from CAP1/Prss8 knockout embryos appears pale and lacks the arborescence of vessels as observed in wildtype embryos at E14.5 (Fig. 2). From day E12.5 to E16.5, wildtype placentas undergo major modifications. The labyrinth of the placenta allows the maternal and embryonic gas and nutrient exchange [31]. Trophoblastic cells develop from cuboidal and proliferative cytotrophoblasts to thin and highly differentiated syncytiotrophoblast cells, thereby increasing the exchange surface and capacity (Fig. 2). The placenta from CAP1/Prss8 knockouts apparently failed to undergo this maturation step, and histological observation indicated that cytotrophoblasts are still mainly present and are not replaced by syncytiotrophoblasts, preventing the development of blood vessels. The development of knockout placentas appeared to be modified from E12.5 onwards. We cannot exclude, however, that functional impairment may precede histopathological changes in the knockout placentas.
To evaluate the importance of CAP1/Prss8 expression in the placenta, we measured mRNA levels by quantitative RT-PCR. We first noticed that placental CAP1/Prss8 expression in wildtypes peaks at E13.5, at a time point where only less than 50% of knockout embryos survived (Fig. 3). It is worth noting that the maternal placenta shows only neglectable expression of CAP1/ Prss8 (data not shown). The peak of CAP1/Prss8 expression (at E13.5) parallels the Gcm1 expression peak in wildtype. Gcm1 is a transcription factor that promotes the late phase of differentiation of the trophoblast lineage [32] by regulating the activity of the placental-specific enhancer of the aromatase gene (CYP19) [33]. Interestingly, it was not significantly different between the experimental and control group (Fig. 3C), and might thus not represent the primary cause of lethality. The expression level of Eomes that is involved in the development of the trophoblast lineage did not differ in the knockout and control placenta (Fig. 3C). Moreover, the mRNA transcript expression of the protease-activated receptor PAR2, recently identified as down-  stream target of CAP1/Prss8 [30], was not altered at that developmental stage. Hepatocyte growth factor activator inhibitor (HAI-1) can form inhibitor complexes with several trypsin-like serine proteases and is required for mouse placental development and embryo survival [34]. HAI-1 is coexpressed with matriptase and CAP1/Prss8 in chorionic trophoblasts, and ablation of HAI-1 led to a complete loss of undifferentiated chorionic trophoblasts after E9.5 [34]. In contrary, genetic ablation of matriptase activity in HAI-1-deficient embryos restored the integrity of chorionic trophoblasts and development to term [34]. Till E12.5, we did not detect any alterations in trophoblast development in our CAP1/ Prss8-deficient embryos, suggesting that their functions may overlap, but they do not seem to be part of the same serine protease cascade.
A known target of CAP1/Prss8 is the epithelial sodium channel ENaC. Whereas a recent study suggested a role of ENaC in decidual cells for embryo implantation, mice lacking ENaC activity seem not affected during embryonic development and are born according to the expected Mendelian distribution [35,36]. Thus, we consider a major implication of ENaC in placental development as rather unlikely.
The close relationship between the embryo and placenta still leaves the question of the primary cause, leading to the death of both the placenta and the embryo. We here hypothesize that the placenta fails to develop which results in embryonic death, but on the other hand, when the embryo dies, the placental development may stop likewise. To test this hypothesis, we deleted CAP1/Prss8 exclusively in the epiblast, while placental CAP1/Prss8 expression was maintained. At birth, all expected genotypes were present following the Mendelian distribution (Fig. 5A), while no constitutive knockout developed to term (Fig. 5). mRNA analyses showed that we indeed successfully abolished CAP1/Prss8 expression in the embryo when it was maintained in placenta (Fig. 4B). The results presented in the paper thus clearly show that CAP1 is required for prenatal development, although not within the embryo itself. Pups lacking CAP1 only in the embryo (but not in the placenta) were even born according to a Mendelian distribution (Fig. 5). This clearly shows that the lack of CAP1/ Prss8 during embryonic development is without any consequence for the embryo. Nevertheless, those pups were born with an hyperkeratosis as seen in newborns lacking CAP1/Prss8 expression in the epidermis [29] (Fig. 5), thus causing death shortly after birth. None of these pups reached the weaning age (Fig. 5), and the phenotype thus recapitulates the features of an epidermalspecific CAP/Prss8 knockout.
In conclusion, our study shows that constitutive knockout of CAP1/Prss8 leads to embryonic lethality due to placental failure. In addition, CAP1/Prss8 expression within the embryo seems not required for its development. CAP1/Prss8 is detected in primate [37] and in human placenta in villi at the level of the trophoblasts and placental cell lines [38]. Our data strongly suggest that balanced placental expression of human CAP1/PRSS8 might be relevant in the context of placental maturation.

Ethics Statement
All animal work was conducted according to Swiss national guidelines. All mice were kept in the animal facility under UNIL animal care regulations. They were housed in individual cages at 2361uC with a 12-h light/dark cycle. All animals were supplied with food and water ad libitum. This study has been reviewed and approved (authorization no. 1793.1 to EH) by the ''Service de la consommation et des affaires vétérinaires'' of the canton of Vaud, Switzerland.

Extraction of Embryonic and Perinatal Tissues
Breedings were started in the evenings, and females were checked for vaginal plugs the next morning, with the morning of plug defined as E0.5. Embryos were recovered at indicated time points, and embryos and placentas were dissected individually. Visceral yolk sacs of each embryo were washed in PBS and processed for genomic DNA extraction and genotyping. Placenta and embryos were separated, rinsed in PBS, and either fixed in formalin over night prior to paraffin embedding, or snap-frozen in liquid nitrogen prior to protein/RNA extractions. Embryos were counted as alive when a beating heart was observed (from E10.5 onwards). Additionally, from E12.5 onwards, viability of embryos was assessed by color and visible pulsative vascularization. Histology 2-3 mm sections were obtained from paraffin blocks and stained with H&E. Labyrinth thickness was measured using ImageJ. Quantitative RT-PCR Embryonic and placental total RNAs were extracted with Qiazol (Qiagen, Basel, Switzerland). The reverse transcription reaction was performed with 1 mg of RNA mixed in a final volume of 20 ml with oligo dT (20) primer (5 mM), dNTP (500 mM), buffer (50 mM TrisHCl pH 8.3, 75 mM KCl, 3 mM MgCl2; 5 mM DTT), and 40 U of SuperscriptII (Invitrogen, Basel, Switzerland). The reaction was incubated at 42uC for 1 h and finally stopped by denaturing the enzyme for 10 min at 70uC. The cDNA was diluted five times before measurements. Quantitative real-time PCRs were performed by TaqMan PCR using the Applied Biosystems 7500 (primers and probes used detailed in table 1). The universal TaqMan mix (2x) and SybR green mix (2x) were purchased and used according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). The quantifications were calculated by measuring the DDCt normalized to ß-actin level. All measurements were done in duplicate. All values are expressed in relative expression (in % of the actin expression level). Primers were used as listed in Table 1.

Calculation and Statistics
All data are expressed as means 6 SEM. Individual groups were compared by using the t test for unpaired comparisons. The Mendelian distributions were analysed by chi-square tests. A level of P,0.05 was accepted as statistically significant for all comparisons. *P,0.05, **P,0.01, ***P,0.001.