A time frame permissive for Protein Kinase D2 activity to direct angiogenesis in mouse embryonic stem cells

The protein kinase D isoenzymes PKD1/2/3 are prominent downstream targets of PKCs (Protein Kinase Cs) and phospholipase D in various biological systems. Recently, we identified PKD isoforms as novel mediators of tumour cell-endothelial cell communication, tumour cell motility and metastasis. Although PKD isoforms have been implicated in physiological/tumour angiogenesis, a role of PKDs during embryonic development, vasculogenesis and angiogenesis still remains elusive. We investigated the role of PKDs in germ layer segregation and subsequent vasculogenesis and angiogenesis using mouse embryonic stem cells (ESCs). We show that mouse ESCs predominantly express PKD2 followed by PKD3 while PKD1 displays negligible levels. Furthermore, we demonstrate that PKD2 is specifically phosphorylated/activated at the time of germ layer segregation. Time-restricted PKD2-activation limits mesendoderm formation and subsequent cardiovasculogenesis during early differentiation while leading to branching angiogenesis during late differentiation. In line, PKD2 loss-of-function analyses showed induction of mesendodermal differentiation in expense of the neuroectodermal germ layer. Our in vivo findings demonstrate that embryoid bodies transplanted on chicken chorioallantoic membrane induced an angiogenic response indicating that timed overexpression of PKD2 from day 4 onwards leads to augmented angiogenesis in differentiating ESCs. Taken together, our results describe novel and time-dependent facets of PKD2 during early cell fate determination.

Herein, PKD2 has been recently shown to regulate cardiac valve formation 13 but also erythropoiesis 14 . However, only a handful of studies report on the expression of PKDs in various stem cell populations. We recently demonstrated that distinct PKD isoforms, dominated by PKD2, are expressed in undifferentiated myoblasts and regulate their differentiation 15 . Similarly, a BMP-PKD2 axis regulates osteoblast differentiation from human mesenchymal stem cells 16 . However, PKD2 activity is not only present in normal stem cells but also in tumour stem cells as shown for CD133(+ ) glioblastoma-initiating cells 17 . A recent study identified PKD1 as an anti-differentiate, pro-proliferate signal in the skin tissue 18 . This observation is not only limited to physiological skin formation but also to cancer initiation. The expression of CD34 in cutaneous cancer stem cells is required for stem cell activation and tumour formation. Furthermore, PKD1 was found to be strongly expressed in CD34(+ ) cells and that inhibition of PKD1 could be preventive in skin cancer development 18 .
As one of the early events during gastrulation, definitive endoderm (DE) and anterior mesoderm derivatives, including cardiovascular and head mesenchyme progenitors, are generated from a transient precursor cell population located in the region of the anterior primitive streak. This cell population is commonly referred to as mesendoderm giving rise to mesoderm and endoderm and is marked by the expression of marker genes such as Brachyury (T) and Foxa2 [19][20][21] . Soon afterwards, the development of the circulatory system is initiated from a common multipotent progenitor cell type, the so-called hemangioblast. This process of de novo formation of blood vessels is called vasculogenesis and is prevalent in the mouse embryo until E8.5. Vasculogenesis is accompanied by a complementary process called angiogenesis, an event that defines vessel formation from pre-existing endothelial cells that undergo sprouting and that is shown to commence in the embryo at E9.5 22 . Various laboratories, including ours, have delineated the role of PKDs during physiological and tumour angiogenesis [23][24][25][26][27][28] . In fact, recent data indicate that in endothelial cells PKD2 is the predominant PKD isoform that is required for proliferation, migration, in vitro angiogenesis and expression of vascular endothelial factor receptor-2 (VEGFR2) as well as fibroblast growth factor receptor-1 (FGFR1) 26 . Moreover, our group identified PKD2 as a novel, essential mediator of tumour cell-endothelial cell communication 29 and as a critical modulator of hypoxia-induced VEGF expression/secretion by the tumour cells 30 . Other recent studies from our laboratory described PKD1 and − 2 isoform-selective effects on cancer cell invasion and angiogenesis 17,31,32 .
The only data linking PKDs to vasculogenesis come from a recent study in zebrafish. Herein, PKD1 deletion moderately reduced the formation of the intersomitic vessels and the dorsal longitudinal anastomotic vessel. In addition, the formation of the parachordal lymphangioblasts, a precursor for the developing thoracic duct, is perturbed upon PKD depletion. By contrast, PKD induced tumour angiogenesis in zebrafish xenografts 33 . This indicates a time-restricted PKD-responsive window during distinct developmental stages and a strong PKD effect during angiogenesis. However, such a hypothesis has never been explored due to the lack of appropriate model systems.
Pluripotent stem cells represent a powerful tool to investigate embryonic development in mouse and human [34][35][36][37] . Moreover, these cells provide a unique platform for dissecting the distinct mechanisms underlying pluripotency and subsequent lineage commitment 37 . Given the high corroboration between embryonic development (in vivo) and pluripotent stem cell differentiation (in vitro), various factors can be investigated in a time-and stage-dependent manner regarding their impact on cell fate determination, differentiation and/or tissue formation 21,38,39,43 .
In this study, we examined the contribution of PKD2 during cell fate determination and cardiovascular lineage formation. To investigate the timed PKD2 activation during ESC differentiation, we used mouse embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) for in vitro cell culture experiments as well as for in vivo assays with a CAM (chorioallantoic membrane) xenograft. PKD2 is dynamically expressed/activated during the first days of differentiation. Functionally PKD2 represses mesendoderm formation and subsequent cardiovascular lineage commitment when activated during germ layer segregation. At later stages, PKD2 promotes exclusively the vascular lineage. In line with these findings, a genetic loss-of-function approach based on the expression of a kinase-dead variant showed a reciprocal differentiation pattern as evident by an increase in mesendodermal differentiation. Thus, our data define for the first time a PKD2-responsive time-window to drive predominantly angiogenesis instead of vasculogenesis.

Experimental Procedures
Ethics statement. All  Generation of iPKD2-ESCs. One day before the nucleofection procedure, A2lox.cre cells were incubated with 1 mg/ml of doxycycline to induce the recombination by Cre-recombinase. ESCs were transfected using the Nucleofector Technology (Lonza, USA) according to the manufacturer's procedures. 10 μ g DNA (iPKD2-p2lox vector) was used to transfect 5 million parental A2lox.cre ESCs. After transfection cells were plated on neomycin-resistant, Mitomycin-C (Sigma, USA) inactivated MEFs. Two days later cells were subjected to selection with neomycin (400 μ g/ml) for additional 7 to 10 days. Several Scientific RepoRts | 5:11742 | DOi: 10.1038/srep11742 clones were picked, expanded and analysed for correct recombination of the HPRT locus using the "lox in" PCR.

Generation of induced pluripotent stem cells (iPSCs) from PKD2-kinase-dead-mice.
Mouse embryonic fibroblasts (MEFs) were isolated according to standard protocols from wild type C57B6 mice (WT) and from a previously reported mouse strain in C57B6 background, where the two critical serine residues (Ser 706, 710) of PKD2 (PKD2-KD) have been replaced by alanine leading to the expression of a kinase-dead mutant. This mouse strain lacks the catalytic activity of PKD2 and recapitulates the phenotype of a PKD2 gene trap line 40,41 . Reprogramming virus was generated according to standard protocols as described previously 34,37,42 . 100,000 WT and PKD2-KD MEFs per well of a 6well-plate or 40,000 cells per well of a 12well-plate, respectively, were seeded one day before infection with virus. Next day, 8 μ l per well of a 6well-plate or 4 μ l per well of a 12well-plate, of 100 × concentrated virus (= 3,5 × 10 7 proviral hOSKM copies/μ l) were added to the cells as a master mix. After 8 h of incubation at 37 °C, medium was changed to ES-Feeder and refreshed every day. After one week, infection rate was assessed by FACS analysis (data not shown). At day 7, medium was changed to ESC culture conditions and changed every day. ESC culture conditions have been described previously 43 , but for reprogramming instead of fetal calf serum, knock out serum replacement (Invitrogen, Germany) was used. On day 13, cultures were fixed and stained for alkaline phosphatase (AP)-expression according to standard protocols. Several independent iPSC clones derived from WT and PKD2-KD MEFs were randomly picked based on typical ESC like morphology at day 13 of reprogramming. After expansion of individual clones, pluripotency analysis was carried out at early passage (1)(2)(3)(4)(5). All analysed clones were stained positive for pluripotency markers and capable to in vitro differentiate to the three germ layers.
EB formation: Iscove's modified Dulbecco's medium (IMDM; Invitrogen) supplemented with 10% FCS (Lonza, USA), 1% Penicillin/Streptomycin, 1% GlutaMax, 1% NEAA and freshly prepared 450 μ M Monothioglycerol (Sigma) were used for differentiation. Briefly, 600 (or as indicated in figure legends) cells per 20 μ l differentiation medium were dropped on petri dish covers, placed upside down on petri dishes filled with 10 ml Dulbecco's Phosphate Buffered Saline (DPBS; Invitrogen) and were cultivated for 2 days in hanging drops. Subsequently, embryoid bodies (EBs) were transferred onto non-adherent petri dishes and were cultivated for additional two days. On day 4, EBs were plated on 0.1% gelatine-coated 6-well dishes or cover slips for RNA or immunofluorescence analysis respectively and assayed at specific time points as described in figure legends. Doxycycline was added as indicated in figure legends.
Immunocytochemistry. Cells at different time points of differentiation were fixed using 4% paraformaldehyde (PFA). Samples were then subjected to treatment with NH 4 Cl and blocking with 0. (1:20,000). Images were captured using an upright fluorescence Zeiss Axioimager Z1 microscope and analysed using Axiovision software (Zeiss, Germany).
Quantitative real-time RT-PCR. qPCR analysis was performed either as one-step or two-step real-time PCR depending on the amount of analysed genes. One-step real-time qPCR was carried out with the Rotor Gene RT-PCR Cycler (Qiagen) using the QuantiTect SYBR Green RT-PCR kit (Qiagen). Each RNA preparation was tested for genomic DNA contamination by replacing reverse transcriptase with water. Internal standards (house-keeping gene) and samples were simultaneously amplified. Details have been described elsewhere 35,43,44 . For two-step real-time PCR iScript ™ cDNA Synthesis Kit (Biorad, #170-8891) and SensiMix ™ SYBR ® No-ROX Kit (Bioline, QT650-05) have been used according to supplier's instruction manual using 100 ng cDNA as a template. PCR conditions included denaturation at 95 °C for 20 s, followed by 50 cycles of 94 °C for 1 min and 60 °C for 20 s, then continuation with dissociation stage. QuantiTect Primer Assays were used throughout this study (Qiagen, Germany, http://www.qiagen.com/ de/products/catalog/assay-technologies/real-time-pcr-and-rt-pcr-reagents/quantitect-primer-assays/).

Statistical Analysis.
If not stated otherwise, SEMs are indicated by error bars. Generally levels of significance were calculated with the two-sides Student's t test or Anova (*p < 0.05; **p < 0.01; ***p < 0.001) using Prism5 (GraphPad, USA). If not otherwise specified three biologically independent experiments done in replicates were evaluated.

PKD2 expression in undifferentiated and differentiating embryonic stem cells.
PKDs are significantly expressed in the developing mouse embryo and in particular, PKD2 shows a differentially regulated expression pattern 12 . In embryonic stem cells, PKCs (the upstream activators of PKDs) drive primitive endoderm formation, while their inhibition facilitates pluripotency 45 . However, the contribution of PKDs during cell fate determination is not clearly understood. Employing murine ESCs as a tool, we sought to characterize the role of PKDs in early stages of differentiation. First, we investigated the mRNA and protein expression of PKD1, PKD2 and PKD3 in undifferentiated mouse ESCs. As shown in Fig. 1A and B, Pkd2 was highly expressed in ESCs, followed by Pkd3. Pkd1 was barely detectable at mRNA and protein level (Fig. 1A,B). Transcriptome data from a public database confirmed our findings (http://biit.cs.ut.ee/fungenes/). Furthermore, the examination of the abundance of PKD1-3 via immunochemical staining (ICC) indicated predominant cytoplasmic localization of PKD2 in both single ESCs and compact ESC colonies. In line with the mRNA data, PKD1 staining was hardly visible, while PKD3 immunoreactivity confirmed an intermediate expression ranking between PKD1 and PKD2 (Fig. 1C). Accordingly, we next examined the expression of PKD2 during ESC differentiation using embryoid body (EB)-based assays. We observed prominent and significant PKD2 expression peaks at day 2 and day 6 of differentiation on mRNA levels (Fig. 1D). Interestingly, protein levels of PKD2 increased continuously until day 6 ( Fig. 1E,G). Of note, the PKD2 antibody used in this study specifically detects PKD2 and does not show cross reactivity with either PKD1 or PKD3 15 . Notably, PKD1 showed moderate protein levels starting at day 4 and day 6 with negligible levels on day 0 and day 2 (Fig. 1E). Next we aimed to assess PKD activity. Western blot analysis with a phospho-specific PKD antibody targeting the activation loop of the kinase demonstrated that undifferentiated ESCs exhibited a moderate basal PKD activity that peaked on day 2 and persisted on lower levels through day 6 of differentiation (Fig. 1F,H). Of note, PKD activity on day 2 mostly reflects PKD2 activity as PKD1 is virtually absent at this stage (Fig. 1B,E). PKD2 specific phosphorylation is further confirmed by immunoprecipitation of the PKD2 protein and subsequent immunoblotting with a phospho-specific PKD antibody (Fig. 1I). PKDs are activated by PKCδ and PKCε , two serine threonine kinases belonging to the novel PKC family 4,5 . In mouse embryonic stem cells PKCδ has been shown to be functionally expressed 46 . Indeed, we found phosphorylated PKCδ coincide with PKD2 activity suggesting that in differentiating ESCs, PKCδ could activate PKD2 (Fig. 1J). Together, these data suggest that PKD2 expression and activity may play a role during ESC differentiation. Based on these observations, we decided to focus on PKD2 in further experiments.
A doxycycline conditional PKD2 knock-in allele. In order to dissect the role of PKD2 in ESC differentiation, we targeted an inducible PKD2-knock-in allele in ESCs to allow the temporally regulated and dose-dependent expression of PKD2 (iPKD2 ESCs; dose-dependence not shown, Fig. 2A). We used a rapid and efficient recombination system in the HPRT locus, where a doxycycline (Dox)-inducible promoter regulates expression of the PKD2 mRNA. This overexpression system has been extensively applied in a series of studies to delineate the function of a particular gene during differentiation 21,39,[47][48][49][50][51][52] . Phosphorylation at three critical serine residues of PKD2 has been reported to solely achieve maximum activation (Ser 706, 710 and 244) and not via gene expression alone 4 . Consequently, we targeted a constitutively active PKD2 mutant (PKD2 S244/706/710E, PKD2-3SE) that mimics phosphorylation 4 . The correct gene recombination in the HPRT locus is shown in Fig. 2B. The PKD2 allele was effectively induced upon Dox-exposure as assessed by mRNA and western blot (Fig. 2C,D). Notably, PKD2 overexpression did not affect the expression levels of the SSEA1 protein indicating a negligible role of PKD2 for the pluripotency circuitry (Fig. 2E).
Timed PKD2 overexpression during ESC differentiation. In pluripotent stem cells, germ layer segregation occurs around days 2/3 19,53 , while the onset of vasculogenesis is ignited around day 3 by the emergence of the hemangioblast, a common progenitor of the endothelial and hematopoietic lineage 54 . The hemangioblast is defined by the co-expression of the vascular endothelial growth factor receptor (Flk1/VEGFR-2) and the mesodermal transcription factor brachyury (T). A second progenitor population arises shortly after the development of the hemangioblast 55 . These cardiovascular progenitors are tripotent giving rise to cardiomyocytes, endothelial cells and smooth muscle cells and co-express Mesp1, brachyury, PDGFRa and c-kit 56 . Using embryoid body (EB) -based differentiation, mimicking early cell fate decision in the embryo, we assessed the effects of overexpression of PKD2 on marker genes of all three germ layers, but also on above mentioned progenitor population markers. In a first approach, PKD2 was activated during the first 4 days of differentiation (Fig. 3A) to investigate changes in germ layer formation and vasculogenesis 54 . In PKD2 overexpressing cells, we observed a downregulation of early markers for mesoderm and endoderm [Brachyury, FoxA2 (Fig. 3B,C)] and upregulation of the early ectoderm marker Pax6 (Fig. 3D). In line, markers labelling both the hemangioblast and the tripotent cardiovascular progenitor were repressed [Flk1, cKit, Pdgfr ( Figure 3E-G)]. In addition, early cardiac and endothelial markers were downregulated at day 4 [Tbx5, Isl1 (Fig. 3H,I)]. These findings were confirmed with a second, independently targeted iPKD2 ESC line (Supplemental Fig. 1A-F).
Next, we analysed whether late cardiovascular-derived cell types are similarly repressed upon ectopic PKD2 expression in a prolonged ESC differentiation culture. At late time-points of differentiation, EBs overexpressing PKD2 showed diminished cardiac differentiation as demonstrated by the reduced number of beating clusters under the respective culture conditions (data not shown). This was accompanied by decreased mRNA levels of cardiac-specific genes Myl2a, Myh6 (Fig. 3K,L) and a decreased number of cells expressing α -actinin protein (Fig. 3M). Similarly, endothelial differentiation appeared to be suppressed as shown by downregulated expression of CD34 and CD31 mRNA (Fig. 3N,O). CD31 positive vessel-like structures were also reduced in cultures overexpressing PKD2 between day 0 and day 4 (Fig. 3P), the predominant time frame/window when vasculogenesis occurs 54 . β -tubulin 3 (Tubb3) expression was enhanced suggesting an inductive effect of PKD2 on the ectodermal germ layer (Fig. 3J). The latter became particularly evident at late time-points of differentiation after PKD2 activation from day 0 to day4 (Fig. 3J, data not shown).
The primary vascular plexus in the EB is remodelled by vessel sprouting starting at day 6 and later, thus marking angiogenesis 54 . Given our primary aim to define a PKD2-responsive time-window during   vascular development, we triggered the overexpression of PKD2 via Dox-exposure starting at day 4 of differentiation and maintained doxycycline in the culture medium until day 14 of differentiation (day 4 to day 14). Continuous Dox-treatment (day 0 to day 14) was administered to distinguish between an early and late PKD2 responsive window (Fig. 4A). Given the close association between the developmental course of the vascular and cardiac lineage, we next sought to investigate the cardiac differentiation potential. Quantification of beating clusters in differentiating EBs showed no significant difference in cardiac differentiation (data not shown). This observation was further mirrored by the Myh6, Myl2a gene expression analyses at days 6, 9 and 14 of EB differentiation (Fig. 4B,C) and immunostaining of α -actinin (Fig. 4D). Consistent with an early inhibitory effect of PKD2, when activated from day 0 to day 4, continuous activation led to reduced expression of cardiac markers (Fig. 4B,C). Next, we assessed the development of our primary lineage of interest, namely endothelial cell differentiation. qPCR analysis revealed a significant upregulation of vascular marker genes (vWF, CD34, CD31) in Dox-treated compared to untreated cultures particularly at days 6, 9 and also 14 though less pronounced (Fig. 4E-G). These findings were corroborated with immunostaining for CD31 (Fig. 4H). Of note, Tubb3 expression was reduced upon later PKD2 induction (data not shown).

Continuous PKD2 loss-of-function during ESC differentiation.
To substantiate the above findings, we made use of a previously reported PKD2 loss-of-function mouse model. In this mouse strain two critical serine residues (i.e. Ser 706, 710) in the kinase domain of PKD2 have been replaced by alanine leading to the expression of a kinase-dead mutant. Therefore, catalytic activity of PKD2 cannot be detected and the mouse recapitulates the phenotype of a PKD2 gene trap line 40,41 . Mouse embryonic fibroblasts isolated form this mouse strain served as a somatic template to generate mutant induced pluripotent stem cell lines. Albeit the experiment did not aim to investigate the impact of PKD2 on pluripotency, we noted fewer Alkaline-Phosphatase (AP)-postive cultures in the reprogrammed kinase-dead fibroblasts (Fig. 5A). However, further studies need to focus on this aspect in more detail. We manually selected several induced pluripotent stem cell clones according to their dome-shaped compact colony morphology (Fig. 5B). These iPSC lines displayed all hallmarks of pluripotency such as marker expression of SSEA1 and Oct3/4 (Fig. 5C) and were capable to differentiate into all three germ layers: endoderm (Sox17), mesoderm (T) and ectoderm (Tubb3) (Fig. 5D). The presence of the mutated PKD2 locus was confirmed in two independent clones via PCR (Fig. 5E). Next, we applied two clones to subsequent spontaneous differentiation as outlined in Fig. 6A. qPCR analyses revealed increased mesendodermal marker expression in the early differentiation phase until day 4 (Brachyury, FoxA2, Fig. 6B,C), while ectodermal markers Pax6, Tubb3) showed decreased expression (Fig. 6D,E). Thus, PKD2-KD iPSCs display the expected, entirely opposite germ layer segregation pattern compared to an early activation of PKD2 (see Fig. 3B-J). Markers of the hemangioblast and the tripotent cardiovascular progenitor showed a similar pattern (Fig. 6F-I). These findings were confirmed with a second PKD2-KD iPSCs clone (Supplemental Fig. 2A-F). Later on, a trend towards decreased terminal differentiation of the cardiovascular lineage was observed as demonstrated by various markers gene expression (Fig. 6J-N). On protein level, CD31 positive cells were fewer and more scattered distributed in PKD2-KD EBs compared to WT, but appeared more organized in structure (data not shown, Fig. 6O).

PKD2 directs angiogenesis in an in vivo model of ESC differentiation.
Finally, we aimed to confirm the basic findings of our data in an in vivo model system. The CAM assay is a widely used method to study angiogenesis in vivo 57 and previous studies have already successfully applied this tool to assess EB-derived angiogenesis 58 . Based on our in vitro observation that PKD2 induction at later time points of differentiation directs angiogenesis from ESCs, we transplanted untreated 4 days old EBs and subsequently triggered the PKD2 overexpression upon ectopic application of doxycycline onto EBs grafted on the CAM (Fig. 7A). After 5 days, a pronounced sprouting of chicken-derived blood vessels into the tumour-like structures of PKD2-induced EBs was observed (Fig. 7B). Moreover, we observed more vessel-like structures inside the differentiating EBs compared to untreated EBs, as demonstrated by enhanced immunoreactivity for the vWF protein (Fig. 7C,D).

Discussion
Previous studies have shown distinct but also overlapping functions of the different PKD family members (PKD1-3) during mouse embryogenesis 11,12 . PKD1 kinase-dead knock-in mice die in utero around day E9.5 while mice featuring a PKD2 deletion are viable and fertile 40,41 . However, a detailed phenotypic analysis of this early lethality in PKD1 null remains elusive but primitive haematopoiesis and vasculogenesis start around this time period. PKD2 exhibits unique characteristics among the PKD family members, due to its pronounced role during physiological and tumour angiogenesis. Yet, implications of PKD2 during cell fate determination of embryonic stem cells remain elusive. Moreover, despite an intimate connection between vasculogenesis and angiogenesis during early development, currently available knowledge is limited to the role of PKD2 during angiogenesis. The current study addresses this by the application of mouse embryonic stem cell differentiation using both complementary gain-and loss-of-function approaches. Thereby we are able to define a time-responsive window being accessible for PKD2 activity and show that PKD2 limits mesendodermal commitment and vasculogenesis but later on induces branching angiogenesis.   Table 2.
Interestingly, the three PKD isoforms act in a context-and cell-type dependent, similar or opposing manner 17,31 : on the one hand, PKD2 gain-of-function in pancreatic cancer promotes invasion and angiogenesis while PKD1 limits these processes 31 ; on the other hand, angiogenesis in the CAM is similarly   promoted by the PKD isoforms 29 . Moreover, PKD1 and 2 share various structural and functional similarities while PKD3 is not as homologous to the two other family members. Particularly, the N-terminus of both PKD1 and PKD2 starts with a hydrophobic domain, which is absent in PKD3. Also, PKD3 lacks the autophosphorylation sites at S916/S876, which have been shown to regulate PKD conformation and activation length (for review see 2 ). Herein, our own data obtained in skeletal muscle stem cells, so-called myoblasts, indicate PKD2 to be expressed at considerably higher levels than PKD1. In line, PKD2-depleted myoblasts exhibited impaired, while forced-PKD2 expression increased myoblast differentiation. Of note, PKD1 and PKD3 had only minor effects. This observation indicates that assessing expression levels of PKDs in the parental stem cell population can predict its relevance during differentiation 15 . This together with the fact that PKD1 was virtually undetectable in undifferentiated ESCs, prompted us to focus on PKD2 in the current study. Intriguingly, atypical PKCs direct differentiation of human and mouse embryonic stem cells leading to PKC inhibitor based culture systems to keep pluripotent stem cells in an undifferentiated stage 45,59 . While we did not observe an obvious impact upon forced PKD2 expression on pluripotency, we noticed fewer AP positive areas in EBs derived from iPSCs expressing the kinase-dead mutant. This urges for further detailed investigations of PKD2 as one possible subordinate player to affect pluripotency.
De novo blood vessel formation defines vasculogenesis while formation of sprouting blood vessels from pre-existing vessels is commonly considered as angiogenesis. While embryonic vasculogenesis is difficult to model in vitro, common adult angiogenesis models to study e.g. formation of vessel tube formation comprise primary endothelial cells or endothelial cell lines such as HUVECs. Interestingly, both embryonic processes can be studied in differentiating embryonic stem cells 54,60 , a unique feature of ESCs not offered by any other cellular model system. Differentiating ESCs are not only able to display vasculogenic features during the first 4 days of EB differentiation. At later stages, after formation of early blood vessels/-angioblasts, they also exhibit embryonic branching angiogenesis. Thus, a particular factor can be investigated with respect to its impact on both processes in the same culture system 54,60 . Nevertheless, pure in vitro vasculogenesis and angiogenesis models lack complex processes that e.g. require progression through several developmental stages 54 . For that, we made use of a combined in vitro/in vivo CAM (chorioallantoic membrane) assay. This assay comprises the advantages of physiological vessel formation with an easy and fast possibility to transplant embryoid bodies derived from gene-manipulated, differentiating ESCs.
To integrate these advantages into one system, we engineered an inducible knock-in allele targeting PKD2 to the HPRT locus and complemented this tool with a genetic loss-of-function system, the PKD2 kinase-dead expressing iPSCs 21,39,47,49,50 . To our surprise, early PKD2 activation did not induce vasculogenesis, but rather repressed mesendodermal differentiation, while inducing a neuroectodermal fate. In line, abrogation of PKD2 activity using genetic loss-of-function tools in pluripotent stem cells triggered opposing effects mirrored by the induction of mesendodermal derivatives and cardiovascular progenitors in expense of a neuroectodermal fate. This is particularly interesting in light of our data showing a PKD2 activation peak around day 2 of differentiation. At this stage, PKD1 remains virtually undetectable. On the other hand, we observed an induction of vascular networks in late PKD2 overexpressing cultures, when PKD2 was induced after day 4 of differentiation. This period coincides with embryonic branching angiogenesis in the ESC-based EB differentiation system 54,60 . To our surprise, there was a trend towards reduced terminal cardiovascular differentiation albeit only some markers turned out to be significantly changed, despite the fact that the mesendodermal germ layer was induced. This discordance between gain-of-function and loss-of-function data might be explained by the inability of timed manipulation in the PKD2 loss-of-function system. Spatiotemporal differences in PKD2 function/requirement during embryonic stem cell differentiation, as suggested by our gain-of-function data, could attenuate the inductive effect of PKD2 loss after germ layer segregation has occurred. However, the CD31 positive vascular structures appeared, despite overall reduced, more branched compared to control cultures. This observation needs further investigation in future studies.
How could these phenotypic observations be explained? Several reports suggest PKD2 to act as the predominant isoform promoting physiologic and tumour angiogenesis 26 . Second, the PKD2 knock-out mouse is viable and fertile without any gross abnormalities, thus the catalytic activity appears not to be crucial for in vivo mouse embryogenesis 41 , (see below). In line, knock-out mouse models of genes being highly critical for proper vasculogenesis, usually undergo embryonic lethality [61][62][63] . Thus, our data showing enhanced mesendodermal commitment upon PKD2 loss is consistent with a viable PKD2 knock-out mouse. Third, a recent manuscript using the zebrafish as model system reported only minimal changes with respect to vasculogenesis, but showed strong effects when PKD1 was knocked-out during xenograft angiogenesis 33 . Fourth, a loss-of-function approach in Drosophila showed wing vein defects, but no other vascular malformations were reported 64 .
However, there is still a remaining discordance between a viable and fertile PKD2 knock-out mouse and our in vitro data reporting alterations during germ layer formation and cardiovascular differentiation. We cannot exclude that compensational events might mask various effects during development in case of loss of a particular PKD isoforms. Indeed, overlapping expression between PKD1, 2 and 3 have been reported in several studies. In particular, PKD1 and PKD2 isoforms display very similar structural properties and phenotypes. Briefly, our own group has gathered deep insights into PKD isoform specificity and defined various overlapping functions in different cancers types including control of proliferation, angiogenesis or cell motility 2,29,30,65-68 and differential expression in various stem cell populations 15 . Of note, we recently reported PKD actions with isoform specific behaviour and even opposing effects 15,31 . In fact, specific events such as gastrulation but also vasculogenesis leading to early lethality in the developing embryo are more frequently protected. Thus, compensation is an established issue throughout development and differentiation [69][70][71][72][73][74][75][76][77] , particularly of vascular development [78][79][80][81] . Apparently this compensation is attenuated in our in vitro system, improving the chances to identify factor specific functions 21 and isoform specific effects of a particular protein family 21 .
In summary, the current study suggests a role of PKDs during embryonic development using a complementary gain and loss-of-function approach. For the first time, we can show that PKD2 limits mesendoderm formation and cardiovascular progenitor formation, but preformed vessels arising after vasculogenesis become permissive for PKD2 activity to allow embryonic branching angiogenesis. Still, there are remaining questions to be answered to gather deeper insights into the complex isoform specific interplay of PKDs during development. Conditional but also combinational knock-out mouse models complemented with tightly to regulate gain-and loss-of-function in vitro systems will support future studies.