Pkd1 Regulates Lymphatic Vascular Morphogenesis during Development

Initial lymphatic precursor sprouting is normal in lyc1 mutants, but ongoing migration fails. Loss of Pkd1 in mice has no effect on precursor sprouting but leads to failed morphogenesis of the subcutaneous lymphatic network. Individual lymphatic endothelial cells display defective polarity, elongation, and adherens junctions. This work identiﬁes a highly selective and unexpected role for Pkd1 in lymphatic vessel morphogenesis during development. of collagen fibers in the medial layer of peri-notochordal region in WT (n=3), WT/MO- pkd2 (7.5 ng MO) (n=3), WT/MO- pkd1b (5 ng MO) (n=3) and lyc1/MO-pkd1b embryos (5ng MO)(n=3). No change was observed in peri-notochordal collagen.


INTRODUCTION
The lymphatic vasculature forms in the embryo as a result of specification of lymphatic endothelial cell (LEC) fate, followed by coordinated sprouting, morphogenesis, and network elaboration. LEC fate is specified through key transcription factors, which act in embryonic veins (Franç ois et al., 2008;Srinivasan et al., 2010;Wigle and Oliver, 1999). LEC precursors subsequently sprout from veins and migrate through the embryo (reviewed in Koltowska et al., 2013). This process is under the control of VEGFC/VEGFR3 signaling (Karkkainen et al., 2004) and its modulators (reviewed in Koltowska et al., 2013). In mouse, lymphatic precursors form lymph sacs in the anterior of the embryo (Franç ois et al., 2012;Yang et al., 2012), which likely remodel into major lymphatic vessels (Hä gerling et al., 2013). Superficial LECs (sLECs) migrate dorsally as loosely attached individual cells to form the subcutaneous lymphatic network (Hä gerling et al., 2013). Although several guidance molecules, cellular interactions, and extrinsic forces pattern embryonic lymphangiogenesis (reviewed in Koltowska et al., 2013), much remains to be understood about the cellular mechanisms that regulate LEC polarization, adhesion, outgrowth, remodeling, and morphogenesis.
In zebrafish, there are strong parallels with mammals in the processes that regulate lymphatic vascular development (Hogan et al., 2009b;Kü chler et al., 2006;Yaniv et al., 2006). We have used forward genetic screens to identify zebrafish mutants that lack lymphatic vessels. Here, one zebrafish mutant uncovers a surprising role for the ADPKD gene Pkd1 in lymphatic vascular development. We show that this function of Pkd1 is conserved and cell autonomous in endothelial knockout mice. Our findings suggest a uniquely staged role for PKD1 in the regulation of lymphatic vascular morphogenesis.

lyc1 Mutants Fail to Form a Lymphatic Vasculature
We identified a zebrafish mutant dubbed lymphatic and cardiac defects 1 (lyc1). lyc1 mutants exhibited a reduction or loss of the main axial lymphatic vessel, the thoracic duct (TD) at 4 days postfertilization (dpf) as well as mild cardiac edema, while retaining blood circulation ( Figures 1A-1D and 1I). By 5 dpf, mutant blood flow was reduced and cardiac edema increased in severity (Figure S1; data not shown). To determine the origins of the phenotype, we examined gene expression for arteriovenous genes, lymphangiogenesis regulators (including chemokines and receptors), and flow-induced pathways at 32 hr postfertilization (hpf), during the initiation of lymphatic development. These markers were unchanged in lyc1 embryos ( Figure S1). In the zebrafish, precursor LECs emerge from the posterior cardinal vein (PCV) during secondary angiogenesis and migrate dorsally to the horizontal myoseptum to form parachordal lymphangioblasts (PLs). Concomitantly, venous sprouts form intersegmental veins (vISVs). Strikingly, the numbers of vISVs and PLs were normal in lyc1 mutants ( Figures 1E-1H, 1J, and 1K). This phenotype differs from described mutants for vegfc, vegfr3, or ccbe1 (Hogan et al., 2009a(Hogan et al., , 2009bLe Guen et al., 2014;Villefranc et al., 2013), which lack all venous sprouting. Time-lapse imaging showed that the lymphatic defect resulted from a block in the migration of PLs out of the horizontal myoseptum (Movies S1 and S2). Quantitative analysis of cell behavior spanning this period of altered migration revealed that mutant precursor LECs remain mobile but show altered exploratory behavior and filopodial extension dynamics, consistent with impaired directional migration (Movies S3 and S4; Figure S2).
A Loss-of-Function Mutation in pkd1a Is Responsible for the lyc1 Phenotype Meiotic mapping (see the Experimental Procedures) was used to identify a region of chromosome 1 containing the lyc1 locus. The critical interval (Figure 2A) contained two genes, tuberous sclerosis 2 (tsc2) and polycystic kidney disease Ia (pkd1a). In the zebrafish genome, pkd1 (encoding Polycystin1) is present as duplicate genes, with pkd1a coding for a conserved 4281 amino acid protein. Sequencing revealed a mutation in pkd1a, introducing a premature stop codon (R3607X) ( Figure 2B). This mutation was predicted to result in the failed translation of six of the 11 transmembrane domains and essential C-terminal cytoplasmic tail of the protein.
In humans, PKD1 and PKD2 (encoding POLYCYSTIN2) are the most commonly mutated genes in ADPKD (for review, see Chapin and Caplan, 2010;Zhou, 2009). PKD1 haploinsufficiency and loss of function have also been frequently associated with cardiovascular complications (reviewed in Rossetti and Harris, 2013). In mammals, POLYCYSTIN1 protein localizes to primary cilia, apical membranes, adherens, and desmosomal junctions. It can act as a mechanosensory signaling protein, transducing extracellular signals through its cytoplasmic C-terminal domain (reviewed in Zhou, 2009). POLYCYSTIN1 binds to POLYCYSTIN2 (a calcium pump) at the membrane to regulate Ca 2+ influx and signaling but also binds to E-cadherin, b-catenin, and components/effectors of the planar cell polarity pathway (Castelli et al., 2013;Lal et al., 2008;Roitbak et al., 2004).
(legend continued on next page) phenotype, confirming that the lyc1 mutation is a loss-of-function allele ( Figures 2C and S3). Pkd1 and Pkd2 can modulate extracellular matrix (ECM) formation (Mangos et al., 2010). Importantly, even the most phenotypically penetrant pkd1a mutants for lymphangiogenesis do not display the body curvature associated with altered ECM. We examined several markers and knockdown scenarios but found no evidence for increased ECM or a role of altered matrix in the lyc1 lymphatic phenotype ( Figure S4).
pkd1a Is Expressed in Migrating LECs and Loss of Function in the ADPKD Complex Mimics lyc1 Defects We found that pkd1a expression was ubiquitous in the 24 hpf embryo but was enriched in the trunk during secondary angiogenesis at 32 hpf ( Figure 2D). We saw no evidence for nonsense-mediated decay in mutants using in situ hybridization at 32 hpf (n = 130 embryos from a carrier incross analyzed; data not shown). As in situ hybridization has proved insensitive in LECs in older zebrafish (post 3 dpf), we isolated LECs using fluorescence-activated cell sorting (FACS). Taking advantage of a new transgenic line Tg(lyve1:DsRed2) nz101 (Okuda et al., 2012) labeling embryonic veins and lymphatic vessels, crossed onto the Tg(kdrl:egfp) s843 line (restricted to blood vessels; Jin et al., 2005), we isolated LECs and venous ECs (VECs). We performed quantitative PCR (qPCR) for known markers, validating the specificity of cell populations (Figures 2E and S3). Consistent with the timing of the lyc1 phenotype, pkd1a and pkd2 were expressed in VECs and LECs, with pkd1a in both populations at 3 dpf but reduced in LECs at 5 dpf. pkd1b was expressed at low, almost undetectable levels at all stages (Figures 2F and S3).
In endothelial cells, Polycystin1 can regulate calcium signaling through Polycystin2 activity (Chapin and Caplan, 2010;Nauli et al., 2003). To investigate this potential mechanism, we knocked down Pkd2. Embryos depleted for Pkd2 exhibited a phenotype similar to that of lyc1/MO-pkd1b embryos and reduced TD extent (Figures 2G-2I and S3). We next treated embryos with previously validated Ca 2+ signaling antagonist and agonists (North et al., 2009). These treatments generated phenotypes highly reminiscent of the lyc1 phenotype ( Figure S3). Cacna1s, an L-type calcium channel targeted by the antagonist Nifedipine, was expressed in ECs ( Figure S3). Taken together, these observations are consistent with Pkd1 functioning in the canonical ADPKD complex.

Pkd1 Cell-Autonomously Regulates Development of the Subcutaneous Lymphatic Vascular Network in Mice
Although most previous studies in mammalian models focus on the role of Pkd1 in epithelia, Pkd1-null mice have been shown to exhibit cardiovascular, skeletal, and renal defects (Boulter et al., 2001;Kim et al., 2000;Piontek et al., 2004). Embryos devoid of Pkd1 die after 15.5 days post coitum (dpc) displaying severe hemorrhaging and subcutaneous edema (Kim et al., 2000;Muto et al., 2002), but a role for this gene in lymphangiogenesis has yet to be reported.
We generated Pkd1 knockout embryos and examined their overall morphology. We observed the previously described subcutaneous edema, but not hemorrhaging ( Figures 3A-3C). Embryonic lymph sacs were present but were blood filled in Pkd1 KO embryos ( Figures 3D-3I). This phenotype suggests that lymphatics in this mutant would not sustain fluid drainage and may explain the subcutaneous edema. Interestingly, we did not find any defect in lymphovenous valves at 14.5 dpc (Figure S5) perhaps suggesting that blood enters the mutant lymph sacs early during morphogenesis, before valve maturation (Franç ois et al., 2012). We next examined the developing subcutaneous lymphatic vasculature in dorsal embryonic skin, a useful system to quantify lymphatic vascular phenotypes (James et al., 2013;Kartopawiro et al., 2014). We found that Pkd1 KO embryos exhibit defects in the morphogenesis of the lymphatic network, with increased width of sprouting vessels, increased cell number per vessel, and a significant reduction in network branching (Figures 3L,3M,3R,3T,and 3U).
Previous studies reported that Tie2:Cre-mediated deletion of Pkd1 did not lead to vascular abnormalities, and these knockout mice did not display the edema observed in full knockout animals (Garcia-Gonzalez et al., 2010;Hassane et al., 2011). This implies that the phenotypes that we observed may not reflect endothelial autonomous function. To investigate this further, we crossed the Tie2:Cre strain into a ROSA26r-LacZ background and examined Cre activity. Although active in blood vessels, we could not detect activity throughout subcutaneous lymphatic vessels ( Figure S6). Hence, previous work would not have uncovered function in these vessels. We generated Tie2:Cre-mediated knockout embryos for Pkd1 and found no subcutaneous lymphatic phenotype ( Figure S5). Therefore, we utilized Sox18:GFP-Cre-ErT2(GCE) as an additional endothelial CRE strain (Kartopawiro et al., 2014). We validated the use of Sox18:GCE on a Rosa26r-LacZ background, which demonstrated activity throughout the vasculature ( Figure S6). We also used an inducible tdTomato reporter to quantify activity in subcutaneous lymphatics by costaining with LEC markers NRP2 and PROX1. We found that induced Sox18:GCE was active in 58% of sprouting subcutaneous LECs at 13.5 dpc and frequently in clonal regions spanning whole vessels ( Figures 3J, 3K, and S6H-S6Q; Movie S5).
We generated induced Pkd1 endothelial cell knockout (iDECKO) embryos using this line. Pkd1 iDECKO embryos displayed either mild or no subcutaneous edema at 14.5 dpc (Figure 3C), with lymph sacs present but not containing blood ( Figures 3I and 3F). In the subcutaneous lymphatic vasculature, Pkd1 iDECKO embryos displayed similar dramatic defects to germline KO animals, if marginally milder on quantification (Figures (T) Quantification of the average width of lymphatic vessels (mm) across the whole skin in WT (n = 15 embryos), Pkd1 KO (n = 7 embryos), and Pkd1 iDECKO (n = 6 embryos) embryos. The average is shown of n = 773, n = 354, and n = 250 measurements, respectively, across leading lymphatic vessels from both sides of the midline at 14.5 dpc. (U) Quantification of nuclei/100 mm of vessel in WT (n = 12 embryos), Pkd1 KO (n = 3 embryos), and Pkd1 iDECKO (n = 6 embryos) (n = 5 representative leading edge vessels counted per embryo) at 14.5 dpc.
Error bars indicate SEM. See also Figures S5 and S6. 3N, 3Q, 3T, and 3U). We examined the blood vasculature of Pkd1 KO embryos. Although we saw defects in Pkd1 KO embryos, these were at the dorsal midline associated with edema and considered secondary to altered tissue architecture ( Figure S5). In contrast, Pkd1 iDECKO embryos displayed normal blood vasculature, including normal vessel width and branching ( Figure S5). Interestingly, Pkd1 iDECKO embryos did not show reduced LEC migration toward the midline ( Figure 3N). This would be expected for mutants in known pathways such as VEGFC/ VEGFR3.

PKD1 Regulates Sprouting and Cell-Cell Junctions In Vitro in Human LECs
Next, we examined the sprouting of human LECs in vitro in response to VEGFC using a spheroid outgrowth assay. Small interfering RNA (siRNA)-mediated knockdown of PKD1 in LECs resulted in a reduced number of cells within individual spheroid sprouts, with extensions exhibiting reduced length and abnormal morphology (Figures 4A-4H; Figure S7). The efficacy of knockdown with the siRNA mix was validated by qPCR, and the specificity was verified with an independent small hairpin RNA (shRNA) knockdown ( Figure S7). We examined the phenotype of LECs in cultured monolayers and observed a rapid change in morphology following PKD1 knockdown ( Figures 4I-4P). Stress fibers were disorganized in these cells (Figures 4I and 4M), and analysis of cell junctions revealed reduced VE-cadherin and b-catenin and disorganized junctions following knockdown ( Figures 4J, 4K, 4N, and 4O). ZO-1 localization at tight junctions was relatively unaffected in these assays, despite altered cell morphology, suggesting a level of selectivity to adherens junctions ( Figures 4L and 4P). The levels of VE-cadherin were not altered by western blot although b-catenin showed a mild reduction ( Figure S7), probably indicative of destabilized junctional complexes.
Pkd1 Regulates Polarity and Cell-Cell Junctions during Lymphatic Vessel Morphogenesis in Mice Pkd1 has been implicated in the regulation of polarity in epithelial cells and shown to regulate cellular convergent extension and polarity during kidney tubule morphogenesis through planar cell polarity (PCP) signaling (Castelli et al., 2013). PKD1 binds to PAR3 and aPKC as well as E-cadherin and b-catenin therefore being associated with both polarity and junctional components (Castelli et al., 2013;Lal et al., 2008;Roitbak et al., 2004). Recently, the PCP pathway has been shown to regulate junctional rearrangements in developing LECs, at least during valve morphogenesis (Tatin et al., 2013).
We examined cell polarity in sprouting embryonic lymphatic vessels. The Golgi apparatus orients toward the migration front relative to the nucleus in many cell types including LECs (Figures 5A and 5C), serving as an ideal readout for polarity. We quantified Golgi orientation in Pkd1 KO embryos and found it to be significantly randomized in 14.5 dpc lymphatic vessels compared with siblings ( Figures 5A-5D and 5G). Furthermore, this loss of polarity was associated with increased nucleus sphericity in mutant vessels, a previously described proxy for polarity and migratory behavior (Hä gerling et al., 2013) ( Figure 5H).
To determine the earliest defect, we performed detailed phenotypic analysis at 10.5 and 11.5 dpc. At 10.5 dpc, analysis of PROX1 expression indicated that cell migration from the cardinal vein and nuclear morphology was normal in mutants (Figures 5I,S6G,and S6H). However, at 11.5 dpc, although the blood vasculature was grossly normal ( Figure S5), mutant LECs at the sprouting vessel front displayed increased nucleus sphericity (decreased elipticity) compared with wild-type ( Figure 5J). We assessed Golgi orientation at these stages, but the direction of individual cell migration events was not regular, and the midline cannot be used as a direction of migration until later in development (data not shown). These early leading vessels also exhibited increased width and numbers of nuclei relative to vessel length similar to later Pkd1 KO vessels ( Figures S5I-S5J).
Finally, we investigated cell shape and the morphology of junctions within lymphatic vessels. At 14.5 dpc, VE-cadherin highlighted cell shape and showed that mutant cells failed to elongate along the plane of migration toward the midline compared with wild-type vessels (Figures 5K, 5L, 5O, 5P, and 5S). At the level of individual junctional morphology, both VEcadherin and b-catenin expression identified junctions that displayed immature morphology with irregular intracellular protrusions (arrowheads in Figures 5M, 5N, 5Q, and 5R). These phenotypes were only seen in phenotypically mutant vessels and not morphologically wild-type mutant vessels (data not shown; phenotypic variability shown in Figure 3). Quantification of the number of cells displaying immature junctions showed a significant phenotype from as early as 12.5 dpc ( Figures 5T-5V).

DISCUSSION
Our results, along with those of Outeda et al. (2014) published in this issue of Cell Reports demonstrate the surprising finding that Pkd1 is a regulator of lymphatic vessel development. In zebrafish, at the cellular level, Pkd1 regulates LEC migration out of the horizontal myoseptum but not initial sprouting from veins that is regulated by ccbe1/vegfc/vegfr3 (Hogan et al., 2009a(Hogan et al., , 2009bLe Guen et al., 2014;Villefranc et al., 2013). pkd1a is expressed in lymphatic precursor cells when they are actively migrating, consistent with the earliest cellular defects in the mutant.
It was important, given the highly studied nature of Pkd1, to ask if this function was conserved in mammals. In knockout mice, early specification and initial sprouting of LECs occurs normally. However, defects are seen in the morphology of migrating LECs at 11.5 dpc with morphological defects in the subcutaneous lymphatic network prominent by 14.5 dpc. This uniquely timed requirement is distinct from phenotypes in known pathways, suggesting that Pkd1 may act by an uncharacterized mechanism in LECs. Interestingly, the lymph sacs were blood filled in full knockout but not in endothelial knockout mice, which displayed only mild edema. This may be due to the staging of tamoxifen treatment to knockout Pkd1 function from 9.5 or 11.5 dpc, when lymph sacs are already establishing (Hä gerling et al., 2013). The observation that the lymphatic phenotype was reproduced by deletion with Sox18:GCE, active in LECs, but not Tie2:Cre, which we observed acts in BECs, suggests that Pkd1 functions in the LECs themselves during vessel morphogenesis.
Given the diverse functions of the protein, several hypotheses could explain the observed migration and morphogenesis defects. PKD1 has been previously reported to function at the primary cilium in endothelial cells (Nauli et al., 2008). However, we found lymphatic vessels developed normally in a ciliogenesis mutant (ift88; Huang and Schier, 2009), we saw no evidence for altered ciliogenesis in lyc1 mutants, and overexpression of a Pkd1a-YFP fusion protein, driven by the pkd1a promoter (BAC clone), did not lead to cilium enrichment ( Figure S8). Hence, we find no supportive evidence that Pkd1 in zebrafish lymphatic development functions at the cilium. Because Pkd1 can also localize to adherens junctions, desmosomal junctions, and intracellular organelles and has a number of binding partners, it has the potential to act at diverse locations.  The earliest consequences of loss of function are changes in cell morphology during morphogenesis, including altered polarity and adhesion. Cell polarity and adhesion are intimately associated and must be carefully regulated to control tissue morphogenesis. It is hard to determine which defect is primarily regulated by Pkd1. However, parallels can be drawn with recent findings in kidney tubule development where Pkd1 regulates cellular convergent extension during tube formation through the PCP pathway (Castelli et al., 2013). Although it will take further work to delineate the pathways modulated by Pkd1 in LECs, the finding of a crucial role in lymphatic vascular development is unexpected and serves as a unique entry point to understand lymphatic vascular morphogenesis.

Imaging and Analysis
For confocal and spinning disk imaging, embryos were mounted as previously described (Hogan et al., 2009b). Imaging was performed on a LSM Zeiss 510 NLO, META, or Zeiss 710 FCS confocal microscope with a 103, 203, and 403 dry objective and 633 oil objective. Images were analyzed with the Zen software, Biplane IMARIS, Photoshop, and ImageJ.

Additional animal procedures
For the induction of Cre-mediated recombination in embryos, 1.5 mg tamoxifen suspension in sunflower oil were injected intra-peritoneally into pregnant females at 9.5, 10.5 and 11.5 or 11.5, 12.5, 13.5 dpc in two separate regimes.

GOLPH4 processing -subtraction of background Golgi staining
The nucleus-Golgi angle relative to the dorsal midline was measured by first using the NRP2 expressing tissues as a mask to remove non-endothelial GOLPH4 staining during processing. The angle was subsequently measured between the perpendicular to the midline and the nucleus-Golgi orientation vectors in endothelial cells. Sphericity was measured in nuclei located within 150 µm of the leading edge, on both side of the midline in 14.5 dpc embryos. At earlier stages (10.5 and 11.5dpc), polarity and nuclear sphericity were assessed in cells at the lymphatic vascular migratory front only.

Statistical analysis
We used a Mann-Whitney rank sum t-test using Prism (GraphPad software), for all figures except in

Whole-mount in situ hybridization and immunochemistry
Primers used to amplify templates for riboprobe production are presented below. All probe template cDNAs were amplified from stage mixed WT cDNA by PCR and all PCR products, except for pkd1a, were subsequently cloned into pCS2+ plasmid (Turner and Weintraub, 1994). In situ hybridization was performed essentially as described in (Habeck et al., 2002;Thisse et al., 1993), with NBT/BCIP staining solution (Roche). Expression analysis and plasmid probes for flt4, dab2, couptfII, ephrinb2a, vegfc probes has been previously described in (Aranguren et al., 2011;Hogan et al., 2009;Lawson et al., 2001;Song et al., 2004;Thompson et al., 1998). Antibodies and primers used in this study are reported below.

Primers and antibody
The  Table 1.

Quantitative real time PCR analysis
Cell isolation, RNA extraction and cDNA synthesis: Zebrafish at 3 dpf and 5 dpf were deyolked by Quantitative PCR: qPCR was performed using an Applied Biosystems Viia 7 384 well qPCR machine, Applied Biosystems). Each qPCR reaction mixture contained 7.5 µl 2 x ABI SYBR green master mix (Applied Biosystems), 5ul cDNA (80-fold dilution), and 500 nM each primer to a final volume of 15 µl.
Amplification was performed in duplicate in 384 well plates (Applied Biosystems) with the following thermal cycling conditions: initial UDG treatment 50°C for 10 minutes, followed by 40 cycles of 15 s at 95°C, 60 s at 60°C. Control reactions included a no template control (NTC) and a no reverse transcriptase control (-RT). Dissociation analysis of the PCR products was performed by running a gradient from 60 to 95°C to confirm the presence of a single PCR product. The efficiency of PCR amplification was determined using LinReg PCR (Ruijter et al., 2009). The stability of several reference genes was analysed including hprt1, ef1a, rps29, rpl13 and β-actin. Reference gene stability was determined using GeNorm (Vandesompele et al., 2002). The geometric average of rps29, rpl13 and ef1a was used for normalisation of gene expression, except in Figure 2F where rpl13 was used, these genes being validated as the most stable across the sample population.
Primer design: Primers were designed using Primer blast (http://www.ncbi.nlm.nih.gov/tools/primerblast/) to have Tm of 60°C and to cross an exon-exon junction to avoid amplification of genomic DNA, whenever possible. Primers were used at a final concentration of 500 nM.

Human lymphatic endothelial cells
Human LECs were isolated and cultured as described (Norrmen et al., 2010). All experiments were performed with confluent cells.

siRNA transfection
Human LECs were transfected with 50 nM of control siRNA (Qiagen, AllStars Negative Control siRNA) or PKD1 siRNA (ThermoScientific, SmartPool containg 4 different siRNAs) using Lipofectamine RNAiMAX (Invitrogen). Photoshop softwares. The number of sprouts, the cumulated length of sprouts, the average of sprout length and the density of nuclei composing the sprouts were measured for each spheroid using ImageJ software. A two-tailed unpaired Student's t test was used to analyze the statistical significance of the difference between BSA-and VEGFC-treated, or Control and PKD1-knockdown groups.
Cells were imaged using Zeiss LSM 510 META scanning confocal microscope. The confocal images were processed using Bitplane IMARIS Suite 6.3.1 and Photoshop softwares.

Calcium drug treatment
Zebrafish embyos were exposed to DMSO, Nifedipine and Bayk8644 (Tocris Bioscience ref 1075 and 1544 respectively) at the concentration indicated in E3 media, from 24 hpf to 4 dpf. The treatment media was changed twice a day. Nifedipine and Bayk8644 were stored according to manufacturer recommendations.

Time
(A-C) Whole mount X-gal staining of β-gal in Tie2:Cre -/-, Rosa26R -/and Tie2:Cre + , Rosa26R + embryos shows that the Cre activity in the subcutaneous vasculature is strong in the blood vascular endothelium but not the lymphatic vascular endothelium at 14.5 dpc.

(F).
Whole mount X-gal staining indicates CRE activity, which is selective for endothelial cells, recapitulating the Sox18 expression pattern. Note that there is no co-stain to determine vessel identity in this experiment. individual co-stained nuclei (PROX1 and tdTOMATO (CAG-tdTOMATO localizes to nuclei)) allowing precise cell counting. Quantification of whole skin as described in Figure 3 and related text, using this approach revealed that n=663/1138 LECs were tdTOMATO, PROX1 and NRP2 positive (58.2%).
(P-Q) Alternative masking in Imaris using PROX1 expression as the mask identified the same colocalisation as using an NRP2 mask (inset are individual nuclei). Scale bar: 10 µm    Cre coding sequence gentoyping for the presence of Cre F 5'-CGAACGCACTGATTTCGACC-3' R