Phospholipase Cγ2 is necessary for separation of blood and lymphatic vasculature in mice

The lymphatic vasculature originates from the blood vasculature during development. Previous studies have demonstrated that the differentiation of lymphatic endothelial cells (LECs) is initiated by expression of the transcription factor prospero-related homeobox 1 (Prox1), in a subpopulation of venous endothelial cells (Wigle and Oliver, 1999). Vascular endothelial growth factor (VEGF) C, which is a ligand for vascular endothelial growth factor receptor (VEGFR) 2 and 3(Joukov et al., 1996), is not required for Prox1-induced LEC specification, but is necessary for lymphatic vessel formation (Karkkainen et al.,2004).

During later development, the lymphatic vasculature separates from the blood vasculature and acquires specialized structures. Previous studies have shown that mice lacking either the spleen tyrosine kinase (Syk) or Src-homology 2 (SH2) domain-containing leukocyte protein of 76 kDa (SLP76)exhibited blood-lymph shunts and their lymphatic vessels contained lymphatic vessel endothelial hyaluronan receptor 1-positive (Lyve1⁺)(Prevo et al., 2001) LECs and blood endothelial cells (BECs) (Abtahian et al., 2003). The blood-lymph separation may also be regulated by hematopoietic cell-derived circulating lymphatic endothelial progenitor cells(Sebzda et al., 2006). However, the precise roles of Syk and SLP76 in the separation process remain unknown.

Here, we have employed a genetic approach to facilitate further understanding of the vascular separation. Positional candidate cloning revealed that a spontaneous mutant mouse line exhibiting blood-filled lymphatic vessels carries a null mutation of the Plcg2 gene, which encodes phospholipase Cγ2 (PLCγ2). We also demonstrate that the blood-filled lymphatic vessels in PLCγ2-null mice were caused by the aberrant separation of blood and lymphatic vasculature during development. Analysis of the expression pattern of a Plcg2 reporter-knock-in and bone marrow reconstitution studies were performed to evaluate the separation process.

All mice were housed under pathogen-free conditions. C57BL/6J and CAST/Ei were purchased from CLEA Japan (Tokyo, Japan) and the Jackson Laboratory (Bar Harbor, ME, USA), respectively. The mutant mouse strain and 129/SvEv mice were kind gifts from Dr Motoya Katsuki (National Institute for Basic Science,Japan). The mutant mice were backcrossed eight times to 129/SvEv mice for genetic analysis. An FLP-deleter strain, FLP66(Takeuchi et al., 2002), was provided by RIKEN BRC (Tsukuba, Japan). All of the work with mice conformed to the guidelines approved by the Institutional Animal Care and Use Committee of the University of Tokyo.

Genomic DNA (0.1 μg) was used for PCR and the amplification conditions were: 94°C for 2 minutes, 30-35 cycles of 94°C, 60°C and 72°C for 1 minute each, with a final extension at 72°C for 7 minutes. PCR products were electrophoresed on 3-4% agarose gels in TBE buffer. Newly designed SSLP-PCR primer pairs located in the region between D8Mit48 and 120 are as follows: D8Ims9, forward 5′-CCACAGTATACCCACATAGATT-3′ and reverse 5′-AGCGGACTGGTGACAGCACA-3′; D8Ims10, forward 5′-CTCACTGAACCATCTCACCA-3′ and reverse,5′-AGGTGCCTGTGTACAATAGA-3′; D8Ims11, forward 5′-GATCTAGTGTAGTAGCAGCA-3′ and reverse 5′-TTCTGGCCTCTGTGAGAGTTTG-3′; D8Ims1, forward 5′-CCTCCATGGACACTGCACTC-3′ and reverse 5′-GTGAGTTCAGTGCCAGCCAG-3′; D8Ims2, forward 5′-TCTCACATTAAGTGCGTGCC-3′ and reverse 5′-AGGAGGAGTCGGATGGAAGC-3′; D8Ims18, forward 5′-ACACTACACTCAATGCACATG-3′ and reverse 5′-ACAATGATGGTCTTCAGAGC-3′; and D8Ims19, forward 5′-CAAGGTGGAGACTAAGAAGC-3′ and reverse 5′-CTGCTGCCACTTCATGTAAG-3′.

A BAC clone encompassing the Plcg2 gene RPCI23-308L22 was purchased from Invitrogen (Carlsbad, CA, USA). Part of the coding region of exon 2 was replaced with EGFP cDNA (Clontech/TAKARA Bio, Shiga, Japan)followed by a murine PGK polyadenylation signal sequence and an FRT-flanked PGK-gb2-neo cassette (Gene Bridges GmbH, Dresden, Germany) by homologous recombination in E. coli. A DNA fragment containing the modified exon 2 was subcloned into pUC-DT-A (a gift from Dr Takeshi Yagi, Osaka University,Japan) (Yagi et al., 1993). The linearized vector was electroporated into a Plcg2^+/alES cell line established from Plcg2^+/al blastocysts (T.I.,unpublished). All correctly targeted G418-resistant clones possessed the Plcg2^+/EGFP genotype. B6- or 129-congenic knock-in mice harbouring the neo cassette, except for embryos in Fig. 3D, were used in this study.

Specimens were fixed in PBS containing 4% paraformaldehyde (PFA) at 4°C overnight and 10 μm frozen sections were prepared. For EGFP immunostaining,trypsinization of sections was used for antigen retrieval. All sections were incubated with 3% H₂O₂ in PBS prior to immunostaining. Sections were incubated with a primary antibody, followed by incubation with the Histofine reagent (Nichirei Biosciences, Tokyo, Japan). Prior to detection of the primary/secondary antibody complexes, sections were incubated with a biotinylated antibody for double immunostaining. A Streptavidin/Biotin blocking kit (Vector laboratories, Burlingame, CA, USA) was used. Following detection of the first antibody complexes, sections were incubated with 3%H₂O₂ to quench the peroxidase activity of the Histofine reagent, and then incubated with streptavidin-HRP (NEN/Perkin-Elmer, Waltham,MA, USA). The TSA HRP detection system (NEN/Perkin-Elmer) and a DAB solution were used. Frozen sections of fresh tissues were acetone fixed and are shown in Fig. S4A (F4/80, CD11b, normal IgG) in the supplementary material. Paraffin wax-embedded sections (7 μm) are shown in Fig. 1 and Fig. S2C (in the supplementary material), and were antigen-retrieved for PLCγ2 immunostaining by boiling in the antigen-retrieval buffer (R&D systems,Minneapolis, MN, USA). Micrographs in figures are representative of 2-20 independent sections from two to six independent specimens. For whole-mount immunostaining, specimens were pre-fixed in 4% PFA and then fixed in a methanol/DMSO solution. After quenching with methanol containing 5%H₂O₂ and rehydration, immunostaining was performed. Micrographs in figures are representative of two independently stained specimens from more than two mice.

We verified that the auto-fluorescent EGFP signal or false-positive staining by normal IgG or isotype control antibodies did not affect the staining results. Images were acquired with an Olympus microscope (Olympus,Tokyo, Japan). Antibodies used were as follows: rat anti-mouse Lyve-1 (R&D Systems); rat anti-GFP (Nacalai-Tesque, Kyoto, Japan); biotinylated rat anti-mouse CD31 (BD-Pharmingen, Franklin Lakes, NJ, USA); biotinylated goat anti-mouse Lyve-1 (R&D Systems); biotinylated rat anti-mouse F4/80(eBioscience, San Diego, CA, USA); rat anti-mouse F4/80 (eBioscience); rat anti-mouse CD11b (eBioscience); and rabbit polyclonal anti-PLCγ2 (Santa Cruz Biotechnology, Santa Cruz, CA, USA).

Human umbilical vein ECs (HUVECs) and human dermal microvascular LECs(HdMLECs) from pooled donors (Lonza, Basel, Switzerland) were cultured using an EGM-2 MV bullet kit (Lonza) according to the manufacturer's protocol. Mouse mesenteric ECs were prepared and used for analysis as described previously(Yamaguchi et al., 2008). ECs and mouse primary embryonic fibroblasts were cultured in EBM-2 basal medium(Lonza) and Dulbecco's modified Eagle's medium (DMEM), respectively,containing 0.5% serum without supplemental growth factors for 16 hours. Recombinant human VEGF₁₆₅ (Peprotech EC, London, UK), rat VEGFC(R&D Systems) or platelet-derived growth factor-BB (PDGF-BB) (Peprotech EC) were added to the medium at a final concentration of 100 ng/ml. Cells were incubated for 10 minutes and harvested for analysis.

Cell lysates (20 μg) were resolved by SDS-PAGE and semi-dry-blotted onto PVDF membranes (Millipore, Billerica, MA, USA). Western blot analysis was performed using a rabbit primary antibody and HRP-conjugated anti-rabbit IgG. Antibody-labelled bands were visualized with an ECL detection system (GE Healthcare Bio-Sciences, Piscataway, NJ, USA) and X-ray film (Fujifilm, Tokyo,Japan). Proteins immunoprecipitated by anti-PLCγ2 were assayed using anti-phosphotyrosine (4G10), according to a previously described method(Yamaguchi et al., 2008). Antibodies used were as follows: rabbit polyclonal anti-PLCγ1,PLCγ2 and VEGFR2 (Santa Cruz Biotechnology); anti-phosphorylated PLCγ1 (Tyr 783), PLCγ2 (Tyr 759) and VEGFR2 (Tyr 1175) (Cell Signaling Technology, Danvers, MA, USA); and 4G10 (Upstate/Millipore,Billerica, MA, USA). Results shown in figures are representative of duplicate or triplicate experiments.

Bone marrow (BM) cells were obtained from 8- to 12-week-old Plcg2/EGFP knock-in or wild-type female mice, and were intravenously injected at a total number of 2.5-5.0×10⁶ cells in DMEM into recipient syngenic 8- to 12-week-old mice that had received a total body irradiation of 1200 rad prior to transplantation. Mice were examined 1-6 months post-transplantation, and more than 10 wild-type mice receiving Plcg2^+/+ or Plcg2^+/EGFP BM cells, 12 wild-type mice receiving PLCγ2-null BM cells and five PLCγ2-null mice (four Plcg2^EGFP/EGFP and one Plcg2^al/al mice) receiving Plcg2^+/+ or Plcg2^+/EGFP BM cells were used for analysis.

We identified spontaneous mouse mutants among offspring of a sibling pair of mice from a mixed genetic background of C57BL/6J and 129/SvEv. These mutants exhibited chylous ascites (Fig. 1A) and blood-filled lymphatic vessels in the intestine, heart,diaphragm and skin (Fig. 1B,C,data not shown for other tissues). Progeny tests indicated that the mutation is inherited in an autosomal recessive manner. We named the strain`abnormal lymphatics (al)'. The homozygous mutants(al/al mice) have blood-filled lymphatic vessels at mid-gestation,and die spontaneously by bleeding during development or in adulthood (data not shown).

In order to map a candidate gene causing the phenotype, we designed a strategy of inter-subspecific backcross mapping between the 129-congenic mutant mice (Mus musculus domesticus) and CAST/Ei mice (Mus musculus castaneus) (see Fig. S1 in the supplementary material). We obtained (129 +/al×CAST) F1 mice and then backcrossed them to 129 +/al mice, because al/al mice were not useful for backcrossing owing to reproductive abnormalities. We identified affected mice by the presence of blood-filled intestinal lymphatic vessels(Fig. 1B) and then performed PCR genotyping to find genomic regions with homozygous M. m. domesticus genotypes. Genetic mapping using SSLP markers that distinguish the two subspecies mapped the mutation to the distal half of chromosome 8,between D8Mit48 and the terminus (data not shown). Further mapping using SSLP markers that distinguish three inbred strains identified a ∼1 Mb non-recombinant B6-derived region between D8Ims9 and D8Mit120. High-resolution mapping narrowed the mutation to a ∼220 kb region between D8Ims10 and D8Ims18 (Fig. 1D). This region contains the Plcg2 gene and one uncharacterized gene homologous to the NAD(P)H steroid dehydrogenase-like (Nsdhl) gene. Sequencing of PCR-amplified cDNA and genomic DNA identified a single A-G substitution located in exon 2 of the Plcg2 gene(Fig. 1E). The mutation results in a translational stop at amino acid 54 which is a tryptophan in the N-terminal pleckstrin homology (PH) domain of PLCγ2. Western blot analysis confirmed the translational stop mutation(Fig. 1F). These data indicate that the mutation in Plcg2 results in a loss-of-function and leads to the blood-filled lymphatic vascular phenotype. We therefore renamed this allele Plcg2^al. A previous study reported that PLCγ2-null mice showed haemorrhaging during development and in adulthood(Wang et al., 2000), but did not report any lymphatic vascular abnormalities. Haemorrhaging may be caused by rupture of blood-filled fragile lymphatic vessels by mechanical stress or pressure of the blood flow.

To identify PLCγ2-expressing cells responsible for the phenotype, we generated Plcg2/EGFP knock-in mice by replacing exon 2 of Plcg2 with EGFP cDNA (see Fig. S2A in the supplementary material). Plcg2^EGFP/EGFP mice did not express PLCγ2 (see Fig. S2B,C in the supplementary material) and were phenotypically indistinguishable from Plcg2^al/al mice (see Fig. S2D,E in the supplementary material). We used the knock-in mice as PLCγ2-null mice for the analysis described below.

First, we characterized the blood-lymph shunt phenotype in detail. Developing lymphatic vessels of PLCγ2-null embryos became blood filled and dispersed peripherally during development (see Fig. S2D in the supplementary material). At E13.5, blood-filled lymph sacs consisted of both Lyve1⁺ LECs and CD31⁺, Lyve1^- BECs(Fig. 2, arrowheads), and the ECs of the lymph sacs remained close to those of the cardinal veins(Fig. 2B,E, arrows). These results were commonly observed in PLCγ2-null embryos, but not in wild-type embryos (six embryos for each genotype; Fig. 2A,D). In PLCγ2-null embryos, blood cells may remain in developing lymph sacs during vascular separation, or transmigrate where the endothelial sheets remain in contact between veins and lymph sacs (Fig. 2B,E). Fused vessels would allow blood to flow directly into the lymph sac (one out of two lymph sacs in one out of six embryos; Fig. 2C,F). Blood accumulation in developing lymph sacs/vessels may lead to lymph sac/vessel over-expansion and LEC attachment to developing blood vessels, followed by occasional fusion between blood and lymphatic vessels.

We next performed an immunohistochemical analysis using an anti-GFP antibody. Plcg2/EGFP was expressed in a variety of embryonic and adult tissues. In addition to F4/80⁺ monocytes/macrophages(Fig. 3A), and platelets(Fig. 3B), Plcg2/EGFP was expressed in a subset of ECs (Fig. 3C), but not in vascular smooth muscle cells (data not shown). Plcg2/EGFP was expressed predominantly in arterial ECs, rarely in venous ECs and not in LECs (Fig. 3D,E). By contrast, Western blot analysis showed that PLCγ2 was expressed in both BECs and LECs in vitro (see Fig. S3A in the supplementary material).

PLCγ1, a highly conserved homologue of PLCγ2, is required for VEGFA/VEGFR2 signalling (Liao et al.,2002) (Sakurai et al.,2005), and is phosphorylated in response to VEGFA and VEGFC (see Fig. S3B in the supplementary material). By contrast, PLCγ2 in ECs was not phosphorylated by VEGFs (see Fig. S3B,C in the supplementary material). It remains to be elucidated which growth factors activate PLCγ2 and how PLCγ2 functions in ECs. We examined the discrepancy of PLCγ2 expression in LECs using primary LECs from Plcg2^+/EGFPmice, and found that Plcg2/EGFP was expressed in a subpopulation of LECs (see Fig. S3D in the supplementary material). This expression pattern may be due to altered gene expression caused by changes in the EC microenvironment(Amatschek et al., 2007). These results suggest that PLCγ2 may not function in differentiated LECs that have become separated from blood vessels; however, we cannot exclude the possibility that transient expression of PLCγ2 in EC-lineage cells in a temporal and spatial manner is involved in the separation process.

As BM-derived cell participation in vascular separation has been suggested previously (Sebzda et al.,2006), we evaluated the BM-cell contribution to the phenotype of PLCγ2-null mice. PLCγ2-null BM cells were used to reconstitute the BM of lethally irradiated wild-type mice, and these mice developed blood-lymph shunts in the intestines (Fig. 4A), in which heterogeneous vessels consisting of CD31⁺BECs and Lyve-1⁺ LECs (Fig. 4B) were found. Accumulation of blood- and chyle-containing ascites was also observed (10 out of 12 mice), but intravenously injected fluorescein-conjugated isolectin-B4 bound to intestinal LECs (two mice; data not shown). Additionally, lymphatic vessels cast by intravenously injected fluorescein-conjugated gelatin (two mice; data not shown) confirmed that the blood-lymph shunt was due to connections between blood and lymphatic vessels.

We next examined whether wild-type BM cells could rescue PLCγ2-null mice from the blood-lymph shunt phenotype. One to 2 months after wild-type BM transplantation (three mice), blood-filled lymphatic vessels were distributed throughout the intestine, and one mouse had peritoneal haemorrhaging (data not shown). However, at 6 months (two mice), blood-lymph shunts were not observed in most of the intestine (Fig. 4A,d) and a remarkable lymphatic vascular re-organization had occurred (Fig. 4A,f), implying that the fused vessels could be repaired. However, blood-filled lymphatic vessels still remained in the heart (data not shown) and in a few areas of the intestine (Fig. 4A,e), and connections between the two vasculatures were still found in the intestine(Fig. 4B). These results indicate that BM-derived cells contribute to the vascular separation process,even though they were not able to completely rescue the phenotype.

In the intestine, BM-derived cells included many F4/80⁺CD11b⁺ monocytes/macrophages in the mucosa and submucosa (see Fig. S4A,B in the supplementary material) and a few ECs (see Fig. S4C in the supplementary material). Monocytes/macrophages, via a Syk/SLP76/PLCγ2 signalling pathway (Wilde and Watson,2001), may contribute to lymphangiogenesis(Cursiefen et al., 2004; Maruyama et al., 2005). Several reports have shown, or suggested, the presence of BM-derived LECs(Religa et al., 2005; Kerjaschki et al., 2006; Sebzda et al., 2006), although their importance in lymphangiogenesis has been debated(He et al., 2004). We did not detect PLCγ2-expressing LECs in vivo(Fig. 3D), but found that PLCγ2 expression was induced in LECs in vitro (see Fig. S3D in the supplementary material). Additionally, a few PLCγ2-null, EGFP-expressing ECs were found in PLCγ2-null BM-reconstituted mice (see Fig. S4C in the supplementary material). BM- or hematopoietic cell-derived LECs transiently expressing PLCγ2 during their differentiation may be also candidates for cells that mediate the vascular separation process.

Future studies using cell type-specific knockouts and transgenic animals will address the issues of which types of BM-derived cells are necessary for vascular separation, and how PLCγ2 is involved in the intracellular signalling which mediates the separation.