Abnormal embryonic lymphatic vessel development in Tie1 hypomorphic mice

The lymphatic vascular system is a blind-ended network of endothelial cell-lined vessels essential for the maintenance of tissue fluid balance, immune surveillance and absorption of fatty acids in the gut. The lymphatic vessels are also involved in the pathogenesis of diseases such as tumor metastasis, lymphedema and various inflammatory conditions. Despite central roles in both normal and disease physiology, our understanding of the development and molecular regulation of the lymphatic vasculature lags far behind that of the parallel blood vascular system (Oliver, 2004; Oliver and Alitalo, 2005).

Many details about the development of the lymphatic vasculature have been described only within the past decade, largely as a result of studies of gene-targeted mice (Oliver and Srinivasan, 2008). For mammals, the development of the lymphatic vessels in embryos is initiated when a subset of endothelial cells in the cardinal vein sprout to form the primary lymph sacs (Oliver and Detmar, 2002), confirming speculation of a venous origin made over a century ago (Sabin, 1909). In mice, the initiation of lymphatic differentiation is first discernible at embryonic day 10.5 (E10.5), when a subpopulation of endothelial cells (ECs) expressing Lyve1, Prox1 (Wigle and Oliver, 1999), Sox 18 (Francois et al., 2008; Hosking et al., 2009) and Vegfr3 are detected on one side of the anterior cardinal vein. These differentiating lymphatic endothelial cells (LECs) sprout, migrate and proliferate to form primary lymph sacs in the jugular region (Oliver, 2004). Subsequently, several lymph sacs are formed close to major veins in different regions of the embryo. The primary lymphatic vascular plexus undergoes remodeling and maturation to create a mature lymphatic network composed of large lymph vessels as well as an extensive lymphatic capillary network. Numerous signaling proteins have been identified as important in these later stages of the lymphatic development including ephrin B2 (Makinen et al., 2005), neuropilin 2 (Yuan et al., 2002), angiopoietin 2 (Gale et al., 2002; Dellinger et al., 2008), podoplanin (Schacht et al., 2003), integrin alpha 9 (Huang et al., 2000), and the transcription factors Foxc2 (Petrova et al., 2004), Net (Ayadi et al., 2001), Vezf1 (Kuhnert et al., 2005), adrenomedullin (Fritz-Six et al., 2008) and Aspp1 (Hirashima et al., 2008).

The orphan receptor tyrosine kinase (RTK) Tie1 shares a high degree of homology and is able to form heterodimers with Tie2, the receptor for the angiopoietins (Yancopoulos et al., 2000; Peters et al., 2004), and is known to play a major role in vascular development. Genetic studies in mice have demonstrated that Tie1 is required for development and maintenance of the vascular system as mice lacking Tie1 die in mid-gestation of hemorrhage and defective microvessel integrity (Puri et al., 1995; Sato et al., 1995). Expression of Tie1 is restricted to ECs and to some hematopoietic cell lineages (Partanen et al., 1992; Korhonen et al., 1994; Dumont et al., 1995; Hashiyama et al., 1996; Taichman et al., 2003; Yano et al., 1997). Interestingly, there is evidence that Tie1 is also expressed by the initial and collecting lymphatic vessels in adult mice (Iljin et al., 2002) and the first visible defect in the Tie1 mutants is edema (Sato et al., 1995). These observations led us to hypothesize that Tie1 might serve a unique function in the development or maintenance of the lymphatic vasculature.

In this study, we found that Tie1 was expressed in the Prox1-positive venous LEC progenitors and lymphatic vessels throughout embryonic and postnatal life. To circumvent early embryonic lethality observed in homozygous mutant animals, we generated mice with a conditional Tie1 allele that fortuitously resulted in hypomorphic expression of Tie1. We were able to take advantage of this Tie1 hypomorphic allele to demonstrate a unique role for Tie1 in lymphatic development that was not observed in the arterial or venous vasculature. Furthermore, the severity of the phenotypes observed correlated with the expression level of Tie1. Our studies show that Tie1 is required for the early stages of normal lymphangiogenesis and is also involved in the later remodeling and stabilization of lymphatic vessels in a dosage-dependent manner.

To generate Tie1^loxP/loxP mice, a Tie1 floxed targeting vector was constructed based on the 129-Sv mouse genomic fragment used by Puri et al. (Puri et al., 1995). A 940 bp HpaI-SacI fragment containing a Tie1 minimal promoter (Korhonen et al., 1995; Iljin et al., 2002) and exon 1 (containing the initial ATG codon) was inserted into the floxed KpnI-ClaI sites of the pDELBOY plasmid (pDELBOY-3X), which contains an Frt-site-flanked neomycin gene (see Fig. S1 in the supplementary material). The targeting vector was electroporated into 129 R1 embryonic stem (ES) cells (Nagy et al., 1993) and targeting was confirmed by Southern blotting with both 5′ and 3′ external probes and PCR with the primers 5′-ATGCCTGTTCTATTTATTTTTCCAG-3′ and 5′-TCGGGCGCGTTCAGAGTGGTAT-3′. Correctly targeted cells were then injected into C57/BL6 blastocysts and two separate clones were found to transmit the targeted allele through the germline. Both lines were maintained on a 129-Sv and C57/BL6 background and demonstrated identical phenotypes.

Hearts, diaphragm, limbs and head skin were collected from embryos and processed as previously described (Gale et al., 2002). The following primary antibodies were used: rat anti-mouse Pecam1 (Pharmingen, monoclonal MEC13.3), goat anti-mouse Vegfr3 antiserum (R&D Systems, #AF743), rabbit anti-Lyve1 (Upstate, #07-538) and Cy3-conjugated anti-α-smooth muscle actin (SMA; Sigma, C-6198).

For fluorescence immunostaining, the following secondary antibodies were used: Alexa Fluor 488-conjugated goat anti-rabbit (Molecular Probes, A-11008), Cy3-conjugated donkey anti-goat and Cy2-conjugated donkey anti-rat IgG (Jackson ImmunoResearch Laboratories, #705-165-147 and #712-225-150). Tissues were mounted in Vectashield (Vector Laboratories) and analyzed using a fluorescent microscope (Olympus) while images were captured using a Zeiss LSM 510 confocal microscope.

To visualize anti-Lyve1 staining with light microscopy, biotinylated goat anti-rabbit IgG (Vector Laboratories, BA-1000) and biotinylated rabbit anti-goat IgG (Vector Laboratories, BA-5000) secondary antibodies were used in horseradish peroxidase stainings with the Vectastain Kit (Vector Elite PK-6100) and DAB Kit (Vector Laboratories, SK-4100).

X-gal staining was performed on frozen sections of tissues fixed in 4% PFA/PBS from Tie1^+/lacZ embryos as previously described (Puri et al., 1995) and tissues were then immunostained for lymphatics with primary antibody goat anti-mouse Vegfr3 and fluorescently labeled secondary antibody Cy3-conjugated donkey anti-goat IgG (Jackson ImmunoResearch Laboratories, #705-165-147).

BrdU incorporation was assessed in embryos following intraperitoneal injection of pregnant females with BrdU. For immunohistochemistry, mouse anti-BrdU monoclonal antibody (clone G3G4, Developmental Studies Hybridoma Bank, University of Iowa, USA) and Alexa Fluor Cy2-conjugated donkey anti-mouse IgG (Jackson ImmunoResearch, #715-225-151) were used. Simultaneously, slides were also incubated with goat anti-mouse Vegfr3 antibody or rabbit anti-Prox1 and Alexa Fluor 594-conjugated donkey anti-goat IgG (Molecular Probes, A-11058) or Alexa Fluor 555 goat anti-rabbit IgG (Molecular Probes, A-21428). Lymphatic endothelial cell (EC) proliferation was also determined by labeling with Prox1 and Ki67 (BD Biosciences, 550609). Other primary antibodies used in tissue section immunohistochemistry were rat anti-mouse CD34 (eBioscience, 14-0341), rat anti-mouse Icam1 (eBioscience, 14-0542), rat anti-mouse endoglin (eBioscience, 14-1051) and rat anti-mouse VE-cadherin (BD Biosciences, 555289). The percentage of proliferative lymphatic ECs in the cardinal vein and jugular lymph sac areas of embryos at E11.5 to E13.5 was defined as the number of BrdU⁺/Prox1⁺ cells divided by the total number of Prox1⁺ cells in each field at similar level.

Apoptosis in lymphatic ECs was assed by TUNEL assay using ApopTag Plus Fluorescein in situ Cell Death Detection Kit (CHEMICON International, S7111). In addition, cleaved caspase 3 (Cell Signaling Technology, 9661) and Vegfr3 double staining was used to detect apoptotic LECs on frozen sections. Images were acquired on an Olympus fluorescent microscope and processed in Adobe Photoshop. Percent apoptotic cells was determined by blinded quantification of TUNEL-positive cells in the lymphatic or vascular endothelium divided by total Prox1 or DAPI-stained nuclei in each comparable field.

In situ hybridization was performed on paraffin sections of 6 μm with radiolabeled single-stranded RNA probes. Probes were 0.42 kb mouse Tie1 fragments corresponding to nucleotides 2221-2639 of NM_011587.2 obtained by RT-PCR on mouse E18.5 lung RNA.

RNA from the lungs of wild-type littermates, Tie1^+/neo and Tie1^neo/neo embryos at E18.5 was isolated using TRIzol Reagent (Invitrogen) with additional DNase treatment (Promega). cDNA was then generated using the SuperScript II Reverse Transcriptase Kit (Invitrogen). Quantitative PCR was performed on the LightCycler machine (Roche) using the LightCycler DNA Master SYBR Green I Kit (Roche) and rodent Gapdh control reagents (Applied Biosystems). Sequences for the mouse Tie1 primers have been published (Taichman et al., 2003). All assays were repeated at least twice and all samples were run in triplicate.

For northern analysis, RNA was hybridized with a probe based on the same 0.42 kb mouse Tie1 fragment used for in situ hybridization following gel electrophoresis and blots were stripped and rehybridized with a mouse glyceraldehyde-3-phosphate dehydrogenase (G3pdh) cDNA probe.

Western blotting was performed using standard protocols from homogenized E18.5 lungs using primary antibody (rabbit anti-Tie1, Santa Cruz Biotechnology) detected with secondary antibody (Alexa Fluor 488-conjugated goat anti-rabbit, Molecular Probes) and imaged with the Odyssey infrared scanner. Mouse β-actin was used as a control.

To visualize functional lymphatic vessels, FITC-dextran (Sigma, MW=2000 kDa, 8 mg/ml in PBS) was injected intradermally into the back of the embryonic forelimb at E17.5 and E18.5 as previously described (Hirashima et al., 2008). Lymphatic flow carrying FITC-dextran in embryos was analyzed by Olympus fluorescence microscopy. For quantitative analysis of lymphangiography, the total length of lymphatic vessels carrying dye in embryos at E17.5 and E18.5 was measured using ImageJ software based on the photos taken after 2 minutes of injection.

The relative lymph sac size was quantified as described previously (Fritz-Six et al., 2008) using both Lyve1-stained and Hematoxylin and Eosin (H&E)-stained paraffin sections. Briefly, transverse sections of the jugular region of mutant and wild-type embryos were imaged on an Olympus SZX16 dissecting microscope. Sections containing jugular lymph sacs that were matched for the same anteroposterior level were blindly measured using ImageJ software. In order to normalize for section variability, the area of the lymph sac was divided by the area of the adjacent jugular vein.

Quantitative analysis of dermal lymphatic parameters (vessel diameter, vessel density and total vessel area per field) of wild-type, Tie1^neo/neo and Tie1^–/– embryos at E13.5 to E17.5 was performed as described (François et al., 2008), based on Vegfr3/Lyve1 wholemount immunostaining.

Data are expressed as the means ± standard deviation. Statistical analysis was conducted using the two-tailed Student’s t-test. A P-value <0.05 was considered significant.

We examined Tie1 expression at E10.5-E11.5 by X-gal staining of Tie1^+/lacZ mice (Puri et al., 1995) and immunostaining with Prox1 antibody. The Tie1^+/lacZ mice carry a β-gal reporter cassette inserted into exon 1 of the Tie1 locus under the control of the Tie1 promoter. This insertion allowed us to use β-gal detection to precisely follow Tie1 expression throughout development. We detected that Tie1 is co-expressed with Prox1 in LECs at all stages examined. As seen in Fig. 1A-C, a few endothelial cells lining the wall of the anterior cardinal vein were Prox1-positive. Some endothelial cells budding from the anterior cardinal vein were also Prox1-positive, which were located toward the outer margin of the vein. These cells represented a subpopulation of endothelial cells – the LECs. Almost all of these Prox1-positive cells co-expressed Tie1 (X-gal positive). The rest of the X-gal-positive/Prox1-negative cells corresponded to ECs of the vascular system.

Next, we determined whether Tie1 is expressed in LECs at later stages. As the first detectable lymphatic vessels (jugular lymph sacs) in the mouse embryo can be morphologically distinguished from the blood vessels at E13.5, we examined Tie1 expression at E13.5 by X-gal staining of Tie1^+/lacZ mice. As seen in Fig. 1D-E, at E13.5, X-gal (Tie1) stains not only the arteries, veins and blood capillaries, but also lymphatics, as judged initially by their anatomical position: in the neck region, the major lymphatic vessels are formed in close proximity to the cardinal vein. To precisely determine whether these vessels were lymphatics, we performed immunolabeling with antibody against Vegfr3, which is largely restricted to LEC after mid-gestation (Kaipainen et al., 1995). We detected strong Vegfr3 expression in the jugular lymphatic vessels (Fig. 1E). Arteries did not express Vegfr3 but were clearly X-gal-positive. Veins were X-gal-positive but Vegfr3-negative. Therefore, Tie1 is expressed in the lymphatic endothelium during embryogenesis at E13.5. The expression of Tie1 in the lymphatic vessels at E13.5 was further confirmed by in situ hybridization (Fig. 1F).

To determine whether Tie1 expression continues in lymphatic vessels throughout development, we examined skin, intestine, mesentery and lung (Fig. 1G-J) from E17.5 embryos. As observed at earlier stages, X-gal staining in ECs colocalized with antibodies against Vegfr3 or Lyve1 (data not shown). Thus, Tie1 is expressed in the Prox1-positive venous LEC progenitors and lymphatic endothelia throughout embryonic development.

To circumvent the embryonic and perinatal lethality observed in homozygous-null mutant mice (Puri et al., 1995; Sato et al., 1995), we created a conditional allele of mouse Tie1 gene (see Fig. S1 in the supplementary material). Once the neomycin selection cassette was removed, the levels of Tiel expression in Tie1^loxP/loxP mice were indistinguishable from those of wild-type animals. When these mice were crossed with E2A-Cre transgenic mice, which express Cre recombinase in the germ line (Lakso et al., 1996), homozygous-deleted pups (Tie1^–/–) died of hemorrhage and severe edema in utero or at birth, displaying the same phenotypes of the conventional knockout mice (Sato et al., 1995).

Interestingly, when the neomycin selection cassette used in the initial targeting construct was not removed, the resultant Tie1^neo allele appeared to be hypomorphic. Tie1^+/neo mice were bred to obtain homozygous pups, and mice homozygous for this allele (Tie1^neo/neo) displayed a more variable degree of severity than detected in the null mutant animals. Before E16.5, the majority (99/108, 91.7%) of the hypomorphic homozygous embryos exhibited mild to severe edema without signs of hemorrhage (Fig. 2B,D). By contrast, all of the Tie1^–/– mutant embryos with the same background manifested severe edema and 16.7% (4/24) also displayed localized hemorrhages distributed throughout the body surface. In Tie1^–/– mutant embryos, the hemorrhagic defect was usually observed from E13.5 onward (Fig. 2C), occasionally even as early as E13.0 (data not shown). After E16.5, 34.6% (9/26) of hypomorphic embryos also developed hemorrhage but it was localized to the tip of the tail, and occasionally (4/26) also at the tips of the toes (Fig. 2E). By contrast, the vast majority (92.5%) of the null mutant Tie1^–/– embryos exhibited extensive hemorrhage from E16.5 (Fig. 2F) to E18.5 or embryonic demise (Fig. 2G). Rare hypomorphic homozygous embryos died before E18.5 and a few (7 mice from 28 litters) survived to adulthood.

To determine if Tie1 expression in the hypomorphic embryos was different from that in wild-type embryos, we performed northern blot analysis using embryonic lungs, a site of robust Tie1 expression during embryogenesis (Taichman et al., 2003). As seen in Fig. 2H, Tie1 mRNA level is reduced to about half of that detected in wild-type embryos, whereas only a weak band was detected in homozygous mutant lungs. We measured Tie1 mRNA by quantitative PCR analysis of the same samples, which showed that the Tie1 mRNA expression in the E18.5 lungs of Tie1^+/neo and Tie1^neo/neo embryos was reduced by 50.4% and 79.4%, respectively (Fig. 2I). Consistently, western blot analysis showed that Tie1 protein levels in the E18.5 lungs of Tie1^+/neo and Tie1^neo/neo embryos were reduced by 43.8% and 76.0%, respectively (Fig. 2J).

We next examined sequential stages of lymphatic development in wild-type, Tie1-hypomorphic and Tie1-null mutant embryos. Histology of Tie1^neo/neo and Tie1^–/– embryos confirmed the presence of extensive interstitial edema (Fig. 3C,E). Lymph sacs were identified in wild-type, Tie1^neo/neo and Tie1^–/– embryos at E13.5 by immunohistochemical recognition of the lymphatic-specific makers Lyve1 (Fig. 3B,D,F) and Vegfr3 (data not shown), indicating normal lymphatic differentiation from venous vasculature. However, we observed that the jugular lymph sacs of Tie1^neo/neo and Tie1^–/– embryos appeared strikingly larger than those of litter-matched wild-type embryos. Computerized morphometry was used to calculate lymph sac area. At E13.5, the jugular lymphatic sacs in the wild-type embryos are located lateral to the internal jugular vein and dilate to a maximal diameter of approximately 2-3 times the size of the diameter of the internal jugular vein (Fig. 3A,B) (Gittenberger-De Groot et al., 2004). In Tie1^neo/neo and Tie1^–/– embryos at E13.5, the maximal diameter of the jugular lymphatic sacs was approximately 10 times the size of the diameter of the internal jugular vein (Fig. 3D,F,G). This phenotype persists until at least E15.5 (data not shown).

To explore why the lymph sacs were dilated, we used a BrdU incorporation assay to assess proliferating cells in both lymph sacs and adjoining jugular veins. At E13.5, we observed no difference in the percentage of proliferative Prox1-positive ECs in lymphatic sacs (27.1±4.9% versus 25.6±4.1%) or jugular veins (data not shown) between the hypomorphic and wild-type embryos. However, when embryos at E11.5 and at E12.5 were compared, the percentage of proliferative Prox1-positive cells in Tie1^neo/neo embryos at E11.5 and E12.5 was significantly higher than that in wild-type littermates (P<0.05), by 17.9% and 22.2%, respectively (Fig. 4). This increase of proliferative LECs was confirmed by co-labeling of adjacent sections with Prox1 and Ki67 (Fig. 4A,B).

We then quantified the total number of LECs in the cardinal vein and jugular lymph sac areas of embryos at sequential stages of lymphatic development. At E10.5, the number of Prox1-positive LECs in Tie1^neo/neo (Fig. 5B,J) and Tie1^–/– (Fig. 5C,J) embryos were increased by 23.2% and 29.1%, respectively (Fig. 5A). Similarly, at E11.5, the number of LECs in Tie1^neo/neo (Fig. 5E,J) and Tie1^–/– (Fig. 5F,J) embryos were increased by 35.0% and 38.9%, respectively, and at E12.5, the number of LECs in Tie1^neo/neo (Fig. 5H,J) and Tie1^–/– (Fig. 5I,J) embryos were increased by 44.2% and 53.5%, respectively. These results suggest that an increased production of LECs at the initiation of lymphangiogenesis contributes to the dilated lymph sacs seen in the Tie1 hypomorphic and mutant mice.

To evaluate whether Tie1 attenuation resulted in a global defect in endothelial differentiation, we used dual immunoflourescence at E11.5 (Fig. 5D-F) and E12.5 (see Fig. S2 in the supplementary material) with specific vascular EC markers – Pecam1, CD34, Icam1, endoglin or VE-cadherin and Prox1. We observed no difference among wild-type, Tie1^neo/neo and Tie1^–/– embryos in the expression of these five markers in vascular endothelium or at the dorsal part of the cardinal vein, the site where lymphatic EC differentiation occurs during embryogenesis. Thus, the abnormal dilated lymph sac defects in Tie1^neo/neo and Tie1^–/– embryos could not be ascribed to a primary endothelial defect.

To determine whether the abnormal lymphatic patterning of Tie1^neo/neo mice could also be associated with abnormal EC apoptosis, we used TUNEL detection in conjunction with Vegfr3 labeling and DAPI nuclear localization to identify apoptosis. We detected no significant difference in the percentage of apoptotic LECs in the wall of the cardinal vein and jugular lymph sac areas between wild-type (E11.5, 3.6±0.5%; E12.5, 3.2±0.8%) and Tie1^neo/neo (E11.5, 3.4±0.9%; E12.5, 2.9±1.1%) embryos (see Fig. S3 in the supplementary material). However, at E13.5, we identified an increase in apoptosis in LECs of hypomorphic embryos. Only 2.5±0.6% TUNEL-positive cells were detected in the lymphatic endothelium of the wild-type jugular lymphatic sacs (Fig. 6A-C,G), whereas we observed 5.7±0.9% TUNEL-positive LECs in the corresponding area of hypomorphic embryos (Fig. 6D-F,G). Interestingly, almost no EC apoptosis was seen in the jugular vein or arteries of wild-type or hypomorphic embryos. Taken together, these results demonstrate that reduction of Tie1 during embryonic development leads to a specific increase in lymphatic, but not venous, EC apoptosis at later stages of lymphangeogenesis.

We next performed wholemount immunohistochemistry to examine the development of dermal lymphatic vasculature in Tie1^neo/neo embryos at E13.5. In wild-type embryos, a few, but clear, Vegfr3-positive lymphatic vessels (Fig. 7A) and an extensive network of lymphatic capillaries (Fig. 7D) was detected in the head dermis near the developing ear and in limb skin. In the Tie1^neo/neo embryos, however, the lymphatic capillaries in the corresponding regions were dilated and disorganized (Fig. 7B,E). The lymphatic defects in the null mutant embryos (Fig. 7C,F) were more severe compared with those in the hypomorphic embryos. Importantly, wholemount staining of E13.5 forelimb skin for Pecam1 revealed a relatively normal blood vessel pattern in the hypomorphic embryos and only mildly dilated blood vessels in the null embryos (Fig. 7G-I), indicating a preferential effect on lymphatic vessels. These results suggested that the severity of lymphatic defects in Tie1 mutant embryos is dependent on the dosage of Tie1 and that the development of lymphatic vasculature is more sensitive to Tie1 reduction than the development of blood vasculature.

To further investigate network formation of blood vessels and lymphatic vessels, we performed wholemount double-fluorescence confocal microscopy of embryonic dorsal skin at later stages of development. As seen in Fig. S4 in the supplementary material, at E14.5, the Vegfr3-positive lymphatic capillary network was detected in wild-type control embryos, whereas lymphatic vessels were disorganized and mildly dilated in Tie1^neo/neo embryos; at E15.5, lymphatic vessels in Tie1^neo/neo embryos were dilated and lymphatic vessels started to regress in some areas, and compared with the phenotype in Tie1^neo/neo embryos, the lymphatic defects in Tie1^–/– embryos were even more severe. This was confirmed by Lyve1 immunoreactivity of skin from the three genotypes at E15.5 (see Fig. S5 in the supplementary material).

By E16.5, the severity of dermal lymphatic defects in Tie1 hypomorphic embryos was even more evident: Vegfr3-positive lymphatic vessels were dramatically dilated in some areas or severely disrupted in other areas (see Fig. S6A-C in the supplementary material). At E17.5 (Fig. 9; see Fig. S6D-F in the supplementary material), the Lyve1-positive lymphatic networks were dense and well organized in wild-type embryos but were markedly dilated, disorganized and irregular in Tie1 hypomorphic embryo lymphatics. In addition, most of the wild-type lymphatic capillaries were interconnected and only a few blind beginning lymphatic capillaries were detected. By contrast, the formation of lymphatic capillary networks was impaired in the head skin of the hypomorphic mice. The skin in the hypomorphic embryos exhibited incomplete network patterning with an increased number of blind beginnings of lymphatics. Some regions were largely bereft of lymphatics (Fig. 8B,F; see Fig. S5 and Fig. S6 in the supplementary material), indicating regression of the developing lymphatic vessels.

To quantify the effect of Tie1 attenuation on lymphatic vessel profiles, we measured lymphatic vessel area, density and diameter (Fig. 8M-O). These studies documented an increase in vessel diameter in the hypomorphic and mutant embryos at all developmental time points examined. The dilated dermal lymphatics seen in the Tie1 hypomorphic and mutant mice could be ascribed to the abnormally high production of LECs early in development, rather than an increase in LEC size. The total number of LECs (Prox1-positive nuclei) in each dilated dermal lymphatic vessel was higher in Tie1^neo/neo embryos than in wild-type embryos. This increase in LEC number was associated with an increase in lymphatic vessel area such that total LEC density (Prox1-positive nuclei/lymphatic vascular area) was not significantly altered in mutant embryos (see Fig. S7 in the supplementary material). However, density of lymphatic vessels (lymphatic vessel area/field) in the hypomorphic and mutant embryos was reduced from E15.5, indicating vessel regression. The total vessel area in the hypomorphic and mutant embryos was significantly higher owing to vessel dilation at E13.5, but started to decrease compared with wild-type controls from E15.5 owing to vessel disruption or regression. Similar to the jugular lymphatic sacs, we also detected accentuated Vegfr3-positive EC apoptosis in the dermis of the hypomorphic embryos at E15.5 and at E17.5 with both TUNEL detection and antibodies against cleaved caspase 3, with minimal Vegfr3-positive EC apoptosis detected in wild-type embryos (see Fig. S8 in the supplementary material).

By contrast, blood vascular patterning was relatively unaffected in the hypomorphic embryos at E14.5, E15.5 (data not shown), E16.5 (see Fig. S6A-C in the supplementary material) or E17.5 (Fig. 8C-L). Thus, attenuation in Tie1 was sufficient to alter the patterning of the lymphatic vasculature without affecting the non-lymphatic vasculature, suggesting that lymphatic dysfunction contributes to the edema seen in the Tie1 hypomorphic and mutant mice. This finding also supports the hypothesis that developing lymphatics are more sensitive to Tie1 reduction than the arterial or venous vasculature.

To assess whether structural defects seen in dermal lymphatic vessels lead to impaired lymphatic drainage, we evaluated lymphatic function by injecting high-molecular-weight FITC-dextran into embryonic limbs. Lymphangiography showed that dye uptake was rapid in the collecting lymphatic vessels and network of capillaries in wild-type mice at E17.5 (Fig. 9A). However, the transport function of lymphatic vessels in the hypomorphic embryos was impaired, consistent with the severity of the subcutaneous edema defect. As seen in Fig. 9B, FITC-dextran labeled only a few disorganized lymphatic capillaries in the hypomorphic embryo that had severe edema and localized hemorrhages at E17.5, indicating attenuation and/or poor function of the lymphatic capillary network. We obtained similar results using wild-type and hypomorphic embryos at E18.5 (Fig. 9C).

It is known that lymphatic capillaries progressively cover the heart surface and the diaphragm from E15 onwards (Yuan et al., 2002). Wholemount staining of these hearts with Lyve1 or Vegfr3 antibody revealed regression of the developing lymphatic vessels at the surface of the heart in the hypomorphic embryos compared with control littermates (see Fig. S9A-F in the supplementary material), which was first observed at E16.5. The epicardial lymphatic vessels in the hypomorphic embryos formed a primitive continuous network but were thinner than those in wild-type control embryos. In addition, the lymphatic vessels of the mutant embryos had begun to regress in some areas. At E17.5, the entire lymphatic network was disrupted in the hypomorphic embryos and by E18.5, only a few single disrupted lymphatic vessels were detected on the surface of the heart. In the central portion of the diaphragm at E16.5, some of the larger collecting lymphatics in the hypomorphic embryos were dilated and disorganized compared with wild-type embryos (see Fig. S9G-L in the supplementary material). By E17.5 and E18.5, there was a striking reduction in the number of lymphatics in the whole diaphragm of the hypomorphic embryos. Furthermore, all of the lymphatics were much thinner than those in wild-type control embryos and the network was disrupted, indicating regression of the developing lymphatics over the entire diaphragmatic surface. Taken together, the Tie1^neo/neo mice demonstrated abnormally patterned lymphatic vessels in all internal organs examined.

Genetic studies in mice have demonstrated that Tie1 is required for development and maintenance of the vascular system. Deletion of Tie1 leads to embryonic lethality due to edema, hemorrhage and microvessel rupture (Puri et al., 1995; Sato et al., 1995). Interestingly, the timing of previously reported embryonic demise at E13.5 correlated with the earliest time when lymphatic vessels can be morphologically distinguished from the blood vessels. The previous report (Sato et al., 1995) and our study showed that the first visible defect in the Tie1 mutants is dorsal subcutaneous edema from E13.5 onward. In this report, we have documented that Tie1 is expressed in LEC progenitors at E10.5 and E11.5 before lymphatic vessels are formed. This expression of Tie1 in the lymphatic endothelium is maintained throughout embryonic development and persisted in the postnatal animal. The early expression in the lymphatic endothelium suggests a unique role for Tie1 in the development and function of the lymphatic vessels.

The intronic insertion of a neo cassette causes aberrant splicing of Tie1 transcripts, reducing the amount of mRNA encoding functional Tie1 protein to ∼20% of normal levels. This Tie1 hypomorphic allele enabled us to demonstrate that a dosage-dependent reduction of Tie1 results in progressive defects in lymphatic patterning not seen in other EC populations. Tie1 attenuation resulted in significantly enlarged jugular lymphatic vessels. Consistent with structural defects seen in the lymphatics, lymphatic function was also impaired in hypomorphic embryonic skin. We detected no apparent lymphatic defects in the heterozygous Tie1^+/neo embryos at any developmental stage.

Interestingly, although the Tie1 hypomorphic embryos display severe generalized edema, there was almost no apparent hemorrhage during the entire period of embryonic development. A few embryos developed hemorrhage at the tips of the tail or digits and this occurred at a much later time point than the diffuse hemorrhage observed in the complete null embryos. In addition, although lymphatic defects were obvious in most of the hypomorphic embryos, the pattern of blood vessels appeared relatively normal. In fact, when we examined expression of the specific blood vascular EC markers in the major blood vessels and in particular at the dorsal part of the cardinal vein, the site where lymphatic EC differentiation occurs during the initiation of lymphangiogenesis, we could detect no differences in the pattern of expression between wild-type and hypomorphic mutant embryos. These results further suggest that the development of lymphatic vasculature is more sensitive to Tie1 reduction than the development of blood vasculature.

It is commonly thought that most of the lymphovascular system in mice is of venous origin (Oliver, 2004; Oliver and Alitalo, 2005). However, recent data argues for a dual origin of LECs in the mouse with incorporation of mesenchymal lymphangioblast-derived cells into the lining of lymph vessels during development providing a minor contribution to the lymphatics (Buttler et al., 2006; Buttler et al., 2008). Although our work does not directly address whether all lymphatics are of venous origin or whether there is an additional contribution from a mesenchymal cell population, the fact that we observed similar abnormalities in both the central lymphatics as well as in the dermis of the embryo would suggest that Tie1 is required for the normal development of both the central lymphatics and the peripheral lymphatics.

Jugular lymphatic sac dilation was a prominent feature in the Tie1 mutant embryos. Attenuation of Tie1 leads to an increased LEC production before mid-gestation in the Tie1 hypomorphic and mutant mice. Consistently, we detected significantly more proliferative LECs in the cardinal vein and jugular lymph sac areas of Tie1 hypomorphic and mutant embryos at E11.5 and E12.5, when specification of lymphatic vasculature occurs (Francois et al., 2008). Although the proliferation rate of LECs in jugular lymphatic sacs of Tie1 hypomorphic and mutant embryos at E13.5 returns to normal, the absolute total number of proliferative LECs is still higher than that in wild-type embryos because the absolute number of LECs is higher (Fig. 5J).

This early period of increased LEC proliferation was followed by an accentuation of apoptosis in the latter stages of lymphatic vascular development. We found normal LEC apoptosis in Tie1 hypomorphic and mutant embryos at early stages (E11.5 and E12.5) but a selective increase in LEC apoptosis from E13.5. This is consistent with the observed regression of the already-formed developing lymphatics and the previous report that Tie1 inhibits apoptosis and is required for EC survival (Kontos, et al., 2002). From these findings, we conclude that attenuation of Tie1 in lymphatics results in dilation due to an elevated production of LECs during specification of lymphatic vasculature, and regression after mid-gestation due to increased apoptosis of LECs. This process leads to a disorganized and disrupted lymphatic network.

The lymphatic defects of Tie1 mutant mice are reminiscent of the lymphatic phenotypes of the Tie2 ligand, Ang2, mutant mice, although the latter exhibits only dilated lymphatic vessels without regression and with no defects in the size of jugular lymphatic sacs (Gale et al., 2002; Dellinger et al., 2008). Ang2 is involved in the remodeling and stabilization of lymphatic vessels. Ang2^–/– mice exhibit defects in the remodeling of the blood vasculature, but surprisingly, even more severe defects of the lymphatic system. Ang2^–/– mice display chylous ascites, peripheral lymphedema and hypoplasia of the lymphatic vasculature (Gale et al., 2002). Lymphatic vessels in Ang2^–/– mice fail to mature and do not exhibit a collecting vessel phenotype. Furthermore, dermal lymphatic vessels in Ang2^–/– pups prematurely recruit smooth muscle cells and do not undergo proper postnatal remodeling (Dellinger et al., 2008).

To date, the Tie1 receptor remains primarily an orphan receptor, as no ligand for Tie1 has been identified and only ligand-independent signal transduction pathways have been characterized (McCarthy et al., 1999; Yabkowitz et al., 1999). However, Tie1 phosphorylation can be induced by overexpression of multiple angiopoietin proteins (Ang1 and Ang4) in vitro and activation is amplified via association with Tie2 (Saharinen et al., 2005). Tie1 and Tie2 receptor heterodimers had been shown to exist in cultured ECs (Marron et al., 2000) but the cellular function of heterodimerization remains to be determined. As Tie1 is expressed in the endothelium of the same lymphatics as Ang2 during embryogenesis and enhanced Tie1 mRNA expression occurs together with Ang2 mRNA in the adults (Gale et al., 2002; Iljin et al., 2002), it is tempting to speculate that Tie1 might be modulating Ang2 function via interaction with the Tie2 receptor in the lymphatic ECs as has been documented in cultured ECs (Yuan et al., 2007) and endothelial progenitor cells (Kim et al., 2006). Interestingly, investigators have recently shown that Vegf can cause a 4-fold increase in phosphorylation of Tie2 that is independent of angiopoietin expression but dependent on proteolytic cleavage of Tie1 and subsequent transphosphorylation of Tie2, further supporting the possibility of Tie1 modulation of Tie2 signaling (Singh et al., 2009).

However, the expression of Tie2 and its role in LEC development has not been well defined. Adult Lyve1-positive lymphatic vessels have previously been shown to express Tie2 by immunofluorescence labeling of mouse ear skin and the small intestine using an antibody against Tie2 (Morisada et al., 2005; Tammela et al., 2005). However, other groups failed to detect GFP expression by lymphatic vessels in the adult (Dellinger et al., 2008) or embryos (Srinivasan et al., 2007) from Tie2-GFP transgenic mice using either immunohistochemical (GFP) analyses or in situ hybridization. One possible explanation for this discrepancy is that the regulatory elements controlling the expression of Tie2 by lymphatic vessels are not included in the Tie2-GFP transgenic construct. Further delineation of the expression pattern of Tie2 should resolve this discrepancy.

In summary, it is now evident that the Tie receptors and their ligands demonstrate context-dependent regulation of vascular remodeling (Eklund et al., 2006) and our work suggests a unique dosage-dependent role for Tie1 in lymphatic ontogeny. Whether Tie1 signals independently or in association with Tie2 is a focus of ongoing investigation. However, the development of a hypomorphic, conditional Tie1 allele in combination with the available lymphatic-specific Cre deletor lines (Srinivasan et al., 2007) should provide the necessary reagents for the temporal- and lymphatic-specific deletion required for further investigation.

We thank Dr Lawrence Price (Vanderbilt University) for providing the antibodies against Prox1 and valuable discussions and Dr Christopher Brown (Vanderbilt University) for critical reading of the manuscript. This work was supported by NIH grants R01 HL086964 (S.B.) and DK038517 and a grant-in-aid from the March of Dimes. Deposited in PMC for release after 12 months.