Vascular Endothelial Growth Factor D Is Dispensable for Development of the Lymphatic System

Vascular endothelial growth factor receptor 3 (Vegfr-3) is atyrosine kinase that is expressed on the lymphatic endotheliumand that signals for the growth of the lymphatic vessels (lymphangiogenesis).Vegf-d, a secreted glycoprotein, is one of two known activatingligands for Vegfr-3, the other being Vegf-c. Vegf-d stimulateslymphangiogenesis in tissues and tumors; however, its role inembryonic development was previously unknown. Here we reportthe generation and analysis of mutant mice deficient for Vegf-d.Vegf-d-deficient mice were healthy and fertile, had normal bodymass, and displayed no pathologic changes consistent with adefect in lymphatic function. The lungs, sites of strong Vegf-dgene expression during embryogenesis in wild-type mice, werenormal in Vegf-d-deficient mice with respect to tissue massand morphology, except that the abundance of the lymphaticsadjacent to bronchioles was slightly reduced. Dye uptake experimentsindicated that large lymphatics under the skin were presentin normal locations and were functional. Smaller dermal lymphaticswere similar in number, location, and function to those in wild-typecontrols. The lack of a profound lymphatic phenotype in Vegf-d-deficientmice suggests that Vegf-d does not play a major role in lymphaticdevelopment or that Vegf-c or another, as-yet-unknown activatingVegfr-3 ligand can compensate for Vegf-d during development.

INTRODUCTION

Top

Introduction

Vessels of the lymphatic system are highly permeable and specializedfor the uptake of fluid and macromolecules from the interstitiumand their return to the venous circulation (33). Consequently,the lymphatic vasculature is critical for the maintenance oftissue fluid volume (for a review, see reference 8). It alsoplays an important role in immune responses, since leukocytetraffic via the lymphatic vessels to lymph nodes is essentialfor antigen presentation, and in cancer biology, since tumorcells can spread via the lymphatics to establish distant metastases(for reviews, see references 40 and 41). Recent studies withanimal models suggested that the growth of the lymphatics (lymphangiogenesis)in and around solid tumors can facilitate the metastatic spreadof tumor cells via the lymphatics (21, 29, 39, 42). Lymphedemais another clinical condition in which the lymphatics play acritical role, as inadequate lymphatic drainage of tissue leadsto the accumulation of lymph and subsequent swelling of theaffected area, a characteristic of this disease (for a review,see reference 8). Elucidation of the molecular mechanisms controllinglymphatic vessel development will underpin approaches to stimulatingor inhibiting lymphangiogenesis in clinical contexts, such aslymphedema or cancer, respectively.

Vascular endothelial growth factor receptor 3 (Vegfr-3), a cellsurface receptor tyrosine kinase (also known as Flt4), is expressedon lymphatic endothelial cells in adult tissues (18, 27), andits activation is sufficient to stimulate lymphangiogenesis(45). The two known activating ligands for Vegfr-3 are Vegf-c(17) and Vegf-d (also known as c-fos-induced growth factor [Figf])(1, 35), secreted homodimeric glycoproteins that are closelyrelated in structure. In humans, both VEGF-C and VEGF-D activateVEGFR-3 and another receptor, VEGFR-2 (1, 17), which is implicatedin signaling for angiogenesis (32); in mice, Vegf-c activatesboth receptors (23), but Vegf-d activates only Vegfr-3 (6).Thus, the receptor binding specificity of Vegf-d suggests thatit may be lymphangiogenic but not angiogenic in the mouse. Indeed,the ectopic expression of VEGF-D in the skin of transgenic miceunder the control of the keratin 14 gene promoter induced thegrowth of lymphatic vessels in the dermis, whereas blood vesselswere unaffected (45). In addition, the expression of VEGF-Din a mouse tumor model stimulated lymphangiogenesis in solidtumors; this process promoted the metastatic spread of tumorcells via the lymphatics but could be blocked by a neutralizingVEGF-D antibody (2, 3, 42). The importance of Vegf-c and Vegf-dwas further illustrated by the finding that the expression ofa soluble form of VEGFR-3 that acts to sequester both growthfactors in the skin of transgenic mice caused impaired fetallymphatic development and a lymphedema-like phenotype (28).

During early embryonic development, Vegfr-3 is expressed widelyon endothelial cells of blood vessels but subsequently becomesrestricted to venous endothelium, at a stage prior to the emergenceof lymphatic vessels, and then becomes restricted to the endotheliumof lymphatic vessels (18, 27). A key role for VEGFR-3 in lymphaticdevelopment was demonstrated by the identification of mutationsin VEGFR-3 in several cases of human congenital lymphedema (19).However, mutant mice deficient for Vegfr-3, generated by homologousrecombination in embryonic stem (ES) cells, were not informativewith respect to lymphatic vessel development or function, asthese mice died at embryonic day 9.5 (E9.5) due to defectiveremodeling and maturation of blood vessels into larger vessels,prior to the emergence of lymphatic vessels from the veins (11).Further studies showed that the Vegfr-3 ligand Vegf-c is requiredfor the initial steps of lymphatic development (20). In Vegf-c^–/–mice, endothelial cells commit to the lymphatic lineage butdo not sprout to form lymphatic vessels. The lack of lymphaticvessels results in prenatal death due to fluid accumulationin tissues. Vegf-c⁺^/^– mice develop cutaneous lymphatichypoplasia and lymphedema, demonstrating that both Vegf-c allelesare required for normal lymphatic development (20).

To assess the role of Vegf-d in the development of the lymphaticsystem, we have generated mutant mice that lack Vegf-d by homologousrecombination in ES cells. Vegf-d-deficient mice are viableand have a functional lymphatic system. The lymphatic vesselsin these mice develop normally and are able to transport fluidto draining lymph nodes. Furthermore, Vegf-d-deficient micedo not display evidence of perturbed lymphatic vessel function.The lack of a profound phenotype associated with Vegf-d deficiencysuggests that Vegf-d does not play an important role in lymphaticdevelopment or that there is compensation for Vegf-d functionduring lymphatic development by Vegf-c or another, as-yet-unknownligand for Vegfr-3.

MATERIALS AND METHODS

Top

Materials and Methods

Construction of the Vegf-d gene-targeting vector. A genomic clone, denoted M18, containing the first and secondcoding exons of the mouse Vegf-d gene was isolated as previouslydescribed (7). The short arm of homology (ShA) was derived fromM18 and consisted of a 2.5-kb region of the Vegf-d gene immediatelyupstream of the translation initiation codon. ShA was insertedinto pZero-2 (Invitrogen) downstream of the cDNA encoding theA fragment of the diphtheria toxin (DTA) gene under the controlof the thymidine kinase (TK) gene promoter (the DTA cassettewas a gift from S. Tajbakhsh, Departement de Biologie du Developpement,Institut Pasteur, Paris, France). The TK-DTA-ShA DNA then wasinserted into the unique BamHI site of pß-gal-pAlox-neo(a gift from Tracy Willson and Warren Alexander, Walter andEliza Hall Institute, Melbourne, Victoria, Australia). The resultingplasmid was denoted pDTA-ShA-ß-gal-pAlox-neo and consistedof the TK-DTA cassette upstream of ShA, which was in frame withthe LacZ gene and upstream of the mouse phosphoglycerate kinasegene promoter and the cDNA encoding neomycin phosphotransferase(PGK-neo^r). The long arm of homology, which consisted of a 6.6-kbEcoRI fragment containing exon 2 and a region of the intronbetween exons 1 and 2 of the Vegf-d gene, was inserted intopSP73 (Promega Corporation) at the EcoRI site to create pSP73-LA.A DNA fragment consisting of TK-DTA-ShA, LacZ, and PGK-neo^rwas isolated from pDTA-ShA-ß-gal-pAlox-neo by digestionwith XhoI and was inserted into pSP73-LA at the SalI site, therebycreating the Vegf-d gene-targeting construct, denoted pVDKO-construct.

Generation of Vegf-d-deficient mice. Electroporation of W9.5 wild-type ES cells (2.5 x 10⁷ cells)with 30 µg of pVDKO-construct that had been linearizedwith XhoI was performed essentially as described previously(9). ES cells were grown on a primary embryonic fibroblast (PEF)feeder layer in the presence of 1,000 U of leukemia inhibitoryfactor (Chemicon)/ml. Selection with G418 (Invitrogen) at aconcentration of 175 µg/ml began 24 h following electroporationand was continued for 9 days, at which time individual colonieswere picked and screened for targeting events by Southern hybridization.Genomic DNA from resistant colonies was digested with each ofthe enzymes HindIII, PvuII, and BglII individually and evaluatedby Southern blotting. Blots were hybridized with a 681-nucleotide(nt) fragment of genomic DNA, denoted E1F2R1, that is homologousto a region of the mouse Vegf-d gene 5' to ShA (Fig. 1). E1F2R1was amplified by PCR with the M18 genomic DNA clone as a template,a sense primer with the sequence 5'-TTTAGTTAGCCCTTTCCTCCA, andan antisense primer with the sequence 5'-TTCTGGGTTGTAGCATAGTCA.Targeted clones were identified at a frequency of 1 of 21, and2 of them (clones 2.2.36 and 2.88) were injected into C57BL/6Jblastocysts and transferred into pseudopregnant mice. Chimericmice were mated, and the offspring were genotyped by Southernhybridization to identify females carrying a single targetedallele (denoted X^–X⁺ because the Vegf-d gene is on theX chromosome) and Vegf-d-deficient males (X^–Y). Subsequentmating predominantly involved the breeding of X⁺X⁺ and X⁺Y miceto generate litters consisting entirely of wild-type mice andthe breeding of X^–X^– and X^–Y mice to generatelitters consisting entirely of Vegf-d-deficient mice. Such offspringwere used for experimental phenotypic analyses. Mice designatedbreeders were the offspring of crosses of wild-type and Vegf-d-deficientmice or were the offspring of crosses of X⁺X^– and eitherX⁺Y or X^–Y mice. Breeders were generated in this way toprevent the genetic divergence of wild-type and Vegf-d-deficientmice.

View larger version (33K):
[in this window]
[in a new window]
FIG. 1. Generation of Vegf-d-deficient mice. (A) Homologous recombination of the gene-targeting vector at the Vegf-d locus (wild-type allele; top) was designed to replace the signal sequence for protein secretion and the remaining sequence of the first coding exon with a LacZ-PGK-neo^r cassette. Construction of the gene-targeting vector (middle) is described in Materials and Methods, and the structure of the targeted Vegf-d allele (bottom) is shown. The Vegf-d gene consists of seven coding exons (7), but only the first two are shown here. Arrows indicate the positions of primers used to genotype mice by PCR. The position of the PCR-amplified genomic DNA probe, E1F2R1, used to screen ES cell colonies by Southern blotting is shown. The E1F2R1 probe is 5' to the region of the Vegf-d gene that underwent homologous recombination and hybridizes to digested fragments of diagnostic sizes in wild-type and targeted ES cells (the sizes of the relevant fragments are shown at the top and bottom). Restriction sites used to construct the gene-targeting vector or digest genomic DNA for screening are also shown: H, HindIII; B, BglII; P, PvuII. (B) Southern blot analysis of genomic DNA extracted from wild-type (W9.5) and targeted (lines 2.2.36 and 2.88) ES cell clones. Genomic DNA from each ES cell line was digested with HindIII (H), BglII (B), and PvuII (P) and analyzed with the E1F2R1 DNA probe. The asterisks in the 2.88 panel indicate hybridization of the E1F2R1 probe to wild-type DNA fragments derived from copurification of PEF DNA with ES cell DNA. The migration of DNA standards (in kilobases) is shown to the left of each panel. (C) Genotypic analysis of mice by PCR. Primers VEGFCM-58 and VEGFCM-36 amplified a 461-bp DNA fragment from a wild-type Vegf-d allele, whereas primers VEGFCM-58 and GenoRC2Alt amplified a 270-bp DNA fragment from a targeted Vegf-d allele. Genotypic analysis of wild-type female (X⁺X⁺) and male (X⁺Y), homozygous Vegf-d-deficient female (X^–X^–) and male (X^–Y), and hemizygous female (X⁺X^–) Vegf-d mice is shown. The migration of DNA size standards (in base pairs) is shown to the right of the panel.

Mice were genotyped by PCR with genomic DNA extracted from thetail tip as a template in a 25-µl PCR mixture consistingof 200 µM deoxynucleoside triphosphates (Amersham PharmaciaBiotech), 2.0 mM MgCl₂, 1x AmpliTaq Gold buffer (Applied Biosystems),and 1.25 U of AmpliTaq Gold. The following primers were alsoadded to the PCR mixture: VEGFCM-58 (5'-CTTTCTCCCATACTAAGATTG;added to a final concentration of 762 nM), which is a forwardprimer homologous to a region of the Vegf-d gene 221 nt upstreamof the translation initiation codon; VEGFCM-36 (5'-TCAAACATTCATGGTAAGAAA;added to a final concentration of 1.4 µM), which is areverse primer homologous to a region 158 nt downstream of exon1; and GenoRC2Alt (5'-CGCCAGGGTTTTCCCAGTCAC; added to a finalconcentration of 94.2 nM), which is a reverse primer homologousto a region of the LacZ gene 46 nt downstream of the Vegf-dtranslation initiation codon in Vegf-d gene-targeted cells (Fig.1A shows primer positions). The wild-type Vegf-d allele wasidentified with a 461-nt DNA fragment (amplified with primersVEGFCM-58 and VEGFCM-36). The targeted Vegf-d allele was identifiedwith a 270-nt DNA fragment (amplified with primers VEGFCM-58and GenoRC2Alt) (Fig. 1C).

All experiments performed with mice were carried out in accordancewith the guidelines set by the National Health and Medical ResearchCouncil of Australia.

Derivation of PEF cell lines. PEF cell lines for Northern and Western blot analyses were generatedfrom litters of wild-type and Vegf-d-deficient mice. Embryoswere collected at E13.5; the heads and livers were removed,digested with 0.1% trypsin-0.02% EDTA, and sheared by passagethrough needles of increasing gauge (18, 19, 21, 23, and 25gauge). Cells were pelleted, resuspended in Dulbecco modifiedEagle medium containing 10% fetal calf serum, 2 mM glutamine,and 10^–6 M ß-mercaptoethanol, and transferredto gelatin-coated flasks. Cells were incubated at 37°C in10% CO₂, and PEFs were passaged 1:2 when a confluent, adherentmonolayer had grown. For Northern and Western blot analyses,PEF supernatants and PEFs were collected when the cells wereconfluent at passage 5.

Northern blot analysis. RNA was extracted from mouse organs and PEF cell lines by theguanidine thiocyanate method essentially as described previously(38). Northern blots were hybridized with a 209-nt PCR product(VDEx7) that had been amplified from Vegf-d cDNA by using asense primer specific to exon 7 (VEGFCM-87: 5'-CTGTGAGGACAGATGTCCTTTTC)and an antisense primer specific to the 3' untranslated regionlocated in exon 7 (VEGFCM-8: 5'-GGTAGTGGGCAACAGTGACAG). In addition,Northern blots were hybridized with a 425-nt PCR product (VD-VHD)amplified from Vegf-d cDNA by using a sense primer specificto exon 3 (VEGFCM-22: 5'-GAAGAATGGCAGAGGACCCA) and an antisenseprimer specific to exon 5 (VEGFCM-17: 5'-CCAGGCAGTGGAGTCTC).VDEx7 and VD-VHD hybridized to transcripts of the same sizes.Northern blotting for Vegf-c mRNA was carried out essentiallyas described previously (23).

Western blot analysis. Antiserum A2 (raised against a peptide in the VEGF homologydomain [VHD] of human VEGF-D) (43), which binds to the VHD ofmouse Vegf-d, or antiserum raised against the DokR protein (26)was incubated with protein A-Sepharose (Amersham Pharmacia Biotech)with rotation for 1 h at 4°C. PEF conditioned medium wasadded and incubated with rotation for $~$ 1.5 h at 4°C. ProteinA-Sepharose then was pelleted by centrifugation and washed threetimes with cold radioimmunoprecipitation buffer (20 mM Tris-HCl[pH 7.5], 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate,0.1% sodium dodecyl sulfate [SDS]). Proteins were eluted bythe addition of reducing SDS sample buffer (2% SDS, 10% glycerol,60 mM Tris-HCl [pH 6.8], 0.02% bromophenol blue, 2% ß-mercaptoethanol)and heating to 95°C for 5 min. Proteins were subjected toSDS-polyacrylamide gel electrophoresis (PAGE), transferred toImmobilon P membranes (Millipore), and analyzed by using biotinylated,affinity-purified antiserum (BAF469; 0.2 µg/ml; R&DSystems) raised against amino acids 98 to 206 of mouse Vegf-d(which corresponds to the VHD) and streptavidin-horseradishperoxidase conjugate (Roche Molecular Biochemicals). Developmentwas carried out by chemiluminescence (SuperSignal System; Pierce).

Histological analysis. Excised embryos or mouse tissues were fixed for 3 to 24 h in4% paraformaldehyde-phosphate-buffered saline, dehydrated, andembedded in paraffin wax. Paraffin sections (5 µm) werestained with hematoxylin and eosin or subjected to immunohistochemicalstaining with a rat anti-mouse monoclonal antibody to the extracellulardomain of mouse Vegfr-3 (22), biotin-conjugated anti-rat immunoglobulin(Dako Corporation), and a tyramide signal amplification (TSA)system according to the manufacturer's instructions (NEN LifeScience Products). As a control, the Vegfr-3 primary antibodywas omitted from the procedure. The number of Vegfr-3-positivelymphatic vessels immediately adjacent to the external surfaceof the muscular layer surrounding bronchioles in age-matchedadult lungs was counted by an investigator who was unaware ofwhich sections were derived from wild-type or Vegf-d-deficientmice. The tissue sections used for this analysis were matchedfor position within the lungs. Three tissue sections were countedper mouse and encompassed at least 130 bronchiole sections peranimal.

Measurement of lung fluid volume. Mouse lungs were excised and lightly blotted onto gauze to removeblood acquired during separation from the heart, and the masswas measured (lung wet mass). The lungs then were dried in anoven at 60°C for 3 days, and the mass was measured dailythereafter. The lung dry mass was recorded when the lung massremained constant over two consecutive measurements. Lung fluidcompositions were calculated by subtracting the lung dry massfrom the lung wet mass.

Microlymphangiography. Patent Blue 5 dye (Rhone Poulenc Rorer) was injected intradermallyinto mice anesthetized with xylazine (8 mg/kg)-ketamine (44mg/kg). The injection site was lightly massaged for 15 min,and the mice were sacrificed by cervical dislocation and dissectedto reveal the underlying surface of the ventral skin and theaxillary, superficial inguinal, and paraaortic lymph nodes.Uptake of Patent Blue 5 dye into lymphatic vessels was monitoredby using a Leica dissecting microscope (model MZ6) and a digitalcamera. Fluorescence microlymphangiography was performed onmouse tails essentially as described previously (15, 25). Hairwas removed from tails 24 h prior to analysis by using depilatorycream. On the day of the experiment, mice were anesthetizedwith xylazine (8 mg/kg)-ketamine (44 mg/kg). Tetramethylrhodamine-dextran(rhodamine-dextran; molecular weight, 2,000,000; 8 mg/ml; MolecularProbes) was injected subdermally ( $~$ 50 µl) into the mousetail tip by using a 30-gauge needle. Mice then were placed onthe heated stage (32°C) of a Nikon confocal microscope (modelEclipse TE 300). Uptake of rhodamine-dextran into the superficiallymphatic networks of mouse tails was monitored by confocalmicroscopy for a period of $~$ 40 min per mouse. The width of thelymphatic vessels and the widest distance between the vesselsin each component of the honeycomb lymphatic pattern were measuredfrom the confocal images.

RESULTS

Top

Results

Generation of Vegf-d-deficient mice. As a prelude to the generation of Vegf-d-deficient mice, Southernblot analysis of W9.5 ES cell genomic DNA was carried out withDNA probes homologous to the 5'- and 3'-end regions of the mouseVegf-d gene. This analysis revealed that each probe hybridizedto one genomic DNA fragment in each of the digested samples,suggesting that there is only one copy of the Vegf-d gene inthe mouse genome (data not shown). This finding is consistentwith the interspecific backcross analysis, which localized themouse Vegf-d gene to the X chromosome (16). These studies didnot reveal any evidence for a second Vegf-d gene or a Vegf-dpseudogene.

To generate Vegf-d-deficient mice, a gene-targeting vector wasused to replace the signal sequence for protein secretion, immediatelydownstream from the translation initiation site, and the remainderof the first coding exon (Fig. 1A). Correctly targeted ES cellswere identified by Southern blot analysis with a DNA probe,E1F2R1, that hybridizes to HindIII, BglII, and PvuII fragmentsof diagnostic sizes in wild-type and Vegf-d-targeted ES cells(Fig. 1B). Eighteen correctly targeted ES cell colonies, outof a total of 384 screened, were identified (targeting frequency,1 of 21); 2 of these were used to generate independent lines(2.2.36 and 2.88) (Fig. 1B). The 2.2.36 and 2.88 ES cell cloneshad incorporated a single copy of the neo^r gene in their genomes,as indicated by Southern blot analysis with a DNA probe homologousto a region of the neo^r gene (data not shown).

Molecular analysis of Vegf-d-deficient mice. To investigate whether replacement of the first coding exonwith a PGK-neo^r cassette affected the generation of transcriptscontaining the remaining exons of the Vegf-d gene, Northernblot analysis was performed (Fig. 2A). The expression of Vegf-din mice was previously shown to be high in E15.5 embryonic lungs(43). Total RNA was purified from the lungs of wild-type andVegf-d-deficient E15.5 embryos and subjected to Northern blottingwith a probe directed to the 3' end of the Vegf-d gene (VDEx7).The major Vegf-d transcripts, of 3.7 and 2.4 kb, in additionto the lower-abundance transcripts, of 3.3 and 2.0 kb, weredetected in wild-type E15.5 mouse lungs but not in Vegf-d-deficientE15.5 lungs (Fig. 2A). Because the Vegf-d gene is also stronglyexpressed in PEFs, we confirmed this lack of expression in Vegf-d-deficientmice by generating multiple PEF lines from matings of X⁺X⁺ andX⁺Y mice (wild-type PEFs; this designation is used because theVegf-d gene is on the X chromosome) and matings of X^–X^–and X^–Y mice (Vegf-d-deficient PEFs). In agreement withthe Northern analysis of embryonic lungs, a high level of Vegf-dexpression was detected in all of the wild-type PEF lines, whereasno transcripts of any size could be detected in RNA from Vegf-d-deficientPEFs with this probe (Fig. 2A). Northern blot analysis witha probe homologous to a region of the VHD of Vegf-d (VD-VHD)also detected high levels of Vegf-d expression in wild-typeE15.5 lungs and PEF lines but did not detect any transcriptsin the RNA derived from Vegf-d-deficient E15.5 lungs or PEFlines (data not shown). These data were also consistent withthe results of Northern blot analysis of RNA from mouse line2.88 and PEFs derived from this line (data not shown).

View larger version (46K):
[in this window]
[in a new window]
FIG. 2. Northern and Western blot analyses of tissues and cell lines derived from Vegf-d-deficient mice. (A) Analysis of Vegf-d expression in E15.5 lungs (left panel) and PEF conditioned media (right panel) derived from wild-type (+/+) and Vegf-d-deficient (–/–) mice (line 2.2.36). Total RNA was analyzed by Northern blotting with the VDEx7 cDNA probe (the same results were obtained with the VD-VHD probe; data not shown). The signal obtained with a cDNA probe for the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) housekeeping gene is shown as a loading control for E15.5 lung RNA, and methylene blue staining of the 28S rRNA transferred to membranes is indicated as a loading control for PEF RNA. Sizes of Vegf-d transcripts (in kilobases) are indicated to the left of the panels. (B) Western blot analysis of conditioned media of PEFs derived from wild-type (+/+) and Vegf-d-deficient (–/–) embryos (line 2.2.36). Conditioned media were incubated with antiserum A2, which binds to the VHD of Vegf-d, or control antiserum raised against the mouse DokR protein (26). The immunoprecipitate was subjected to SDS-PAGE (reducing condi-tions), transferred to membranes, and blotted with affinity-purified, biotinylated serum raised against the VHD of mouse Vegf-d. The $~$ 21-kDa mature form, consisting of the VHD, and the $~$ 31-kDa derivative, consisting of the N-terminal propeptide and the VHD, were detected in conditioned media of wild-type PEFs but not in media of PEFs from Vegf-d-deficient mice. The specificity of A2 and biotinylated VEGF-D antisera for Vegf-d is indicated by the lack of the $~$ 31- and $~$ 21-kDa proteins when DokR antiserum was used in the immunoprecipitation reaction. The band of $~$ 55 kDa detected in all lanes represents the immunoglobulin heavy chain. The migration of molecular mass markers (in kilodaltons) is shown to the right of the panel. (C) Northern blot analysis of mouse tissues for Vegf-c mRNA. Ten micrograms of total RNA from tissues of adult wild-type (+/+) and Vegf-d-deficient (–/–) mice (line 2.2.36) was hybridized with the Vegf-c probe. The position of the 2.4-kb Vegf-c mRNA is marked to the left of the upper panel, and methylene blue staining of 28S rRNA on the filters is shown in the lower panel as a loading control.

To further confirm that the targeting strategy resulted in thegeneration of a null Vegf-d allele, conditioned media derivedfrom wild-type and Vegf-d-deficient PEF lines were subjectedto immunoprecipitation followed by Western blot analysis withantisera raised against peptides in the VHD of human and mouseVegf-d (see Materials and Methods). The Western blots demonstratedthat the $~$ 21-kDa fully processed subunit of Vegf-d, consistingof the VHD (31, 43), was present in the conditioned medium ofwild-type PEFs but not in the conditioned medium of Vegf-d-deficientPEFs (Fig. 2B). In reducing SDS-PAGE, this fully processed derivativeconsisting of the VHD migrates as two species at $~$ 21 kDa, dueto variations in glycosylation. Previous studies indicated thatcleavage at the C terminus of the VHD is more rapid than thatat the N terminus (43). Consequently, the partially processedderivative consisting of the VHD and the N-terminal propeptideshould be relatively abundant compared to the fully processedVHD derivative and other partially processed forms. For humanVEGF-D, the size of the polypeptide consisting of the N-terminalpropeptide and the VHD is $~$ 31 kDa, consistent with the migrationof the N-terminal propeptide at $~$ 10 kDa and of the VHD at $~$ 21kDa under reducing conditions. This $~$ 31-kDa partially processedform of mouse Vegf-d was abundant in wild-type PEF conditionedmedium but could not be detected in that of Vegf-d-deficientPEFs (Fig. 2B). The $~$ 21-kDa mature form and the $~$ 31-kDa partiallyprocessed form of Vegf-d were abundant in conditioned mediaof multiple wild-type PEF lines, whereas Vegf-d could not bedetected in conditioned media of any of the Vegf-d-deficientPEFs derived from either of the Vegf-d-deficient mouse lines.The lack of expression of Vegf-d protein in Vegf-d-deficientmutant mice indicates that the gene-targeting strategy createda null Vegf-d allele.

Vegf-d is closely related in primary structure to Vegf-c, andthese proteins are both able to induce lymphangiogenesis viathe activation of Vegfr-3 (45). It is therefore feasible thatthe up-regulation of Vegf-c could compensate for Vegf-d function.However, Vegf-d deficiency did not significantly alter the expressionof the Vegf-c gene, as assessed by Nothern blot comparison ofVegf-c mRNA levels in livers, hearts, and lungs from wild-typeand Vegf-d-deficient mice (Fig. 2C).

General observations of Vegf-d-deficient mice. Neither female nor male Vegf-d-deficient mice (denoted X^–X^–and X^–Y, respectively) could be distinguished from wild-typemice by morphological appearance. Newborn Vegf-d-deficient micewere observed to suckle and, postweaning, to feed in a mannersimilar to that of control mice. This finding is reflected inthe body masses, which were not different in wild-type and Vegf-d-deficientmice measured at up to 1 year of age (Table 1 shows measurementsat 16 weeks). The similar body masses of Vegf-d-deficient andwild-type mice suggest that Vegf-d-deficient mice are able toeffectively access, digest, and metabolize food to the sameextent as wild-type mice. This suggestion is consistent withobservations that Vegf-d-deficient mice displayed levels ofactivity similar to those of wild-type mice. Vegf-d-deficientmice were fertile, as visibly healthy mice were born from crossesof X^–X^– and X^–Y mice in litters with a mean± standard deviation number of offspring that was notsignificantly different (as assessed by the Student t test)from that in litters derived from wild-type matings (for wild-typemice, 8.0 ± 2.5 [n = 26]; for Vegf-d-deficient mice,9.3 ± 3.0 [n = 34]) (P = 0.078). Perturbations in thedevelopment and function of the lymphatic system may be reflectedin an alteration in interstitial fluid volume leading to swellingof tissue or accumulation of fluid in body cavities. However,there was no visual evidence of altered interstitial fluid volumein the subcutaneous, intraperitoneal, or thoracic regions ofVegf-d-deficient mice, nor was there any accumulation of chylein the intestines or peritoneal cavity. None of the tissuesor organs of Vegf-d-deficient mice appeared swollen or distended.

View this table:
[in this window]
[in a new window]
TABLE 1. Body and lung masses of mice at 16 weeks of age^a

Histological analysis of lymphatic vessels in the lungs of Vegf-d-deficient mice. Because the Vegf-d gene is highly expressed in mouse lungs duringembryonic development and in adults (1, 5, 43, 47), we hypothesizedthat it may have a role in the development and modulation oflymphatic vessel growth in these tissues. The lungs of Vegf-d-deficientmice were comparable in mass (Table 1) and morphology to thoseof wild-type mice and appeared to function normally, as Vegf-d-deficientmice did not suffer difficulties with respiration. Hematoxylinand eosin staining of lung sections indicated a normal alterationof alveolar density relative to the midline in both Vegf-d-deficientand wild-type mice, reflecting bronchus and alveolar branchingin the pulmonary tree (33). The alveolar densities in sectionsmatched for position from the midline were not obviously differentin Vegf-d-deficient and wild-type mice. In addition, the tissueand cellular organizations of these matched lung tissue sections,taking into consideration the apparent number and structureof bronchioles in a tissue section, were not different in Vegf-d-deficientand wild-type mice.

Immunohistochemical staining with an antibody that was raisedagainst the extracellular domain of mouse Vegfr-3 and that detectslymphatic endothelium in normal adult tissues indicated thatthe lymphatic vessels of Vegf-d-deficient mouse lungs were similarin structure, appearance, and location to those of wild-typemouse lungs (Fig. 3). Typically, Vegfr-3-positive lymphaticvessels are clustered adjacent to bronchioles and the bloodvessels associated with the bronchioles in the lungs (Fig. 3E and F).Counting of Vegfr-3-positive vessels immediately adjacentto the external surface of the muscular layer of bronchiolesrevealed a small but statistically significant reduction (asassessed by the Student t test) in the number of Vegfr-3-positivevessels per bronchiole in Vegf-d-deficient mice in comparisonto wild-type control mice (mean ± standard deviationfor Vegf-d-deficient mice, 0.80 ± 0.16 [n = 10]; forwild-type mice, 1.15 ± 0.14 [n = 10]) (P = 0.00005).

View larger version (135K):
[in this window]
[in a new window]
FIG. 3. Immunostaining of lungs for Vegfr-3. (A and C) Immunostaining of vessels (arrows) in the lungs of wild-type and Vegf-d-deficient mice, respectively, with Vegfr-3 monoclonal antibody. (B and D) Serial sections from panels A and C, respectively, in which the Vegfr-3 antibody was omitted from the immunostaining protocol. (E and F) Higher-power images of lung tissues obtained from wild-type and Vegf-d-deficient mice, respectively, and immunostained for Vegfr-3 show bronchioles (asterisks), Vegfr-3-positive lymphatic vessels (arrows), and blood vessels (arrowheads).

Analysis of lung fluid volume in Vegf-d-deficient mice. Because we observed a slight difference in the abundance oflymphatic vessels adjacent to bronchioles in the lungs of Vegf-d-deficientmice, we compared the volume of fluid within the lungs of wild-typeand Vegf-d-deficient mice because this parameter can be usedas an indicator of lung edema (13, 30). Lungs were excised fromthe chest cavity, weighed (wet mass), and dried in an oven forat least 3 days, and the mass was remeasured (dry mass). Therewas no significant difference in total lung masses or lung fluidcompositions in wild-type and Vegf-d-deficient mice (Table 1),suggesting that Vegf-d-deficient mice have functional pulmonarylymphatic vessels.

Structure and function of large lymphatic vessels in Vegf-d-deficient mice. The mouse Vegf-d gene is strongly expressed immediately underthe skin during embryonic development (4, 5). We therefore analyzedthe function of large lymphatic vessels associated with theskin in adult Vegf-d-deficient mice. Patent Blue 5 dye, whichenters lymphatic vessels but not blood vessels (14), was intradermallyinjected into the ventral skin of wild-type mice. The dye wasspecifically taken up from the interstitium into a very largelymphatic vessel that is located on the underlying surface ofthe ventral skin (Fig. 4A). As expected, the dye did not enterthe blood vascular system during the course of this experiment,as shown by the lack of dye in the accompanying blood vessel(Fig. 4A). Furthermore, this large lymphatic vessel drainedto the superficial inguinal (Fig. 4C) and axillary (Fig. 4E)lymph nodes. The locations of this lymphatic vessel, the accompanyingblood vessel, and the draining lymph node were similar in allof the wild-type mice studied (n = 12). The uptake of PatentBlue 5 dye into this large lymphatic vessel was also observedin all of the Vegf-d-deficient mice analyzed (n = 10) (Fig.4B). Drainage of the dye to the superficial inguinal (Fig. 4D)and axillary (Fig. 4F) lymph nodes also occurred in the Vegf-d-deficientmice. Drainage to the paraaortic lymph nodes was also observedin wild-type and Vegf-d-deficient mice (Fig. 4G and H), suggestingthat the drainage of lymph from large lymphatic vessels to lymphnodes is not significantly affected by Vegf-d deficiency.

View larger version (76K):
[in this window]
[in a new window]
FIG. 4. Visualization of lymphatic vessels and lymph nodes by dye injection. Patent Blue 5 dye was intradermally injected into the ventral skin of wild-type (+/+) and Vegf-d-deficient (–/–) mice. The dye was taken up into a large lymphatic vessel (arrow) but not an accompanying blood vessel in the ventral skin of both wild-type (A) and Vegf-d-deficient (B) mice. The dye was transported to a superficial inguinal lymph node (arrow) of wild-type (C) and Vegf-d-deficient (D) mice as well as the draining axillary lymph node (arrow) of wild-type (E) and Vegf-d-deficient (F) mice. Uptake of the dye into the paraaortic lymph node (arrows) was also observed in wild-type (G) and Vegf-d-deficient (H) mice.

Following injection, uptake of the dye into the large lymphaticvessel under the skin was rapid (within 5 min) in both wild-typeand Vegf-d-deficient mice; no obvious difference in uptake ratewas observed. Leakage of the dye from this lymphatic vessel,which would be an indicator of an abnormality of lymphatic functionor endothelial cell association and/or permeability, was notobserved in either wild-type or Vegf-d-deficient mice. The locationof this lymphatic vessel in Vegf-d-deficient mice was also comparableto that in control mice.

Structure and function of dermal lymphatic vessels. The dermal lymphatic plexus of the mouse tail is arranged ina regular honeycomb-like pattern that allows abnormalities oflymphatic vessel density, location, and function to be readilyassessed by fluorescence microlymphangiography (15, 25, 44).Rhodamine-dextran, which is specifically taken up by lymphaticvessels, was injected into the tips of mouse tails, and itsdistribution was monitored by fluorescence microscopy. As expected,a regular superficial lymphatic network was observed in allwild-type mice examined (n = 20) (Fig. 5A), and rhodamine-dextranwas transported at least 3 cm along the tail from the site ofinjection within 15 min. No difference was observed in the patternor rate of dye uptake into the superficial lymphatic networkin Vegf-d-deficient mice (n = 17) (Fig. 5D). In addition, lymphaticvessel width and the maximum distance between vessels in eachcomponent of the honeycomb lymphatic pattern were not significantlydifferent in Vegf-d-deficient mice and control mice (the P valuefor both parameters, as determined by the Student t test, was>0.05). At higher magnification, comparably abundant precollectorsor ducts were observed in wild-type and Vegf-d-deficient mice(Fig. 5C and F). These ducts drain to deeper precollectors that,in turn, drain to deep lymphatic vessels beneath the skin (25).These findings indicate that the dermal lymphatic vasculatureof Vegf-d-deficient mice is arranged in an orderly, functionalconfiguration that is comparable to that of wild-type mice.

View larger version (60K):
[in this window]
[in a new window]
FIG. 5. Rhodamine-dextran uptake into the superficial lymphatic network of the tails of wild-type and Vegf-d-deficient mice. Rhodamine-dextran was injected into the tips of the tails of wild-type (+/+) (A) and Vegf-d-deficient (–/–) (D) mice, and uptake into the superficial lymphatic network was monitored by confocal microscopy. Uptake of the dye into lymphatic drainage ducts (arrows) of wild-type (C) and Vegf-d-deficient (F) mice was observed at higher magnifications. Autofluorescence in the tails of wild-type (B) and Vegf-d-deficient (E) mice prior to injection of the dye is also shown.

DISCUSSION

Top

Discussion

We hypothesized that Vegf-d may have a role in the embryonicdevelopment of the lymphatic vasculature because it is lymphangiogenic(45), it is an activating ligand for Vegfr-3 (1, 6), and itis expressed in a range of tissues during embryogenesis, whenthe formation of the lymphatics is taking place (5, 43). Perturbationsin lymphatic development or function may be manifested as tissueedema and accumulation of chyle in the intestines or peritonealcavity, as was observed in mutant mice deficient in Vegf-c (20),the homeobox protein Prox1 (46), or angiopoietin 2, a ligandfor endothelial cell receptor Tie2 (12). Therefore, it was surprisingthat there was no evidence of edema in the skin or organs ofVegf-d-deficient mice, nor was there an abnormal accumulationof chyle in the intestines or peritoneal cavity. The lymphaticvessels and lymph nodes appeared to develop and function normally,as indicated by immunohistochemical and microlymphangiographicstudies. Furthermore, the lymphatics of Vegf-d-deficient micewere similar in width, location, and structural integrity tothose of wild-type mice and were capable of transporting fluidto lymph nodes in the axillary and inguinal regions and to deeplymph nodes surrounding the descending aorta. Therefore, inactivationof the Vegf-d gene alone does not significantly perturb lymphaticdevelopment and clearly does not phenocopy the overt phenotypeexhibited by Vegfr-3-deficient mice, which die during earlyembryonic development from defects in the remodeling of largeblood vessels, before the lymphatic vessels emerge (11).

The lack of a profound lymphatic vessel defect in Vegf-d-deficientmice may reflect a subtle, redundant, or nonexistent role forthis growth factor during embryonic development. Nonetheless,Vegf-d may induce lymphatic vessel growth in adult life in responseto pathological conditions. Vegf-d-deficient mice representan opportunity to study such putative roles by analyzing theeffects of physiological or pathological challenges. This studyalso suggests that there may be functional redundancy withinthe VEGF family. The most likely candidate for compensatingfor Vegf-d deficiency is Vegf-c. Vegf-c is highly homologousto Vegf-d (1, 17), is the only other known activator of Vegfr-3(17), has an expression pattern partly overlapping that of Vegf-d(1, 4, 5, 43, 23, 24), and is critical for the development ofthe lymphatic vasculature (20). We did not, however, find anyevidence of Vegf-c up-regulation in the tissues of Vegf-d-deficientmice by Northern blotting, suggesting that if such compensationoccurs, then it does so without a requirement for significantlyenhanced levels of Vegf-c mRNA.

Inactivation of the Vegfr-3 gene in mice causes death at E9.5in association with defective remodeling of blood vessels (11),whereas inactivation of the mouse Vegf-c gene leads to deathof embryos between E15.5 and birth due to fluid accumulationin tissues as a consequence of a lack of lymphatic vessels (20).As Vegf-c and Vegf-d are the only known ligands for Vegfr-3,we anticipated that Vegf-d function might account for the differencebetween the phenotypes of Vegf-c^–^/^– and Vegfr-3^–^/^–mice. The lack of a profound phenotype in Vegf-d-deficient miceraises the possibility that Vegf-d may not function in Vegf-c-deficientmice to allow them to survive longer than Vegfr-3-deficientmice but rather that other, as-yet-unknown signaling mechanismsvia Vegfr-3 could be involved during the early development ofblood vessels. Such mechanisms could involve other soluble Vegfr-3ligands or cell surface-ssociated molecules. The generationof mice doubly deficient for Vegf-c and Vegf-d will be requiredto explore this issue more conclusively.

The findings presented here indicate that Vegf-c-mediated signalingis more important for the development of the lymphatic vasculatureduring embryogenesis than is signaling induced by Vegf-d. Itis noteworthy that a range of in vivo delivery approaches havedemonstrated that Vegf-d can induce lymphangiogenesis and angiogenesisin normal tissues and pathological conditions, such as cancer(10, 36, 37, 42, 45). Therefore, the biological functions ofVegf-d may be revealed in the response to disease or tissuedamage rather than in the context of embryonic development.

ACKNOWLEDGMENTS

This work was supported by grants from the National Health andMedical Research Council of Australia (NH&MRC) and the CancerCouncil of Victoria, Australia (CCV). M.E.B. was supported bya postdoctoral fellowship from the CCV, S.A.S. was supportedby a Pfizer Foundation senior research fellowship, and M.G.A.and M.L.H. were supported by senior research fellowships fromthe NH&MRC.

We thank Shin-Ichi Nishikawa for antibody against Vegfr-3, PatriciaLai, Dimitria Vranes, and Gillian Thornton for technical assistance,Peter Lock for DokR antiserum, and Tony Burgess for constructivecomments.

FOOTNOTES

* Corresponding author. Mailing address: Ludwig Institute for Cancer Research, Royal Melbourne Hospital, P.O. Box 2008, Parkville, Victoria 3050, Australia. Phone: (613) 9341-3155. Fax: (613) 9341-3107. E-mail: Marc.achen{at}ludwig.edu.au.

Present address: Genentech Inc., South San Francisco, CA 94080-4990.

Present address: Center for Developmental Biology and Kent WaldrepFoundation Center for Basic Research on Nerve Growth and Regeneration,University of Texas Southwestern Medical Center, Dallas, TX75390-9133.

REFERENCES

Top