|Home | About | Journals | Submit | Contact Us | Français|
Disruption of the precise balance of positive and negative molecular regulators of blood and lymphatic vessels can lead to myriad diseases that affect one in four people worldwide. Although dozens of natural inhibitors of hemangiogenesis have been identified, an endogenous selective inhibitor of lymphatic vessels has not yet been described. We report the existence of a secreted, splice variant of vascular endothelial growth factor receptor-2 (sVegfr-2) that inhibits developmental and reparative lymphangiogenesis by blocking Vegf-c. Tissue-specific loss of sVegfr-2 in mice induced, at birth, spontaneous lymphatic invasion of the normally alymphatic cornea and hyperplasia of skin lymphatics without accompanying changes in blood vasculature. sVegfr-2 inhibited lymphangiogenesis but not hemangiogenesis induced by corneal suture injury or transplantation, enhanced corneal allograft survival, and suppressed lymphangioma cellular proliferation. Naturally occurring sVegfr-2 is a molecular uncoupler of blood and lymphatic vessels whose modulation might have a therapeutic role in lymphatic vascular malformations, transplantation, and potentially in tumor lymphangiogenesis and lymphedema.
The blood and lymphatic vessel networks are jointly essential for development, wound healing, and immune surveillance and activation. Pathological responses of these parallel circulatory systems can lead to diseases as varied as age-related macular degeneration, atherosclerosis, cancer, lymphedema, rheumatoid arthritis, and tumor metastasis1. Collectively, exuberant or inadequate responses of hemangiogenesis and lymphangiogenesis are estimated to affect nearly two billion people2,3. The prevalence of such diseases has fueled intense efforts to develop proand anti-angiogenic therapeutics. Several anti-hemangiogenic drugs are FDA-approved but not one specific lymphangiogenesis inhibitor is even in clinical trials. As such there is great interest in identifying such specific inhibitors both to alleviate disease burden and to better understand lymphatic vascular biology. However, despite concerted efforts, it has been challenging to approach lymphangiogenesis selectively due to the difficulty in mechanistically disassociating it from hemangiogenesis.
The vascular endothelial growth factors (Vegf) family of molecules is indispensable for growth of blood4,5 and lymphatic6 vessels. Hemangiogenesis requires a precise balance of positive and negative regulators7; however, the mechanisms governing lymphangiogenesis equilibria remain nebulous. Soluble splice variants of Vegf receptor-1 (sVegfr-1) act as potent natural inhibitors of hemangiogenesis by trapping the blood endothelial mitogen Vegf-a (refs. 8-10) and are the only reported secreted splice variants of any of the Vegf receptor tyrosine kinases since their discovery in the early 1990s. Here we describe the detection and function of a new secreted splice variant within the Vegf receptor family, which we demonstrate is a critical endogenous antagonist of Vegf-c and a physiological regulator of lymphatic vessels.
During our studies uncovering the non-redundant function of sVegfr-1 in corneal avascularity11, which is critical for optimal vision, we observed, on western blotting, anomalous migration of a 75 kDa protein species that was immunoreactive to an antibody (T014; ref. 12) recognizing the amino-terminus of Vegfr-2 (Supplementary Fig. 1a). Given the homology in the exon-intron structure between Vegfr1 and Vegfr2, we hypothesized that this unidentified protein represented a novel truncated splice variant of the 230 kDa membrane-bound form of Vegfr-2 (mbVegfr-2) and that it might have anti-lymphangiogenic activity because the lymphatic mitogen Vegf-c (ref. 13) is capable of binding mbVegfr-2 (ref. 14).
By modeling the putative alternative splicing event of Vegfr2 to that of Vegfr1 (ref. 8), we found that retention of intron 13 would yield a truncated transcript variant whose protein product would lack the transmembrane and intracellular tyrosine kinase domains of mbVegfr-2 due to the presence of an in-frame early termination TAA codon 39 nucleotides downstream from the aforementioned exon/intron junction (Supplementary Fig. 1c). To verify the existence of this novel soluble splicing variant in the mouse cornea, we devised primers targeting intron 13 and exon 12 (Supplementary Fig. 1c and Supplementary Fig. 2a). Targeting exon 12 allowed us to distinguish between amplification of mRNA-derived cDNA and genomic DNA contamination based on amplicon size. PCR yielded a 393-bp product encompassing the locus of the splicing event (Supplementary Fig. 1d). Bioinformatic analysis of intron 13 revealed three potential polyadenylation (polyA) signal sequences (Supplementary Fig. 1c). Using rapid amplification of cDNA 3′ ends PCR, we found the third potential polyA signal at position 3956-61 to be active (Supplementary Fig. 1c,e and Supplementary Fig. 2a). From mouse cornea cDNA, we cloned the 2022-bp open reading frame of sVegfr2 encoding a polypeptide of 673 amino acids (Supplementary Fig. 2b, for detailed sequence of the sVegfr2 transcript see Supplementary Fig. 2a). This novel protein contained a unique 13-aa carboxyl-terminus sequence (Supplementary Fig. 1b) not present in mbVegfr-2 or any other known protein and against which we raised a rabbit polyclonal antibody (AA21127; Supplementary Fig. 3).
This sVegfr2 transcript was localized by in situ hybridization principally to the corneal epithelium (Fig. 1a). Immunolocalization using AA21127 in the newborn mouse demonstrated the presence of sVegfr-2 in the corneal epithelium and stroma (Supplementary Fig. 4a). In the adult cornea, sVegfr-2 was more abundant in the epithelium than in the stroma (Fig. 1b and Supplementary Fig. 4a). sVegfr-2 was distributed uniformly in the cornea with enhanced expression near the limbus (Supplementary Fig. 4b). In contrast, sVegfr-2 was not immunolocalized in the conjunctiva (Supplementary Fig. 4b,c). sVegfr-2 was identified in the cornea as a 75 kDa species by western blotting both by AA21127 and T014 (Supplementary Fig. 4d). However, 230 kDa mbVegfr-2 was not detected in the cornea by western blotting using T014 (Supplementary Fig. 1a) or by immunofluorescence using an antibody targeting the carboxyl-terminus of mbVegfr-2, which is not present in sVegfr-2 (Supplementary Fig. 4e). Neither was the mbVegfr2 transcript detected in the cornea by RT-PCR (Supplementary Fig. 4f). Thus, the mouse cornea expresses sVegfr-2 but not mbVegfr-2.
To define the function of sVegfr-2 in the cornea, we targeted it using multiple strategies. Because Vegfr2–/– mice die in utero15, we conditionally ablated corneal sVegfr-2 using a Cre-loxP strategy. This strategy enabled the specific targeting of sVegfr-2 because mbVegfr-2 is not expressed in the cornea (Supplementary Fig. 4e,f). We created Vegfr2loxP/loxP mice by targeting exon 1 (Supplementary Fig. 5) and interbred them with LeCre mice that constitutively and uniformly express Cre recombinase in the cornea16. Mice of all genotypes were born at the expected mendelian ratios and were macroscopically indistinguishable from one another. Strikingly however, all LeCre/Vegfr2loxP/loxP mouse corneas (n = 30), which verifiably lacked sVegfr-2 expression (Supplementary Fig. 6a), were densely covered with lymphatic vessels at birth (P0) (Fig. 1c). These vessels were identified as lymphatics by virtue of their intense Lyve-1 reactivity, moderate Cd31 reactivity, nuclear Prox1 expression, and blind-ended morphology (Fig. 1c-e and Supplementary Fig. 6b). Although Lyve-1+ macrophages have been described in the cornea17, co-expression of Prox1 in these vessels confirms their lymphatic endothelial fate (Fig 1d,e). Furthermore, ultrastructural examination showed that these vessels lacked erythrocytes, did not have a continuous basement membrane, and contained partly overlapping thin endothelial cells free of pericyte coverage, all features typical of lymphatics (Fig. 1f). Surprisingly these corneas were not invaded by blood vessels as demonstrated by the absence of Cd31+ Lyve-1– vessels (Fig. 1c,g). We confirmed this independently by demonstrating that the vessels in these corneas did not express Meca-32, a blood vessel-specific marker (Supplementary Fig. 6c,d). All littermate control corneas (n =30) were, like wild-type mice, devoid of both lymphatic and blood vessels. LeCre/Vegfr2loxP/loxP mice had normal limbal blood vessel morphology and an unimpaired corneal hemangiogenesis response to suture injury (Supplementary Fig 6d,e), confirming intact vascular endothelial cell Vegfr-2 function.
These results indicate that the developing mouse cornea is exposed to pro-lymphangiogenic stimuli that are counterbalanced by sVegfr-2 to create an alymphatic cornea. Indeed wild-type P0 corneas, unlike adult corneas, expressed Vegf-c (Supplementary Fig. 7). We reasoned that sVegfr-2 ablation led to spontaneous corneal lymphangiogenesis because sVegfr-2 trapped Vegf-c. Indeed, the levels of sVegfr-2 (1,655.8±44.62 pg per mg of total protein) in the wild-type P0 cornea were sufficiently in excess of Vegf-c (45.4±3.2 pg per mg of total protein) to fulfill a trapping function. sVegfr-2, which contains the Vegf-c binding Ig-like domain 2 of mbVegfr-2 (ref. 18), interacted with Vegf-c by immunoprecipitation and inhibited both Vegfr-3 phosphorylation and proliferation of lymphatic endothelial cells (LECs) stimulated by Vegf-c (Fig. 1h-j). Also, Vegf-c-induced corneal lymphangiogenesis was inhibited by intracorneal administration11 of a plasmid encoding for sVegfr-2 (psVegfr-2) (Supplementary Fig. 8). Collectively, these data are consistent with a model of sVegfr-2 acting as an endogenous sink for Vegf-c during corneal development and thereby establishing an alymphatic cornea.
Consistent with the report that mouse Vegfr-2 binds human VEGF-D but not mouse Vegf-d (ref. 19), psVegfr-2 inhibited corneal lymphangiogenesis induced by human VEGF-D but not mouse Vegf-d (Supplementary Fig. 8). Also, unlike Vegf-c, Vegf-d was not expressed in the newborn mouse cornea (Supplementary Fig. 9), thus accounting for the ability of sVegfr-2 to maintain the alymphatic nature of the cornea.
To determine the function of sVegfr-2 in the adult, we studied a clinically relevant mouse model of suture-induced corneal neovascularization20. Suture injury increased corneal sVegfr-2 expression (in the epithelium and not in infiltrating macrophages) in wild-type mice (Supplementary Fig. 10). Ablating sVegfr-2 by intracorneal administration11 of pCre in Vegfr2loxP/loxP mice markedly increased suture-induced lymphangiogenesis but not hemangiogenesis compared to pNull administration by 161±9% (P < 0.05, n = 5) (Fig. 2a-c). This outstripping of hemangiogenesis by lymphangiogenesis suggests that induction of endogenous sVegfr-2 by injury is a compensatory anti-lymphangiogenic response. Both pCre and pNull-injected corneas of wild-type mice showed similar degrees of lymphangiogenesis following suture injury, excluding a non-specific effect of Cre recombinase (Fig. 2c). pCre induced enhancement of suture-induced lymphangiogenesis in Vegfr2loxP/loxP mice was reduced by a Vegfr-3 tyrosine kinase inhibitor21, supporting the concept that endogenous sVegfr-2 is an in vivo Vegf-c antagonist (Fig. 2c). Conversely, augmenting sVegfr-2 via in vivo transfection11 of wild-type mouse corneas with psVegfr-2, but not pNull, reduced suture injury-induced lymphangiogenesis but not hemangiogenesis by 76±6% (P < 0.05, n = 5) (Fig. 2d,e).
Because lymphangiogenesis has been implicated in corneal allograft rejection22, we studied the function of sVegfr-2 in a mouse model of corneal transplantation. A single intracorneal administration of dimeric Vegfr-2/Fc more than doubled transplant survival rate (105% increase compared to the IgG/Fc treatment and 101% compared to no treatment; P < 0.05, n = 10–13; Fig. 2f). The dramatic reduction in rejection rate induced by Vegfr-2/Fc administration (75% compared to the IgG/Fc treatment and 74% compared to no treatment) is consistent with the observed reduction in blood vessel and lymphatic sprouting through the donor-recipient interface (Fig. 2g,h). Surprisingly, a single intracorneal administration of monomeric sVegfr-2 induced the same degree of corneal allograft survival as dimeric Vegfr-2/Fc despite reducing sprouting only of lymphatic but not blood vessels through the donor-recipient interface (Fig. 2f-h). These data suggest that impairment of the establishment of the afferent arm of the immune-arc-reflex23,24 alone is sufficient to enhance corneal allograft survival. Further, the rate of allograft survival following a single local administration of sVegfr-2 was at least as great as the rate of survival induced by multiple systemic administrations of a Vegfr-3 antagonist in a prior study25. Apart from reducing lymphangiogenesis, it also is possible that sVegfr-2 promoted allograft survival by inhibiting Vegf-c-induced Vegfr-3 signaling in corneal dendritic cells and preventing their transmigration into the draining lymph node25. Collectively, these data provide a novel experimental strategy to uncouple hemangiogenesis from lymphangiogenesis and a new endogenous therapeutic target for improving survival of this most common solid transplant in humans.
Selective lymphangiogenesis inhibition by sVegfr-2 is consistent with the selective effect of endogenous Vegf-c on the lymphatic vasculature6,13,26-28. However, implantation of pellets containing 160–200 ng of recombinant VEGF-C into the 2 μl mouse cornea has been reported to induce hemangiogenesis29-32. To reconcile these findings, we injected neutralizing anti-Vegf-c antibodies into the cornea of wild-type mice after suture injury and found that this resulted in inhibition of lymphangiogenesis (P < 0.05, n = 8) but not hemangiogenesis (Fig. 3a,b). These data directly demonstrate that endogenous Vegf-c selectively promotes lymphangiogenesis in the cornea, and help explain the selective effect of sVegfr-2. The lack of Vegfr-3 expression by conjunctival and corneal blood vessels33,34 also support a model where endogenous Vegf-c preferentially binds Vegfr-3 on lymphatic vessels over Vegfr-2 on blood vessels in these tissues, based on relative receptor affinities35. Also, Vegf-c is far less efficient at activating Vegfr-2 than Vegf-a36, the principal driver of corneal hemangiogenesis11,32.
We quantified VEGF-C levels following 160 ng VEGF-C pellet implantation and found that this resulted in corneal VEGF-C levels that were 100–125 times greater than the levels observed after suture injury and 45–66 times greater than neonatal levels (Fig. 3c). Such exaggerated levels that are supra-physiological and far greater than those achieved in pathophysiologically relevant states could explain the reported hemangiogenic effects of exogenous VEGF-C in this artificial context.
These selective anti-lymphangiogenic effects of sVegfr-2 also were unexpectedly specific because mbVegfr-2 is capable of binding Vegf-a and promoting hemangiogenesis. To explain this selectivity, we first sought to determine whether sVegfr-2 existed in monomeric or dimeric form because a recombinant form of the ectodomain of mbVegfr-2 has been shown to be a monomer that has little or no affinity for Vegf-a compared to a dimeric recombinant Vegfr-2/Fc fusion protein37-39. Expression studies revealed that sVegfr-2 secreted by mouse corneal epithelial, human embryonic kidney, and Chinese hamster ovary cells, as well as recombinant sVegfr-2, all migrated at equivalent apparent molecular masses under both non-reducing and reducing conditions of western blotting, data consistent with its existence as a monomer (Fig. 3d,e). In contrast, the migration of Vegfr-2/Fc was consistent with it being a dimer.
Next we tested the effects of monomeric sVegfr-2 and dimeric Vegfr-2/Fc on models of corneal neovascularization induced by injury or Vegf-a. Suture-induced corneal hemangiogenesis, which is driven principally by upregulation of endogenous Vegf-a (ref. 32), was inhibited by Vegfr-2/Fc but not sVegfr-2 (Fig. 3f). Similarly, Vegf-a-induced phosphorylation of mbVegfr-2 in porcine aortic endothelial cells was inhibited by Vegfr-2/Fc but not sVegfr-2, and Vegf-a-induced corneal hemangiogenesis was inhibited by Vegfr-2/Fc but not by sVegfr-2 or psVegfr-2 (Fig. 3g,h). These functional data provide a mechanistic basis for the absence of an anti-hemangiogenic effect of endogenous sVegfr-2, and corroborate the previously reported in vitro differential Vegf-a binding avidity between monomeric and dimeric forms of Vegfr-2.
We found sVegfr2 expression by Northern blotting of poly A-enriched RNA isolated from various mouse organs (Supplementary Fig. 11), suggesting that sVegfr-2 might have functions outside the eye. sVegfr2 mRNA was abundant in the alymphatic epidermis of the skin as well as in the hair follicles of wild-type mice (Fig. 4a). sVegfr-2 protein was immunolocalized in the epidermis, hair follicles, and, consistent with its ability to diffuse, also in the dermis of wild-type mice (Fig. 4b). In contrast, mbVegfr-2 was expressed in the skin vasculature, but not in the epithelial cells or hair follicles (Supplementary Fig. 12a). As in the cornea, Vegf-c was expressed in the P0 wild-type mouse skin but undetectable in the adult (Supplementary Fig. 12b). Excess Vegf-c in the skin, achieved by either transgenic overexpression or implantation of overexpressing cells, leads to hyperplasia but not sprouting of lymphatic vessels13,40. To determine the function of sVegfr-2 in the skin, we interbred Vegfr2loxP/loxP mice with K14Cre mice that constitutively and uniformly express Cre recombinase in the epidermis and hair follicles41. This strategy specifically targets sVegfr-2 because the epidermis and hair follicles express sVegfr-2 but not mbVegfr-2. Strikingly, in all P0 K14Cre/Vegfr2loxP/loxP mouse skin, which lacked sVegfr-2 expression (Fig. 4c), there was dramatic enlargement of lymphatic vessels compared to those in littermate control skin (Fig. 4d-f). These dilated skin lymphatics in K14Cre/Vegfr2loxP/loxP mice also were hyperplastic (Fig. 4e,g). However, the density of lymphatic structures, as quantified by branch point analysis, was not greater in K14Cre/Vegfr2loxP/loxP mice (Fig. 4h), just as in mouse skin exposed to excess Vegf-c (refs. 13,40). In contrast to the lymphatic architectural changes, there was no increase in skin blood vessel diameter or density in K14Cre/Vegfr2loxP/loxP mice (Fig. 4d,e). The recapitulation of Vegf-c overexpression induced selective lymphatic hyperplasia in mice lacking sVegfr-2 supports the concept that sVegfr-2 is an in vivo antagonist of Vegf-c.
A protein immunoreactive to an antibody recognizing the amino-terminus of Vegfr-2 has been detected in plasma and shown to be a surrogate biomarker of tumor growth42-44. However, its molecular identity (whether it is encoded by a splice variant of Vegfr2 or derived from ectodomain shedding or proteolytic cleavage of mbVegfr-2) and cellular source have been elusive. Circulating Vegfr-2 in plasma was immunoreactive to both AA21127 and T014 but not an antibody recognizing the carboxyl-terminus of mbVegfr-2 (Fig. 5a), suggesting that it was encoded by the sVegfr2 splice variant. Mouse blood endothelial cell (BEC) lines derived from the microvasculature of the brain, pancreas, or skin all synthesized and secreted sVegfr-2, far in excess of LEC production (Fig. 5b,c). Also sVegfr-2 was detected by immunofluorescence using AA21127 in mouse lung microvasculature, identifying BECs as sources of plasma sVegfr-2 (Fig. 5d). The production of anti-lymphangiogenic sVegfr-2 by BECs could be one of the mechanisms underlying the observation that lymphangiogenesis typically lags behind hemangiogenesis in many neovascular models. Surprisingly, plasma levels of sVegfr-2 were significantly lower in K14Cre/Vegfr2loxP/loxP mice compared to littermate controls (Fig. 5e), suggesting that the skin epithelium also is a source of circulating sVegfr-2. An earlier report showing that K14-driven recombinant soluble Vegfr-3 led to measurable circulating levels of the engineered protein27 supports the concept that endogenous skin-derived sVegfr-2 also can enter the circulation.
We confirmed the existence of sVEGFR2 in humans using RT-PCR to clone the 2159-bp open reading frame of a transcript in human umbilical vein endothelial cells (Fig. 6a), that, as with mouse sVegfr2, resulted from retention of intron 13 also containing an in-frame early termination stop codon in human VEGFR2. Translation of sVEGFR2 results in a protein containing 679 amino acids with a unique 16-aa carboxyl-terminus sequence not present in mbVEGFR-2 and against which we developed a polyclonal antibody (AA21129, Supplementary Fig. 13). sVEGFR-2 was expressed in human cornea with a distribution similar to that in mouse (Fig. 6b), suggesting that it plays a similar role in maintaining an alymphatic state. The potential anti-lymphangiogenic property of sVEGFR-2 in humans was tested in vitro. Lymphangioma is a common disfiguring childhood neoplasia whose etiology is unknown. Human lymphangioma endothelial cells (LaECs) produce VEGF-C and express VEGFR-3 (refs. 45-47), suggesting that their growth could be inhibited by disrupting this axis. Previously we described the isolation of LaECs from axillary lymphangiomas in two human infants47. We found that sVEGFR-2 abolished VEGF-C-induced proliferation of both these LaECs (Fig. 6c,d), raising the exciting possibility that these pediatric tumors could be molecularly targeted.
As an endogenous uncoupler of the intertwined blood and lymphatic circulatory systems, sVegfr-2 represents a new molecular tool to selectively dissect the individual contributions of these parallel vasculatures in development and disease. More importantly the identification of sVegfr-2 can enable new therapeutic strategies by selectively modulating aberrant lymphatic proliferation without encountering the potential adverse effects of non-specific anti-angiogenic therapy.
The importance of pure lymphangiogenesis suppression is nowhere more apparent than in the setting of lymphangioma, which occurs in roughly 1 in 50 children48. These tumors, although classified as benign, can be locally invasive and terribly disfiguring. Their pediatric preponderance highlights the need to develop selective anti-lymphangiogenic therapy. Our finding that sVEGFR-2 can prevent VEGF-C induced proliferation of human lymphangioma endothelial cells raises the exciting possibility that these pediatric tumors could be targeted on a molecular level.
Our discovery of sVegfr-2 in the cornea endorses this tissue as a rich source of angiogenesis modulators, and introduces the yin-yang concept of opposing natural molecular forces that regulate blood vessel growth7 to lymphangiogenesis. Just as sVegfr-1 is required for corneal avascularity11,49, we now show that sVegfr-2 is essential for corneal alymphaticity. The dual presence of sVegfr-1 and sVegfr-2 attests to the premium placed by the cornea on impeding both blood and lymphatic vessel growth to preserve optical clarity and relative immunological privilege, and illustrates the conservation of alternative splicing in generating these tandem traps.
The surprising finding that sVegfr-2 enhanced corneal allograft survival despite not inhibiting hemangiogenesis suggests that impairment of the establishment of the afferent arm of the immune-arc-reflex23,24 alone is sufficient to enhance corneal allograft survival. Interestingly, the rate of allograft survival following a single local administration of sVegfr-2 was at least as great as the rate of survival induced by multiple systemic administrations of a Vegfr-3 antagonist in a prior study25. Apart from reducing lymphangiogenesis, it also is possible that sVegfr-2 promoted allograft survival by inhibiting Vegf-c-induced Vegfr-3 signaling in corneal dendritic cells and preventing their transmigration into the draining lymph node25. In addition to identifying a new endogenous therapeutic target for improving survival of this most common solid transplant in humans, our data also provide a mechanistic basis for evaluating selective lymphangiogenesis suppression in other transplants, such as kidney, whose rejection is associated with lymphangiogenesis50.
Unlike collagen XVIII fragments, semaphorin 3F, and transforming growth factor-β, the only other reported endogenous inhibitors of lymphangiogenesis51-54, native sVegfr-2 is a specific and direct lymphangiogenesis inhibitor that did not inhibit hemangiogenesis in multiple models, which often is intimately intertwined with lymphangiogenesis. One reason for this specificity is that native sVegfr-2 as a monomer, unlike dimeric mbVegfr-2, has poor avidity for Vegf-a. Nevertheless, we cannot exclude a minor anti-hemangiogenic role for sVegfr-2 that might be masked by the dominant role of coexisting molecules such as sVegfr-1 that have much greater affinity for Vegf-a, or in experimental contexts where high levels of exogenous Vegf-c promote hemangiogenesis. A complete molecular resolution of why endogenous sVegfr-2 binds Vegf-c and not Vegf-a must await structure solutions of the Vegfr-2 ectodomain alone and in complex with Vegf-a as well as of Vegf-c. Collectively our findings demonstrate that sVegfr-2 is a broad and non-redundant physiological regulator of lymphatic vessels. Further studies will reveal whether sVegfr-2 can arrest solid tumor lymphangiogenesis and metastasis, which can be promoted by Vegf-c overexpression55-57 and inhibited by blocking Vegf-c activity58,59.
Balb/C, C57Bl/6J, and K14Cre mice were purchased from The Jackson Laboratory. Vegfr2loxP/loxP (flox) mice were generated as described below. LeCre mice16, which constitutively express Cre recombinase in the cornea were a gift of R. Ashery-Padan and P. Gruss via D.C. Beebe. For all procedures, anesthesia was achieved by intraperitoneal injection of 50 mg/kg ketamine hydrochloride (Fort Dodge Animal Health, Wyeth) and 10 mg/kg xylazine (Phoenix Scientific, San Marcos, CA). All animal procedures were approved by the Animal Care and Use Committees at the University of Kentucky or Yokohama City University, and conformed to the Association for Research in Vision and Ophthalmology Statement on Animal Research.
We achieved embryonic conditional genetic ablation of Vegfr2 in the cornea by cross-breeding Vegfr2loxP/loxP mice with LeCre mice16. Alternatively, to target Vegfr2 in the adult mouse cornea, we performed intrastromal injections of naked plasmids11 containing the sequence for Cre recombinase (pCre; 20 μg; gift of R.K. Nordeen, University of Colorado) or an empty plasmid (pNull; 20 μg) in fellow eyes of Vegfr2loxP/loxP or BALB/c mice 3 days prior to suture placement. To achieve embryonic conditional genetic ablation of Vegfr2 in the epidermis, we cross-bred Vegfr2loxP/loxP mice with K14Cre mice that constitutively express Cre recombinase in the skin epidermis and hair follicle.
Eyes were enucleated from wild-type and LeCre/Vegfr2loxP/loxP mice and fixed in 3.5% glutaraldehyde/4% paraformaldehyde for 2 h followed by preparation of uranyl acetate- and lead citrate-stained ultrathin sections for transmission electron microscopy studies (Phillips Biotwin).
Two intrastromal 11-0 sutures (Mani, Japan) were placed in the mouse cornea 180° from each other. All sutures were placed in the midpoint between the limbus and the corneal apex. They were left in place for up to 14 days. Naked plasmids coding for Cre-recombinase (gift of R.K.Nordeen, University of Colorado), mouse sVegfr-2 (psVegfr-2), were utilized for in vivo enforced expression studies as shown earlier11. To block Vegfr-3 activity, daily intraperitoneal injections of Vegfr-3 tyrosine kinase inhibitor (MAZ51, EMD Chemicals, 8 mg/kg) were performed in pCre treated Vegfr2loxP/loxP mice after corneal suture placement for 14 consecutive days. Vehicle only (DMSO) was administered as control treatment.
Neutralizing rabbit anti-Vegf-c (Angio-Proteomie, pV1006R-r, 16 μg) antibody and control rabbit IgG (16 μg) were injected intrastromally into fellow eyes on the day of suture placement and every 3 days thereafter for 14 days.
Porcine aortic endothelial cells stably transfected with Vegfr-2 (PAE-KDR) were cultured in DMEM (Invitrogen) containing 10% FBS, penicillin G (100 units/ml), streptomycin sulfate (0.1 mg/ml) (all from Sigma Aldrich) at 37 °C, 10% CO2 and 90% room air. Upon attaining 80% confluence these cells were serum starved for 24 h, after which time they were exposed to Vegf-a (R&D Systems, 2.5 nM) or media alone (negative control) and monomeric sVegfr-1 (Reliatech, 15 nM) , monomeric sVegfr-2 (Reliatech, 75 nM), dimericVegfr-2/Fc (R&D Systems, 75 nM) or IgG/Fc control (Jackson ImmunoResearch, 75 nM) for 15 min. Cell lysates were processed in RIPA buffer and resolved by SDS 8% polyacrylamide gradient gel electrophoresis (PAGE) and transferred to nitrocellulose (NC) membranes. Immunoblotting was performed using a rabbit antibody against phosphorylated Vegfr-2 (1:500, Upstate, 36-019), and loading was assessed using an antibody against Vegfr-2 (1:1000, R&D Systems, AF357) and Vinculin (1:2000, Sigma, V4139).
Vegfr-2/Fc or sVegfr-2 were resolved by SDS 10% PAGE and stained using SilverSNAP (R) Stain (Pierce) according to manufacturer instructions.
Immunostaining and flat mounting were performed as previously shown11 using rabbit anti-mouse Lyve-1 antibody (Abcam; 1:333), rat anti-mouse Cd31 antibody (BD Biosciences; 1:50), rat-anti-mouse Meca-32 antibody (BD Biosciences; 1:10), goat anti-mouse Lyve-1 antibody (R&D, 1:100), and rabbit anti-mouse Prox-1 (Angiobio;1:500) for 48 h at 4 °C. Alexa Fluor 488 (goat anti-rabbit; 1:200) and Alexa Fluor594 (goat anti-rat; 1:200), Alexa Fluor488 (donkey anti-goat; 1:200, Invitrogen), or Cy3 conjugated donkey anti-rabbit (Jackson ImmunoResearch; 1:400) for 24 h were used as secondary antibodies. Tissue mounts were visualized under fluorescent microscopy (Leica SP-5) and analyzed with ImageJ (NIH). Mean percentage Lyve-1+ (lymphatic vessels) or Cd31+/Lyve-1– (blood vessels) areas were calculated for corneal flat mounts and skin whole mounts using ImageJ software. The number of Prox1+ nuclei within Lyve-1+ skin lymphatic vessels were counted in 12 random fields and expressed as LEC density per 100 μm. The density of lymphatic structures was determined by counting lymphatic vessel branch points per unit area (750 μm × 750 μm) on standard low-magnification Lyve-1-stained images of the mouse skin.
Mouse blood endothelial cells from brain (Bend3, gift of C.D. Kontos, Duke University), pancreas (MS1) and skin (Py4, both gifts from J.L. Arbiser, Emory University) and mouse lymphatic endothelial cells (mLEC; ref. 60) were cultured in DMEM (Invitrogen) containing 10% FBS, penicillin G (100 units/ml), streptomycin sulfate (0.1 mg/ml) (all from Sigma Aldrich) at 37 °C, 10% CO2 and 90% room air. Upon attaining 80% confluence these cells were serum starved for 24 h when supernatant fractions were collected for sVegfr-2 protein quantification by ELISA. Total RNA was also extracted for RT-PCR.
Cultured human lymphatic microvascular endothelial cells (Cambrex) were maintained in EGM-2 MV (Clonetics) supplemented with 10%FBS and antibiotics at 37 °C under 5% CO2. Once confluent cells were plated at 96-well plate at a density of 30,000 cells/well. Cells were serum starved for 2 h and then exposed to media alone (MCDB 131 + 5% FBS), VEGF-C (200 ng/ml; R&D Systems) enriched media or VEGF-C enriched media with sVEGFR-2 (13.6 μg/ml). This concentration of sVEGFR-2 corresponds to a molar ratio of approximately 11-13:1 compared to VEGF-C, and is therefore physiological in view of our finding that sVegfr-2 levels in the cornea are ~20-fold higher than Vegf-c levels on a molar basis. Proliferation was quantified using BrdU uptake (Chemicon International) at 36 h after incubation with VEGF-C.
Lymphatic endothelial cells were isolated from lymphangiomas from two individuals, 4 and 10 months of age47, with the approval of the ethics committee of the Georg-August-University and with the informed consent of the patients’ parents. These cells were grown in EGM2-MV growth media containing 5% FBS. Cells were passaged onto a 96-well plate (5000 cells/well) in basal media (MCDB131) containing 2% FBS, and allowed to adhere overnight. Cultures were then treated with 200 ng/ml recombinant human wild-type (WT) VEGF-C (Reliatech) alone or together with 25 μg/ml of sVegfr-2 (Reliatech) in basal media with 0.1% FBS. Cell proliferation was measured after 24 h by using BrdU cell proliferation kit (Chemicon) according to the manufacturer's instructions.
Mouse cornea lysates as well as cell culture supernatant were resolved by SDS 8% or 4-20% polyacrylamide gradient gel electrophoresis (PAGE) and transferred to nitrocellulose (NC) membranes. Immunoblotting was performed using a rabbit antibody against the amino terminus of mouse Vegfr-2 (1:1000; clone T014; ref. 12), custom made sVegfr-2 specific antibody (1:1000; AA21127) and loading was assessed using rabbit antibody against human GAPDH (1:2000; Abcam).
Recombinant mouse sVegfr-2 (2 μg) was incubated with mouse recombinant Vegf-c (100 ng, Biovision) in PBS at 4 °C for 1 h. Immunoprecipitation was carried out with 2 μg of an anti-Vegfr-2 antibody or an isotype control IgG. Immobilized protein A/G beads (20 μl, Pierce) were used for precipitation. Samples were boiled, resolved by SDS-PAGE with respective positive controls (rsVegfr-2 and rVegf-c) and transferred to a NC membrane. Rabbit antibodies against Vegf-c (1:1000, Santa Cruz) and Vegfr-2 (1:1000, T014) were used to probe for Vegf-c and Vegfr-2 respectively. Immunoprecipitation of Vegfr-2 from mouse plasma was performed as described previously by others 42. Rabbit antibody against Vegfr-2 (T014, 2 μg) was employed for immunoprecipitation. T014 (1:1000), goat anti-Vegfr-2 (C) antibody (1:1000, Abcam-ab2349) or rabbit anti-sVegfr-2 antibody (1:1000, AA21127) was used for immunoblotting. Mouse LECs were incubated with media only or Vegf-c (200 ng/ml, Biovision) with or without sVegfr-2 (13.8 μg/ml) or Vegfr-1/Fc (20 μg/ml, R&D Systems) for 15 min. The sVegfr-2 and Vegfr-1/Fc concentrations are equimolar. The lysates were immunoprecipitated with anti-Vegfr-3 antibody (Santa Cruz, C-20), immunoblotted with anti-phosphotyrosine (Millipore, 4G10, 1:1000), and reblotted with anti-Vegfr-3 antibody (eBioscience, AFL4, 1:500).
Mann Whitney U test with Bonferroni correction was used for statistical comparison of multiple variables. Comparison of corneal transplant survival was performed by Kaplan-Meier Survival. The null hypothesis was rejected at P < 0.05.
We thank various collaborators for gifts of transgenic mice, cell lines, and plasmids; R. King, L. Xu, M. McConnell, K. Emerson, A. Blanford, M. Baker, S. Furlow, M. LaFalce and C. Long for technical assistance, and R. Mohan, F. Cambi, S. Bondada, M. Detmar, M.W. Fannon, T.V. Getchell, R.K. Jain, T.S. Khurana, B.J. Raisler, J.E. Springer, P.A. Pearson, J.G. Woodward, A.M. Rao, G.S. Rao, K. Ambati, and L. Garcia for valuable discussions. This work was supported by U.S. National Institutes of Health/National Eye Institute grants EY015422, EY018350 and EY018836 to J.A. and EY017182 and EY017950 to B.K.A.; Research to Prevent Blindness Lew R. Wassermann Merit Award (J.A.), Physician Scientist Award (J.A., B.K.A.), Medical Student Fellowship (R.J.C.A.), and departmental unrestricted grant (J.A.); University of Kentucky University Research Professorship (J.A.); Fight for Sight (R.J.C.A.); Japan Society for the Promotion of Science for Young Scientists (A.T.); VA Merit Award (B.K.A.); Department of Defense (B.K.A.). J.A. is also supported by Doris Duke Distinguished Clinical Scientist Award, Burroughs Wellcome Fund Clinical Scientist Award in Translational Research, and Dr. E. Vernon Smith and Eloise C. Smith Macular Degeneration Endowed Chair. The sVegfr2 and sVEGFR2 sequences are deposited in GenBank as accession numbers 1113431 and 1206797.