We have shown that lymphatic vessels positive for the key lymphatic endothelial cell markers Lyve-1, podoplanin and VEGFR-3 are largely restricted to the connective tissue between the longitudinal and circular layers of the myometrium in the mouse uterus; only very few lymphatic vessels were present within the mouse endometrium and when observed were nearly always situated adjacent to the myometrial-endometrial border. We have also demonstrated that the presence of tumor cells over-expressing human VEGF-D has differential effects on the uterine vasculature in a mouse model of endometrial cancer. Unexpectedly, 293EBNA cells expressing human VEGF-D did not stimulate growth of new (lyve-1 positive) lymphatic vessels into the mouse endometrium, although they did stimulate changes in the pre-existing lymphatic vessels of the myometrium and blood vessels of the endometrium; no significant changes were observed in the myometrial blood vessels.
Until relatively recently, it was necessary to conduct detailed histological, ultrastructural or functional studies to describe the distribution of lymphatic vessels. The available techniques all have associated limitations and it is not always possible to accurately differentiate lymphatic vessels from blood vessels or tissue spaces. Specific lymphatic endothelial cell markers (including Lyve-1, podoplanin and VEGFR-3) are now available enabling detailed studies of lymphatic growth and function. In the current study, we have used three markers (Lyve-1, podoplanin, VEGFR-3) to examine the distribution of lymphatics in the mouse uterus. Our observation that the majority of lymphatic vessel profiles positive for Lyve-1, podoplanin and VEGFR-3 are situated within the myometrium is consistent with several earlier detailed histological and functional studies in mice, rats, rabbits and sheep [35–40]. In these studies, a well developed lymphatic vasculature was observed in the connective tissue separating the two myometrial muscle regions with smaller lymphatic capillaries extending into the longitudinal and circular muscle layers. Most of these studies report few lymphatic vessels in the endometrium. However, some studies report contrary results suggesting the presence of extensive endometrial lymphatic capillaries in mice and rats [35, 40]. There have also been conflicting results in studies using lymphatic-specific endothelial cell markers. While Donoghue et al.  describe a distinct population of lymphatic vessels in the human endometrium based on podoplanin (D2-40) immunohistochemistry, other studies using Lyve-1 report an absence of lymphatics in the endometrium [33, 34]. It is apparent that caution is still required when examining lymphatic vessels within uterine tissues, even when lymphatic endothelial cell markers are used. In addition, it will be important for future researchers to investigate the functional differences in uterine vessel populations exhibiting differential lymphatic-endothelial specific marker expression.
In this study, we used a VEGF-D over-expressing tumor cell line as a model of VEGF-D over-expression in endometrial cancer. We chose to use the 293EBNA-VEGF-D cells as they had previously been shown to induce lymphangiogenesis and angiogenesis in NOD/SCID mice . In addition, the original 293EBNA cells do not express detectable levels of VEGF-A, VEGF-C or VEGF-D, reducing any confounding effects of other key VEGF family members on our analyses. The 293EBNA-VEGF-D cells express full-length VEGF-D (53 kD), some of which is proteolytically processed to intermediate and fully processed forms with higher affinities for its cognate receptors VEGFR-2 and VEGFR-3 on endothelium. The VEGF-D produced by these cells was previously shown to induce tumor angiogenesis and lymphangiogenesis and to promote tumor metastasis to local lymph nodes when cells were injected subcutaneously . VEGF-D has also been shown to induce lymphangiogenesis in several other models following cell-mediated over-expression , adenoviral delivery [41, 42] and transgenic expression .
Although the presence of VEGF-D-expressing 293EBNA tumor cells did not induce endometrial lymphangiogenesis, they did cause differential effects on the existing uterine vasculature. Increased proliferation and vessel size was observed in the myometrial lymphatic and endometrial blood vessels. Similar vessel enlargement has been observed in rabbit and mouse hind limb [24, 42], rat cremaster muscle , pig heart  and mouse skin  in response to VEGF-D. In contrast, no significant changes were noted in the myometrial blood vessels. This differential response to VEGF-D might be due to differences in the local tissue microenvironment or the particular response of these tissue-specific vessels. Adenoviral VEGF-D (AdvVEGF-D) over-expression in mouse skin induces increased angiogenesis and lymphangiogenesis with vessel enlargement [41, 42], while in the lung AdvVEGF-D induces lymphangiogenesis without angiogenesis . Alternatively, the differential responses may reflect the location of blood or lymph vessels on the vascular tree. The small endometrial capillaries, which have less mural cell (pericytes and vascular smooth muscle cells) support than the more stable blood arterioles present in the connective tissue of the myometrium , are likely to be more responsive to angiogenic promoters. However, this latter hypothesis does not explain the lack of response in small capillaries present within the muscle itself.
It has been proposed that VEGF-D is crucial to carcinoma-associated lymphangiogenesis, as well as the process of metastasis [12, 25]. In the current study, 293-EBNA-VEGF-D cells or tumor emboli were observed within myometrial lymphatic vessels. In contrast, no invasion of lymphatic vessels was observed in mice receiving control 293EBNA cells. The presence of emboli in the vasculature is a prognostic factor in several cancers, including endometrial cancer (see [8, 46, 47] and references therein). Invasion of the vasculature is also thought to be an early step in the metastatic process. The mechanism by which VEGF-D might induce vascular invasion and ultimately metastasis is not known. Although VEGF-D over-expression caused enlargement and proliferation of uterine lymphatic vessels, these effects in themselves are not sufficient to explain the observed migration of tumor cells into uterine vessels. Future research will be needed to determine whether vessel invasion reflects characteristics of the endometrial VEGF-D expressing tumor cells or because of changes induced in the vessels affected.
In addition to our focus on endometrial cancer, the observed effects of VEGF-D over-expression on uterine vasculature raises questions about the function of this growth factor in normal uterine remodelling. In this study, no new growth of lymphatic vessels into the endometrium was induced. A possible explanation for this lack of growth is that lymphangiogenesis is actively inhibited within the endometrium. Such inhibition has been observed in the cornea, which must remain avascular to maintain the transparency required for vision. It has been shown that constitutive expression of VEGFR-3 on the corneal epithelium acts as a sink for VEGFR-3 ligands, thereby suppressing vascular growth into the cornea . In addition, a recent study has identified a soluble splice variant of VEGFR-2 in mice that inhibits lymphangiogenesis by blocking VEGF-C. This isoform specifically blocks injury-induced lymphangiogenesis, but not hemangiogenesis, in the mouse cornea . Several studies have described varying amounts of non-lymphatic VEGFR-3 immunostaining within the endometrial epithelium and stroma [50–52], however, the activity of this non-lymphatic VEGFR-3 remains to be elucidated. Whether a soluble isoform of VEGFR-2 is expressed in the uterus has not yet been investigated.
In the current study, mouse uterus was treated with human VEGF-D which interacts with both mouse and human VEGFR-2 and VEGFR-3. It should be noted, however, that mouse VEGF-D does not bind with VEGFR-2  and it is unlikely that similar blood vessel responses would have been observed if tumour cells had been transfected with mouse VEGF-D. However, as human VEGF-D does bind with human VEGFR-2, future studies need to address the functional role of VEGF-D in human endometrium with emphasis on the differential effects on different vascular components within the tissue.
Future studies should also consider the role of VEGF-C in endometrial vascular remodeling. As with VEGF-D, VEGF-C can interact with both VEGFR-2 and VEGFR-3, and has been shown to induce both angiogenesis and lymphangiogenesis in various tissue and pathology models [53–55]. Depending on the nature of the ligand and the presence of co-receptors, the VEGF receptors can form both homo and heterodimers with subsequent auto-phosphorylation of the receptor complex and downstream signaling; the associated variation in signaling pathways associated with homo versus heterodimers has yet to be elucidated (reviewed in [56, 57]). The specific interactions of VEGF-D with VEGF receptors has not been examined in the current model, but future research in this area will highlight the mechanisms by which VEGF-D acts to promote lymphangiogenesis and metastasis in endometrial cancer. While this study has focused on the effects of VEGF-D overexpression on vessel morphology and endothelial cell proliferation, future research will need to consider the other processes influenced by VEGF receptor activation (cell survival, cell migration, vessel permeability). Modification of any of these processes would also likely contribute to VEGF-D's role in endometrial cancer progression.
The NOD SCID mice used in the current study have various defects of immune function including decreased numbers of circulating T lymphocytes, B lymphocytes and natural killer cells. This means that any effects of VEGF-D overexpression on the uterine lymphocyte population could not be investigated in this model. The profiles of VEGF-D and VEGFR-3 expression may also vary in the uterus of these mice relative to those in the C57/CBA mice used for the initial analysis. Despite the above, our study has shown a clear effect of VEGF-D over-expression on uterine vasculature.