Prox1 is a master control gene in the program specifying lymphatic endothelial cell fate
Abstract
Early during development, one of the first indications that lymphangiogenesis has begun is the polarized expression of the homeobox gene Prox1 in a subpopulation of venous endothelial cells. It has been shown previously that Prox1 expression in the cardinal vein promotes and maintains the budding of endothelial cells that will form the lymphatic vascular system. Prox1-deficient mice are devoid of lymphatic vasculature, and in these animals endothelial cells fail to acquire the lymphatic phenotype; instead, they remain as blood vascular endothelium. To investigate whether Prox1 is sufficient to induce a lymphatic fate in blood vascular endothelium, Prox1 cDNA was ectopically expressed by adenoviral gene transfer in primary human blood vascular endothelial cells and by transient plasmid cDNA transfection in immortalized microvascular endothelial cells. Transcriptional profiling combined with quantitative real-time reverse transcription-polymerase chain reaction and Western blotting analyses revealed that Prox1 expression up-regulated the lymphatic endothelial cell markers podoplanin and vascular endothelial growth factor receptor-3. Conversely, genes such as laminin, vascular endothelial growth factor-C, neuropilin-1, and intercellular adhesion molecule-1, whose expression has been associated with the blood vascular endothelial cell phenotype, were down-regulated. These results were confirmed by the use of specific antibodies against some of these markers in sections of embryonic and adult tissues. These findings validate our previous proposal that Prox1 is a key player in the molecular pathway leading to the formation of lymphatic vasculature and identify Prox1 as a master switch in the program specifying lymphatic endothelial cell fate. That a single gene product was sufficient to re-program the blood vascular endothelium toward a lymphatic phenotype corroborates the close relationship between these two vascular systems and also suggests that during evolution, the lymphatic vasculature originated from the blood vasculature by the additional expression of only a few gene products such as Prox1. © 2002 Wiley-Liss, Inc.
INTRODUCTION
The lymphatic vascular system consists of a dense network of blind-ending, thin-walled lymphatic capillaries and collecting lymphatics that drain extravasated protein-rich fluid from most organs and that transport the lymph by means of the thoracic duct to the venous circulation (Witte et al., 2001). A century ago, Sabin proposed that the lymphatic system develops by sprouting of endothelial cells from embryonic veins, leading to the formation of primitive lymph sacs from which lymphatic endothelial cells then sprout into surrounding organs to form mature lymphatic networks (Sabin, 1902, 1904). However, in contrast to the rapid progress made in elucidating the formation and molecular control of the blood vascular system (Gale and Yancopoulos, 1999; Carmeliet, 2000), the mechanisms controlling the normal development of lymphatic vessels have remained poorly understood.
Recent studies in Prox1-null mice have provided critical evidence supporting Sabin's original hypothesis (Wigle and Oliver, 1999). The homeobox gene Prox1 was originally cloned by homology to the Drosophila gene prospero, a gene required for the normal differentiation of neuronal lineages (Hassan et al., 1997). In mice, Prox1 is expressed in the developing central nervous system, retina, lens, pancreas, liver, and heart (Oliver et al., 1993). Beginning at embryonic day (E) 9.5 of mouse development, Prox1 expression is specifically localized in a subpopulation of yet uncommitted endothelial cells present on one side of the anterior cardinal vein (Wigle and Oliver, 1999). At this stage, the venous endothelium also expresses the hyaluronan receptor LYVE-1, a CD44 homologue (Banerji et al., 1999), and vascular endothelial cell growth factor receptor-3 (VEGFR-3)—a receptor for the lymphangiogenic factors vascular endothelial growth factor-C (VEGF-C) and VEGF-D (Wigle and Oliver, 1999; Alitalo and Carmeliet, 2002; Wigle et al., 2002). The expression of both of these receptors later becomes restricted to lymphatic endothelium (Kaipainen et al., 1995; Wigle et al., 2002). The polarized expression of Prox1 is one of the earliest indications that the molecular program leading to lymphatic vasculature formation has started. This initial expression is followed by polarized budding and migration of Prox1-expressing lymphatic progenitor endothelial cells that progressively down-regulate the expression of blood vascular genes such as CD34 and laminin, while increasing the expression levels of lymphatic markers such as VEGFR-3 (Wigle et al., 2002). This expression pattern is an indication that lymphatic vasculature formation is proceeding in the developing embryo. Importantly, in Prox1 null mice, budding and sprouting of lymphatic endothelial cells from the veins is arrested at around E11.5–E12.0 and these mutant mice represent, until now, the only known animal model completely devoid of a lymphatic vascular system (Wigle and Oliver, 1999). Together, these findings highlight the essential role of Prox1 in the specification of the lymphatic fate and subsequent development of the lymphatic vasculature (Wigle and Oliver, 1999; Wigle et al., 2002), and they suggest that its functional role is established by promoting a lymphatic fate in otherwise undifferentiated endothelial cells. There is ample agreement that the blood and lymphatic vasculature systems are highly similar: both are lined by endothelial cells and share the expression of many common markers (Oliver and Detmar, 2002). Could it be possible that the lymphatic system has a venous origin during evolution? This concept would facilitate the understanding of the close morphologic relationship between both systems. Moreover, the expression of a few additional gene products might not only be sufficient to promote the lymphatic fate in uncommitted venous endothelial cells but also to convert, whenever required, the blood vascular phenotype into a lymphatic phenotype. In this study, we determined that Prox1 is a master regulator of this proposed molecular switch; its activity is sufficient to override the blood vasculature phenotype in primary human dermal microvascular endothelial cells by promoting a lymphatic endothelial phenotype instead.
RESULTS AND DISCUSSION
Expression of the homeobox gene Prox1 is required for the early development of the lymphatic system from embryonic veins (Wigle and Oliver, 1999). In Prox1 null embryos, budding Prox1-deficient endothelial cells maintain a default blood vascular phenotype instead of acquiring the normal lymphatic phenotype (Wigle et al., 2002). Although these results demonstrated that Prox1 expression was necessary for the induction and development of the lymphatic vasculature, the question as to whether its expression was also sufficient to promote a lymphatic fate remained unanswered (Oliver and Detmar, 2002). Therefore, to determine whether expression of Prox1 might indeed be sufficient for re-programming blood vascular endothelial cells, we transduced primary human dermal microvascular endothelial cells (HDMEC) with an adenovirus containing human Prox1 cDNA or with a control adenovirus containing the gene for green fluorescent protein (GFP). Cultured HDMEC are derived from cutaneous blood vessels (Richard et al., 1998) and do not normally express any detectable amounts of Prox1 (Fig. 1E,G). At 48 hr after infection, both Prox1- and control virus-transduced HDMEC maintained their characteristic cobble stone-like morphology in two-dimensional cell culture (Fig. 1A,C). Efficient adenoviral gene transfer was confirmed by the detection of high levels of GFP expression in more than 90% of both Prox1 and control virus-transduced HDMEC (Fig. 1B,D). Northern blot analyses detected Prox1 mRNA expression already 3 hr after adenoviral Prox1 gene transfer, with steadily increasing expression levels during the first 48 hr (Fig. 1E). As expected, Prox1 mRNA expression was not detected in HDMEC transduced with control virus (Fig. 1E). This result validates our cell isolation protocol by confirming that no lymphatic endothelial cells were present in HDMEC cultures. Quantitative real-time reverse transcriptase-polymerase chain reaction (RT-PCR) using Prox1-specific TaqMan probes further demonstrated a steady, time-dependent increase of Prox1 mRNA expression levels after adenoviral Prox1 gene transfer (Fig. 1F). Western blot analysis of cell lysates detected high levels of Prox1 protein in Prox1-transduced HDMEC cultures after 48 hr, whereas it was undetectable in control transfected cultures (Fig. 1G). Taken together, these results confirmed the efficiency of the viral transduction assay and the absence of lymphatic endothelial cells in cultured HDMEC.
Primary human dermal microvascular endothelial cells (HDMEC, passage 7) were transduced with either control adenovirus (A,B) or with adenovirus containing human Prox1 cDNA (C,D). At 48 hr after infection, HDMEC maintained their characteristic cobble stone-like phenotype (A,C). Efficient transduction was confirmed by strong expression of green fluorescent protein (B,D). A,C: brightfield photomicrographs, ×10. E: Northern blot demonstrates time-dependent increase of Prox1 mRNA expression in HDMEC transduced with Prox1 adenovirus (P), whereas no Prox1 expression was detected in control cultures (C). F: Quantification of the ratio of Prox1 mRNA expression in Prox1-adenovirus–transduced HDMEC vs. control virus transduced HDMEC (blue), using TaqMan real-time reverse transcriptase-polymerase chain reaction. Time-dependent induction of Prox1 mRNA expression was also obtained in immortalized human microvascular endothelial cells (HMEC-1) that were transiently transfected with pcDNA vector containing human Prox1 cDNA (red). G: Western blot analysis of cell lysates obtained from primary HDMEC confirmed efficient Prox1 protein expression in cells transduced with Prox1-adenovirus but not in cells transduced with control virus (1) at 48 hr after infection.
As an initial approach to identify possible gene expression changes promoted by Prox1 transduction of HDMEC, we performed differential gene array analyses of Prox1-transduced and of control virus-transduced HDMEC cultures, by using the human U135 Affymetrix gene arrays followed by data analysis with the Affymetrix Gene Spring version5 software. These studies were focused on the comparative expression analysis of genes with previously reported expression in blood vascular and lymphatic endothelium. We found that adenoviral Prox1 gene transfer was sufficient to induce expression of the two well-characterized lymphatic markers podoplanin (Breiteneder-Geleff et al., 1999; also known as T1-alpha; Rishi et al., 1995) and VEGFR-3 (Flt4) in the transduced blood vascular endothelium (Table 1). Conversely, several genes whose expression has been associated with the blood vascular endothelial cell phenotype, such as endoglin, laminin, intercellular adhesion molecule-1 (ICAM-1), low density lipoprotein receptor, neuropilin-1, thrombomodulin, VEGFR-2, and VEGF-C, were strongly repressed (Table 1). These results were confirmed by the use of quantitative RT-PCR assays of total RNA extracted from Prox1- or control-transduced HDMEC (Table 1). No major changes of expression of the pan-endothelial marker CD31 (PECAM-1) were observed. Western blot analyses of HDMEC cell extracts confirmed that forced Prox1 expression in HDMEC induced protein expression of the lymphatic markers podoplanin and VEGFR-3/Flt-4 but repressed the expression levels of neuropilin-1 and ICAM-1 (Fig. 2). To further confirm the biological effects of Prox1 expression, we also transiently transfected the immortalized human microvascular endothelial cell line HMEC-1 with a human Prox1 cDNA vector. Transient Prox1 transfection resulted in strong, time-dependent up-regulation of Prox1 mRNA expression (Fig. 1F) and induced changes in the expression of blood vascular and lymphatic markers similar to those observed after adenoviral gene transfer to primary vascular endothelial cells (Table 1).
| Gene names | Accession no. | Gene array adenovirus | Real-time adenovirus | RT-PCR transient |
|---|---|---|---|---|
| Genes repressed by Prox1 | ||||
| Activin A receptor | NM_001105.2 | −2.64 | −3.09 | ND |
| CD31 | M37780.1 | −2.46 | 1.13 | −2.67 |
| CD34 | M81104.1 | −1.41 | 1.07 | ND |
| Connexin 43 | NM_000165.2 | −1.41 | −2.85 | −1.76 |
| CX3C | U84487 | −6.06 | −5.21 | ND |
| Endoglin | NM_000118.1 | −1.52 | −1.71 | −1.07 |
| ICAM-1 | NM_000201.1 | −13.00 | −47.18 | −1.67 |
| VEGFR-2/KDR | NM_002253.1 | −3.73 | −3.36 | −10.34 |
| Laminin (b1 chain) | M20206.1 | −2.46 | −1.96 | −2.57 |
| LDL receptor | NM_003693.1 | −1.74 | −3.36 | ND |
| Neuropilin-1 | BE620457 | −5.28 | −3.27 | −1.84 |
| PAI-1 | NM_000602.1 | −4.29 | −4.12 | ND |
| PPAR-Gamma | NM_016109.1 | −2.30 | −4.98 | −1.42 |
| Syndecan-2 | AI380298 | −3.03 | −2.21 | −3.48 |
| Thrombomodulin | NM_000361.1 | −1.74 | −2.13 | −2.22 |
| *UPAR | AY029180.1 | −13.93 | −6.45 | −1.37 |
| VE-cadherin | NM_001795.1 | −2.46 | −1.65 | −3.81 |
| VEGF-C | U58111.1 | −1.41 | −3.03 | ND |
| Genes induced by Prox1 | ||||
| Podoplanin | NM_006474.1 | 1.30 | 2.26 | ND |
| VEGFR-3 (Flt-4) | X69878 | 2.7 | 2.1 | ND |
- a ICAM-1, intercellular adhesion molecule-1; VEGFR-2, vascular endothelial growth factor receptor-2; LDL, low density lipoprotein; UPAR, urokinase plasminogen activator receptor.
Western blot analysis demonstrates induction of podoplanin and vascular endothelial cell growth factor receptor-3 (Flt-4) after adenoviral Prox1 gene transfer (Adenovirus) to primary human dermal microvascular endothelial cells and after transient Prox1 transfection of immortalized human microvascular endothelial cells (Transient). In contrast, Prox1 repressed expression of neuropilin-1 and intercellular adhesion molecule-1 (ICAM-1). P, Prox1-expressing cells; C, control cells. Molecular masses are indicated at the left.
We next investigated the in vivo expression of selected lymphatic and blood vascular genes identified by gene expression profiling. In adult murine skin, all Prox1-positive lymphatic endothelial cells also expressed podoplanin (Fig. 3A–C) and VEGFR-3 (data not shown); all podoplanin-positive lymphatic endothelial cells also coexpressed the lymphatic-specific (Prevo et al., 2001) hyaluronan receptor LYVE-1 (Fig. 3D–F). In contrast, neuropilin-1 was found to be expressed on blood vascular endothelium but was undetectable on LYVE-1–positive lymphatic vessels (Fig. 3G–I). Moreover, the vascular adhesion molecule ICAM-1 was detected on blood vessels in inflamed skin (Fig. 3J) but was completely absent from LYVE-1–positive lymphatic vessels (Fig. 3K,L).
Differential immunostains for lymphatic and blood vascular markers in adult murine skin. A–C: All Prox1-positive lymphatic endothelial cells also express podoplanin. D–F: Coexpression of podoplanin and the hyaluronan receptor LYVE-1 in cutaneous lymphatic vessels. Nuclear Hoechst counterstain (blue) was used to depict skin morphology in F. G–I: Neuropilin-1 is selectively expressed by blood vessels but not by LYVE-1–positive lymphatic vessels. J–L: Mutually exclusive expression of intercellular adhesion molecule-1 (ICAM-1) and LYVE-1 in cutaneous blood vessels and lymphatics in inflamed murine skin. Green signals in K were pseudocolored (pink) for merged image.
The process of lymphatic vasculature development is initiated by the budding of a subpopulation of venous endothelial cells (Wigle and Oliver, 1999). As embryogenesis progresses, the level of expression of blood and lymphatic vasculature markers is likely to reflect the state of differentiation of endothelial cells toward the lymphatic phenotype (Oliver and Detmar, 2002; Wigle et al., 2002). VEGFR-3, a gene that is expressed at similar levels in blood and lymphatic vasculature in early embryos, becomes down-regulated in the blood vasculature later on (Kaipainen et al., 1995; Wigle et al., 2002). Similar to Prox1, LYVE-1 expression is restricted to lymphatic endothelial cells already in the early embryo (Wigle et al., 2002). Therefore, we next studied the lineage-specific expression of selected Prox1-modulated genes during embryonic development of the lymphatic system by using heterozygous Prox1-deficient LacZ knock-in mice (Wigle et al., 2002). The endothelial junction molecule CD31 is a general marker for blood and lymphatic endothelial cells (Fig. 4A); and this was reflected by the lack of any major changes in its expression levels after the forced expression of Prox1 in the cultured cells. Unlike blood vasculature, lymphatic capillaries and vessels lack a distinct continuous basal membrane (Ezaki et al., 1990). These differences are normally reflected by the very weak to absent levels of laminin expression observed in the early developing lymphatic system (Fig. 4B). In accordance, levels of laminin expression were found to be down-regulated in the cultured Prox1 transduced vascular endothelial cells. Finally, at E12.5, podoplanin expression is clearly detected in the Prox1-positive lymphatic progenitors located in the cardinal vein as well as in those budding from the vein (Fig. 4C); this is also in agreement with the observed induction of its expression by Prox1 in cultured blood vascular endothelium.
A comparison of selected vascular and lymphatic endothelial markers during embryonic development reflects the modulation of expression observed in human blood vascular endothelial cells infected with Prox1 cDNA. A: At embryonic day (E) 10.5, pan-endothelial marker CD31 (PECAM-1) is expressed in all endothelial cells. B: Conversely, the expression of laminin is barely detectable on endothelial cells of the cardinal vein and budding lymphatic endothelial cells, indicated by Prox1/β-gal expression (green). C: Podoplanin expression is up-regulated on lymphatic endothelial cells at E12.5. At this stage, podoplanin (red) is expressed on endothelial cells of the cardinal vein which express Prox1 (green) and also on budding lymphatic endothelial cells (arrowheads). CV, cardinal vein; DA, dorsal aorta; NT, neural tube.
These results provide the first direct evidence that the homeobox gene Prox1 is sufficient to induce the lymphatic phenotype in blood vascular endothelial cells by initiating the lymphatic lineage-specific gene program in conjunction with the repression of genes specifically associated with blood vascular endothelium. These findings, together with those reporting the absence of a lymphatic vascular system in Prox1-null mice (Wigle and Oliver, 1999), position Prox1 as a master regulator in the molecular pathway leading to the formation of lymphatic vasculature. That a single gene product was sufficient to re-program the vascular endothelium toward a lymphatic phenotype does not only corroborate the close relationship between these two vascular systems, but also suggests that during evolution, the lymphatic vasculature originated from the blood vasculature by the expression of only a few additional gene products such as Prox1. Remarkably, Prox1 expression was able to re-program the phenotype of endothelial cells that had already undergone blood vascular differentiation and that were beyond the embryonic stage of “lymphatic competence” (Oliver and Detmar, 2002), suggesting that, in the future, it could be possible to convert blood vessels into lymphatic vessels even in adult organisms by selective gene transfer. In addition, these findings also suggest that, at later stages, lymphatic outgrowth occurs not only from preexisting lymphatic vessels but also by transdifferentiation of surrounding blood vessels. Ongoing studies will reveal whether ectopic expression of Prox1 in blood vascular endothelium will also be able to re-program the lineage-specific vascular differentiation in vivo.
EXPERIMENTAL PROCEDURES
Cell Culture and Transfections
Primary human dermal microvascular endothelial cells (HDMEC) were isolated from neonatal foreskins and were cultured as previously described (Richard et al., 1998). HDMEC at passages 5 to 8 were transduced with either control- or Prox1-adenovirus and were further incubated for up to 48 hr. Immortalized human microvascular endothelial cells (HMEC-1) were maintained in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS), 2 mM L-glutamine, 100 U/ml penicillin G, 100 μg/ml streptomycin, and 0.25 μg/ml amphotericin B (Life Science, Grand Island, NY). HMEC-1 were transiently transfected by using SuperFect (Qiagen, Chatsworth, CA) and were further incubated for 48 hr before analysis.
Generation of Adenovirus Containing Human Prox1 cDNA
We used the AdEasy system to generate Prox1-containing adenovirus (He et al., 1998). A NotI fragment containing human Prox1 coding sequence (Zinovieva et al., 1996) was cloned into the NotI site of the pAdTrack-CMV vector. After linearization with PmeI, the vector was used to transform Escherichia coli strain BJ5183 that had been previously transformed with the pAdEasy-1 vector to induce homologous recombination between the two vectors (He et al., 1998). Several kanamycin-positive colonies were chosen and screened by NotI digestion. Successfully recombined vectors were linearized with PacI and were transfected into HEK293 cells. Transfection and subsequent amplification were monitored by assessing GFP signals. Expression of Prox1 was confirmed by both Northern and Western blot analyses. Corresponding control adenovirus was obtained from Dr. Sam Lee (Harvard Medical School).
Real-Time RT-PCR
The ABI Prism 7000 Sequence Detection System was used to perform either SYBR-Green based or dual-labeled probe based real-time RT-PCR reactions as described (Hawighorst et al., 2002). Sequences of the primers used for this study are provided upon request. For SYBR-Green based reactions, at least three sets of primers were used for each gene of interest. SYBR Green PCR Master Mix was used for all SYBR Green reactions with the addition of MultiScribe reverse transcriptase (Applied Biosystems, Foster City, CA). For dual-labeled probe-based real-time RT-PCR reactions, probes labeled with 6-FAM and TAMRA were multiplexed with GAPDH primers and probe, labeled with JOE and TAMRA, as an internal control. TaqMan EZ RT-PCR Core Reagent was used for dual-labeled probe-based reactions. Total RNAs were isolated by using Tri-reagent (Sigma, St. Louis, MO) and were treated with RNAse-free RQ-DNAse (Promega, Madison, WI) before analysis. Twenty to 100 ng of total RNA were used for each reaction.
Northern and Western Blot Analyses and Immunostains
Northern blot analyses were performed as described (Hawighorst et al., 2002), by using 15 μg of total RNA and a 2.2-kb human Prox1 cDNA probe. For Western analyses, cell lysates were obtained as described (Skobe et al., 2001) and 30 μg of protein per sample were analyzed by denaturing sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotted with antibodies against neuropilin-1 (Oncogene, San Diego, CA), human podoplanin (antibody gp40 [Zimmer et al., 1997], kindly provided by Dr. G. Herrler), human ICAM-1 (BD Pharmingen, San Diego, CA), and human VEGFR-3/Flt4 (Santa Cruz Biotechnology, Santa Cruz, CA).
Immunofluorescence stains were performed on 6-μm cryostat sections of adult murine back skin and on 10-μm sections of mouse embryos as described (Detmar et al., 1998; Wigle et al., 2002), by using polyclonal rabbit antibodies to Prox1 (Wigle et al., 2002), LYVE-1 (kindly provided by Dr. D. Jackson [Prevo et al., 2001]), VEGFR-3/Flt4 (Zymed, San Francisco, CA), neuropilin-1 (R&D Systems, Minneapolis, MN), a monoclonal rat antibody to CD31 (BD Pharmingen), hamster antibodies to podoplanin (clone 8.1.1; Developmental Studies Hybridoma Bank, University of Iowa), and murine ICAM-1 (BD Pharmingen), and corresponding secondary antibodies labeled with AlexaFluor488 or AlexaFluor594 (Molecular Probes, Eugene, OR). Cell nuclei were counterstained with 20 μg/ml Hoechst bisbenzimide. Sections were examined by using a Nikon E-600 microscope (Nikon, Melville, NY) and images were captured with a SPOT digital camera (Diagnostic Instruments, Sterling Heights, MI).
Acknowledgements
We thank D. Jackson for providing the LYVE-1 antibody, G. Herrler for providing the gp40 antibody, and S. Tomarev for providing human Prox1 cDNA. M.D. received funding from the NIH/NCI, the American Cancer Society Research Project, the Susan Komen Breast Cancer Foundation, the Cutaneous Biology Research Center through the Massachusetts General Hospital/Shiseido Co. Ltd. Agreement; G.O. received funding from the NIH, a NCI Cancer Center Support (CORE) grant, and the American Lebanese Syrian Associated Charities; and V.S. received funding from the Deutsche Forschungsgemeinschaft.
NOTE ADDED IN PROOF
While this manuscript was in press, another paper also reported lymphatic endothelial reprogramming of cultured vascular endothelial cells by Prox1 (Petrova et al. [2002] EMBO J 21:4593–4599).
