Main

Sox18 belongs to the SRY-related HMG domain family of developmental transcription factors6,7. During embryogenesis, Sox18 is expressed in developing vasculature and hair follicles8, consistent with the reported cardiovascular and coat texture phenotype of naturally occurring mouse mutants of the ragged (Ra) allelic series9,10. The phenotype of ragged mutants is evidently due to dominant-negative action of the mutant Sox18 protein (Sox18Ra) produced in these mice11, resulting from the normal ability of Sox18Ra to occupy binding sites in the promoters of Sox18 target genes, but a failure to engage partners and/or cofactors required to initiate transcription of these targets12.

To explore the role of Sox18 in lymphatic vascular development, we first re-examined the phenotype of ragged-opossum (RaOp ), the most severely affected class of Ra mutant11,13. Most homozygous RaOp /RaOp embryos showed gross subcutaneous oedema at 13.5 days post coitum (d.p.c.; Fig. 1a). We also bred mice with a targeted inactivation (‘knockout’) of Sox18 from their original mixed 129-CD1 background14 onto a pure C57BL/6 (B6) background by 11 generations of backcrossing to inbred B6 mice; the resulting strain, designated Sox18-/- (B6), was used throughout this study. This backcrossing revealed gross subcutaneous oedema (Fig. 1a) and fetal lethality in all homozygotes, and mild subcutaneous oedema around the head in some heterozygotes (2 of 4; not shown), features not evident on the mixed background.

Figure 1: Lymphatic vascular defects resulting from Sox18 dysfunction.
figure 1

a, Sox18-/- (B6) and RaOp /RaOp mutant embryos show oedema at 13.5 d.p.c. (arrows). b, c, Lymphatic defects in adult Sox18 mutant mice. b, Immunofluorescence for the lymphatic marker LYVE-1 (green) on whole ear segments revealed clear differences between the large-diameter, blunt-ended lymphatic vessels in wild-type (upper row) and the smaller, highly branched vessels in Sox18+/- (B6) (middle row) and RaOp /+ mice (lower row). Immunostaining for the general endothelial cell marker PECAM-1 (blue) showed a similar patterning of the blood vasculature in both types of mice. c, Quantitative analysis of lymphatic parameters in RaOp /+ versus wild-type ears based on LYVE-1 immunostaining. RaOp /+ heterozygotes showed a statistically significant difference in all parameters observed compared to wild type. The averaged field counts for each parameter were collated for all sections and their respective statistical significance determined by Student’s t-test (n = 3). Data represent mean + s.e.m.; ***P < 0.001, **P < 0.01. d–i, Loss of lymphatic endothelial differentiation in Sox18-mutant embryos. Prox1 (green)/PECAM-1 (red) double immunofluorescence on transverse sections. d, At 10.5 d.p.c. and e, 13.5dpc, Prox1/PECAM-1-positive lymphatic endothelial cells can be seen in the cardinal vein and jugular lymph sac areas in wild-type embryos (arrows). f, g, Markedly fewer Prox1/PECAM-1-positive cells were detected in RaOp /RaOp embryos (arrows). h, i, No Prox1-positive lymphatic endothelial cells could be detected at all in Sox18-null embryos. CV, cardinal vein; DA, dorsal aorta; LS, lymph sac; D, dorsal, M, medial. Scale bars, 1.6 mm (a); 40 μm (b, 20×); 20 μm (b, 40×, d–i).

PowerPoint slide

Full size image

RaOp /RaOp and Sox18-/- (B6) embryos die shortly after 14.5 d.p.c., precluding study of lymphatic physiology beyond that stage. We therefore examined the structure of the mature lymphatic vascular network in adult mice with heterozygous Sox18 mutations (RaOp /+ and Sox18+/- B6), by immunofluorescence analysis of whole ear tissue, using antibodies specific for the marker LYVE-1. The lymphatic vessels in Sox18-mutant mice were finer, denser and more branched, with more loops and fewer blind-ended sacs, compared to wild-type mice (Fig. 1b, c). In contrast, blood vascular patterning was unaffected in RaOp /+ and Sox18+/- (B6) mice, as revealed by PECAM-1 staining (Fig. 1b). Thus, defects in Sox18 function are sufficient to alter the patterning of the lymphatic vasculature, and imply that lymphatic dysgenesis contributes to the oedema seen in the Sox18 mutant mice and in hypotrichosis-lymphoedema-telangiectasia (HLT).

We next examined the timing and location of Sox18 expression during mouse embryo development, with reference to the known molecular and cellular hallmarks of developmental lymphangiogenesis. At 9 d.p.c., expression of Sox18, but not Prox1, was observed in endothelial cells lining the dorsolateral sector of the cardinal vein (Fig. 2a, arrows). At 10.5 d.p.c., Sox18 expression continued in a population of cells migrating from this area; these cells also expressed Prox1 and the endothelial marker PECAM-1, indicating that they are precursors of the lymphatic vasculature15 (Fig. 2b, c, yellow nuclei). Sox18 continued to be expressed in most cells of the primary lymph sacs at 13.5 d.p.c. (Fig. 2d, e, arrows). These cells also expressed PECAM-1 and the lymphatic marker podoplanin, indicating their lymphatic nature. At this stage, Sox18 expression was also maintained in some blood endothelial cells, expressing PECAM-1 but not podoplanin, outside the lymph sacs (Fig. 2f, asterisks). By 14.5 d.p.c., Sox18 expression had subsided in the lymph sacs (Fig. 2g), but could still be detected in mesenteric and dermal blood vessels (Fig. 2h, i, red). Sox18 was not detectable in mesenteric or dermal lymphatics during their early development (Fig. 2h, i, arrows) or in newborn mice (Fig. 2j, k), or in mature lymphatic vessels in the adult (Fig. 2l), signifying that this protein is not required for maintenance of lymphatic endothelial cells once they have differentiated.

Figure 2: Sox18 expression during lymphatic vascular development.
figure 2

a–c, Immunofluorescence for Sox18 (red), endothelial marker PECAM-1 (blue) and Prox1 (green). Lymphatic endothelial progenitors (yellow) co-express Sox18 and Prox1 at 10.5 d.p.c. (c, boxed area in b, arrows). d–l, Immunofluorescence for Sox18 (red), endothelial marker PECAM-1 (blue) and lymphatic marker podoplanin (green). e, Enlargement of boxed area in d, arrows indicate triple-positive lymphatic endothelial cells in lymph sacs at 13.5 d.p.c. f, Some blood endothelial cells expressing Sox18 and PECAM-1 but not podoplanin are visible outside the lymph sacs (asterisks). g, Lack of Sox18 expression in lymph sacs, mesenteric lymphatics (h, arrows) and dermal lymphatics (i, arrows) by 14.5 d.p.c. j, Newborn mesentery, k, newborn skin and l, mature lymphatic vessels at adulthood, lack Sox18 expression. C, caudal V, ventral; M, medial; D, dorsal; CV, cardinal vein; LS, lymph sac; DA, dorsal aorta. Scale bars: 50 μm (a, b, d, g); 20 μm (j, k, l); 10 μm (c, e, f, h, i).

PowerPoint slide

Full size image

To investigate whether Sox18 is able to modulate the expression of Prox1 and other lymphatic markers in an in vitro model system, we overexpressed wild-type Sox18 or dominant-negative Sox18RaOp in mouse embryonic stem (ES) cells by lentiviral infection and examined the resulting expression of lymphatic marker genes. In the system studied, the ES cells form embryoid bodies and spontaneously differentiate into endothelial cells that express Prox1 and podoplanin16,17. Expression of both markers was significantly increased through overexpression of Sox18, while overexpression of Sox18RaOp strongly inhibited induction of lymphatic markers (Fig. 3a). Moreover, the inhibition of Prox1 and podoplanin expression by Sox18RaOp was rescued by infection with high levels of Sox18-expressing lentivirus (Fig. 3a), indicating that the effects of Sox18RaOp were mediated by direct interference with Sox18 function in this system. Overexpression of Sox18 and Sox18RaOp did not affect general blood vascular endothelial cell differentiation markers such as Vegfr2, Tie2 and VE-cadherin (Fig. 3b). Therefore, Sox18 stimulates expression of lymphatic markers during the differentiation of ES cells along an endothelial-specific pathway.

Figure 3: Sox18 induces lymphatic markers in differentiating ES cells and blood vascular endothelial cells.
figure 3

a, ES cells were infected with lentivirus expressing GFP (an inert negative control; black bars), Sox18 (grey bars), Sox18RaOp (white bars), or Sox18RaOp together with an excess of Sox18 (hatched bars) and induced to differentiate in vitro. Control cells expressed the lymphatic markers Prox1 and podoplanin, as assessed by quantitative RT–PCR, within 7 days of culture. Both markers were further upregulated by Sox18 and downregulated by Sox18RaOp; the effect of Sox18RaOp could be rescued by overexpression of Sox18. b, Sox18 or Sox18RaOp overexpression did not affect blood vascular endothelial cell differentiation markers Tie2, Vegfr2 or VE-Cadherin. c, H5V blood vascular endothelial cells were infected with lentivirus expressing GFP (black bars), Sox18 (grey bars) or Sox18RaOp (white bars) and assayed for marker gene expression by quantitative PCR. Lymphatic markers Prox1, EphrinB2 and Vegfr3 were induced by Sox18 but not by Sox18RaOp. For each set of studies the figure represents a typical experiment of at least three performed. Data are means ± s.d. of six replicates and are expressed relative to GFP. ***P < 0.01, *P < 0.05 versus GFP expressing cells by Student’s t-test.

PowerPoint slide

Full size image

To further investigate the possible role of Sox18 at the nexus of vascular and lymphatic endothelial cell specification, we used an additional cell-based model system. H5V mouse blood endothelial cells18 were infected with the Sox18, Sox18RaOp or GFP-expressing lentiviruses described above, and lymphatic marker expression was analysed by quantitative RT–PCR. Sox18 overexpression in these cells resulted in significant upregulation of the lymphatic markers Prox1, ephrinB2 and Vegfr3 (Fig. 3c). Enhanced Prox1 protein expression in nuclei of H5V cells after Sox18 overexpression was confirmed by immunofluorescence (Supplementary Fig. 1). In contrast, cells overexpressing the mutant Sox18RaOp protein did not show any significant change in the expression level of these lymphatic-specific markers relative to GFP-infected cells (Fig. 3c), probably because H5V cells express only basal levels of lymphatic markers that cannot be reduced further by suppressing Sox function. The induction of lymphatic markers in this assay demonstrates that Sox18 overexpression causes blood vascular endothelial cells to activate the lymphatic endothelial cell pathway. In addition, levels of the blood vascular markers N-cadherin and CD44, but not Nrp1, endoglin or cd34, were reduced after Sox18 overexpression (Supplementary Fig. 2), supporting the concept that Sox18, acting via Prox1, downregulates at least some blood endothelial markers while promoting the lymphatic endothelial differentiation program.

To determine whether Sox18 regulates Prox1 expression directly, we cloned 3,952 base pairs (bp) of the Prox1 promoter and identified two putative Sox binding sites at -1,135 to -1,130 bp (SoxA) and -813 to -808 bp (SoxB) relative to the translation start site. We tested the binding of Sox18 to these putative response elements by electrophoretic mobility shift assays (EMSA). Bacterially expressed, purified GST–Sox18 fusion protein bound strongly to both SoxA and SoxB sites (Fig. 4a), while the GST-only control protein bound neither site. Specificity of binding was confirmed by competing the interaction with increasing amounts of unlabelled oligonucleotide. Sox18RaOp was able to bind to the same sites in the Prox1 promoter (Fig. 4a) with a similar binding affinity to that of wild-type Sox18, supporting previous evidence that ragged forms of Sox18 act in a dominant-negative fashion by binding to Sox target sequences but failing to activate transcription of target genes11,19.

Figure 4: Sox18 activates the Prox1 promoter in vitro and in vivo.
figure 4

a, EMSA showing binding of Sox18 and Sox18RaOp to SoxA (left panel) and SoxB (right panel) binding sites in the Prox1 promoter (arrows). Specificity was confirmed using a 25- or 50-fold molar excess of non-radioactive competitor oligonucleotide (Comp.). b, Chromatin immunoprecipitation (ChIP) of endogenous Sox18 in mlEnd cells. Negative controls omitting PCR template (Neg.), or using anti-MYC or anti-LBX2 antibodies are shown. Prox1 fragment is arrowed. c, Reporter gene assays using constructs containing either wild-type Prox1 promoter (left graph) or versions with the corresponding Sox responsive element mutated (right graph). Data are means ± s.e.m. of three independent experiments each performed in triplicate; ***P < 0.001. d, In vivo analysis of the 4 kb Prox1 promoter-GFP construct in transgenic embryos. GFP signal (left and upper right panels, arrows) was detected in cells adjacent to the cardinal vein in 10.5 d.p.c. embryos. This expression pattern coincided with endogenous Prox1 expression at 10.5 d.p.c. (lower right panel, arrows). GFP and Prox1 expression were also observed in the sympathetic ganglia (asterisks). e, GFP expression in the lymphatic network of the skin of 16.5 d.p.c. transgenic embryos (top panel), co-incident with podoplanin (centre panel; merged pictures, bottom panel). f, Genetic ablation of SoxA and SoxB sites by site-directed mutagenesis (ΔSoxA/B-Prox1-GFP) abolishes Prox1 promoter activity in vivo. At 10.5 d.p.c., transgenic embryos generated with the wild-type Prox1-promoter display GFP expression in PECAM-1 positive cells migrating from the dorsal part of the cardinal vein (left, arrows), whereas transgenic embryos (right) lack these cells. Asterisk indicates residual GFP expression in PECAM-1-negative neural cells. DA, dorsal aorta; CV, cardinal vein; D, dorsal; M, medial. Scale bars: 100 μm (d, left panel); 50 μm (d, right panels, f); 10 μm (e).

PowerPoint slide

Full size image

As an independent confirmation of these data, chromatin immunoprecipitation (ChIP) experiments were performed using chromatin from the mouse mesenteric lymph node endothelial cell line mlEnd (ref. 20), which expresses endogenous Sox18 and Prox1 (Supplementary Fig. 3). DNA fragments precipitated by anti-Sox18 antibody were purified and tested for the presence of the SoxA/B binding sites in the Prox1 promoter by PCR-amplifying a 521 bp fragment. After 22 PCR cycles, a clear enrichment of this fragment was detected (Fig. 4b), confirming that Sox18 protein binds to the Prox1 promoter in mammalian cells.

We next determined whether Sox18 could trans-activate the Prox1 promoter driving a luciferase reporter gene after transfection into mlEnd cells. Overexpression of Sox18 led to a significant increase in luciferase activity (Fig. 4c). This activation was observed using the original 4 kb Prox1 promoter fragment that contains both SoxA and SoxB binding sites. Overexpression of the mutant Sox18RaOp did not enhance the luciferase activity driven by the Prox1 promoter (Fig. 4c). Site directed mutagenesis of the Prox1 promoter, which altered key sequence elements of either the SoxA site or the SoxB site, abolished the enhanced activity produced by Sox18 overexpression (Fig. 4c). Hence, activation of the Prox1 promoter in this assay system requires cooperation of the identified SoxA and SoxB sites.

To confirm the importance of the 4 kb promoter fragment in the in vivo regulation of Prox1 expression, we generated transgenic mice carrying a GFP reporter construct driven by this fragment. In transgenic embryos at 10.5 d.p.c., GFP-expressing cells were found in the region of the cardinal vein corresponding to the location of the Prox1-positive lymphatic progenitors at that stage (Fig. 4d, Supplementary Fig. 4a, arrows). By 16.5 d.p.c., a complex network of GFP-positive vessels was observed, identical to the pattern of endogenous Prox1 expression in the lymphatic vascular network (Supplementary Fig. 4b). Co-localization of GFP with the lymphatic marker podoplanin in transgenic embryos confirmed transgene expression in lymphatic endothelial cells in these mice (Fig. 4e). These observations indicate that the Sox18-responsive 4 kb Prox1 promoter fragment identified in this study is sufficient to drive faithful expression of GFP during developmental lymphangiogenesis in vivo. This promoter fragment also drove reporter gene expression in other tissue types such as the sympathetic ganglia, where endogenous Prox1 is also normally expressed21 (Fig. 4d, asterisks). This fragment evidently contains elements required for the expression of Prox1 in those sites and also for maintenance of lymphatic endothelial expression of Prox1 after Sox18 expression has subsided. Ablation of the SoxA and SoxB sites in the transgene abolished Prox1 promoter activity specifically in the cardinal vein area where lymphatic endothelial precursor cells are generated (Fig. 4f, arrows), but not in the sympathetic ganglia (Fig. 4f, asterisk), indicating that Sox18 binding to the Prox1 promoter is specifically required in the context of lymphangiogenesis in vivo.

Finally, to test whether Sox18 is also necessary for stimulating Prox1 expression in lymphatic precursor cells in vivo, we examined the expression pattern of Prox1 in Sox18-mutant embryos, in comparison to their wild-type littermates. At 10.5 d.p.c., when extensive migration of Prox1-expressing lymphatic endothelial precursor cells normally occurs in the mouse embryo (Fig. 1d), the number of Prox1-positive cells within and around the cardinal vein was greatly reduced in RaOp /RaOp embryos (Fig. 1f). Moreover, these cells were completely absent in Sox18-null (B6) embryos at this stage (Fig. 1h, Supplementary Fig. 5). This total lack of lymphatic endothelial cells was confirmed by complete serial section analysis of 10.5 d.p.c. Sox18-null (B6) embryos. Expression of Prox1 in non-lymphatic cell types such as the myocardium was not affected by the lack of Sox18 (Supplementary Fig. 5). The blockade of lymphatic development cannot be ascribed to a primary deficiency in blood vascular endothelial cells; no defects in blood vascular structure or marker expression were observed in the area of the cardinal vein, or anywhere else in mutant embryos (Supplementary Fig. 6 and data not shown).

By 13.5 d.p.c., linear groups of Prox1-positive cells were clearly visible lining the primary lymph sacs in wild-type embryos (Fig. 1e), as described previously4. Similar cells were present in only low numbers in RaOp /RaOp embryos (Fig. 1g), and completely absent in Sox18-null embryos (Fig. 1i). Both RaOp /RaOp and Sox18-null embryos exhibited a hypertrophic jugular vein and absence of lymph sacs (Supplementary Fig. 7). These observations establish that lack of Sox18 function results in a complete blockade of Prox1-mediated lymphangiogenesis from the pool of vascular endothelial cell precursors during mouse embryogenesis, confirming the pivotal role of Sox18 in this system.

In summary, our data suggest a model in which Sox18 acts at the nexus of differentiation of lymphatic endothelial progenitor cells from blood vascular precursors in the embryo (Supplementary Fig. 8). The role of Sox18 in promoting lymphatic endothelial cell specification is consistent with the known roles of other Sox transcription factors that act as developmental switches6,22. Moreover, Sox18, like other Sox factors, clearly acts in a number of developmental systems—hair follicle development, vasculogenesis and lymphangiogenesis—and is likely to depend on context-specific partner proteins and/or cofactors23,24. Further studies will focus on defining the molecular milieu in which Sox18 operates to activate Prox1 transcription, and may reveal new strategies for therapeutic stimulation or suppression of lymphangiogenesis.

Methods Summary

Antibodies were obtained commercially, except anti-Sox18 (Supplementary Fig. 9), which was raised in rabbits against amino acids (a.a.) 228–254 of mouse Sox18 and biotinylated as described25. For transgenesis, CBA/C57BL6 F1 zygotes were injected with Prox1–GFP and reimplanted into outbred foster mothers. EMSA was carried out as described19. For ChIP, Prox1 primers 5′-GGCAAGGGTGTTTCCTTTTT-3′ and 5′-GGCACAGACATGCACTGTCC-3′ were used with 50 °C annealing and 22 PCR cycles.

Online Methods

Mouse strains

Heterozygous ragged-opossum (RaOp ) C57BL6/J mice were obtained from the Jackson Laboratory (Bar Harbor) and interbred to generate RaOp /RaOp embryos, RaOp /+ adults and wild-type littermates for analysis. Sox18-null embryos were genotyped as previously described14. Backcrossing of Sox18 null mice from the original mixed background (129/B6) was accomplished by 11 generations of backcrossing onto a pure B6 background. Prox1–GFP transgenic mouse embryos were produced by pronuclear injection into CBA/C57BL6 F1 zygotes and screened by PCR. Procedures involving animals and their care conformed to institutional guidelines (University of Queensland Animal Ethics Committee).

Immunofluorescence

Immunofluorescence was performed as previously described26. Skin and ear samples were dissected from 16.5 d.p.c. transgenic embryos and adult mice respectively, and fixed overnight in 4% PFA. Primary antibodies in blocking solution were added and incubated for 24 h at 4 °C. Samples were washed for 6 h in washing solution (1% BSA, 1% DMSO in PBSTx) and incubated for 16 h with secondary antibodies in blocking solution. Confluent H5V cells were fixed in 4% PFA for 15 min at room temperature. Cells were permeabilized with PBSTx (0.3%) for 5 min at 4 °C and then blocked in PBS/5%BSA for 30 min at room temperature.

Quantitation of lymphatic vasculature

After immunofluorescence using anti-LYVE-1 or anti-PECAM-1 antibody, apical photographs of mouse ears were taken at 10× magnification. The numbers of lymphatic vessel branching points, lymphatic vessel loops, and ‘blind ended’ lymphatic sacs were determined, in a blinded fashion, and averaged for an entire image, as described previously27.

Antibodies

Antibodies were used at the following dilutions: rabbit polyclonal anti-mouse Prox1 (Covance), 1:2,500; hamster anti-mouse podoplanin (RDI), 1:500; rabbit polyclonal anti-mouse Sox18 (Supplementary Fig. 9) raised against peptides PLAPEDCALRAFRA and RAPYAPELARDPSFC (a.a. 228–241 and 240–254 respectively of mouse Sox18), biotinylated as described25, 1:1,000; rabbit polyclonal anti-mouse LYVE-1 (Fitzgerald Industries), 1:1,000; rat anti-mouse CD31 (BD Pharmingen), 1:200; mouse monoclonal anti-α-tubulin (clone B-5-1-2, Sigma) 1:2,000; mouse monoclonal NG2 antibody (Chemicon) 1:200; mouse monoclonal anti-MYC Tag (9B11, Cell Signalling), 1:2,000; rabbit polyclonal anti-GFP (Molecular Probes), 1:1,000; rabbit polyclonal anti-Lbx2 (Millipore) 1:1,000; rat monoclonal anti-endoglin (BD Pharmingen), 1:200; goat polyclonal anti-neuropilin-1 (R&D Systems) 1:50; hamster polyclonal anti-ICAM-1 (Millipore) 1:200; mouse monoclonal anti-N-cadherin, 1:200 (kindly provided by A. Yap); rat monoclonal anti-CD44 (BD Pharmingen) 1:200; rat monoclonal anti-mouse VE-cadherin (BD Pharmingen) 1:300; and rat monoclonal anti-CD34 as previously described28. Secondary antibodies anti-rat IgG Alexa 594, anti-rabbit IgG Alexa 488, anti-hamster IgG Alexa 488, anti-biotin IgG Alexa 594, and anti-biotin IgG Alexa 488 (Molecular Probes) were used at a dilution of 1:200, and anti-rabbit IgG HRP conjugated (Zymed Laboratories) used at a dilution of 1:2,000.

ES cell culture

ES cells were differentiated to endothelial cells as previously described16,17. ES cells were mildly trypsinized and suspended in Iscove’s modified Dulbecco medium with 15% FBS, 100 U ml-1 penicillin, 100 μg ml-1 streptomycin, 450 μM monothioglycerol, 10 μg ml-1 insulin, 50 ng ml-1 human recombinant VEGF-A165 (Peprotech Inc.), 2 U ml-1 human recombinant erythropoietin (Cilag AG), and 100 ng ml-1 human bFGF (Genzyme). Cells were seeded in bacteriological Petri dishes and cultured for 3, 7 or 11 days at 37 °C with 5% CO2 and 95% relative humidity.

Quantitative PCR analysis

cDNA was amplified in triplicate in a reaction volume of 15 μl using TaqMan Gene Expression Assay (Applied Biosystems) and an ABI/Prism 7900 HT thermocycler, using a pre-PCR step of 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 60 s at 60 °C. Preparations of RNA template without reverse transcriptase were used as negative controls. Ct values were normalized to Gapdh expression29.

Electrophoretic mobility shift assay

EMSA was carried out as previously described19. Two double-stranded oligonucleotides were constructed, containing either of the SOX binding sites (underlined) and surrounding sequences from the Prox1 promoter at -1,135 bp (SoxA) 5′-GGTTCCCCCGCCCCCAGACAATGCTAGTTTGCATACAAAG-3′ and at -813 bp (SoxB) 5′-ACAGTCCATCTCCCTTTCTAACAATTGAAGTCACCAAATG-3′.

Plasmids

The mouse Prox1 promoter fragment encompassing nucleotides -3,952 to -1 from the ATG was generated by PCR. Hot-start PCR was performed (annealing temperature at 58 °C). The resulting fragment cloned into the KpnI and XhoI sites of the pGL2Basic luciferase reporter vector (Promega). The 4 kb Prox1 promoter fragment was subcloned into pEGFP-N1 vector (Clontech, BD Bioscience) which had the CMV promoter removed. The integrity of the reporter construct was verified by sequencing. SoxA and SoxB mutant clones were generated by site directed mutagenesis (QuickChange, Stratagene) using mutated primers (5′-CCCCCGCCCCCAGATGATGCTAGTTTGC-3′) at position -1,135 or a different mutated primer (5′-GGCAACAGTCCATCTCCCTTTCTAATGATTGAAGTCACC-3′) at position -813.

Cell culture and transfection experiments

Mouse mesenteric lymph node endothelial cells (mlEnd) were created as previously described20 and obtained from J. Gamble. Cells were transfected using a combination of Lipofectamine 2000 (Invitrogen) and CombiMag (Chemicell) for 20 min (according to manufacturer’s instructions) on a magnetic plate at room temperature. Cells were harvested 24 h after transfection, and luciferase assays were carried out as previously described12. H5V murine endothelial cells were cultures as previously described18.

Lentiviral vector preparation

To generate HIV3-PGK vector, the Pgk promoter from pRetro/SUPER (ref. 30) was ligated into the lentiviral plasmid pRRLsin.PPTs.hCMV.GFPwpre (ref. 31) in place of the CMV promoter. Sox18 and Sox18RaOp cDNAs were subcloned in HIV3-PGK and lentiviral vectors produced as described32. Lentiviral and packaging plasmids were kindly donated by L. Naldini.

Chromatin immunoprecipitation

A total of 1 × 107 mlEnd cells were cross-linked with 1% formaldehyde for 10 min at room temperature. Chromatin immunoprecipitation was performed as previously described26. Primers 5′-GGCAAGGGTGTTTCCTTTTT-3′ and 5′-GGCACAGACATGCACTGTCC-3′ used in a 22 cycle PCR reactions (annealing temperature 50 °C).

X-gal staining

X-gal staining was performed as previously described33.

Western blotting

TM3 cells were transfected with Sox7, Sox17 or Sox18 expression constructs. Proteins were extracted in 2× sample buffer (125 mM Tris, pH 6.8; 4% SDS; 20% glycerol and 5% β-mercaptoethanol). Primary antibody was used overnight at 4 °C. HRP-conjugated secondary antibody followed by chemiluminescence detection with Super Signal West Pico Chemiluminescence reagent (Pierce) were used to reveal protein expression.