Orphan nuclear transcription factor TR3/Nur77 regulates microvessel permeability by targeting endothelial nitric oxide synthase and destabilizing endothelial junctions

Abstract

Low-level basal vascular permeability (BVP) provides nutrients to normal tissues, and increased vascular permeability is characteristic of inflammation and cancer. We recently reported that VEGF-A, a potent vascular permeabilizing and angiogenic factor, exerts much of its angiogenic activity by up-regulating expression of TR3/Nur77, an orphan nuclear transcription factor, in vascular endothelial cells (EC). To determine whether TR3/Nur77 had a more general role in regulating vascular permeability, we found that histamine, serotonin, and platelet-activating factor, small molecule vascular permeabilizing agents, also increased TR3/Nur77 expression acutely in EC. BVP and the acute vascular hyperpermeability (AVH) induced by these vascular permeabilizing factors were greatly decreased in Nur77^−/− mice, and both BVP and AVH correlated with Nur77 expression levels in several different mouse strains. BVP and AVH were enhanced in transgenic mice in which Nur77 was selectively overexpressed in vascular EC, whereas both were suppressed in mice overexpressing dominant-negative Nur77. Chronic vascular hyperpermeability (CVH) was induced long before the onset of angiogenesis in a modified, in vivo Matrigel assay that included PT67 cells packaging retroviruses expressing Nur77-sense, whereas inclusion of cells packaging viruses expressing Nur77-antisense prevented VEGF-A–induced CVH. TR3/Nur77 modulated vascular permeability by increasing endothelial nitric-oxide synthase expression and by downregulating several EC junction proteins that maintain vascular homeostasis. Both functions required TR3/Nur77 transcriptional activity. Taking these data together, TR3/Nur77 is up-regulated by several vascular permeabilizing agents and has critical roles in mediating BVP, AVH, and CVH.

TR3 (mouse homolog Nur77, rat homolog NGFI-B) is a member of the class IV subfamily of the orphan nuclear receptor superfamily of transcription factors (1). TR3/Nur77 plays important roles in tumor, lymphocyte, and neural growth and survival (2–5). Although Nur77^−/− mice develop a normal vasculature and lack a developmental phenotype, we recently reported the unexpected finding that TR3/Nur77 is strongly up-regulated by VEGF and has a critical role in pathological angiogenesis (6).

VEGF-A was originally described as a potent vascular permeabilizing factor (7), and increased vascular permeability is a characteristic property of “mother” vessels, the first new blood vessels induced to form in pathological angiogenesis by VEGF-A–secreting tumors, healing wounds, and chronic inflammation (8). Furthermore, endothelial growth factors, such as bFGF and PDGF, which do not affect vascular permeability also did not induce TR3/Nur77 expression (6). Therefore, we considered the possibility that TR3/Nur77 might have an important role in regulating vascular permeability in its three different contexts (9): (i) Basal vascular permeability (BVP), the low-level leakage of plasma protein-poor fluid from normal capillaries that provides tissues with nutrients; (ii) Acute vascular hyperpermeability (AVH), the acute, explosive, plasma protein-rich exudate that proceeds from postcapillary venules in response to short-term exposure to vascular permeabilizing agents, such as VEGF-A; and (iii) Chronic vascular hyperpermeability (CVH), the long-term, plasma protein-rich exudate that extravasates from mother vessels in pathological angiogenesis. We now report that TR3/Nur77 has important roles in regulating vascular permeability in all three of these contexts and that its expression is regulated not only by VEGF-A but also by the small molecule vascular permeabilizing agents histamine, platelet-activating factor (PAF), and serotonin. Furthermore, TR3/Nur77, like VEGF, acts at least in part by regulating the expression of endothelial nitric-oxide synthase (eNOS) and endothelial cell (EC) junction proteins.

Results

TR3/Nur77 Is Up-Regulated by Several Vascular Permeabilizing Agents in Addition to VEGF-A.

We hypothesized that TR3/Nur77 might be selectively up-regulated by other agents that increased microvascular permeability. Testing this hypothesis, we found that, like VEGF-A¹⁶⁵, three well-known, potent, small-molecule vascular permeabilizing factors, histamine, PAF and serotonin, strongly up-regulated TR3 mRNA and protein expression in human umbilical vein endothelial cells (HUVEC) (Fig. 1 A and B).

Fig. 1.

TR3/Nur77 expression is increased by several different permeability factors and regulates BVP and AVH. (A) HUVEC were stimulated with VEGF-A¹⁶⁵ (10 ng/mL), histamine (10 μM), PAF (0.1 μM), or serotonin (10 μM) for 0.5 to 2 h. TR3 mRNA expression was determined by quantitative real-time RT-PCR (mean ± SD). (B) Immunoblots demonstrating TR3 protein expression in HUVEC stimulated with histamine (10 μM), PAF (0.1 μM), or serotonin (10 μM) for indicated times. (Right) MAPK protein loading control. (Ci) Representative macroscopic images of Miles vascular permeability assay in flank skin and mesentery of C57BL/6^+/+ (wild-type) mice (Upper) and C57BL/6 Nur77^−/− mice (Lower) following intradermal injection of VEGF-A¹⁶⁵, histamine, PAF and serotonin, or intraperitoneal injection of VEGF-A¹⁶⁵. Photos were taken 30 min after intravenous injection of 0.2 mL of 0.5% EB dye in saline. (Cii) Quantification of Miles assays in Ci. EB dye was extracted and quantified as microliter per gram (mean ± SD, four mice per group) (6). ***P < 0.001, Tukey–Kramer multiple comparisons test (flank skin) or unpaired t test (mesentery), comparing Nur77^+/+ vs. Nur77^−/− mice. (D) Western blot demonstrating expression of Nur77 and eNOS in Nur77^+/+ vs. Nur77^−/− mice and in several different mouse strains. β-Actin, loading control. (E) Quantification of Miles assays in flank skin of indicated mouse strains following HBSS, VEGF-A¹⁶⁵, histamine, or serotonin injection. EB dye was extracted and quantified (mean ± SD, four mice per group). *P < 0.05, **P < 0.01, ***P < 0.001, Tukey–Kramer test.

BVP and AVH Are Severely Compromised in Nur77^−/− Mice in Vivo.

The finding that several vascular permeabilizing agents induced TR3/Nur77 expression suggested that this gene product might exert a regulatory effect on vascular permeability in vivo. We therefore performed the Miles assay in Nur77^−/− vs. wild-type mice. The low-level BVP of control skin, or after HBSS injection, was significantly diminished in Nur77^−/− mice, as was the AVH response induced by VEGF-A, histamine, serotonin, and PAF, even when administered at supraphysiological doses (Fig. 1C). Permeability induced in the peritoneal cavity by intraperitoneal injection of VEGF-A¹⁶⁵ was similarly reduced in Nur77^−/− mice (Fig. 1C). In addition, protein expression levels of Nur77 (Fig. 1D), and both BVP and AVH (Fig. 1E), differed significantly in four different mouse strains, and in the same order: C57BL/6 Nur77^+/+ > A/J > BALBc >129SV>> C57BL/6 Nur77^−/−.

Overexpression of TR3/Nur77 Is Sufficient to Induce CVH.

We used a modified Matrigel assay to determine whether TR3/Nur77 was able to induce CVH in vivo (6). SKMEL-2 melanoma cells transfected to overexpress VEGF-A¹⁶⁵, and retrovirus-packaging PT67 cells engineered to express Nur77-related cDNAs, were incorporated into Matrigel plugs that were implanted subcutaneously in nude mice. In this assay, the expectation is that VEGF-A¹⁶⁵ secreted by SKMEL/VEGF cells will induce nearby vascular EC to divide and therefore to become susceptible to infection with retroviruses secreted by PT67 packaging cells (6). As expected, inclusion of VEGF-A¹⁶⁵-expressing SKMEL/VEGF cells stimulated angiogenesis and the formation of leaky mother vessels at 3 d, but not earlier (Fig. 2A, lanes 2 and 3 vs. lane 1; quantification in Fig. S1). Both angiogenesis and microvascular permeability were strikingly inhibited when PT67 cells encoding Nur77-antisense (AS) were included (lane 4). However, when PT67 cells packaging Nur77-sense (S) were also included (lane 5), permeability developed within 12 h, long before the onset of EC division and 60 h before the permeability increase induced by SKMEL/VEGF cells.

Fig. 2.

TR3/Nur77 expression and vascular permeability (A) Macroscopic images illustrate leakage of EB dye at 0.5, 1, and 3 d after implantation of Matrigel plugs with indicated contents of VEGF-A¹⁶⁵–secreting SKMEL/VEGF cells and PT67 cells packaging indicated cDNAs. Dye from 1- and 3-d time points was extracted and quantified (Fig. S1, eight animals per group). At 1 d, lanes 5 and 6 differed significantly from all other lanes (P < 0.001, Tukey–Kramer test); at 3 d, lanes 1 and 4 differed significantly from all other lanes (P < 0.001). (B) BVP, 30 min after intravenous EB dye injection, in selected organs of wild-type (Upper) and EC-Nur77-S mice (Lower), and AVH after VEGF-A¹⁶⁵ injection in flank skin. EB dye extravasation was quantified and, in all organs except uterus and brain, EC-Nur77-S was significantly greater than wild-type Fvb mice (P < 0.001 for mesentery, testis, lung, kidney, heart, and skin ± HBSS or VEGF-A¹⁶⁵; P < 0.01 for liver, Mann–Whitney test). (C) Liver and heart sections immunostained for CD31. Mother vessels that formed in EC-Nur77-S transgenic mice were significantly enlarged (P < 0.001), but vascular density vs. control wild-type mice was unchanged (P = 0.15) (unpaired t test). Four mice per group, based on three different transgenic founders. (D) EB dye leakage induced by HBSS or VEGF-A¹⁶⁵ in flank skin and mesentery was significantly greater in wild-type Fvb mice than in EC-Nur77-DN mice 6 d after withholding tetracycline (**P < 0.01, ***P < 0.001, unpaired t test).

PT67/Nur77-S–expressing cells also rendered vessels hyperpermeable in a 12-h time-frame, even in the absence of SKMEL/VEGF cells (Fig. 2A, lane 6). Furthermore, whereas the VEGFR-2/KDR inhibitor, SU1498, strikingly inhibited VEGF-A¹⁶⁵–induced microvascular permeability, it had no effect on the vascular permeability induced by PT67 cells packaging retroviruses expressing Nur77-S (Fig. S2). Taken together, these data indicate that Nur77 induces vascular hyperpermeability independently and downstream of VEGF-A, and therefore of VEGFR-2/KDR, the VEGF-A¹⁶⁵ receptor primarily responsible for both angiogenesis and permeability.

Vascular Permeability in EC-Nur77-S and EC-Nur77-DN Mice.

To determine whether overexpression of Nur77 in mouse endothelium was sufficient to regulate microvessel permeability, we generated Fvb mice that selectively overexpressed Nur77 full-length cDNA (Nur77-S), or a dominant-negative mutant (Nur77-DN), in vascular EC under the control of a tetracycline-sensitive promoter (Fig. S3 A–C). Expression levels of Nur77-S and Nur77-DN mRNAs and protein in different organs of these mice are shown in Fig. S3 D–F. Following withdrawal of tetracycline, Fvb mice overexpressing Nur77-S exhibited a striking, highly significant (P < 0.001) increase in BVP and AVH in multiple organs (Fig. 2B; quantification in Fig. S4). In addition, vessel size increased as mother vessels formed (P < 0.001) but vascular density was not changed 2 d after withdrawing tetracycline (Fig. 2C; quantification in Fig. S5). In Fvb Nur77-DN mice, BVP in skin and AVH in both skin and peritoneum (mesentery) were significantly reduced from that of wild-type mice (Fig. 2D).

TR3/Nur77 Regulates eNOS Expression Through Transcriptional Activity.

VEGF-A regulates eNOS, the enzyme responsible for synthesizing nitric oxide (NO), an important mediator of vascular permeability (10, 11). To determine whether TR3 also regulated eNOS expression, we transduced HUVEC with Flag-TR3-S, as described (6) (Fig. S6), and found that TR3-overexpressing cells expressed greatly increased levels of eNOS, comparable to those induced by stimulating control HUVEC with VEGF-A¹⁶⁵ (Fig. 3A). Overexpression of TR3 antisense cDNA (TR3-AS) or TR3 siRNA (6) (Fig. S6) strikingly inhibited VEGF-A¹⁶⁵-induced eNOS expression (Fig. 3A).

Fig. 3.

TR3 regulates eNOS expression through transcriptional activity. (A) HUVEC transduced with indicated TR3 constructs and LacZ (control) were stimulated with VEGF-A¹⁶⁵ for 0 to 24 h. Cell extracts were immunoblotted with antibodies against eNOS (Left) and β-actin (loading control, Right). (B) Transwell permeability assay in HUVEC transduced with LacZ or TR3-S ± l-NAME (mean ± SD, n = 4). At the 8.5-h time point, LacZ and TR3-S/l-NAME permeability differed significantly (P < 0.001) from TR3-S but not from each other (Tukey–Kramer test). (C) Miles permeability assays quantified as in previous figures. (Ci) BVP in peritoneum (mesentery) of EC-Nur77-S mice, greatly increased compared with wild-type mice, was significantly inhibited by l- but not d-NAME. (Cii) l-, but not d-NAME, greatly inhibited vascular permeability in Matrigel assays implanted in Nu/Nu mice in which P67 cells packaging retroviruses expressing Nur77-S were incorporated. (Ciii) Nur77-S packaging cells induced expected increased permeability in Matrigel assays in wild-type but not in eNOS^−/− mice. ***P < 0.001, Mann–Whitney test.

Like other orphan nuclear receptors, TR3 is comprised of ligand-binding (LBD), transactivation (TAD), and DNA-binding (DBD) domains, and both TAD and DBD are required for transcriptional activity (12). HUVEC transduced with TR3 constructs lacking the LBD domain (Fig. S6) expressed increased levels of eNOS, similar to those transduced with TR3-S, and eNOS expression was not further stimulated by added VEGF-A¹⁶⁵ (Fig. 3A). In contrast, HUVEC transduced with TR3 constructs lacking either the TAD or DBD domain (Fig. S6) only expressed low levels of eNOS that were not increased by VEGF-A¹⁶⁵. eNOS protein expression was greatly reduced in Nur77^−/− mice (Fig. 1D, lane 1 vs. 2). In addition, eNOS expression levels paralleled those of Nur77 (Fig. 1D, lanes 3–5) and both BVP and AVH in several different mouse strains (Fig. 1E).

TR3-S cells exhibited significantly greater permeability than LacZ-transduced cells in EC transwell monolayer assays, an effect reversed by N-nitro-l-arginine methyl ester (l-NAME) (Fig. 3B). Furthermore, the peritoneal (mesentery) hyperpermeability of Fvb EC-Nur77-S mice was inhibited >80% by l-NAME but not by d-NAME (Fig. 3Ci). In addition, l-NAME (but not d-NAME) significantly inhibited the permeability response observed in vivo in Matrigel plugs in which PT67 cells packaging Nur77-S had been incorporated (Fig. 3Cii), and permeability was greatly reduced when such Matrigel plugs were implanted in eNOS^−/− mice (Fig. 3Ciii).

TR3/Nur77 Suppresses Expression of Adherens and Tight Junction Proteins.

EC junction proteins have also been implicated in altering vascular permeability (13–15). Like control HUVEC, LacZ-transduced HUVEC expressed increasing levels of the mRNAs of both VE-cadherin and claudin 5 as they approached confluence, and this effect was strongly inhibited by Flag-fused TR3-S transduction (Fig. 4A and Fig. S7). We performed Western blots to determine whether TR3 affected the expression of HUVEC junction proteins. Expression of the adherens junction-associated proteins VE-cadherin, β-catenin, γ-catenin, and p120 was strikingly down-regulated in TR3-S HUVEC (Fig. 4B, lane 4 vs. lanes 2 and 3). Expression of the tight junction protein claudin 5 was also substantially down-regulated, and that of occludin to a lesser extent; however, JAM-B, ZO-1, and CD31 expression levels were not affected. Transduction of TR3-overexpressing HUVEC with Flag-fused VE-cadherin led to increased expression of VE-cadherin (Fig. S7), and partial restoration of β-catenin expression, but expression levels of the other junction proteins were unaffected (Fig. 4B, lanes 4–6). l-NAME did not prevent the down-regulation of VE-cadherin, and β- and γ-catenin, but did allow partial restoration of claudin 5 (Fig. 4B, lane 1 vs. 4).

Fig. 4.

TR3/Nur77 regulates EC junction proteins and permeability. (A) LacZ and TR3-S–transduced HUVEC plated at 30% density were harvested at indicated times for quantitative real-time RT-PCR assay of VE-cadherin and claudin 5 mRNAs, expressed as fold-change vs. the 12-h time point (mean ± SD, n = 2). (B) HUVEC transduced with indicated constructs were cultured for 48 h and immunoblotted for cell junction proteins. (C) Serum-starved HUVEC transduced with LacZ, TR3-AS, or TR3-siRNA were stimulated without (Left) or with 10 ng/mL VEGF-A¹⁶⁵ for indicated times and stained with DAPI and antibodies against VE-cadherin. (D) HUVEC transduced with LacZ or Flag-TR3-S were stained after 48-h culture with DAPI and indicated antibodies. (E) Transwell permeability assay on HUVEC that had been transduced with indicated constructs. Data (mean ± SD, n = 4). At the 8.5-h time point, LacZ and TR3-S+VE+Cl5 did not differ significantly from each other, but both differed significantly (P < 0.01–0.001, Tukey–Kramer test) from the three other conditions. (F) Matrigels containing P67 cells packaging viruses expressing indicated constructs were implanted subuctaneously in nude mice (n = 4). Matrigels were harvested at 24 h, 30 min after intravenous EB dye injection. Dye was extracted and quantified. Lane 1 differed significantly (P < 0.001) from lanes 2 to 4; lanes 2 and 3 differed significantly from lane 4 (P < 0.01); lane 2 vs. lane 3, NS, not significant (Tukey–Kramer test).

We used immunofluorescence to determine the effects of VEGF-A and TR3 on the cellular distribution of junction proteins. VEGF-A caused a rapid loss of junctional VE-cadherin in confluent HUVEC transduced with LacZ, but this effect was largely inhibited in HUVEC transduced with TR3-AS or TR3-siRNA (Fig. 4C). A similar loss of VE-cadherin and of several other junction proteins occurred when HUVEC were transduced with TR3-S in the absence of added VEGF-A (Fig. 4D). Expression of both VE-cadherin and claudin 5 mRNAs were variably increased or decreased in different organs of Nur77^−/− and EC-Nur77-S mice (Fig. S8).

Effects of TR3, VE-Cadherin, and Claudin 5 Expression on EC Monolayer Permeability.

Macromolecules extravasate across cultured EC monolayers, at least in part, by an inter- or paracellular pathway that requires opening of adherens and tight junctions (13–15). The effects of TR3/Nur77 on junction protein expression suggested that TR3 would affect EC monolayer permeability. HUVEC transduced with full-length TR3-S exhibited a substantial increase in monolayer permeability (Fig. 4E). TR3’s hyperpermeabilizing effect was partially abolished when TR3-S HUVEC were additionally transduced with either VE-cadherin or claudin 5 (Fig. S7), and transduction with both completely inhibited the effects of TR3 transduction (Fig. 4E). VE-cadherin and claudin 5 had similar, permeability-inhibiting effects in Matrigel assays in vivo. Inclusion of P67 cells packaging retroviruses expressing Nur77 cDNA rendered microvessels highly permeable to the Evans blue (EB) dye-albumin complex (Fig. 4F, lane 1). However, additional transduction with either VE-cadherin or claudin 5 reduced permeability by ∼50%, and, when both VE-cadherin and claudin 5 were transduced, permeability was reduced by 90% (Fig. 4F, lanes 2–4).

TR3/Nur77 Transcriptional Activity Is Required for Regulating VE-Cadherin Expression and Microvascular Permeability.

Both the transcriptional and DNA binding domains are required for VEGF-induced angiogenesis (6) and were essential for regulating eNOS expression (Fig. 3A). HUVEC transduced with Flag-tagged constructs exhibited, as expected, strong expression and nuclear staining (Fig. S9). TR3-S and TR3-ΔLBD–transduced HUVEC exhibited striking reductions in junctional VE-cadherin staining; however, cells transduced with either TR3-ΔTAD or TR3-ΔDBD exhibited normal, strong junctional VE-cadherin staining (Fig. 5A). Similar expression data were obtained for VE-cadherin in Western blots of HUVEC transduced with these constructs (Fig. 5B).

Fig. 5.

TR3/Nur77 transcriptional activity is required for regulating EC junction protein expression and permeability. (A) HUVEC transduced with indicated constructs were stained with DAPI and anti–VE-cadherin antibodies. (B) Cell extracts from HUVEC transduced as in A were cultured for 3 d and subjected to immunoblotting. β-Actin, protein loading control. (C) Matrigel plugs, with PT67 cells packaging retroviruses expressing Nur77-S, or indicated mutant constructs, were implanted, without or with SKMEL cells expressing VEGF-A¹⁶⁵, subcutaneously in nude mice and harvested 3 d later, 30 min after intravenous injection of EB dye. Dye was extracted and quantified. **P < 0.01, ***P < 0.001; NS, not significant (Tukey–Kramer test).

Finally, we performed Matrigel assays to determine whether transcriptional regulation was also essential for Nur77’s effects on vascular permeability in vivo (Fig. 5C). Inclusion of PT67 cells packaging retroviruses that expressed full-length Nur77-S, or Nur77 lacking the LBD domain, caused increased EB dye leakage, whether or not SKMEL/VEGF-A¹⁶⁵ cells were included. However, inclusion of cells packaging retroviruses that expressed Nur77 cDNAs lacking either the TAD or DBD domains strikingly inhibited vascular permeability, even when VEGF-A¹⁶⁵-secreting SKMEL cells were present.

TR3 Regulates eNOS and VE-Cadherin mRNA Stability.

TR3 regulates eNOS and VE-cadherin expression through transcriptional activity (Figs. 3 and 5). However, eNOS and VE-cadherin promoters lack TR3 consensus binding elements. Therefore, we hypothesized that TR3 might regulate eNOS and VE-cadherin by affecting their mRNA stability. To test this possibility, we transduced HUVEC with full-length TR3 or LacZ (control). After culture for 24 h, cells were treated with actinomycin D to inhibit mRNA synthesis. RNA was then collected at various times thereafter and subjected to real-time PCR with eNOS and VE-cadherin primers. Overexpression of TR3 increased eNOS mRNA stability but decreased VE-cadherin mRNA stability (Fig. 6).

Fig. 6.

 TR3 regulates eNOS and VE-cadherin mRNA stability. HUVEC transduced without or with LacZ (control) or TR3 were treated with actinomycin D for indicated times. RNA was subjected to real-time RT PCR with eNOS (Left) and VE-cadherin (Right) primers (n = 2 for real-time PCR assay).

Discussion

Our data indicate that several vascular permeabilizing agents, VEGF-A, as well as the small molecules histamine, serotonin, and PAF, strongly and rapidly induce TR3 mRNA and protein expression in vascular EC. These findings are of interest because these agents act through very different receptors: KDR/VEGFR-2 in the case of VEGF-A and different G protein-coupled receptors in the case of histamine, serotonin, and PAF (16). Furthermore, TR3/Nur77 had important roles in regulating vascular permeability in all three of its contexts, BVP, AVH, and CVH. Moreover, it did so downstream and independent of VEGF-A and, at least in part, by transcriptionally regulating eNOS and EC junction protein expression (Figs. 3 and 5). However, eNOS and VE-cadherin promoters lack TR3 consensus binding elements, indicating that TR3 regulates eNOS and VE-cadherin expression indirectly. TR3/Nur77 could act through additional transcription factors that do interact with the eNOS and VE-cadherin promoters. However, TR3 acts at least in part by regulating eNOS and VE-cadherin mRNA stability (Fig. 6), perhaps through microRNAs or RNA binding proteins. Further work will be required to sort out these mechanisms.

Our data raise several conundrums. The first is that Nur77^−/− mice are viable and develop a normal vasculature (17), whereas deletion of even one copy of the VEGF gene is lethal (18). This difficulty can be addressed by making the following reasonable assumptions: (i) Pathological angiogenesis and developmental angiogenesis differ in many respects (9) and TR3/Nur77 has an essential role in the former but not in the latter; (ii) VEGF-A's permeability-enhancing function, mediated through TR3/Nur77, is dispensable in developmental angiogenesis in that normally developing blood vessels do not exhibit increased permeability; and (iii) VEGF-A's endothelial cell viability function is mediated through pathways that do not involve TR3/Nur77; closely related TR3 family members (Nurr1, Nor-1) may play a compensatory role (19, 20).

A second conundrum concerns differences in vascular permeability in vivo and in vitro. The altered permeability induced by eNOS or alterations in EC junction proteins in transwell assays takes place over a period of hours (Figs. 3B and 4E). This finding stands in contrast to the AVH induced by VEGF-A¹⁶⁵, histamine, serotonin, or PAF in vivo. Single injections of any of these permeability factors into skin, peritoneal cavity, and so forth induce a dramatic local increase in vascular permeability that begins within a few minutes and is complete by 20 to 30 min. Furthermore, AVH in vivo is mediated, at least in part, by a transendothelial cell route through vesiculo-vacuolar organelles (VVOs), an interconnecting collection of vesicles and vacuoles that spans the cuboidal venular endothelium from lumen to albumin (21, 22). Cultured endothelial cells, in contrast, are flattened cells that contain few VVOs and that have lost the majority of their cytoplasmic vesicles and caveolae (23). They thus closely resemble the mother vessel EC associated with tumors and healing wounds (i.e., flattened, vesicle-poor EC lacking pericyte coverage) (8, 24, 25). We suggest, therefore, that transwell assays do not provide a useful model for BVP or AVH, but may closely model the CVH associated with mother vessel permeability in pathological angiogenesis.

The final conundrum also involves the role of TR3/Nur77 in AVH and is the most puzzling. On the one hand, VEGF-A¹⁶⁵ and the several other permeability factors tested were unable to increase permeability when injected into TR3/Nur77^−/− mice, even at supraphysiological doses (Fig. 1C); therefore, TR3/Nur77 clearly has an essential role in AVH. However, the kinetics of AVH is much too rapid to be explained by activation of a transcription factor that acted to induce expression of downstream genes or regulate mRNA stability. One possibility is that these permeability factors interact with TR3/Nur77 by an additional second mechanism that does not involve transcription. There is precedence for this in that another TR3 function, that of inducing apoptosis in tumor cells, is independent of transcription and occurs when TR3 transmigrates from the nucleus to the cytoplasm (12, 26, 27). Another possibility is that TR3/Nur77, although necessary, is not by itself sufficient for inducing the AVH induced by acutely acting permeability factors. Perhaps these vascular permeabilizing factors act additionally at steps downstream of TR3/Nur77, for example to increase eNOS expression, destabilize EC junctions, open VVOs, and so forth. At present we have no evidence to distinguish among these possibilities and sorting them out will require additional investigation.

Materials and Methods

See SI Materials and Methods for further details. Experiments were repeated at least three times.

Animals.

Nur77^−/− mice were originally obtained from J. Millbrandt (Washington University School of Medicine, St. Louis, MO) (17). Doubly transgenic mice overexpressing Nur77-sense (S) or dominant-negative Nur77 (Nur77-DN) in vascular endothelial cells under the control of a tetracycline-regulated promoter were derived as described in SI Materials and Methods. Tetracycline, l-NAME and d-NAME were administered in drinking water.

Cell Culture and Monolayer Permeability Assay.

For cell culture, 3 × 10⁵ HUVEC (Clonetics) were seeded on polycarbonate membranes in 24-well transwell chambers and transduced with retroviruses (6). Three days later, after cells achieved confluence, 70 KDa fluoresceinated-dextran (Molecular Probes) was added to the upper chamber. Dye passing through the monolayer from the upper to the lower chamber was measured at successive intervals by fluorimetry (28).

Other Assays.

The Miles assay, modified in vivo Matrigel assays, Western blots, and immunohistochemistry were performed as previously described (6). Statistical tests are indicated in the figure legends.