Vascular Abnormalities and Deregulation of VEGF in Lkb1-Deficient Mice

See allHide authors and affiliations

Science  17 Aug 2001:
Vol. 293, Issue 5533, pp. 1323-1326
DOI: 10.1126/science.1062074


The LKB1 tumor suppressor gene, mutated in Peutz-Jeghers syndrome, encodes a serine/threonine kinase of unknown function. Here we show that mice with a targeted disruption ofLkb1 die at midgestation, with the embryos showing neural tube defects, mesenchymal cell death, and vascular abnormalities. Extraembryonic development was also severely affected; the mutant placentas exhibited defective labyrinth layer development and the fetal vessels failed to invade the placenta. These phenotypes were associated with tissue-specific deregulation of vascular endothelial growth factor (VEGF) expression, including a marked increase in the amount of VEGF messenger RNA. Moreover, VEGF production in culturedLkb1−/− fibroblasts was elevated in both normoxic and hypoxic conditions. These findings place Lkb1in the VEGF signaling pathway and suggest that the vascular defects accompanying Lkb1 loss are mediated at least in part by VEGF.

Germ line mutations of theLKB1 gene cause Peutz-Jeghers syndrome, which is characterized by gastrointestinal polyposis, abnormal melanin pigmentation, and increased risk of cancer (1). TheLkb1 gene encodes a serine/threonine kinase of unknown function with no identified in vivo substrates and a ubiquitous expression pattern during mouse development (2). To study the function of Lkb1, we generated Lkb1-deficient mice.

Two independent gene-targeting strategies were used to functionally disrupt Lkb1 in the murine germ line (3, 4). In Lkb1 heterozygous (Lkb1+/− ) intercrosses, bothLkb1+/+ (n = 87) andLkb1+/− (n = 177) animals were observed at expected frequencies, whereas noLkb1−/− animals were obtained. Analysis ofLkb1−/− embryos throughout embryonic development revealed no abnormalities before embryonic day 7.5 (E7.5), and most embryos appeared to develop normally up to E8.0. Macroscopic analysis of Lkb1−/− embryos beyond E8.25 revealed multiple abnormalities, including a failure of the embryo to turn, a defect in neural tube closure, and a hypoplastic or absent first branchial arch (Fig. 1A). No viable embryos were recovered after E11.0, indicating that Lkb1 is essential for embryonic development.

Figure 1

Characterization of developmental arrest inLkb1−/− embryos. (A) Light microscopy of E9.25 Lkb1+/+ (+/+) andLkb1−/− (−/−) embryos showing unturned embryos with open neural folds, a missing branchial arch (arrow), and aberrant somites (arrowhead). (B and C) Whole-mount in situ hybridization of E9.5 (+/+) and (−/−) embryos with Brachyury (T) expressed in the notochord andEngrailed1 (En1) expressed at the mid-hindbrain junction and in the somites.

Whole-mount in situ hybridization was used to study the integrity of various developmental lineages in theLkb1−/− embryos at E8.5 and E9.5 (5). The expression of the mesodermal marker brachyury (6) (T; Fig. 1B) showed that although the notochord developed along the full length of the anterior/posterior axis in the mutant embryos, it was misaligned and contorted. This was accompanied by defective somitogenesis, which resulted in dysmorphic protrusive somites (arrowhead in Fig. 1A), which failed to express engrailed 1 (7) at E9.5 (En1; Fig. 1C). No pronounced changes in the expression of Wnt3A(E8.5), Fgf8 (E9.0), or Krox-20(E9.0) were noted (8), suggesting that there was normal development of the mesoderm of the tail bud, forebrain, and primitive streak, and normal segmentation of the hindbrain.

Mutant embryos at E9.25 had a translucent appearance, suggesting the possibility of vascular defects (Fig. 1A). To visualize endothelial cells, we subjected E8.5 and E9.5 embryos to whole-mount immunostaining for the platelet endothelial cell adhesion molecule–1 (PECAM-1) (9). Both mutant and wild-type embryos developed a paired dorsal aorta, but by E8.5, the mutant aorta was thin and discontinuous, particularly in the anterior part of the vessel (arrowhead in Fig. 2A). At E9.5, the lumen of the mutant aorta remained thin (arrows in Fig. 2B), with intersomitic branches terminating prematurely in the mesenchyme.

Figure 2

(A and B) Whole-mount PECAM-1 immunostaining of E8.5 (A) and E9.5 (B)Lkb1+/+ (+/+) andLkb1−/− (−/−) embryos. Discontinuities of the aortic vessel are indicated with an arrowhead in (A), and the decrease in the luminal diameter of the dorsal aorta is noted by arrows in (B). (C) Hematoxylin and eosin (H&E)–stained transverse sections from E9.5 (+/+) and (−/−) head folds demonstrating large cystic degenerations (asterisks) and aberrant mesenchyme in the mutant. (D) Cell death in E9.5 (+/+) and (−/−) cephalic mesenchyme as assessed by TUNEL labeling (T). Sections were counterstained with Hoechst (H). (E) Whole-mount smooth-muscle actin immunostaining of E9.5 (+/+) and (−/−) embryos. The absence of signal surrounding the mutant dorsal aorta and the ectopic VSMC signal in the head folds of the mutant embryo is apparent.

Histological sections of E9.5 Lkb1−/− neural folds often revealed large cystic degenerations near the dorsal aorta that occasionally contained embryonic blood cells (asterisks in Fig. 2C). Additionally, the surrounding cephalic mesenchyme had a lower cell density and fewer developing capillaries than controls (Fig. 2C). The decreased cell density was due to increased cell death in the mesenchyme as determined by terminal deoxynucleotidyl transferase–mediated dUTP nick end-labeling (TUNEL) (Fig. 2D) (10). This phenotype was limited to the E9.5 embryos; the E8.5 embryos showed similar numbers of TUNEL-positive nuclei regardless of genotype (8).

In addition to the endothelial and mesenchymal defects, abnormalities were also observed in vascular smooth muscle cells (VSMCs) in E9.5 embryos as revealed by staining with antibodies to smooth-muscle actin (Sm-A) (9). Although Sm-A expression in the developing heart was comparable in mutant and control embryos (Fig. 2E), Lkb1−/− embryos showed a complete absence of VSMC staining in the dorsal aorta and somites. In addition, an unusual, strong ectopic Sm-A signal was detected in head folds of mutant embryos (Fig. 2E). This ectopic VSMC staining was evenly distributed within the cephalic mesenchyme and did not appear to contribute to the supportive vascular structures (8). These observations suggest that abnormal VSMC development may be one of the factors contributing to the vascular and mesenchymal defects.

Analysis of extraembryonic tissues at E9.5 revealed that the mutant yolk sacs failed to develop large vitelline vessels and an extensive capillary network (Fig. 3A), and they contained large cavities that sequestered vast pools of embryonic blood (arrow in Fig. 3A). There were rudimentary vessels between parietal and visceral leaves of the yolk sac, which were congested by nucleated embryonic blood cells (Fig. 3B). The vitelline artery was completely atretic inLkb1−/− yolk sacs, effectively disconnecting the embryo from the yolk sac (vitelline) circulation.

Figure 3

Characterization of E9.5Lkb1+/+ and Lkb1−/− yolk sacs (A and B) and placentas (A, C to E). (A) Yolk sac vasculature of E9.5 (+/+) and (−/−) conceptuses. The rudimentary vasculature, lack of vitelline vessels, and pools of embryonic blood (arrow) are apparent in the mutant. (B) Cross section of H&E-stained wild-type yolk sac, demonstrating congestion of mutant blood islands with nucleated embryonic blood cells (−/−). Magnification, ×400. (C) Low-power (×40) cross section of (+/+) and (−/−) placentas with labyrinthine (la), giant cell (gc), spongio (sp), and maternal (ma) layers, demonstrating shallow invasion of decidua and rudimentary labyrinth layer in the mutant. (D and E) High magnification (×100) of fetal vessels invading the labyrinth layer in (+/+) [arrows in (D), HE], but not in (−/−) placenta. In situ hybridization of adjacent sections with flk-1 and flt-1, showing the absence of flk-1 in the mutant labyrinth layer (E) (flk-1, la) and comparable expression of flt-1 in wild-type and mutant spongiotrophoblast layers [flt-1 in (D) and (E)].

High expression of Lkb1 mRNA in the placenta (4) suggested that placental development might also be compromised in the Lkb1−/− conceptuses. Mutant placentas at E9.5 were oedematous, hemorrhagic, and small in diameter (Fig. 3A). The connection between placenta and the embryo (chorio-allantoic fusion) in Lkb1−/− conceptuses was delayed compared with controls, but had occurred by E9.5. However, invasion of embryonic blood vessels (arrows in Fig. 3D) into the placenta did not occur in the mutant (Fig. 3E), and embryo-derived (nucleated) erythrocytes were not detected in mutant placentas. The lacunae in Lkb1−/− placentas were abnormally congested with maternal erythrocytes and by E10.5 the blood lacunae had ruptured, leading to massive hemorrhaging. In situ hybridization (10) with a probe for VEGF receptorflk-1, which detects migrating allantoic vessels in the placenta (11), confirmed the lack of fetal blood vessels in the rudimentary labyrinth layer (la in Fig. 3E). The spongiotrophoblast layer (sp in Fig. 3, D and E) was less organized in the mutant and contained more VEGF receptor flt-1 (12) negative giant cells than did the controls. As expected, flt-1 was expressed in the maturing fetal vessels in the labyrinth layer of the wild-type (la in Fig. 3D) but not in the mutant placentas (Fig. 3E). The spongiotrophoblast marker sna1 and the vascular development marker Tgf-β1 were expressed normally in the mutant placentas (8).

In the embryo flk-1, flt-1,sna1, and Tgf-β1 were expressed at comparable levels (8). However, expression of another key regulator of embryonic vascular development, VEGF(13, 14), was found to be deregulated in both the embryonic and extraembryonic compartments at E8.5 (8) and E9.5 (Fig. 4A). The Lkb1−/− placentas exhibited markedly diminished VEGF mRNA expression (p in Fig. 4A) particularly in the trophoblast giant cells (asterisk in Fig. 4A), whereas the Lkb1−/− embryos expressed abnormally elevated levels of VEGF in several tissues including the mesenchyme, heart, and yolk sac (y and e in Fig. 4A).

Figure 4

Deregulation of VEGFexpression after loss of Lkb1 in embryos (A) and in cell culture (B and D). (A) VEGF in situ hybridization analysis of cross sections of E9.5 (+/+) and (−/−) conceptuses with embryo (e), yolk sac (y), placenta (p), and trophoblast giant cells (asterisks) indicated in the dark-field images (×20). (B) VEGF levels in supernatants of (+/+) and (−/−) MEF cultures subjected to 1% O2 for 24 hours. Error bars indicate the standard deviation in independent VEGF ELISA measurements. (C) Western blot analysis of glucose transporter 1 (Glut-1) in normoxia (21% O2) and hypoxia (1% O2) in lysates from (+/+) and (−/−) MEFs, showing comparable induction of Glut-1 in both. (D) VEGF levels in supernatants of wild-type (w1, w2, w3) and mutant (m1, m2, m3) cultures, each generated from independent littermate embryos. Error bars as in (B).

To investigate whether Lkb1 directly regulates VEGF expression, we isolated primary mouse embryonic fibroblasts (MEFs) from E8.5 wild-type and mutant embryos, expanded them in culture for 2 weeks (15), and then subjected the cells to hypoxic conditions (1% O2) for 24 hours to induce VEGF expression (16). Analysis of VEGF levels in culture supernatants (Fig. 4B) revealed that the mutant MEFs produced significantly higher levels of VEGF than controls (165.0 ± 31.9 versus 44.5 ± 6.2 pg/ml, respectively) in hypoxic conditions.

The elevated VEGF levels in the mutant MEF could have resulted from an increased HIF-1–mediated hypoxia response or from a deregulation of basal VEGF expression. Because the HIF-1 levels remained undetectable in MEFs, we assayed the expression of Glut1 (17), a HIF-1–responsive marker of hypoxia, before and after exposure of the cells to hypoxia (Fig. 4C). There was a comparable induction of Glut1 in wild-type and mutant MEFs, suggesting that loss of Lkb1 had not changed the HIF-1 response to hypoxia. In addition, VEGF secretion in normoxic conditions was significantly higher in the mutant than in wild-type MEFs (80.3 ± 13.9 versus 12.3 ± 5.4 pg/ml) (Fig. 4D). These results indicate that loss of Lkb1 leads to increased basal and induced expression of VEGF in fibroblasts.

In summary, we have demonstrated that disruption ofLkb1 results in multiple developmental defects resulting in embryonic lethality and establishes Lkb1 as a critical regulator of mammalian vascular development. This phenotype was associated with tissue-specific deregulation of VEGFexpression. Whereas VEGF was significantly down-regulated in extraembryonic tissues, embryonic VEGF was markedly up-regulated. Disruption of the murine VHL tumor suppressor gene exhibits a similar extraembryonic down-regulation ofVEGF (18), and while the expression ofVEGF in VHL−/− embryos has not been reported, VHL loss in other systems has been found to lead to up-regulation of VEGF (19, 20). Exploring well-characterized pathways regulating VEGFexpression, we found unexpectedly that HIF-1 does not appear to play a role in Lkb1-mediated VEGF expression based on unaltered Glut-1 levels (Fig. 4C) and undetectable HIF-1α inLkb1−/− MEF cultures (8). Similarly, neither the p42/44MAPK nor the p38 kinase pathways appear to be deregulated based on unaltered levels of activated p42/44MAPK or p38 kinases inLkb1−/− MEF cultures (8).

Our findings have established a genetic link betweenLkb1 and VEGF regulation, thereby placing Lkb1 in the VEGF signaling pathway. This suggests that the vascular defects accompanying Lkb1 loss that we describe are mediated at least in part by VEGF, thereby providing a basis for the vascular phenotype. Most important, our results also provide a rationale for the increased risk of cancer incidence in Peutz-Jeghers patients (21) by demonstrating that loss of Lkb1 confers an increased angiogenic potential in certain cell types by up-regulation of VEGF.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: tomi.makela{at}


View Abstract

Stay Connected to Science

Navigate This Article