1. Volume 41, Issue 2, Pages 274-283 (August 2004)
Disruption of hepatocellular tight junctions by vascular endothelial growth factor (VEGF): a novel mechanism for tumor invasion 
Marcus Schmitt, Axel Horbach, Ralf Kubitz, Andrea Frilling, Dieter Häussinger 
Journal of Hepatology 
Volume 41, Issue 2, Pages (August 2004)
DOI: /j.jhep

2. Fig. 1
Localization of occludin, ZO-1 and MRP2 in Hep G2 cells. Hep G2 cells grown to subconfluency (G), are differentiated and possess apical plasma membrane domains, forming pseudocanaliculi. Immunostaining of the apical ABC-transporter MRP2 (green) outlines the apical poles of the Hep G2 cells (A). These MRP2 containing pseudocanaliculi are delineated by small, belt-like occludin bands (red) representing the tight junction border between apical and basolateral plasma membrane domains (B, merged C). Costaining of the transmembrane tight junction protein occludin (red, D) and of the cytoplasmatic protein ZO-1 (green, E) results in colocalized, belt or basket like staining patterns representing the tight junctions. Colocalization of occludin and ZO-1 results in a yellow signal (F).
Journal of Hepatology  , DOI: ( /j.jhep )

3. Fig. 2
VEGF-induced disappearance of occludin-delineated pseudocanaliculi in Hep G2 cells. Subconfluent Hep G2 cells were incubated in culture medium containing VEGF (□) at concentrations of 5 ng/ml (0.12nmol/l; Fig. 2A), 50 ng/ml (1.2nmol/l; Fig. 2B) or 500 ng/ml (12nmol/l; Fig. 2C) Cells were fixed at the time points indicated and immunostained for occludin. Cell nuclei were stained with DAPI. The ratio of ring- or belt-like occludin structures and nuclei represents the ‘canalicular index’, i.e. a measure for the number of pseudocanaliculi delineated by tight junctions. After 45 min of VEGF-stimulation the canalicular index was significantly reduced compared to controls (●). VEGF-induced occludin disappearance is prevented by Gö6850 (1μmol/l), an inhibitor of most PKC isoforms (▴). Results are expressed as means±SD; n=3 independent experiments.
Journal of Hepatology  , DOI: ( /j.jhep )

4. Fig. 3
Effect of PMA on occludin-delineated pseudocanaliculi in Hep G2 cells and its inhibition by Gö6850. Hep G2 cells were exposed to PMA (100nmol/l)-containing or control medium. PMA significantly (P<0.05) reduced the canalicular index (♦), whereas PMA-free control medium was ineffective (■). In the presence of Gö6850 (1μmol/l), the PMA (100nmol/l) effect on the canalicular index, i.e. tight junction disappearance was fully abolished (▵). Results are expressed as means±SD; n=3 independent experiments.
Journal of Hepatology  , DOI: ( /j.jhep )

5. Fig. 4
Dose dependence of PMA-induced tight junction loss in Hep G2 cells. Hep G2 cells were stimulated with increasing PMA concentrations (♦) for 4 h or with medium alone as control condition (■). A further control was an experiment with only Gö6850 (1μmol/l) added to the medium (▵). PMA significantly (P<0.0001) reduced the canalicular index in a concentration dependent manner. The half maximal effect on tight junctions was observed between 0.1 and 1nmol/l PMA. Gö6850 has by itself no effect the canalicular index. Results are expressed as means±SD; n=3 independent experiments.
Journal of Hepatology  , DOI: ( /j.jhep )

6. Fig. 5
VEGF and PMA have no short term effects on occludin protein levels in HepG2 cells. (A) Neither VEGF (50 ng/l) nor Gö6850 (1μmol/l) have any significant effect on occludin protein levels over a time period of 45 min, which is sufficient to induce tight junctional disintegration (compare Fig. 2). (B) Also PMA (100nmol/l) stimulation resulted in no significant changes of occludin protein levels as detected by immunoblotting. Data from densitometric analysis from 5 experiments are given by the numbers below the blots (relative protein levels) and show no significant difference in the occludin protein levels. For densitometric analysis the values found for control conditions (15 min in A) were arbitrarily set to 1.
Journal of Hepatology  , DOI: ( /j.jhep )

7. Fig. 6
Effect of VEGF on pseudocanalicular sealing and on PKC-α membrane translocation. (A–D): HepG2 cells were incubated with CMFDA (2.5μmol/l) and were allowed to accumulate fluorescence in the pseudocanaliculi for 30 min, which was detected by confocal laser scanning microscopy as described in methods (A,C). The fluorescence accumulated in the pseudocanaliculi was maintained for another 45 min under control conditions (B), but was completely lost when VEGF (50 ng/ml) was added for 30 min (D), suggesting a VEGF-induced loss of secreted glutathione-methylfluorescein from the pseudocanaliculi due to disintegration of the tight junction barrier (inserts show corresponding Hep G2 cells, representative pictures from three independent experiments). (E,F) Subconfluent monolayers of isolated rat hepatocytes were incubated with the fluorescent bile acid analog cholyl-lysyl-fluorescein (CLF 2μmol/l) for 15 min. The number of pseudocanaliculi retaining the accumulated CLF was visualized by confocal microscopy. Under control conditions (E) many fluorescence-containing pseudocanaliculi are detected (arrow), but not after stimulation with PMA (100nmol/l) (F). (G,H) Subconfluent Hep G2 cells were stimulated for 15 min with control medium (G) or a medium containing VEGF (50 ng/ml; H). Cells were fixed and immunostained for the PKC-α isoform (green) and counterstained with propidium iodide for detection of cell nuclei. Confocal laser scanning microscopy shows a VEGF-induced membrane translocation of the PKC-α isoform (H), whereas a diffuse cytoplasmatic PKC-α staining is found under control conditions (G).
Journal of Hepatology  , DOI: ( /j.jhep )

8. Fig. 7
(A) Immunoblots of membrane-bound PKC-α. Membrane and cytosolic fractions of VEGF-stimulated HepG2 cells were probed for PKC-α: 15 min VEGF (50 ng/ml) addition, an increase of PKC-α in the membrane-fraction is found, indicating a translocation and activation of PKC-α. In line with this the cytosolic-fraction of PKC-α decreased. Efficacy of fractionating is shown by immunodetection of GAPDH and annexin V as cytosolic and membrane markers, respectively. (B) Immunoblots of serine- and tyrosine-phosphorylation of occludin immunoprecipitated from the membrane fraction of HepG2 cells treated with VEGF. Hep G2 cells were stimulated for 15 min with control medium (lane 1), or with VEGF (50 ng/ml) (lane2). Occludin was immunoprecipitated from the membrane fraction, separated by SDS-PAGE, transferred to nitrocellulose and immunoblotted with an anti-phosphoserine or anti-phosphotyrosine monoclonal antibody to detect the presence of serine- or tyrosine-phosphorylated occludin. 15 min after VEGF stimulation a decrease of serine-phosphorylated occludin was detected. Representative blots are shown.
Journal of Hepatology  , DOI: ( /j.jhep )

9. Fig. 8
Effect of VEGF on MAP kinase activation in Hep G2 cells and VEGF protein expression in hepatocellular carcinoma and surrounding normal liver tissue. (A,B) Hep G2 cells were incubated without (0 min) or with VEGF (50 ng/ml) for the indicated time and lysed. ERK1/2 (A) and p38MAPK (B) activities were measured by immune complex assays. VEGF has no effect on MAPK activities. Representative experiments from a series of 3 experiments are shown.(C) VEGF expression (arrow) in normal (tumor surrounding) liver and hepatocellular carcinoma tissue specimens from three different patients as detected by immunoblotting. VEGF overexpression is found in all 3 HCCs studied compared to the corresponding normal liver tissue.
Journal of Hepatology  , DOI: ( /j.jhep )

10. Fig. 9
Distribution of the tight junction protein occludin in human hepatocellular cancer and surrounding liver tissue. Snap frozen liver tissue was immunostained for the tight junction proteins occludin and claudin1 and analyzed by confocal laser scanning microscopy. Specimens were sectioned from the tumor and the surrounding liver of three different patients. Representative images of one patient are shown. In (A,C,E) magnification with the 20× objective, in B,D,F magnification with the 63× objective was used to obtain higher resolution. (A,B) Tumor: hepatocellular carcinoma tissue shows very small, irregular tight junctional complexes (occludin red; claudin1 green; colocalization resulting in yellow signals). (C,D) Adjacent zone between tumor and normal liver tissue. For details the area within the white rectangle in C is enlarged (D). Normal liver tissue some micrometers away from the tumor shows a regular tight junction staining pattern outlining the canalicular structures of the normal liver (D, arrow 1) Tight junctions of hepatocytes in the immediate vicinity of the tumor show a condensed, irregular staining pattern (arrows 2 and 3). Some of the surrounding hepatocytes lack any regular costained tight junction structure (arrows 4). (E,F) Normal corresponding liver tissue with regular tight junctional staining patterns.
Journal of Hepatology  , DOI: ( /j.jhep )