Increased angiogenesis and permeability in the mesenteric microvasculature of rats with cirrhosis and portal hypertension: an in vivo study

Basic Studies

Increased angiogenesis and permeabilityin the mesenteric microvasculature of ratswith cirrhosis and portal hypertension: anin vivo study

Anja M. Geerts1, An S. De Vriese2,Eline Vanheule1, Hans VanVlierberghe1, Siska Mortier2, Kin-JipCheung1, Pieter Demetter3, NorbertLameire2, Martine De Vos1, IsabelleColle1

Departments of 1Hepatology &

Gastroenterology, 2Department of Nephrology,

and 3Department of Pathology, Ghent University

Hospital, Ghent, Belgium

Geerts AM, De Vriese AS, Vanheule E, Van Vlierberghe H, Mortier S,Cheung KJ, Demetter P, Lameire N, De Vos M, Colle I. Increasedangiogenesis and permeability in the mesenteric microvasculature of rats withcirrhosis and portal hypertension: an in vivo study.Liver International 2006: 26: 889–898.r 2006 The Authors. Journal compilation r 2006 Blackwell Munksgaard

Abstract: Background: In vivo evidence for angiogenesis in the splanchnicvasodilation in portal hypertension (PHT) and cirrhosis is lacking. Vascularendothelial growth factor (VEGF) and endothelial nitric oxide synthase(eNOS) are mediators of angiogenesis. The present study visualises in vivostructural changes (angiogenesis and vascular hyperpermeability) andexamines the presence of VEGF and eNOS in the mesentericmicrovasculature of animal models of PHT with and without cirrhosis.Methods: Portal hypertension was induced by partial portal vein ligation(PPVL) and cirrhosis was induced by common bile duct ligation (CBDL) inrats. The mesenteric microcirculation was examined by intravital microscopy.Expression of VEGF, eNOS and CD31 in mesenteric tissue were studied byimmunohistochemistry. Results: An increased mesenteric angiogenesis wasobserved in PPVL and CBDL rats compared with Sham-operated andcontrol rats, as shown by intravital microscopy and CD 31 staining. VEGFand eNOS expression was higher in CBDL and PPVL rats compared withcontrol groups and correlated positively with vascular density.Macromolecular leakage was increased in cirrhotic rats compared withcontrol and PPVL rats. Conclusion: Our study provides in vivo evidence of anincreased angiogenesis in the mesenteric microvasculature of animal modelsof PHT and cirrhosis. Increased VEGF and eNOS expression in themesentery of PPVL and CBDL rats may suggest their contribution.Microvascular permeability in the mesenteric vessels was only increased incirrhotic rats.

Key words: angiogenesis – cirrhosis –

endothelial nitric oxide synthase – intravital

microscopy – portal hypertension – vascular

endothelial growth factor

Anja M. Geerts, MD, Department of Hepatology

& Gastroenterology, Ghent University Hospital,

De Pintelaan 185, 9000 Ghent, Belgium.

Tel: 132 9 2402371

Fax: 132 9 2404984

e-mail: [email protected]

Received 20 January 2006,

accepted 4 May 2006

��������������������������������������������������

��������������������������������������������������

Portal hypertension (PHT) is characterized by asplanchnic vasodilation. Different pathophysio-logical mechanisms have been proposed to beinvolved in this vasodilatory state. Most studiesfocused their attention on the potential role offunctional changes in the splanchnic territory,such as an increased production of nitric oxide(NO) predominantly by the endothelial nitricoxide synthase (eNOS) isoform (1). Nevertheless,the role of structural changes like angiogenesishas been suggested by experimental studies in thepathophysiology of portal hypertension and cir-rhosis (2–7). Sumanovski et al. (3, 4) reported an

increased angiogenesis in the abdominal cavity ofportal hypertensive rats without cirrhosis.Furthermore, recently published data suggesteda role for angiogenesis by diminishing the devel-opment of the hyperdynamic splanchnic circula-tion and portal-systemic collateral vessels inportal hypertensive rats by blockade of the vas-cular endothelial growth factor (VEGF) signal-ling pathway (6, 7).However, in vivo evidence of structural changes

in the splanchnic system directly related to cir-rhosis is lacking. Therefore, the first goal of thepresent study is to visualise in vivo structural

Liver International 2006: 26: 889–898 r 2006 The AuthorsJournal compilation r 2006 Blackwell Munksgaard

DOI: 10.1111/j.1478-3231.2006.01308.x

889

changes in the mesenteric microvasculature ofanimal models of portal hypertension with orwithout cirrhosis. Furthermore, an immunohis-tochemical analysis of VEGF and eNOS expres-sion is performed to elucidate the underlyingresponsible mechanisms contributing to angio-genesis. VEGF is a well-known mediator ofphysiological and pathological angiogenesis andvascular permeability (8). An upregulation ofVEGF in cirrhotic livers of bile-duct ligated ratshas been previously shown by Rosmorduc et al.(9, 10) These papers suggest that VEGF might beinvolved in cirrhosis-associated angiogenesis.Several lines of evidence suggest that NO mayalso regulate angiogenesis and vascular perme-ability (11, 12). Thereby, VEGF is known tostimulate eNOS expression in endothelial cellsand there is evidence that NO acts as a down-stream mediator of VEGF-induced hyperperme-ability and neoangiogenesis (13).The second aim of the present study is to

explore in vivo whether blood vessels in themesenteric microvasculature of portal hyperten-sive and cirrhotic rats have an increased perme-ability. An increased extravascular leakage in thelung and kidney of rats with biliary cirrhosis hasbeen shown previously (14). A possible role ofenhanced microvascular leakage in the splanchnictree is also suggested by Wirth et al. (15) but isnot yet demonstrated in vivo in the mesentericmicrocirculation.

Materials and methods

Animals

The experiments were performed in 48 maleWistar rats (Iffa Credo, Brussels, Belgium) thatreceived care in accordance with the nationalguidelines for animal protection. The EthicalCommittee of experimental animals at the Fa-culty of Medicine and Health Sciences, GhentUniversity, Belgium, approved the protocols.

Animal model of portal hypertension (n 5 12)

Induction of prehepatic portal hypertension with-out cirrhosis was performed by partial portal veinligation (PPVL), as previously described (16).Briefly, rats were anaesthetized under halothaneinhalation (Fluothanes, Zeneca NV, Destelber-gen, Belgium). A midline abdominal incision wasperformed and the portal vein was separatedfrom surrounding tissue. A ligature (silk cut 3-0) was tied around both portal vein and adjacent20-gauge blunt-tipped needle. Subsequent re-moval of needle yielded a calibrated stenosis ofthe portal vein. Afterwards, the abdominal wall

was closed by suturing abdominal muscle (silk cut5-0) and clipping skin. An intramuscular injectionof buprenorphine (Temgesics, 0.1ml/kg/12 h,Schering-Plough, Brussels, Belgium) was givenduring a period of 48 h after surgery to providea good postoperative analgesia.

All studies were performed 14 days after PPVL,a period where hyperdynamic circulation andportal hypertension are known to be fully estab-lished (16).

Animal model of cirrhosis (n 5 12)

Secondary biliary cirrhosis was induced by com-mon bile duct ligation (CBDL), as describedpreviously (16, 17). In brief, under halothaneinhalation anaesthesia, a midline abdominal inci-sion was made and the common bile duct wasisolated. The common bile duct was occludedwith a double ligature of a non-resorbable suture(silk cut 7-0). Before tying the upper part of thecommon bile duct, the duct was rinsed with salineand 150 ml formalin 10% was injected. Formalin10% caused a sclerosing cholangitis and pre-vented the formation of a bilioma. The commonbile duct was resected between the two ligatures.Postoperative care was given as described above.

The experiments in this group were done 4weeks after surgery, when the rats had developedcirrhosis (16, 17).

Control (n 5 12) and Sham-operated (n 5 12) animals

Control rats were not operated to exclude apotential influence by surgical manipulation.

In sham-operated rats, the abdominal cavitywas opened and the portal vein or common bileduct was isolated, but no ligature was placed.Postoperative care was given as described above.The data shown were obtained after 2 weeks inthe control and Sham group.

Preliminary experiments showed no significantdifferences in vascular density and permeabilitybetween Sham-operated rats 2 or 4 weeks afterinduction (data not shown). This was also ob-served for controls rats (data not shown).

Intravital microscopy

All experiments were performed after a 12 hfasted state. Rats were anaesthetized with thio-butabarbital (Inactins 100mg/kg, Sigma, St.Louis, MO). The trachea was intubated to main-tain a patent airway throughout the experiment.A polyethylene catheter was inserted in the jugu-lar vein for continuous infusion of isotonic saline,and a second catheter was placed in the carotidartery for blood pressure recording. A midline

890

Geerts et al.

abdominal incision was made and a loop of smallbowel was exteriorized. The mesentery wasspread over a plexiglas plate and superfusedcontinuously with an Earle’s balanced salt solu-tion (EBSS, MP Biomedicals Inc., Aurora, OH)heated at 37 1C (Fig. 1A). Observations weremade with an Axiotech Vario 100 HD micro-scope (Zeiss, Jena, Germany) using water immer-sion objectives (Achroplan � 10, � 40, Zeiss)(Fig. 1B). The microscopical stage was drivenby a stepping motor control MCL-2 (Lang,Huttenberg, Germany), operated by a joystickor a software program (Wincommander, Marz-hauser-Wetzlar, Germany) via a RS-232 inter-face. The tissue was transilluminated via afiberoptic using a light source (KL 1500, Schott,Wiesbaden, Germany) equipped with a 150Whalogen lamp. Epifluorescence was performedwith a mercury lamp HBO 50W and a filter set(excitation filter BP 450–490 and emission filterLP 520, Zeiss). The resulting image was displayedon a television monitor by a TK-1281 camera(Victor Company of Japan LTD-JVC, Tokyo,Japan) and recorded by a video recorder (S-VHSPanasonic AG-7350, Matsushita, Japan) for off-line analysis. The video images were analyzedwith an image analysis software program (Cap-image, Ingenieursburo Zeintl, Heidelberg, Ger-many) (16–19). The technique allowed the studyof the peritoneal microcirculation including smallarteries, arterioles, capillaries, postcapillary ve-nules, venules and small veins. Large vessels weresurrounded by fat and were not routinely visua-lized. To evaluate microvascular density, a pre-defined segment was chosen to avoid selectionbias. The objective (� 10) was positioned atrandom in the segment of the distal ileum, prox-imal to the caecum. The microscopical stage wasdriven through a meander consisting of five stepsof 1mm in the X direction and five steps of 1mmin the Y direction. The microscopical image was

recorded at each of these positions. The totalvessel length (arterioles, venules and capillaries)per area (L/A, cm/cm2) was determined for eachmicroscopical image with image analysis soft-ware, as previously described (18, 19).To evaluate macromolecular leakage, a venular

segment with a diameter of 20–40mm and anunbranched length of about 150mm was selectedfor study. Fluorescent isothiocyanate-bovineserum albumin (FITC-albumin, 50mg/kg, Sigma)was given as an intravenous bolus. The experi-ment was discontinued when traumatic leakswere observed. Epifluorescence recordings weremade every 10min during a period of 60min. Onthe recorded image of the venule, two intralum-inal areas and two contiguous areas of perivenu-lar interstitium were defined. The average greyscale value, ranging from 0 for black to 255 forwhite, was calculated for each area. The intra-luminal grey scale value falls and the perivasculargrey scale value rises when the FITC moleculeleaves the circulation. Macromolecular leakagewas defined as the ratio between the average greyscale value within the venule (Gv) and the averagegrey scale value in the perivenular interstitium(Gi) (18, 19).To evaluate portal hypertension, the portal

pressure was measured in each rat, throughcannulation of an ileocolic vein with a 24-gaugecatheter (Becton Dickinson, Erembodegem-Aalst, Belgium) at the end of the experiment,with zero level at the midpoint of the animal.The presence of cirrhosis was confirmed by

routine histology (haematoxylin & eosin andsirius red staining).

Immunohistochemistry

At the end of the experiments, tissue samples ofthe mesentery were taken from the most distalloop of the small bowel and were immediately

Fig. 1. Intravital microscopy technique. A loop of the distal ileum is exteriorized and spread over a plexiglass plate (A). Themicrovascular density and permeability is studied in such a thin translucent membrane (arrow). Observations are made using waterimmersion objectives (B).

891

Angiogenesis in PHT and cirrhosis

fixed in a 4% phosphate-buffered formaldehydesolution (Sigma) and embedded in paraffin. Fivemicrometer sections were cut for immunohisto-chemistry. Sections were deparaffinized, rehy-drated and incubated in triphosphate-bufferedsaline (TBS) with sodium azide for 20min toblock endogenous peroxidase. Subsequently,they were incubated for 1 h at room temperaturewith primary antibody, goat antimouse CD 31(Santa Cruz Biotechnology, Santa Cruz, CA),mouse antihuman VEGF (Santa Cruz Biotech-nology) (20, 21) and mouse antihuman eNOS(Transduction Laboratories, Lexington, KY)(20, 21), respectively. The specificity of the im-munolabeling was confirmed by incubation with-out primary antibody. Thereafter streptavidin-biotinylated links (LSAB1system-HRP, DakoCytomation, Carpenteria CA, USA) were appliedfor 15min each. 3,30-diaminobenzidine (DAB kit,Dako Cytomation) was used as chromogenicsubstrate to visualize immunolabeling, resultingin a brown precipitate.Microscopic evaluation for CD31, VEGF and

eNOS was carried out blinded by two indepen-dent investigators unaware of the status of theanimals. Vascular density in mesentery was as-sessed by determining the count of CD31-labelledendothelial cells in 10 successive high-magnifica-tion (� 100) fields. The intensity of VEGF andeNOS staining in the mesenteric tissue was semi-quantitative scored as follows: 05no staining,15weak, 25moderate, 35 strong intensity. Theexpressions of VEGF and eNOS are evaluated ineach individual blood vessel on 10 successivefields in the mesentery and scored on the scalefrom 0 to 3. A median value of intensity (range0–3) on individual blood vessels is obtained.

Statistical analysis

Data were given as the mean� standard error ofthe mean (SEM). An unpaired or paired Stu-dent’s t-test or a Mann–Whitney test was used tocompare the different groups as appropriate.Bonferroni’s corrections for multiple compari-sons were done. A P value of o0.05 was con-sidered statistically significant.

Results

Characteristics of laboratory animals

The main characteristics of the different groupsare given in Table 1. Body weights were compar-able between all groups at the time of induction.At the time of the final experiment, the bodyweight was significantly lower in the PPVL and

CBDL group compared with the control andSham-operated groups.

As expected there was a significantly higherportal venous pressure (PVP) and lower meanarterial blood pressure (MAP) in the PPVL andCBDL groups, compared with the control andSham-operated groups. These measurements(PVP and MAP) were performed at the end ofthe intravital microscopy studies.

The presence of cirrhosis in the CBDL modelwas confirmed with a picro-sirius red staining. AllCBDL rats developed cirrhosis, without ascites,at the time of 4 weeks (data not shown).

Intravital microscopy: microvascular density

The mean total vessel length (L/A, cm/cm2)per area was quantified with the image analysissoftware, as described. The microvascular densitywas not different between the control (13.8�1.0 cm/cm2) and Sham-operated (14� 1.1 cm/cm2) rats. Compared with control and Sham-operated rats, the microvascular density wassignificantly increased in PPVL (57.6� 4.1 cm/cm2) (Po0.0001; Po0.0001, respectively) andCBDL (124.9� 5.0 cm/cm2) rats (Po0.0001;Po0.0001, respectively). CBDL rats had also asignificantly higher microvascular density thanPPVL rats (Po0.0001). Irregular arranged, tor-tuous and dense vascular networks were seen inPPVL and CBDL rats, even more pronounced inthe cirrhotic animal model (Fig. 2B and C). Thesealterations were absent in control and Sham-operated rats (Fig. 2A).

Intravital microscopy: microvascular permeability of FITCalbumin

The hyperpermeability for albumin increasedover time in CBDL rats in contrast to control,Sham-operated and PPVL rats where no signifi-cant increase of permeability was observed overtime (Fig. 3). Leakage of albumin was more

Table 1. Main characteristics of animals body weight (at time of

experiment) (g), mean arterial blood pressure (mmHg) and portal

venous pressure (mmHg) are shown (mean � SEM)

Body weight (g)

Mean arterialblood pressure(mmHg)

Portal venouspressure (mmHg)

Control 312 � 9 127 � 5 5 � 1Sham-operated 300 � 9 130 � 10 4.5 � 1PPVL 276 � 7n,w 101 � 14n,w 9 � 1n,wCBDL 274 � 8z,§ 99 � 11z,§ 9 � 1z,§

nPo0.05 PPVL vs Control. wPo0.05 PPVL vs Sham-operated. zPo0.05CBDL vs Control. §Po0.001 CBDL vs Sham-operated. PPVL, partialportal vein ligation; CBDL, common bile duct ligation.

892

Geerts et al.

pronounced in the CBDL rats as compared withthe other groups (Fig. 4).

Immunohistochemical analysis of the mesentery

The analysis of the expression of CD31, VEGFand eNOS are given in Table 2.Expression of CD31 was more pronounced in

the CBDL (65� 5.6) and PPVL rats (40� 3.2)compared with the control (17� 2.3; Po0.0001;Po0.0001, respectively) and sham-operatedgroup (15� 3; Po0.0001; Po0.0001, respec-tively). CBDL rats had a significantly higherexpression of CD31-labelled endothelial cellsthan PPVL rats (Po0.0001).VEGF expression was more pronounced in

PPVL (2.5� 0.4) and CBDL (2.9� 0.3) animalscompared with control (1.5� 0.5; Po0.005,Po0.001, respectively) and sham rats (1.6� 0.5;Po0.005; Po0.001, respectively). Furthermore,the VEGF expression in the CBDL rats wassignificantly higher compared with the PPVLgroup (Po0.05). Representative samples of im-munostaining of VEGF are shown in Fig. 5.

Fig. 2. Intravital microscopy: Microvascular density in the visceral peritoneum. In Sham rats (A), a normal vasculature is presentwhere arterioles divide into a capillary network, that form postcapillary venules. In contrast, in partial portal vein ligation (PPVL) (B)and common bile duct ligation (CBDL) (C) rats, vascular networks are irregular, tortuous and dense arranged. Areas of intensecapillary proliferation (arrows) are even more pronounced in CBDL rats (C).

0 10 20 30 40 50 60 701

2

3Control

Sham

PPVL

CBDL

minutes

Rat

io G

v/G

i

a, c, d, eb, c, d, e

Fig. 3. Microvascular permeability. Leakage of fluorescent iso-thiocyanate-albumin as defined by the ratio of grey scale withinthe vessel (Gv) and the grey scale value within the interstitium(Gi) in control, Sham-operated, partial portal vein ligation(PPVL) and common bile duct ligation (CBDL) rats. Thehyperpermeability for albumin increased over time in CBDLrats in contrast to control, Sham-operated and PPVL rats whereno significant increase in permeability is observed over time. Themicrovascular permeability in CBDL rats becomes significantlyhigher after 30 and 60min compared with control, Sham-operated and PPVL rats.aPo0.001 CBDL at t5 0min vsCBDL at t5 30min, bPo0.001 CBDL at t5 0min vs CBDLat t5 60min, cPo0.05 CBDL vs Control, dPo0.05 CBDL vsSham-operated, ePo0.05 CBDL vs PPVL.

893

Angiogenesis in PHT and cirrhosis

VEGF expression was predominantly present inthe endothelial cell layer but also in smoothmuscle cells of blood vessels.eNOS expression was also more pronounced in

PPVL (2.5� 0.5) and CBDL (2.9� 0.1) animalscompared with control (1.1� 0.3; Po0.005;Po0.001, respectively) and sham rats (1.1� 0.3;Po0.005; Po0.001, respectively). Also, the

eNOS expression in the CBDL rats was signifi-cantly higher compared with the PPVL group(Po0.05). Representative samples of immunos-taining of eNOS are shown in Fig. 6. eNOSexpression was detected predominantly in en-dothelial cells.

Vascular density correlated positively withboth eNOS (Spearman’s r5 0.812, Po0.0001)

Fig. 4. Microvascular permeability: leakage of the fluorescent isothiocyanate-albumin is seen in the common bile duct ligation(CBDL) group as an increased grey level within the interstitium (arrow) and a decreased grey level in the vessel (open arrow) after60min compared with 0min. PPVL, partial portal vein ligation.

Table 2. Immunohistochemical analysis of the mesentery CD31 expression is expressed as the number of positive cells counted in 10 successive

fields (magnification � 100). eNOS and VEGF staining are semi-quantitative scored (grading 0–3) and expressed as the median intensity on

individual blood vessels

Control Sham operated PPVL CBDL

CD31 expression (positive cells/10 fields) 17 � 2 15 � 3 40 � 3n,w 65 � 6z,§,zeNOS staining intensity (median value) 1.1 � 0.3 1.2 � 0.3 2.5 � 0.5n,w 2.9 � 0.1z,§,zVEGF staining intensity (median value) 1.5 � 0.5 1.6 � 0.5 2.5 � 0.4n,w 2.9 � 0.3z,§,z

nPo0.01 PPVL vs Control. wPo0.01 PPVL vs Sham-operated. zPo0.001 CBDL vs Control. §Po0.001 CBDL vs Sham-operated. zPo0.05 CBDL vsPPVL. PPVL, partial portal vein ligation; CBDL, common bile duct ligation; eNOS, endothelial nitric oxide synthase; VEGF, vascular endothelial growthfactor.

894

Geerts et al.

Fig. 5. Immunostaining for vascular endothelial growth factor (VEGF) of the mesentery (magnification � 20). A weak signal ofVEGF is observed in the Sham rats (A) (1.6 � 0.5). The signal intensity of VEGF was more pronounced in the partial portal veinligation (PPVL) (B) (2.5 � 0.4) and common bile duct ligation (CBDL) (C) (2.9 � 0.3) rats. Immunostaining for VEGF expressionwas predominantly present in the endothelial cell layer but also in smooth muscle cells of blood vessels (arrows).

Fig. 6. Immunostaining for endothelial nitric oxide synthase (eNOS) of the mesentery (magnification � 20). A faint signal for eNOSwas located in the endothelium that lined mesenteric blood vessels of Sham rats (A) (1.2 � 0.3). The signal intensity of eNOS on theendothelial layer of bloodvessels was increased in partial portal vein ligation (PPVL) (B) (2.5 � 0.5) and common bile duct ligation(CBDL) (C) (2.9 � 0.1) rats (open arrow). The immunostaining of eNOS showed also the density of stained capillaries and focal areasof capillary proliferation (arrows) were observed in the PPVL and CBDL group.

895

Angiogenesis in PHT and cirrhosis

and VEGF expression (Spearman’s r5 0.824,Po0.0001). Furthermore, there was a positivecorrelation between eNOS and VEGF expression(Spearman’s r5 0.826, Po0.0001).

Discussion

Our intravital microscopy study of the mesenteryshows in vivo an increased microvascular densityin rats with portal hypertension without cirrhosis(PPVL). These results are in agreement withpreviously published data using other visualiza-tion techniques (3, 4). However, the present studyshows for the first time in vivo evidence for anincreased angiogenesis in the mesenteric micro-vasculature of an experimental model of portalhypertension with cirrhosis (CBDL). Moreover,the mesenteric vascular density is significantlyhigher in CBDL rats compared with PPVL rats.Focal areas of intense vascular proliferation andirregular arranged, tortuous vascular networksare observed in the mesentery of cirrhotic ani-mals. We exclude an influence of surgical manip-ulation by comparing control and Sham-operatedrats. Endothelial cell proliferation is demon-strated by immunostaining for CD31 (also knownas PECAM-1). CD31 is a highly abundant cellsurface protein constitutively expressed on almostall endothelia, and its expression has been corre-lated with the extent of angiogenesis. An impress-ive higher number of CD31 positive cells are seenin the mesentery of CBDL and PPVL rats com-pared with control and Sham rats. These resultsdemonstrate clearly that the increased microvas-cular density is at least partially, a result ofangiogenesis.The VEGF is considered as a major primary

stimulus of angiogenesis in pathologic conditions(8). VEGF is a secreted protein that is a specificgrowth factor for endothelial cells and it has beenshown to increase vascular density and vascularpermeability in in vivo and in vitro assays. Anincreased VEGF expression is previously shownin vitro in the mesentery of portal hypertensivemice (6, 7). These results are confirmed in ourstudy by an increased expression of VEGF in themesentery of PPVL rats, which is significantlyhigher compared with the control groups. Recentstudies by Fernandez et al. (6, 7) showed adecrease of the intestinal neovascularization,splanchnic blood flow and formation of porto-systemic collaterals in portal hypertensive rats bythe administration of anti-VEGF receptor-2monoclonal antibodies. These data suggest thatVEGF-dependent angiogenesis plays a role in themaintenance of the increased portal inflow andformation of collaterals in portal hypertension.

For the first time, the present study shows thatthe mesenteric VEGF expression is also increasedin cirrhotic rats and this expression is even highercompared with PPVL rats. These results providefurther evidence for an important role of VEGFin the development of mesenteric angiogenesis inPHT with cirrhosis. More research is necessary toinvestigate whether angiogenesis in PHT withcirrhosis represents a beneficial mechanism tocontrol the portal venous pressure or exerts anadditional pathogenetic role in maintaining in-creased splanchnic blood flow.

Several lines of evidence suggest that NO mayalso regulate angiogenesis and vascular perme-ability (11, 12). A central role of NO in thedevelopment of the hyperdynamic circulationhas been suggested (1). However, studies showedthat after NO inhibition (22) or in eNOS or iNOSportal hypertensive knock out mice (23), thehyperdynamic circulation persisted. Recently,also an eNOS knockout mice induced by bileduct ligation developed fibrosis and a hyperdy-namic circulation associated with portal hyper-tension (24). This does not exclude a role foreNOS in the maintenance and regulation ofportal hypertension since a large body of evidencesuggests that eNOS is involved in portal hyper-tension. Our study shows for the first time anincreased expression of eNOS in the mesentery ofPPVL and CBDL rats compared with the controlgroups. An even higher expression is seen in theCBDL compared with the PPVL group. Theseresults are in agreement with previous data inabdominal aortic endothelium where the eNOSproduction is also higher in cirrhotic rats than inportal vein-stenosed rats (25, 26). This observeddifference in eNOS production may contribute tothe more profound vasodilation and sodiumretention of cirrhotic rats vs PPVL rats.

Several studies have pointed to the critical roleof NO in VEGF-induced angiogenesis as well asvascular permeability and showed that eNOSacted as a downstream mediator for VEGF (11–13). Angiogenesis, vessel diameter, blood flowrate and vascular permeability were proportionalto NO levels and were most impaired in eNOSknockout mice (27, 28). Sieber et al. (4) showedthat the increased mesenteric angiogenesis couldbe prevented by chronically inhibiting NO for-mation in portal hypertensive rats. These data,together with our findings, support an importantrole of NO as a mediator of angiogenesis in portalhypertension and cirrhosis.

In the second part of our study we provide forthe first time in vivo evidence of an increasedmacromolecular leakage in the mesenteric venulesof CBDL rats compared with Sham and PPVL

896

Geerts et al.

rats. This increased permeability in the CBDLrats may be related to a higher VEGF and eNOSexpression in the mesenteric microvasculaturecompared with Sham and PPVL rats. Theseresults support the hypothesis that indeed anincreased eNOS production, and even an in-creased VEGF production as shown in the pre-sent study, may be responsible for the moremarkedly systemic haemodynamic changes seenin patients or animals with cirrhosis than in thosewithout liver disease. Although our cirrhotic ratshave no ascites, we may hypothesize that thisobservation of an increased microvascular pro-tein leakage may contribute additionally to theformation of ascites in cirrhosis. The developedtechnique of intravital microscopy makes furtherinvestigation of this proposed mechanism possi-ble in cirrhotic models with ascites.In conclusion, our study provides for the first

time in vivo evidence of an increased angiogenesisin the intact mesenteric microvasculature in anexperimental animal model of portal hyperten-sion and cirrhosis. Portal hypertension alone isable to induce an increased angiogenesis in thesplanchnic microvasculature but additional pre-sence of cirrhosis increases this phenomenon.Furthermore, the present study shows for thefirst time in vivo the presence of an increasedmacromolecular leakage in the mesenteric venulesof rats with cirrhosis but not in rats with portalhypertension alone. An increased mesentericVEGF and eNOS expression suggests their im-portant role in the mesenteric angiogenesis andvascular permeability in PHT and cirrhosis.

Acknowledgements

The authors thank Julien Dupont, Tommy Dheuvaert and NeleNica for their expert technical assistance.Financial support and grants: The work is supported by a grantfrom Astra Zeneca Belgium, Roche Belgium, Schering-PloughBelgium and UCB Belgium.Isabelle Colle is supported by a grant from the Fund forScientific Research-Flanders (1.5.083.03). Anja Geerts receiveda grant form the Flemish Society of Gastroenterology.

Reference

1. Vallance P, Moncada S. Hyperdynamic circulation incirrhosis: a role for nitric oxide? Lancet 1991; 337: 776–8.

2. Perez-Ruiz M, Ros J, Morales-Ruiz M, et al. Vascularendothelial growth factor production in peritoneal macro-phages of cirrhotic patients: regulation by cytokines andbacterial lipopolysaccharide. Hepatology 1999; 29: 1057–63.

3. Sumanovski LT, Battegay E, Stumm M, van der Kooij

M, Sieber C C. Increased angiogenesis in portal hyperten-sive rats: role of nitric oxide. Hepatology 1999; 29: 1044–9.

4. Sieber C C, Sumanovski LT, Stumm M, van der Kooij

M, Battegay E. In vivo angiogenesis in normal and portal

hypertensive rats: role of basic fibroblast growth factor andnitric oxide. J Hepatol 2001; 34: 644–50.

5. Cejudo-Martin P, Ros J, Navasa M, et al. Increasedproduction of vascular endothelial growth factor in perito-neal macrophages of cirrhotic patients with spontaneousbacterial peritonitis. Hepatology 2001; 34: 487–93.

6. Fernandez M, Vizzutti F, Garcia-Pagan J C, Rodes J,Bosch J. Anti-VEGF receptor-2 monoclonal antibody pre-vents portal-systemic collateral vessel formation in portalhypertensive mice. Gastroenterology 2004; 126: 886–94.

7. Fernandez M, Mejias M, Angermayr B, Garcia-Pa-

gan J C, Rodes J, Bosch J. Inhibition of VEGF receptor-2 decreases the development of hyperdynamic splanchniccirculation and portal-systemic collateral vessels in portalhypertensive rats. J Hepatol 2005; 43: 98–103.

8. Carmeliet P, Jain RK. Angiogenesis in cancer and otherdiseases. Nature 2000; 407: 249–57.

9. Rosmorduc O, Wendum D, Corpechot C, et al. Hepa-tocellular hypoxia-induced vascular endothelial growth fac-tor expression and angiogenesis in experimental biliarycirrhosis. Am J Pathol 1999; 155: 1065–73.

10. Corpechot C, Barbu V, Wendum D, et al. Hypoxia-induced VEGF and collagen I expressions are associatedwith angiogenesis and fibrogenesis in experimental cirrhosis.Hepatology 2002; 35: 1010–21.

11. Fukumura D, Gohongi T, Kadambi A, et al. Predomi-nant role of endothelial nitric oxide in vascular endothelialgrowth factor-induced angiogenesis and vascular perme-ability. Proc Natl Acad Sci USA 2001; 98: 2604–9.

12. Ziche M, Morbidelli L, Masini E, et al. Nitric oxidemediates angiogenesis in vivo and endothelial cell growthand migration in vitro promoted by susbstance P. J ClinInvest 1994; 94: 2036–44.

13. Ziche M, Morbidelli L, Choudhuri R, Zhang H,Donnini S, Granger H J. Nitric oxide synthase liesdownstream from vascular endothelial growth factor-in-duced but not basic fibroblast growth factor-induced angio-genesis. J Clin Invest 1997; 99: 2625–34.

14. Ohara N, Voelkel N F Chang S W. Tissue eicosanoidsand vascular permeability in rats with chronic biliaryobstruction. Hepatology 1993; 18: 111–8.

15. Wirth K J, Bickel M, Hropot M, et al. The bradykininB2 receptor antagonist Icatibant (HOE 140) corrects avidNa1 retention in rats with CCl4-induced liver cirrhosis:possible role of enhanced microvascular leakage. Eur JPharmacol 1997; 337: 45–53.

16. Colle I, De Vriese A,Van Vlierberghe H, Lameire N,De Vos M. Vascular hyporesponsiveness in the mesentericartery of rats with cirrhosis and portal hypertension: an invivo study. Eur J Gastroenterol Hepatol 2004; 16: 139–45.

17. Colle I, De Vriese A,Van Vlierberghe H, Lameire N,De Vos M Systemic and splanchnic haemodynamic effectsof sildenafil in an in vivo animal model of cirrhosis, supportfor a risk in cirrhotic patients. Liver Int 2004; 24: 63–8.

18. De Vriese A S Lameire N H. Intravital microscopy: anintegrated evaluation of peritoneal function and structure.Nephrol Dial Transplant 2001; 16: 657–60.

19. DeVriese A S, Tilton RG, Stephan C C Lameire N H.

Vascular endothelial growth factor is essential for hyper-glycemia-induced structural and functional alterations ofthe peritoneal membrane. J Am Soc Nephrol 2001; 12:1734–41.

20. Mortier S, Faict D,Lameire NH DeVriese AS. Benefitsof switching from a conventional to a low-GDP bicarbo-nate/lactate-buffered dialysis solution in a rat model. Kid-ney Int 2005; 67: 1559–65.

21. Mortier S, Faict D, Schalkwijk C G, Lameire N H De

Vriese A S. Long-term exposure to new peritoneal dialysis

897

Angiogenesis in PHT and cirrhosis

solutions: effects on the peritoneal membrane. Kidney Int2004; 66: 1257–65.

22. Garcia-Pagan J C, Fernandez M, Bernadich C, et al.Effect of continued NO inhibition on portal hypertensivesyndrome after portal vein stenosis in rat. Am J PhysiolGastrointest Liver Physiol 1994; 267: G984–90.

23. Theodorakis N G, Wang Y, Skill N J, et al. The role ofnitric oxide synthase isoforms in extrahepatic portal hyper-tension: studies in gene-knockout mice. Gastroenterology2003; 124: 1500–8.

24. Koshy A,Gottardi A D Ledermann M, et al. Endothe-lial nitric oxide synthase is not essential for the developmentof fibrosis and portal hypertension in bile duct ligated mice.Liver Int 2005; 25: 1044–52.

25. Niederberger M, Gines P, Martin P Y, et al. Compar-ison of vascular nitric oxide production and systemichemodynamics in cirrhosis versus prehepatic portal hyper-tension in rats. Hepatology 1996; 24: 947–51.

26. Tazi K A, Barriere E, Moreau R, et al. Role of shearstress in aortic eNOS up-regulation in rats with biliarycirrhosis. Gastroenterology 2002; 122: 1869–77.

27. Lee P C, Salapongse A N, Bragdon G A, Shears L L,Watkins S C Edington HD. Impaired wound healing andangiogenesis in eNOS-deficient mice. Am J Physiol 1999;277: H1600–8.

28. Zhao L X, Lu X, Feng O. Deficiency in endothelial nitricoxide synthase impairs myocardial angiogenesis. Am JPhysiol Heart Circ Physiol 2002; 283: H2371–8.

898

Geerts et al.