Inhibition of VEGF-A prevents the angiogenic switchand results in increased survival of Apc�/min miceNina Korsisaari, Ian M. Kasman, William F. Forrest, Navneet Pal, Wei Bai*, Germaine Fuh, Franklin V. Peale,Ron Smits†, and Napoleone Ferrara‡

Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080

Contributed by Napoleone Ferrara, May 6, 2007 (sent for review February 9, 2007)

Anti-VEGF-A monoclonal antibodies, in combination with chemo-therapy, result in a survival benefit in patients with metastaticcolorectal and non-small cell lung cancer, but little is knownregarding the impact of anti-VEGF-A therapy on benign or prema-lignant tumors. The Apc�/min mice have been widely used as amodel recapitulating early intestinal adenoma formation. To in-vestigate whether tumor growth in Apc�/min mice is mediated byVEGF-A-dependent angiogenesis, we used two independent ap-proaches to inhibit VEGF-A: monotherapy with a monoclonalantibody (Mab) targeting VEGF-A and genetic deletion of VEGF-Aselectively in intestinal epithelial cells. Short-term (3 or 6 weeks)treatment with anti-VEGF-A Mab G6–31 resulted in a nearly com-plete suppression of adenoma growth throughout the small intes-tine. Growth inhibition by Mab G6–31 was associated with adecrease in vascular density. Long-term (up to 52 weeks) treatmentwith Mab G6–31 led to a substantial increase in median survival.Deletion of VEGF-A in intestinal epithelial cells of Apc�/min miceyielded a significant inhibition of tumor growth, albeit of lessermagnitude than that resulting from Mab G6–31 administration.These results establish that inhibition of VEGF-A signaling issufficient for tumor growth cessation and confers a long-termsurvival benefit in an intestinal adenoma model. Therefore, VEGF-Ainhibition may be a previously uncharacterized strategy for theprevention of the angiogenic switch and growth in intestinaladenomas.

adenoma � angiogenesis � gastrointestinal � Wnt-pathway � polyp

Angiogenesis is essential for many physiological processes (1).Several pathological conditions, particularly tumor growth andmetastasis, also depend on angiogenesis (2). One of the key positiveregulators of angiogenesis is vascular endothelial growth factor(VEGF)-A (reviewed in ref. 3). VEGF-A is part of a gene familythat includes VEGF-B, VEGF-C, VEGF-D, and PlGF (4).VEGF-A primarily binds two high affinity receptor tyrosine kinases(RTKs), VEGFR-1 (Flt-1) and VEGFR-2 (Flk-1/KDR), the latterbeing the major mediator of mitogenic and angiogenic signals ofVEGF-A. Additionally, neuropilin-1 is a coreceptor for heparin-binding VEGF-A isoforms (reviewed in ref. 4).

Early studies showed that an anti-VEGF-A mouse monoclonalantibody (Mab) called A.4.6.1 inhibits the growth of human tumorcell lines transplanted in nude mice (5). Interestingly, the same Mabinhibited tumor angiogenesis in a model of liver metastasis ofcolorectal cancer (6). Furthermore, several other VEGF inhibitors,including soluble VEGF receptors (7, 8), anti-VEGFR-2 Mabs (9),and small molecule VEGF RTK inhibitors (10), have been shownto inhibit tumor growth.

VEGF-A inhibition with a humanized variant of Mab A.4.6.1(bevacizumab), in combination with chemotherapy, results in asurvival advantage in patients with previously untreated metastaticcolorectal cancer (11) and nonsquamous non-small-cell lung car-cinoma (12) relative to chemotherapy alone. The small moleculeVEGF RTK inhibitors sunitinib (13) and sorafenib (14) have shownefficacy in metastatic renal cell cancer patients and were recentlyapproved by the Food And Drug Administration.

Investigating the mechanisms of tumor angiogenesis in xeno-grafts has limitations, because these models do not recapitulatetumor development in a natural setting. Furthermore, little isknown regarding the role of angiogenesis in the growth andprogression of benign or premalignant tumors.

The syndrome of Familial Adenomatous Polyposis (FAP) andthe majority of sporadic colorectal cancers are caused by mutationsin the APC gene (reviewed in ref. 15). FAP patients develophundreds to thousands of adenomatous polyps in their lowergastrointestinal tract, in addition to extracolonic tumors. APC hasbeen reported to be involved in numerous cellular processesincluding proliferation, apoptosis, signal transduction, and cellmigration, but its best studied function is the regulation of �-cateninin the Wnt signaling pathway (15).

Apc�/min mice with a heterozygous truncation allele at codon 850mimic some features of the polyposis of FAP patients with germ-line APC mutation (16, 17). The onset of tumor formation inApc�/min mice is in early adulthood, and the animals typicallydevelop 60–150 intestinal polyps in a C57BL/6 genetic background.Tumor development results in a severely compromised longevity ofthe mice, usually resulting in death from anemia and/or hypopro-teinemia (16) at the age of �5 months. Whereas humans with FAPtypically develop colonic adenomas, the vast majority of polyps inthe Apc�/min mice develop in the small intestine. These polyps reacha size of 1–2 mm in diameter, and larger polyps (up to 4 mm indiameter) arise at a lower frequency.

We sought to determine the role of VEGF-A-dependent angio-genesis in the Apc�/min model. Tumors were analyzed after short-and long-term treatment with anti-VEGF-A Mab or after geneticdeletion of VEGF-A by Cre-LoxP technology in intestinal epithelialcells.

ResultsExpression of VEGF-A in the Apc�/min Intestinal Adenomas. To inves-tigate the expression pattern of VEGF-A in intestinal tumors ofApc�/min mouse, we performed in situ hybridization [ISH, seesupporting information (SI) Text]. VEGF-A expression was ob-served in the epithelial cells with varying intensity compared with

Author contributions: N.K. and N.F. designed research; N.K., I.M.K., N.P., W.B., F.V.P., andR.S. performed research; G.F. and R.S. contributed new reagents/analytic tools; N.K., I.M.K.,W.F.F., and F.V.P. analyzed data; and N.K. and N.F. wrote the paper.

Conflict of interest statement: The authors are employees and shareholders ofGenentech, Inc.

Freely available online through the PNAS open access option.

Abbreviations: EMH, extramedullary hematopoiesis; FAP, Familial Adenomatous Polyposis;RTK, receptor tyrosine kinase.

*Present address: Department of Medicine, New York Downtown Hospital, New York,NY 10038.

†Present address: Department of Experimental Pathology, Josephine Nefkens Institute,Erasmus University Medical Center, Dr. Molewaterplein 50, P.O. Box 2040, 3000 CA,Rotterdam, The Netherlands.

‡To whom correspondence should be addressed. E-mail: nf@gene.com.

This article contains supporting information online at www.pnas.org/cgi/content/full/0704213104/DC1.

© 2007 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0704213104 PNAS � June 19, 2007 � vol. 104 � no. 25 � 10625–10630

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

June

18,

202

1

http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1

normal intestinal villus epithelium, whereas a focally prominentsignal was observed in stromal cells of the adenomas and the stromaof the normal villi (Fig. 1 A–F).

VEGF-A in situ hybridization signal in Apc�/min intestinal ade-nomas was variable but was focally greater than in the surroundingnormal epithelium in all animals examined. In three animals withmeasured small intestinal adenomas (n � 12), the average adenomasignal intensity was 40–60% above the surrounding normal epi-thelium [range of individual adenoma signals: 0.9–3.0 times controllevels; Student’s t test P values (per animal) � 0.02–0.03]. In thesingle large intestinal adenoma examined, the average epithelialsignal intensity was 2.6 times that of the adjacent normal epithelium(Student’s t test P value � 0.002).

Inhibition of VEGF-A Lowers Tumor Burden of Apc�/min Mice. Wesought to determine whether anti-VEGF-A Mab therapy would beeffective at lowering the tumor burden in Apc�/min mice. We choseMab G6–31 because of its ability to potently block VEGF-A acrossspecies (18). This is unlike the well characterized anti-VEGF MabA.4.6.1, which inhibits human but not mouse VEGF-A (7, 18). Toassess the short-term effects of Mab G6–31 on tumor burden,treatment of 10 mice per cohort was started at 13 weeks of age andcontinued for 3 or 6 weeks. To determine the tumor phenotype atthe age of treatment onset, an untreated control group (n � 12) wasanalyzed at 13 weeks of age (day 0).

Treatment with anti-VEGF-A Mab for either 3 or 6 weekssignificantly reduced overall tumor burden in Apc�/min mice. At day0, the mean tumor burden of Apc�/min mice was 39.3 mm3 (rangingfrom 12.3 mm3 to 97.0 mm3) (Fig. 2A). The mean tumor burden ofmice treated with control IgG for 3 weeks was 96.8 mm3 (47.1–299.9mm3), whereas the mean tumor burden of mice treated for 3 weeks

with Mab G6–31 was 23.5 mm3 (4.5–58.2 mm3). This was astatistically significant 76%, or 4-fold reduction in mean tumorburden, with a P � 0.008. After 6 weeks of administration of controlIgG, the tumor burden reached a mean of 198.6 mm3 (40.5–315.7mm3), whereas the tumor burden in mice treated with Mab G6–31remained at 28.4 mm3 (3.2–75.9 mm3), exhibiting a significant 86%,or 7-fold reduction in mean tumor burden, with a P � 5.3 � 10�5(Fig. 2A).

After 3 weeks of treatment with control IgG or Mab G6.31, themean tumor numbers were respectively 116 � 9 (� SEM) and107 � 11 (P � 0.28). After 6 weeks, the mean tumor number was120 � 11 in the control IgG group and 100 � 10 (P � 0.09) in theMab G6–31 group. At day 0, mice had an average of 100 � 9tumors. Thus, the decrease in tumor burden after either 3 or 6weeks of anti-VEGF-A treatment was due to a decreased adenomasize, rather than to a decrease in the number of adenomas.

There was no evidence of adenoma growth escape duringanti-VEGF-A treatment of 3 or 6 weeks. Tumors in mice treatedwith Mab G6–31 had a more compact size distribution (Fig. 2B,middle and bottom graph) compared with the broader size distri-bution of tumors from mice treated with control IgG (Fig. 2B,graphs 2 and 4 from the top). The mean polyp diameter in micetreated for 3 weeks with control IgG was 1.28 mm; in the MabG6–31 group it was 0.85 mm (P � 0.0001). After 6 weeks oftreatment, the mean polyp diameters were 1.64 mm in the controlIgG group and 0.86 mm in the Mab G6–31 group (P � 0.0001).Mean tumor diameter at day 0 was 0.97 mm.

Interestingly, anti-VEGF-A treatment appeared to inhibit thegrowth of tumors of all sizes. After a 3-week treatment with MabG6–31, the frequency of small tumors, 0.3–1.0 mm in diameter (for6-week treatment 0.3–1.2 mm) was greater than in the control

A B C D

I

E F G H

L M N

J K

Fig. 1. VEGF-A expression in Apc�/min adenomas and normal villus. In situ hybridization with VEGF-A probe on an intestinal adenoma from large (A, E, I, andL) and small (B, F, J, and M) bowel as well as adjacent normal colon (C and G) and small intestine (D, H, K, and N) of 14-week-old Apc�/min mice demonstratesincreased expression of VEGF-A in adenomas relative to normal intestine. Representative images are shown. (A–D and I–K) Brightfield. (E–H and L–N) Darkfield.VEGF signal in adenomas (E and F) is focally stronger than in adjacent normal intestinal epithelium (G and H). VEGF signal arises from both epithelial cells (I–K,arrows) and stromal cells (I–K, arrowheads). (Scale bars: A–H, 100 �m; I–N, 25 �m.)

10626 � www.pnas.org�cgi�doi�10.1073�pnas.0704213104 Korsisaari et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

1

treated group, whereas the frequency of tumors with a diameter�1.0 mm (for 6 weeks �1.2 mm) was decreased (Fig. 2C Top andMiddle). A comparison to the tumor size distribution at day 0 (Fig.2C Bottom) suggested that the growth of the adenomas had virtuallyarrested after Mab G6–31 administration.

Mab G6–31 was effective at suppressing adenoma growth in allsmall-intestinal areas as a significantly lower mean tumor diameterwas observed after 3 or 6 weeks of therapy (Fig. 2D). Furthermore,the mean adenoma diameter in the first intestinal quarter of micetreated with Mab G6–31 was significantly smaller compared withthat observed in mice at day 0 (double asterisk in Fig. 2D). Thereduction in mean tumor diameter of the colonic adenomas did notreach statistical significance (Fig. 2D). The mean diameter of thelarge bowel polyps in mice treated with Mab G6–31 for 3 weeks was1.3 � 0.3 mm (� SEM), whereas the mean diameter in the controlIgG-treated mice was 2.5 � 0.4 mm, with a P � 0.064. The meandiameter of large bowel tumors after 6 weeks of treatment with Mab

G6–31 was 2.2 � 0.3 mm and 2.6 � 0.3 mm after administrationwith control IgG, with a P � 0.37.

Deletion of VEGF-A in Intestinal Epithelial Cells Reduces Mean TumorDiameter. We next sought to dissect the contribution of VEGF-Aoriginating from intestinal epithelial sources to adenoma develop-ment. To this end, tumor diameter and number were assessed in13-week-old Apc�/min mice that were crossed to mice in whichVEGF-A was conditionally deleted in intestinal epithelial cells withCre/loxP technology (VEGFlox;Villin-Cre mice).

The expression of Villin, an actin-binding protein and a majorstructural component of the brush border of specialized absorptivecells, begins during embryogenesis in the intestinal hindgutendoderm and later extends throughout the small- and large-intestinal endoderm (19, 20). In the adult, Villin distributionbecomes diffuse with moderate apical polarization in cells of thecrypts and polarization in brush borders of fully differentiated cells

Fig. 2. Inhibition of VEGF-A lowers tumor burden. (A) Tumor burden is indicated by bars from the largest to the smallest value of tumor burden of individualmice in the group. Black squares indicate group averages. *, P � 0.008; **, P � 5.3 � 10�5. n, number of animals. (B) Distribution of tumors by diameter and shownas percent of the total number of tumors. n, number of tumors in a group. (C) Overlaid tumor size frequencies after 3 weeks of treatment (Top), after 6 weeksof treatment (Middle), and in comparison to day 0 (Bottom). Vertical bars illustrate the size smaller or equal of which tumor frequency is greater in MabG6–31-treated animals: 1 mm in 3-week-treatment and 1.2 mm in 6-week-treatment group. (D) Mean tumor diameter plotted against the intestinal location.n, number of tumors per group in the first, second, third, and fourth intestinal quarter, respectively. Day 0 group contained 12 animals; other groups 10. S,stomach; C, caecum; R, rectum. Bars represent SEM. *, P � 1.0 � 10�10; **, P � 0.002 compared with Mab G6–31 3 or 6 weeks. (E) Mean tumor diameter of 14Apc�/min mice;Villin-Cre (black columns) and Apc�/min VEGFlox;Villin-Cre (gray columns) mice presented in a descending order. Bars represent SD.

Korsisaari et al. PNAS � June 19, 2007 � vol. 104 � no. 25 � 10627

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

June

18,

202

1

lining the villi of the small intestine (21). The expression of Crerecombinase driven by Villin promoter (Villin-Cre) has been shownto recapitulate the expression pattern of the Villin gene in every cellof the intestinal epithelium (22).

The mean tumor diameter of control Apc�/min Villin-Cre micewas 1.02 � 0.3 mm (�SEM), whereas the mean tumor diameter ofApc�/min VEGFlox;Villin-Cre mice was 0.82 � 0.3 mm (Fig. 2E),demonstrating a 19.8% reduction (P � 0.001). Tumor number wasnot significantly different between the two groups. WhereasApc�/min Villin-Cre mice had 137 � 11 intestinal adenomas,Apc�/min VEGFlox;Villin-Cre mice had 150 � 17 adenomas(P � 0.27).

These data indicate that deletion of VEGF-A from all intestinalepithelial cells from duodenum through colon, and crypt to villustip results in a significant inhibition of tumor growth, albeit of areduced degree compared with that resulting from systemic ad-ministration of anti-VEGF-A antibody. These data suggest thatextraepithelial sources of VEGF-A contribute to the growth ofintestinal adenomas of Apc�/min mice.

Inhibition of VEGF-A Extends the Median Survival of Apc�/min Mice.Given the effectiveness of anti-VEGF-A treatment in tumor growthinhibition, we wished to investigate whether treatment with MabG6–31 could yield a long-term survival benefit for Apc�/min mice.To this end, administration with Mab G6–31 or control IgG wascontinued for up to 52 weeks or until the mice were observed to bemoribund. The median survival in the control IgG group was 24.0weeks. In the Mab G6–31 group it was 33.6 weeks with log-rank P �2.4 � 10�3 (Fig. 3).

Normal Serum Total Protein, Albumin, and Triglycerides Level andReduced Splenic Extramedullary Hematopoiesis in Apc�/min Micetreated with anti-VEGF-A. Apc�/min mice treated with Mab G6–31appeared considerably more alert and responsive than those treatedwith control IgG. Moreover, pale paws, suggestive of the progres-sive anemia as initially reported by Moser et al. (16), were observedfrequently in animals treated with control IgG, but not in animalstreated with Mab G6–31. Consistent with this observation, themean total serum protein and serum albumin of Apc�/min miceadministered with control IgG was decreased, whereas total proteinand albumin levels were within normal range in mice treated withMab G6–31 (Table 1). As previously reported for Apc�/min mice(16) and consistent with hypoproteinemia, mean triglyceride levelwas elevated in animals treated with control IgG, although it waslowered to a level comparable to a reference value upon treatmentwith Mab G6–31 (Table 1). Although there were no treatment-

related differences in body masses after 3 or 6 weeks of treatment,the mean spleen masses were significantly (P � 2.3 � 10�3)increased in mice treated with control IgG. After 3 weeks withcontrol IgG, the mice had a mean spleen mass of 0.26 g, or 1.17%of body mass, whereas the mean spleen mass was 0.11 g (0.49% ofbody mass) in mice treated with Mab G6–31 for the same duration.The increase in mean spleen mass in mice treated with control IgGis consistent with extramedullary hematopoiesis (EMH), possiblysecondary to intestinal bleeding. This was confirmed by histologicexamination of the spleens (data not shown). Ten of 10 mice treatedfor 6 weeks with control IgG showed marked EMH, whereas 2 micetreated with Mab G6–31 had moderate EMH, 5 had mild EMH,and 3 had no diagnostic changes in their spleens.

The lower degree of EMH in the spleens of mice treated withMab G6–31 short-term suggests that anti-VEGF-A therapy reducesintestinal bleeding.

Kidney Changes After Long-Term Treatment with Mab G6–31. Toinvestigate potential toxicities related to administering high-affinityanti-VEGF-A Mab G6–31, pancreas, liver, and kidney were ana-lyzed histologically after short- (3–6 weeks) and long-term (18–53weeks) treatment. No significant toxicity was noted in animalstreated for 3–6 weeks. After long-term treatment with Mab G6–31,mice showed variable (mild to severe) diffuse global glomerulo-sclerosis and moderate stromal edema of the pancreas (reflectinghypoproteinemia). These observations are consistent with previ-ously observed toxicity resulting from long-term administration ofMab G6–31 (23). Importantly, the adverse effects were outweighedby the overall improvement of health reflected by the increasedmedian survival. Four of five mice treated with Mab G6–31 for18–53 weeks were diagnosed with mild to extensive EMH in thespleen.

Altered Tumor Morphology upon Mab G6–31 Treatment Was NotAccompanied by a Change in Proliferative Index. To further charac-terize intestinal polyps in Apc�/min mice, macroscopic and histologicanalyses were performed (see SI Text). The gross morphology ofpolyps treated with Mab G6–31 differed noticeably from that of thepolyps treated with control IgG (SI Fig. 5 A and B). Whereas tumorsfrom control mice typically had a relatively unbroken, smoothsurface, tumors from animals treated with Mab G6–31 appearedwith deep invaginations on their surface. Histologic analysis con-firmed that tumors from Mab G6–31 and control IgG-treated miceare tubular adenomas (SI Fig. 5 C–F). Adenomas from controlIgG-treated mice had marked intravillous epithelial proliferation,with vertical and lateral expansion, and were typically widened�2-fold from their base to luminal surface. There was minimalfibrous stroma. Adenomas from mice treated with Mab G6–31characteristically had fewer intravillous epithelial cells, were lessbroad at the luminal surface, shallower, and involved fewer adjacentvilli. Histologic analysis of the colonic polyps in both treatmentgroups showed pendunculated tubular adenomas with abundantfibrovascular stroma and a variable amount (up to 100%) ofdysplastic epithelium (data not shown).

Fig. 3. Extended median survival in mice treated with Mab G6–31. Kaplan–Meier of Mab G6–31 (gray line) or control IgG (black line) -treated mice isshown. Open arrow designates the duration of the treatments. Mediansurvival is indicated with gray arrows. *, P � 2.4 � 10�3. n, number of mice ina group.

Table 1. Serum chemistry

Group (n)Total protein,

g/dl*Albumin,

g/dl†Triglycerides,

mg/dl‡

Control IgG 3 weeks (10) 3.7 � 0.2 1.9 � 0.1 268.9 � 82.5G6–31 3 weeks (10) 4.9 � 0.2 2.6 � 0.1 75.5 � 4.9control IgG 6 weeks (10) 3.0 � 0.3 1.6 � 0.2 591.1 � 81.3G6–31 6 weeks (10) 4.9 � 0.1 2.7 � 0.1 71.1 � 4.3

Data are � SEM.*Reference value 3.9–5.5 g/dl.†Reference value 2.3–3.2 g/dl.‡Reference value 35–244 mg/dl.

10628 � www.pnas.org�cgi�doi�10.1073�pnas.0704213104 Korsisaari et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

1

http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1

To assess the extent of proliferation in the tumor tissue and innormal mucosa, an indirect immunohistochemical staining withKi-67 antibody was performed (SI Fig. 5 G–J). Quantitative analysisrevealed similar amounts of Ki-67-positive cells in tumors frommice treated with either control IgG or with Mab G6–31. Likewise,the proliferative index of the normal adjacent mucosa was compa-rable between both treatments (SI Fig. 5K).

Reduced Vascular Density in Mab G6–31 Treated Tumors. Given thatVEGF-A is known to be a mitogen for vascular endothelial cellsthrough VEGFR-2 signaling, we examined the tumor vascularnetworks in mice treated with Mab G6–31 and control IgG byimmunohistochemical staining of thick tissue sections with anti-bodies for three independent vascular markers, CD31, CD105, andvon Willebrand Factor (vWF) (Fig. 4 A–H) (see SI Text). Quanti-fication of the vessel density indicated that the vascular componentof tumors from Mab G6–31-treated mice was significantly reducedcompared with that seen in control IgG-treated mice, both at the3 and 6 weeks time points (Fig. 4I). A similar reduction wasobtained with all three markers used.

DiscussionWe used Apc�/min mice to investigate the role of VEGF-A in benignintestinal tumorigenesis. Our in situ analysis documented an up-regulation of VEGF-A in adenomas compared with the normalvilli. Our experiments were designed to measure the effects ofshort- and long-term anti-VEGF-A treatment on established in-testinal adenomas undergoing robust growth. We show that treat-ment with anti-VEGF-A Mab G6–31 significantly lowers the tumorburden and extends the survival of the Apc�/min mice.

Several studies have been conducted on the effect of dietary andchemopreventive agents on tumor burden of Apc�/min mice (re-viewed in ref. 24), of which an updated list exists at http://corpet.net/min. Many of these studies report a significant decrease in tumornumber. Nonsteroidal antiinflammatory drugs such as piroxicamand sulindac, which target both COX-1 and COX-2, have beenamong the most potent agents in suppressing tumor formation inApc�/min mice (25–28). More specifically, the important role ofCOX-2 in intestinal polyposis was demonstrated by using differentselective COX-2 inhibitors (29–31). Recently, Goodlad et al. (32)reported that short-term administration of the RTK inhibitorAZD2171 resulted in reduced tumor burden in the Apc�/min model.AZD2171 inhibits several RTKs including, but not limited to,VEGFR-1, -2, and -3 (33). The authors (32) noted that earliertreatment onset (at 6 weeks) with AZD2171 was able to reducetumor number, whereas later intervention (at 10 weeks) onlyreduced tumor size. That anti-VEGF-A Mab G6–31 did not reducethe number of tumors potentially reflects its unique mechanism oftumor inhibition, by antiangiogenesis. Moreover, it is possible thata tumor prevention approach (with an earlier treatment onset) ismore effective at reducing tumor number than a tumor interventionapproach (with a later treatment onset) that was used in our study.

Our study demonstrates that targeting VEGF-A is sufficient toachieve profound therapeutic effects in the Apc�/min model. Com-paring systemic VEGF-A inhibition by Mab G6–31 to geneticdeletion of VEGF-A in the intestinal epithelial compartmentsuggests that, in addition to epithelial cells, other cellular sources ofVEGF-A play an important role in Apc�/min adenoma growth.These additional sources of VEGF-A potentially include mononu-clear cells (34) and stromal fibroblasts (35, 36). Our in situ analysisindicates extraepithelial VEGF-A expression within the adenomasand normal villi, supporting this notion.

It is conceivable that much of the observed antitumor effects ofMab G6–31 is mediated by suppression of VEGFR-2-dependentangiogenesis (4, 37). Indeed, a reduced vascular supply in responseto anti-VEGF-A monoclonal antibody has been observed in severaltumor xenograft studies (38). In agreement with these observations,using three independent vascular markers, we were able to dem-

onstrate a significant reduction in vessel area density of the Apc�/minintestinal adenomas after 3 or 6 weeks of administration of MabG6–31 compared with control IgG treatment. However, in contrastto our study, treatment with AZD2171 was reported to have noeffect on vascular density (32). Whether such lack of effects ofAZD2171 on blood vessel density reflects qualitative and/or quan-titative differences in the mechanism of tumor suppression com-pared with the anti-VEGF Mab, or other experimental variables,remains to be established.

Recent studies have raised the possibility that VEGF-A also mayhave direct effects on intestinal epithelial cells through VEGFR-1(39, 40). Future studies are required to determine whether such amechanism plays a role in the growth of Apc�/min adenomas.

Fig. 4. Distribution of vascular markers and quantification of blood vesseldensity after Mab G6–31 treatment. (A and B) Fluorescence micrographs com-paring the distribution of vascular endothelial cells positively stained for CD105(red, isolated in C and D), CD31 (green, isolated in E and F), or von WillebrandFactor (blue, isolated in G and H) in tumors from Apc�/min mice after 6 weeks oftreatment with G6–31 or control IgG. (Scale bar: 100 �m, applies to all panels). (I)After 3 or 6 weeks of G6–31 treatment, vascular density expressed as the percentarea positive for vascular markers CD31, CD105, or von Willebrand Factor (vWF)relative to total tumor area is significantly reduced in Apc�/min mice when com-pared with IgG control. Columns represent mean vessels area density (n � 3–6tumors per mouse, 2 mice per group); bars represent SEM. *, P � 0.01; **, P �0.005 significant difference compared with IgG control treatment.

Korsisaari et al. PNAS � June 19, 2007 � vol. 104 � no. 25 � 10629

MED

ICA

LSC

IEN

CES

Dow

nloa

ded

by g

uest

on

June

18,

202

1

http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1http://www.pnas.org/cgi/content/full/0704213104/DC1

A long-standing concept postulates that a tumor requires avascular support to grow larger than 1 mm (41). Recent studies alsoindicate that tumors �1 mm in diameter remain in a nonangiogenicstate for 100 days or more (42). In the present study, however, weobserved significant accumulation of adenomas smaller than 1 mmupon inhibition of VEGF-A. This suggests that, in the intestinaladenomas of the Apc�/min mice, the angiogenic switch (43) mayoccur earlier than generally believed for tumor development, as hasbeen seen in the Apc716 model (35).

An important and unexpected conclusion of our study is that amonotherapy targeting a single angiogenic factor may be highlyeffective at suppressing tumor growth and yield a survival benefit.In contrast, in advanced malignant tumors antiangiogenic therapyseems to be most useful when combined with cytotoxic chemo-therapy (11, 12, 44). Therefore, our data suggest the possibility ofa nonsurgical treatment for benign tumors, even without the needof chemotherapeutic agents. However, it should be noted that asystemic VEGF-A blockade can be associated with significant sideeffects in some patients, including hypertension, proteinuria, andarterial thromboembolism (45). Therefore, further understandingof the epidemiology and mechanism of such side effects will beneeded to better define the risk/benefit ratio before clinical trialswith VEGF-A inhibitors in benign tumors can be considered.Finally, selection of an anti-VEGF agent with the appropriatebalance of efficacy/toxicity may be critical for such long-termtreatments (23).

Materials and MethodsAnimal Acquisition and Husbandry. Apc�/min mice (stock number002020; ref. 16) and 12.4KbVilCre mice (stock number 004586),hereafter VillinCre (22), were obtained from The Jackson Labo-ratory (Bar Harbor, ME). VEGFlox/lox mice (hereafter VEGFlox)have been described in ref. 46. Mice were housed in micro isolatorcages in a barrier facility and fed ad libitum. Maintenance of

animals and experimental protocols were conducted by followingfederal regulations and approved by Institutional Animal Care andUse Committee.

Treatment of Mice with anti-VEGF-A or Control IgG Antibodies. Theanti-VEGF-A Mab G6–31 was derived from human Fab phagelibraries as described in ref. 18. To generate an antibody suitable forlong-term administration in mice, the variable domains weregrafted into murine IgG2a constant domain. Mab G6–31 (18) orisotype matched control murine IgG2a (anti-gp120), both at thedose of 5 mg/kg, was administered i.p. once a week in a 90- to 140-�lvolume in PBS. Treatment durations were 3 weeks, 6 weeks, up to1 year, or until the mice were found moribund. Treatment of 10–14mice per each group was started at 91 � 3 days of age.

Analysis of Tumor Size and Number. The gastrointestinal tract, fromglandular stomach to rectum, was opened longitudinally, rinsed,and spread flat on a filter paper. After overnight fixation with NotoxHisto Fixative (Scientific Design Laboratory, Inc., Des Plaines, IL)and staining with methylene blue 0.1% aqueous solution, thenumber, location, and diameter of each intestinal adenoma of thesmall and large bowel was scored by a single observer, blinded tothe treatment, through an ocular scale under �20 magnification ona Leica dissection microscope. By this method, polyps with adiameter 0.3 mm or greater were recorded reliably. Tumor volumeswere calculated as hemispheres. Tumor burden for each mouse wascalculated as a sum of its tumor volumes. P values were calculatedby a two-tailed Student’s t test. A nontreated group of mice (day 0)was analyzed at the age of treatment onset (13 weeks) as a controlto the antibody treated mice. Apc�/min;Villin-Cre andApc�/min;VEGFlox;Villin-Cre mice were likewise analyzed at 13weeks of age.

We thank Laurie Leong for clinical chemistry analysis.

1. Risau W (1997) Nature 386:671–674.2. Folkman J (1995) Nat Med 1:27–31.3. Ferrara N, Gerber HP, LeCouter J (2003) Nat Med 9:669–676.4. Ferrara N (2004) Endocr Rev 25:581–611.5. Kim KJ, Li B, Winer J, Armanini M, Gillett N, Phillips HS, Ferrara N (1993) Nature

362:841–844.6. Warren RS, Yuan H, Matli MR, Gillett NA, Ferrara N (1995) J Clin Invest

95:1789–1797.7. Gerber HP, Kowalski J, Sherman D, Eberhard DA, Ferrara N (2000) Cancer Res

60:6253–6258.8. Holash J, Davis S, Papadopoulos N, Croll SD, Ho L, Russell M, Boland P, Leidich

R, Hylton D, Burova E, et al. (2002) Proc Natl Acad Sci USA 99:11393–11398.9. Prewett M, Huber J, Li Y, Santiago A, O’Connor W, King K, Overholser J, Hooper

A, Pytowski B, Witte L, et al. (1999) Cancer Res 59:5209–5218.10. Manley PW, Martiny-Baron G, Schlaeppi JM, Wood JM (2002) Expert Opin Investig

Drugs 11:1715–1736.11. Hurwitz H, Fehrenbacher L, Novotny W, Cartwright T, Hainsworth H, Helm W,

Berlin J, Baron A, Griffing S, Holmgren E, et al. (2004) N Engl J Med 350:2335–2342.12. Sandler A, Gray R, Perry MC, Brahmer J, Schiller JH, Dowlati A, Lilenbaum R,

Johnson DH (2006) N Engl J Med 355:2542–2550.13. Motzer RJ, Hutson TE, Tomczak P, Michaelson MD, Bukowski RM, Rixe O,

Oudard S, Negrier S, Szczylik C, Kim ST, et al. (2007) N Engl J Med 356:115–124.14. Escudier B, Eisen T, Stadler WM, Szczylik C, Oudard S, Siebels M, Negrier S,

Chevreau C, Solska E, Desai AA, et al. (2007) N Engl J Med 356:125–134.15. Polakis P (2000) Genes Dev 14:1837–1851.16. Moser AR, Pitot HC, Dove WF (1990) Science 247:322–324.17. Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, Luongo C, Gould KA,

Dove WF (1992) Science 256:668–670.18. Liang WC, Wu X, Peale FV, Lee CV, Meng YG, Gutierrez J, Fu L, Malik AK,

Gerber HP, Ferrara N, Fuh G (2006) J Biol Chem 281:951–961.19. Braunstein EM, Qiao XT, Madison B, Pinson K, Dunbar L, Gumucio DL (2002)

Dev Dyn 224:90–102.20. Maunoury R, Robine S, Pringault E, Leonard N, Gaillard JA, Louvard D (1992)

Development (Cambridge, UK) 115:717–728.21. Robine S, Huet C, Moll R, Sahuquillo-Merino C, Coudrier E, Zweibaum A,

Louvard D (1985) Proc Natl Acad Sci USA 82:8488–8492.22. Madison BB, Dunbar L, Qiao XT, Braunstein K, Braunstein E, Gumucio DL (2002)

J Biol Chem 277:33275–33283.23. Gerber HP, Wu X, Yu L, Weissman C, Liang XH, Lee CV, Fuh G, Olsson C,

Damico L, Xie D, et al. (2007) Proc Natl Acad Sci USA 104:3478–3483.

24. Corpet DE, Pierre F (2003) Cancer Epidemiol Biomarkers Prev 12:391–400.25. Boolbol SK, Dannenberg AJ, Chadburn A, Martucci C, Guo XJ, Ramonetti JT,

Abreu-Goris M, Newmark HL, Lipkin ML, DeCosse JJ, Bertagnolli MM (1996)Cancer Res 56:2556–2560.

26. Chiu CH, McEntee MF, Whelan J (1997) Cancer Res 57:4267–4273.27. Hansen-Petrik MB, McEntee MF, Johnson BT, Obukowicz MG, Masferrer J,

Zweifel B, Chiu CH, Whelan J (2002) Cancer Lett 175:157–163.28. Ritland SR, Gendler SJ (1999) Carcinogenesis 20:51–58.29. Oshima M, Dinchuk JE, Kargman SL, Oshima H, Hancock B, Kwong E, Trzaskos

JM, Evans JF, Taketo MM (1996) Cell 87:803–809.30. Jacoby RF, Seibert K, Cole CE, Kelloff G, Lubet RA (2000) Cancer Res 60:5040–

5044.31. Wagenaar-Miller RA, Hanley G, Shattuck-Brandt R, DuBois RN, Bell RL,

Matrisian LM, Morgan DW (2003) Br J Cancer 88:1445–1452.32. Goodlad RA, Ryan AJ, Wedge SR, Pyrah IT, Alferez D, Poulsom R, Smith NR,

Mandir N, Watkins AJ, Wilkinson RW (2006) Carcinogenesis 27:2133–2139.33. Wedge SR, Kendrew J, Hennequin LF, Valentine PJ, Barry ST, Brave SR, Smith

NR, James NH, Dukes M, Curwen JO, et al. (2005) Cancer Res 65:4389–4400.34. Sunayama K, Konno H, Nakamura T, Kashiwabara H, Shoji T, Tsuneyoshi T,

Nakamura S (2002) Carcinogenesis 23:1351–1359.35. Seno H, Oshima M, Ishikawa TO, Oshima H, Takaku K, Chiba T, Narumiya S,

Taketo MM (2002) Cancer Res 62:506–511.36. Williams CS, Tsujii M, Reese J, Dey SK, DuBois RN (2000) J Clin Invest 105:1589–1594.37. Shibuya M, Claesson-Welsh L (2005) Exp Cell Res 312:549–560.38. Borgstrom P, Bourdon MA, Hillan KJ, Sriramarao P, Ferrara N (1998) Prostate

35:1–10.39. Bates RC, Goldsmith JD, Bachelder RE, Brown C, Shibuya M, Oettgen P, Mercurio

AM (2003) Curr Biol 13:1721–1727.40. Lesslie DP, Summy JM, Parikh NU, Fan F, Trevino JG, Sawyer TK, Metcalf CA,

Shakespeare WC, Hicklin DJ, Ellis LM, Gallick GE (2006) Br J Cancer 94:1710–1717.

41. Gimbrone MA, Jr, Leapman SB, Cotran RS, Folkman J (1972) J Exp Med136:261–276.

42. Naumov GN, Bender E, Zurakowski D, Kang SY, Sampson D, Flynn E, Watnick RS,Straume O, Akslen LA, Folkman J, Almog N (2006) J Natl Cancer Inst 98:316–325.

43. Hanahan D, Folkman J (1996) Cell 86:353–364.44. Jain RK (2005) Science 307:58–62.45. Ferrara N, Kerbel RS (2005) Nature 438:967–974.46. Gerber HP, Hillan KJ, Ryan AM, Kowalski J, Keller G-A, Rangell L, Wright BD,

Radtke F, Aguet M, Ferrara N (1999) Development (Cambridge, UK) 126:1149–1159.

10630 � www.pnas.org�cgi�doi�10.1073�pnas.0704213104 Korsisaari et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

1