Microvascular Research biology and pathology - part 9 doc

Growth Patterns and Angiogenesis in Primary Lung Cancer Bronchiolo-alveolar lung adenocarcinomas have a cal “lepidic” growth pattern: Tumor cells replace the normalpneumocytes, thereby p

944 P ART III Pathology

metastases with a replacement growth pattern expressed

the hypoxia marker CA IX or had fibrin depositions at

the tumor–liver interface (Table II) Probably, the

well-described mechanisms of invasive tumor growth, such as

fibroblast–myofibroblast transdifferentiation, TGFb

path-ways, proinflammatory signaling, hyaluronic acid action,

and hypoxia-responsive gene activation, are not involved

The search for gene sets that are responsible for the

pheno-type of nonangiogenesis-dependent colonization of a

distant site is ongoing Selective induction of apoptosis in

hepatocytes at the interface by tumor cells might be one

of the mechanisms of growth of blood-vessel-coopting

metastases The other growth patterns in the liver were

characterized by destruction of the architecture of the liver

parenchyma and were associated with desmoplasia and new

blood vessel formation The metastases were (desmoplastic

growth pattern) or were not (pushing growth pattern)

sur-rounded by a fibrotic capsule

The consequences of this heterogeneity of human liver

metastases are the limited value of model systems that

selec-tively reproduce the well-studied angiogenesis-dependent

growth of metastases and the difficulties in analyzing the

results of clinical trials applying biomodulatory drugs

Imaging of the vascular flow and leakage by

contrast-enhanced CT or MR might be helpful in selecting patients

with angiogenic versus nonangiogenic liver metastases The

existence of different growth patterns also stresses the

supe-rior value of the endothelial cell proliferation (ECP) fraction

for angiogenesis quantification compared to microvessel

density [1] Liver metastases with a replacement growth

pat-tern indeed have a high microvessel density as a result of

cooption of the liver vasculature but have a low ECP due to

lack of ongoing angiogenesis

Growth Patterns and Angiogenesis in Primary Lung Cancer

Bronchiolo-alveolar lung adenocarcinomas have a cal “lepidic” growth pattern: Tumor cells replace the normalpneumocytes, thereby preserving the stromal component ofthe alveolar wall and coopting the capillary blood vessels.The structure of the lung parenchyma remains intact, whichprobably explains the growth of satellite lesions in the lungdue to transportation of tumor cells by airflow This nonan-giogenic growth has been shown to be present in more com-mon types of nonsmall-cell lung carcinomas (NSCLC) and

typi-in lung metastases [3] In 16 percent of 500 NSCLC, thetumors were growing without parenchymal destruction andwithout the formation of desmoplastic stroma In this “alve-olar” growth pattern, tumor cells were filling the alveoli assolid nests The only blood vessels present were those in thepreserved alveolar septa The coopted blood vessels did notexpress the integrin alpha-v-beta-3 necessary for angiogen-esis In another study, outcome of 283 patients with opera-ble NSCLC was studied and linked to the growth pattern ofthe lung tumors (Sardari Nia P et al., 2004) Whereas themajority of the patients had a tumor with associated desmo-plasia and angiogenesis in which the alveolar architecture ofthe lung was not preserved, 18 percent of the patients had aNSCLC with an alveolar, nonangiogenic growth at thetumor–lung interface The alveolar growth pattern was notassociated with a specific histiotype (30% adenocarcinoma,40% squamous cell carcinoma, and 30% large cell carci-noma and other types) In univariate analysis, T-stage, N-stage, and growth pattern predicted overall and disease-freesurvival Multiple logistic regression showed that TN-stageand growth pattern were independent prognostic factors.Hazard ratios for the alveolar growth pattern were 2.0 (95%confidence interval: 1.3 to 3.2) for overall survival and 2.4(95% confidence interval: 1.5 to 3.8) for disease-free sur-vival, if compared to NSCLC with associated desmoplasiaand angiogenesis When stage I tumors were analyzed sepa-rately (174 patients), growth pattern retained its independentprognostic value This confirms the study of Pastorino et al.[8], which described 137 pT1N0 patients Both a nonangio-genic type of vascular pattern and epidermal growth factorreceptor expression were associated with a poorer survivalrate

Assessing the growth pattern in primary NSCLC ispotentially important since it can predict prognosis, butprobably also the response to different treatment modalities.The growth pattern is indeed an integrative parameter con-taining information of the relationship between tumor cellsand stromal cells Surgical pathologists can easily determinethe growth pattern on a standard hematoxylin–eosin stainedtissue section and integrate it in the pathology report.Another consequence of the growth patterns of lung carci-nomas is that the prognostic value of microvessel density inNSCLC can only be investigated within the subgroup ofangiogenic tumors

Figure 2 Replacement growth pattern in a liver metastasis of a breast

adenocarcinoma The tumor cells (left) are replacing the hepatocytes in the

liver plates (right), thereby coopting the sinusoidal blood vessels There is

close apposition of tumor cells and hepatocytes at the tumor–liver interface

(arrows) without induction of inflammation or fibrosis.

CHAPTER 139 Tumor Growth Patterns and Angiogenesis 945

Conclusion

Different growth patterns of primary and metastatic

tumors are a reflection of different interactions of the cancer

cells with the surrounding tissue structures The observation

of nonangiogenic growth patterns in human carcinomas

challenges the hypothesis that tumor growth is always

dependent on angiogenesis It would be more correct to say

that neoplastic growth depends on an adequate blood supply

If this can be obtained from a vascular bed that already

exists, the tumor can grow without the formation of new

blood vessels

Glossary

Alveolar growth pattern: Nonangiogenic growth pattern described in

primary nonsmall-cell lung cancer and in pulmonary metastases; tumor

cells fill the alveoli and exploit the interalveolar capillaries for their blood

supply.

Fibrotic focus: Focus of exaggerated reactive tumor stroma formation

in the center of a carcinoma consisting of collagen, a variable number of fibroblasts, blood vessels, and inflammatory cells; practical histopatholog- ical surrogate marker of hypoxia-driven angiogenesis in breast cancer.

Nonangiogenic growth: Tumor growth without induction of

angio-genesis in which tumor cells obtain adequate blood supply by exploiting a preexisting vascular bed.

Replacement growth pattern: Nonangiogenic growth pattern

described in liver metastases; tumor cells replace the hepatocytes in the liver plates and exploit the sinusoidal blood vessels for their blood supply.

References

1 Vermeulen, P B., Gasparini, G., Fox, S B., Colpaert, C G., Marson, L., Gion, M., Beliën, J A M., de Waal, R M W., Van Marck, E A., Magnani, E., Weidner, N., Harris, A L., and Dirix, L Y (2002) Second international consensus on the methodology and criteria of evaluation of

angiogenesis quantification in solid human tumours Eur J Cancer 38,

1564–1579 Guidelines for the estimation of ongoing angiogenesis and

the amount of blood vessels in a solid tumor, integrating new concepts and mechanisms of tumor vascularization.

Table II Comparison of Glandular Differentiation, Fibrin Deposition, CAIX Expression, and the Macrophage Content of Breast Cancer and Colorectal

Cancer Liver Metastases.

Fibrin deposition: Detected immunohistochemically with NYB.T2G1monoclonal antibody 0, No staining;

1, minimal staining; 2, moderate staining; 3, extensive staining.

Global CA IX score: Carbonic anhydrase IX is an endogenous marker of hypoxia Its expression is tively scored as the product of the percentage of immunostained cells with an immunostaining intensity score rang-

semiquantita-ing from 0 (no stainsemiquantita-ing) to 3 (strong stainsemiquantita-ing).

CA IX: absent, no immunostaining; present, immunostaining in any percentage of tumor cells.

Macrophage count: the relative area occupied by CD68 immunostained macrophages, quantified with the Chalkley morphometric point counting method.

946 P ART III Pathology

2 Hasebe, T., Tsuda, H., Hirohashi, S., Shimosato, Y., Iwai, M., Imoto, S.,

and Mukai, K (1996) Fibrotic focus in invasive ductal carcinoma:

An indicator of high tumor aggressiveness Jpn J Cancer Res 87,

385–394 First report presenting the fibrotic focus as a prognostic

factor in invasive breast carcinoma.

3 Pezzella, F., Pastorino, U., Tagliabue, E., Andreola, S., Sozzi, G.,

Gasparini, G., Menard, S., Gatter, K C., Harris, A L., Fox, S., Buyse,

M., Pilotti, S., Pierotti, M., and Rilke, F (1997) Non-small-cell

lung carcinoma tumor growth without morphological evidence of

neo-angiogenesis Am J Pathol 151, 1417–1423 Investigation of the

pat-tern of vascularization in a series of 500 lung carcinomas with the

description of an alveolar growth pattern characterized by the lack of

parenchymal destruction and absence of both tumor associated stroma

and new vessels.

4 Vermeulen, P B., Colpaert, C G., Salgado, R., Royers, R., Hellemans,

H., Van de Heuvel, E., Goovaerts, G., Dirix, L Y., and Van Marck, E A.

(2001) Liver metastases from colorectal adenocarcinomas grow in three

patterns with different angiogenesis and desmoplasia J Pathol 195,

336–342 Description of different patterns of vascularization in liver

metastases with identification of a growth pattern characterized by

tumor cells replacing the hepatocytes in the liver plates and exploiting

the preexisting sinusoidal blood vessels.

5 Colpaert, C G., Vermeulen, P B., Fox, S B., Harris, A L., Dirix, L Y.,

and Van Marck, E A (2003) The presence of a fibrotic focus in

inva-sive breast carcinoma correlates with the expression of carbonic

anhy-drase IX and is a marker of hypoxia and poor prognosis Breast Cancer

Res Treat 81, 137–147.

6 Colpaert, C G., Vermeulen, P B., van Beest, P., Goovaerts, G., Weyler, J., Van Dam, P., Dirix, L Y., and Van Marck, E A (2001) Intratumoral hypoxia resulting in the presence of a fibrotic focus is an independent predictor of early distant relapse in lymph node-negative breast cancer

patients Histopathology 39, 416–426.

7 Colpaert, C G., Vermeulen, P B., van Beest, P., Goovaerts, G., Dirix,

L Y., Harris, A L., and Van Marck, E A (2003) Cutaneous breast cer deposits show distinct growth patterns with different degrees of

can-angiogenesis, hypoxia and fibrin deposition Histopathology 42,

530–540.

8 Pastorino, U., Andreola, S., Tagliabue, E., Pezzella, F., Incarbone, M., Sozzi, G., Buyse, M., Menard, S., Pierotti, M., and Rilke, F (1997) Immunocytochemical markers in Stage I lung cancer: Relevance to

prognosis J Clin Oncol 15, 2858–2865.

Capsule BiographyCecile G Colpaert is a pathologist working in a teaching hospital Her main interest is breast cancer and the application of research findings from the field of tumor biology in diagnostic pathology practice.

Peter B Vermeulen is a diagnostic pathologist doing translational breast cancer research mainly focused on angiogenesis and tumor–stroma interactions.

C HAPTER 140

Breast Cancer Resistance Protein in

Microvessel Endothelium

Hiran C Cooray and Margery A Barrand

University of Cambridge, Cambridge, United Kingdom

organisms, including bacteria, plants, and animals tively, they are responsible for transporting a multitude ofdiverse substrates including sugars, ions, lipids and phos-pholipids, peptides, bile acids, sterols, pigments, and xeno-biotics across membranes, thus affecting the distribution ofmolecules at subcellular, cellular, and tissue levels Thereare many different ABC transporters The mammalian ones have now been classified into subfamilies, termedABCA through ABCG (http://www.humanabc.org/) In this scheme, BCRP is named as ABCG2 The reader isdirected to http://www.ncbi.nlm.nih.gov/books/bookres.fcgi/mono_001/mono_001.pdf and http://arjournals.annualreviews.org/doi/pdf/10.1146/annurev.biochem.71.102301.093055 for two comprehensive reviews on ABCtransporters

Collec-Structural Organization of ABC Transporters

Many of the ABC transporters are constructed as a dem repeat of a basic unit containing two domains: an N-terminal transmembrane domain (TD) of 6 to 11 a-helices

tan-that provide substrate specificity to the protein, and a terminal nucleotide-binding domain (NBD) located in thecytoplasm that binds and cleaves ATP in order to generatethe energy for substrate transport Binding and subsequenthydrolysis of ATP result in a conformational change thatcauses the substrate to be translocated across the membrane.This general structure is typified by the mammalian mul-tidrug transporter, P-glycoprotein, otherwise named ABCB1(Figure 1) Some ABC transporters including the MultidrugResistance-associated Protein, MRP1 (ABCC1), have anadditional N-terminal TD However others contain only one NBD and one TD This “half transporter” structure is

C-Introduction

Breast Cancer Resistance Protein (BCRP) is a transporter

recently identified as a member of the ATP Binding Cassette

(ABC) superfamily of transmembrane proteins (discussed in

the following section) It was recognized initially in several

drug-resistant tumor cell lines (as discussed later) and was

subsequently identified in certain tumor tissues where its

presence has been putatively associated with poor clinical

response to chemotherapy However, BCRP is found on

many normal, that is, nonmalignant, cells at a number of

dif-ferent sites in the body (see later discussion) These include

endothelial cells lining various vasculature Clues to the

possible physiological role or roles of BCRP in these

loca-tions may be derived from the many studies that are now

being conducted, which examine its mechanisms of action

and its substrate (discussed later) and inhibitor profiles in

different cell systems or that analyze its influence on the

dis-tribution of drugs in whole animals

ABC Transporters

Transporters belonging to the ATP-Binding Cassette (or

ABC) family of proteins have the ability to transport

sub-strates across cellular membranes against a concentration

gradient using the energy of ATP hydrolysis They are so

called because of their distinctive ATP-binding domains,

which contain highly conserved sequences, Walker A and

Walker B and an additional ABC “signature” sequence

ABC proteins constitute the largest subclass of

transmem-brane proteins and are expressed ubiquitously in all living

Copyright © 2006, Elsevier Science (USA).

948 P ART III Pathology

exemplified by members of several ABC subfamilies

includ-ing the ABCG subfamily of which BCRP is a member In

addition, members of this subfamily have their TD and NBD

in reverse orientation (see Figure 1); hence, BCRP is

referred to as a “reverse” half-transporter Half transporters

have to dimerize with a partner protein (either with itself to

form a homodimer or with another protein to form a

het-erodimer) in order to be functional Currently, the bulk of

the evidence points to BCRP being a homodimer

Interest-ingly however, BCRP shows closest homology with the

white, brown, and scarlet proteins that heterodimerize

(white/brown and white/scarlet) to transport pigment

pre-cursors (guanine and tryptophan) in the Drosophila eye.

ABC Transporters Associated with Multidrug Resistance

Drug resistance can be a serious obstacle to successfulanticancer treatment with tumors often failing to respondeither to initial chemotherapy (intrinsic resistance) or tosubsequent rounds of treatment (acquired resistance) Stud-ies conducted in the laboratory on tumor cell lines cultured

in the presence of cytotoxic drugs reveal that resistance candevelop, not only to the selecting drug, but to a number

of structurally and functionally dissimilar drugs as well,hence providing multidrug resistance (MDR) This MDR

Figure 1 Putative membrane topology of the three main multidrug transporters BCRP consists

of one transmembrane domain (TD) and one nucleotide binding domain (NBD) and is termed a

half-transporter P-glycoprotein, like many ABC proteins, is a full transporter with a TD1-NBD1

-TD2-NBD2structure Several of the MRP family members, including MRP1, have an additional N-terminal TD linked to the P-gp-like core by a linker region (L0) Note that the TD and NBD in

BCRP are in reverse orientation to those of the other two ABC proteins (N and C refer to the N and

C termini of the transporter, respectively.) (see color insert)

CHAPTER 140 Breast Cancer Resistance Protein in Microvessel Endothelium 949

phenomenon may involve several different types of

mecha-nisms, but a common cause is the presence of multidrug

ABC transporters, which prevent access of the drugs to their

intracellular targets sites They accomplish this task by

either effluxing the drug out of the cell via the plasma

mem-brane, or by sequestering the drug within intracellular

organelles such as the endoplasmic reticulum, lysosome, or

peroxisome

Identification of BCRP as a Multidrug Transporter

P-gp was the first ABC transporter to be associated with

MDR (around the mid-1970s) with the MRPs being later

discoveries (MRP1 sequenced in 1992) It was not until

1998 that BCRP was revealed as another MDR-associated

ABC transporter It was becoming apparent that there were

certain tumor cell lines that showed resistance to several

drugs (mainly mitoxantrone, bisantrene, topotecan, and

dox-orubicin) yet did not overexpress P-glycoprotein or MRPs

Recognition of this atypical, non-P-gp, non-MRP resistance

phenotype initiated a search for another multidrug

trans-porter, leading to the ultimate discovery of BCRP, otherwise

termed Mitoxantrone Resistance Protein (MXR) or ABC

Transporter in Placenta (ABCP) The different names of this

protein derive from the fact that it was characterized and

cloned from three different sources in independent

laborato-ries at about the same time: from a multidrug-resistant

breast cancer cell line selected in doxorubicin (MCF7/

AdrVp3000, hence BCRP), from a colon cancer line

selected in mitoxantrone (S1-M1-80, hence MXR), and

from human placenta (hence ABCP) [1]

Though commonly called BCRP by virtue of its initialdiscovery in a drug-selected breast cancer cell line, it is notclear that the protein is actually often overexpressed inbreast cancers in vivo Indeed only weak BCRP expressionhas been found in breast tumors There are, however, manydifferent human tumors in which BCRP expression has beenclearly demonstrated, including tumors of the kidney, ovary,stomach, colon, thyroid, brain, endometrium, and testis;squamous tumors (lung, head and neck, and esophagus);soft tissue sarcomas; pheochromocytomas; and hepatocarci-nomas This may reflect the distribution of BCRP in normaltissues (see later discussion)

Substrate and Inhibitor Profiles of BCRP Compared

to Other Multidrug Transporters

The ABC transporters associated with MDR vary what in their mechanisms of action, substrate and inhibitorprofiles, and in their tissue locations Nevertheless there is asignificant overlap in these characteristics between thetransporters (Figure 2)

some-Functional studies on both BCRP overexpressing drug-selected and BCRP-transfected cell lines show that thetransporter can confer resistance to anthracyclines (dox-orubicin, daunorubicin), anthracenediones (mitoxantrone),camptothecins (topotecan, irinotecan, and its active metabo-lite, SN-38), and etoposide, but not to vincristine, taxol, orcolchicine, which are classical P-gp substrates Substratescurrently recognized to be transported by BCRP are shown

in Table I

Figure 2 Actions of multidrug transporters BCRP, P-gp, and MRP1 in efflux of substances from cells Drugs can enter and leave cells by passive diffusion along a concentration gradient Multidrug transporters provide a second route for drug exit and can drive drugs out of the cell against a con- centration gradient by exploiting the energy of ATP hydrolysis This additional efflux reduces drug concentrations inside cells to sublethal levels P-gp and BCRP (in the form of a dimer) can efflux drugs unmodified (mainly hydrophobic, amphipathic compounds) The MRPs require the presence of reduced glutathione (GSH) to transport unmodified drugs (predominantly organic anions) but can transport drugs following their conjugation (GST, Glutathione transferase) (see color insert)

950 P ART III Pathology

A point mutation affecting substrate specificity has

recently been reported in BCRP Though wild-type BCRP

has an arginine at the 482nd position (at the start of the third

transmembrane segment), BCRP overexpressed in certain

drug-selected cell lines was shown to contain either a

glycine or threonine at this site This point mutation causes

a paradigm shift in the protein’s substrate

specificity—wild-type BCRP is incapable of effluxing the fluorescent dye

rho-damine 123 or anthracycline drugs such as doxorubicin, but

the mutants can handle both these substrates Conversely,

only wild-type BCRP transports the antifolate cytotoxic

methotrexate [1] Polymorphisms of the BCRP gene have

also been described in human populations

Subcellular Location of BCRP

BCRP seems to be predominantly localized to the plasma

membrane of both drug-selected and transfected cell lines

This sets it apart from other half-transporters that are

local-ized mainly to intracellular membranes such as the

mito-chondrion (ABCB7), the peroxisome (ALD subfamily), and

the endoplasmic reticulum (Tap1/Tap2) Such a location for

BCRP is consistent with a putative role in the efflux of strates from the cell Furthermore, it is apparent in polarizedcell lines such as BCRP-transfected MDCK-II Madine-Darby canine kidney cells that the transporter becomeslocalized primarily to the apical aspect of the plasma mem-brane, where it mediates the translocation of substrates frombasal to apical side However, a role for BCRP in the intra-cellular trafficking of molecules cannot be ruled out as someimmunocytochemical studies have reported perinuclearstaining for BCRP in several topotecan- and mitoxantrone-resistant cell lines [1]

sub-Tissue Distribution of BCRP Compared to Other Multidrug Transporters

The apical siting of BCRP on polarized cells is of ular relevance to the possible role or roles of BCRP in nor-mal tissues The protein is found in many tissues, includingbarrier sites, as outlined in Table II

partic-The highest BCRP expression is found in the placenta onthe syncytiotrophoblast facing the maternal circulation Thissuggests a role for the protein in the elimination of sub-strates from the fetus This has been established for mouseBcrp1; in both wild-type and P-gp knockout mice, inhibition

of Bcrp1 by GF120918 (a common inhibitor of human andmouse P-gp and BCRP) resulted in at least a twofoldincrease in the fetal uptake of orally administered topotecan,

a BCRP substrate BCRP is also expressed at more modestlevels in the colon, small intestine, liver, ovary, and breast,where it may be concerned with elimination of materialfrom these tissues In a recent clinical study utilizingGF120918, it was shown that the oral bioavailability oftopotecan more than doubled (from 40% to 97%) when thedrug was coadministered with the inhibitor, thus underliningthe functional significance of BCRP expression in the intestine

Table I Substrate Profile of BCRP

Compared with P-gp and MRP1.

E2-GLU, 17 b-Estradiol 17-(b- D -glucuronide); GSH, reduced

glu-tathione; LTC4, leukotriene C4 Rhodamine 123 and Lysotracker Green are

fluorescent dyes.

aMethotrexate is effluxed only by the wild-type BCRP.

Table II Tissue Distribution and Putative Functions of BCRP at These Locations.

Localization Putative function

Placenta—syncytiotrophoblast Protection of fetus, excretion of

substrates Apical membrane of epithelium Reduced uptake/excretion of

of small intestine and colon substrates into maternal circulation Liver canalicular membrane Excretion of substrates by the liver

into bile Apical membrane of lobules Unknown and lactiferous ducts of breast

Endothelium of capillaries and Unknown veins

“Side” population of Unknown hematopoietic stem cells

CHAPTER 140 Breast Cancer Resistance Protein in Microvessel Endothelium 951

This distribution of BCRP shows similarities to that of

the multidrug transporter P-glycoprotein, which is also

expressed in various epithelia, particularly in organs

associ-ated with drug absorption and disposition, such as

hepato-cyte canalicular membrane and the intestinal mucosa P-gp

is thought to provide a first line of defense against the entry

of many types of xenobiotics into the body Knockout mice

deficient in functional P-gp, although viable, fertile, and

without obvious histological or developmental

abnormali-ties, show significantly altered pharmacokinetics (and

toxicity of several drugs) [2] The third subfamily of

multidrug transporters, the MRPs, are widely distributed

throughout the body in tissues including the choroid plexus,

oral mucosa, small intestine, testis, and respiratory tract

Because there are several MRP homologs with overlapping

substrate specificities, the importance of each for the

elimi-nation of particular substances is difficult to assess

Both BCRP and P-gp are to be found on the endothelium

lining the blood–brain barrier (see later discussion) The

presence of multidrug transporters at such barrier sites

creates “pharmacological sanctuaries” within the body,

per-mitting certain organs and tissues to function in relative

isolation from the rest of the body Indeed, in P-gp knockout

mice, the integrity of the blood–brain barrier is shown to be

significantly compromised, with much higher brain

penetra-tion of P-gp substrates such as vinblastine and ivermectin

being demonstrated The relevance of BCRP at these sites is

still under investigation (see later discussion)

The generation of the Bcrp1 knockout mouse [3] has

thrown new light on the putative physiological function of

this transporter Though these mice were anatomically

nor-mal and fertile, a defect was seen in their ability to handle a

metabolite of chlorophyll, pheophorbide a, resulting in

severe phototoxicity in mice exposed to light They also

exhibited a previously uncharacterized form of porphyria

Thus it became known that BCRP performs an essential

function at the gut epithelium in effluxing toxic products of

chlorophyll metabolism BCRP knockout mice generated

independently by Zhou et al [4] were used to demonstrate

that this transporter, rather than P-gp, is responsible for the

dye efflux in the cells This allows analysis of the

“side-population,” enriched in murine hematopoietic stem cells,

which have high bone-marrow repopulating activity The

role BCRP plays at this location is still to be elucidated

Expression of BCRP in Endothelia of Normal Tissues

and of Tumors

BCRP differs from P-gp in being expressed on the

endothelial lining of vascular beds in many tissues, not just

at the blood-brain barrier Interestingly, BCRP is evident in

venules and capillaries (see Table II) but not in arterioles [5]

Hence the transporter is distributed in the regions of the

vasculature where the bulk of the exchange of materials

between blood and tissues occurs On endothelial cells of

vasculature-supplying tumors (for example, testicular

germ-cell tumors, endometrial, ovarian and colon mas, and brain tumors), antibody staining for BCRP hasbeen described as moderate to strong, stronger indeed than

carcino-on the vascular endothelium in the surrounding normalregions [6] This raises the interesting possibility that BCRPexpression is perhaps upregulated in the endothelium ofblood vessels during neoplastic vasculogenesis

Localization of BCRP in the Specialized Endothelium

of the Blood–Brain Barrier

The presence of multidrug transporters is of particularimportance in vascular endothelial cells at special barriersites such as the blood–brain and blood–testis barriers Herethe vessels possess tight junctions that place severe restric-tions on the free paracellular diffusion of many substancesseen in peripheral endothelia

Recent studies have explored the localization of BCRP inhuman brain material using fresh-frozen samples of bothnormal and tumor brain (meningiomas and gliomas) [7].Western blot results show a higher degree of expression ofBCRP protein in the gliomas over the normal and menin-gioma samples It could be seen by immunostaining thatBCRP is primarily localized to blood vessels within thebrain In the case of two meningioma samples, notable heterogeneous staining for BCRP was seen in brainparenchymal cells in addition to endothelial cells Diestra

et al [6] also reported a higher expression of BCRP in several unspecified brain tumors over normal brainparenchyma using immunohistochemical staining with awell-characterized anti-BCRP antibody

By exploiting the powerful resolving capabilities of the confocal microscope, it has been possible to gain some understanding of the subcellular distribution of BCRPwithin brain microvessels Utilizing the fact that the brainendothelial glucose transporter GLUT-1 is localized on bothsides of brain endothelial cells (both luminal and abluminalmembranes), dual-staining with antibodies for GLUT-1 andfor BCRP revealed the main sites of BCRP expression inmicrovessels in both normal and tumor brain sections Thedistribution of BCRP staining was seen to be inner to that ofGLUT-1 in all microvessels viewed, which suggests thatBCRP is localized toward the luminal membrane of humanbrain endothelial cells in the in vivo blood–brain barrier [7]

It is probable therefore that BCRP, localized strategically atthe luminal membrane of endothelial cells, has a protectivefunction at the blood–brain barrier in limiting entry of sub-strates into the brain

P-gp also has been localized to the luminal aspect of thebrain capillary endothelium It is already well documentedthat in this situation it performs what has been described as

a “gatekeeper” role at the blood–brain barrier, pumping out

a variety of xenobiotics that would otherwise gain access

to the brain via the transcellular pathway due to theirlipophilicity [2] The number of drugs known to be excludedfrom the brain by P-gp is large, ranging from nonsedating

952 P ART III Pathology

antihistamines, antiepileptics, and beta-blockers to anti-HIV

reverse transcriptase inhibitors What additional protection

BCRP may bring to bear is as yet not well defined and

would require the advent of knockout mice lacking several

of the MDR transporters or use of combinations of specific

inhibitors so that the influence of BCRP can be

distin-guished from that of P-gp in vivo

The presence and importance of MRPs at the

blood–brain barrier is even less clear This is due both to the

multiplicity of transporters in this family and to existing

controversies in the literature In contrast to BCRP, MRP1

is known to be functionally active in vivo at the epithelium

of the choroid plexus, regulating the distribution of several

xenobiotics into the CSF But it has not been definitively

localized in vivo at the blood–brain barrier MRP1 does,

however, become upregulated in cultured brain endothelial

cells [8] This has allowed its functionality to be explored

in vitro in brain endothelial cells cultured from several

sources including human brain In the case of MRP2,

results of recent studies using in vivo microdialysis hint

at a functional role for this protein at the rat blood–brain

barrier, limiting the brain uptake of the

anticonvul-sant phenytoin However, these observations need further

investigation

A homolog of BCRP has been described in porcine brain

endothelial cells by Eisenblatter et al [9] This protein,

named Brain Multidrug-Resistance Protein (BMDP), shows

86 percent amino acid identity with human BCRP and is

predicted to have the typical architecture of a “reverse”

half-transporter BMDP was shown, using

immunohistochem-istry, to be localized to the cell membrane of cultured

porcine brain endothelial cells A blood vessel location in

vivo for the message was inferred from RNA isolation

experiments in which the mRNA of BMDP appeared to be

concentrated in the brain microvessels, with the levels of

transcript higher in isolated capillaries than in homogenized

brain tissue

High levels of BMDP transcript were also detectable in

cultured porcine brain endothelial cells These appeared

approximately 30 times higher than equivalent P-gp

expres-sion, suggesting that at least in the porcine endothelium,

BMDP plays a more prominent role than P-gp This was

corroborated via functional studies using the radiolabeled

substrate 3H-daunorubicin (a substrate common to both

P-gp and BCRP) performed on cultured porcine brain

endothelial cells grown as monolayers—GF120918 (which

inhibits both P-gp and BCRP) abrogated almost completely

the transport of daunorubicin from the basolateral to the

api-cal side of the porcine brain endothelial cell monolayer

However, specific P-gp inhibitors gave only moderate

inhibition This strongly suggests that the contribution

by BMDP to transport of substrates across the porcine

blood–brain barrier may be greater than by P-gp These

studies are the first to report functional BCRP in cells

derived from the blood–brain barrier

Many questions still remain regarding the in vivo

func-tionality of BCRP Its ubiquitous expression in the

endothe-lium of veins and capillaries of every tissue so far examinedsuggest that it might efflux substrates that are potentiallytoxic to many tissues but are incapable of passing betweenendothelial cells In particular, its expression at theblood–brain barrier may also be of paramount significance

to limiting the brain penetration of substrates The vast body

of research available on P-gp and members of the MRPfamily has pointed to a number of specific–roles performed

by these transporters at various sites in the body There isstill much to be learned about BCRP and the function it mayperform in the microvessel endothelium

AcknowledgmentsThe authors thank the Cancer Research Campaign for their contribu- tions to the authors’ own research work and the Cambridge Commonwealth Trust for assistance toward a studentship for HCC, who also holds an award from Universities UK.

References

1 Allen, J D., and Schinkel, A H (2002) Multidrug resistance and macological protection mediated by the Breast Cancer Resistance Pro-

phar-tein (BCRP/ABCG2) Mol Can Ther 1, 427–434 A very useful and

comprehensive recent review on all aspects of BCRP.

2 Schinkel, A H (1999) P-Glycoprotein, a gatekeeper at the blood–brain

barrier Adv Drug Deliv Rev 36, 179–194.

3 Jonker, J W., Buitelaar, M., Wagenaar, E., van der Valk, M A., Scheffer, G L., Scheper, R J., Plosch, T., Kuipers, F., Oude Elferink,

R P J., Rosing, H., Beijnen, J H., and Schinkel, A H (2002) The breast cancer resistance protein protects against a major chlorophyll-

derived dietary phototoxin and protoporphyria Proc Natl Acad Sci.

USA 99, 15649–15654.

4 Zhou, S., Morris, J J., Barnes, Y., Lan, L., Schuetz, J D., and Sorrentino, B P (2002) BCRP1 gene expression is required for normal numbers of side-population stem cells in mice, and confers relative

protection to mitoxantrone in hematopoietic cells in vivo Proc Natl.

Acad Sci USA 99, 12339–12344.

5 Maliepaard, M., Scheffer, G L., Faneyte, I F., van Gastelen M A., Pijnenborg, A C L M., Schinkel, A H., van de Vijver, M J., Scheper,

R J., and Schellens, J H M (2001) Subcellular localization and tribution of the breast cancer resistance protein transporter in normal

dis-human tissues Cancer Res 61, 3458–3464.

6 Diestra, J E., Scheffer, G L., Catal, I., Maliepaard, M., Schellens,

J H M., and Scheper, R J (2002) Frequent expression of the multidrug resistance associated protein BCRP/MXR/ABCP/ABCG2 in human tumors detected by the BXP-21 monoclonal antibody in paraffin-

embedded material J Pathol 198, 213–219.

7 Cooray, H C., Blackmore, C G., Maskell, L., and Barrand,

M A (2002) Localization of Breast Cancer Resistance Protein in

microvessel endothelium of human brain Neuroreport 13, 2059–2063.

Establishes a luminal localization for BCRP at the human blood–brain barrier.

8 Seetharaman, S., Barrand, M A., Maskell, L., and Scheper, R J (1998) Multidrug resistance–related transport proteins in isolated human brain

microvessels and in cells cultured from these isolates J Neurochem 70,

1151–1159.

9 Eisenblatter, T., Huwel, S., and Galla, H J (2003) Characterisation

of the brain multidrug resistance protein (BMDP/ABCG2/BCRP)

expressed at the blood–brain barrier Brain Res 971, 221–231 The first

paper to report a porcine homolog of BCRP highly expressed in cultured endothelial cells derived from the blood–brain barrier.

CHAPTER 140 Breast Cancer Resistance Protein in Microvessel Endothelium 953

Further ReadingLitman, T., Brangi, M., Hudson, E., Fetsch, P., Abati, A., Ross, D D.,

Miyake, K., Resau, J H., and Bates, S E (2000) The

multidrug-resistant phenotype associated with overexpression of the new ABC

half-transporter, MXR (ABCG2) J Cell Sci 113, 2011–2021.

Capsule BiographyHiran C Cooray is in the final year of his doctorate studying the expres-

sion and putative roles of BCRP in human brain material and in cultured

endothelial cells.

Dr Margery Barrand is a Senior Lecturer in the Department of macology in the University of Cambridge Her group has strong research interests in multidrug transporters and, in particular, transport systems at the blood–brain barrier.

Phar-C HAPTER 141

Tumor versus Normal Microvasculature

Moritz A Konerding

Johannes Gutenberg-Universität Mainz, Germany

injected into the vascularity either systemically or locally.After polymerization of the casting medium the tissue iscorroded in an alkaline solution After drying and mountingthe vessel system replicas can be studied in detail Intravitalmicroscopy, especially after injection of fluorescent dyes, iseffective in the observation of two-dimensional vascularnetworks as given in most angiogenesis assay systems Thepossibility of examining the vasculature in the time course

is the biggest advantage; the limited resolution the majorshortcoming Classical injection methods with light micro-scopic evaluation are time consuming, require laboriousreconstruction work, and are still of limited value, unlesssophisticated computerized techniques are used The variety

of morphological techniques for assessing microvascularstructure from light to confocal scanning and transmissionmicroscopy was increased significantly by immunostaining,allowing for further structure–function correlations

Structural Differences between Tumor and

Normal Microvasculature

Normal microvasculature is hierarchically organized.The capillary network branches from arterioles with contin-uous or metarterioles with discontinuous layers of smoothmuscle cells In both cases precapillary sphincters may reg-ulate blood flow Blood flow regulation is also facilitated byarteriovenous anastomoses The capillaries are usually made

up by a continuous layer of endothelial cells that are joinedtogether by tight junctions In some organs, such as in parts

of the gut and in endocrine glands, fenestrated endotheliumprevails Pericytes originating from mesenchymal cells and a basement membrane regularly embrace continuousendothelium-type capillaries, whereas pericytes are rarelyfound in fenestrated capillaries Gaps between endothelialcells are seen only in discontinuous, sinusoidal vessels in the

Introduction and Definitions

Therapeutic concepts in oncology such as antiangiogenic

or antivascular approaches have reawakened the scientific

community’s interest in comparative microvascular

anatomy and biology Microvascularity and angiogenesis

play a significant role during normal growth, in

physiologi-cal conditions, and in a variety of pathologiphysiologi-cal conditions

such as inflammation, diabetes, macular degeneration, and

wound healing Angiogenesis is thus, not a specific

phe-nomenon in tumors, but instead an integral element of

numerous different normal and pathological conditions

After the short early embryonic phase of primary

angio-genesis, which comprises the formation of poorly

differ-entiated vessels composed of endothelial cells derived

from angioblasts of blood islands in the extraembryonal

mesoderm of the yolk sac, all further vascular growth in

physiological and pathological conditions is summarized as

secondary angiogenesis In secondary angiogenesis

forma-tion of new vessel segments can be accomplished either by

vessel sprouting, which requires endothelial cell mitoses, or

by intussusceptive growth without need for endothelial cell

mitoses Primary angiogenesis never works without mitoses

When describing microvessel morphology, we have to

discern between structure and architecture, both of which

determine functional properties in terms of blood flow

Structure in morphological terms means vessel wall

build-ing, cellular differentiation, cell shape, cell contact structure

and cell surface differentiation, and organelle and

cytoskele-ton contents Microvessel architecture encompasses pattern

formation, vessel course, density, and all parameters

defin-ing the 3D arrangement of the microvascular unit

Microvessel architecture can be studied best by means of

scanning electron microscopy of corrosion casts, which

allows also for 3D imaging, reconstruction, and quantitative

analysis For microvascular corrosion casting a resin is

Copyright © 2006, Elsevier Science (USA).

956 P ART III Pathology

spleen, liver, and bone marrow In these vessels basement

membranes are missing

Tumor microvasculature is characterized by a lack of

hierarchy and differentiation Even large caliber vessels are

mainly composed of no more components than an

endothe-lium and a basement membrane (Figure 1) Because of the

usually higher interstitial pressure the endothelium may

decrease in height and appear flattened Only sprout

forma-tions and early vessel forms, such as in the tumor invasion

front, reveal an organelle-rich and comparatively high

endothelium (Figure 1b,c) Villus-like protrusions of the

luminal surfaces indicate the young age of these vessels

Abluminal protrusions of cell ramifications, again, indicate

active migration and new sprouting The thinning of the

endothelium sets in with further stabilization and increase in

the luminal diameter However, only a part of the early

forms persists: Many early forms undergo degeneration and

destruction just like those sprouts that did not fuse with

oth-ers Until now no reliable morphometric data have been

available on the fate of sprouts and early forms We have the

impression that at least two thirds of all formed sprouts will

not differentiate into early forms or established vessels

Fenestrations shut by diaphragms are rarely seen (Figure

1d), whereas discontinuities or gaps that allow for

hemor-rhage and facilitate permeability are common features

irre-spective of the origin of the tumor (Figure 1e) Cell contacts

are usually poorly differentiated, and no complex contact

structures exist even in well-established vessels and late

forms Pericytes are frequently missing

Comparisons of the structural features of different human

sarcomas, melanomas, and carcinomas xenografted onto

nude mice as well as a variety of human primary tumors

(colorectal, renal, and larynx carcinomas) did not show

tumor cell line or tumor entity specific cell structure

pat-tern This might well have consequences for targeting

approaches In general, the number of intracytoplasmic

con-tractile filaments is reduced, which contradicts any possible

blood flow regulation by pericyte or endothelial cell

con-tractility Blood flow is regulated primarily on the level of

preexisting arteries and arterioles nourishing the tumor,

resulting in heterogeneous blood flow Apart from

incorpo-rated vessels no clearly demarcated arterial sphincters are

visible within the newly formed tumor vascular network

Autonomous nerves are also detectable only in the tumor

periphery as incorporated nerve fibers Thus, low structural

stability is accompanied by lacking regulation

The difficulties frequently seen when using

panendothe-lial cell markers such as anti-CD31 or anti-FVIII for

label-ing tumor microvessels indicate a functional and structural

heterogeneity of lumen-confining cells in tumors This is

at least in part due to the participation of nonendothelial

cells in the vessel wall formation Newly formed vessels

composed of cells with extremely different cell organelle

contents indicate the involvement of different cell types

(Figure 1d) Sometimes “endothelial cells” could be

observed that engulfed bundles of collagen fibers

It has been well known for 30 years that pericytes may

be formed not only by the angioblastic pathway, but also

from mesenchymal cells Likewise, since then evidence hasaccumulated that tumor cells themselves may be involved invessel wall building Interstitial plasma flow may take place

in channels completely lined by nonendothelial cells (Figure

1f), for which the term vascular mimicry was recently

In tumors, the organ- and tissue specific vascular tecture is not retained but is replaced by newly formed vessels without significant hierarchy (Figure 2e–h) Vesseldensities may vary considerably; the highest vessel densitiesare usually found in the periphery within the invasion front.Vessel density within “hot spots,” that is, densely vascular-ized areas, may be even higher than in the autochthonoustissue In desmoplasias next to the neoplastic tissue, sprout-ing, dilatation, and structural adaptation may change theoriginal architecture of the preexisting vessels as result ofthe release of proangiogenic factors

archi-Common features of human primary and of experimentaltumor vascularities—irrespective of origin, size, and growthbehavior—are missing hierarchy, the formation of large-caliber sinusoidal vessels (Figure 2e, f), and markedlyexpressed vessel density heterogeneity (Figure 2e, g) Thediameters within individual tumor vessels vary significantly.Sinusoids originating from and draining into venous vesselsincrease vessel densities, but do not contribute to nutritiveblood flow Capillary elongations by more than tenfoldexplain the low intratumoral pO 2 Vessel compressions andblind ends are found close together

Tortuous courses and elongated sinusoidal vessels mayoccur also in other forms of secondary angiogenesis; however, in wound healing and chronic inflammation a realremodeling of the newly formed vasculature takes placewith differentiation into arteries and veins and elimination

of ineffective vessel segments

CHAPTER 141 Tumor versus Normal Microvasculature 957

Figure 1 Transmission electron microscopy of tumor vascularity (a) Established sinusoidal vessel (late

form) of a renal cell carcinoma transplanted under murine renal capsule with flattened endothelium (arrows)

without medial layer Note the edema (e) within some endothelial cell protrusions (b) Intercellular bud

forma-tion through endothelial cell migraforma-tion in a human melanoma xenografted onto nude mice The eccentrically located perikaryon of cell 2 and a pseudopodium of cell 1 veer out from the endothelial structure, forming a new

lumen, which is not fully connected to the primary lumen (c) Early form in a melanoma Note the height of the endothelium and the unusual overlapping of the three lumen-confining cells (d) Partial compression of an early

form by extravascular tumor cells (tc) next to a lumen-confining cell (1), which results in a thinning of the

endothelium (Inset) Fenestrated endothelium is superimposed by continuous endothelium (e) Structural defects

in sinusoidal vessel of a xenografted squamous cell carcinoma with evasions The endothelial cells (left) show

some bleb formation indicative of cytoskeleton damage, whereas the pericytes appear normal (f ) Interstitial

channel (ic) lined completely by hypoxic tumor cells in a xenografted spindle cell sarcoma All bars = 5 mm.

958 P ART III Pathology

CHAPTER 141 Tumor versus Normal Microvasculature 959

Last, it should be pointed that the features seen in mary tumor vascularity can be seen qualitatively as well inprecancerous lesions—although to a lesser extent—longbefore the transformation to the malignant phenotype

pri-Glossary

Microvascular corrosion casting: Replication of vessel systems by

injection of low-viscosity casting media and digestion of all surrounding tissues, allowing for scanning electron microscopic examination.

Microvascular unit: Capillary bed segment fed by an individual

metarteriole.

Tumor sinusoids: Relatively undifferentiated vessels with capillary

wall building and increased diameter.

Vascular mimicry: Nonendothelial cells lining vessels or perfused

channels.

Further ReadingBurri, P H., and Tarek, M R (1990) A novel mechanism of capillary

growth in the rat pulmonary microcirculation Anat Rec 228, 35–45.

The first paper describing intussusceptive angiogenesis.

Cliff, W J (1981) Endothelial structure and ultrastructure during growth,

development and aging In Structure and Function of the Circulation,

C J Schwartz and N T Werthessen, eds., Vol 2, pp 695–718 New York: Plenum Press.

Hammersen, F., Endrich, B., and Messmer, K (1985) The fine structure of tumor blood vessels I Participation of nonendothelial cells in tumor

angiogenesis Int J Microcirc Clin Exp 4, 31–43 The first article

proving the involvement of tumor cells in angiogenesis.

Konerding, M A., Fait, E., and Gaumann, A (2001) 3D microvascular architecture of pre-cancerous lesions and invasive carcinomas of the

colon Br J Cancer 84, 1354–1362.

McDonald, D M., and Choyke, P L (2003) Imaging of angiogenesis:

From microscope to clinic Nat Med 9, 713–725 Comprehensive

review.

Capsule Biography

Dr Konerding has been Professor of Anatomy at the Institute of Anatomy in Mainz, Germany, since 1993 His group focuses on secondary angiogenesis and antiangiogenesis as well as on clinical anatomy His work

is mainly supported by grants of the DFG and EC.

Frequently xenografted tumors induce the formation of a

vascular envelope of sprouting and preexistent host vessels

embedded in connective tissue, in which a certain hierarchy

is retained (Figure 2h) This is not true for human primary

tumors since the invasive growth prevents the formation of

such a vascular and/or fibrous tissue capsule

Despite the common features, which are expressed to

different extents, the architecture of tumor vasculature is

tumor-type specific This is in contrast to the structure of the

tumor vascular cells Morphometry of parameters

determin-ing the architecture of the microvascular unit such as

inter-vessel and interbranch distances as well as diameter and

variability of diameter prove that individual tumor entities

express characteristic vascular patterns The inherent

archi-tecture of the tumor seems to be primarily determined by the

tumor cells themselves; experiments involving transfected

tumor cells with different capacity to produce and release

FGFII showed significant differences in tumor growth, but

not in microvascular architecture

Figure 2 Scanning electron microscopy of normal (a–d) and tumor

microvascular corrosion casts (e–h) (a) The human colonic mucosal

capil-lary plexus (c) is arranged in a honeycomb pattern and shows

numer-ous intercapillary connections The supplying arterioles (a) and draining

veins (v) take a straight course from the underlying submucosal vessels

(b) Mouse kidney vessels with cortical vessels and glomeruli (lower left

and inset) and parallel-oriented medullary vessels (upper part) (c)

Subcu-taneous vessel plexus on the fascia with undulating vessel courses

(d) Hairpin loops of skin capillaries oriented along the dermal papillae

(sheep ear skin) (e) Loss of original vessel architecture in human colonic

carcinoma (pT3, pNo, pMx; G2) Note the heterogeneous distribution of

the sinusoids and the loss of hierarchy (f) Murine RenCa-tumor sinusoids

with varying diameters and numerous blind ends (g) Xenografted sarcoma

with compressed main veins (v) and avascular areas (+) next to hot spots.

(h) Tumor vascular envelope in a xenografted squamous cell carcinoma.

C HAPTER 142

Endothelial Cell Heterogeneity

Targeting the Tumor Vasculature

Robert R Langley

Department of Cancer Biology, The University of Texas M.D Anderson Cancer Center, Houston, Texas

malignant cell populations embedded within the underlyingtissue parenchyma However, although a number of angio-genic inhibitors have advanced from the laboratory into theclinical setting, their success in the patient population has, todate, been less than remarkable One potential explanationfor the limited therapeutic response is that an extensive body

of evidence clearly indicates that the microcirculation of ferent tissues exhibits diversity at the structural, molecular,and functional levels Determining which angiogenic factorsare specific for the vessels associated with a given tumor iscentral for designing appropriate therapeutic regimes.Emerging evidence indicates that hematopoietic stemcells (HSCs) and endothelial progenitor cells (EPCs) thatoriginate from either the bone marrow or other tissues maycontribute to the neovascularization of tumors HSCs andEPCs have also been found in newly developed blood vessels associated with wound healing, limb ischemia, atherosclerosis, and post–myocardial infarction Prelimi-nary experimental studies indicate that the homing proper-ties of precursor cells may be exploited to produce therapy

dif-of tumors Considerable effort is now directed toward mining the factors that mobilize and direct these cells tosites of neovascularization

deter-The extent to which tumors are dependent on the cess of angiogenesis varies among different types of tumors Some neoplasms, such as certain nonsmall-cell lung tumors, have been shown to meet their metabolicrequirements by proliferating along the surface of preexist-ing blood vessels and, thus, are independent of angiogene-sis Therefore, this form of tumor growth could beimpervious to pharmacologic interventions that are directed

pro-The growth and dissemination of malignant tumors are

dependent on a blood supply In response to increasing

metabolic pressures, tumors enhance their vascular supply

by inducing resident endothelial cells to form new vascular

networks (angiogenesis), by mobilizing populations of

endothelial precursor cells to the tumor site

(vasculogene-sis), or by modifying the structural architecture of

preexist-ing blood vessels (vascular remodelpreexist-ing) Understandpreexist-ing

the molecular mechanisms that regulate these complex

processes in different organs is essential for developing

antivascular therapy

Introduction

In his 1945 report examining the vascular response to

tumor implantation, Algire concluded “an outstanding

char-acteristic of the tumor cell is its capacity to elicit

continu-ously the growth of new capillary endothelium from the

host” [1] Some 25 years later, enough supportive evidence

had accumulated to advance the hypothesis that the

progres-sive growth of tumors is, indeed, dependent upon the

induc-tion of angiogenesis Since that time, considerable effort has

been extended toward defining the molecular mechanisms

that regulate tumor neovascularization and characterizing

the phenotype of tumor blood vessels Targeting of

tumor-associated endothelium is an attractive approach to control

tumor growth in that the endothelial component of the tumor

is considered to be genetically stable and, therefore, not

prone to develop resistance to therapy Moreover, the tumor

endothelium provides an accessible target as compared to

Copyright © 2006, Elsevier Science (USA).

962 P ART III Pathology

against dividing endothelial cell populations Perhaps a

more effective approach for the management of this type of

tumor growth would be to deliver therapy in a regional

man-ner by targeting tissue-specific receptors that are present on

the vascular endothelium and, thereby, alleviate much of the

systemic toxicity associated with chemotherapy

Angiogenesis

Angiogenesis refers to the development of new blood

vessels from the existing vascular bed During embryonic

development, angiogenesis provides ectoderm- and

mesoderm-derived tissues, including the brain and kidney,

with a vascular supply In the adult, angiogenic events occur

less frequently and are primarily restricted to tightly

regu-lated processes that occur during the female reproductive

cycle (formation of corpus luteum and placenta) and in

response to wound healing However, angiogenesis does

play a prominent role in the progression of several

patho-logic conditions including arthritis, diabetic retinopathy,

psoriasis, and tumor growth In contrast to the transient

highly regulated vascular development that occurs during

normal tissue growth, the angiogenesis associated with the

aforementioned disease states is persistent and unrelenting

The onset of angiogenesis is presumed to be the result of

an imbalance between inhibitor and stimulator molecules

Normal tissues are exposed to an excess of inhibitor

mole-cules and, consequently, the vascular endothelium remains

in a quiescent state Studies examining the kinetics of

endothelial cell proliferation in several normal tissues

esti-mate that the turnover time of endothelial cells can be

meas-ured in years Tumors may disrupt the equilibrium between

inhibitor and stimulator molecules, however, by inducing

alterations in the local microenvironment, by

downregu-lating tumor-suppressor genes, or through the activation of

oncogenes Microenvironmental changes appear to result

from increasing metabolic demands by an expanding tumor

cell population that are not satisfactorily met by available

blood delivery Insufficient blood flow, in turn, leaves

resi-dent tumor cells exposed to an inhospitable milieu that is

characteristically devoid of oxygen and nutrients and rich in

metabolic waste products For some tumors, such as

carci-nomas of the head and neck, the decline in oxygen tension

functions as a selection pressure that leads to the

prolifera-tion of cells with enhanced metastatic potential However,

for tumor cells subjected to severe hypoxia, the anaerobic

environment may have a detrimental effect on tumor cell

survival Indeed, Fidler and colleagues have demonstrated

that if a tumor cell resides more than 170mm from a

vascu-lar supply, there is a strong probability that this cell’s destiny

is one of death (Figure 1)

Tumor Blood Flow

Studies conducted in experimental animal tumors

indi-cate that insufficient blood delivery is a common feature of

a number of tumor types For some, the reduction in bloodflow can be profound when compared to measurementsobtained in adjacent, nonaffected tissue For example, meas-urements of tumor blood flow in ovarian implants show thatblood flow rates in these tumors are some 50 times less thanthat observed in normal ovarian tissue In rodent models,blood flow to tumors continues to decline as the size of thetumor increases, irrespective of the intensity of the angio-genic response Many tumors display a progressive rarefac-

tion of the vascular bed (i.e., a decrease in the number of

patent vessels per gram of tissue) Because the rate of vascularization is considerably reduced in comparison totumor cell proliferation, the vascular space becomes a pro-gressively smaller component of the total mass To deter-mine how the vascular surface area of experimental tumorscompared to that of normal tissues, we injected a radio-labeled monoclonal antibody directed against CD31 into thesystemic circulation of mice harboring subcutaneous tumorsand compared antibody binding on tumor vessels to uptake

neo-on vessels supplying other tissues (Figure 2) The resultsshow that the tumor vascular surface area is a fraction of thatfound in normal tissues

Hypoxia Inducible Factor

Recent studies indicate that many tumors rely on thesame hypoxia-sensing mechanism utilized by normal cellsduring conditions of low oxygen availability to activategenes associated with angiogenesis Specifically, hypoxia

inducible factor-1 (HIF-1) has been shown by in situ

hybridization to be expressed in hypoxic regions of tumors,and overexpression has been documented in several primaryhuman cancers and their metastases HIF is composed of an

ab heterodimer in which the a subunits are inducible by

hypoxia, whereas the b subunits are constitutively expressed

nuclear proteins Declining oxygen tension stabilizes

HIF-1a and promotes its accumulation in the cell nucleus, where

it heterodimerizes with HIF-1b and initiates the

transcrip-tion of a number of genes implicated in angiogenesis, sion, and metastasis

inva-One of the most intensely studied genes targeted by HIFactivation is vascular endothelial cell growth factor/vascularpermeability factor (VEGF/VPF) Although originally char-acterized based on its ability to induce protein extravasationfrom tumor vessels, VEGF has since been determined to beessential for the establishment of new blood vessels that are associated with embryonic development and maturation

of the ovarian follicle, as well as a number of pathologicprocesses At present, five VEGF ligands (VEGFs A–E)have been identified that possess binding potential for three distinct VEGF receptors: VEGFR1 (Flt-1), VEGFR2(KDR/Flk-1), and VEGFR3 (Flt-4) Of the VEGF ligands,the various VEGFA isoforms appear to be key mediators ofnew blood vessel growth, whereas VEGFC and VEGFDhave been implicated in lymphatic angiogenesis

VEGF can elicit cell migration, protease production, and proliferation necessary for angiogenesis VEGF also

CHAPTER 142 Endothelial Cell Heterogeneity 963

functions as a survival factor for endothelial cells by upregulating the phosphatidylinositol-3 OH kinase/Akt signal transduction pathway and enhancing the expression

of the antiapoptotic proteins Bcl-2 and A1 Overexpression

of VEGF is a hallmark of a variety of human tumors, andinhibition of Flk-1 signaling using several differentapproaches can significantly impair tumor growth in exper-imental models Considering the important role that VEGFand its receptors play in the neovascularization process ofmalignant tumors, it is not surprising that this signaling cas-cade is at the forefront of antiangiogenic research efforts.Also included among the genes induced by HIF-1a acti-

vation are those that encode for the polypeptide chains ofplatelet-derived growth factor (PDGF) To date, four PDGFpolypeptide chains have been identified that combine toform five PDGF isoforms: PDGF-AA, -AB, -BB, -CC, and-DD These isoforms exert their effects by activating twoprotein tyrosine kinase receptors, denoted the a-receptor

and b-receptor Although each of the different PDGF

iso-forms has been shown to be capable of contributing to tumorneovascularization, their specific role in this process appears

to be dependent upon the location of the tumor Indeed, theresults obtained from studies examining PDGF-R signalingcascades in different experimental tumors underscore the

Figure 1 Location of dividing and apoptotic tumor cells in relation to blood vessels in brain metastases

(a) Autochthonous human lung-cancer brain metastases BrdU-positive cells (cells engaged in cell division)

are stained red and are located within 75mm of the nearest blood vessel (arrow) (b) Autochthonous human

lung-cancer brain metastasis TUNEL-positive (apoptotic) cells stained bright green and are positioned 160–170 mm

from the nearest blood vessel The vessel is marked in red (c) Human KM12C metastasis in the brain of nude mice BrdU-positive nuclei stained red Arrow points to blood vessel stained for CD31 (brown) (d) Human

KM12C metastasis in the brain of nude mice CD31 staining (red) identified blood vessels and bright green nuclei were TUNEL-positive Bar = 100 mm Reproduced with permission from Fidler et al (2002) Lancet

Oncology 3, 53–57 (see color insert)

CD31 Expression 15.1

3.5 3.0 0.5 0.6

Figure 2 CD31 expression levels in lung, heart, small intestine, tumor,

and skin of C57Bl/6 mice Radiolabeled monoclonal antibody (mAb)

directed against CD31 was injected intravenously into mice harboring

subcutaneous murine RM1 tumors, and binding on tumor blood vessels

was compared to values obtained from blood vessels supplying normal

tissues Values are expressed as mg mAb/g tissue This technique has

been shown to provide a reliable index of microvascular density in

differ-ent tissues Modified from Langley et al (1999) Am J Physiol 277,

H1156–H1166.

964 P ART III Pathology

diverse roles that tyrosine kinase receptors may assume in

different vascular beds For example, in pancreatic tumors,

PDGF signaling has been shown to provide integrity to

developing vascular networks by recruiting mural cells to

support the immature vessel wall In gliomas, PDGF BB

promotes the neovascularization of cerebral lesions, in part,

by stimulating the release of VEGF from tumor-associated

endothelium Tumors growing in the subcutaneous space,

however, rely on PDGF BB to regulate the level of

inter-stitial fluid pressure within the tumor In the bone

microen-vironment, activation of PDGF-R on tumor-associated

vessels has been shown to be essential for progression of

androgen-independent prostate tumors With the advent of

small-molecule inhibitors that selectively target PDGF-R,

such as STI571 (Glivec), much attention has been directed

toward determining whether other malignancies are

criti-cally dependent on this tyrosine kinase

Role of the Microenvironment and Angiogenesis

The anatomical location of a tumor may be an important

determinant in regulating the phenotype of its vasculature

Unfortunately, many preclinical studies neglect to account

for influence of the microenvironment on tumor growth and

vascularization by utilizing models in which different types

of tumors are all implanted into the subcutaneous space

One may argue that, with the exception of those tumors

originating from or metastasizing to skin, subcutaneously

implanted tumors provide relatively little useful information

regarding antiangiogenic intervention This is perhaps true

in that this model evaluates only one type of endothelial cell

(dermal) and, equally important, because frequently the

tumor is located in an ectopic site Recently, Blouw and

coworkers [2] utilized an elegant strategy involving

HIF-1a-deficient transformed astrocytes to illustrate how

HIF-1a can act as either a negative or a positive regulator of

astrocytoma progression, depending on the

microenviron-ment in which the tumor is located Specifically, the loss of

HIF-1a impaired the growth of astrocytomas in the

sub-cutis, but not in the brain This finding was attributed to the

inability of HIF-1a-deficient cells to mobilize VEGF that,

consequently, blocked their capacity to recruit new vessels

in the inherently vessel-poor subcutaneous compartment

(see Figure 2), whereas the same HIF-1a-deficient cells

exploited the preexisting rich vascular networks of the brain

The results of this study are noteworthy in several respects

First, had the examination been conducted solely within the

confines of the subcutaneous space, one may have reached

the conclusion, albeit an incorrect one, that targeting

HIF-1a may be an effective approach to control the growth

of glioblastomas In addition, recently there has been an

appreciable interest in the development of pharmaceuticals

capable of targeting HIF-1a in tumors If this interest is

extended to the production phase, it would be very

informa-tive to determine whether agents directed against HIF-1a

possess efficacy in the treatment of cancer metastases

located in lung, brain, bone, or liver tissue, which are vascularized tissues

well-Studies evaluating angiogenesis in human tumors alsoconclude that there is considerable heterogeneity regardingthe intensity of endothelial cell proliferation among dif-ferent types of tumors Eberhard and colleagues utilized adouble-labeling immunohistochemical approach to measureproliferating endothelial cells in a broad panel of humantumors The data from this study indicate that glioblastomasand renal cell carcinomas possess the highest number ofcapillary beds with evidence of endothelial cell proliferation(9.6% and 9.4%, respectively), whereas lung and prostatetumors possess the fewest (2.6% and 2.0%, respectively).These results caution against accepting rapidly dividingmurine tumors that grow exponentially over a period of afew weeks as the model for the more measured expansionthat takes place in human malignancies

Recent improvements in the ability to selectively isolateand propagate endothelial cells from different organs cannow permit more detailed examinations into the tissue-specific factors that regulate angiogenesis Indeed, our laboratory has applied a series of multicolor flow cytometryselection strategies that target inducible endothelial cell

adhesion molecules to H-2K b -tsA58 murine tissues in order

to generate microvascular endothelial cell lines from a

num-ber of different organs [3] Cells derived from H-2K b -tsA58

mice all harbor a temperature-sensitive SV40 large T gen and, thus, permit detailed molecular examinations onboth activated and differentiated phenotypes of endothelialcells Preliminary studies conducted on several endothelialcell lines reveal that each possesses distinct growth factorprofiles For example, we have noted that for uterine-derived endothelial cells, the most potent endothelial cellmitogen is epidermal growth factor (EGF) However, forendothelial cells originating from bone tissue, basic fibrob-last growth factor (bFGF) elicits the most robust increase inproliferation (unpublished data) Similarly, a recent studyconducted on microvascular endothelial cells from thehuman intestine has determined that interleukin-8 can pro-mote both cell division and migration cDNA expressionarrays have also been utilized to construct gene expressionprofiles on human endothelial cells obtained from brain,lung, and lymphatic tissue Additional reports have providedevidence that some tissues elaborate organ-restrictedendothelial cell mitogens and that, moreover, these growthfactors can be detected in tumors arising from theseanatomic regions

anti-Targeting Angiogenesis in Tumors

Kerbel and Folkman have proposed a classificationscheme that places agents that target tumor neovasculariza-tion into two classes Direct inhibitors, such as endostatin,angiostatin, and vitaxin, function by preventing endothelialcell proliferation, migration, and initiation of antiapoptoticprograms in response to various angiogenic proteins Indi-

CHAPTER 142 Endothelial Cell Heterogeneity 965

rect inhibitors, on the other hand, block production of

pro-tein products (bFGF, VEGF) from tumor cells or interfere

with endothelial cell receptor tyrosine kinase signaling

cascades Perhaps one of the most valuable lessons learned

from the preclinical evaluations of these agents is that their

effectiveness at limiting tumor growth is much more

pronounced if they are administered in combination with

chemotherapy or radiotherapy

One of the most active areas of antiangiogenic

investiga-tion has centered on the VEGF system Although several of

the small-molecule inhibitors of VEGFR2, such as SU5416

and SU6668, demonstrated efficacy in limiting human

tumor xenograft growth in mice, these agents were found to

produce unacceptable toxicity profiles in humans that has

halted their development Adverse side effects observed in

patients receiving these agents have included pulmonary

emboli, myocardial infarction, and cerebrovascular events

rhuMAB VEGF (avastin) is a recombinant humanized

monoclonal antibody to VEGF that has undergone initial

evaluation in the treatment in two different types of human

tumors In a Phase II study conducted with patients

harbor-ing stage IIIB/IV nonsmall-cell lung cancer, individuals

were randomized to standard chemotherapy with

carbo-platin and paclitaxel alone or in combination with either 7.5

or 15 mg/kg rhuMAB Although the response rates were

slightly increased in the group receiving chemotherapy and

high-dose rhuMAB, median survival was only improved by

3 months in this group A randomized trial employing

rhuMAB has also been completed in a group of 116 patients

with metastatic clear-cell renal carcinoma [4] Interestingly,

in this study design, patients were assigned to groups

receiving placebo or 3 or 10 mg/kg rhuMAB that was

administered every 2 weeks Clear-cell renal carcinoma is

perhaps an ideal candidate for VEGF therapy in that the von

Hippel–Lindau tumor suppressor gene is usually mutated

in this cancer and, as a result, VEGF is overproduced

because of a mechanism involving HIF1-a Unfortunately,

monotherapy with rhuMAB failed to demonstrate an

increase in survival in this patient population

Many preclinical studies are based on rapidly growing

subcutaneous neoplasms Moreover, treatment of human

tumor xenograft models is usually initiated at early time

points This is in sharp contrast to the clinical setting, where

patients usually present with advanced disease and are

eligi-ble to receive antiangiogenic therapies only after failing one

or more conventional treatment modalities In addition,

because tumors are made up of heterogeneous populations

of cells, treatment with only one angiogenic agent may

result in the selective expansion of cells expressing different

proangiogenic proteins

Although there remain many unanswered questions

regarding which angiogenic inhibitor is appropriate for a

given tumor, there have been some recent advances in

deter-mining which signaling cascades are operative in some of

the preferred sites of metastasis The bone is the fourth most

common site of tumor metastasis, and recent estimates

pre-dict that approximately 350,000 cancer patients die each

year with evidence of tumor in the skeleton Consequently,much effort has been extended toward identifying thosecomponents in the bone microenvironment that supporttumor growth Recently, members of our laboratory evalu-ated the efficacy of STI571, a small-molecule inhibitor ofthe PDGF-R tyrosine kinase, in a human tumor xenograftmodel of androgen-independent prostate cancer implantedinto the bone [5] In that model, inhibition of PDGF-R acti-vation with STI571, when administered in combination withtaxotere, produced a profound reduction in both tumor massand lymphatic metastases It was determined that the effec-tiveness of the combination therapy was directly related toapoptosis of tumor-associated endothelial cells The pro-mising results obtained in this study have prompted the initiation of a clinical trial to evaluate the efficacy of thistreatment regime in men with refractory prostate cancerbone metastasis In addition, these findings may have impor-tant implications for other cancers that metastasize to bonesuch as renal, breast, and lung tumors However, it will benecessary to determine whether these tumors are operatingthrough a PDGF-dependent pathway Indeed, a recent reporthas shown that renal cell carcinoma may preferentiallyexploit EGF signaling cascades in the bone microenviron-ment in order to promote their vascularization and ensuretheir growth [6]

Paracrine EGF signaling from tumor cells to cular endothelial cells has also been shown to be an impor-tant component for the growth of pancreatic metastases inthe liver Using either a monoclonal antibody blocking strat-egy (C225) or small molecule inhibitor of EGFR signaling

microvas-in combmicrovas-ination with gemcitabmicrovas-ine, Bruns and coworkerswere able to dramatically reduce both primary pancreatictumor growth and liver metastatic burden In both tumormodels, combination therapy was associated with down-regulation of VEGF in the tumor region and apoptosis

of tumor-associated endothelial cells Independent studieshave also identified VEGF as an important proangiogeniccytokine in the liver Reports examining liver regeneration

in rats following partial hepatectomy have shown thatVEGF plays a critical role in the revascularization process

by stimulating the proliferation of hepatic sinusoidalendothelial cells VEGF also appears to play a role in coloncancer liver metastasis, as blockade of either VEGF orVEGFR2 in a human tumor xenograft model results in theinduction of both tumor cell and endothelial cell apoptosis[7]

One of the most elusive organs for targeted treatment ofmetastasis is the brain Cerebral blood vessels possesshighly resistant tight junctions that are further reinforced bythe end feet processes of astrocytes In addition, this highlyspecialized vascular network is also enriched with a number

of transport proteins, such as the multidrug-resistant proteinP-glycoprotein, which restrict the passage of severalchemotherapeutic agents into the tissue parenchyma In fact, it has been shown that P-glycoprotein can also mediatethe efflux of small-molecule inhibitors such as Glivec.Adding to the complexity of the cerebral vasculature, Fidler

966 P ART III Pathology

and colleagues demonstrated that neovascularization of

metastases in the brain occurs by a mechanism that is

distinct from traditional sprouting angiogenesis In brain

metastasis, blood vessel expansion occurs by the insertion of

dividing endothelial cells into the preexisting blood vessel,

a process that is referred to as angioectasia In addition,

because of the detrimental consequences that ensue upon

cessation of cerebral blood flow, it is likely that there is

considerable redundancy in angiogenic proteins in this

anatomic compartment

Tumor Vasculogenesis

Although most of the examinations into the

neovascular-ization of tumors have focused on angiogenesis, new

evi-dence indicates that hematopoietic stem cells (HSCs) and

endothelial precursor cells (EPCs) may also play an

impor-tant role in this process Asahara et al [8] were the first to

demonstrate the EPCs could be isolated from human

periph-eral blood and incorporate themselves into active areas of

angiogenesis in animal models of ischemia EPCs may

be differentiated from mature circulating endothelial cells

that have been shed from the vascular wall by their

signifi-cantly enhanced capacity to proliferate, and also by their

unique expression of cell-surface markers that include

VEGFR2, AC133, CXCR4, and CD146 Recent studies

are beginning to provide insight into the mechanisms

that facilitate the mobilization of HSCs and EPCs from

the bone marrow and to sites of neovascularization The

recruitment process appears to be initiated by angiogenic

products that are released from tumor cells and lead to the

activation and secretion of matrix metalloproteinase-9

(MMP-9) by hematopoietic and stromal elements in

the bone marrow MMP-9 activation, in turn, leads to

liber-ation of soluble KIT ligand that promotes cell proliferliber-ation

and also directs the transfer of these cells into the peripheral

circulation

Direct evidence that HSCs and EPCs contribute to tumor

neovascularization has come from examinations conducted

in mice that are deficient in Id proteins Mutant mice with

the Id1+/-Id3-/- phenotype possess defective angiogenic

responses and, thus, are unable to support tumor growth

However, when these mice are transplanted with wild-type

bone marrow or HSCs, tumor growth in the subcutaneous

compartment can be restored This process appears to

require cooperation of both VEGFR1 and VEGFR2 in that

treatment of Id1+/-Id3-/-mice with neutralizing antibodies

against both of these receptors results in vascular disruption

and tumor cell death The contribution of HSCs and EPCs to

the tumor vasculature appears to be influenced by the tumor

type, but most reports suggest that these cells represent a

small fraction (6 to 10%) of tumor-associated blood vessels,

and that they arrive at tumor vessels during the initial phase

of malignant growth Despite the relatively low number of

progenitor cells that lend support to tumor vessels, evidence

suggest that their ability to traffic to tumor sites may be

exploited for therapeutic targeting Genetically modifiedendothelial progenitors that were stably transfected withthymidine kinase or the soluble truncated form of VEGFR2have been shown to home to subcutaneous tumors and significantly impair tumor growth Although these initialreports are promising, more studies are needed to determinewhich types of malignant lesions are critically dependentupon progenitor cells

Conclusion

Over the past decade, the understanding of the cellularand molecular mechanisms that tumors utilize to promotetheir blood delivery and facilitate their growth hasadvanced However, therapies that are directed toward thevascular compartment of malignant lesions are still in theearly developmental phase Continued investigations intothe specific factors that regulate angiogenesis in differentorgans, coupled with detailed characterization of the prod-ucts released by tumors in different tissues, should improvetargeted therapy of angiogenesis Similarly, an enhancedunderstanding of the biology of HSCs and EPCs could pro-vide clinicians with a superior method of delivering therapy

to tumors The identification of tissue-specific receptors, onboth normal and tumor-associated endothelial cells, willalso enhance the development of vascular-based therapies.Employing orthotopic tumor models that more closelyapproximate the clinical reality could well facilitate transla-tion of laboratory findings to the clinic

Glossary

Angioectasia: Blood vessel dilation that is due to endothelial cell

divi-sion occurring within the wall of the blood vessel This process is distinct from angiogenesis and has been reported in brain metastasis.

Angiogenesis: Refers to the generation of new vascular networks from

preexisting blood vessels During this process, endothelial cells elaborate proteolytic enzymes, exhibit migratory behavior, and undergo cell division Ultimately, developing capillary sprouts will coalesce to form new vascu- lar structures.

Endothelium: Specialized simple squamous epithelium that lines the

intimal surface of all blood vessels.

AcknowledgmentThe author thanks Dr I J Fidler for helpful discussions and editorial assistance.

References

1 Algire, G H., Chalkley, H W., Legallais, F Y., and Park, H D (1945) Vascular reactions of normal and malignant tumors in vivo I Vascular reactions of mice to wounds and to normal and neoplastic transplants.

J Natl Cancer Inst 6, 73–85.

2 Blouw, B., Song, H., Tihan, T., Bosze, J., Ferrara, N., Gerber, H P., Johnson, R S., and Bergers, G (2003) The hypoxic response of

tumors is dependent on their microenvironment Cancer Cell 4,

133–146.

CHAPTER 142 Endothelial Cell Heterogeneity 967

3 Langley, R R., Ramirez, K M., Tsan, R Z., Van Arsdall, M., Nilsson,

M B., and Fidler, I J (2003) Tissue-specific microvascular

endothe-lial cell lines from H-2K(b)-tsA58 mice for studies of angiogenesis and

metastasis Cancer Res 63, 2971–2976.

4 Yang, J C., Haworth, L., Sherry, R M., Hwu, P., Schwartzentrube,

D J., Topalian, S L., Steinberg, S M., Chen, H.X., and Rosenberg,

S A (2003) A randomized trial of bevacizumab, an anti-vascular

endothelial growth factor antibody, for metastatic renal cancer N.

Engl J Med 349, 427–434.

5 Uehara, H., Kim, S J., Karashima, T., Shepherd, D L., Fan, D., Tsan,

R., Killion, J J., Logothetis, C., Mathew, P., and Fidler, I J (2003).

Effects of blocking platelet-derived growth factor-receptor signaling in

a mouse model of experimental prostate cancer bone metastases J.

Natl Cancer Inst 95, 458–470.

6 Weber, K L., Doucet, M., Price, J E., Baker, C., Kim, S J., and Fidler,

I J (2003) Blockade of epidermal growth factor receptor signaling

leads to inhibition of renal cell carcinoma growth in the bone of nude

mice Cancer Res 63, 2940–2947.

7 Shaheen, R M., Davis, D W., Liu, W., Zebrowski, B K., Wilson,

M R., Bucana, C D., McConkey, D J., McMahon, G., and Ellis, L M.

(1999) Antiangiogenic therapy targeting the tyrosine kinase receptor

for vascular endothelial growth factor receptor inhibits the growth of

colon cancer liver metastasis and induces tumor and endothelial cell

apoptosis Cancer Res 59, 5412–5416.

8 Asahara, T., Murohara, T., Sullivan, A., Silver, M., van der Zee, R., Li,

T., Witzenbichler, B., Schatteman, G., and Isner, J M (1997) Isolation

of putative progenitor endothelial cells for angiogenesis Science 275,

964–967.

Further ReadingArap, W., Pasqualini, R., and Ruoslahti, E (1998) Cancer treatment by tar-

geted drug delivery to tumor vasculature in a mouse model Science

279, 377–380.

Dvorak, H F (2003) How tumors make bad blood vessels and stroma Am.

J Pathol 162, 1747–1757.

Ferrara, N., Gerber, H P., and LeCouter, J (2003) The biology of VEGF

and its receptors Nat Med 9, 669–676.

Fidler, I J., Uano, S., Zhang, R., Fujimaki, T., and Bucana, C D (2002) The seed and soil hypothesis: Vascularisation and brain metastases.

Lancet Oncol 3, 53–57.

Hood, J C., Bednarski, M., Frausto, R., Guccione, S., Reisfeld, R A., Xiang, R., and Cherish, D A (2002) Tumor regression by targeted

gene delivery to the neovasculature Science 296, 2404–2407 The

authors utilize nanosphere-assisted targeting of the neovasculature with mutant Raf-1 to induce apoptosis of tumor-associated endothelium and promote regression of established primary tumors and metastatic lesions.

McIntosh, D P., Tan, X Y., Oh, P., and Schnitzer, J E (2002) Targeting endothelium and its dynamic caveolae for tissue-specific transcytosis in vivo: A pathway to overcome cell barriers to drug and gene delivery.

Proc Natl Acad Sci USA 99, 1996–2001 Study demonstrates that the

molecular composition of lung microvascular caveolae is distinct from caveolae found in other regional circulations and that, moreover, this discriminating feature can be exploited to selectively transport immunotoxin to the lung.

Pugh, C W., and Ratcliffe, P J (2003) Regulation of angiogenesis by

hypoxia: Role of the HIF system Nat Med 9, 677–684.

Raffi, S., Lyden, D., Benezra, R., Hattori, K., and Heissig, B (2002) Vascular and haematopoietic stem cells: Novel targets for anti-

angiogenesis therapy? Nat Rev Cancer 2, 826–835 This important

review discusses contribution of HSCs to tumor vasculature and the therapeutic potential of precursor cells for targeting tumor-associated blood vessels.

Capsule BiographyRobert R Langley received his Ph.D from the Department of Molecu- lar and Cellular Physiology, Louisiana Health Sciences Center, Shreveport, Louisiana, in 1999 He received postdoctoral training in the Department of Cancer Biology at the University of Texas M.D Anderson Cancer Center

in Houston, Texas, from 2000 to 2003 He is currently serving as an tor in the Department of Cancer Biology and examining the contribution of endothelial cells to the metastatic process.

Instruc-C HAPTER 143

A Metronomic Approach

to Antiangiogenesis

Raffaele Longo and Giampietro Gasparini

Division of Medical Oncology, “S Filippo Neri” Hospital, Rome, Italy

Angiogenesis is necessary to sustain the growth of mary tumor and metastases Metronomic schedules aremore effective in targeting tumor “activated” ECs than largesingle high-dose bolus doses followed by long rest periods[2], because intratumor ECs, in contrast to quiescent matureECs of normal adult tissues, proliferate rapidly and are morevulnerable to cytotoxic agents [5] However, the rest periodbetween cycles of conventional chemotherapy permits thesurvival and regrowth of a fraction of ECs, allowing tumorangiogenesis to persist [5] Indeed, other functions such as

pri-EC motility, invasion, and vessel remodeling may beblocked or altered by metronomic chemotherapy [5] Sup-pression of mobilization of bone marrow–derived EC pro-genitors to sites of angiogenesis is another possibility [2, 3].Tumor-cell heterogeneity may allow the coexistence of cellshaving different sensitivities to the same therapeutic agent,and the ability of those cells to shift among the different sen-sitivity compartments over time, with a “resensitization”effect that may best be exploited using metronomicchemotherapy

A further theoretical advantage of metronomicchemotherapy is that it minimizes the toxic effects, allowingcombinations of potentially synergistic selective inhibitors

of angiogenesis [3, 5]

However, a phenomenon that may limit the advantages oflow-dose metronomic or continuous-dose delivery is athreshold effect for drug activity Specifically, there mayexist a minimum concentration of drug below which notumor inhibition takes place [1]

Recently, Miller et al defined the criteria for genic activity of cytotoxic agents: (a) camptothecin analogs,vinca alkaloids, and taxanes are active against ECs at doseslower than those required for tumor-cell cytotoxicity; (b)

antiangio-Introduction

Chemotherapy has been the mainstay of medical

approaches to the treatment of solid neoplasia It is usually

given in the form of bolus infusion at maximum tolerated

doses (MTDs) with the goal of complete tumor kill With the

exception of a few tumors of the adult, such as lymphoid,

germ cell, and some pediatric cancers, however, eradication

of advanced cancer has been elusive, even with high doses

and autologous bone-marrow rescue [1] It was suggested

that the rest periods between each cycle of therapy provide

the tumor endothelium an opportunity to repair the

chemotherapy-induced damages The harsh side effects and

the ultimate failures of conventional chemotherapy fueled

broad investigation of therapeutic alternatives, including

drugs targeting not only tumor cells, but also genetically

stable cells of tumor stroma, such as endothelial cells

(ECs) The emerging paradigm is that patient survival is

not incompatible with tumor persistence A therapy aimed

at making the cancer a chronic disease, with tumor burden

held at the lowest achievable volume, may prove to be a

more appropriate therapeutic strategy for human solid

tumors (Figure 1)

Metronomic Chemotherapy

Based on the results of experimental studies [2, 3],

Hanahan et al [4] proposed the term metronomic

chemo-therapy for schedules of cytotoxic agents given regularly at

subcytotoxic doses and with suppression of the “activated”

endothelium as the principal target (i.e., the

antiangiogene-sis chemotherapy paradigm)

Copyright © 2006, Elsevier Science (USA).

970 P ART III Pathology

cisplatin, cyclophosphamide, methotrexate, and doxorubicin

interfere with EC function without cytotoxic effects; and (c)

purine analogues, cisplatin, and anthracyclines block

spe-cific steps of the angiogenic cascade [5]

Experimental Studies

Recent studies have demonstrated that the antitumor

activity of metronomic schedules of chemotherapy is

medi-ated through an antiangiogenic effect [5]

Browder et al compared cyclophosphamide given by a

metronomic schedule with conventional single, high-dose

chemotherapy [2] The metronomic schedule was more

effective in eradicating Lewis lung carcinoma and L1210

leukemia When the drug was given at the MTD, which

requires long periods of rest to allow bone marrow recovery,

apoptosis of ECs in the tumor microvasculature was

observed, but this damage was largely repaired during therest periods between the cycles [2]

However, if the same drug was administered chronically

on a once a week schedule, without long breaks, at a lowerdose, the repair process was compromised and the anti-angiogenic effects of the drug were enhanced It was alsoobserved that tumors resistant to conventional dosage ofcyclophosphamide responded to the metronomic schedule.Reversal of acquired resistance was attributed to the shifting

of the target from the cancer cell to the genetically stableand sensitive ECs

Similarly, in human orthotopic breast or ectopic coloncancer xenografts in nude or SCID mice, cyclophosphamidegiven PO at low doses through drinking water showed a rel-evant antitumor effect, particularly when used in combina-tion with an anti-VEGF-R2antibody [6]

In an in vivo mouse corneal model, O’Leary et al.showed a significant antiangiogenic activity of camp-

Standard CT Metronomic CT

Angiogenesis inhibition

Indirect cytotoxic effect

Direct cytotoxic effect

Tumor cell Lymphocytes Macrophages

Mast cells Platelets Fibroblasts

Figure 1 Metronomic chemotherapy interferes with tumor microvasculature and the stromal compartment,

in contrast to standard chemotherapy that presents a direct cytotoxic effect (see color insert)

CHAPTER 143 A Metronomic Approach to Antiangiogenesis 971

tothecin analogs at a dose of 359 nmol/kg (210 mg/kg)

delivered over 6 days [5]

Presta et al reported that 6-methylmercaptopurine

ribo-side, a purine analog, inhibits angiogenesis induced by

FGF-2 in in vitro and in vivo rabbit cornea and chorioallantoic

membrane models at topical doses of 100 nmol twice a day

for 10 days, or at the single dose of 25mmol onto the

implants, respectively [5]

Low doses of vinblastine (0.1–1 pmol/L in vitro, and

0.5–1 pmol/L in vivo) had reversible antiangiogenic activity,

without cytotoxicity [5] The highest antiangiogenic dose

was 1 pmol/L, which is equivalent to a dose of 16mg in a

70-kg adult, a much lower dose than currently used in the

clinic

More recently, Klement et al demonstrated that

vinblas-tine at one tenth to one twentieth of the MTD caused a

sig-nificant inhibition of angiogenesis with partial tumor

regression in subcutaneous implanted tumors This effect

was significantly enhanced by combination with an

anti-VEGF R2(flk-1) antibody [3]

In Swiss male nude mice implanted intracranially with

human glioblastoma cells, Bello et al showed that low and

semicontinuous chemotherapy in combination with the

recombinant human PEX (a fragment of matrix

metallopro-teinase-2) was associated with longer survival, marked

decreases in tumor volume, vascularity, and proliferative

index, and increased apoptosis, without major side effects

[7]

Paclitaxel selectively inhibits the proliferation of human

ECs at noncytotoxic concentrations (0.1 to 100 pM) by

blocking the formation of sprouts and tubes in the

three-dimensional fibrin matrix This activity does not affect the

cellular microtubule structure, and the treated cells do not

show G2/M cell cycle arrest and apoptosis, suggesing a

novel, but as yet unknown, mechanism of action [8]

Available data obtained from studies in tumor-bearing

animals specifically aimed to investigate the antiangiogenic

effect of cytotoxic agents have found two patterns of

anti-angiogenic effect [9] The first “type” is defined by an

angio-genesis inhibition due to a direct action on ECs regardless of

the tumor cell line used and observed at dose levels lower

than that required for the cytotoxic effect (e.g., bleomycin,

vinca alkaloids, paclitaxel) The type-2 antiangiogenic

effect is not consistently present within a variety of tumor

cell lines resulting from an antiproliferative effect on

tumor cells or non-ECs of the tumor host In the alginate

tumor angiogenesis model, the type-2 reaction pattern has

been shown for doxorubicin, epirubicin, etoposide, and

5-fluorouracil However, these effects may be dependent on

the scheduling of drug administration [9]

All these preclinical studies present certain important

problems: (i) the experimental model used: The studies with

subcutaneous transplantation of rapidly proliferating tumors

do not reflect the slow growth of spontaneous tumors and

the characteristics of the stromal component of the organ of

origin [5]; (ii) the lack of systematic comparative studies

with conventional schedules; and (iii) the possible

mecha-nisms of acquired resistance have not been adequatelyinvestigated [5]

In addition, the best results were obtained using nations with selective antiangiogenic inhibitors, leading toadditive and/or synergistic effects

combi-In conclusion, before proceeding to clinical trials, morecompelling and appropriate experimental studies on metro-nomic chemotherapy are needed, and, although severalcytotoxic agents affect ECs in vitro, only a few have beenshown to produce substantial antiangiogenic activity invivo, namely, cyclophosphamide, vinblastine, and paclitaxel[2, 3, 5]

Clinical Studies

There is currently no published clinical study in whichmetronomic chemotherapy is prospectively compared withconventional schedules, or that describes appropriate invitro or in vivo methods of monitoring the antiangiogenicactivity [5]

In a Phase I study, Retain et al explored the continuousinfusion of low doses of vinblastine up to 36 weeks andfound that 0.7 mg/mq daily was the most effective dose,associated with a good tolerability and some clinical benefit[5]

Colleoni et al demonstrated the activity of the tion of low oral doses of cyclophosphamide and methotrex-ate in patients with metastatic breast cancer, withoutrelevant side effects Serum VEGF levels measured at 2months were lower than at baseline, with statistically signif-icant reduction only in the subgroup of responding patients[10] The major weakness of the study are the dosagesadministered, being in the range of cytotoxic effects; the lowresponse rate in previously untreated patients; and, finally,the fact that the method of determination of VEGF is notstandardized

combina-Other Phase I–II studies are investigating low, ous oral doses of cytotoxic agents (uracil/tegafur and leucovorin, etoposide, 5-fluorouracil) or continuous intra-venous infusion (5-fluorouracil, idarubicin, methotrexate,irinotecan) resembling an antiangiogenic schedule [5] Fortypercent of the patients with nonsmall-cell lung cancer notresponsive to standard doses of etoposide responded to thesame drug given orally at a much lower single dose, withonly a 1-week break every month [5]

continu-Promising results have been reported with weekly taxanetreatment in breast and ovarian cancer, even in patients withprogressive disease after the same drug given at the MTDevery 3 weeks [11,12] Thus, reversal of an apparent state ofclinical drug resistance could be achieved by altering thedosing and frequency of the drug However, at present, it isunknown if this effect is really related to an antiangiogenicactivity

At our center, we are evaluating the combination ofweekly paclitaxel, at 80 mg/mq, and celecoxib, at 400 mgbid, in patients with nonsmall-cell lung cancer, as second-

972 P ART III Pathology

line chemotherapy The preliminary data regarding both the

tolerability and efficacy are encouraging and in accordance

with the results of other similar ongoing Phase II studies

In a Phase I dose-finding study, we evaluated the

combi-nation of rofecoxib with intravenous weekly irinotecan on

days 1, 8, 15, 22 and infusional 5-fluorouracil at the fixed

dosage of 200 mg/mq/day, as second-line therapy of

metastatic colorectal cancer The dose levels of irinotecan

explored were from 75 to 125 mg/mp Seven of 15 patients

assessable for response obtained a partial response (46%)

with a median duration of 5 months and another six had a

stable disease with a median duration of 5 months The

acute and subacute hematological toxicity was moderate,

and mucosal side effects were less than those expected with

the same regimen without rofecoxib (submitted) A Phase II

study testing the activity of such a schedule is ongoing

Future Directions and Open Questions

There are several theoretical advantages and

opportuni-ties for metronomic chemotherapy (Table I) However, there

are also potential problems and challenges in terms of

appropriate experimental study design and clinical testing

[5]

First, combined metronomic chemotherapy should be

tested by adequate experimental models, such as orthotopic

and metastatic tumors The EC heterogeneity, which also

extends to morphology, function, and response to

antiangio-genic molecules, suggests that agents that affect

angiogene-sis in one organ may not be effective in other sites Tumor

cells implanted into mice usually produce rapidly growing

lesions, which can double in size every few days and

con-tain a high proportion of dividing ECs

Second, human solid tumors are heterogeneous, with

dif-ferent molecular abnormalities, even in the same tumor

histotype [5] Gene expression profiling and cDNA

microarrays may categorize tumors into biologically

homo-geneous subgroups and may be of help to design

individu-ally tailored treatments [5]

Third, the identification of specific surrogate biomarkers

can allow the selection of patients as well as the possibility

of monitoring tumor response “Biological” response

crite-ria should replace the currently used the clinical end pointbased on the objective response

Fourth, experimental results have emphasized the criticalneed for combining metronomic regimens with selectiveantiangiogenic agents The intrinsic elevated sensitivity ofactivated ECs to metronomic chemotherapy, compared withthat of other cells, may not be related to the presence of highconcentrations of EC-specific survival factors, such asVEGF [12] Such combinations may be particularly effec-tive in inducing higher levels of apoptosis in activated ECscoupled with the inhibition of cell proliferation It is possi-ble that the inhibition of ECs proliferation or induction ofapoptosis may not be a direct effect of the drug, but rather

an induced secondary event: e.g., a change in expression ofgenes or proteins that mediate the antiangiogenic effects

In conclusion, certain cytotoxic drugs with genic properties retain a potent antitumor activity when used

antiangio-in a protracted manner at very low concentrations, beantiangio-ingable to affect both the tumoral parenchyma and vascularcompartments In contrast to the concept of “more is better,”

it appears that survival depends more on a cytostatic effect

of chemotherapy than on rapid tumor shrinkage [5]

It is unlikely that metronomic chemotherapy will lead tosignificant clinical benefit if given alone We suggest that

it should be considered a novel approach making feasiblefor long periods the administration of cytotoxic agents

in combination with selective molecular-targeting pounds directed against specific growth factors or blockingangiogenesis

Even though the initial development of these novel bined treatments is in the context of advanced disease, themajor therapeutic benefits are expected in the adjuvant setting or as maintenance therapy

com-Glossary

Angiogenesis: Intratumor formation of new blood vessels from a

pre-existing vascular network.

Anticancer therapy: Conventional cytotoxic chemotherapy.

Metronomic chemotherapy: Chemotherapy given regularly at

subcy-totoxic soles with the “activated” endothelium as principal target.

References

1 Hahnfeldt, P., Folkman, J., and Hlatky, L (2003) Minimizing term tumor burden: The logic for metronomic chemotherapeutic dos-

long-ing and its antiangiogenic basis J Theor Biol 220, 545–554.

2 Browder, T., Butterfield, C E., Kraling, B M., Shi, B., Marshall, B., O’Reilly, M S., and Folkman, J (2000) Antiangiogenic scheduling of chemotherapy improves efficacy against experimental drug-resistant

cancer Cancer Res 60, 1878–1886 The first experimental study

test-ing an antiangiogenic schedultest-ing of chemotherapy The study also demonstrated that metronomic chemotherapy reverses drug resistance

to cyclophosphamide.

3 Klement, G., Baruchel, S., Rak, J R., Man, S., Clark, K., Hiclin, D J., Bohlen, P., and Kerbel, R S (2000) Continuous low-dose therapy with vinblastine and VEGF receptor-2 antibody induces sustained

tumor regression without over toxicity J Clin Invest 105, 15–24 The

first study suggesting that the combined therapy of metronomic

Table I Potential Advantages of Metronomic

over Conventional Schedules of

Chemotherapy.

• The targets are genetically stable ECs

• Activity against both the parenchymal and stromal tumor components

• Enhanced antiangiogenic and proapoptotic activity

• Reduced likelihood of emergence of acquired resistance

• Fewer systemic side effects

• Feasibility of long-term administration

• Possibility of combination with other cytostatic, molecular-targeted

treatments

CHAPTER 143 A Metronomic Approach to Antiangiogenesis 973

chemotherapy with a selective inhibitor of angiogenesis has a

supra-additive antitumoral effect.

4 Hanahan, D., Bergers, G., and Bergsland, E (2000) Less is more,

regularly: Metronomic dosing of cytotoxic drugs can target tumor

angiogenesis in mice J Clin Invest 105, 1045–1047.

5 Gasparini, G (2001) Metronomic scheduling: The future of

chemotherapy? Lancet Oncol 2, 733–740 A review paper detailing

the rationale, the results of experimental and clinical studies, and

chal-lenges of metronomic chemotherapy.

6 Mann, S., Bocci, G., Francia, G., Green, S K., Jothy, S., Hanahan, D.,

Bohlen, P., Hicklin, D J., Bergers, O., and Kerbel, R S (2002)

Anti-tumor effects in mice of low-dose (metronomic) cyclophosphamide

administered continuously through the drinking water Cancer Res 62,

2731–2735.

7 Bello, L., Carrabba, G., Giussani, C., Iucini, V., Cerntti, F., Scaglione,

F., Landre, J., Pluderi, M., Tomei, G., Villani, R., Carrull, R S., Black,

P M., and Bikfalvi, A (2001) Low-dose chemotherapy combined with

an antiangiogenic drug reduces human glioma growth in vivo Cancer

Res 61, 7501–7506.

8 Wang, J., Lou, P., Lesniewski, R., and Henkin, J (2003) Paclitaxel at

ultra low concentrations inhibits angiogenesis without affecting

cellu-lar microtubule assembly Anticancer Drugs 14, 13–19.

9 Schirner, M (2003) Antiangiogenic chemotherapeutic agents Cancer

Metastasis Rev 19, 67–73.

10 Colleoni, M., Rocca, A., Sandri, M T., Zorzino, L., Masci, G., Nole F.,

Perruzzotti, G., Robertson, C., Orlando, L., Cinieri, S., de B., F., Viale

G, and Goldhirsch, A (2002) A Low-dose oral methotrexate and

cyclophosphamide in metastatic breast cancer: Antitumor activity and

correlation with vascular endothelial growth factor levels Ann Oncol.

13, 73–80.

11 Burstein, H J., Manola, J., Younger, J., Parker, L M., Bunnel, C A., Scheib, R., Matrilonis, U A., Garber, J E., Clarke, K D., Shulman,

L N., and Winer, E P (2000) Docetaxel administered on a weekly

basis for metastatic breast cancer J Clin Oncol 18, 1212–1219.

12 Sweeney, C J., Miller, K D., Sissons, S E., Nazaki, S., Heilman,

D K., Shen, J., and Sledge, GWjr (2001) The antiangiogenic property

of docetaxel is synergistic with a recombinant humanized monoclonal antibody against vascular endothelial growth factor or 2-methoxyestra-

diol but antagonized by endothelial growth factors Cancer Res 61,

3369–3372.

13 Klement, G., Mayer, B.Huang, P., Green, S K., Man, S Bohlen, P., Hicklin, D., and Kerbel, R S (2002) Differences in therapeutic indexes of combination metronomic chemotherapy and an anti- VEGFR-2 antibody in multidrug resistant human breast cancer

xenograft Clin Cancer Res 8, 221–232.

Capsule BiographyRaffaele Longo is medical doctor and physician researcher at the Divi- sion of Medical Oncology, “S Filippo Neri” Hospital, Rone (Italy), since March 2002 Ongoing research topics are tumor angiogenesis, Cox-2 inhibitors in solid cancers, and clinical development of new targeted molec- ular drugs.

Professor Gasparini has been Director of the Medical Oncology Unit at the San Filippo Neri Hospital in Rome since 2000 His scientific interests primarily focus on translational research on angiogenesis and molecular- targeted anticancer treatments He is author of 250 publications and a mem- ber of the editorial boards of 15 international oncological journals.

C HAPTER 144

Platelet–Endothelial Interaction in

Tumor Angiogenesis

1,2Philipp C Manegold and 1,3Marc Dellian

1 Institute for Surgical Research, Klinikum Grosshadern, University of Munich, Germany

2 Department of Surgery, Klinikum Grosshadern, University of Munich, Germany

3 Department of Otorhinolaryngology, Head and Neck Surgery, Klinikum Grosshadern, University of Munich, Germany

microvessels Mosaic blood vessels composed of lial cells and nonendothelial cells forming vascular lumenare frequently found [4] High vascular resistance and highblood viscosity result in sluggish blood flow within thetumor microcirculation Anoxic, acidotic areas within thetumor center appear in consequence of redistribution ofblood perfusion to peripheral areas [5]

endothe-Therefore, tumor angiogenesis might provide a thrombotic environment that favors interactions of plateletswith the angiogenic endothelium Platelet–endothelial interactions might affect tumor angiogenesis by release ofangiogenic growth factors from activated platelets Sponta-neous thrombosis, however, might result in tumor necrosis

pro-Physiology of Platelet–Endothelial Interaction

In differentiated tissues, activation and aggregation ofplatelets on the endothelial surface is abrogated by endo-thelial antithrombotic mediators such as nitric oxide (NO)and prostacyclin (PGI) The small amount of continu-ously expressed adhesion molecules on the endothelial surface does not sufficiently provoke platelet–endothelial interactions

As a result of disruption of the endothelial surface andexposure of subendothelial matrix as well as in response toinflammatory cytokines (TNF, IL-1, IFN) and angiogenicgrowth factors (VEGF), platelet–endothelial interactions are mediated by endothelial and platelet-specific adhesionmolecules Platelet interaction with the endothelial surfaceand subendothelial matrix occurs in a sequence of initialshort-timed contact (rolling) followed by adhesion and

Tumor angiogenesis, the growth of new blood vessels

from preexisting blood vessels, is indispensable for tumor

growth and tumor metastasis Activated blood coagulation

has been linked to tumor angiogenesis since clotting factors

were found to promote endothelial cell proliferation Beside

plasmatic clotting factors, platelets may be involved in

angiogenic processes: they release pro- and anti-angiogenic

growth factors upon activation and aggregation and promote

formation of capillary-like structures in vitro [1, 2]

Tumor angiogenesis is a complex process regulated by a

versatile number of mediators [3] Focusing on general

mechanisms, tumor angiogenesis is induced by

endothe-lium-specific growth factors VEGF and FGF, but also by

cytokines (TNF, IL-1) and clotting factors that are released

from tumor cells as well as from macrophages and

fibrob-lasts On preexisting venules growth factors and cytokines

stimulate endothelial cells and induce loosening of

intercel-lular junctions of endothelial cells (PECAM, VE-cadherin)

and smooth muscle cells Therefore, angiogenic

endothe-lium is highly permeable, leading to extravasation of plasma

proteins such as fibrinogen and clotting factors

Extravascu-lar fibrin meshes provide a provisional tumor matrix

Pro-teases, especially urokinase plasminogen activator (uPA)

and matrix metalloproteinases (MMPs, e.g., progelatinase

A) expressed on proliferating endothelial cells, facilitate

sprouting of blood vessels into the fibrin matrix toward the

stimuli

The established tumor microcirculation is characterized

by its high vascular density, chaotic vascular branching, and

aneurysmatic sacculations High vascular permeability is a

consequence of sustained influence of growth factors and

cytokines and is a result of intercellular gaps in tumor

Copyright © 2006, Elsevier Science (USA).

976 P ART III Pathology

aggregation of activated platelets Initiation of platelet

rolling appears to be independent of platelet activation

Platelet rolling on stimulated endothelium is mediated by

endothelial P- and E-selectin P-selectin and von Willebrand

factor are released from Weibel–Palade bodies of

endothe-lial cells and are presented on the endotheendothe-lial surface within

minutes following stimulation [6] Expression of E-selectin

and a second phase of P-selectin expression depend on de

novo protein synthesis and occur hours after the primary

stimulus The main ligand for endothelial P-selectin on

platelets has been identified to be P-selectin binding

ligand-1 Binding of endothelial P-selectin by glycoprotein (GP)

Ib/IX/V as a second ligand depends on previous platelet

activation Platelets store P-selectin along with adhesion

proteins fibronectin, fibrinogen, and von Willebrand factor

(vWF) in their a-granules and release these factors within

minutes following stimulation P-selectin expression on

activated platelets may also facilitate platelet rolling

The endothelial ligand for platelet P-selectin has not been

identified

Slowing down platelet trafficking along the endothelial

surface ensures stable platelet adhesion by a

fibrin-dependent bridging mechanism between GPIIb/IIIa on

platelets and endothelial ICAM-1 or avb3-integrin on

acti-vated and proliferating endothelial cells Adhesion proteins

vWF, fibrin, and fibronectin are required for further platelet

aggregation by GPIIb/IIIa binding Receptors involved in

platelet interaction with intact endothelial surface are

depicted in Figure 1

At sites of endothelial damage and exposure of

sub-endothelial matrix platelet rolling and aggregation depend

mainly on vWF [7] Plasmatic and endothelial vWF become

immobilized on subendothelial matrix Platelet rolling and

adhesion are initiated by the binding of GPIb/IX/V to

immo-bilized vWF that goes along with platelet activation ATP

and serotonin released from dense granules of activated

platelets stimulate the aggregation of further platelets This

is accompanied by secretion of vWF, P-selectin, fibrinogen,and fibronectin from a-granules of platelets and transforma-

tion of GPIIb/IIIa into its active configuration Thus, plateletadhesion to the subendothelial matrix is enhanced by fibrin-mediated GPIIb/IIIa binding to immobilized vWF Further-more, different types of collagen receptors on platelets cancause a fast platelet activation on contact to the subendothe-lial matrix Depending on the type of vessel wall, collagenreceptors contribute differently to platelet–vessel wall inter-actions Receptors involved in platelet rolling, adhesion, andaggregation on the subendothelial matrix are summarized in Figure 2

Platelet–Endothelial Interaction in Tumor Angiogenesis

The microvascular architecture of tumors has been studied in different experimental tumors However, plate-lets in tumor microcirculation have rarely been des-cribed Ultrastructural studies on different melanomas of the hamster revealed that platelets were attached to themicrovascular endothelium only occasionally at sites oftumor invasion into the microvasculature [8] In vivo obser-vations of platelets in tumor microvessels were initiallydescribed in the amelanotic melanoma (A-Mel-3) implantedinto the dorsal skinfold chamber of hamsters using intravitalmicroscopy and intravenous application of fluorescencemarker for platelets When tumor necrosis became obvious,masses slowing down blood flow and partly occludingtumor blood vessels were detected This phenomenon was interpreted as platelet aggregation and vascular throm-bosis [5] However, extensive ultrastructural studies on themicrovasculature of the amelanotic melanoma revealedinfrequent contact of platelets to the microvascular endothe-lium even at sites of endothelial leakage and erythrocyte

ICAM-1 PSGL-1

fibrinogen

WPB a-granules

Figure 1 Platelet adherence to the intact endothelial surface (Left)

Mechanisms of platelet rolling (Right) Platelet adherence and aggregation

on the endothelial surface GP, glycoprotein; PSGL-1, P-selectin

glycopro-tein ligand-1; vWF, von Willebrand factor; WPB, Weibel–Palade body.

GPIb/IX/V GPIIb/IIIa

vWF SEM

P-selectin

Ca ++

fibrinogen

vWF a-granules

Figure 2 Platelet interactions on the subendothelial matrix (Left)

Ini-tial platelet rolling on immobilized vWF induces activation of glycoprotein

(GP) Ib/IX/V and GP IIb/IIIa (Right) Adherence on the subendothelial

matrix and aggregation of platelets is mediated by glycoproteins and sion proteins SEM, subendothelial matrix.

adhe-CHAPTER 144 Platelet-Endothelial Interaction in Tumor Angiogenesis 977

extravasation Tumor microvessels were regularly

com-posed of endothelial and tumor cells forming the vessel

lumen [4]

We have studied platelet–endothelial interactions in

angiogenesis and growth of Lewis lung carcinoma (LLC-1)

and methylcholanthrene-induced fibrosarcoma (BFS-1)

The study was carried out using the dorsal skinfold chamber

on mice and intravital fluorescence microscopy Ex vivo

flu-orescently labeled syngenic platelets were transfused into

experimental animals Both tumors developed a typical

heterogeneous tumor microcirculation In contrast to the

amelanotic melanoma of hamsters, tissue necrosis did

not appear in LLC-1 carcinoma and BFS-1 fibrosarcoma

within 14 days after tumor cell implantation into the dorsal

skinfold chamber Platelet–endothelial interaction was

rarely observed in tumor microvessels No differences in

platelet–endothelial interactions were seen in tumor

microvessels compared to subcutaneous venules in

tumor-free tissue (Figure 3)

Under physiologic conditions, low interactions between

platelets and endothelial cells have been described in

dif-ferent organ systems that might be due to weak platelet

agonist (ADP, epinephrine, serotonin) We found equivalent

baseline platelet–endothelial interactions in subcutaneous

venules of tumor-free tissue In the early phase of tumor

growth, platelet rolling was slightly enhanced in angiogenic

microvessels in close vicinity to implanted tumor cells The

angiogenic microvessels appeared as dilated and highly

per-meable microvessels due to growth factors and

inflamma-tory cytokines

Since tumor microcirculation was supposed to present a

highly prothrombotic environment, we would have expected

more significant platelet–endothelial interactions in tumor

angiogenesis and tumor microcirculation However, the

compromised hemodynamics within the tumor

microvascu-lature was not associated with activation or aggregation ofplatelets Our results were confirmed by a study indicatingthat radioactively labeled fibrinogen accumulates in Lewis-lung carcinoma; however, the study failed to detect accu-mulation of radioactively labeled platelets within the sametumor Furthermore, electron microscopy showed singleplatelets adhering to endothelial gaps in tumor microvesselsonly occasionally

Platelets Are Potentially Involved in

Tumor Angiogenesis

Platelets are potentially involved in tumor angiogenesisbecause they contain numerous angiogenic factors (e.g.,VEGF, bFGF, PDGF, PAI-1) Clinical studies revealed thatplatelets are the main source of the serum VEGF concentra-tion typically correlated with poor prognosis in cancerpatients

Although the mechanisms of growth factor release fromplatelets have not been entirely resolved, in vitro experi-ments have demonstrated that growth factor release depends

on platelet activation and aggregation Although plateletsecretion occurs independently of platelet aggregation,release of growth factors becomes detectable only abovethreshold doses of thrombin and other platelet agonists atwhich platelet aggregation is inevitable A further prerequi-site for the promoting effect of platelets on the formation ofmicrovascular tubuli seems to be direct cell-to-cell interac-tions between platelets and the proliferating endothelium.Growth factors, especially VEGF, are secreted by a highpercentage of malignant animal and human tumors, and also

by stromal cells adjacent to the tumors Only a few tional tumors are known that express little or no VEGFmRNA and protein These tumors, renal papillary carcinomaand lobular breast cancer, are found to be almost avascularand to have better prognosis because of their minor inva-siveness A positive correlation between platelets and dis-ease progression has only been reported for advanced cancerdiseases characterized by dense vascularization, invasive-ness, and metastatic diseases

excep-VEGF may be taken up by platelets from blood plasmasince there is a strong correlation between plasma VEGFlevels and serum VEGF load per platelet in tumor patients.Therefore, VEGF content in platelets might be not only aresult of increased protein syntheses in megakaryocytes inresponse to tumor released cytokines and growth factors, butalso a consequence of endocytosis of tumor-derived VEGF

by circulating platelets [9]

In conclusion, tumor microvessels do not express ahighly prothrombotic environment favoring platelet aggre-gation Hence, platelets do not become sufficiently activated

to release angiogenic growth factors at sites of tumor genesis It is more probable that platelets take up and pre-serve angiogenic growth factors released by tumor cells.The role of platelets as a reservoir for angiogenic growthfactors in tumor growth has not yet been identified

angio-Day after tumor cell implantation

Figure 3 Platelet–endothelial interaction is not an intense phenomenon

during tumor angiogenesis Platelet rolling was increased on day 1 in

pre-existing microvessels of the host in response to both LLC-1 (gray bars) and

BFS-1 (hatched bars) and also on day 3 in LLC-1 tumor microvessels, but

was only slightly above the low baseline level quantified in postcapillary

venules of controls (open bars) n= 6 experimental animals per group,

*p< 0.05 versus controls; #p< 0.05 versus day 1 and day 3 (Kruskal–

Wallis test).

978 P ART III Pathology

Platelet–Endothelial Interaction Following

Endothelial Stimulation

In a second set of experiments on the microcirculation

of LLC-1 carcinoma and BFS-1 fibrosarcoma, platelet–

endothelial interactions were assessed in response to

endothelial stimulation by calcium ionophore A23187

Calcium ionophore A23187 increases intracellular calcium

concentration and thereby induces the release of adhesion

molecules vWF and P-selectin from endothelial

cell–spe-cific Weibel–Palade bodies sparing endothelial integrity

The secretion of Weibel–Palade bodies from endothelial

cells plays a central role in wound healing, coagulation,

inflammation, and ischemia–reperfusion injury In addition,

angiogenic growth factor VEGF induces the secretion of

Weibel–Palade bodies and adhesion molecule presentation

Hence, tumor cells might influence adhesion molecule

expression in microvessels [10]

The application of calcium ionophore A23187 did not

affect platelet–endothelial interaction in BFS-1

fibrosar-coma, and only a slight increase in platelet rolling was

detected in microvessels of the LLC-1 carcinoma However,

in subcutaneous venules of tumor-free tissue, platelet rolling

increased significantly (Figure 4) Platelet adherence was

only occasionally seen following superfusion with calcium

ionophore The results indicate that tumor microvessels do

not express sufficient amounts of vWF and P-selectin in

response to stimulation with calcium ionophore A23187 In

respect to the stimulus used, this phenomenon can be

attrib-uted to the reduced or missing secretion of adhesion

mole-cules from endothelial Weibel–Palade bodies [10]

Immunohistological studies on vWF and P-selectinstored in Weibel–Palade bodies of endothelial cells describe

a granular staining pattern in the microcirculation of entiated tissues [11] In tumor microvessels, however, vWF expression appears to be extremely heterogeneous.Immunohistological techniques showed reduced or evenabsent staining for vWF alternating with areas highly posi-tive for vWF In follicular thyroid cancer, vWF was notdetectable, whereas it was highly significant in normal thy-roid tissue A similar immunohistological pattern has beenfound for P-selectin expression within the tumor micro-circulation The distribution pattern of vWF and P-selectinwas never associated with angiogenic processes

differ-Hence, immunohistological studies disclose that vWFand P-selectin storage in Weibel–Palade bodies of endothe-lial cells is reduced in tumor microvessels This might be aconsequence of Weibel–Palade body exocytosis due to sustained stimulation by angiogenic growth factors andcytokines preventing redistribution and regeneration ofvWF and P-selectin into their intracellular storage granules.Furthermore, angiogenic growth factors prevent proteinsynthesis of adhesion molecules that are required for interactions of platelets as well as of leukocytes with the microvascular endothelium Distribution of Weibel–Palade bodies within the tumor microcirculation seems todepend on tumor type and species Weibel–Palade bodiesmay be completely missing in the microvasculature of sometumors

Conclusions

Tumor angiogenesis and tumor growth are accompanied

by only a transient increase in platelet rolling in angiogenicmicrovessels in close vicinity to tumor cells This might

be due to endothelial stimulation by growth factors andcytokines released partly from tumor cells Within tumormicrocirculation, adhesion molecules mediating platelet–endothelial interactions appeared to be downregulated oreven absent Platelet–endothelial interaction in response toendothelial stimulation was significantly reduced in tumormicrocirculation

Future studies are needed to identify mechanisms lating adhesion molecule expression in tumor microcircula-tion Further attempts should focus on induction of plateletaggregation by selective transport of prothrombotic agentsinto the tumor microcirculation The results might providefurther therapeutic aspects in targeting tumor angiogenesis

regu-Bibliography

1 Möhle, R., Green, D., Moore, M A., Nachman, R L., and Rafii, S (1997) Constitutive production and thrombin-induced release of vas- cular endothelial growth factor by human megakaryocytes and

platelets Proc Natl Acad Sci USA 94, 663–668.

2 Pinedo, H M., Verheul, H M., D’Amato, R J., and Folkman, J.

(1998) Involvement of platelets in tumour angiogenesis? Lancet 352,

1775–1777 This article proposes a possible role for platelets in tumor

Figure 4 Calcium ionophore A23187 induces platelet rolling favorably

in normal and less in tumor microvessel Superfusion of tumor tissues with

calcium ionophore A23187 (20 mM) over 15 minutes causes a twofold

increase of rolling platelets in LLC-1 (gray bars) and exerts no effects on

BFS-1 (hatched bars) on day 14 after tumor cell implantation In contrast,

the treatment of tumor-free preparations resulted in a threefold increase of

rolling platelets in postcapillary venules of controls (open bars) on

corre-sponding days to tumor groups after chamber preparation n= 6

experi-mental animals per group, *p< 0.05 versus controls, #p< 0.05 versus

baseline (Kruskal–Wallis test and Wilcoxon test).

CHAPTER 144 Platelet-Endothelial Interaction in Tumor Angiogenesis 979

angiogenesis based on a review of the literature, which was the

start-ing point of our experimental work.

3 Carmeliet, P and Jain, R K (2000) Angiogenesis in cancer and other

diseases Nature 407, 249–257.

4 Hammersen, F., Osterkamp-Baust, U., and Endrich, B (1983) Ein

Beitrag zum Feinbau terminaler Strombahnen und ihrer Entstehung

in bösartigen Tumoren In Mikrozirkulation in malignen Tumoren,

P Vaupel and F Hammersen, eds., pp 15–51 Basel, CH: Karger An

early publication that already described the absence of platelets in

tumor microvasculature.

5 Asaishi, K., Endrich, B., Goetz, A., and Messmer, K (1981)

Quanti-tative analysis of microvascular structure and function in the

amelan-otic melanoma A-Mel-3 Cancer Res 41, 1898–1904 A detailed,

quantitative analysis of tumor microcirculation based on intravital

microscopy of a newly developed transparent chamber preparation.

6 Frenette, P S., Johnson, R C., Hynes, R O., and Wagner, D D (1995).

Platelets roll on stimulated endothelium in vivo: An interaction

mediated by endothelial P-selectin Proc Natl Acad Sci USA 92,

7450–7454.

7 Konstantopoulos, K., Kukreti, S., and McIntire, L V (1998)

Biome-chanics of cell interactions in shear fields Adv Drug Deliv Rev 33,

141–164.

8 Warren, B A., Shubik, P., and Feldman, R (1978) Metastasis via the

blood stream: The method of intravasation of tumor cells in a

trans-plantable melanoma of the hamster Cancer Lett 4, 245–251.

9 Salgado, R., Benoy, I., Bogers, J., Weytjens, R., Vermeulen, P., Dirix, L., and van Marck, E (2001) Platelets and vascular endothelial growth

factor (VEGF): A morphological and functional study Angiogenesis 4,

37–43.

10 Manegold, P C., Hutter, J., Pahernik, S A., Messmer, K., and Dellian,

M (2003) Platelet–endothelial interaction in tumor angiogenesis and

microcirculation Blood 101, 1970–1976.

11 Schlingemann, R O., Rietveld, F J., Kwaspen, F., van de Kerkhof, P C., De Waal, R M., and Ruiter, D J (1991) Differential expression

of markers for endothelial cells, pericytes, and basal lamina in the

microvasculature of tumors and granulation tissue Am J Pathol 138,

1335–1347.

Capsule Biography

Dr Dellian has been leading a research laboratory on tumor culation and angiogenesis at the Institute for Surgical Research, University

microcir-of Munich, since 1995 He was introduced to his research by the pioneers

in tumor microcirculation, Prof Konrad Messmer, Prof Alwin Goetz, and Prof Rakesh Jain.

C HAPTER 145

Von Hippel–Lindau Disease: Role of Microvasculature in Development

of the Syndrome

David R Mole and Peter J Ratcliffe

University of Oxford, Oxford, United Kingdom

The von Hippel–Lindau Hereditary

Cancer Syndrome

VHL disease is inherited as an autosomal dominant traitaffecting 1 in 36,000 of the population Affected individualsbear a germ-line mutation in the VHL tumor suppressorgene However, it is the predisposition to cancer rather thanthe cancer itself that is inherited, and the presence of a sin-gle mutation at the susceptibility locus is insufficient fortumor formation Tumors are associated with somatic loss orinactivation of the remaining wild-type allele, in accordancewith the Knudson “two-hit” hypothesis The high likelihood

of somatic mutation affecting the single remaining type allele over the lifetime of an individual with the inher-ited syndrome explains the dominant inheritance pattern

wild-of tumors, the greater than 90 percent penetrance by 65 wild-ofyears age, and the multifocal nature of associated tumors.Sporadic tumors can arise in nonaffected individuals, by theoccurrence of somatic loss of both alleles within the samecell However, on average, it takes longer to accrue two

“hits” within the same cell, accounting for the lower lence (within the much larger at-risk population) and olderage distribution of sporadic tumors, compared to those asso-ciated with the hereditary VHL syndrome

preva-The first report of patients with what was probably VHLsyndrome was by Treacher Collins in 1894, when he re-ported two siblings with retinal angiomas In 1904 Eugenevon Hippel, who gives his name to the syndrome, reportedfamilies with blood-vessel tumors (angiomas) of the retina

Positional cloning of defective genes that underlie rare

hereditary cancer syndromes, such as von Hippel–Lindau

disease, has provided many important insights into more

common nonhereditary malignancies and the fundamental

cellular processes that are involved in oncogenesis In

gen-eral, the tumor suppressor functions defined by analysis

of such genes have involved a direct role in the regulation

of cell proliferation (such as the retinoblastoma gene

prod-uct, Rb), a direct role in DNA repair (such as MSH2

and MLH1), or a role in linking DNA damage to the arrest

of cellular proliferation and apoptosis (such as p53) Such

functions fit readily into a simple genetic model of

cancer whereby the accrual of mutations allows

uncon-trolled growth, and genomic instability reinforces the

gener-ation of mutant clones with an increasingly aggressive

growth advantage

Nevertheless tumor development involves many other

processes, such as the maintenance of adequate blood

oxy-gen delivery by induction of angiooxy-genesis (indeed the Greek

name cancer derives from the crab-like pattern of tumor

neovasculature) Von Hippel–Lindau (VHL) disease is one

of several hereditary cancer syndrome syndromes associated

with excessive angiogenesis, and recent insights into the

role of the VHL protein (pVHL) in the development of the

abnormal microvasculature seen in this syndrome have led

to a greater mechanistic understanding of tumor

angiogene-sis and cellular oxygen sensing

Copyright © 2006, Elsevier Science (USA).

982 P ART III Pathology

The correlation with microscopically indistinguishable

tumors (hemangioblastomas) of the central nervous system

(especially cerebellum and spinal cord) was first described,

in 1926, by Arvind Lindau This description also included

cysts in the kidney, pancreas, and epididymis However, the

first diagnostic criteria to include renal cell carcinoma as

part of the syndrome was that of Melmon and Rosen in

1964 Other tumors now known also to be associated with

the syndrome include pheochromocytoma (a

hormone-secreting tumor of the adrenal medulla), endolymphatic-sac

tumors of the inner ear, and islet-cell tumors of the pancreas

Hemangioblastomas and renal-cell carcinomas are both

highly vascular tumors and demonstrate disordered growth

of the microvasculature Hemangioblastomas are the

com-monest tumors in the VHL syndrome, affecting 60 to 80

per-cent of individuals, and have a predilection for the posterior

fossa and spinal cord They are well-defined thinly

encapsu-lated benign tumors formed from a mixture of “stromal”

cells and a rich plexus of sometimes telangiectatic capillary

blood vessels, and are frequently associated with edema and

cysts It is the stromal cells that are the tumor cells These

cells present a finely vacuolated or foamy cytoplasm, and

sometimes evoke an epitheloid appearance of the tumor

mimicking that of metastatic renal cell carcinoma They lack

functional pVHL and overproduce angiogenic growth

fac-tors such as vascular endothelial growth factor (VEGF) and

platelet-derived growth factor B chain (PDGF-b) These in

turn drive proliferation of endothelial cells and pericytes,

respectively, leading to the angiogenic phenotype

Further-more, VEGF, also known as vascular permeability factor

(VPF), increases capillary leakiness, leading to edema

Renal cell carcinomas, also known as clear-cell

carcino-mas, arise from the renal tubular epithelium and again are

highly vascular, giving them their characteristic red color at

operation The tumors develop from preneoplastic cysts

lined with VHL-/- epithelial cells Importantly, the very

large number of associated benign lesions compared to

malignant lesions suggests a complex oncogenic process,

most probably involving several additional genetic

alter-ations following VHL inactivation However, restoration of

pVHL function in fully transformed VHL-/- renal

carci-noma cells suppresses their ability to form tumors in nude

mice, indicating an ongoing requirement for VHL

inactiva-tion Interestingly, restoration of pVHL does not affect the

ability of tumor cells to grow in adherent tissue culture,

sug-gesting that the tumor suppressor action is in some way

environmentally specific Like the stromal cells in

heman-gioblastomas, renal-cell carcinoma cells also overexpress

VEGF and PDGF-b, leading to their highly vascular

appearance

Although the angiogenic tumor phenotype is most

appar-ent in VHL-associated tumors, this increased angiogenesis

is not unique to them A considerable body of evidence

spanning more than three decades has documented that

tumor growth and metastasis of many cancers requires

per-sistent new blood-vessel growth When tumor cells were

transplanted into avascular sites, such as the cornea, the

implants attracted new capillary growth In the absence ofaccess to an adequate vasculature, tumor cells becamenecrotic and/or apoptotic Furthermore, ingress of newblood vessels in these tumor transplantation models sug-gested that tumors released diffusible activators of angio-genesis that could stimulate a quiescent vasculature to begincapillary sprouting In a variety of transgenic mouse tumormodels, an “angiogenic switch” could be detected during theearly stages of tumorigenesis, preceding the development ofsolid tumors Recent analyses of von Hippel–Lindau diseasehave provided the first direct link between a tumor sup-pressor gene function and the molecular mechanisms ofangiogenesis

The VHL Protein (pVHL)

The VHL gene was first cloned from chromosome

3p25-26 by a large consortium in 1993, and comprises just threeexons coding for a protein of 213 amino acids A second iso-form of 150 residues is produced from an in frame ATG atcodon 54 Notably all disease-causing mutations affectsequences C-terminus to this second start site, so thattumorigenesis is associated with inactivation of both forms.The majority of familial point mutations lie within tworegions of the protein, suggesting two major functionaldomains Although the primary VHL sequence did notimmediately suggest a function, protein association experi-ments defined a series of pVHL interacting proteins, includ-ing elongins B and C, CUL2, and Rbx1, which bound to thefrequently mutated a-domain Cul2, a member of the cullin

family, resembles yeast Cdc53, and elongin C resemblesyeast Skp1 In yeast, Cdc53 and Skp1 bind to one another

to form ubiquitin ligases referred to as SCF complexes(Skp1/Cdc53/F-box protein), which covalently attach aubiquitin polymer to cell-cycle control proteins targetingthem for destruction by the proteasome In such complexes,the target protein destined for polyubiquitination, and hencedestruction by the proteasome, is recognized or bound bythe F-box protein (so named because of a collinear Skp1-binding motif present in cyclin F), suggesting an analogousrole for pVHL This putative role was strengthened byrecognition of a second frequently mutated subdomain ofpVHL, the b-domain, which has features of a substrate-

docking site Experimental evidence confirmed that pVHL immunoprecipitates exhibit ubiquitin ligase activity

anti-in vitro

The discovery that pVHL was acting as a ubiquitin ligasesubstrate recognition module raised the question of whatwas its target protein, and how was this related to the angio-genic phenotype of the tumors? Repetition of the proteinassociation studies in the presence of proteasomal blockadedemonstrated two novel pVHL binding partners, the a-

subunits of the transcription factors hypoxia inducible factor

1 and 2 Furthermore, in vitro ubiquitylation studies showedthat these two proteins were targeted for ubiquitylation bypVHL (see Figure 1)

CHAPTER 145 Von Hippel–Lindau Disease: Role of Microvasculature in Development of the Syndrome 983

pVHL and Oxygen Sensing

Hypoxia inducible factor (HIF) is an ab-heterodimer that

was first recognized as a DNA-binding protein responsible

for mediating the hypoxia-inducible activity of the gene

encoding the hematopoietic growth factor erythropoietin

Both subunits contain basic helix–loop–helix (bHLH)-PAS

domains Whereas the bHLH domain defines a superfamily

of eukaryotic transcription factors, the PAS domain was first

defined in the PER, ARNT, and SIM proteins and defines a

subset of the bHLH family HIF-b subunits are

constitu-tively expressed nuclear proteins that are involved via

other dimerization partners in a variety of transcriptional

responses The HIF-a subunits exist as three isoforms, all of

which are inducible by lack of oxygen (hypoxia) However,

in oxygenated cells these a-subunits are rapidly degraded,

with HIF-1a and HIF-2a having an exceptionally short

half-life of just a few minutes When the a-subunits are

sta-bilized, HIF dimerizes and interacts with cis-acting hypoxia

response elements (HREs) to induce transcriptional activity

A large and rapidly expanding array of genes have now been

shown to contain HREs and to be transcription targets ofHIF Examples of these responses include not only erythro-poietin, but also endothelial nitric oxide synthase andendothelin, which control vascular tone; glucose trans-porters and many key enzymes involved in the metabolism

of glucose; enzymes involved in catecholamine synthesis;proteins concerned with iron transport and handling; andcheckpoints in cell proliferation and quiescence However,most importantly with regard to the angiogenic phenotype

of VHL disease, many genes that play an essential role inangiogenesis are HIF target genes

HIF-a subunits contain two domains that regulate their

activity in response to oxygen availability: an internal oxygen-dependent degradation domain (ODDD), whichregulates protein destruction, and the C-terminal transacti-vation domain (CAD), which regulates transactivating abil-ity through recruitment of the coactivator p300/CBP Innormoxia, enzymatic hydroxylation of two prolyl residues,within the HIF-a ODDD, by a newly defined prolyl hydro-

xylase activity facilitates recognition of HIF-a by the VHL

E3 ubiquitin ligase complex that targets HIF for destruction

by the proteasome In mammalian cells three closely relatedenzymes, each the product of a different gene, have beenshown to have HIF prolyl hydroxylase activity These pro-teins are all nonheme Fe (II) enzymes that are members ofthe superfamily of 2-oxoglutarate dependent dioxygenases

An absolute requirement for molecular oxygen renders theirfunction oxygen sensitive In the absence of pVHL, hydro-xylated HIF-a cannot be targeted for ubiquitination and sub-

sequent proteasomal degradation, leading to constitutiveupregulation of HIF-mediated gene transcription even in thepresence of adequate oxygen for hydroxylase function.Further oxygen-dependent control is regulated by factorinhibiting HIF (FIH) through hydroxylation of an aspar-aginyl residue in the C-terminal transactivation domain.Hydroxylation at this residue interferes with HIF transacti-vating ability by blocking interaction with the transcrip-tional coactivator CBP/p300 Interestingly, the fact thatpVHL loss results in full, rather than partial, activation sug-gests that pVHL could be involved in aspects of the HIFresponse to oxygen beyond ODDD-mediated proteolysis Ithas been suggested that pVHL might be involved directly inthe process of transcriptional inactivation both by promotingthe function of FIH and by recruitment of transcriptionalrepressors

Hypoxia-Inducible Factor and Angiogenesis

It is now clear that vascular architecture in tissues islargely controlled by angiogenic signals from the con-stituent cells Local oxygen tension appears to be a criticalsignal in detecting and responding to local vascular insuffi-ciency; for example, in the retina local ischemia almostalways precedes new vessel growth

HIF and pVHL play a critical role in these responses and many points of interaction with molecular processes

Figure 1 Regulation of HIF- a by pVHL Under oxygen-replete

condi-tions HIF- a is enzymatically modified by prolyl hydroxylase activity to a

form that is recognized by pVHL The pVHL ubiquitin ligase complex

covalently attaches a polyubiquitin tag, which then targets HIF- a for rapid

destruction in by the proteasome, so that it is no longer available to effect

target gene transcription (see color insert)