fundamental understandings how cancer spreads

Our body is a community ofcells, in which each cell occu-pies a place appropriate for

its tasks on behalf of the whole. With theexception of white blood cells, whichpatrol the body for microbial invadersand tissue damage, normal cells stay inthe tissue of which they are part. Can-cer cells, however, are rogues that tres-pass aggressively into other tissues.

Metastasis, the spread of cancer to

distant sites in the body, is in fact whatmakes cancer so lethal. A surgeon canremove a primary tumor relatively easily,but a cancer that has metastasized usu-ally reaches so many places that cure bysurgery alone becomes impossible. Forthat reason, metastasis and the invasionof normal tissue by cancer cells are thehallmarks of malignancy. In countrieswhere health care is primitive, one some-times sees people who live with tumors

as big as a soccer ball; the cells that makeup these so-called benign tumors obvi-ously overproliferate, but unlike malig-nant cancer cells, they do not invade ormetastasize.

Acquiring the capabilities needed toemigrate to another tissue is therefore akey event in the development of a can-cer. To metastasize successfully, cancercells have to detach from their originallocation, invade a blood or lymphaticvessel, travel in the circulation to a dis-tant site and establish a new cellularcolony. At every one of these steps, theymust escape many controls that, in ef-fect, keep normal cells in place.

A fruitful way of understanding howtumor cells evade these controls has con-sequently been to study the signals thatnormally direct cells to their place in thebody and keep them there during adult-hood. When I was a postdoctoral fel-low at the California Institute of Tech-nology from 1968 to 1970, my mentor,William J. Dreyer, had become interest-ed in those questions. Roger W. Sperry,also at Caltech, had found that the light-sensing nerve cells in the retina of the eyegrow orderly extensions into the brainsuch that the extensions from a givenretinal region always project into thesame brain region. These findings in-spired Dreyer and Leroy E. Hood to pos-

How Cancer SpreadsTumor cells roam the body by evading the controls that keep normal cells in place. That fact offers clues to fighting cancer

by Erkki Ruoslahti

Fundamental Understandings

72 Scientific American September 1996 How Cancer Spreads

PRIMARY TUMOR

NORMAL EPITHELIAL CELL

BASEMENT MEMBRANE

BLOOD VESSELINVASIVE TUMOR CELL

Copyright 1996 Scientific American, Inc.

tulate their “area code” hypothesis, thata cell has on its surface an address sys-tem—written in one set of molecules andreadable by molecules on other cells—

that identifies where the cell should be.It seemed to me at the time that if a

molecular address system existed, some-thing had to be wrong with it in cancer,because cancer cells did not stay put. Idecided to try to find such molecules.As the work of many laboratories even-tually showed, area code molecules doexist. They mediate cell adhesion, theanchoring of cells to adjacent structures.

In normal tissues, cells adhere both toone another and to an insoluble mesh-work of protein filling the space betweenthem, known as extracellular matrix.(This arrangement is particularly de-scriptive of the epithelia, which are thecell layers that form the outer surface ofthe skin and the lining of the gut, lungsand some other organs, and from whichmost cancer originates.) The two kindsof adhesion play different critical rolesduring tissue invasion and metastasis.

Cell-cell adhesion molecules appear tohelp keep cells in place; these moleculesseem to be missing or compromised incancer cells. For example, various kindsof cancers lose some or all of an inter-

cellular adhesion molecule called E-cad-herin. By manipulating this molecule incultured cancer cells, one can change thecells’ ability to invade tissues and formtumors. Walter Birchmeier, now at theMax Delbrück Center in Berlin, firstshowed that blocking the function of E-cadherin can turn a cultured lineage ofcells from noninvasive to invasive. Con-versely, restoring E-cadherin to cancercells that lack it can negate their ability toform tumors when they are injected intomice. Thus, loosening of the adhesiverestraint between cells is likely to be animportant early step in cancer invasion.

The Need for Adhesion

Adhesion to extracellular matrix, on the other hand, allows cells to sur-

vive and proliferate. As researchers haveknown for many years, cultured cellscannot reproduce until they attach to asurface, a phenomenon called anchor-age dependence. This attachment is me-diated by cell-surface molecules knownas integrins that bind to the extracellu-lar matrix. As Steven Frisch of the Burn-ham Institute in La Jolla, Calif., MartinA. Schwartz of the Scripps Research In-stitute, also in La Jolla, Calif., and Mina

J. Bissell of the University of Californiaat Berkeley have shown, only attach-ments involving integrins can satisfy therequirements of anchorage dependence.

My laboratory at the Burnham Insti-tute, together with Tony Hunter of theSalk Institute for Biological Studies inSan Diego, Calif., has recently shownthat unattached cells stop growing be-cause one of the nuclear proteins (knownas the cyclin E–CDK2 complex) thatregulates the growth and division of cellsbecomes less active. Inhibitory substanc-es in the nuclei of these cells seem to shutdown this protein.

As Frisch, Schwartz and Bissell alsodiscovered, when many types of cells aredenied anchorage, they not only stopproliferating but commit suicide. Thatis, they spontaneously undergo specificchanges that lead to their own death.This kind of cell death, in which the cellis an active participant, has been termedapoptosis.

My group has demonstrated that forcells to survive, the extracellular matrixto which they adhere must bear the right“area code,” one that is probably foundonly in the extracellular matrix of selecttissues. Moreover, they have to use theappropriate integrin to attach to the ma-

How Cancer Spreads Scientific American September 1996 73

DA

NA

BU

RN

S-P

IZER

INVASION AND METASTASIS are the processes that lethally spread cancer cellsthroughout the body. First, cancer cells detach from the primary site (which is often inan epithelial tissue) and breach the basement membrane separating them from othertissue layers. Some of these invasive cells can penetrate the basement membrane sur-rounding a blood vessel, as well as the layer of endothelial cells lining it. The cells arethen free to circulate via the bloodstream. Eventually a cancer cell may lodge in a cap-illary. If it then adheres to and penetrates the capillary wall again, it can create a sec-ondary tumor. Perhaps fewer than one in 10,000 cancer cells that escape the primarytumor survives to colonize another tissue.

ENDOTHELIAL LINING

BASEMENT MEMBRANE

SECONDARY TUMOR SITE

METASTATIC CELLIN CIRCULATION

TUMOR CELL ADHERINGTO CAPILLARY


trix. As all these results show, a molecu-lar explanation for anchorage depen-dence is beginning to take shape, al-though much more critical detail stillneeds to be filled in.

Cellular suicide from lack of anchor-age or from inappropriate anchorage islikely to be one of the safeguards thatmaintain the integrity of tissues. Cellsusually cannot just float away from theirtissue and establish themselves some-where else, because they will die on theway. Yet cancer cells get around this re-quirement; they are anchorage indepen-dent. The cyclin E–CDK2 complex insuch cells stays active whether the cellsare attached or not.

How cancer cells accomplish this trickis not fully understood, but it seems thatoncogenes can be blamed. (Oncogenesare mutated versions of normal genescalled proto-oncogenes; these mutationscan turn normal cells into malignantones; see “How Cancer Arises,” by Rob-ert A. Weinberg, on page 62.) In effect,as various experiments have shown,proteins made by these oncogenes con-vey a false message to the nucleus thatthe cell is properly attached when it isnot, thereby stopping the cell from ar-resting its own growth and dyingthrough apoptosis.

Anchorage dependence is only one ofthe constraints that a cancer cell mustovercome to roam around the body. Ep-ithelial cells, the most common sourcesof cancers, are separated from the rest

of the body by a basement membrane,a thin layer of specialized extracellularmatrix. Basement membranes form abarrier that most normal cells cannotbreach, but cancer cells can [see “Can-cer Cell Invasion and Metastasis,” byLance A. Liotta; Scientific Ameri-can, February 1992].

This fact can be strikingly demonstrat-ed by giving cells in a test tube an op-

portunity to invade through a naturalor reconstructed basement membrane:cancer cells will penetrate it; normalones will not. Furthermore, in this ex-periment, cells from metastatic cancersgenerally invade faster than those fromnonmetastatic tumors. White bloodcells, in keeping with their role as secu-rity patrol, are an exception to the rulethat normal cells do not invade—they,too, are adept at penetrating tissues, in-cluding basement membranes. Cancercells and white blood cells do so by re-leasing enzymes, called metalloprotein-ases, that dissolve basement membranesand other extracellular matrices. Othercells have less of these enzymes andmore enzyme inhibitors.

After a cancer cell has passed throughthe basement membrane separating itfrom the rest of the tissue at its originalsite, it soon encounters another basementmembrane, one surrounding a smallblood vessel. (A blood vessel is usuallynearby, because to sustain themselvessuccessful tumors induce the growth ofnew blood vessels.) By penetrating thissecond basement membrane barrier andthe layer of endothelial cells that formthe vessel’s inner lining, the cancer cellgains access to the bloodstream and iscarried elsewhere in the body.

New technology makes it possible todetect cancer cells in the blood of pa-

How Cancer Spreads74 Scientific American September 1996


MELANOMACELL

RGD

MELANOMA METASTASIZES NO METASTASIS

RGDTRIPEPTIDE

FIBRONECTIN

TUMOR CELL

INHIBITING METASTASIS by interfering with cancer cell adhesion may someday bea therapeutic option. In mouse experiments, injections of RGD, a fragment of the pro-tein fibronectin, discouraged melanoma cells from spreading to the lungs. Presumably,the RGD molecules blocked receptors that wandering cancer cells needed for bindingto fibronectin in the extracellular matrix of tissues.

EXTRACELLULAR MATRIX

CELL IN WRONG LOCATION

CELLS IN APPROPRIATE LOCATION

ADHESION MOLECULES

“AREA CODES” FOR CELLS take the form of specific surface adhesion moleculesand receptors. During development, a normal cell recognizes its proper place in thebody by fitting its adhesion molecules to those on other cells and on the extracellularmatrix. In cancer, something goes wrong with this address system.

DA

NA

BU

RN

S-P

IZER

DIM

ITR

Y S

CH

IDLO

VSK

Y


tients. Great strides have been made inidentifying telltale marker moleculesthat distinguish a cell as having comefrom a specific tissue or type of tumor.At the same time, researchers have alsodeveloped ultrasensitive assays (basedon such techniques as the polymerasechain reaction and monoclonal-anti-body tagging) for detecting those mole-cules. From studies employing thesemethods, we know that malignant cellsare often circulating even when a clini-cal examination cannot yet find evidenceof the cancer’s distant spread.

The further development of such testsmay eventually improve therapies, byhelping physicians determine whetherthey need to prescribe treatments be-yond surgery for seemingly containedtumors. Detection of micrometastasesin the blood and elsewhere in the bodyis a significant step forward in early di-agnosis, and it is the vanguard of ap-plied research on metastasis.

Some doctors have also wonderedwhether the manipulation of a tumorduring its diagnosis or surgical removalmight be enough to release cells into thecirculation. The new testing methodsshould allow researchers to prove ordisprove this ominous hypothesis, butto my knowledge, that has not yet beendone. But even if the hypothesis provesto be correct, it is clear that the benefitsof diagnostics and surgery far outweighthe possible risks from inaction.

Vulnerable in the Blood

Fortunately, even when cancer cellsdo get into the circulation, the for-

mation of secondary tumors is not in-evitable. The circulating cell still facesseveral more hurdles: it must attach tothe inner lining of a blood vessel, crossthrough it, penetrate the basement mem-brane at this new location, then invadethe tissues beyond and begin multiply-ing. Each of these obstacles makes de-mands of the tumor cell that may gobeyond those it faced in its home tissue.Furthermore, it may also be that manycancers cannot entirely overcome thedefense mechanisms that keep our cellsin the right places—another hindranceto metastasis.

Probably fewer than one in 10,000 ofthe cancer cells that reach the circulationsurvive to found a new tumor at a dis-tant site. The reasons for this apparentvulnerability while in the blood are not

well understood—perhaps the anchorageindependence of the tumor cells is notcomplete, and they sometimes diethrough apoptosis after all. In any case,researchers believe the cells need to at-tach fairly promptly to the inner liningof a small blood vessel.

Blood circulation explains much aboutwhy various metastatic cancers spreadpreferentially to certain tissues. Circu-lating tumor cells usually get trapped inthe first vascular bed (or network ofcapillaries, the finest blood vessels) thatthey encounter “downstream” of their

origin. The first vascular bed encoun-tered by blood leaving most organs is inthe lungs; only the intestines send theirblood to the liver first. Accordingly, thelungs are the most common site of me-tastasis, followed by the liver.

In part, cancer cells lodge in smallblood vessels because these cells tend tobe large. Also, some cancers producechemical factors that cause platelets, thetiny blood cells that initiate blood clot-ting, to aggregate around them. Theseaggregates effectively make the cancercells even larger and stickier. (It is also


CELLULAR ADHESION

CELL-CELL ADHESION MOLECULES

RECEPTOR FOR EXTRACELLULAR MATRIX

EXTRACELLULAR MATRIX

IMPORTANCE OF ADHESIONTO EXTRACELLULAR MATRIX

DETACHED CELL

DEAD CELL

CELL ATTACHEDTO INAPPROPRIATEEXTRACELLULAR MATRIX

IMPORTANCE OF CELL-CELL ADHESION

INVASIVE CELL

CELLULAR ADHESION is a vital brake on the migration of normal cells. Two typesapply to most body cells: cell-cell adhesion and adhesion to the extracellular matrix(top). If a cell cannot adhere to other cells, it may become more invasive and migratethrough the matrix (middle). If a cell lacks adhesion to the extracellular matrix, it candetach from its native tissue (bottom). Usually, if a cell fails to reattach to the extracel-lular matrix or if it attaches to the wrong type of matrix, it dies through apoptosis (cel-lular suicide). Cancer cells, however, can survive without this adhesion.

DA

NA

BU

RN

S-P

IZER


noteworthy that platelets produce theirown rich supply of growth factors, andthese may help the cancer cells to whichthey bind survive in the blood. Thismay be why, in some experimental sys-tems, drugs that interfere with plateletfunctions have anticancer effects.)

Physical trapping of cancer cells in theblood vessels at the site of metastasis isnot the whole story, however. If it were,cancers would not spread so diversely

through the body. Indeed, some types ofcancer show a striking preference for or-gans other than those that receive theirvenous blood—witness the tendency ofmetastatic prostate cancer to move intothe bones. Once again, the explanationseems to rest with the molecular addresssystem on cell surfaces. A specific affin-ity between the adhesion molecules oncancer cells and those on the inner lin-ings of blood vessels in the preferred tis-

sues could explain the predilection ofthe cells to migrate selectively. Differentconcentrations of growth-promotingfactors and hormones in various tissuesmay also play a part.

Recently, in an elegant piece of work,Ivan Stamenkovic of Harvard MedicalSchool and his colleagues showed thathe could direct the metastatic spread oftumor cells: he genetically engineeredmice so that their livers displayed a tar-get for an adhesion molecule found oncertain tumor cells. As predicted, thetumor cells homed in on the liver. Forthese experiments, Stamenkovic bor-rowed receptors and targets from themolecular adhesion system used bywhite blood cells to leave the circulationand enter tissues. Although this systemwas artificial, it may be that cancersnaturally mimic white blood cells inmuch this way—cancer cells do oftenmanufacture certain molecules (calledLex) important to the mobility of whiteblood cells in the body.

Finding the Body’s Area Codes

If, as seems likely, there is much to belearned by identifying the molecular

addresses that white blood cells and tu-mors use to find particular tissues, amethod of doing so that Renata Pasqua-lini, a postdoctoral fellow in my labora-tory, and I have devised should provehelpful. We adapted a technique for iso-lating biologically active molecules fromhuge collections, or “libraries,” of diversecompounds. The theory behind this ap-proach is that if one screens a sufficient-ly large number of compounds, one canfind a molecule for almost any purpose.

We use a large library of peptides(small pieces of protein) as the sourceof our compounds. During the 1980s,George Smith, now at the University ofMissouri, devised a technique for build-ing such a library that employs a phage,a type of virus that infects bacteria. If ashort random piece of DNA is insertedinto the phage’s gene for a surface pro-tein, the phage will thereafter display onits surface a corresponding random pep-tide. Applying Smith’s method, one cancreate an entire library of phages carry-ing a billion different peptides, with eachindividual phage expressing only onepeptide.

Our innovation was to test the affini-ties of peptides in this library by inject-ing the diverse viruses into a living ani-

How Cancer Spreads76 Scientific American September 1996


MELANOMAOFTEN SPREADSTO LUNGS

COLORECTALCANCER OFTENSPREADS TO LIVER

PROSTATE CANCER OFTEN SPREADS TO BONES

ADHESION MOLECULE

RECEPTOR

STROMAL CELL OF BONE

PROSTATE CANCER CELL

DA

NA

BU

RN

S-P

IZER

PATTERNS OF METASTASIS can beexplained in part by the architecture ofthe circulatory system. Tumors in the skinand many other tissues often colonize thelungs first because the lungs contain thefirst capillary bed “downstream” of mostorgans. In contrast, because the intestinessend their blood to the liver first, the liveris often the primary site of metastasis forcolorectal cancers. Yet circulation is notthe only factor: prostate cancer, for ex-ample, usually metastasizes to the bones.This tendency may result from an affinitybetween receptors on prostate tumor cellsand molecules in bone tissues (inset).

JEN

NIF

ER C

. C

HR

ISTI

AN

SEN


mal. Any phage that carried a peptidewith an affinity for molecules on a par-ticular tissue would stick there. Welooked for and found phages that boundpreferentially to blood vessels in amouse’s brain and kidney. That successsuggests that specific addresses for oth-er organs could also be discovered andtested for their involvement in tumorcell homing.

Knowledge of the addresses that tu-mor cells seek may eventually pay off inclinical benefits. Given the vulnerabilityof tumor cells in transit, anything we cando to make it more difficult for tumorcells to attach to tissues may be benefi-cial to patients.

Initial work in that direction has start-ed. In 1984 Michael D. Pierschbacher,who was then a postdoctoral associatein my laboratory and is now at TeliosPharmaceuticals, and I showed that allcells attach to fibronectin and severalother extracellular matrix proteins at astructure made up of just three aminoacids. This result was surprising, giventhat fibronectin is a long chain of 2,500amino acids. We went on to show thatartificial peptides containing this criti-cal tripeptide (arginine-glycine-asparticacid, designated as RGD) can act like adecoy, binding to cells’ receptors forfibronectin and blocking their attach-ment to the matrix.

Martin Humphries and Kenneth M.Yamada, who were then at the Nation-al Cancer Institute, and Kenneth Olden,then at Howard University, subsequentlyshowed that if they injected mice withcells from melanomas (lethal skin can-cers), RGD peptides could prevent thecells from colonizing the animals’ lungs.Such peptides can even prevent metas-tasis from melanoma tumors grown un-der the skin of mice—an experimental

system that more closely re-sembles the human disease.David A. Cheresh of theScripps Research Institutehas shown that RGD com-pounds can also prevent theformation of new blood ves-sels that nurture tumors. Re-lated compounds thereforemay someday augment phy-sicians’ anticancer arsenal,but much work will have tobe done first so that thesepeptides can be taken orallyand will act longer.

Understanding Invasion

Disappointingly little isas yet understood in

molecular detail about themechanisms that turn a can-cer from a locally growingtumor into a metastatic kil-ler. Some of the same geneticchanges that allow cancercells to escape growth control and avoidapoptosis are clearly important in theearly stages of metastatic spread, be-cause they enable cells to survive with-out anchorage. What then turns on theprograms that make the cancer invasiveand metastatic, however, is not reallyknown.

Genetic approaches similar to thoseused in the discovery of oncogenes andtumor suppressor genes have producedsome candidates for genes with a specificrole in metastasis. Further genetic com-parisons of local and metastatic tumorsmay well explain their differences, but itis also possible that entirely new think-ing is needed.

My own bias is that studying resis-tance to cancer invasion at both the tis-

sue and genetic levels may provide im-portant answers. For example, some tis-sues are not invaded by cancer: cartilageand, to an extent, the brain. Cancersoriginating elsewhere in the body canmetastasize to the brain, but they do nottruly invade the brain tissue—they justgrow bigger within and near the bloodvessels. Something about brain tissueseems to repel otherwise invasive tumorcells. Some species of animals also ap-pear to be unusually resistant to devel-oping cancers. I suspect that much couldbe learned if the molecular bases forthese and other phenomena were un-derstood. The fact that metastasis is thedeadliest aspect of cancer adds the ut-most urgency to our quest for thisknowledge.


The Author

ERKKI RUOSLAHTI is president and chief execu-tive officer of the Burnham Institute (formerly the LaJolla Cancer Research Foundation) in La Jolla, Calif.,and adjunct professor in pathology at the University ofCalifornia, San Diego. Born in Finland, he attended theUniversity of Helsinki, where he received his bachelor’sdegree in 1961, his medical degree in 1965 and hismedical doctorate in immunology in 1967. He hasbeen the recipient of many internationally respectedhonors; in 1995 he was a Nobel Fellow at the Karo-linska Institute and delivered a Nobel Forum Lecture.

Further Reading

Molecular Mediators of Interactions with Extracellular MatrixComponents in Metastasis and Angiogenesis. L. R. Bernstein and L. A. Li-otta in Current Opinion in Oncology, Vol. 6, No. 1, pages 106–113; January1994.

Anchorage Dependence, Integrins, and Apoptosis. Erkki Ruoslahti andJohn C. Reed in Cell, Vol. 77, No. 4, pages 477–478; May 20, 1994.

Evolutions: Metastasis. David I. Lewin in Journal of NIH Research, Vol. 8,pages 82–87; June 1996.

Fibronectin and Integrins in Invasion and Metastasis. S. K. Akiyama, K.Olden and K. M. Yamada in Cancer and Metastasis Reviews, Vol. 14, No. 3,pages 173–189; 1996.

PHAGE BINDING TO TISSUE IN BRAIN

PHAGE BINDING TOTISSUE IN KIDNEY

LIBRARY OF PHAGEWITH DIVERSE RECEPTORS

SA

PHAGE LIBRARY, consisting of billions of virusessporting diverse receptor molecules, can help identifythe area codes of tissues to which cancer cells home.In one experiment, a phage library was injected intoa mouse. Some of the viruses bound uniquely in ei-ther the brain or the kidney.

DIM

ITR

Y S

CH

IDLO

VSK

Y


fundamental understandings how cancer spreads

Documents