News & Views | Published:

Cell biology

Tumour jailbreak

New work shows that a cage-like matrix of protein fibres around cells can inhibit the growth of tumours. But cancer cells producing the enzyme MT1-MMP can cleave this matrix and proliferate freely.

Tumour cells resemble hardened criminals. They defy the body's social restraints, altering their behaviour and interactions to proliferate and spread as they please. The body also has potential physical restraints — the three-dimensional mesh-like matrices that provide support for normal cells — and Hotary et al.1 now report in Cell that these can prevent the proliferation of tumour cells in the laboratory. But if the cells are able to produce a specific matrix-cleaving enzyme, they can escape these physical restraints as well, becoming free to change shape and multiply.

One of the purposes of the 'extracellular matrix' is to provide structural support for the cells that sit on top of it. When normal cells start on the path to malignancy, they can proliferate with ease on top of the matrix, in what is effectively a 'two-dimensional' environment. But once cancer cells start to invade the matrix in an attempt to spread to distant organs, they are hampered in several ways. First, because the matrix provides cells with growth and survival signals, which vary between tissues, tumour cells moving into a foreign tissue have to be able to adapt to the new signals2. Second, the cells must be able to proliferate and expand the tumour mass within a resistant 'three-dimensional' matrix that is rich in fibres made of a protein called type I collagen. Third, the body often encases tumour cells in abnormal amounts of crosslinked fibrin protein, from which the cells need to escape3.

Given the defences provided by the extracellular matrix, it is perhaps not surprising that previous studies of cancer mechanisms have implicated a class of protein-cleaving enzymes, or proteases, known as the matrix metalloproteinases (MMPs). These enzymes are thought to be important for tumour-cell invasion and spread, by opening up paths in the matrix through which the cells can move. They have also been proposed to have indirect effects on cell proliferation, by freeing growth factors sequestered in the matrix and by promoting the growth of new blood vessels to supply the tumour with nutrients and oxygen4. It seemed unlikely that the MMPs would have a more direct effect on proliferation: studies of cells grown in regular two-dimensional culture systems (on top of matrix proteins in cell-culture dishes, for instance) have generally failed to find MMP-stimulated proliferation. But a rapidly growing body of literature has underscored the importance of the physical three-dimensionality of the matrix in regulating normal cell behaviour, including cell proliferation5,6,7,8. So Hotary and co-workers1 compared tumour-cell proliferation in two-dimensional and three-dimensional conditions.

The authors found that the expansion of cell colonies inside three-dimensional collagen gels was suppressed. Yet the same cells could proliferate readily on the two-dimensional top of the gel. Vigorous growth inside the confines of such gels occurred only when the cells were engineered to produce one particular MMP — MT1-MMP — out of eight tested. Tumour cells producing large amounts of this enzyme could even grow well in the relatively dense connective tissue in the skin of mice. Yet MT1-MMP production conferred no growth advantages on cells on a two-dimensional substrate.

The enzyme, which is located in the cell membrane, needs its membrane 'anchor' and protease activity to permit growth under three-dimensional conditions. It seems to work by cleaving the surrounding collagen matrix (Fig. 1): when Hotary et al. grew MT1-MMP-producing cells in gels composed of a mutant collagen that could not be degraded, proliferation was blocked. Moreover, surrounding the tumour cells with crosslinked fibrin also suppressed growth unless MT1-MMP was active. These and other studies suggest that this protease is particularly effective in cleaving local collagen or fibrin, without help from other MMPs.

Figure 1: Tumour cells: prisoners and escapees.

a, Hotary et al.1 have found that cancer cells in vitro that are surrounded by a three-dimensional matrix of type I collagen or crosslinked fibrin are prevented from changing shape and proliferating. b, Production of the enzyme MT1-MMP by the cells results in local cleavage of matrix proteins; the cells then become free to change shape and so to proliferate. c, Using a collagen that cannot be cleaved suppresses proliferation, even in the presence of MT1-MMP. Likewise, treating MT1-MMP-producing tumour cells with the inhibitor TIMP-2 results in growth suppression in three dimensions (but not on flat surfaces — that is, in two dimensions).

How does a three-dimensional matrix restrain tumour-cell proliferation? Previous work9 showed that cell shape — which is linked to the organization of the cell's internal 'skeleton' — is important in regulating proliferation. Hotary et al. found that tumour cells trapped in collagen or fibrin gels had an abnormally round shape, with actin filaments (a component of the cytoskeleton) being assembled only in the cell periphery. But cells producing MT1-MMP were able to stretch out and organize robust actin bundles. So the authors suggest that MT1-MMP promotes tumour-cell proliferation by disrupting an otherwise constricting three-dimensional matrix and permitting the changes in cell geometry that are needed for proliferation.

One of the most pressing questions raised by these exciting findings is whether membrane-type MMPs such as MT1-MMP are essential for the growth of human cancers. Further measurements of these enzymes in various tumours will reveal how often their levels are increased. Studies of the behaviour of tumours developed from cells genetically engineered to lack MT1-MMP or other membrane-anchored MMPs should likewise be enlightening. Also, some normal cells, such as human fibroblasts, can proliferate readily in certain three-dimensional matrices6,8. It will be important to compare the effects of known regulatory properties of these matrices, such as internal tension and pliability, on normal and tumour cells.

Could these findings have implications for treating cancer? Initial clinical tests of broad-spectrum MMP inhibitors have been disappointing4, even though they would have been expected to inhibit membrane-anchored MMPs (unless effective concentrations were toxic, causing side effects before they could be useful). Moreover, Wolf et al.10 have shown that cells in which all proteases are inhibited can still migrate in collagen (although the authors used a form of collagen that does not prohibit changes in tumour-cell shape). But Hotary et al.1 point out that the chemical inhibitors used in such studies are not specific to MMPs, but also inhibit another major class of protease, the adamalysins. This additional inhibition might have inadvertently interfered with therapy. So it will be important to develop inhibitors of greater specificity, to establish the importance of membrane-type MMPs in cancer. If these enzymes are indeed crucial, then such inhibitors could also be useful therapeutically. Clinical experience shows how difficult it is to selectively kill every last cancer cell, by chemotherapy or other means. But a strategy of preventing the escape of any surviving cells from a matrix prison might just work.


  1. 1

    Hotary, K. B. et al. Cell 114, 33–45 (2003).

  2. 2

    Frisch, S. M. & Screaton, R. A. Curr. Opin. Cell Biol. 13, 555–562 (2001).

  3. 3

    Dvorak, H. F., Senger, D. R. & Dvorak, A. M. Cancer Metastasis Rev. 2, 41–73 (1983).

  4. 4

    Coussens, L. M., Fingleton, B. & Matrisian, L. M. Science 295, 2387–2392 (2002).

  5. 5

    Bissell, M. J. & Radisky, D. Nature Rev. Cancer 1, 46–54 (2001).

  6. 6

    Cukierman, E., Pankov, R., Stevens, D. R. & Yamada, K. M. Science 294, 1708–1712 (2001).

  7. 7

    Walpita, D. & Hay, E. Nature Rev. Mol. Cell Biol. 3, 137–141 (2002).

  8. 8

    Grinnell, F. Trends Cell Biol. 13, 264–269 (2003).

  9. 9

    Chen, C. S., Mrksich, M., Huang, S., Whitesides, G. M. & Ingber, D. E. Science 276, 1425–1428 (1997).

  10. 10

    Wolf, K. et al. J. Cell Biol. 160, 267–277 (2003).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.