Cues for migration

Lack of oxygen causes the cells of certain tumours to spread to new locations. It also activates a homing mechanism that enables the migrating cells to target specific organs.

The ability of tumour cells to metastasize — to spread to other parts of the body — is perhaps the main reason that certain types of cancer are often fatal. But how do tumours acquire this characteristic? Starving tumour cells of oxygen seems to be one trigger for metastasis, and researchers are beginning to uncover the molecular pathways that underlie this phenomenon1,2. Writing on page 307 of this issue, for instance, Staller and co-workers3 reveal that a gene called CXCR4 is activated by the lack of oxygen, and that this activation causes tumour cells to migrate and to home in on a specific set of organs.

Highly aggressive tumours rapidly outgrow their blood supply, leaving the cells starved of oxygen — a condition known as hypoxia. Tumour cells adapt to hypoxia by increasing their synthesis of a protein named HIF (hypoxia-inducible factor), which in turn binds to and activates several genes4. The proteins encoded by these HIF-responsive genes have a variety of functions. Some increase tissue oxygenation — such as vascular endothelial growth factor (VEGF), which stimulates the outgrowth of new blood vessels — and some enhance cellular glucose uptake and metabolism to allow energy generation when oxygen is scarce.

Our understanding of how levels of HIF are upregulated during hypoxia is growing rapidly4. The protein encoded by the von Hippel–Lindau tumour suppressor gene (pVHL), which is frequently mutated in cancer, is central to this process. The normal VHL protein is part of a complex that, when oxygen is abundant, targets the α-subunits of HIF (HIF-α) for degradation. Recognition of these subunits by pVHL depends on a modification of HIF-α that can occur only in the presence of oxygen. When oxygen is scarce, this modification does not occur and so HIF-α escapes destruction, causing an increase in HIF levels and enhancing expression of the hypoxia-inducible genes. Mutations of the VHL gene produce an effect rather like hypoxia — mutant forms of pVHL found in cancer cannot destroy HIF-α, and as a result, the hypoxia-inducible genes are persistently activated4 (Fig. 1).

Figure 1: The hypoxic response.

a, Under conditions of normal oxygen, the von Hippel–Lindau tumour suppressor protein (pVHL) modifies the protein HIF, which leads to its destruction. b, When oxygen is scarce (hypoxia), or when pVHL is mutated, HIF accumulates inside cells and activates the expression of certain genes. This triggers two complementary responses. First, tissue oxygenation is stimulated through the activation of genes such as VEGF (which stimulates the outgrowth of new blood vessels) and erythropoietin (EPO, which stimulates the production of red blood cells). Second, tumour cells are stimulated to move away from the site of hypoxia through the activation of genes such as c-Met, which enhance cell motility and invasion. Now, Staller et al.3 show that the gene CXCR4 is also activated by HIF. CXCR4 not only stimulates migration, it also enables tumour cells to home in on specific, distant organs.

In the new study, Staller et al.3 searched for genes that are regulated by pVHL. They introduced VHL into renal carcinoma cells (which lack a normal copy of this gene) and then, using DNA microarray analysis, they looked for changes in the activity of thousands of other genes, under non-hypoxic conditions. Unexpectedly, they found that normal pVHL dramatically reduced the production of a receptor protein called CXCR4. This receptor binds chemokines — secreted proteins, rather like growth factors, that allow migrating cells (immune cells, for example) to navigate to specific organs5. The binding of chemokines to receptors such as CXCR4 on the surface of migrating cells stimulates both cell adhesion and motility, and causes the cells to move towards the source of the chemokine.

But how does pVHL regulate the production of CXCR4? As the authors expected, it does so by downregulating HIF. The authors found a functional HIF-binding site in the regulatory region of the CXCR4 gene. They also found that cells containing normal pVHL produced more CXCR4 when exposed to hypoxia. So CXCR4 seems to be a bona fide hypoxia-inducible gene.

The connection between CXCR4 and hypoxia is revealing, because several studies have indicated that 'chemoattraction' through CXCR4 contributes to organ-specific metastasis in certain forms of cancer. For instance, human breast cancer cells often contain high levels of CXCR4, and these cells preferentially metastasize to sites that produce large amounts of the chemokine SDF-1α (the binding molecule, or ligand, of the CXCR4 receptor), such as the lungs and bone marrow6. In fact, CXCR4 is part of a small set of genes that cooperate to promote bone metastases from breast cancer7.

But why does hypoxia induce CXCR4? Presumably, it doesn't matter to a tumour cell where it migrates. An answer to this question can be deduced from the characteristics of mice that are genetically engineered to lack this gene. Such mice show defects in the branching and/or remodelling of certain blood vessels8. So the activation of CXCR4 in blood-vessel cells might be part of an integrated hypoxic response that allows both the generation of new vessels (through the induction of VEGF) and the remodelling of existing vessels (through CXCR4 induction). The finding that VEGF induces expression of CXCR4 supports the view that CXCR4 plays a part in remodelling the vasculature during hypoxia9.

CXCR4 might be needed to promote the survival of tumour cells in a hypoxic environment10, and it enhances cell motility, allowing the tumour cell to migrate away from areas of low oxygen. Hypoxia-induced metastasis also occurs through other signalling pathways — for instance, the gene encoding the c-Met receptor was recently identified as being hypoxia-inducible1. This receptor protein enhances both cell motility and invasion through binding its ligand, hepatocyte growth factor. But the CXCR4 pathway is different because it enables the migrating cells to navigate to specific organs. So, apart from the well-known effects of hypoxia on blood-vessel growth, it also seems to trigger a second and complementary response, enabling cells to migrate away from areas of low oxygen and to home to specific, distant organs (Fig. 1). That CXCR4 expression stimulates homing to distant sites is probably not relevant to the physiological response to hypoxia, but only an unfortunate side effect of tumour-cell hypoxia. This side effect could, however, explain the generally worse prognosis of patients with a hypoxic tumour2. Consistent with this, Staller et al. show that clear-cell renal cancers that harbour a mutant form of VHL express high levels of CXCR4, which correlates with poor survival3.

The new study also contributes to the continuing debate over whether the ability of tumour cells to metastasize is acquired early or late11,12 (Fig. 2). The data of Staller et al. enable us to envisage how tumour cells might be poised early on to spread to other parts of the body. Incipient tumour cells that acquire a mutation in the VHL gene early on may be predestined to spread to secondary sites at a later stage — when they have acquired additional growth-promoting mutations — as the activation of genes such as c-Met and CXCR4 endows them with the ability to migrate, invade and home in on specific tissues. So as well as providing clues about how tumours metastasize, Staller and colleagues' findings edge us closer to an understanding of the timing of tumour progression.

Figure 2: The development of metastatic capability in cancer cells.

a, The classical view is that tumour cells with a high metastatic capability arise at low frequency within the primary tumour. These rare variants give rise to metastases. b, c, DNA-microarray-based gene-expression profiling has indicated, however, that some tumours are homogeneously of 'good prognosis' and others are uniformly of 'poor prognosis'. One way of explaining these data is to assume that some tumours start off on the wrong foot, their initial mutations predisposing them to become highly metastatic after additional mutations have been acquired12,13. Other incipient tumours have initial mutations that make progression to a metastatic phenotype unlikely. The work of Staller et al.3 provides support for the situation indicated in c. Mutations in the VHL gene may not only stimulate uncontrolled cell division (an early step in cancer development), but also, as the authors show, promote cell motility, invasion and homing to distant organs.


  1. 1

    Pennacchietti, S. et al. Cancer Cell 3, 347–361 (2003).

  2. 2

    Bottaro, D. P. & Liotta, L. A. Nature 423, 593–595 (2003).

  3. 3

    Staller, P. et al. Nature 425, 307–311 (2003).

  4. 4

    Kaelin, W. G. Jr Nature Rev. Cancer 2, 673–682 (2002).

  5. 5

    Homey, B., Muller, A. & Zlotnik, A. Nature Rev. Immunol. 2, 175–184 (2002).

  6. 6

    Muller, A. et al. Nature 410, 50–56 (2001).

  7. 7

    Kang, Y. et al. Cancer Cell 3, 537–549 (2003).

  8. 8

    Tachibana, K. et al. Nature 393, 591–594 (1998).

  9. 9

    Salcedo, R. et al. Am. J. Pathol. 154, 1125–1135 (1999).

  10. 10

    Zeelenberg, I. S., Ruuls-Van Stalle, L. & Roos, E. Cancer Res. 63, 3833–3839 (2003).

  11. 11

    Bernards, R. & Weinberg, R. A. Nature 418, 823 (2002).

  12. 12

    Hynes, R. O. Cell 113, 821–823 (2003).

  13. 13

    Bernards, R. & Weinberg, R. A. Nature 419, 560 (2002).

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.