by ecancer reporter Clare Sansom

Cancer progression is conventionally described using a so-called ‘linear model’ in which the tumour cells acquire new mutations gradually, one by one, eventually becoming able to migrate from the original tumour site and form metastases.

This view, which suggests that cells in a secondary tumour will contain a similar set of mutations to its primary, underlines personalised medicine initiatives such as the US Cancer Moonshot, which recommends understanding advanced cancer through sequencing the primary tumours. 

However, this approach will only be appropriate if the assumption that metastases are genetically similar to the primary tumour at the time of diagnosis is correct.

Recent work has shed doubt on this by suggesting that cancer cells can disseminate through the body soon after an initial tumour arises, and that these disseminated cancer cells (DCCs) can develop independently of the primary tumour – perhaps acquiring a different set of mutations – and may form metastases.

The first evidence for a new theory of cancer development, termed ‘parallel progression’ because it models metastases developing in parallel with the primary tumour, came in studies of human breast cancer.

About 20-30% of patients diagnosed with early, non-invasive breast cancer are found to have disseminated cancer cells in their bone marrow; the fact that about 8% of early breast cancers do recur at distant sites suggests that at least some of these DCCs will have metastatic potential.

Similar results have been observed in mouse models of breast cancer; in one such model, in which cancer development is driven by the over-expression of the Her2 proto-oncogene, DCCs can be observed in the breast cancer at four weeks of age but tumours only become palpable at 18 weeks.

Two studies using this model published back to back in the same issue of Nature have now described some of the molecular mechanisms behind this early migration of breast tumour cells.

The first of these studies, by a large group of researchers led by Christoph Klein of the University of Regensburg and Fraunhofer Institute for Toxicology and Experimental Medicine, Regensburg, Germany looked at gene expression patterns in the epithelial cells lining the breast ducts of these Her2 mice before the tumours had become palpable [1].

Gene expression in these cells was compared to that in healthy mammary glands, more advanced primary tumours and lung metastases.

The gene expression signature of early lesions included increased expression of progesterone receptors, and a similar expression pattern could be stimulated in the early tumour cells in vitro by the administration of progesterone.

However, progesterone receptor expression decreased as the tumours developed into palpable lesions.

The researchers therefore explored the role of progesterone and its receptor in breast cancer development and metastasis further.

They found that the expression of two genes, Rankl and Wnt4, was upregulated in early lesions; these genes are together known as progesterone-induced paracrine signals (or PIPS), and increasing PIPS expression in cells derived from these early lesions mimicked the effect of progesterone.

Progesterone and PIPS both induced migration of tumour cells derived from early lesions and suppressed migration in those from established primary tumours.

PIPS are known to activate mammary stem cells during normal breast development.

The degree of ‘stemness’ of mammary cells correlates with their ability to form clumps of cells termed mammospheres in vitro.

Cells derived from early lesions and primary tumours both form mammospheres when stimulated with progesterone and PIPS, but more are formed from the early lesions.

Chemical suppression of RANLK or WNT4 protein signalling and blocking progesterone with RU486 both inhibited the migratory and sphere-forming response of the early breast cancer cells.

Taken together, these results suggest that HER2 signalling works with progesterone and PIPS to promote epithelial cell proliferation and dissemination in early breast tumours.

Expression levels of the progesterone receptor (PGR) and of HER2 were seen to determine the balance between migration and proliferation, with high HER2 expression correlating with more proliferation and less migration.

The researchers then tested the effect of cell density in culture on HER2 and PGR expression, and showed that PGR expression correlated with low cell density; low cell density also induced other molecular features of early lesions, including low HER2 expression.

These findings suggest a model in which metastatic dissemination of tumour cells is regulated by cell density via HER2 expression and progesterone signalling.

This model was tested in vivo by transplanting mammospheres derived from early lesions and primary tumours into wild type mice of different ages; transplanting early lesions generated DCCs in the bone marrow, whereas transplanting primary tumour cells generated palpable breast tumours.

Transplanting cells from early lesions also stimulated PGR signalling in old (40-week) mice in which this is naturally reduced.

Both cell dissemination and tumour growth were faster in mice with progesterone levels that had been naturally increased through inducing pregnancy.

The metastatic potential of the early and late tumours was assessed by transplanting tissue from the mammary fat pads of transgenic mice (early lesions) and pieces of primary tumour into the fat pads of wild type mice; mice transplanted with fat pads developed primary tumours more slowly but metastases more quickly than those transplanted with primary tumour.

Phylogenetic analysis of primary tumours and metastases indicated that most metastases were derived from cells that disseminated early, before the primary tumours developed 50% of their mutations.

Finally, karyotype analysis of primary tumours and DCCs obtained from breast cancer patients showed a similar pattern, suggesting that these findings will be relevant to the human disease.

The second study, by a group of researchers led by Maria Sosa and Julio Aguirre-Ghiso from Icahn School of Medicine at Mount Sinai, New York, USA, identified a sub-population of invasive cells in early breast lesions in similar HER2 mice.

These cells, which they termed early disseminated cancer cells (eDCCs), originated after breast tumours had initiated on the cellular level but before tumours could be detected.

Overall, both primary tumours and eDCCs showed high expression levels of HER2, as expected, and also low levels of phosphorylated ATF2 (p-ATF2lo) and E-cadherin (E-cadlo); low levels of phosphorylated p38 (p-p38lo) were found in large tumours and disseminated cells.

A similar pattern of protein expression was seen in human breast cancer cells.

Further analysis suggested that expression and phosphorylation of p38 might prevent invasion and that DCCs from early lesions have an Her2 E-cadlop-p38lop-ATF2lo expression profile.

Human and mouse mammary cells that overexpress HER2 are misshapen and may show invasive properties when grown in a 3D-organoid culture.

Reducing HER2 activity in these cells either with an inhibitor, lapatinib, or with a small interfering RNA molecule (siRNA) that binds to the HER2 mRNA and prevents protein expression, restored E-cadherin and p-ATF2 levels and prevented this abnormal behaviour.

Adding a p38 inhibitor, SB203580, eliminated p-ATF2 even in the presence of a HER2 inhibitor, suggesting that increased p-ATF2 levels are dependent on the presence of active p38.

Human and mouse mammary organoids composed of HER2 cells also showed an invasive phenotype without the loss of the cytoskeletal proteins CK8/18.

The researchers then monitored cell dissemination from early lesions in vivo using a HER2 / cyan fluorescent protein (CFP) transgenic mouse model and detected cells migrating into the stroma and entering the lumen of blood vessels of 15 to 18-week-old mice.

Disseminated cancer cells were found in the blood vessels, bone marrow and lungs of these mice from 14-18 weeks of age.

Furthermore, downregulating p38 in early lesions led to a loss of E-cadherin junctions, an increase in beta-catenin levels and induction of the transcription factor TWIST1, which is known to play a role in metastasis.

Beta catenin is a signal transduction protein in the Wnt signalling pathway and can be over-expressed in invasive tumours including breast cancer.

Either upregulating HER2 or suppressing p38 alone upregulates other genes in the Wnt pathway, including TWIST1; combining both changes exacerbates this effect.

These findings suggest that the HER2 eDCCs will activate a Wnt-dependent signalling cascade that resembles the epithelial–mesenchymal transition (EMT) that is necessary for epithelial cells to become invasive.

However, these cells remained dormant or initiated metastasis only slowly, showing that they retained some characteristics of epithelial cells.

Disseminated cancer cells from overt primary tumours varied more in phenotype, with most showing low expression of Twist1.

Taken together, these findings suggest a molecular model of early tumour cell dissemination in which a subpopulation of cells acquire a distinct Her2 CK8/18 Wnthipp38loTwist1hiE-cadlo profile, and that these cells will rapidly disseminate away from the primary site and then eventually – but only slowly – form metastases.

It is interesting that the observed changes in gene expression patterns are fairly similar to those that occur in normal breast development, during ductal branching.

Both these studies can help explain why some patients with ductal carcinoma in situ (DCIS) never develop overt cancer, and why others are first diagnosed with metastatic disease.

This, in turn, may lead to novel, targeted therapies to prevent metastases from breast and other solid tumours.

References
[1]: Hosseini, H., Obradović, M.M.S., Hoffmann, M. and 24 others (2016). Early dissemination seeds metastasis in breast cancer. Nature, published online ahead of print 14 December 2016. doi: 10.1038/nature20785
[2]: Harper, K.L., Sosa, M.S., Entenberg, D. and 11 others (2016). Mechanism of early dissemination and metastasis in Her2 mammary cancer. Nature, published online ahead of print 14 December 2016. doi:10.1038/nature20609
[3]: Ghajar, C.M. and Bissell, M.J. (2016). Nature, published online ahead of print 14 December 2016. doi:10.1038/nature21104