Tumour stem cells are proposed to be the source of tumour cells. It now emerges that they also give rise to the endothelial cells that line the tumour vasculature, mediating tumour growth and metastasis. See Letters p.824 & p.829
Therapeutics Disease Health and medicine
To grow, solid tumours need a blood supply. They recruit new blood vessels mainly by inducing the sprouting of endothelial cells from external vessels and promoting the cells' migration into the tumour. This ability, called the angiogenic switch, is required for tumour cells to invade surrounding tissue and metastasize to distant sites — the deadly hallmarks of cancer1. In this issue, Wang et al.2 and Ricci-Vitiani et al.3 show that, in addition to recruiting vessels from outside, brain tumours produce endothelial cells for vessel formation from within.
Recent research in tumour biology has focused on two main concepts. According to the first concept — vasculogenic mimicry — some tumour cells take on certain characteristics of vascular endothelial cells and line the tumour's blood vessels4. The origin of such tumour cells is ill-defined: whereas one study5 suggested that tumour stem cells show vasculogenic mimicry, it is generally thought that tumour cells in the immediate environment of the nascent vessel are co-opted for the purpose. The co-opted cells are thought to retain most of their tumour-cell characteristics while acquiring a limited number of endothelial-cell features.
The second concept — that some tumours originate from a tumour stem cell — has been controversial. According to this idea, tumour stem cells are both refractory to most traditional therapies and capable of regenerating the tumour following treatment. The deadly brain tumour glioblastoma is thought to arise from tumour stem cells6.
Wang et al.2 (page 829) and Ricci-Vitiani et al.3 (page 824) now reveal data that are relevant to both concepts, and provide strong evidence that a proportion of the endothelial cells that contribute to blood vessels in glioblastoma originate from the tumour itself, having differentiated from tumour stem-like cells.
Both groups note that a subset of endothelial cells lining tumour vessels carry genetic abnormalities found in the tumour cells themselves (Fig. 1). For instance, a comparable proportion of a cell population expressing endothelial-cell markers and a population of neighbouring tumour cells harboured three or more copies of either the EGFR gene or other parts of chromosome 7. Such cell populations also shared a mutated version of the oncogene p53. Another indicator of the tumour origin of some tumour-vessel endothelial cells is that, as well as expressing characteristic endothelial-cell markers — such as von Willebrand factor and VE-cadherin — they expressed the non-endothelial, tumour marker GFAP.The researchers also present evidence that tumour-derived endothelial cells arise from tumour stem-like cells. They find that a glioblastoma cell population that could differentiate into endothelial cells and form blood vessels in vitro was enriched in cells expressing the tumour-stem-cell marker CD133. Moreover, Wang and colleagues show that a clone of cells derived from a single tumour cell, which expressed CD133 but not VE-cadherin, was multipotent: in vitro, the cells differentiated into both neural cells (which eventually form tumour cells) and endothelial cells.
On being grafted into mice, these cells formed highly vascularized tumours. Moreover, even the progenitor cells from these tumours continued to form tumours and tumour-derived endothelial cells, suggesting that the multipotential characteristic had been maintained. Ricci-Vitiani et al. gained further insights by generating undifferentiated cell aggregates from human tumour-derived CD133-expressing cells and grafting them into mice. The internal vessels of the resulting tumours expressed human vascular markers, whereas more external vessels carried mouse-specific endothelial-cell markers. What's more, the authors found human endothelial cells in tumour vessels linking to the mouse vessels and delivering blood to the tumour.
Wang et al.2 suggest that the differentiation of tumour stem-like cells into endothelial cells might be mediated by signalling pathways involving two proteins — vascular endothelial growth factor (VEGF) and Notch. The authors propose that Notch regulates the initial differentiation of tumour stem-like cells to endothelial progenitor cells, whereas VEGF selectively affects the differentiation of endothelial progenitors to tumour-derived endothelial cells (Fig. 1).
Another team7 has also investigated the source of cells contributing to tumour vessels, and has shown that tumour stem-like cells cultured from human glioma tumours form endothelial cells in vitro. The authors detected channels lined with tumour-derived cells in mice transplanted with human tumours — a process they classify as vasculogenic mimicry. However, their analysis of the original human tumours was limited to marker expression, and so they could draw no firm conclusion about the relationship between the tumour cells and the endothelial cells. Similarly, other groups8, 9 have presented evidence of genetic abnormalities common to tumour cells and endothelial cells, but their data did not distinguish among several potential mechanisms for the observations.
What is the functional significance of a tumour origin for vascular endothelium? To address this question, Ricci-Vitiani et al.3 generated tumours in which the tumour-derived vessels were susceptible to drug-mediated destruction. Following drug treatment, these tumours were smaller than control tumours and had fewer blood vessels. This indicates that blood vessels derived from tumours are crucial for tumour survival.
The new work2, 3 also defines the relationship between a tumour and the blood vessels with which it interacts. If a dedicated compartment of some tumours provides a niche for stem cells that can give rise to functional blood vessels, there may be a less urgent need for tumour cells to undergo the angiogenic switch to recruit vessels, and stronger selective pressure on them to differentiate into endothelial cells.
Moreover, these observations challenge the assumption that tumour endothelial cells are normal cells, and therefore lack the genetic instability that may be the basis of drug resistance in tumour cells. Consistent with this suggestion, earlier studies10, 11 showed that tumour endothelial cells over-duplicate centrosomes — cellular organelles involved in cell division — and possess elevated levels of chromosome abnormalities. Moreover, there seems to be a link between increased activity of the signalling cascades that promote blood-vessel formation and chromosome abnormalities in endothelial cells12. Tumour cells may therefore promote genetic instability in tumour endothelial cells through two distinct mechanisms: by giving rise to them directly, or by sending a signal to a nearby endothelial cell. Thus, not only the tumour compartment, but also genetically unstable tumour endothelial cells, may contribute to drug resistance.
Several compelling questions arise from the latest data2, 3. First, how general is the differentiation of tumour stem-like cells into endothelial cells? Both studies focused on glioblastomas, and so the relevance of this pathway in other tumours of suspected stem-cell origin must also be determined. Other cell types of the underlying support tissue (stroma), such as fibroblasts, also play a part in tumour formation and progression. Do tumour stem cells contribute to these non-endothelial stromal lineages, and, if so, under what conditions?
It is also necessary to define the conditions that promote the differentiation of tumour stem-like cells to endothelial cells, and to determine the prevalence of this process within a given tumour environment. For example, does local shortage of oxygen trigger this differentiation? The present studies examine the molecular pathways that regulate the formation of tumour-derived endothelium at a superficial level. Defining the relevant mechanisms thoroughly is an essential prelude to the design of new therapies.
Finally, it will be crucial to determine how tumour-derived endothelial cells and vessels differ from their non-tumour counterparts in both morphology and function. Other studies13 have reported that, when cultured, endothelial cells isolated from tumours exhibit some properties of stem cells, with the assumption that these properties were acquired by signals from the tumour environment. In light of the present work, an intriguing alternative possibility is that endothelium derived from tumour stem-like cells contributes to the observed cell characteristics. This work2, 3 therefore highlights yet another of the numerous ways in which tumours evade destruction: by contributing to their own support system.