Cancer and intestinal stem cells
Renewing tissues and many cancers are maintained by a small number of long-lived stem cells. Most models of stem cell organisation take account of their longevity and assume that they are stable populations carrying unique identifying characteristics. For decades the assays used to test different cell populations for their ‘stemness’ have appeared consistent with such deterministic models. These assays commonly challenge the ability of cells, separated into discrete populations based on the expression of cell surface antigens, to undergo growth when cultured or engrafted. Cells able to support long-term growth are viewed as being synonymous with stem cells.
However, this interpretation now seems too simplistic. For example: cell fate is likely determined by small changes in the expression of regulatory transcription factors in the context of transcriptional networks; the cell surface signatures of stem cells may not be as stable over time as previously thought; the success of stem cell engraftment may be partly determined by properties of the recipient rather than the transplanted cells (Chang et al., Nature 2008; 453: 544; Quintana et al., Nature 2008; 456: 593). Rather stem cell biology may be driven by stochastic switching between different states in response to variations in the balance of signals coming from complex transcriptional networks. In accordance with this view we have previously demonstrated, by following the dynamics of clonal growth in situ, that intestinal stem cell turnover is a constant and rapid stochastic process that follows a pattern of neutral drift (Lopez-Garcia et al., Science 2010; 330: 822).
Our approach is pragmatic: to identify novel ways of assaying stem cells in situ with respect to the functional end-points that are integral to their biology.
How many stem cells?
In 2010 we thought we knew how many stem cells were responsible for maintaining the intestinal epithelium because we could identify them both morphologically and by their expression of stem cell markers such as Lgr5. However, this was an assumption that remained untested. In seeking independent confirmation we have developed a novel method of continuous labelling that acts to mark individual cells with a reporter by detecting replication errors that can probabilistically cause a frame-shift mutation and reporter gene expression. By subsequent analysis of the resultant clonal patterns we were able to show that there was linear accumulation of mutated clones with age. Moreover, intestinal crypts were either wholly or partly populated by clones. We attempted to infer the stem cell dynamics that produced these patterns. Unexpectedly the mathematical simulations and Bayesian inferences used identified fewer stem cells than recognised previously. By revisiting our original data and performing additional analyses we were able to show that the original assumption that all Lgr5+ cells contributed to clone dynamics was incorrect. Rather only about a half to a third of them are acting to maintain the tissue in homeostasis, indicating that there is redundancy or additional factors required. The insights gained from this alternative approach can be best illustrated by applying it to adenomas arising in the intestine following oncogenic transformation: several hundred candidate stem cells are present in each adenoma gland as suggested by expression of a stem cell specific ‘marker’ but less than 10 are actively responsible for their maintenance (Figure 1).
Figure 1. Photomicrograph of an intestinal adenoma containing a reporter positive clone (magenta). Such intratumour clones arise stochastically by spontaneous mutations and are characteristically few in number and relatively large. Quantitative analyses allows the inference that this is a result of small numbers of tumour stem cells in each gland of the adenoma.
Role of quiescent cells
Label retaining cells, identified by their ability to sequester and retain label, have long been considered synonymous with quiescent stem cells. Using inducible expression of nuclear-localised fluorescent protein (Histone H2B-YFP) we have identified a population of crypt-base cells that appear to divide either very slowly or to be quiescent. Conventional views of stem cell organisation would place these cells as potential long-lived cells acting at the apex of a proliferative hierarchy. However, such an interpretation is not compatible with the dynamics that we have documented: rapid stem cell turnover with neutral drift. It now appears that these cells are committed to become secretory Paneth cells and do not normally contribute to stem cell maintenance. In coming to this conclusion we performed a novel lineage tracing experiment based on Cre-complementation that permitted quiescent cells to be clonally marked and their fate established for the first time. This established that quiescent cells do not normally contribute to the stem cell population. However, they can do so following injury illustrating that they can be recalled to the stem cell compartment. Importantly similar quiescent secretory cells are found in tumours and are also clonogenic under regenerative conditions.
Clonal advantage and biased drift
The observation of neutral drift describes the neutral fate of cells that are randomly marked and with a fate determined stochastically. However the initiation of cancer and associated overgrowth of cells suggests that the fate of pro-oncogenic mutations is not neutral. To investigate the role of such mutations on the early clone dynamics we induced low level clonal recombination using conditional, tamoxifen inducible Cre lines to inactivate the Apc and p53 tumour suppressor genes and to activate an oncogenic version of Kras. A fluorescent reporter was simultaneously activated allowing mutated cells and clones to be tracked in detail. Clones expressing only the reporter acted as controls.
As expected the size distribution of clones expressing only a reporter cassette increased with time in a manner characteristic of neutrality and that is explained by 50:50 probability of survival at each round of stem cell replacement. In contrast clones lacking Apc (either due to heterozygous or homozygous loss) or with activated Kras showed a departure from a neutral fate: surviving clones were larger than expected at all times and more rapidly populated whole crypts. In interpreting these altered clone size distributions we inferred how the normal dynamics might be skewed at the level of individual stem cells at the point of divisions that resulted in loss or expansion events. Dramatic effects were observed for Kras and Apc (homozygous loss) with expansion now favouring these mutations in the ratios of 70:30 and 60:40 respectively.
Strikingly our interpretation shows that although such oncogenic mutations are favoured they are not inevitably predestined to becoming fixed within the tissue. Rather they are still subject to loss due to the stochastic nature of the stem cell replacement. This demonstrates a protective effect of the cell organization and tissue architecture of the intestine. One result of this protection is that the rare loss of two Apc alleles in any one crypt will more often require more than two ‘hits’.
With respect to p53 we observed that simple loss of function of this protein (due to expression of a dominant-negative allele) conferred no advantage in homeostasis. However, p53 deficient clones occurring on a background of chemically induced inflammation and colitis do prove to have a significant growth advantage. This reveals that the context in which pro-oncogenic mutations occur can influence their early fate from the point of appearance in the tissue. In the case of p53, mutations are notably associated with the human cancers arising in patients with colitis.