Clone Wars: A New Model
25 September 2018
Stem cell turnover and tissue maintenance is a stochastic process. This means that a randomly occurring mutation has an unknown chance of becoming fixed and spreading within a tissue. Clonal mutations have been observed in apparently healthy tissue, increase in frequency with age and – in some cases – have been described as a pre-malignant state (e.g. clonal haematopoiesis). In certain tissues, such as the colonic epithelium, the contribution of mutations in stem cells to neoplastic transformation remains unclear.
This process is a major interest of Ed Morrissey, who joined the MRC WIMM Centre for Computational Biology in late 2016. His group has recently published a mathematical model that aims to address how functional mutations can contribute to altered stem cell dynamics, with the hope of understanding precisely how these rare mutations accumulate in the lead up to cancer.
The paper builds on work Ed began while working as a post-doc in Cambridge, where he developed a model to explain mutation dynamics in intestinal stem cells in the mouse (Kozar, Morrissey et al, Cell Stem Cell 2013 and Vermeulen, Morrissey, et al Science 2013). This new study applies this model to human colonic epithelium. Within this tissue, stem cells are located at the base of the crypt. Once a mutation has become fixed in an individual crypt, it can expand by fission. While the process is similar between the two organisms, moving from a well-defined mouse model to the human system was no simple task. Human data is much more variable and this additional noise meant that Ed had to carry out substantial optimisation of his existing model. This variability precisely highlights the value of combining the mathematical stem cell model with statistical inference, as it helps extract important insights from apparently noisy data.
To begin to understand how a mutation can impact stem cell dynamics, we need to first understand what fixation would look like for a neutral mutation. A benchmark was created using a visually detectable marker for a known neutral polymorphism (levels of O-acetylation). In a section of colonic epithelium, there can be up to 30,000 crypts; only one or two of these will be positive for a mutation. How can you find these rare crypts?
Ed developed an image processing algorithm to automate the detection of crypts and mutation positive clones and this will soon be released as a tool based on Deep Learning. The age-related increase identified was as predicted by the mathematical model. The rate at which crypts are converted from partial to complete mutation occupancy indicates that this process occurs very slowly – over a median of 6.3 years.
The data gathered from this analysis was fed into the model and used to infer that only a small number of active stem cells (mean = 7) are maintaining the human colonic crypt. These are replaced at a rate of once every 9 months. Therefore, for a neutral mutation, the process of variant fixation occurs over a period of many years and the expansion of neutral clones is expected to be an extremely rare event – 0.7% of all crypts undergo fission in a single year.Having established a mathematical framework and the parameters that govern neutral mutations, the group then looked to use this benchmark to analyse other mutations and check whether the mutation confers a functional effect, something that had been observed in the mouse studies (Vermeulen, Morrissey, et al Science 2013). In collaboration with Doug Winton of CRUK-Cambridge Institute, new visual markers of clonal expansion in the crypt were tested by selecting antibodies for proteins encoded on the X chromosome. Clonal truncating mutations were detected for STAG2, a tumour suppressor that encodes a subunit of the cohesin complex. An age-related increase in crypts wholly occupied by STAG2 mutation was observed. This tells us something. The mathematical analysis using the neutral parameters showed that the mutation rate of STAG2 does not differ significantly from the tested neutral genes; instead, the mutation is predicted to alter stem cell competition dynamics with the mutant stem cells more likely to outcompete the wild type stem cells. The age-related change in mutation-positive patch size was modelled to estimate the crypt fission rate, indicating that crypt fission was elevated 3-fold in the presence of STAG2 mutation versus the predicted rate for neutral mutations.
The altered dynamics of stem cell turnover and clonal expansion in the presence of a functional mutation explains how mutations can reach elevated allele frequencies within a tissue. But is this mechanism specific to STAG2 or can it be observed with other functional mutations? To examine this question, the group looked at KRAS G12D mutation. In the colonic epithelium, elevated levels of the pro-oncogenic KRAS G12D mutation has been described in apparently healthy individuals. Targeted sequencing of colonic tissue was carried out to determine the mutant allele frequencies of all activating mutations at codons 12/13 of KRAS, with mutations detected in the range of 0.2% to 1.8%. Applying the model to this data revealed that these mutant allele frequencies could only be explained by a 10-fold increase in lateral expansion of KRAS mutation positive crypts. While the mutation rate of KRAS is lower than the studied neutral mutations, the fission rate is increased and this results in extensive expansion. This shows that while the mutation itself is rare, it can quickly lead to very large patches of mutated tissue which could enable further mutations to accumulate.
By combining mathematical modelling with biological observation of mutations in the human colon, the Morrissey group were able to study the burden of variant accumulation with age. They showed that pro-oncogenic mutations can significantly alter the dynamics of stem cell replacement and drive accelerated fixation of the mutation and subsequent clonal expansion. They also demonstrated that a mutation takes years to become fixed which may indicate a therapeutic opportunity to intervene prior to the accumulation of mutations that could contribute to the development of colorectal cancer.
This model lays the groundwork for further study of mutation dynamics in stem cells and creates a common framework to enable the comparison of two mutations. Ed plans to continue characterising how mutations alter stem cell dynamics and to try to understand how mutant clones can progress to cancer. One way he plans to do this is to look at patients with conditions that predispose them to colorectal cancer (e.g. Lynch syndrome and FAP). Given the expertise of the MRC WIMM, Ed is also hoping to expand his work into new avenues, such as the blood.
Post written by Hannah Ralph (Wilkie Group)