LKB1 as the ghostwriter of crypt history

Most epithelial tissues self-renew throughout adult life owing to the presence of multipotent stem cells and/or unipotent progenitor cells [14]. The cells of the intestinal epithelium are arranged hierarchically, becoming progressively more differentiated with age and passage along the crypt-villus axis. The relatively straightforward histological set-up and the high proliferative rate has made the mammalian intestinal stem cell compartment one of the most studied of all mammalian stem cell niches [15]. Recently, through the identification of LGR5 as a murine intestinal stem cell marker, the direct analysis of mammalian intestinal stem cells has become possible [16]. These studies show that each intestinal crypt harbors multiple stem cells that together give rise to all other post-mitotic differentiated cells populating the villus. All stem cells generate daughter progeny through continuous self-renewal, and every individual stem cell and its clonal daughter progeny can be considered to represent one lineage in the crypt. Classic studies performed in chimeric animals carrying a mixture of background marker alleles show that stem cell lineages within the intestinal stem cell crypt compartment undergo so-called lineage competition [17]. That is, at birth every crypt in these animals displays a mosaic (polyclonal) composition of marker alleles. However, within several weeks these crypts become monoclonal, retaining just one of the marker alleles. This mechanism of random loss of some lineages with replacement by others is called lineage competition. Elegant studies employing the aforementioned LGR5 stem cell marker have re-affirmed this notion (vide infra). Lineage competition is expected to continue even after the crypt has reached a monoclonal status with respect to its chimeric background. In other words, after one chimeric background has attained dominance, all daughter lineages of this mother lineage will now undergo lineage competition. In time, as a result of random genetic drift, one daughter lineage will again attain dominance, and its daughter lineages will then undergo lineage competition. This is an ongoing cycle of lineage competition within the crypt and this lottery-like process of clonal evolution occurs at steady state in each of the approximately 15 million crypts in the adult human colon. The time between two successive stem cell lineages attaining dominance in one crypt is called crypt niche succession time. Importantly, new potential markers such as genomic methylation changes or somatic mutations arise all the time (Fig. 4). However, as no practical method of prospectively identifying and tracing these markers in the human gut exists, the process of lineage competition is obscured from our eyes.

Recently, Hugo Snippert and co-workers for the first time elegantly visualized the behavior of multiple individual lineages ‘competing’ in the murine intestinal crypt [18]. Their method involved the previously established Brainbow cassette, which recombines in a random fashion to irreversibly label individual stem cells and their future daughters with one of four (green, yellow, red or blue) fluorescent marker proteins. This allows quantification of the number of lineages present over time in individual crypts. Their work firmly establishes lineage competition as the driving force behind clonal evolution. Moreover, it demonstrates that all stem cells in the crypt—and their associated lineages—are equally likely to attain dominance, i.e. all lineages are equipotent. Doug Winton’s group in Cambridge employed a different approach to reach a similar conclusion [19]. Both research groups inferred that contrary to previous thinking the dominant mode of stem cell division in the crypt is symmetric, not asymmetric. Note that on a population level, stem cell turn-over can be analyzed by tracing the loss (or de novo aquisition) of lineages over time, regardless of the dynamics on the level of the singular stem cell with regards asymmetric or symmetric stem cell division. Other recent studies reached a similar conclusion for other stem cell niches, such as the murine epidermal and germ line stem cell compartment [20, 21]. This body of work indicates that lineage competition in the crypt must be understood by investigating clonal evolution from a population perspective. Fortuitously, previous work has paved the way for investigating this complex matter at the human level.

Studies performed by Darryl Shibata and colleagues in FAP patient material previously formulated a statistical approach to crypt niche succession times [22, 23]. These modeling studies are based on the retrospective analysis of genomic differences within a population of intestinal crypt progenitor cells through the concept of a molecular clock. Molecular clocks, in paleontology, record time between a last common ancestor and species divergence based on the random accumulation of genetic changes according to a pre-determined neutral mutation rate between related genomes [24]. Analogously, it should be possible to reveal phylogenies of progenitor lineages within individual crypts based on the genetic changes accumulated over time since progenitor lineages last went through crypt niche succession [25]. A marker under neutral selection is expected to show an increased measure of diversity as lineages diverge over time. Following this reasoning, Darryl Shibata and his co-investigators analyzed the diversity in methylation patterns of the CpG island of the CSX gene—which is not expressed in human colon—within and between individual crypts. These crypt heterogeneity patterns were significantly greater in histologically normal FAP crypts compared to crypts from control patients [23]. This indicates that crypt niche succession times are expanded in FAP crypts, since lineages that persist longer retain a greater number of variants.

Prior to the discovery of the energy-sensing AMPK module as a downstream target, LKB1 had been most prominently linked to the regulation of cellular polarization in model organisms. The alleles encoding the nematode and fly counterpart of the LKB1 tumour suppressor were both retrieved in genomic screens designed to pick mutants defective in cellular polarity regulation [26, 27]. Not withstanding the accumulated data on the LKB1-AMPK module in cellular metabolism [28], these early studies remain some of the strongest evidence linking LKB1 to cellular polarization. Later studies in vertebrate model systems provided additional evidence for a role of LKB1 in cell polarity and directed cell division [29]. Consequently, germline LKB1 mutations may upset stem cell turn-over in PJS patients’ crypts as well. We, therefore, implemented the technique outlined above involving the analysis of CSX crypt diversity patterns on unaffected PJS mucosa and age-matched control crypts. These analyses are difficult to perform because of the relative scarcity of patient material. Nonetheless, through this unbiased technique, we found that PJS crypts retain a statistically increased number of variants when compared to age-matched controls (unpublished data). Thus, akin to FAP, crypt diversity is greater in unaffected PJS crypts indicating that this physiological process of clonal evolution is protracted in PJS. As discussed below, this protracted clonal evolution scenario implicates a cancer-prone state in unaffected PJS mucosa at normal background mutation rates.

Stem cells accumulate mutations stochastically from birth in phenotypically normal epithelia. During life, stem cell populations and their pool of accumulated mutations continuously change as individual lineages become extinct or attain niche dominance. Most mutations in stem cell lineages will be lost, because only one current stem cell lineage attains future dominance [17‐19]. These mutations need not evoke a selective advantage and may be initially neutral. Individual mutations without selective advantage may thus initially ‘hitchhike’ along with the inherent clonal evolution of stem cell niches. The analysis of crypt diversity patterns through neutral markers such as CSX methylation tags is based on this hitchhiking pattern. Eventually though, combinations of mutations confer a visible tumor such as a tubular adenoma. Differences in clonal evolution rate may allow crypt mutations to accumulate at different frequencies even though these mutations arise at a similar rate. This occurs because the rate at which mutations are shed from the crypt depends on the clonal evolution rate [22]. This was also shown in an earlier study comparing crypt-restricted loss of O-acetyltransferase activity between normal colonic epithelia in FAP and Lynch syndrome patients [30]. In this study phenotypically normal FAP epithelium demonstrated more crypts mutant for O-acetyltransferase activity than phenotypically normal age-matched Lynch syndrome epithelium. These findings on O-acetyltransferase activity should extend to other transcripts such as p53 or K-RAS. Thus, in crypts undergoing a protracted clonal evolution scenario such as in FAP [23] or PJS (our unpublished data) genetic clonal diversity is greater.

Differences in the amount of accumulated mutations that persist over time show that the rate of clonal evolution can be protective (or anti-tumorigenic) because mutant stem cells may be lost through lineage competition. Mathematical modelling adapted from population genetics predicts that the chance of loss of novel alleles within a population is given by p_loss = (2 N − 1)/2 N, where N is population size and the chance of fixation of new alleles is p_fix = 1 − p_loss = 1/2 N. In a similar fashion one can derive that the rate of drift (loss or fixation of novel alleles under neutral selection) is inversely proportional to population size (see Table 1 on the dynamics of surnames in Korea). Therefore, population size has a domineering effect on the rate of fixation or loss of novel alleles and this effect becomes more pronounced in smaller populations. Accordingly, with larger population sizes, the time between population bottlenecks or crypt niche succession increases proportionally. A protracted clonal evolution or increased lineage residence time, therefore, decreases the time to cancer presentation by expanding the pool of selectable variants retained over time. Or, put in other words, a protracted clonal evolution will shorten the time to adenoma formation through the increased likelihood of an accidental acquisition of a biallelic gatekeeper mutation. Thus, by definition, demonstration of an expanded progenitor pool in a niche that undergoes lineage competition reflects a cancer-prone state.

We previously proposed that the ‘hamartomatous’ polyps observed in PJS are an epiphenomenon to the cancer-prone condition, and that its characteristic smooth muscle core is indistinguishable from histopathological alterations seen in forms of mucosal prolapse [5]. Analysis of murine PJS polyps in Lkb1 hemizygous mice established independently in several laboratories convincingly shows that the wild-type allele need not be lost during polyp formation [31‐33]. Following our proposal that the polyps are an epiphenomenon to the cancer-prone condition, we are faced with the remarkable situation of a rare polyposis syndrome where polyps and carcinomas occur simultaneously, yet these are not linked in a clinicopathological sense. However, regardless of whether the polyps lack premalignant potential, an understanding of their development will reveal important insight into the co-existent cancer-prone condition, as the molecular mechanism responsible for polyp formation must, under the simplest model, be responsible for the cancer-prone condition in PJS as well.

We previously observed an expanded crypt progenitor zone in normal intestinal mucosa in PJS patients by standard immunohistochemical labeling with the Ki67 proliferation marker [34]. We inferred that germline mutation of LKB1 leads to a defect in the regulation of stem and/or transit amplifying cell turn-over in the mammalian intestine. Importantly, in this study we observed an elongated progenitor tract in the absence of an increased labeling index (that is, the number of Ki67-labeled nuclei divided by the total number of nuclei in the crypt was not significantly different between PJS crypts and matched controls). This shows that, besides the absolute size of the progenitor zone, other basic characteristics of the crypt as a stem cell niche (such as division rate, the pace of maturation and/or differentiation, and niche exit) are unchanged [35]. The observation of an expanded progenitor tract displaying a normal labeling index argues for dysregulation at the level of the intestinal stem cell. Our data derived from the analysis of dynamic CSX methylation patterns in unaffected PJS mucosa are in full agreement with this hypothesis. Definitive analysis of the size of the dedicated stem cell pool in PJS crypts awaits the availability of direct stem cell markers in human tissue.

In PJS crypts the ratio of proliferative over non-proliferative cells remains essentially the same even though more cells (in absolute numbers) are continuously fed onto the villous epithelial carpet [34]. Incipient polyps might be created by stochastic deviations from the normal equilibrium of stem cell divisions. In the study previously discussed by Lina Udd et al. scattered foci of increased Ki67-labeling in unaffected Lkb1 ^+/− gastric mucosa were observed [12]. These stochastic bursts of proliferative activity may form small mucosal elevations which may serve as a leadpoint upon which peristalsis can act. Classically, PJS polyps have been described as numbering from a few to several dozens along the entire length of the gastrointestinal tract. A recent study investigating COX-2 inhibition by sulindac analogues in a small cohort of PJS patients necessitated repeated gastroscopies. This study unexpectedly showed that the gastric epithelium in these patients was in fact studded with ‘hundreds of small hyperplastic lesions’ [36]. Potentially these lesions wax and wane over time (small outpocketings may be extruded or auto-amputated) and never come to clinical attention. In fact, most larger lesions both in the murine system (typically at the level of the pylorus [2]) as well as in the human system (at the level of the descending colon and rectum [1]) occur at sites of greatest mechanical force. This suggests that incipient polyps occur frequently and along the gastrointestinal tract, but a literal tug of war between mucosal elevations and mechanical force would favour a small mucosal elevation to be moulded into a macroscopic polyp. Indeed, an incipient polyp already shares features with mucosal prolapse, in particular smooth muscle proliferation. A small incipient polyp could serve as the leadpoint for further outpouching due to mechanical forces pulling on the nascent polyp. This would culminate in the full-blown macroscopic cauliflower-like lesion with abundant smooth muscle proliferation microscopically. Importantly, in this scenario smooth muscle proliferation is secondary to, and accommodates for, epithelial hyperproliferation. Our observations on epithelial turn-over dynamics in PJS can explain the formation of incipient PJS polyps as small hyperplastic lesions which lack an inherently increased risk of neoplastic transformation.