• familial adenomatous polyposis
• APC
• modifier genes
• genotype-phenotype association
• colon
• polyps

One of the earliest tumour suppressor genes to be identified was APC. Germline mutations in APC are found in familial adenomatous polyposis (FAP) and second hits lead to the development of often many hundreds of adenomas in the colon and rectum, some of which progress to cancer if untreated. Many sporadic adenomas, and their ensuing carcinomas, show APC mutations, and FAP remains an important paradigm for the commoner sporadic form. Thus recent studies from the Tomlinson laboratory¹ show a very close linear relationship between the macroscopic—or naked eye—count of adenomas in excised FAP colons and the count made microscopically from adenomas occupying one crypt (the unicryptal or monocryptal adenoma, fig 1) upwards. Such a close relationship strongly indicates that progression from microadenomas to macroscopic size is essentially random, that variation in disease severity (number of adenomas) results from differences in the number of microadenomas rather than disease progression, and importantly, that the selective advantages provided by different APC mutations act on tumour initiation rather than progression. A paper in this issue of Gut, also from the Tomlinson laboratory,² analyses the effects of putative modifier genes: the severity of the disease was related to the site of the mutation, as might be expected, but first degree relatives showed polyp counts which were more similar than more distant relatives [see page 420]. These observations indicate that modifier genes influence the severity of FAP, again through tumour initiation. Furthermore, the finding of a constant microadenoma density as the colon is traversed¹ suggests that initiation of FAP adenomas are spontaneous events rather than environmentally produced, which of course has considerable potential implications for sporadic adenomas.

These observations underline the pivotal early events in colonic carcinogenesis: establishment of the mutant clone, its evolution to a microadenoma, and its development into a tumour recognisable by the naked eye. The molecular events associated with these stages are clear: in FAP a second hit in the APC gene is sufficient to give microadenoma development.³ But a further recent article from the Vogelstein laboratory⁴ has drawn on some earlier morphological studies to challenge contemporary concepts of how such mutant cells establish themselves and develop into an adenoma. Struck by the appearances in some early non-FAP adenomas (fig 1), dysplastic cells were seen only at the orifices and luminal surface of colonic crypts. Shih et al determined loss of heterozygosity (LOH) for APC, and nucleotide sequence analysis of the mutation cluster region of the APC gene was applied to microdissected well orientated histological sections of these adenomas. Not surprisingly perhaps, half the sample showed LOH in the upper portion of the crypts and most of these had a truncating APC mutation. Those cases without LOH showed a truncating mutation, again confined to the dysplastic epithelium at the crypt apex. Moreover, these cells showed intense proliferative activity, with nuclear localisation of β-catenin, supporting the presence of an APC mutation in these apical dysplastic cells. Several earlier morphological studies have drawn attention to the same appearances,^5–9 including those in FAP.⁹

This morphological and immunohistochemical profile was apparently virtually always present in nearly every crypt of each adenoma studied. Two models for adenoma morphogenesis were proposed (fig 1): in the first, mutant cells appear in the intracryptal zone between crypt orifices and, as the clone expands, cells migrate laterally and downwards to displace the normal epithelium of adjacent crypts. Alternatively, a mutant cell in the crypt base, classically the site of the stem cell compartment,¹⁰ migrates to the crypt apex where it expands as before (fig 1).

This “top down morphogenesis” has profound implications for concepts of stem cell biology in the gut. Most evidence indicates that crypt stem cells are found at the origin of the cell flux, near the crypt base.¹¹ Their repertoire includes all crypt cell lineages, metaplastic and reparative cell lineages, the genesis of new crypts and, as is widely believed, gastrointestinal tumours.¹² These proposals by Shih and colleagues⁴ either re-establish the stem cell compartment in the intracryptal zone or make the intracryptal zone a locus where stem cells, having acquired a second hit, clonally expand.

Where the concepts of Crabtree and colleagues^1,2 and Shih and colleagues⁴ diverge is in the recognition of the earliest lesion, the unicryptal adenoma, where the dysplastic epithelium occupies an entire single crypt.¹ These lesions are very common in FAP,¹³ and while rare in non-FAP patients, have certainly been described.¹⁴ Here, a stem cell apparently acquires a second hit and expands—either stochastically or more probably because of a selective advantage to colonise the entire crypt. Such lesions are thus clonal.¹³ Similar crypt restricted expansion has been well documented in mice after ENU treatment,¹⁵ and also in humans heterozygous for the OAT (O-acetyl transferase) gene where, after LOH, initially half and then the whole crypt is colonised by the progeny of the mutant stem cell.¹⁶ Interestingly, OAT+/OAT− individuals with FAP show increased rates of stem cell mutation with clustering of mutated crypts.¹⁶

In this scenario, in sharp contrast, the mutated clone further expands, not by lateral migration but by crypt fission, where the crypt divides, usually symmetrically at the base, or by budding (fig 1). In several studies, fission of adenomatous crypts is regarded as the main mode of adenoma progression, certainly in FAP where such events are readily evaluated,^17,18 but also in sporadic adenomas.¹⁹ In fact, the non-adenomatous mucosa in FAP, with only one APC mutation, shows a large increase in the incidence of crypts in fission.¹⁷ Aberrant crypt foci, lesions which are putative precursors of adenomas, which can show k-ras and APC mutations,²⁰ grow by crypt fission^21,22 as do hyperplastic polyps.²³ But this concept does not exclude the possibility that the clone also expands by lateral migration and downward spread into adjacent crypts; this model of morphogenesis is conceptually quite different from that proposed by Shih and colleagues.⁴

Finally, there are other reasons for finding crypts containing a mixture of mutant and wild-type cells—or APC−/− and APC+/− cells. Bjerkes and colleagues²⁴ found crypts harbouring cells staining both positively and negatively for APC protein in FAP although these were not spatially distinct, and were construed as crypts containing at least two stem cell lineages. Moreover, at the margins of FAP adenomas, serial section reconstruction has shown normal crypts in continuity with two or three adenomatous crypts,¹⁷ interpreted as adjacent normal crypts transforming into adenomatous crypts. Since crypts are clonal units,¹³ this would explain the observation that some 75% of microadenomas in an FAP patient and in Min mice appear polyclonal.^13,25

The concept that the severity of the disease, or adenoma number, depends on initiation rather than progression,^1,2 brings these early events into sharp focus. The debate also extends into how clonal patches of dysplasia spread in the colon in ulcerative colitis²⁶—“top down” by lateral migration or “bottom up” by crypt fission, or both? Which management structure prevails will have considerable implications for gut biology.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

(A) A monocryptal or unicryptal adenoma. (B) A three dimensional reconstruction of a unicryptal adenoma (inset) from serial sections, showing the adenoma in blue. Note that the adenomatous epithelium extends to the base of the crypt. (C) The mechanism of crypt fission in the normal colon whereby a crypt divides into two by this fission process. (D) A larger adenoma showing expansion by basal fission and budding. (E) Lateral migration at the margins of an adenoma, with adenomatous epithelium invading crypt territories (reproduced with permission from Shih and colleauges,⁴ copyright 2001 National Academy of Sciences, USA). (F) “Top down models” of adenoma morphogenesis where either a single cell incurs APC inactivation, passes to the top of the crypt and proliferates, or transforms in situ at the top of the crypt. Both concepts lead to expansion of the clone in the intercrypt zone (from Shih and colleauges⁴). (G) How mutated clones expand in the colorectal epithelium by crypt fission.