We often forget what marvellous machines our bodies are. Each of our organs is constructed from diverse cell types arranged in a complex, stereotyped pattern, allowing them to carry out their assigned tasks — propelling blood, composing a paragraph, or absorbing nutrients. Perhaps more remarkably, these organs operate continuously for decades, requiring constant remodelling to replace cells lost to attrition. In two landmark papers in Cell1, 2, Clevers and colleagues begin to explain the structure of one organ, the intestine, identifying the architect that directs its development and renewal, contractors that supervise different aspects of the process, and skilled labourers that do the heavy lifting. In doing so they touch on many of the hottest issues in biology, including stem cells and microarrays, as well as perennial favourites such as tumour suppressors and cell-cycle regulators.
The intestine is an interface with the outside world, serving to protect the body and absorb nutrients. During development this organ must produce an array of cell types with different roles and must position each cell properly — a task that is complicated by the harsh environment inside the intestine. As a result, cells live only three to five days, so throughout life the intestine must constantly recreate its complex architecture, with new cells taking the place of their deceased predecessors. Intestinal cells are arranged in a folded sheet known as an epithelium (Fig. 1; reviewed in ref. 3). The bases of the folds, the crypts, contain stem cells and their transiently proliferating daughters. As the cells so produced become specialized (differentiate), most migrate upward along the lengths of the folds — the villi — to form various cell types that absorb nutrients and perform the epithelium's other functions. So-called Paneth cells, meanwhile, migrate downward to the crypt base.
Figure 1: The architects, contractors and labourers in colon construction.
The intestine is lined by a folded epithelial sheet. Proliferating cells are confined to the crypts; most of the differentiated cells are in villi, although Paneth cells are found at the crypt base. a, Clevers and colleagues1, 2 suggest that Wnt proteins, presumably released from mesenchymal cells beneath the crypt, maintain the proliferative potential of crypt cells by turning on the gene encoding the Myc protein, which in turn represses the cell-cycle inhibitor p21. In villi, no Wnt signal is received and p21 mediates cell-cycle arrest, thus allowing differentiation. b, The authors also find that ephrin proteins are expressed on cells of villi and at the top of crypts, whereas their receptors, Eph proteins, are expressed in overlapping patterns in the crypt. The authors thus propose that repulsive interactions between ephrins and Eph receptors allow proliferating and differentiating cells to separate out. c, This idea is supported by the fact that in mice with mutations in ephB2 and ephB3, cells are no longer positioned correctly. (Figure derived from one kindly provided by E. Sancho and H. Clevers.)
High resolution image and legend (57K)How is this complex architecture produced and maintained? Over the years, beginning with genetic analyses in model animals, scientists have identified and examined the function of key molecules that direct developmental decisions, revealing an outline of how cells throughout the embryo choose their fates. As this work has matured, it has become clear — often through the discovery of unexpected connections between developmental regulators and cancer — that the machinery used to establish tissues is also used to maintain them in adults.
One superb example involves the Wnt proteins (reviewed in ref. 4), extracellular signalling molecules that are key to the patterning of diverse tissues in fly and mammalian development. Wnt signals are transmitted into cells by a signal-transduction pathway that ultimately activates gene-expression programmes. This occurs through inactivation of a multiprotein complex, including the tumour suppressor APC, that otherwise targets another protein, 
-catenin, for destruction. Inactivation of this complex stabilizes -catenin, which can then enter the nucleus, where it converts DNA-binding proteins of the TCF/LEF family from gene repressors to gene activators.
Inappropriate activation of Wnt signalling underlies most cases of colon cancer (reviewed in ref. 5). For instance, loss of APC leads to the formation of benign polyps, which become malignant tumours if they undergo additional genetic changes. Without APC, -catenin is stabilized as described above, and Wnt target genes are activated, including the Myc and cyclin D1 genes6, 7, 8, which drive cell proliferation and are thought to help drive tumour formation. These findings suggested that normal Wnt signalling might regulate normal colon cell proliferation. Supporting that idea, mice lacking the DNA-binding protein TCF-4 die soon after birth, and their colons lack crypts, as they fail to maintain intestinal cell proliferation9.
To explore this idea further, the Clevers lab began1 by studying proliferating colon cell lines carrying mutant versions of -catenin or APC, which constitutively activate Wnt signalling. When the authors turned off Wnt signalling in these cells, the cells ceased dividing, consistent with the notion that Wnts are indeed needed for proliferation. Next, Clevers and colleagues looked for regulators of the cell-division cycle that could mediate this switch. Their data point to the protein p21, which inhibits certain enzymes that drive cell-cycle progression. Wnt signalling downregulates p21 expression (Fig. 1a), but when such signalling is blocked, p21 is expressed and causes cell-cycle arrest. In an interesting aside, blocking the cell cycle by overexpressing p21 induces differentiation, suggesting an intriguing coupling between cell division and differentiation. Thus, p21 is the labourer that shapes the cell cycle.
The authors next investigated whether the TCF–-catenin complex supervises p21 directly, or whether it farms out this task to a sub-contractor. The latter appears to be the case. Wnt signals turn on Myc, which encodes a gene-transcription factor. Myc then mediates the proliferative effects of Wnts, apparently by directly repressing p21 expression.
Clevers and colleagues then looked for other effects of Wnts on colon cells, using microarrays to examine the expression of some 24,000 genes in the presence or absence of Wnt signalling1. The response was surprisingly simple — 120 genes were activated by Wnt signalling and 115 were repressed. They next tested the hypothesis that Wnt signalling promotes cell proliferation in colon crypts. This was confirmed in striking fashion. Genes activated by Wnts were expressed in the proliferative compartment of crypts (and are overexpressed in colorectal tumours) (Fig. 1a). Conversely, genes downregulated by Wnts were expressed in specialized cells at the tops of crypts and in villi. Together, these data suggest that Wnts are colon architects, drafting plans that direct the switch from proliferation to differentiation. Wnt signals, presumably emanating from mesenchymal cells beneath the crypt, maintain crypt cells in a proliferative state. Cells exiting this environment are no longer exposed to Wnts, and differentiate rather than proliferate. But how do cells position themselves properly in this complex epithelium? The Clevers group aimed to find out.
The success of a microarray experiment depends on creativity in mining surprising gems from gene lists. Clevers and colleagues noticed that the receptor proteins EphB2 and EphB3 are upregulated by Wnt signalling, whereas their ligand, ephrin-B1, is downregulated2. These proteins have properties (reviewed in ref. 10) that caught the eye of the investigators. Unlike many ligands, the ephrins remain tethered to the cell that makes them. Thus, both ligands and receptors are found on cell membranes, and interactions between them often lead to cell repulsion, so that cells expressing ligand on their surface are sorted out from those expressing receptor. This helps set segmental boundaries in the brain and shapes retinal axon guidance, for example.
Eph receptors are expressed in overlapping regions of crypts, consistent with their activation by Wnt signalling, and ephrins are expressed in a complementary domain in differentiated cells (Fig. 1b). Clevers and colleagues2 hypothesized that repulsive Eph–ephrin interactions might establish a boundary between proliferating and differentiated cells, and thus position cells properly. To test this, they turned to mice with mutations in EphB2 and EphB3, and found that many different cell types were mispositioned all along the crypt–villus axis (Fig. 1c). Proliferative cells were no longer restricted to crypts; differentiated cell types strayed from villi; and Paneth cells left the crypt base.
These data firmly establish Wnt signals as colon architects, identify sub-contractors such as Myc that supervise particular parts of the job, and highlight skilled labourers, such as p21 and Eph receptors, that carry out specialized tasks. They also raise new questions. How much of the programme is run directly by TCF–-catenin and how much is parcelled out to sub-contractors such as Myc? How are proliferation and differentiation coupled? Perhaps most intriguing, how do intestinal stem cells generate progeny with asymmetries in developmental potential, and what are the signals or intrinsic factors that cooperate with Wnts in this process? The known role of Wnt signalling in asymmetric cell divisions in other contexts is intriguing in this regard. Finally, do other tissues use similar mechanisms? Studies of the skin hint that this may be the case. Wnts, presumably emanating from mesenchymal cells at the base of the hair follicle, are key to cell-fate decisions there (reviewed in ref. 11) — it will be interesting to see whether these proteins regulate skin-cell positioning, and whether Eph receptors are involved.
