Keeping neuroendocrine cells in check: roles for TGFβ, Smads, and menin?

The endocrine cells of the gastrointestinal epithelium sense the luminal contents and through secretions at their basolateral side signal both to other epithelial cells and to subepithelial cells, including smooth muscle, neurones, and inflammatory cells.¹ Some of the features of these cells are clearly neurone-like and for a time it was thought that during development they might be derived, like enteric neurones, from the neural crest. This now seems unlikely, and instead it is thought that normally they arise from the pluripotent stem cells that also give rise to the other epithelial cell lineages.² However, in some circumstances at least, these cells appear to have the capacity for proliferation, and in extreme cases this gives rise to tumours that are called “neuroendocrine” as they exhibit some of the features of neurones and endocrine cells. There are many similarities between neuroendocrine tumours of the gastrointestinal tract and pancreas. In general, these tumours grow slowly and the reasons for this are unknown. Wimmel and colleagues³ now present evidence that transforming growth factor β (TGFβ) is produced by neuroendocrine tumours and through autocrine and paracrine mechanisms restrains tumour cell proliferation [see page 1308].

There are over a dozen major enteroendocrine cell (EEC) types, most with a restricted distribution along the gut.¹ The cellular mechanisms that normally determine the differentiation of these cells, and their numbers relative to other epithelial cells in each region of the gut, are only now becoming clear. For example, the basic helix-loop-helix (bHLH) transcription factor neurogenin 3 is required for the development of intestinal and pancreatic endocrine cells and for the main pyloric antral endocrine cells (G and D cells), but not for endocrine cells of the gastric corpus such as enterochromaffin-like (ECL) and X cells.^4,5 Another bHLH transcription factor, BETA2/NeuroD, is required for the development of intestinal secretin and cholecystokinin cells.⁶ In contrast, the bHLH transcriptional repressor Hes1 is a negative regulator of endocrine cell numbers and in mice with deletion of the Hes1 gene there is hyperplasia of pyloric antral and intestinal endocrine cell populations.⁷

The extent to which the mechanisms determining EEC differentiation also play a part in the EEC hyperplasias found in different clinical conditions remains uncertain. Clear examples of EEC hyperplasia in patients include ECL cell hyperplasia in hypergastrinaemia,⁸ G cell hyperplasia in achlorhydria,⁹ and rectal EEC hyperplasia in Campylobacter enteritis.¹⁰ Of these, ECL cell hyperplasia in the gastric corpus is probably the best understood. Thus hypergastrinaemia in several different clinical settings, including gastrinoma, pernicious anaemia, and prolonged acid suppression with proton pump inhibitors, is associated with ECL cell hyperplasia.^11,12 Similarly, ECL cell hyperplasia occurs in rats with prolonged hypergastrinaemia (either endogenous or exogenous).^13,14 There is direct experimental evidence in the rat to indicate that in hypergastrinaemia ECL cells have the capacity to proliferate.¹⁵ Whether this occurs normally in people is uncertain. However, it is clear that in both patients and experimental animals, hypergastrinaemia is also associated with ECL cell dysplasia and with a tendency to develop ECL cell carcinoid tumours. Moreover, there is evidence that in the setting of pernicious anaemia these tumours may regress after antrectomy compatible with the view that gastrin provides a primary drive to proliferation.¹⁶

At the cellular level, a clue to the mechanisms that might regulate the proliferation of neuroendocrine tumour cells is provided by observations in multiple endocrine neoplasia type 1 (MEN-1). Endocrine tumours of the MEN-1 syndrome may arise in several organs, particularly the pancreas, parathyroid, and pituitary glands. In addition, loss of heterozygosity (LOH) at the locus of the menin gene occurs in about 75% of ECL cell carcinoid tumours in patients with gastrinoma on a background of MEN-1, compared with <15% of patients with ECL cell carcinoids on a background of hypergastrinaemia due to chronic atrophic gastritis.¹⁷ Interestingly, LOH at this locus was not observed in mid and hindgut carcinoid tumours.¹⁷ These observations implicate the product of the menin gene in the inhibition of proliferation of both pancreatic and gastric endocrine tumours. The relevant protein, menin, binds several signalling proteins, including the transcription factors Jun-D and Smad3.^18,19 Smad3 is a downstream mediator of TGFβ signalling, and as loss of menin appears to downregulate Smad3 function, it seems reasonable to suppose that TGFβ might be a negative regulator of proliferation in at least some neuroendocrine tumours. The idea is attractive not least because TGFβ is known to inhibit the proliferation of other cells.

The data reported by Wimmel et al support the idea that TGFβ inhibits neuroendocrine tumour cell proliferation. The authors showed by immunohistochemistry that TGFβ1 was expressed in 50–80% of fore, mid, and hindgut neuroendocrine tumour cells as well as by mesenchymal cells, and that the two relevant receptors, TGFβR I and TGFβR II, were also highly expressed by these tumours. There was similar expression in two neuroendocrine cell lines (BON cells, from a functional human pancreatic neuroendocrine tumour, and LCC-18 cells from a non-functional colorectal neuroendocrine tumour) and in these cells TGFβ was shown to increase p21^(WAF1) and decrease c-myc, causing arrest in the G1 phase of the cell cycle. Moreover, neutralising antibodies to TGFβ, or transfection with a dominant negative receptor, increased proliferation of responsive neuroendocrine cell lines.³

Taken as a whole, these findings provide direct evidence for the importance of TGFβ as a paracrine/autocrine inhibitor of neuroendocrine tumour cell proliferation. The findings are generally compatible with data in other systems that indicate inhibition of proliferation by TGFβ mediated by the Smad pathway and directed at decreased expression of c-myc and induction of p21^(WAF1) and p15^INK4B.^20,21 The role of TGFβ in tumorigenesis is however more complicated. In particular, in other cancers it is now clear that TGFβ can act both as an enhancer of tumour progression as well as a suppressor. The picture emerging over the last few years indicates that TGFβ also stimulates tumour cell migration, promotes epithelial to mesenchymal transition (EMT), and increases the production of matrix metalloproteinases (MMPs); together these effects lead to tumour cell invasion and metastasis.²¹ Interestingly, while the Smad signalling pathway appears to be required for inhibition of proliferation, other signalling systems including the MAPkinase, PI-3-kinase, and protein phosphatase2A/p70s6k pathways are implicated in the pro-oncogenic effects of TGFβ.²⁰ The mechanisms responsible for the shift in TGFβ signalling from a tumour suppressor mode to a tumour enhancer are still unclear. Wimmel et al did not specifically address the question of whether TGFβ stimulates invasion, EMT, or expression of MMPs in neuroendocrine tumour cells. However, as these cells appear to retain the inhibitory effects of TGFβ on proliferation, they may provide a useful model for further studies of the relative importance of the tumour suppressor and pro-oncogenic actions of TGFβ. Recent reports have suggested possible ways to block TGFβ signalling by delivery of soluble TGFβ receptor protein constructs.^22,23 In experimental models, this approach appears to inhibit tumour cell invasion, and so may be valuable in preventing cancer progression. However, because suppression of neuroendocrine tumour cell proliferation by TGFβ appears to be relatively well preserved, a primary objective in this case should be the maintenance and enhancement of this action of TGFβ, and care should be taken before considering whether inhibition of TGFβ is worthwhile.