CANCER AND CELL ADHESION
The superfamily of cadherins comprises nearly 200 transmembrane proteins that are related by the presence in their extracellular domains of multiple copies of the cadherin motif. A sub-group (T-cadherins) attach to the plasma membrane via a glycosylphosphatidylinositol (GPI) anchor. Through the interactions of identical or closely related sub-types, they mediate cell-cell adhesion (in contrast to integrins that mainly associate with the extracellular matrix). This means that they play important roles in development, e.g. during epithelial-mesenchymal transitions in embryonic growth, and we have already noted changes in the expression of endothelial cadherin during the corresponding pathological change as epithelial cells become malignant. In addition, cadherins can signal intercellular status to intracellular pathways. This is best understood for E-cadherin, expressed on many epithelial cells, which binds β-catenin via a phosphorylated intracellular domain. Through its interaction with α-catenin, β-catenin can regulate the actin cytoskeleton but it is also an indirect regulator of MYC transcription. These activities may explain why the loss of E-cadherin function is associated with tumour cell metastasis and why it is one of the ‘invasion suppressor’ genes.
The integrins are a family of over 20 transmembrane proteins that bind specifically to extracellular matrix proteins (e.g. laminin or fibronectin) as α/β heterodimers (Fig. 1).
1. Signal integration by an integrin scaffold protein.
They respond to signals from the extracellular matrix to modulate cell shape, motility and division. Integrins do not possess enzymatic activity but when activated they recruit cytosolic molecules and can transmit signals bidirectionally. The intracellular tail of integrin-β links to filamentous actin via talin. Talin (or paxillin for some integrin-α subunits) recruits focal adhesion kinase (FAK), a tyrosine kinase that acts as a major scaffold protein for relaying integrin signals. FAK has docking sites for SRC, GRB2, SHC, PLCγ and PI3K and can thus interact with all the major signalling pathways. This includes the angiogenic system of which integrins αvβ3 and αvβ5 are apart and they too signal primarily through the RAS-RAF-ERK1/2 and PI3K pathways.
Integrin signalling is principally mediated by the effectors BCAR1 (p130CAS), NEDD9, CRK, SRCIN1 (p140CAP) and the IPP complex (comprising ILK, LIMS1/pinch and Parvin). These have different tissue expression patterns but they are all scaffold proteins with multiple binding sites for signal propagation. A striking feature of these adaptors is that they receive signals not only from integrins but also from RTKs, cytokine receptors and steroid hormones. They thus integrate a wide range of extracellular signals. The emerging picture of integrin signalling in cancers is of aberrant expression of receptors and adaptors, rather than of mutational changes. Thus, for example, integrins α2, α3, α4, α5, β1 and β2 are over-expressed in melanomas levels of BCAR1, CRK and ILK have been detected in a range of tumours.
FAK (PTK2) over-expression has been found in many types of cancers, particularly in carcinomas but also in haematological malignancies, and there is considerable evidence associating this with metastatic disease. The stimulation of integrin-FAK signalling activates SRC that in turn phosphorylates BCAR1, a step that is required for metastasis to the human breast cancer cells expressing oncogenic RAS or PI3K. This is consistent with the role of FAK/SRC signalling in LKB1-deficient tumours. However, SRC may also be activated by disrupted integrin-FAK signalling (e.g. by ablation of FAK). Although SRC can promote tumorigenesis, as with other oncoproteins, unrestrained SRC signalling can lead to cell death. This implies that cell fate is determined by precise regulation of SRC in focal adhesions (protein complexes that connect the extracellular matrix with the cytoskeleton). The sequestration and destruction of SRC thereby permit cell survival and tumour development as a response to compromised signalling through the integrin-FAK pathway. Inhibition of the autophagic pathway leads to accumulation of active SRC at focal adhesions, which is toxic to the cell. These effects point to the promise combinations that include SRC/FAK inhibitors.
The integrative nature of these systems is illustrated by their direct effects on RTK signalling. Thus, for example, α5β1-integrin regulates re-cycling of the EGFR to the cell surface: the rate is increased through the transcriptional effects of a mutant form of p53, thereby enhancing signalling from the RTK and increasing invasive potential.
The BCR-ABL1 translocation occurs in almost all cases of CML and in about 25% of adult ALL, generating a fusion protein that is a constitutively active tyrosine kinase. In addition, BCR-ABL1 acts as a scaffold for a sequential assembly of protein complexes that are regulated by phosphorylation (Fig. 2).
2. BCR-ABL1 signalling.
This generates a large network of interacting proteins emanating from three sets of adaptors that bind directly to BCR-ABL1: the GRB2, DOK and CRK complexes. Adaptor proteins in the core complexes may also act as scaffolds (e.g. GAB2 binds PI3K and INPPL1/SHIP2) so that BCR-ABL1 may be viewed as a ‘super scaffold’. It is notable that some of the players are familiar from the numerous other scaffolds that we have already encountered and so the story is essentially one of the variations on a theme: these signalling platforms activate pathways that can affect all aspects of cellular function. Signalling from the GRB2 complex generally promotes proliferation whereas that from DOK is inhibitory. SHC1 interacts with both the DOK and GRB2 complexes, accounting for the evidence that it can exert opposing effects on haematopoietic cell proliferation depending on cellular context.
Sonic hedgehog (SHH) is a ~45 kDa precursor auto-catalytically cleaved to release the N-terminal signalling domain (SHH-N, ~20 kDa) and a ~25 kDa C-terminal domain with no known signalling role (Fig. 3).
3. Sonic hedgehog (SHH) signalling.
During processing, cholesterol to the carboxyl end of the N-terminal domain. This modification mediates trafficking, secretion and receptor interaction of the ligand. SHH can signal in an autocrine manner or, after secretion, by a paracrine mechanism (i.e. after secretion it binds to receptors on nearby cells) that requires the participation of the 12-transmembrane protein Dispatched.
SHH binds to the Patched receptors (PTCH1, PTCH2) that, in the absence of ligand, inhibit Smoothened (SMO), a member of the Frizzled family of GPCRs that do not interact directly with their endogenous ligand, SHH. In the unstimulated state, SMO interacts with and is repressed by PTCH. When SHH interacts with PTCH, SMO is released and phosphorylated to activate a signalling cascade that controls the activity of GLI transcription factors.
Hedgehog signalling plays a major role in development but it is abnormally activated in a variety of cancers. In a familiar pattern, the activity of GLI signalling is positively regulated by both the RAS-MAPK and AKT pathways. Activated by Hedgehog signalling include insulin-like growth factor-binding protein, cyclin D2 and osteopontin. The first two of these contribute to cell proliferation: osteopontin promotes malignancy in melanoma cells. Ninety per cent of basal cell carcinomas, the most common skin cancer in humans, have PTCH mutations: in the remaining 10%, SMO mutations block repression by PTCH. PTCH and SMO mutations also, occur in 20% of the malignant brain tumour medulloblastoma. Mutations have not been reported in pancreatic, prostate, colorectal, oestrogen receptor α-negative breast and ovarian cancers but Hedgehog signalling is commonly up-regulated in these cancers.
The small molecule inhibitor GDC-0449 that blocks the PTCH/SMO receptors is in clinical trials for a range of cancers. Drug-resistant metastases have arisen after GDC-0449 treatment as a result of the acquisition of SMO coding mutations that were not present in a primary tumour.
The term ‘transforming growth factor’ was coined when two completely unrelated TGFs (TGFα and TGFβ) were discovered in the medium of transformed cells growing in culture. The transformation had been by infection with a sarcoma virus and the TGFs released could transform normal fibroblasts – that is, make these ‘normal’ cells lose contact inhibition and the requirement for substrate attachment, so that they formed colonies in soft agar. In the fullness of time TGFβ has emerged as one of the most pleiotropic of all molecules – i.e. it causes a wide range of responses in essentially all the major cell lineages. Thus, despite its name, it is a potent inhibitor of epithelial cell growth and its capacity to cause cell cycle arrest, differentiation and apoptosis enable it to play a key role in the development and tissue homeostasis. In these contexts, TGFβ behaves essentially as a tumour suppressor. In addition, TGFβ regulates the response to injury, one component of which is activation of the epithelial to mesenchymal transition (EMT) and the associated activation of cell invasiveness and motility. This step is also generally considered to be part of the induction of metastasis and thus reflects the capacity of TGFβ to promote carcinogenesis. So TGFβ can switch from tumour suppressor to promoter when its growth arrest and apoptotic functions are lost but other signalling responses are retained. This is consistent with the fact that most tumour cells are refractory to growth arrest by TGFβ. How does it do that?
Like all protein hormones, TGFβ binds to its cognate trans-membrane receptors (Fig. 4).
4. TGFβ signalling in endothelial cells.
Rather than RTKs, they are serine/threonine kinases that phosphorylate a family of signal proteins, SMADs that are specific to TGFβ receptors (the canonical signalling pathway). The SMAD family are transcription factors and, depending on the proteins they interact with, they can regulate the expression of at least 100 genes. In normal cells, two major targets of this pathway are MYC, which is repressed, and INK4B, which is turned on by TGFβ signalling. As MYC is essential for cell proliferation and INK4B is a CDK inhibitor, two complementary mechanisms for bringing about cell cycle arrest and indeed apoptosis are activated. Thus TGFβ is a tumour suppressor. However, in contrast with this benign, protective behaviour, abnormal activity of the TGFβ signalling system is widespread in cancer. TGFβ is secreted by a variety of tumour cells and somatic mutations occur in the receptors and in SMADs. Mutations in TGFBR1 are rare but there is a significant incidence of TGFBR2 mutations in the liver and ovarian tumours and in colorectal and gastric carcinomas that are replication error positive (fail to accurately replicate repetitive nucleotide sequences). Colorectal and gastric tumours also have a substantial frequency of SMAD4 mutations. Mutations in this pathway have not been reported in primary breast tumours but have been found in relapses after tamoxifen therapy.
Although some receptor mutations inhibit catalytic activity, their overall effect is to redirect pathway signalling rather than to block it. Thus secreted TGFβ is essential for maintenance of the EMT. Transformed cells retain an active SMAD signalling pathway, consistent with their continued requirement for TGFβ, and expression of SMAD2 is required for tumour extravasation and metastasis. The switch in signalling activates transcription of, for example, matrix metalloproteinases (MMPs) and, in endothelial cells, or angiogenic factors, while selectively inhibiting responses associated with tumour suppressor activity. One mechanism appears to be that TGFβ can also switch on the RAS-MAPK pathway and the changes in cancer may shift the balance in favour of MAPK so that growth inhibitory/apoptotic signalling is overwhelmed by an invasive/metastatic phenotype. Accordingly, TGFβ is also shown in Fig. 4 as a mixed tumour suppressor and oncogene.