The RASSF (Ras-association domain family) has recently gained several new members and now contains ten proteins (RASSF1–10), several of which are potential tumour suppressors. The family can be split into two groups, the classical RASSF proteins (RASSF1–6) and the four recently added N-terminal RASSF proteins (RASSF7–10). The N-terminal RASSF proteins have a number of differences from the classical RASSF members and represent a newly defined set of potential Ras effectors. They have been linked to key biological processes, including cell death, proliferation, microtubule stability, promoter methylation, vesicle trafficking and response to hypoxia. Two members of the N-terminal RASSF family have also been highlighted as potential tumour suppressors. The present review will summarize what is known about the N-terminal RASSF proteins, addressing their function and possible links to cancer formation. It will also compare the N-terminal RASSF proteins with the classical RASSF proteins and ask whether the N-terminal RASSF proteins should be considered as genuine members or imposters in the RASSF family.
- N-terminal Ras-association domain family (RASSF7–10)
- tumour suppressor
- ubiquitin fold
Ras proto-oncogenes form part of a superfamily of small GTPases comprising of five families: Ras, Rho, Rab, Ran and Arf . They play a pivotal role in a myriad of cellular processes, including cell growth, apoptosis, adhesion, migration and differentiation [2,3]. Unsurprisingly, defects in Ras signalling can result in disease progression, in particular oncogenesis. Indeed, Ras mutations, resulting in signalling aberrations, frequently occur in human tumours, particularly in pancreatic and lung adenocarcinomas [see the COSMIC (Catalogue of Somatic Mutation in Cancer) database at http://www.sanger.ac.uk/genetics/CGP/cosmic/]. Ras proteins carry out their diverse functions by binding to a broad range of Ras effectors and blocking these interactions has been highlighted as an important therapeutic opportunity that could be exploited for cancer treatments . However, this requires a better understanding of the effector pathways utilized by Ras .
Each Ras effector contains one of a number of Ras-binding domains, an example is the RA [RalGDS (Ral guanine nucleotide dissociation stimulator)/AF6/MLLT4 (mixed-lineage leukaemia translocated to 4) Ras association] domain. This conserved domain is the defining feature of RASSF (Ras association domain family) members. The family now contains ten members (RASSF1–10) which are split into two groups, the classical (RASSF1–6) and the N-terminal (RASSF7–10) RASSF proteins . Members of the classical RASSF proteins have been implicated in a range of biological processes, including the regulation of cell death, cell cycle control and microtubule stability, and they are generally regarded as tumour suppressors. This has prompted great interest in these proteins and there are excellent reviews which mainly focus on the classical RASSF family [6–8] and, in particular, RASSF1A [9–11]. Recently, four other proteins have been added to the family  and renamed RASSF7–10 (Table 1). These N-terminal RASSF proteins represent a new group of potential Ras effectors and they may have important biological functions, some of which could well be distinct from previously studied Ras effectors. They may also have a role in cancer progression. In the present review we will focus on the N-terminal RASSF proteins. We will summarize what is known about this newly described group of proteins and ask if there is any evidence to suggest a role for these proteins in cancer formation. We will also address the question of whether they should be considered as long-lost members or imposters in the RASSF family.
RASSF PROTEINS ARE DEFINED BY THE PRESENCE OF A RA DOMAIN/UBIQUITIN FOLD
The defining feature of the RASSF proteins is the presence of a RA domain. This domain was identified by comparing sequences from different Ras-binding proteins  and is present in over 50 human proteins [see the SMART (Simple Modular Architecture Research Tool) database at http://smart.embl-heidelberg.de/]. However, the RA domain nomenclature is potentially misleading as it implies that a protein with this domain will bind Ras. In fact, the binding affinities of RA domains for members of the Ras family show a huge variation and not all members will bind Ras [13,14]. A good example of a RA domain which does not bind Ras is found in the class IX myosin protein, Myr5 . All RA domains are believed to form a similar three dimensional structure called an ubiquitin fold ; however, the RA domain in Myr5 lacks the positively charged amino acids which are required for Ras binding . It is not surprising that only a subset of RA domains bind Ras, as the sequences of different RA domains are highly divergent . There are also other ubiquitin-fold-containing proteins, such as FERM (4.1/ezrin/radixin/moesin)-domain-containing proteins and ubiquitin, which do not interact with Ras . Another possible cause of confusion is the fact that other Ras effectors such as Raf and PI3K (phosphoinositide 3-kinase) interact with Ras through a RB (Ras-binding) domain. Despite the difference in nomenclature this domain also forms an ubiquitin fold . Thus Raf, PI3K, RASSF proteins, FERM domain proteins and ubiquitin all share a common structural domain and can be considered part of an ubiquitin-fold family . The variation in ability to bind Ras means that a key step in studying RA domain/ubiquitin fold proteins, such as the RASSF family members, is to establish whether the proteins function as Ras effectors, something which will be discussed below.
THE CLASSICAL AND N-TERMINAL RASSF PROTEINS HAVE DIFFERENT DOMAIN ARCHITECTURES
The RA domain/ubiquitin fold of classical RASSF members is found near the C-terminal of the protein, adjacent to a protein–protein interaction domain called the SARAH domain (Figure 1). This domain is named after the three types of proteins that contain it; salvador (also known as WW45 in vertebrates), RASSF and hippo [MST1/2 (mammalian STE20-like kinase 1/2) in vertebrates] . SARAH domains have two α-helices, which form a novel dimeric anti-parallel helix . Dimerization between SARAH domains allows salvador, RASSF and hippo to form homo- and hetero-dimers. RASSF1 and 5 also contain a DAG (diacylglycerol/phorbol ester)-binding domain (Figure 1), known as C1 (protein kinase C conserved region). In RASSF5 (also known as Nore1) the C1 domain can form an intramolecular complex with the RA domain/ubiquitin fold and, when free, bind the lipid phosphatidylinositol 3-phosphate .
The N-terminal RASSF proteins have different domain architecture to the classical RASSFs (Figure 1). The RA domain/ubiquitin fold of the N-terminal members is located at the opposite end to the C-terminal location found in the classical RASSF proteins. The RA domains/ubiquitin folds of the two groups also have quite different sequences which form phylogenetically distinct groups (Figure 2). In addition to the differences in the RA domains/ubiquitin folds the N-terminal RASSF members lack an identifiable SARAH motif [5,17]. However, some caution may be required on this point. The SMART database predicts that RASSF7, 8 and 10 have extensive coiled-coil regions, which, like SARAH domains, can form dimers mediated by hydrophobic residues . Structural studies are required to confirm there is no similarity between the coiled-coils of the N-terminal RASSF proteins and SARAH domains of the classical proteins.
The RASSF7, 8 and 10 genes are all located close to members of the Ras family in the genome [5,21,22]. This suggests that the N-terminal RASSF proteins may have co-evolved with members of the Ras family. We have not found a similar association between the classical RASSF genes and members of the Ras family, so this unusual juxtaposition of a Ras gene and a potential Ras effector represents another distinction between the two groups. The separation between N-terminal and C-terminal RASSF genes is not a recent event, Drosophila and Caenorhabditis elegans have both classical (dmRASSF  and T24F1.3 respectively ) and N-terminal RASSF (Table 1 and ) homologues. The differences between classical and N-terminal RASSF proteins prompted us to suggest they are distinct families, with the N-terminal RASSF proteins representing a new group of RA-domain/ubiquitin-fold-containing proteins .
CLASSICAL RASSF PROTEINS ACT AS TUMOUR SUPPRESSORS
The focus of the present review is the N-terminal RASSF proteins; however, before covering these proteins in detail we summarize what is known about the six classical RASSF (RASSF1–6) proteins. This is not intended to replace comprehensive reviews of the family [6–8], but to allow a comparison with the N-terminal RASSF members.
RASSF1 was originally identified in a yeast two-hybrid screen and the gene was found to reside at chromosome 3p21.3 , a region long suspected to contain at least one tumour suppressor . The expression of one of the RASSF1 transcripts, RASSF1A, was found to be repressed by promoter hypermethylation in lung tumours . Subsequent studies found that RASSF1A was inactivated by methylation in a wide range of tumours (e.g. see ) and it quickly emerged that RASSF1A is one of the most frequently methylated genes in cancer [9–11]. Restoring expression of RASSF1A reduces tumour growth and knocking out Rassf1A in mice causes an increased frequency of tumour formation [28,29]. These studies provide convincing evidence that RASSF1A is a tumour suppressor, which is inactivated in a wide range of cancers. Inactivation by promoter methylation occurs in several other members of the classical RASSF family in human tumour cells, including RASSF2, 4 and 5 [6–8], suggesting that they may also be tumour suppressors. Understanding why RASSF1A and other members of the classical RASSF family act as tumour suppressors has been far from straightforward due to the variety of biological roles they possess. Classical RASSF proteins have been linked to a range of processes, particularly the regulation of apoptosis, cell-cycle progression and microtubule stability.
CLASSICAL RASSF PROTEINS ARE KEY REGULATORS OF APOPTOSIS
RASSF family members have been linked to promoting apoptosis through a number of effectors [6–8]. One group of effectors are the pro-apoptotic kinases MST1 and MST2, which bind members of the classical RASSF family [24,30,31]. Hippo, the Drosophila homologue of MST1, forms part of an important tumour suppressor network which is crucial for growth control [32,33]. Hippo functions by regulating the kinase warts which, in turn, regulates the transcriptional activator yorkie, which controls apoptosis-associated genes. Recent work shows that RASSF1A-induced apoptosis acts via a similar pathway, which involves MST2 activating LATS (large tumour suppressor), causing the release of YAP (yes-associated protein 1) which promotes transcription of p73 . NDR (nuclear Dbf2-related) kinases, which are related to LATS kinases, can also function downstream of MST1 to promote apoptosis . In addition to MST1/2, classical RASSF proteins bind another positive regulator of apoptosis, MOAP1 (modulator of apoptosis 1) [36–38]. After death receptor signalling, MOAP1 and RASSF1A are recruited to the death receptor where the interaction of RASSF1A with MOAP1 allows MOAP1 to activate Bax and promote apoptosis [36,39].
CLASSICAL RASSF PROTEINS ARE IMPORTANT FOR MICROTUBULE STABILITY AND CELL-CYCLE PROGRESSION
A second function of the classical RASSF proteins is to regulate the cell-cycle. Expression of RASSF1A blocks cell-cycle progression at a number of stages, including G1, G2–M and in prometaphase [40–43]. The RASSF1A-induced arrest in G1 is associated with reduced activity of JNK (c-Jun N-terminal kinase)  and AP1 (activator protein-1) , which both promote cell-cycle progression. RASSF1A also up-regulates expression of the cyclin-dependent kinase inhibitor p21Cip1/Waf1 . These effects are likely to be mediated by a number of effectors. RASSF1A can bind MDM2 (murine double minute 2) and DAAX (death-domain-associated protein). This prevents degradation of the tumour suppressor p53, which would allow p53 to promote cell-cycle arrest . RASSF1A can also bind and increase the activity of the transcription factor p120(E4F), a transcriptional repressor of cyclin A2 [48,49].
The RASSF1A-induced arrest in mitosis is tightly associated with the ability of RASSF1A to associate with microtubules. RASSF1A can bind and stabilize microtubules [41,43,50,51], probably via interacting with a number of microtubule-associated proteins [50,52]. Once bound to microtubule-associated proteins RASSF1A appears to function as a scaffold, recruiting multiple regulators of mitosis. Current data suggests these may include MST1/2, CDC20 (cell division cycle protein 20), Aurora-A and Ran. MST1 could signal through the hippo pathway (see above) to regulate mitotic progression . RASSF1A can also bind and inhibit CDC20, which activates the anaphase-promoting complex , although it should be noted that this interaction is controversial . RASSF1A is phosphorylated by the mitotic kinase, Aurora-A , and can regulate the activity of Aurora-A . Finally it has recently been shown that Ran can act as a RASSF1A effector to regulate microtubule stability . In addition to mitosis, RASSF1A has been linked to cell migration, which is consistent with a role in regulating microtubules .
Other members of the classical RASSF family have been linked to cell-cycle progression. An example is RASSF5/Nore1, which shows striking similarities to RASSF1A. RASSF5/Nore1 can associate with microtubules  and suppress growth by a mechanism which involves p53 activating the expression of p21Cip1 . In summary, classical RASSF proteins have been linked to apoptosis, cell-cycle control and the regulation of microtubule stability, all of which may contribute to the tumour suppressor function of these proteins.
CLASSICAL RASSF PROTEINS AND RAS
The presence of a RA domain/ubiquitin fold suggests that the classical RASSF proteins will act as Ras effectors. However, as discussed above, not all RA domains bind Ras, and for many of the classical RASSF proteins it is not clear whether they function as Ras effectors in order to mediate the processes described above. RASSF5/Nore1 is perhaps the best documented Ras effector of the RASSF family. The splice variant RASSF5A/Nore1A was identified as a Ras-interacting protein by yeast two-hybrid screens and the endogenous protein interacts with Ras following addition of EGF (epidermal growth factor) . RASSF5A/Nore1A is also the first member of the RASSF family to have the crystal structure of its RA domain/ubiquitin fold determined . This was carried out in complex with Ras and demonstrated that the region which interacts with Ras is extended compared with other Ras effectors. This lengthened interface provides the RASSF5A/Nore1A–Ras complex with a prolonged lifetime compared with other Ras effectors. However, a physiological role for a RASSF5A/Nore1A–Ras complex has yet to be identified. The splice variant RASSF5B/Nore1B (also known as RAPL) is a Ras effector with a well-documented physiological role in T-cell signalling, where it associates with the Ras protein Rap1 .
RASSF1A can bind Ras in a GTP-dependent manner . However, it binds with a much lower affinity than RASSF5/Nore1 [63,66]. This raises the question, is RASSF1A a genuine Ras effector? An andogenous RASSF1A–Ras complex has been described [66a], but, similar to the RASSF5A/Nore1A–Ras complex, the physiological role of this complex is not known. A recent twist to this story is that RASSF1A can bind to the small GTPase Ran  and regulate microtubule organization (discussed above). The binding appears to be direct and can be seen with endogenous protein. Ran is not a member of the Ras family but is part of the larger superfamily of related small GTPases . This suggests that RASSF1A functions by binding Ran in addition to, or instead of, binding Ras. It also raises the possibility that RASSF1A and other RASSF proteins might bind other small GTPases in addition to Ras and Ran. Future work is required to untangle the biology of the classical RASSF proteins and the role that Ras and other small GTPases play in their function.
THE N-TERMINAL RASSF FAMILY
The difference in domain architecture and sequence of the RA domains prompted us to propose that the N-terminal RASSF proteins are a distinct family from the classical RASSF proteins . This makes the decision to add them to the RASSF family look questionable. However, one crucial benefit of the renaming is to group the N-terminal RASSF proteins together for the first time. This makes it possible to compare what is known about each member. To achieve this we have searched the literature for studies relating to each N-terminal RASSF protein. We used the 11 different names which have been given to the vertebrate members (Table 1); it is important to point out that many of the references cited in the present review use the older nomenclature. Given the importance of the work on the Drosophila classical RASSF protein , we have also looked at the three potential N-terminal RASSF members in Drosophila. In the following sections we will summarize what is known about each of the N-terminal RASSF proteins and, where appropriate, relate that back to what is known about the classical RASSF proteins.
RASSF7: THE FIRST RASSF PROTEIN TO BE DESCRIBED?
The RASSF7 gene was originally identified by a study which set out to sequence genes which are located close to H-Ras in the genome . The authors found an unstudied gene and called it HRC-1 (H-Ras1 cluster 1). HRC-1 was recently renamed RASSF7, presumably because the protein it encodes contains an RA domain/ubiquitin fold and was not part of a recognized family. The hypothesis of the authors who identified RASSF7/HRC-1 was that it might be a growth regulator because it was close to H-Ras in the genome . They also suggested, on the basis of Southern blotting experiments, that RASSF7/HRC-1 might be part of a large family of related proteins. This was six years before RASSF5/Nore1 was identified as a potential Ras effector  and it is only recently that the authors' predictions about RASSF7 have begun to be confirmed.
The genomic position of RASSF7 places it in close proximity to the HRAS1 minisatellite which is immediately downstream of H-Ras. Rare alleles of this minisatellite were shown to be associated with cancer risk [67,68], and it was proposed that altered expression of RASSF7 might contribute to the increased risk . This generated a great deal of interest in the region; however, subsequent studies using improved technology failed to find a link [70,71] and the idea that rare alleles of the minisatellite are associated with cancer risk has fallen from favour.
RASSF7: A HYPOXIC-RESPONSE GENE WHICH IS UP-REGULATED IN CERTAIN CANCERS
The advent of genomic screening technology has made it possible to ‘interrogate’ the entire genome for genes which are misregulated in cancer. Several microarray studies have shown that RASSF7 is up-regulated in cancer (Table 2). One example is in pancreatic cancer. Two independent studies have found that RASSF7 expression is increased in pancreatic ductal adenocarcinoma relative to normal tissue [72–74]. In addition to ductal adenocarcinoma, RASSF7 has increased expression in a second type of pancreatic cancer, islet cell tumours ; the study in fact selected RASSF7 as a key gene whose expression can be used to identify islet cell tumours. RASSF7 expression is also increased in endometrial cancer . Similar to the situation in islet cell tumours, RASSF7 showed a large increase in expression and was selected as one of the top 50 genes that distinguish malignant from normal endometrium . Finally RASSF7 lies in a genomic region which is amplified in ovarian clear cell carcinoma and its expression is increased in these cancers, correlating with the genomic amplification .
Interestingly, recent work offers plausible explanations as to why RASSF7 may be up-regulated in cancer samples. Hypoxia, which occurs in solid tumours, is known to cause a large number of gene-expression changes  and RASSF7 expression was found to be up-regulated by hypoxia in the MCF7 breast cancer cell line  and in human umbilical vein endothelial cells . This predicts that the hypoxic environment found in solid tumours would cause an increase in RASSF7 expression. Furthermore, RASSF7 is also down-regulated by the tumour suppressor, BRCA1 (breast cancer 1), suggesting its expression would be increased in cancer cells which have lost BRCA1 function .
RASSF7 IS REQUIRED FOR CELL DEATH AND PROLIFERATION
An important question is what role RASSF7 plays in these cancerous cells. There is currently no evidence to suggest that increased expression of RASSF7 promotes cancer formation. However, RASSF7 function has been linked to some key biological processes including the regulation of cell death and proliferation. RASSF7 has been shown to be required for necroptosis , a regulated form of necrosis which is distinct from apoptosis. A large-scale siRNA (small interfering RNA) screen was carried out to find proteins required for necroptosis and this identified RASSF7 and RASSF8.
We identified Xenopus RASSF7 in a microarray screen  and subsequently found that in Xenopus RASSF7 is essential for cell-cycle progression and cell survival . In cells where RASSF7 is knocked-down, mitotic spindles fail to form and cells arrest in mitosis. This causes nuclear fragmentation and apoptosis. Consistent with a role in mitotic progression, Xenopus RASSF7 is localized at the centrosome. However, Xenopus RASSF7 is not a core component of the centrosome, rather it appears to be enriched at the centrosome because it interacts with the minus ends of microtubules. Preliminary results from a large-scale screen suggests that the Drosophila homologue may also be required for cell proliferation as knockdown by RNA interference caused a reduction in the mitotic index and weak spindle defects .
RASSF7, like other RASSF proteins, contains no catalytic domain so to understand its function it is crucial to identify the proteins it interacts with. Yeast two-hybrid studies have identified potential binding partners for human RASSF7. These include CHMP1B (chromatin modifying protein 1B), which is associated with endosomal membrane trafficking, and DISC1 (disrupted in schizophrenia 1), which interestingly interacts with microtubules [85,86].
RASSF8 IS LOCATED IN A GENOMIC REGION ASSOCIATED WITH LUNG CANCER RISK
The genomic sequence for RASSF8 was first deposited into the NCBI (National Center for Biotechnology Information) database by Hoon and Yuzuki in 1996 (accession number Q8NHQ8) and was called human carcinoma-associated HoJ-1. However, there appears to have been no publication associated with this submission. Subsequently, RASSF8 was characterized as a gene involved in a chromosomal translocation that is associated with a complex type of synpolydactyly [87,88]. RASSF8 was found to be located on chromosome 12 and referred to as C12orf2. The chromosomal translocation fused RASSF8 with the FBLN1 (fibulin-1) gene (22q13.3). It is believed that this disrupts a FBLN1 splice variant, causing the synpolydactyly. C12orf2 was then renamed RASSF8, presumably because it contains a RA domain/ubiquitin fold and was not part of a recognized family. This occurred at the same time as HRC-1 was renamed RASSF7 and both proteins were added to the RASSF family.
RASSF8 is located approx. 700 kb from the K-Ras2 gene  and both genes lie in a region called Pals1. This region has been identified as a major susceptibility locus in a mouse model for lung carcinogenesis . There are a number of genes in this region, but it is mutations in the K-Ras2 gene that are believed to be responsible for the increased risk . The homologous region in humans has also been associated with increased lung adenocarcinoma risk . However, in humans it is not clear whether it is K-Ras2 that is responsible. Analysis of the region in a Japanese population, identified the D12S1034 microsatellite as being most tightly associated with lung cancer risk . The D12S1034 locus showed a bigger difference between cases and controls than the microsatellite adjacent to K-Ras2. This argues that in some human cancers, susceptibility may be due to a mutation in a gene adjacent to D12S1034 rather than in K-Ras2 itself. RASSF8 lies within 20 kb of D12S1034 making it a good candidate gene, particularly as RASSF8 has also been described as a potential tumour suppressor in lung cancer (see below). However, common polymorphisms in RASSF8 are not associated with cancer risk in an Italian population . Thus it is not clear currently if there is a link between RASSF8 and the increased lung cancer risk associated with the Pals1 region.
RASSF8 IS A POTENTIAL TUMOUR SUPPRESSOR
There are several lines of evidence to suggest that RASSF8 may be a tumour suppressor (altered expression levels are summarized in Table 2). The best characterized example is in lung cancer . In lung adenocarcinoma RASSF8 transcript levels were reduced compared with normal tissue. Overexpression of RASSF8 protein in lung cancer cell lines also inhibited anchorage-independent growth, which has been correlated with tumour progression and metastasis. In addition to lung adenocarcinoma, RASSF8 expression is also down-regulated in male germ cell tumours , despite the fact that the gene lies in a genomic region which shows gain in almost 100% of these cancers. Finally, RASSF8 was identified as a candidate gene involved in leukaemia and lymphoma formation in a study on retroviral-induced blood cancers in mice . This model assumes that oncogenes and tumour suppressors often lie near common retroviral insertion sites. A genomic region next to RASSF8 was targeted seven times, making it one of the most frequently hit sites in the study. This suggests that misregulation of RASSF8 may contribute to leukaemia and lymphoma formation, so it is interesting that RASSF8 has higher expression in human haematopoietic stem cells and is required for blood cell development in zebrafish . It is not known why RASSF8 might be a tumour suppressor, but it is interesting that, similar to RASSF7, it is required for cell death by necroptosis . This form of cell death may be particularly important in cells with deficiencies in their apoptotic machinery, such as tumour cells, so a role in necroptosis would be consistent with a tumour suppressor function.
Mass spectrometry and yeast two-hybrid screens have identified a number of potential binding partners for RASSF8, including the scaffolding protein, 14-3-3γ, which binds phosphoproteins to modulate their function , FRMD4A (FERM-domain-containing 4A), a protein that links membrane domains to actin, and PSMD4 [proteasome (prosome, macropain) 26S subunit, non-ATPase, 4], a component of the proteosome . These potential binding partners offer interesting leads for future work aimed at understanding why RASSF8 may function as a tumour suppressor.
RASSF9 IS A RAS-BINDING PROTEIN THAT HAS BEEN LINKED TO VESICLE TRAFFICKING
RASSF9 was first identified by a yeast two-hybrid screen as a protein that interacted with the cytoplasmic domain of PAM (peptidylglycine α-amidating monooxygenase) . On the basis of this interaction it was originally named P-CIP1 (PAM C-terminal interactor 1). PAM is a transmembrane protein found in secretory vesicles of neurons and endocrine cells, where it catalyses the α-amidation of bioactive peptides such as oxytocin and vasopressin. This modification is essential for the activity of these peptides . We realized that P-CIP1 contains an RA domain/ubiquitin fold and is closely related to RASSF7 and 8 and suggested it should be renamed RASSF9 . The binding of RASSF9/P-CIP1 to PAM was confirmed and RASSF9/P-CIP1 was found to associate with recycling endosomes . This led to the model that it might bind the cytoplasmic domain of PAM during recycling of the enzyme , an interaction that may be regulated by phosphorylation, as the cytoplasmic domain of PAM is known to be multiply phosphorylated . However, RASSF9 mRNA is expressed much more widely than that of PAM, so RASSF9 might have additional roles and perhaps binds other transmembrane proteins.
Interestingly RASSF9 is the one member of the N-terminal RASSF proteins which has been shown to bind Ras proteins. Pull-down experiments with RASSF9 and Ras family GTPases showed that RASSF9 binds N-Ras, K-Ras and R-Ras . An issue that remains to be addressed is whether RASSF9 binds endogenous Ras proteins, or other small GTPases, something which has not been straightforward to answer for the classical RASSF proteins (see above).
RASSF10 IS A CANDIDATE TUMOUR SUPPRESSOR IN CHILDHOOD LEUKAEMIA
We discovered a predicted protein was similar in sequence to RASSF9 and named this protein RASSF10 . This gene was completely unstudied prior to recent work showing that it is a candidate tumour suppressor in childhood leukaemias . The transcript of RASSF10 was characterized and found to be shorter than the predicted version. The protein encoded by the shorter version is more similar to RASSF7–9 and is used in Figure 1. RASSF10 contains a large CpG island and, given the frequent inactivation of classical RASSFs by promoter hypermethylation (see above), the authors examined the methylation status of this gene in childhood leukaemia. They found that RASSF10 was frequently methylated in leukaemia cell lines (100%) and T-cell ALL (acute lymphocytic leukaemia) (88%), but not in normal bone marrow and blood samples. RASSF10 was also rarely methylated in B-cell ALL (16%). Inhibiting this methylation caused an up-regulation of expression in the leukaemia cell lines. These results strongly suggest that RASSF10 expression is inhibited by promoter methylation in a high percentage of T-cell ALL, raising the possibility that RASSF10 might function as a tumour suppressor in these cancers. ESTs (expressed sequence tags) for RASSF10 are present in a number of tissues  and it will be interesting to see whether RASSF10 is methylated in tumours derived from these tissues.
The function of RASSF10 remains unstudied in vertebrates. There is a potential Drosophila homologue (Table 1). However, little is known about this gene except that it is expressed in precursors of the peripheral nervous system , and knocking down its function impairs hedgehog signalling . RASSF10 offers exciting opportunities for future study.
The N-terminal RASSF proteins have a different domain architecture from the classical RASSF proteins and so we proposed that they should be considered as a separate family . Donninger and colleagues came to a similar conclusion for RASSF7 and 8, suggesting that they are a separate sub-family distinct from the ‘true’ RASSF proteins . If the N-terminal and classical RASSF proteins are members of different families then one might expect that there will be little overlap between their biology. However, RASSF7 and RASSF1A show similar centrosomal localization and mitotic defects when knocked-down, and RASSF10 and members of the classical RASSF family both show promoter hypermethylation. These similarities might suggest that the N-terminal RASSF proteins are genuine RASSF proteins. However, we feel the differences between them still outweigh the similarities and that the N-terminal RASSF proteins are not true RASSF proteins, but a separate family. Emerging evidence presented in this review suggests that the N-terminal RASSF proteins might play a role in tumour formation (summarized in Figure 3). There is now an exciting opportunity to study this new group of proteins in more detail and confirm whether they are important in oncogenic progression.
The work of our laboratory is supported by a Marie Curie PhD studentship [grant number MCEST-CT-2005-019822 to A.R.]; a Medical Research Council Co-operative Group Grant [grant number 66812 to A.W.]; a Medical Research Council Fellowship [grant number G120/844 to A.C.]; and by Cancer Research U.K. [grant number C26932/A9548].
We apologize to those whose work we could not cite due to space restrictions. We thank Dr Eric O'Neill (Gray Institute for Radiation Oncology and Biology, University of Oxford, Oxford, U.K.) and Dr Paul Whitley (University of Bath, Bath, U.K.) for critical comments on the manuscript prior to submission.
Abbreviations: ALL, acute lymphocytic leukaemia; BRCA1, breast cancer 1; C1, protein kinase C-conserved region; CDC20, cell division cycle protein 20; FBLN1, fibulin-1; FERM, 4.1/ezrin/radixin/moesin; HRC-1, H-Ras1 cluster 1; JNK, c-Jun N-terminal kinase; LATS, large tumour suppressor; MOAP1, modulator of apoptosis 1; MST, mammalian STE20-like kinase; NDR, nuclear Dbf2-related; PAM, peptidylglycine α-amidating monooxygenase; P-CIP1, PAM C-terminal interactor 1; PI3K, phosphoinositide 3-kinase; RA, RalGDS (Ral guanine nucleotide dissociation stimulator)/AF6/MLLT4 (mixed-lineage leukaemia translocated to 4) Ras-association; RASSF, Ras-association domain family; RB, Ras-binding; SARAH, salvador, RASSF and hippo; SMART, Simple Modular Architecture Research Tool
- © The Authors Journal compilation © 2010 Biochemical Society