The development of sporadic colorectal cancer (CRC) involves genetic and epigenetic changes, including allelic losses in specific chromosomal arms, mutations of oncogenes, tumour suppressor genes and mismatch repair genes, micro-satellite instability, and methylation changes in gene promoters [reviewed 1]. Additionally, CRC is frequently associated with altered splicing of tumour suppressor genes, and genes encoding enzymes, growth factors, cytoskeletal and cell adhesion proteins, hormone and growth factor receptors, and transcription factors [see 2345]. Importantly, since different splice variants of a given gene can have different or even antagonistic effects on diverse cellular functions, including apoptosis, proliferation, differentiation, angiogenesis and cell motility, a change in the splice variant ratio may actively contribute to tumour progression. Indeed, positive selection for splice variants that encode isoforms with a selective advantage in tumour progression is of potential diagnostic value and could provide therapeutic targets.

Ras proteins are low molecular weight (~21 kD) GTPases which cycle between the GDP-bound (inactive) and the GTP-bound (active) state at the plasma membrane, and thereby regulate cell growth, apoptosis, motility and differentiation. K-ras activating point mutations occur in about 50% of human sporadic CRC cases and act by stabilising the active GTP-bound configuration, and so promote cellular transformation by constitutive activation of downstream effector pathways, including Raf kinases, phosphatidylinositol 3-kinases (PI3-K), and RalGDS family members [reviewed 6]. K-ras activating mutations play a key role in tumour progression and metastasis in CRC by regulating angiogenesis and protease expression, and cell polarity, adhesion and motility [see 789]. The K-ras gene encodes two splice variants, K-ras 4A and 4B, and activating mutations that usually arise at codons 12, 13 or 61, jointly affect both isoforms [reviewed 6]. Importantly, since K-Ras oncoproteins differentially promote transformation, cell migration, and anchorage-independent growth, they most probably act in a cooperative manner to drive neoplastic progression 10. The ratio of the K-ras 4A/4B splice variants is reduced in human sporadic CRC in both primary adenocarcinomas and colon cancer cell lines that harbour K-ras activating mutations, including homozygous mutations 1112. Since mutationally activated K-Ras 4B has an anti-apoptotic action 1314 and, unlike K-Ras 4A, can promote cell migration 10, and K-Ras 4B can drive expression of matrix metalloproteinase 2 (MMP-2) which specifically cleaves type IV collagen, and is involved in cell detachment and migration 15, the altered splicing of the K-ras oncogene in CRC in favour of K-ras 4B could contribute to neoplastic progression by enabling the survival of cells with DNA damage and facilitating tumour invasion and, ultimately, metastasis. Indeed, tumour growth and metastasis in human CRC is linked with increased expression of MMP-2 [see 7]. However, the finding that the K-ras 4A/4B ratio is also reduced in CRC cell lines that lack K-ras activating mutations raises the possibility that a regulated switch in alternative splicing of the K-ras proto-oncogene may also have a causal role in tumour progression 12. The mechanism could involve increased expression of MMP-2 (see above), and/or reduced apoptosis given that the K-Ras 4A proto-oncoprotein exerts a pro-apoptotic action in mouse intestine following etoposide-induced DNA damage, and evidence that the K-Ras proto-oncoproteins have antagonistic effects on apoptosis in embryonic stem (ES) cells: Ras 4A promotes, whereas K-Ras 4B inhibits, apoptosis 16. Further, while K-Ras 4B, and probably K-Ras 4A, promote ES cell differentiation following withdrawal of leukaemia inhibitory factor 16 it is unlikely they do so with similar efficiency since the Raf/MAPK pathway regulates stem cell differentiation [reviewed 17] and K-Ras 4A and 4B differ in their ability to activate Raf-1 10. Therefore, in accordance with the 'stem cell model' for cancer formation we proposed that a change in the ratio of K-Ras proto-oncoproteins may further contribute to neoplastic progression by perturbing stem cell differentiation [see 12]. Thus, altered splicing of the K-ras proto-oncogene could drive tumour progression in sporadic CRC by promoting MMP-2 expression, and inappropriate stem cell survival and self-renewal.

To address the hypothesis that alteration in the ratio of the K-Ras proto-oncoproteins in favour of K-Ras 4B can affect tumour formation in the small intestine in the absence of K-ras activating mutations, K-ras^{tmΔ4A/tmΔ4A} mice, which express the K-ras 4B splice variant only 18, were crossed with Apc^Min/+ (Min) mice. The latter mice harbour a heterozygous germ-line nonsense mutation in the Apc (adenomatous polyposis coli) tumour suppressor gene, and are predisposed to developing multiple intestinal tumours initiated by loss of the wild-type Apc allele [reviewed 19]. By this means tumorigenesis in the small intestine was compared between Apc^Min/+, K-ras^+/+ and Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A} mice that can, and cannot, express K-ras 4A respectively. This approach was selected since K-ras^{tmΔ4A/tmΔ4A} mice are healthy 1618, intestinal tumours in Min mice do not harbour K-ras activating mutations 20, and K-Ras 4A deficiency does not affect K-ras 4B expression in the small intestine 16 where, importantly, most (> 95%) intestinal tumours form in Min mice [reviewed 19]. Thus, the effect of K-Ras 4A on tumorigenesis can be examined in the absence both of its oncogenic allele and of alteration in K-ras 4B expression as a consequence of K-Ras 4A deficiency.

Both K-ras splice variants are expressed in mouse small intestine 1824. In the present study the relative levels of the K-ras 4A and 4B transcripts were quantified by real-time RT-qPCR analysis. Min mice were examined at 8 weeks-old before they showed any visible sign of intestinal tumorigenesis and, for control purposes, comparisons were made with age-matched wild-type mice. The K-ras 4A/4B transcript ratio did not differ significantly (P = 0.95) between wild-type (1.07 ± 0.09) and Min (1.07 ± 0.15) mice, and in both cases the K-ras 4A transcript accounted for about 50% of K-ras expressed in the small intestine (Figure 1).

The reduced K-ras 4A/4B splice variant ratio in sporadic human CRC involves down-regulation of K-ras 4A expression 11. In Min mice (Figure 2), while there was no significant difference between K-ras 4A and 4B transcript levels in normal small intestine (P = 0.383), tumours of the small intestine showed significantly reduced levels of K-ras 4A relative to K-ras 4B transcripts (P = 0.0015). However, the reduced K-ras 4A/4B splice ratio was due to an increase in the level of K-ras 4B transcript expression, while the level of K-ras 4A transcripts was not appreciably altered. While this provides no support for the hypothesis that K-ras 4A plays a role in modulating intestinal tumorigenesis, it does not exclude this hypothesis, which is worthy of further examination given the high expression levels and established pro-apoptotic action of K-ras 4A in mouse intestine and the antagonistic effects of the 4A and 4B isoforms 16. Therefore, to determine if the K-Ras 4A proto-oncoprotein can modulate tumorigenesis in the small intestine comparisons were made between age-matched Apc^Min/+, K-ras^+/+ and Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A} mice. Since gender may influence tumour development in the Min mouse [see 25] males and females were considered separately. Mice were sampled randomly from the stock at 106 days when the stock was outwardly healthy, and at 152 days when 17% (19/112) of the stock had been either found dead or necessitated necropsy due to sickness [20% (12/59)Apc^Min/+, K-ras^+/+ and 13% (7/53) Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A}]. Examination of the small intestine revealed a similar overall tumour number, and no difference in tumour size for both male and female mice at either time point (Table 2). Histological analysis of intestinal preparations from Apc^Min/+, K-ras^+/+ (n = 8) and Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A} (n = 7) mice with overt signs of neoplasia revealed that all tumours from both genotypes were adenomas of similar, mild to moderate, dysplasia, with no evidence of invasive carcinoma in any of them. There was no statistically significant difference between the genotypes in mitotic counts (Apc^Min/+, K-ras^+/+: 25.3; Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A}: 26.8; P = 0.468) or apoptotic counts (both genotypes: 2.1; P = 1.0) in the tumours. Consistent with the tumour data Apc^Min/+, K-ras^+/+ and Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A} mice exhibited similar survival (Figure 3) for both males (χ²₁ = 1.91, P = 0.167) and females (χ²₁ = 0.44, P = 0.506).

Previous studies have established that intestinal tumours from Apc mutant mice, including adenocarcinomas, do not harbour K-ras activating mutations 202627. In confirmation, and to establish this applies in compound mutant Min mice, K-ras codons 12, 13 and 61 were sequenced in 27 small intestinal tumours, that included 8 tumours from Apc^Min/+, K-ras^+/+ mice, 15 tumours from Apc^Min/+, K-ras^tmΔ4A/+ mice, and 4 tumours from Apc^Min/+, K-ras^{tmΔ4A/tmΔ4A} mice. No tumours were found to harbour a K-ras activating mutation (data not shown).

Because intestinal tumours develop spontaneously in Apc mutant mice they present a powerful model system to study factors that modulate intestinal tumour development, including genetic modifiers, which may be important in the design of potential therapeutic strategies [reviewed 28]. To date, inactivating mutations in genes involved in arachidonic acid and sphingosine metabolism, DNA methylation, and genes encoding the matrix mellalloproteinase MMP-7, VEGF-A and matricellular proteins have been found to suppress intestinal tumorigenesis in Apc mutant mice. Conversely, inactivating mutations in genes related to genomic stability and DNA mismatch repair, and genes that regulate PI3-K/AKT signalling, and proliferation and differentiation in the crypt-villus axis, promote tumorigenesis [see 19252829303132333435363738].

Because homozygosity for the K-ras null allele is embryonic lethal 39 nothing is known about the effect of inactivation of the K-ras proto-oncogene on intestinal homeostasis. The present study found the K-Ras 4A proto-oncoprotein, whose transcript accounts for some 50% of K-ras expressed in the small intestine, does not affect tumour development in the Min mouse intestine where tumours lack K-ras activating mutations. The result is perhaps unexpected given that the K-ras proto-oncogene can promote ES cell differentiation, exert a pro-apoptotic action in ES cells following DNA damage induced by etoposide or cisplatin 164041 and, importantly, can suppress tumour development in the absence 41 as well as the presence 42 of its oncogenic allele. Moreover, the K-Ras 4A proto-oncoprotein exerts a pro-apoptotic action in both ES cells and mouse small intestine following etoposide-induced DNA damage 16. The failure of K-Ras 4A to affect tumorigenesis may reflect the fact that K-Ras 4A deficiency does not affect baseline levels of apoptosis in the crypt 16 and/or abolish completely the apoptotic response in the small intestine following DNA damage, but rather it is reduced and delayed, which implies activation of K-Ras 4A-independent apoptotic pathways. The complex relationship between apoptosis and cancer development is highlighted by recent evidence that the tumour suppressor activity of p53 in radiation-induced lymphoma is not dependent on its ability to promote apoptosis and so eliminate mutant cells 43. Therefore, the K-Ras 4A-mediated pro-apoptotic response in the small intestine following etoposide-induced DNA damage could be irrelevant for tumour formation and, indeed, it does not necessarily follow that K-Ras 4A has a pro-apoptotic action at physiological levels of DNA damage, or with other types of DNA damage. Thus, while the pattern of K-ras 4A expression is strongly conserved in human and mouse tissues 1218 its role in intestinal homeostasis remains unclear. However, given that K-Ras 4B can promote ES cell differentiation and suppress mammary carcinogenesis in the absence of its oncogenic allele 1644 it remains to be determined whether K-Ras 4B can suppress intestinal tumorigenesis. This would necessitate a comparison between Apc^Min/+ mice that can, and cannot, express K-ras 4B. However, since K-ras is essential for mouse development 39 and K-ras 4B is the major splice variant and, unlike K-ras 4A, is expressed ubiquitously 1218, it is doubtful that K-Ras 4B-deficient mice would be viable, and therefore such a study would require conditional inactivation.

In sporadic CRC the K-ras 4A/4B splice variant ratio is reduced in colon cancer cell lines, regardless of whether they harbour K-ras activating mutations 1112. Since the K-ras 4A/4B ratio is reduced in primary CRC tumours 12 the altered ratio in cell lines, including those that lack K-ras activating mutations, is more likely to be symptomatic of the tumours from which they were derived rather than an in vitro artefact of the growth conditions. While altered splicing of the K-ras proto-oncogene in favour of K-ras 4B could, conceivably, modulate tumour progression, and possibly by effects on MMP-2 expression, apoptosis and/or differentiation (see Background), the present study found that alteration in the ratio of the K-Ras proto-oncoproteins in favour of K-Ras 4B (by targeted deletion of K-ras exon 4A) does not affect mouse survival or tumour number, size or histopathology (including mitotic and apoptotic counts). These observations are unlikely to reflect the fact that Min mice die before intestinal tumours can progress, since adenocarcinomas do develop in compound mutant Min mice [see 1929303136]. They could relate to the fact that the altered ratio in human CRC involves not only an decrease in K-ras 4A expression but also an increase in K-ras 4B expression 11: indeed, in Min mouse tumours, K-ras 4B expression is elevated while K-ras 4A expression is not appreciably altered. K-ras 4B expression was unaffected in the comparisons shown in Table 2 and Figure 3 because K-ras 4A deficiency does not affect K-ras 4B expression in the small intestine 16. Given that K-ras 4B, unlike K-ras 4A, can promote cell migration and MMP2 expression 1015, the possibility that increased expression of K-ras 4B, with or without reduced expression of K-ras 4A, is an essential component for tumour progression in cases that lack K-ras activating mutations remains to be addressed. Although the present study establishes that a reduction in the K-ras 4A/4B ratio does not affect Apc-driven intestinal tumorigenesis per se the finding that the ratio is reduced in Min adenomas that lack K-ras activating mutations raises the intriguing possibility that K-ras may have a more widespread role in tumorigenesis in addition to that in lung, colon and pancreatic cancers that normally harbour K-ras activating mutations [reviewed 6]. Thus, given that K-ras 4A and 4B are co-expressed widely in mammalian tissues 1218 it remains to be determined whether their ratio is altered in other types of tumours that routinely lack K-ras activating mutations.

Even though the K-Ras 4A proto-oncoprotein exerts a pro-apoptotic action in the small intestine following etoposide-induced DNA damage, it does not affect tumour development, albeit when expressed in the absence of its oncogenic allele.

The authors declare that they have no competing interests.

CEP, study design, performed experimental studies, drafted the manuscript; MJA & FL, real-time RT-qPCR analysis, histopathology, edited the manuscript; LR, MW, PSD, RLB, NJL & RAR, mouse breeding, genotyping and tumour studies; OJS, contribution to study design, provision of cDNAs, edited the manuscript; MLH, statistical analysis, edited the manuscript. All authors read and approved the final manuscript.