PIK3CA, the gene encoding the p110α catalytic subunit of PI3K (phosphoinositide 3-kinase), is mutated in approximately 20% of sporadic CRCs (colorectal cancers), but the role of these mutations in the pathogenesis of CRC remains unclear. In the present study we used a novel mouse model to investigate the role of the Pik3caH1047R mutation, the most common PIK3CA mutation in CRC, during the development and progression of intestinal cancer. Our results demonstrate that Pik3caH1047R, when expressed at physiological levels, is insufficient to initiate intestinal tumorigenesis; however, in the context of Apc (adenomatous polyposis coli) loss, which is observed in 80% of CRCs and by itself results in benign intestinal adenomas, the Pik3caH1047R mutation promotes the development of highly aggressive and invasive adenocarcinomas in both the small and large intestines. The results of the present study show that an activating Pik3ca mutation can act in tandem with Apc loss to drive the progression of gastrointestinal cancer and thus this disease may be susceptible to therapeutic targeting using PI3K pathway inhibitors.

INTRODUCTION

CRC (colorectal cancer) is the third most common cancer worldwide with poor prognosis and survival accounting for more than 600000 deaths globally in 2008 [1]. Most CRCs undergo a well-described multistage pathogenesis, known as the adenoma–carcinoma sequence, involving the progression of benign adenomas to carcinoma over 10–17 years [2]. Investigation into the underlying molecular mechanisms and genetic events occurring at each stage of pathogenesis has revealed this process to be multifactorial, involving chromosomal abnormalities, loss of tumour-suppressor genes and activation of oncogenes [3].

FAP (familial adenomatous polyposis) accounts for ~1% of all CRC cases and is the result of inherited mutations in the APC (adenomatous polyposis coli) tumour-suppressor gene [4]. APC mutation leads to the formation of many benign dysplastic polyps and adenomas in the small and large intestines and a 90% lifetime chance of progression to CRC [3]. APC loss is also detected in ~80% of sporadic CRCs making it one of the most common initiating genetic alterations in CRC [5]. The molecular mechanisms driving the progression of APC-initiated adenomas to adenocarcinomas are poorly understood.

The PI3K (phosphoinositide 3-kinase)/Akt pathway plays a critical role in regulating numerous cellular processes that are vital in cancer development, including growth, proliferation, migration, apoptosis and survival [6]. Mutations in PIK3CA (PI3K catalytic, α polypeptide), the gene encoding the p110α catalytic subunit of PI3K, have been identified in many human cancers including 20–30% of sporadic CRCs [7,8]. Approximately 80% of these mutations cluster in three hotspots: two in the helical domain (Glu542 and Glu545) and one in the kinase domain (His1047) [7,8]. PIK3CA mutations have been detected in CRCs, but not adenomas [8], suggesting their involvement in the later stages of CRC progression.

We have generated mice harbouring the most commonly detected PIK3CA mutation in human cancer (H1047R) conditionally knocked into the endogenous Pik3ca locus to explore the consequences of Pik3ca mutation in the gastroin-testinal tract. In a system ideal for mimicking sporadic disease, we demonstrate that the Pik3caH1047R mutation, by itself, is not tumorigenic. However, when expressed in combination with the loss of Apc gene function, we observe invasive intestinal adenocarcinomas. This demonstrates a direct link between mutations in the PI3K/Akt and Wnt/β-catenin pathways and the development of these adenocarcinomas.

EXPERIMENTAL

Mice

Mice harbouring a single latent Cre-inducible knockin of the Pik3caH1047R mutation (Pik3caLat−H1047R) [9] and/or two Cre-inducible Apc loss-of-function alleles (Apc580S/580S) [10] were crossed with mice heterozygously expressing the RU486-activatable Cre–progesterone receptor ligand-binding domain fusion knocked into the intestinal epithelial cell-specific Gpa33 (glycoprotein A33) locus (Gpa33CrePR2) [11]. The mice generated (equal numbers of male and females) expressed heterozygous mutant Pik3caH1047R and/or homozygous Apc loss of function (now Apc580D/580D when alleles were deleted) in the gastrointestinal tract. Transgenic B6;129S4-Gt(ROSA)26Sortm1Sor/J (RosalacZ) reporter mice were used for verification of Cre-mediated recombination [12]. All mice strains were maintained on C57BL/6 backgrounds.

All animal experiments were approved by the Peter MacCallum Cancer Centre Animal Experimental Ethics Committee and conducted in accordance with the NHMRC (National Health and Medical Research Council) Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. Mice were closely monitored and humanely killed when they exhibited >10% mass loss, rectal bleeding or other signs of distress.

Induction of conditional Cre expression

To induce recombination, all mice were fed on chow containing RU486 at a concentration of 20 mg/kg of body mass (dissolved in ethanol, mixed with peanut oil and added to mouse food softened with sucrose water) for 4 weeks from 6 weeks of age.

β-Galactosidase staining

Intestinal tissues from Gpa33CrePR2:RosalacZ and RosalacZ control mice were harvested, rolled, fixed, embedded in optimal cutting temperature medium for cryosectioning and stained for expression of β-galactosidase as described previously [13].

Histology and immunohistochemistry

The small and large intestines of mice were removed and flushed with PBS. A small piece of the proximal end of each, and any tumours if present, were snap-frozen for RNA extraction. The remaining intestinal tissues were fixed as a ‘Swiss roll’ on surgical tape in 10% neutral-buffered formalin for 24 h before paraffin embedding.

Sections were stained with H/E (haematoxylin and eosin) for histological analysis or used for IHC (immunohistochemistry). Primary antibodies for IHC included anti-β-catenin (BD Biosciences), anti-phospho-Akt (Ser473) (catalogue number CS3787), anti-phospho-Akt (Thr308) (catalogue number CS4056) and anti-phospho-RPS6 (ribosomal protein S6; catalogue number CS2211) (Cell Signalling Technology).

Sequencing

To confirm knockin of the Pik3caH1047R mutation, RNA was extracted from tissues and tumours and reverse-transcribed to cDNA which was sequenced with Pik3ca-specific primers (forward, 5′-CAAGAGTACACCAAGACCAGAGAGTT-3′ and reverse, 5′-TCCAATCCATTTTTGTCGTCC-3′) using PCR cycle conditions of 95°C for 30 s, 55°C for 30 s and 72°C for 45 s for 35 cycles.

RESULTS

Confirmation of Gpa33-driven Cre-mediated recombination in the intestinal tract by β-galactosidase expression

The efficiency of Gpa33-driven Cre-mediated recombination, and thus conditional induction of Pik3caH1047R expression in the gastrointestinal tract, was assessed using the Cre-inducible Rosa26lacZ reporter mouse strain (Figure 1A). Cre recombination was observed in 1–5% of crypts in the small and large intestines, whereas no staining was detected in other tissues such as the liver (Figure 1B). As the mouse intestinal epithelium renews every 4–5 days, only recombination that occurs in the stem cell at the base of the crypt is reflected in positive β-galactosidase staining beyond a few days.

Confirmation of Cre excision using the Rosa26lacZ reporter mouse

Figure 1
Confirmation of Cre excision using the Rosa26lacZ reporter mouse

(A) Schematic diagram of the genetic strategy used to assess lacZ expression using the lox-stop-lox Rosa26lacZ mouse strain crossed with the RU486-inducible Gpa33Cre mouse strain. Upon exposure to RU486, Cre excises the floxed stop codon only in intestinal cells which express Gpa33 resulting in lacZ expression (blue) in these cells which is detectable by β-galactosidase staining. (B) lacZ expression in tissues 90 days post-induction of Cre expression by RU486 from a reporter positive RosalacZ (Gpa33Cre+/−:Pik3caH1047R/WT:RosalacZ/WT) mouse and from a control no Gpa33Cre (Gpa33Cre−/−:Pik3caH1047R/WT:RosalacZ/WT) mouse. Colon and small intestinal tissues showed positive staining in our test mouse (insert shows a higher magnification) and no staining in the control mouse. Liver was stained as a negative tissue control and showed no positive staining in either the RosalacZ reporter or the Gpa33Cre control mice. Scale bars, 50 μm. (C) Schematic of the exon-switch strategy used to engineer a conditional knockin of mutant Pik3caH1047R. LoxP sites were placed on either side of exon 20 of the Pik3ca gene and a duplicate copy of exon 20 containing the mutant sequence (red) placed downstream of the wild-type exon (lat-H1047R). Before Cre exposure the wild-type exon is expressed as normal and only after Cre expression was the floxed wild-type exon excised and replaced by the duplicate exon resulting in expression of the Pi3kcaH1047R mutation [9]. (D) The Apc conditional knockout mouse strategy used involves the flanking of exon 14 with loxP sites (purple), which is illustrated before (580S) and after (580D) Cre expression [10]. Ex, exon.

Figure 1
Confirmation of Cre excision using the Rosa26lacZ reporter mouse

(A) Schematic diagram of the genetic strategy used to assess lacZ expression using the lox-stop-lox Rosa26lacZ mouse strain crossed with the RU486-inducible Gpa33Cre mouse strain. Upon exposure to RU486, Cre excises the floxed stop codon only in intestinal cells which express Gpa33 resulting in lacZ expression (blue) in these cells which is detectable by β-galactosidase staining. (B) lacZ expression in tissues 90 days post-induction of Cre expression by RU486 from a reporter positive RosalacZ (Gpa33Cre+/−:Pik3caH1047R/WT:RosalacZ/WT) mouse and from a control no Gpa33Cre (Gpa33Cre−/−:Pik3caH1047R/WT:RosalacZ/WT) mouse. Colon and small intestinal tissues showed positive staining in our test mouse (insert shows a higher magnification) and no staining in the control mouse. Liver was stained as a negative tissue control and showed no positive staining in either the RosalacZ reporter or the Gpa33Cre control mice. Scale bars, 50 μm. (C) Schematic of the exon-switch strategy used to engineer a conditional knockin of mutant Pik3caH1047R. LoxP sites were placed on either side of exon 20 of the Pik3ca gene and a duplicate copy of exon 20 containing the mutant sequence (red) placed downstream of the wild-type exon (lat-H1047R). Before Cre exposure the wild-type exon is expressed as normal and only after Cre expression was the floxed wild-type exon excised and replaced by the duplicate exon resulting in expression of the Pi3kcaH1047R mutation [9]. (D) The Apc conditional knockout mouse strategy used involves the flanking of exon 14 with loxP sites (purple), which is illustrated before (580S) and after (580D) Cre expression [10]. Ex, exon.

Expression of the Pik3caH1047R mutation alone in the intestines does not initiate tumorigenesis

To explore the role of the Pik3caH1047R mutation in intestinal cancer in vivo, we crossed Pik3caLat−H1047R mice [9] (Figure 1C) with Gpa33CrePR2 mice [11]. The resulting Gpa33CrePR2/+:Pik3caLat−H1047R/+ mice expressed wild-type Pik3ca until fed on the progesterone agonist RU486 which induces Cre expression leading to the heterozygous expression of Pik3caH1047R in the gastrointestinal tract. Following induction, mice expressing heterozygous Pik3caH1047R were monitored for up to 2 years at which time they were killed and their gastrointestinal tracts analysed. No evidence of intestinal tumours or adenomas were detected (Figure 2A, blue line, and Table 1).

Mutant Pik3caH1047R and Apc loss leads to intestinal adenocarcinoma

Figure 2
Mutant Pik3caH1047R and Apc loss leads to intestinal adenocarcinoma

(A) Results are the percentage of mice which were intestinal adenocarcinoma free at the end point for each genotype. Kaplan–Meier analysis reveals a median intestinal tumour-free survival of 506 days in Pik3caH1047R:Apc580D/580D mice compared with Apc580D/580D, Pik3caH1047R and control mice, all of which remained tumour free (n is given in parentheses). (B) Macroscopic images of large tumours (arrows) observed in both the small (i) and large (ii) intestines of Pik3caH1047R:Apc580D/580D mice. Scale bar, 1 cm. (C) Graphical representation of the location proportions of adenocarcinomas observed across the small intestine and colon in Pik3caH1047R:Apc580D/580D mice. (D) H/E staining demonstrated that the tumours observed in Pik3caH1047R:Apc580D/580D mice were invasive adenocarcinomas. A higher magnification of the boxed area in panel i is shown in panel ii. The additional lesions observed included adenomas with early invasion. A higher magnification of the boxed area in panel iii is shown in panel iv. Apc580D/580D mice only presented with adenomas. A higher magnification of the boxed area in panel v is shown in panel vi. Scale bars, 100 μm. (E) RNA was extracted from the tumours, reverse-transcribed into cDNA and Sanger sequenced. Sequencing from exon 19 into exon 20 of the Pik3ca transcript confirmed the introduction of the mutant exon by the presence of all silent base changes engineered into the mutant exon 20 of the mouse [9] (indicated by orange boxes) as well as the base change responsible for the Pik3caH1047R mutation (indicated by the blue box). The portion of sequence spanning the mutation site (labelled H1047R) indicates the A>G base change resulting in the histidine (H) to arginine (R) amino acid change.

Figure 2
Mutant Pik3caH1047R and Apc loss leads to intestinal adenocarcinoma

(A) Results are the percentage of mice which were intestinal adenocarcinoma free at the end point for each genotype. Kaplan–Meier analysis reveals a median intestinal tumour-free survival of 506 days in Pik3caH1047R:Apc580D/580D mice compared with Apc580D/580D, Pik3caH1047R and control mice, all of which remained tumour free (n is given in parentheses). (B) Macroscopic images of large tumours (arrows) observed in both the small (i) and large (ii) intestines of Pik3caH1047R:Apc580D/580D mice. Scale bar, 1 cm. (C) Graphical representation of the location proportions of adenocarcinomas observed across the small intestine and colon in Pik3caH1047R:Apc580D/580D mice. (D) H/E staining demonstrated that the tumours observed in Pik3caH1047R:Apc580D/580D mice were invasive adenocarcinomas. A higher magnification of the boxed area in panel i is shown in panel ii. The additional lesions observed included adenomas with early invasion. A higher magnification of the boxed area in panel iii is shown in panel iv. Apc580D/580D mice only presented with adenomas. A higher magnification of the boxed area in panel v is shown in panel vi. Scale bars, 100 μm. (E) RNA was extracted from the tumours, reverse-transcribed into cDNA and Sanger sequenced. Sequencing from exon 19 into exon 20 of the Pik3ca transcript confirmed the introduction of the mutant exon by the presence of all silent base changes engineered into the mutant exon 20 of the mouse [9] (indicated by orange boxes) as well as the base change responsible for the Pik3caH1047R mutation (indicated by the blue box). The portion of sequence spanning the mutation site (labelled H1047R) indicates the A>G base change resulting in the histidine (H) to arginine (R) amino acid change.

Table 1
Incidence of intestinal tumours in control, Pik3caH1047R, Apc580D/580D and Pik3caH1047R:Apc580D/580D mice

*P<0.0001 compared with the wild-type and single mutant controls using Fisher's exact test.

Genotype Number of mice Number of mice with adenomas Number of mice with adenocarcinomas Total number of adenomas Total number of adenocarcinomas 
Control 19 
Pik3caH1047R 33 
Apc580D/580D 28 1 (4%) 
Pik3caH1047R:Apc580D/580D 46 3 (7%) 27* (59%) 28 
Genotype Number of mice Number of mice with adenomas Number of mice with adenocarcinomas Total number of adenomas Total number of adenocarcinomas 
Control 19 
Pik3caH1047R 33 
Apc580D/580D 28 1 (4%) 
Pik3caH1047R:Apc580D/580D 46 3 (7%) 27* (59%) 28 

Expression of the Pik3caH1047R mutation and Apc loss in the intestine synergize to cause tumorigenesis

To explore if the Pik3caH1047R mutation could promote intestinal tumorigenesis in the context of homozygous Apc loss, we crossed the Gpa33CrePR2/+:Pik3caLat−H1047R/WT mice with the conditional Apc580S mice in which exon 14 of the Apc gene is flanked by loxP sites [10]. RU486 exposure to both the control Gpa33CrePR2/+:Apc580S/580S mice and the double mutant Gpa33CrePR2/+:Pik3caLat−H1047R/WT:Apc580S/580S mice induces Cre resulting in deletion of exon 14 and a frameshift at codon 580 in both alleles of the Apc gene, leading to the generation of a truncated non-functional Apc protein (Figure 1D) either alone or alongside concurrent knockin of the Pik3caH1047R mutation. After Cre induction, floxed exons in Apc and Pik3ca were deleted and control and double mutant mice were named as Gpa33CrePR2/+:Apc580D/580D and Gpa33CrePR2/+:Pik3caH1047R/WT:Apc580D/580D mice respectively.

Upon killing, intestinal tumours were identified macroscopically in 59% of the Pik3caH1047R:Apc580D/580D (Gpa33CrePR2/+:Pik3caH1047R/WT:Apc580D/580D) mice which was dramatically increased compared with the control (Gpa33CrePR2/+), Apc580D/580D (Gpa33CrePR2/+:Apc580D/580D) and Pik3caH1047R (Gpa33CrePR2/+:Pik3caH1047R/WT) mice which did not present with any intestinal tumours (Figure 2A, P<0.0001). The double mutant Pik3caH1047R:Apc580D/580D mice presented with up to three tumour masses greater than 4 mm×4 mm in size in both the small intestine (Figure 2B, panel i) and colon (Figure 2B, panel ii, and Table 1).

Histological analysis identified tumours in the Pik3caH1047R:Apc580D/580D mice as aggressive adenocarcinomas with extensive invasion into the submucosa (Figure 2D). Areas of dysplasia were also observed adjacent to the adenocarcinomas. Adenocarcinomas were distributed throughout the intestine (nine in the duodenum, six in the jejunum, seven in the ileum and five in the distal colon; Figure 2C). Three Pik3caH1047R:Apc580D/580D mice (7%) presented with one to four small (<4 mm) polyps observed macroscopically in addition to adenocarcinoma. These were pathologically confirmed as adenomas, although some exhibited early signs of invasion (Figure 2D). No single mutant mice developed adenocarcinomas and only one Apc580D/580D mouse (4%) developed adenomas.

Sanger sequencing confirmed the presence of the Pik3caH1047R mutation and the silent base changes engineered into the mutant Pik3ca exon 20 in the adenocarcinomas (Figure 2E). IHC staining for phospho-RPS6, a downstream target of PI3K/Akt activity, demonstrated consistent staining across the tumours (Figure 3A, panel i) which was more intense than in the adjacent normal intestine (Figure 3A, panel ii) confirming increased PI3K pathway activity in the tumours.

Intestinal adenocarcinomas from Pik3caH1047R:Apc580D/580D mice show high PI3K and β-catenin signalling

Figure 3
Intestinal adenocarcinomas from Pik3caH1047R:Apc580D/580D mice show high PI3K and β-catenin signalling

Representative histological images and IHC staining of adenocarcinomas observed in the small intestine and colon of Pik3caH1047R:Apc580D/580D mice. (A) Staining for phospho-RPS6 in an invasive adenocarcinoma (i) and the adjacent normal small intestine (ii) (higher resolution images are shown in the inserts). Colon (B) and small intestine (C) adenocarcinomas were stained for β-catenin to evaluate the loss of Apc activity and Cdx2, an intestinal cell marker, was used to demonstrate the intestinal origin of the tumours. The boxed areas are shown at a higher resolution below each image. Areas which stained differentially for β-catenin and Cdx2 are separated by a broken line. Scale bars, 100 μm.

Figure 3
Intestinal adenocarcinomas from Pik3caH1047R:Apc580D/580D mice show high PI3K and β-catenin signalling

Representative histological images and IHC staining of adenocarcinomas observed in the small intestine and colon of Pik3caH1047R:Apc580D/580D mice. (A) Staining for phospho-RPS6 in an invasive adenocarcinoma (i) and the adjacent normal small intestine (ii) (higher resolution images are shown in the inserts). Colon (B) and small intestine (C) adenocarcinomas were stained for β-catenin to evaluate the loss of Apc activity and Cdx2, an intestinal cell marker, was used to demonstrate the intestinal origin of the tumours. The boxed areas are shown at a higher resolution below each image. Areas which stained differentially for β-catenin and Cdx2 are separated by a broken line. Scale bars, 100 μm.

The Pik3caH1047R mouse was also crossed with a second Apc mouse strain, the ApcMin mouse, which possessed a heterozygous germline Apc mutation predisposing it to the development of multiple adenomas [14]. As demonstrated previously, the latency for intestinal tumour-free survival in Pik3caH1047R:Apc580D/580D mice is ~200 days (Figure 2A). The ApcMin mouse strain had a reduced lifespan (<150 days) as a result of anaemia associated with polyp burden in the small intestine (Figure 4A). No difference in the proportion of adenomas observed across the small and large intestines was detected between ApcMin and Pik3caH1047R:ApcMin mice (Figure 4B). Yet despite this narrow window in which tumorigenesis can occur, one of the 12 (8%) Pik3caH1047R:ApcMin mice developed a large invasive adenocarcinoma with a confirmed presence of Pik3caH1047R mutation and Apc loss (Figures 4C–4E). The single Pik3caH1047R:ApcMin mouse which possessed an intestinal adenocarcinoma also developed a colonic adenoma that did not express the Pik3caH1047R mutation (Figure 4D), but expressed higher β-catenin levels characteristic of Apc loss of function (Figure 4E).

An intestinal adenocarcinoma was observed in a single Pik3caH1047R:ApcMin mouse

Figure 4
An intestinal adenocarcinoma was observed in a single Pik3caH1047R:ApcMin mouse

(A) Kaplan–Meier curve of survival in a cohort of control ApcMin alone (Gpa33Cre+/−:ApcMin/WT) mice compared with a cohort of mutant Pik3caH1047R:ApcMin (Gpa33Cre+/−:Pik3caH1047R/WT:ApcMin/WT) mice (the n is shown in parentheses). (B) Graphical representation of the proportion of adenomas observed across the small intestine and colon of Pik3caH1047R:ApcMin and ApcMin alone mice. (C) Upon dissection of the gastrointestinal tract, a large polyp was observed in the jejunum of the small intestine (arrow). A smaller polyp was observed in the colon (arrow head). Scale bar, 1 cm. (D) RNA was extracted from polyps as well as normal colon and small intestine (SI) tissue. The RNA was reverse-transcribed to cDNA and Sanger sequenced for the Pik3caH1047R mutation. The wild-type Pik3ca gene (WT Pik3ca) sequence is shown as well as the silent base changes engineered into the mutant Pik3caH1047R which are identified by the orange boxes. (E) Pathological analysis of H/E sections from these polyps revealed that the jejunum growth was an invasive adenocarcinoma, whereas the colon polyp was an adenoma. IHC staining for β-catenin of FFPE (formalin-fixed paraffin-embedded) sections of these tumours or adenomas was conducted to confirm the loss of Apc activity (a higher magnification image is shown in the insert). The broken line separates area of normal colon (Co) from the adenoma (Ad). Scale bars: bracketed, 25 μm and unbracketed, 100 μm.

Figure 4
An intestinal adenocarcinoma was observed in a single Pik3caH1047R:ApcMin mouse

(A) Kaplan–Meier curve of survival in a cohort of control ApcMin alone (Gpa33Cre+/−:ApcMin/WT) mice compared with a cohort of mutant Pik3caH1047R:ApcMin (Gpa33Cre+/−:Pik3caH1047R/WT:ApcMin/WT) mice (the n is shown in parentheses). (B) Graphical representation of the proportion of adenomas observed across the small intestine and colon of Pik3caH1047R:ApcMin and ApcMin alone mice. (C) Upon dissection of the gastrointestinal tract, a large polyp was observed in the jejunum of the small intestine (arrow). A smaller polyp was observed in the colon (arrow head). Scale bar, 1 cm. (D) RNA was extracted from polyps as well as normal colon and small intestine (SI) tissue. The RNA was reverse-transcribed to cDNA and Sanger sequenced for the Pik3caH1047R mutation. The wild-type Pik3ca gene (WT Pik3ca) sequence is shown as well as the silent base changes engineered into the mutant Pik3caH1047R which are identified by the orange boxes. (E) Pathological analysis of H/E sections from these polyps revealed that the jejunum growth was an invasive adenocarcinoma, whereas the colon polyp was an adenoma. IHC staining for β-catenin of FFPE (formalin-fixed paraffin-embedded) sections of these tumours or adenomas was conducted to confirm the loss of Apc activity (a higher magnification image is shown in the insert). The broken line separates area of normal colon (Co) from the adenoma (Ad). Scale bars: bracketed, 25 μm and unbracketed, 100 μm.

Loss of Cdx2 (caudal type homeobox 2) expression observed in adenocarcinomas in areas of enhanced nuclear β-catenin expression

β-Catenin localization was used to evaluate the loss of Apc function. High levels of nuclear β-catenin staining, corresponding to regions of Apc depletion, are clearly visible in tumours compared with the predominately cytoplasmic staining observed in the surrounding normal tissue (Figures 3B and 3C). Interestingly, in Pik3caH1047R:Apc580D/580D mice tumour tissue with high nuclear β-catenin staining we observed a complete loss of Cdx2 expression (Figures 3B and 3C).

Development of HCCs in all mice as a result of hypomorphic unrecombined Apc alleles

All Apc580D/580D mice, regardless of their Pik3ca status, developed HCCs (hepatocellular carcinomas) that required the mice to be killed (Supplementary Figures S1A–S1C at http://www.biochemj.org/bj/458/bj4580251add.htm) in spite of the fact that the RU486 induction of Gpa33CrePR2 does not express Cre in the liver (Figure 1B). The lack of expression of the Pik3caH1047R mutation in the HCCs and normal liver in these mice was confirmed by sequencing RNA extracted from these tissues (Supplementary Figure S1D). Previous research has demonstrated the presence of loxP cassettes in this strain to be hypomorphic and sufficient to reduce activity of the floxed Apc alleles prior to recombination (Apc580S/580S) [15]. The liver is particularly sensitive to this resulting in HCC development even with unrecombined alleles. In the present study, HCC growth resulted in the mice being killed before intestinal tumour development, accounting for the incomplete (59%) penetrance of tumour formation in the Pik3caH1047R:Apc580D/580D mice.

DISCUSSION

Even though mutations in the PIK3CA gene are commonly found in human CRC, the underlying molecular role they play in the mechanisms of intestinal carcinogenesis have not been fully elucidated. In the present study, we described a model in which the physiological expression of the Pik3caH1047R mutation synergizes with the targeted loss of Apc to induce intestinal tumorigenesis. We demonstrated that Pik3caH1047R expression alone is insufficient to initiate intestinal tumorigenesis, yet it is able to drive progression of adenomas, initiated by the inactivation of Apc, to aggressive and invasive intestinal adenocarcinomas. This result is consistent with our previous findings that Pik3caH1047R expression alone does not induce tumorigenesis in the ovary [9] or mammary gland [13].

Mutations in the APC gene are observed in over 80% of CRCs [5] and ~30% of sporadic CRCs possess PIK3CA mutations [7,8]. Analysis of the TCGA (The Cancer Genome Atlas) Network database of 212 CRCs found 14% harboured mutations in both PIK3CA and APC [16] (Supplementary Figure S2 at http://www.biochemj.org/bj/458/bj4580251add.htm). This co-operative interaction between the Wnt/β-catenin and PI3K pathways in cancer is supported by previous studies showing that the loss of Apc together with Pten (phosphatase and tensin homologue deleted on chromosome 10), a negative regulator of the PI3K enzyme, synergize in the progression of intestinal [17], ovarian [18] and bladder [19] tumorigenesis. Furthermore, PIK3CA mutations are commonly co-expressed with APC loss/β-catenin mutation in 14% of CRCs [16]. When extended to include other genes in the PI3K/Akt and Wnt/β-catenin pathways [PIK3R1 (PI3K regulatory subunit 1) and PTEN and CTNNB1 (catenin β1) respectively], this increases to ~22% of cases which possess a mutation or homozygous deletion in one or more genes in both of these pathways [16], highlighting further the importance of these two signalling cascades in CRC tumorigenesis.

Cdx2 is required for the regulation of intestinal epithelium homoeostasis, development and differentiation [20]. Loss of CDX2 mRNA and CDX2 protein expression has been associated previously with increased invasiveness in human CRC cell lines and poor prognosis in ~80% of poorly differentiated advance-stage adenocarcinoma samples from patients [20,21]. Thus loss of Cdx2 expression in Pik3caH1047R:Apc580D/580D tumours is consistent with them being advanced-stage aggressive carcinomas as observed in late-stage CRCs in humans.

Cdx2 is also considered to be a tumour suppressor in the adult colon. This is considered to be the result of down-regulation, not genetic alterations, due to the low number of CDX2 mutations observed in CRCs [2023]. In vitro studies demonstrated that PTEN stimulated the expression of Cdx2 in colon cancer cell lines and inhibition of PI3K by wortmannin also produced a similar effect, suggesting that Cdx2 is a target of the PTEN/PI3K signalling pathway [24]. As such, we postulate that constitutive activation of the PI3K pathway by the Pik3caH1047R mutation in our system may be down-regulating Cdx2 and its action as a tumour suppressor contributing further to the intestinal tumorigenesis observed.

The results of the present study contrast with a recent study where the transgenic expression of a constitutively active PI3K protein was sufficient to initiate aggressive tumour formation in the mouse colon within 60 days [25], which became slightly more aggressive upon spontaneous loss of Apc [26]. The tumorigenesis observed with PI3K activation alone in their model may be due to the different mutant protein used to dominantly activate the PI3K enzyme [a fusion of p110α and the iSH2 (inter-Src homology 2) domain of p85 compared with our single amino acid substitution]. In addition, they used a transgenic approach to express their mutant PI3K, which is likely to result in the overexpression of PI3K and, potentially, its expression in cells that may not normally express PI3K [25]. In our model, both the expression of the Pik3caH1047R mutation and loss of both Apc alleles are targeted to the same cell in the intestine and, importantly, the Pik3caH1047R mutation is knocked into the endogenous gene and thus expression is under the control of the endogenous promoter. As a result, the mutant protein is expressed at physiological levels and only in cells that normally express Pik3ca. Thus our model more accurately reflects the scenario of a heterozygous somatic mutation, as occurs in human CRC.

Evidence is growing that the Pik3caH1047R mutation is a weak oncogene and that multiple hits within the PI3K pathway itself, or in other oncogenic pathways, are necessary for it to contribute to tumorigenesis [9,13]. The results of the present study support further these findings, indicating that Pik3ca mutations alone, when expressed at physiological levels, are unable to initiate tumorigenesis in the intestine, but have a role in driving the progression of adenomas to invasive adenocarcinomas.

The ability of Pik3caH1047R to act in tandem with Apc loss to drive progression to adenocarcinoma implies that therapeutic targeting of the PI3K pathway could be beneficial in limiting disease progression in patients with CRC. Furthermore, this novel model of intestinal carcinoma will provide a valuable tool for further examining the role of the PI3K signalling cascade in CRC.

Abbreviations

     
  • APC

    adenomatous polyposis coli

  •  
  • Cdx2

    caudal type homeobox 2

  •  
  • CRC

    colorectal cancer

  •  
  • Gpa33

    glycoprotein A33

  •  
  • HCC

    hepatocellular carcinoma

  •  
  • H/E

    haematoxylin and eosin

  •  
  • IHC

    immunohistochemistry

  •  
  • PI3K

    phosphoinositide 3-kinase

  •  
  • PIK3CA

    PI3K catalytic, α polypeptide

  •  
  • PTEN

    phosphatase and tensin homologue deleted on chromosome 10

  •  
  • RPS6

    ribosomal protein S6

AUTHOR CONTRIBUTION

Wayne Phillips conceived the idea for the present study. Lauren Hare, Toby Phesse, Karen Montgomery and Kevin Mills performed the experiments. Robert Ramsay, Matthias Ernst and Joan Heath made contributions to the conception and design of in vivo experiments and provided reagents. Paul Waring provided pathological analyses of tissues. All authors participated in analysis, discussion and interpretation of results. Lauren Hare wrote the paper. All authors read, commented on and approved the paper.

We thank the Peter MacCallum Cancer Centre core histology and animal house facilities and technicians.

FUNDING

This work was supported by the National Health and Medical Research Council of Australia [grant numbers 628621 (to W.A.P.) 1007523 (to M.E.), 487922 (to M.E. and J.K.H.) and 603127 (to T.J.P.)] and L.M.H was the recipient of an Australian Postgraduate Award.

References

References
1
Ferlay
J.
Shin
H.
Bray
F
Forman
D.
Mathers
C.
Parkin
D.
Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 10
2010
Lyon
World Health Organization, International Agency of Research on Cancer
2
Fearon
E. R.
Vogelstein
B.
A genetic model for colorectal tumorigenesis
Cell
1990
, vol. 
61
 (pg. 
759
-
767
)
3
Fearon
E. R.
Molecular genetics of colorectal cancer
Annu. Rev. Pathol.
2011
, vol. 
6
 (pg. 
479
-
507
)
4
Kinzler
K. W.
Nilbert
M. C.
Su
L. K.
Vogelstein
B.
Bryan
T. M.
Levy
D. B.
Smith
K. J.
Preisinger
A. C.
Hedge
P.
McKechnie
D.
, et al. 
Identification of FAP locus genes from chromosome 5q21
Science
1991
, vol. 
253
 (pg. 
661
-
665
)
5
Rowan
A. J.
Lamlum
H.
Ilyas
M.
Wheeler
J.
Straub
J.
Papadopoulou
A.
Bicknell
D.
Bodmer
W. F.
Tomlinson
I. P.
APC mutations in sporadic colorectal tumors: a mutational “hotspot” and interdependence of the “two hits”
Proc. Natl. Acad. Sci. U.S.A.
2000
, vol. 
97
 (pg. 
3352
-
3357
)
6
Vanhaesebroeck
B.
Stephens
L.
Hawkins
P.
PI3K signalling: the path to discovery and understanding
Nat. Rev. Mol. Cell Biol.
2012
, vol. 
13
 (pg. 
195
-
203
)
7
Campbell
I. G.
Russell
S. E.
Choong
D. Y. H.
Montgomery
K. G.
Ciavarella
M. L.
Hooi
C. S. F.
Cristiano
B. E.
Pearson
R. B.
Phillips
W. A.
Mutation of the PIK3CA gene in ovarian and breast cancer
Cancer Res.
2004
, vol. 
64
 (pg. 
7678
-
7681
)
8
Samuels
Y.
Wang
Z.
Bardelli
A.
Silliman
N.
Ptak
J.
Szabo
S.
Yan
H.
Gazdar
A.
Powell
S. M.
Riggins
G. J.
, et al. 
High frequency of mutations of the PIK3CA gene in human cancers
Science
2004
, vol. 
304
 pg. 
554
 
9
Kinross
K. M.
Montgomery
K. G.
Kleinschmidt
M.
Waring
P.
Ivetac
I.
Tikoo
A.
Saad
M.
Hare
L.
Roh
V.
Mantamadiotis
T.
, et al. 
An activating Pik3ca mutation coupled with Pten loss is sufficient to initiate ovarian tumorigenesis in mice
J. Clin. Invest.
2012
, vol. 
122
 (pg. 
553
-
557
)
10
Shibata
H.
Toyama
K.
Shioya
H.
Ito
M.
Hirota
M.
Hasegawa
S.
Matsumoto
H.
Takano
H.
Akiyama
T.
Toyoshima
K.
, et al. 
Rapid colorectal adenoma formation initiated by conditional targeting of the Apc gene
Science
1997
, vol. 
278
 (pg. 
120
-
123
)
11
Malaterre
J.
Carpinelli
M.
Ernst
M.
Alexander
W.
Cooke
M.
Sutton
S.
Dworkin
S.
Heath
J. K.
Frampton
J.
McArthur
G.
, et al. 
c-Myb is required for progenitor cell homeostasis in colonic crypts
Proc. Natl. Acad. Sci. U.S.A.
2007
, vol. 
104
 (pg. 
3829
-
3834
)
12
Soriano
P.
Generalized lacZ expression with the ROSA26 Cre reporter strain
Nat. Genet.
1999
, vol. 
21
 (pg. 
70
-
71
)
13
Tikoo
A.
Roh
V.
Montgomery
K. G.
Ivetac
I.
Waring
P.
Pelzer
R.
Hare
L.
Shackleton
M.
Humbert
P.
Phillips
W. A.
Physiological levels of Pik3caH1047R mutation in the mouse mammary gland results in ductal hyperplasia and formation of ERα-positive tumors
PLoS ONE
2012
, vol. 
7
 pg. 
e36924
 
14
Moser
A. R.
Pitot
H. C.
Dove
W. F.
A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse
Science
1990
, vol. 
247
 (pg. 
322
-
324
)
15
Buchert
M.
Athineos
D.
Abud
H. E.
Burke
Z. D.
Faux
M. C.
Samuel
M. S.
Jarnicki
A. G.
Winbanks
C. E.
Newton
I. P.
Meniel
V. S.
, et al. 
Genetic dissection of differential signaling threshold requirements for the Wnt/β-catenin pathway in vivo
PLoS Genet.
2010
, vol. 
6
 pg. 
e1000816
 
16
Cancer Genome Atlas Network
Comprehensive molecular characterization of human colon and rectal cancer
Nature
2012
, vol. 
487
 (pg. 
330
-
337
)
17
Marsh
V.
Winton
D. J.
Williams
G. T.
Dubois
N.
Trumpp
A.
Sansom
O. J.
Clarke
A. R.
Epithelial Pten is dispensable for intestinal homeostasis but suppresses adenoma development and progression after Apc mutation
Nat. Genet.
2008
, vol. 
40
 (pg. 
1436
-
1444
)
18
Wu
R.
Hendrix-Lucas
N.
Kuick
R.
Zhai
Y.
Schwartz
D. R.
Akyol
A.
Hanash
S.
Misek
D. E.
Katabuchi
H.
Williams
B. O.
, et al. 
Mouse model of human ovarian endometrioid adenocarcinoma based on somatic defects in the Wnt/β-catenin and PI3K/Pten signaling pathways
Cancer Cell
2007
, vol. 
11
 (pg. 
321
-
333
)
19
Ahmad
I.
Morton
J. P.
Singh
L. B.
Radulescu
S. M.
Ridgway
R. A.
Patel
S.
Woodgett
J.
Winton
D. J.
Taketo
M. M.
Wu
X. R.
, et al. 
β-Catenin activation synergizes with PTEN loss to cause bladder cancer formation
Oncogene
2011
, vol. 
30
 (pg. 
178
-
189
)
20
Baba
Y.
Nosho
K.
Shima
K.
Freed
E.
Irahara
N.
Philips
J.
Meyerhardt
J. A.
Hornick
J. L.
Shivdasani
R. A.
Fuchs
C. S.
Ogino
S.
Relationship of CDX2 loss with molecular features and prognosis in colorectal cancer
Clin. Cancer Res.
2009
, vol. 
15
 (pg. 
4665
-
4673
)
21
Bonhomme
C.
Duluc
I.
Martin
E.
Chawengsaksophak
K.
Chenard
M. P.
Kedinger
M.
Beck
F.
Freund
J. N.
Domon-Dell
C.
The Cdx2 homeobox gene has a tumour suppressor function in the distal colon in addition to a homeotic role during gut development
Gut
2003
, vol. 
52
 (pg. 
1465
-
1471
)
22
Sivagnanasundaram
S.
Islam
I.
Talbot
I.
Drummond
F.
Walters
J. R.
Edwards
Y. H.
The homeobox gene CDX2 in colorectal carcinoma: a genetic analysis
Br. J. Cancer
2001
, vol. 
84
 (pg. 
218
-
225
)
23
Choi
B. J.
Kim
C. J.
Cho
Y. G.
Song
J. H.
Kim
S. Y.
Nam
S. W.
Lee
S. H.
Yoo
N. J.
Lee
J. Y.
Park
W. S.
Altered expression of CDX2 in colorectal cancers
APMIS
2006
, vol. 
114
 (pg. 
50
-
54
)
24
Kim
S.
Domon-Dell
C.
Wang
Q.
Chung
D. H.
Di Cristofano
A.
Pandolfi
P. P.
Freund
J.-N.
Evers
B. M.
PTEN and TNF-α regulation of the intestinal-specific Cdx-2 homeobox gene through a PI3K, PKB/Akt, and NF-κB-dependent pathway
Gastroenterology
2002
, vol. 
123
 (pg. 
1163
-
1178
)
25
Leystra
A. A.
Deming
D. A.
Zahm
C. D.
Farhoud
M.
Olson
T. J.
Hadac
J. N.
Nettekoven
L. A.
Albrecht
D. M.
Clipson
L.
Sullivan
R.
, et al. 
Mice expressing activated PI3K rapidly develop advanced colon cancer
Cancer Res.
2012
, vol. 
72
 (pg. 
2931
-
2936
)
26
Deming
D. A.
Leystra
A. A.
Nettekoven
L.
Sievers
C.
Miller
D.
Middlebrooks
M.
Clipson
L.
Albrecht
D.
Bacher
J.
Washington
M. K.
, et al. 
PIK3CA and APC mutations are synergistic in the development of intestinal cancers
Oncogene
2013
 
doi: 10.1038/onc.2013.167

Supplementary data