Protein kinase A antagonist inhibits β-catenin nuclear translocation, c-Myc and COX-2 expression and tumor promotion in ApcMin/+ mice

Background The adenomatous polyposis coli (APC) protein is part of the destruction complex controlling proteosomal degradation of β-catenin and limiting its nuclear translocation, which is thought to play a gate-keeping role in colorectal cancer. The destruction complex is inhibited by Wnt-Frz and prostaglandin E2 (PGE2) - PI-3 kinase pathways. Recent reports show that PGE2-induced phosphorylation of β-catenin by protein kinase A (PKA) increases nuclear translocation indicating two mechanisms of action of PGE2 on β-catenin homeostasis. Findings Treatment of ApcMin/+ mice that spontaneously develop intestinal adenomas with a PKA antagonist (Rp-8-Br-cAMPS) selectively targeting only the latter pathway reduced tumor load, but not the number of adenomas. Immunohistochemical characterization of intestines from treated and control animals revealed that expression of β-catenin, β-catenin nuclear translocation and expression of the β-catenin target genes c-Myc and COX-2 were significantly down-regulated upon Rp-8-Br-cAMPS treatment. Parallel experiments in a human colon cancer cell line (HCT116) revealed that Rp-8-Br-cAMPS blocked PGE2-induced β-catenin phosphorylation and c-Myc upregulation. Conclusion Based on our findings we suggest that PGE2 act through PKA to promote β-catenin nuclear translocation and tumor development in ApcMin/+ mice in vivo, indicating that the direct regulatory effect of PKA on β-catenin nuclear translocation is operative in intestinal cancer.


Findings
The adenomatous polyposis coli (APC) gene is thought to play a gate-keeping role in the tumor formation and progression and is the most commonly mutated gene in all colorectal cancers. In humans, APC mutations can be acquired (spontaneous CRC) or inherited as in the autosomal, familiar adenomatous polyposis (FAP), characterized by the formation of multiple colonic adenomatous polyps [1]. Inactivation of both APC alleles (APC -/-) is considered necessary for tumor formation. The APC protein forms a destruction complex with Axin, glycogen synthase kinase 3β (GSK3β) and casein kinase 1 (CK1) which phosphorylates β-catenin at multiple sites [2], and targets β-catenin for ubiquitination and to degradation by the proteasome system [3]. A defective APC protein leads to cytoplasmic accumulation and translocation of β-catenin to the nucleus [4]. β-catenin, originally discovered as a cadherin-binding protein, has been shown to interact with and function as a coactivator of T-cell factor/lymphocyte enhancer factor (TCF/LEF) transcription factors. Human transcription factor 4 (hTCF-4), a TCF family member that is expressed in human colonic epithelium and colon carcinoma cells, transactivates transcription only when associated with β-catenin [5]. The result is expression and production of mitogenic and survival genes including c-Myc [6], cyclin D1 [7] and cyclooxygenase-2 (COX-2) [8].
COX-2 levels are elevated in as many as 85% of human CRCs and approximately 50% of colorectal adenomas [8]. Studies have shown that COX inhibition by non-steroidal anti-inflammatory drugs (NSAIDS) or aspirin reduces the risk of CRC and may be beneficial in large population groups at risk [9]. Selective COX-2 inhibitors are also associated with a decline in the incidence of CRC and reduced mortality rate, although COX-2 inhibitors have been associated with serious cardiovascular events in this context [10]. Prostaglandin E 2 (PGE 2 ) has been shown to be an important mediator of COX-2 associated effects, and PGE 2 levels are elevated in CRC biopsies compared with normal mucosa and even in patient blood samples [11]. Beside an anti-angiogenic effect [12], COX inhibition promotes apoptosis and alters tumor growth [13]. PGE 2 and COX-2 over-expression also correlates with CRC risk and metastasis of CRC [14], making this pathway relevant also in follow-up after treatment of the primary cancer. Furthermore, our observations show that the PGE 2 produced also inhibits anti-tumor immunity through the EP2 prostanoid receptor -cAMP -protein kinase A (PKA) -Csk pathway in effector T cells that inhibit T cell activation [11].
Both the Wnt-Frz and the PGE 2 -EP3 pathway acting through phosphoinositide 3-kinase (PI3K) and protein kinase B (PKB) negatively regulates the APC destruction complex that controls β-catenin proteosomal degradation. COX inhibitors are thought to reverse the inhibitory effect of PGE 2 -EP3 receptor signaling on the APC destruction complex promoting β-catenin degradation and reversing the mitogenic effects. However, homozygous deletion of the gene for the PGE 2 receptor EP2 also reduced the number and size of colorectal polyps in a polyposis mouse model [15]. Furthermore, recent reports have shown that PKA can phosphorylate β-catenin at Ser552 [16] and Ser675 [16,17] and that the effect of βcatenin phosphorylation at the latter site is mediated by non-canonical mechanism(s) that does not involve regulation of the formation of the destruction complex. While Taurin et al. show that Ser675 phosphorylation promotes β-catenin interaction with the transcriptional coactivator CREB-binding protein in the nucleus and does not affect β-catenin stability and intracellular location [16], Hino et al. report that PKA phosphorylation of the same site stabilizes β-catenin and affects its intracellular localization [17]. These differences highlight the complexity of regulation of Wnt-β-catenin signaling and may relate to the experimental conditions and system examined. Finally, PGE 2 has been shown to control βcatenin homeostasis in zebrafish stem cells by signaling through both the EP3 receptor to the destruction complex and through the EP2 and EP4 receptors via cAMP to PKA affecting β-catenin stability [18]. Given the importance of β-catenin as a trans-activator in CRC and the interest in COX chemoprevention, the question of whether the PGE 2 -EP2/4-cAMP-PKA pathway is also active in controlling β-catenin levels in CRC is highly relevant [19].
The Apc Min/+ mouse is a well-established model of FAP with a germline mutation in one APC allele, thus increasing the probability of a double allele mutation and tumor formation. Apc Min/+ mice develop multiple adenomas in the intestinal tract, mainly in the small intestine, at an early age which can be blocked effectively by COX inhibition through NSAIDS. Here, we asked whether perturbation of the EP2/4 but not the EP3 pathway by inhibition at the level of PKA, could affect β-catenin levels and tumor formation. We show that treatment of Apc Min/+ mice with a PKA antagonist, Rp-8-Br-cAMPS, reduces tumor load, β-catenin levels and nuclear translocation as well as expression of β-catenin target genes in Apc Min/+ mice in vivo.

Differential effects of COX and PKA inhibition on tumor formation in Apc Min/+ mice
To more closely delineate the effect of PKA in the COX-2 -PGE 2 pathway active in colorectal cancer, we treated Apc Min/+ mice with the PKA antagonist Rp-8-Br-cAMPS for 6 weeks using earlier established doses (see Additional file 1, Supplementary information) and compared the result with that of treatment with the COX inhibitor indomethacin, previously shown to inhibit tumor development in the Apc Min/+ model [20]. Phosphate buffered saline (PBS) was used as vehicle control for the Rp-8-Br-cAMPS. Examination revealed that indomethacin reduced the number and area of tumors in the small intestine of the Apc Min/+ mice compared to PBS (from 47 to 3 tumors per mouse and from 0.44 mm 2 to 0.10 mm 2 tumor area; P < 0.001; Figure 1A, B). The PKA antagonist Rp-8-Br-cAMPS did not significantly reduce the number of adenomas (47 versus 43 tumors; P = 0.368, Figure 1A, B), but reduced the tumor area by 36% (from 0.44 mm 2 to 0.28 mm 2 ; P < 0.001; Figure 1). Specifically, tumor load was reduced in the distal part of the small intestine ( Figure 1C). The differential effect of COX inhibitor and PKA antagonist on tumor numbers and tumor size indicated to us that the mechanisms of action could be distinct and were examined in more detail in the following.

Inhibition of PKA does not affect lymphocytic tumor infiltration or HIF-1a expression in Apc Min/+ mice tumors
Lymphocytic tumor infiltration affects the course of human CRC where type, density and location of immune cells are shown to have higher prognostic power than the classical UICC-TNM staging [21]. Furthermore, the hypothesis of adaptive regulatory T cells (Treg) inhibiting anti-tumor immune responses has been subject to considerable interest [22]. Previously, we found that upon activation, Tregs express COX-2 and suppress effector T cells by PGE 2 -cAMP dependent mechanisms that may be of clinical relevance in patients with CRC [11]. However, immunohistochemical characterization of small intestinal tumors from PKA antagonist Rp-8-Br-cAMPS treated animals did not reveal any significant changes in the number of CD3 + T cells, CD8 + cytotoxic T cells, Foxp3 + Tregs or CD56 + natural killer (NK) cells (Figure 2A, B and Additional file 2, Figure S1). In contrast, levels of granzyme B (GZMB), a protein expressed in the cytotoxic T lymphocytes (CD8 + T cells) and NK cells, were reduced in Rp-8-Br-cAMPS treated animals which may indicate more degranulated cytotoxic cells post activation ( Figure 2B). Our observations indicate that intestinal immune responses play a minor role in the development of the Apc Min/+ mice tumor load, consistent with other observations [23]. PGE 2 also affects angiogenesis and up-regulates vascular endothelial growth factor receptor-1 (VEGFR-1) in a human colon cancer cell line [12] whereas indomethacin inhibits the expression of VEGF and thereby angiogenesis [24]. To assess treatments effects on angiogenesis, we examined levels of the hypoxia-inducible transcription factor (HIF)-1α which regulates the expression of target genes important in angiogenesis by accumulation and translocation to the nucleus under hypoxic conditions. While apical regions of all tumors showed higher cytoplasmic intensity and nuclear staining of HIF-1α, no differences between treatment groups were observed ( Figure 2C and Additional file 2, Figure S1).
PKA antagonist treatment of Apc Min/+ mice decreases the levels b-catenin signaling to the nucleus and of COX-2 and c-Myc expression in Apc Min/+ mice tumors We next examined the effect of treatment on the activity of the PGE 2 -β-catenin pathway in tumor cells. As evident from image analysis of immunohistochemically stained sections, levels of β-catenin were significantly decreased in tumors from the animals treated with the PKA antagonist Rp-8-Br-cAMPS compared to tumors from the controltreated group (138 versus 117 median inverse grayscale intensity units; P < 0.001, Figure 3A, B) Furthermore, the median number of β-catenin positive nuclei were reduced from 20% in tumors in the control group to 10% in tumors in the Rp-8-Br-cAMPS treated group (P = 0.024, Figure 3A, B and Additional file 3, Figure S2). In addition, expression of the β-catenin/TCF/LEF transcription complex-regulated genes c-Myc and COX-2 were reduced in tumor cells upon treatment with the PKA antagonist Rp-8-Br-cAMPS as evident from the median number of nuclei positive for c-Myc (reduced from 60% in control to 20%; P < 0.001, Figure 3A, B and Additional file 3, Figure S2) and from the cytoplasmic expression levels of the COX-2 enzyme (reduced from 136 in control to 106 median inverse grayscale intensity units in the treated group; P < 0.001, Figure 3A, B and Additional file 3, Figure S2).  Cytoplasmic β-catenin may be targeted to proteosomal degradation through the destruction complex consisting of GSK3β, Axin, CK1 and APC ( Figure 4C). However, in the presence of active Wnt signaling, β-catenin accumulates in the cytosol and translocates to the nucleus to act in a mitogenic fashion by transactivation of TCF/ LEF leading to expression of target genes in a cell proliferation and survival program [5]. As is well established, the Wnt-Frz pathway inhibits the destruction complex at the level of GSK3β, leading to less proteosomal degradation and more nuclear translocation and activation of β-catenin [2]. Similarly, the up-regulation of COX-2 in colorectal cancer leads to production of PGE 2 which binds to the EP3 receptor leading to PI3K and PKB activation, phosphorylation and dissociation of GSK3β and thereby inhibition of the destruction complex [26] (Figure 4C). In zebrafish stem cells, PGE 2 acting through an EP2/4-cAMP-PKA pathway was recently shown to induce direct phosphorylation of β-catenin, thereby stimulating its translocation to the nucleus and mitogenic effect [18]. Here, we tested whether this second pathway was providing a mitogenic drive in intestinal cancer. Using Apc Min/+ mice with a disturbed β-catenin degradation, we specifically inhibited the PGE 2 -cAMP pathway at the level of PKA by treating mice with Rp-8-Br-cAMPS for 6 weeks (see, Figure 4C for point of action). We show that this not only reduces tumor load but also specifically inhibits β-catenin nuclear translocation and the activation of β-catenin target genes such as c-Myc and COX-2 which may indicate that the direct regulatory effect of PKA on β-catenin nuclear translocation is also operative in intestinal cancer cells. Furthermore, the fact that COX inhibitors may block the effect of PGE 2 both in the β-catenin degradation and β-catenin nuclear translocation pathways while Rp-8-Br-cAMPS only affects the latter may explain why inhibitory effect of the PKA antagonist on tumor promotion is comparably weaker than that of indomethacin. Finally, our observation that COX inhibitor abolishes tumor numbers whereas PKA antagonist reduces tumor load but not tumor numbers may indicate that the antitumorigenic and anti-proliferative effects are distinct and relate to different points of action in PGE 2 signal pathways. It is interesting to speculate that stem cells in crypt foci that give origin to adenomas may be more sensitive to regulation via the Wnt-Frz and PGE 2 -EP3 pathways than via the EP2/4-cAMP-PGE 2 pathway whereas this balance may shift during tumor development.
Additional file 2: Figure S1. Immunohistochemical staining with indicated antibodies of tumor and normal mucosa from small intestines from Apc Min/+ mice treated with PBS or Rp-8-Br-cAMPS.