Axin, which was initially identified from analysis of the mouse Fused locus, plays a critical role in controlling axis formation during embryonic development in that it is a negative regulator of Wnt signaling pathway.1 In the absence of Wnt signaling pathway, Axin serves as a scaffold protein that coordinates the action of APC, casein kinase, GSK-3β and PP2A to down-regulate the function of oncogenic β-catenin. Through the formation of multiprotein complex, Axin links these components together and facilitates the phosphorylation of β-catenin by GSK-3β, leading to the ubiquitin-mediated degradation by the proteasome system. Stimulation of the Wnt signaling pathway activates Dishevelled, resulting in activation of GSK-3β binding protein (GBP). As a result, GSK-3β is recognized and inhibited by GBP, attenuating phosphorylation of β-catenin. Unphosphorylated β-catenin is stable and translocated into the nucleus where it associates with the lymphoid enhancer factors to activate target genes encoding proliferative factors such as c-MYC and cyclin D, resulting in oncogenesis.2, 3, 4, 5
In addition, Axin also exerts other functions including activation of the JNK mitogen-activated protein kinase and apoptosis.6 In the JNK signaling pathway, Axin interacts with MEKK1 through the MEKK1 interacting domain (MID) for activation of JNK via MKK4/7. It is intriguing that in addition to the MEKK1 binding, homodimerization via the DIX domain, sumoylation at the C-terminal are also critical for the activation of JNK by Axin.7 More perplexing yet is that the protein sequence in the PP2A binding region of Axin is also required for JNK activation (Jin and Lin, unpublished observations). These data all indicate that Axin requires many functional domains for different biological functions, and that any alteration in the Axin sequence could lead to functional abnormality.
Alterations of the Wnt signaling pathway appear to contribute to pathogenesis of several human cancers including cerebellar medulloblastomas,8, 9 endometrioid adenocarcinomas,10 hepatocellular carcinomas,11, 12 hepatoblastomas12 and primitive neuroectodermal tumors.13 Mutations of AXIN1 have been identified previously in these tumors. Furthermore, there have been identified 2 sequence variants that cause amino acid substitutions in 4 colon cancer cell lines.14 Colorectal cancer is one of the most frequent cancers in the world; it has been shown that more than 80% of the colorectal cancers have mutations of the APC gene.15 To find out if Axin is also involved in the pathogenesis of colorectal cancers, we analyzed 54 colorectal cancer tissues for point mutations in the AXIN1 gene.
We obtained 54 surgically removed colorectal cancer tissues from the Xijing Hospital, the affiliated hospital of the Fourth Military Medical University, China. All of the 54 samples were pathologically confirmed adenocarcinoma tissues located either in recta or colon. Clinical information of the patients with Axin sequence alteration is shown in Table I. Tissue specimens had previously been fixed in 10% formalin and embedded in paraffin for routine histological examination. Five healthy blood samples were analyzed as normal control.
Table I. Clinical Profile of the Cases With Missense Alteration
WHO histological classification
Poorly differentiated adenocarcinoma
Preparation of genomic DNA
Individual tissue samples were examined microscopically on H&E stained sections to exclude contaminating necrotic or normal colon tissue. Twenty sections of 5 mm thick paraffin-embedded colorectal cancer sections were used for genomic DNA extraction. The method for extraction was a modification of a method previously described.16 Briefly, the sections were transferred to 1.5 ml Eppendorf tubes, deparaffinated in 3 changes of 1.0 ml xylene and then washed 3 times in 1.0 ml absolute ethanol. DNA was extracted by standard proteinase K digestion and phenol/chloroform extraction after tissues were briefly dried.17 Genomic DNA was extracted from the 5 healthy blood samples as described as follows. Fresh blood samples (0.5 ml each) were centrifuged in 1.5 ml Eppendorf tubes; leukocytes were collected and subjected to vortex in 5 mM potassium iodine and then centrifuged at the maximal speed in a microfuge. The DNA in the supernatant was extracted with phenol/chloroform and precipitated with isopropanol and then was dissolved in TE for use.
From each extracted DNA, approximately 50 ng was used as template for PCR followed by nonradioactive SSCP analysis. The complete coding region of the AXIN1 gene was amplified using 23 sets of primers, which were prepared according to primer sequences reported by Satoh et al.11 PCR reactions were performed in a T3 Thermocycler (Biometra, Göttingen, Germany) and the amplifications were carried out in a volume of 25 ml containing 50 ng of DNA, 50 mM KCl, 10 mM Tris-HCl, 2 mM of each dNTP, 20 pmol of each primer, 1.5 mM MgCl2 and 1 U Taq polymerase, for 5 min at 94°C for initial denaturing, followed by 30 cycles of 94°C for 30 sec, 48–61°C (depending on the primers melting temperatures) for 30 sec and 72°C for 1 min and a final incubation at 72°C for 5 min. The PCR products were diluted with formamide dye solution (95% formamide, 1 mM EDTA (pH 8.0), 0.05% bromide blue and 0.05% xylene cyanol) and electrophoresed on 10% and 14% polyacrylamide (29:1, 39:1) gels that had been pre-run at 80 V for 1 hr. The electrophoresis was run at 4°C, 120 V for 5 hr. The gels were then visualized by silver staining.18 Aberrant bands were excised and eluted in elution buffer (0.5 mM acetic ammonium, 10 mM acetic magnesium, 1 mM EDTA and 0.1% SDS) and the DNA was reamplified. The resulting PCR products were subcloned into pBluescript and were sequenced.
Genomic DNA samples were extracted from 54 colorectal cancer tissues and were analyzed for mutations in the AXIN1 gene by PCR-SSCP method. Five healthy blood samples were included as normal control. The target sequence covered the whole coding region and their flanking portions of the intron sequences of the AXIN1 gene. Aberrantly migrating bands appeared in 27 different colorectal cancer cases and 1 blood sample (Fig. 1). After subcloning and sequencing analysis of PCR products of single strand DNA eluted from these bands, 18 out of the 27 aberrant bands contained sequence alterations in the intron 4, all of which were at the 17th nucleotide (Table II). The remaining 9 DNA samples contained nucleotide changes distributed among exons 1, 2, 4, 6 and 10, with 3 having silent mutations and 6 missense mutations (Table II). The 6 tissue samples with missense mutations were all well-differentiated adenocarcinoma located in recta except one (case 26) that is poorly differentiated and located in colon. Among the missense mutations, K203M (Genbank Accession Number AE006463-4) was in the RGS domain of exon 1, the binding domain of APC; Y305-stop and N307K were in the MID domain of exon 2, the binding domain of MEKK1; H394N was in exon 4, the binding domain of GSK-3β and 2 mutations, P848L and E852G, were in the DIX domain of exon 10, the C-terminal region of Axin for its oligomerization (Fig. 2).
Table II. Sequence Alterations of the Axin1 Gene in Colorectal Cancers
In addition to the missense mutations, we confirmed one polymorphism, g36883a (GenBank Accession Number AE006463), the nucleotide 17 in intron 4, which was reported previously in 18 cases out of 50 healthy Japanese individuals;19 Three silent alterations were newly detected at positions c1132t (Genbank Accession Number XM-027520) for Asp 256, t1192c for Ala 276 and t2190c for Ala 609, respectively. All the mutations were confirmed in their respective tumor tissues by another round of PCR-SSCP analysis and sequencing analysis.
Axin is a multidomain protein on which many important cellular factors bind, indicating that its sequence alterations would potentially lead to functional abnormalities. Its DIX domain is highly conserved from Xenopus to mammalian. It is very important in Axin's multifunction. Injection of wild-type rAxin can inhibit axis formation in early vertebrate development in Xenopus embryos; histological examination and gene expression of ΔDIX mutant of rAxin revealed weaker ventralizing activity than wild-type rAxin.20 The DIX domain of Axin was found to be important for association with Dishevelled through mutational and binding analysis of murine Axin and Disheveled proteins. It was reported that the DIX domain of Disheveled also controlled the ability of Disheveled to induce the accumulation of cytosolic β-catenin.21 The C-terminal region was required for binding to Axin itself. Substitution of the C-terminal with an unrelated dimerizing molecule, the retinoid X receptor, restored its inhibitory effect on Lef-1-dependent transcription; this also suggested that oligomerization of Axin through its C-terminus was important for its function in regulation of β-catenin-mediated response.22 Previously, we also found that Axin needs strict homodimerization through its DIX domain to activate JNK.23, 24
In our study, we identified 2 genetic alterations in the DIX domain of AXIN1 in colorectal cancers: P848L and E852G. Considering the importance of secondary structure of the DIX domain for Axin oligomerization, these mutations would perhaps affect heterodimerization between Disheveled and Axin or homodimerization of Axin itself through secondary structure change of Axin. Therefore, they would interfere with functions of Axin in both the Wnt signaling pathway and the JNK signaling pathway.
APC is the suppressor of colorectal cancers; mutations of the APC gene play a major role in the early development of colorectal neoplasms.25 In fact, approximate 80% of colorectal cancers were due to the mutations of APC gene. Axin can regulate the function of APC during its transcriptional regulation process through binding with it.6 It has been shown that overexpression of human Axin strongly promoted the downregulation of wild-type β-catenin in colon cancer cells.26 In contrast, Axin that lacked the APC-binding domain cannot accelerate the degradation of β-catenin, suggesting that APC is critical for Axin downregulation of β-catenin.26 In fact, APC can facilitate the phosphorylation of β-catenin by GSK-3β, which is a prerequisite for β-catenin ubiquitination and subsequent degradation.27 The domain responsible for APC binding lies in the conserved RGS domain.6 The crystal structures of the Axin/APC complex have revealed that APC/Axin interaction occurs at a conserved groove of the RGS domain. Lys203 that was found to be mutated to Met in case 26 lies in the eighth α-helix of this structure, which is in the exterior of the groove corner.28 It is possible that this mutation may alter the secondary structure of Axin and accordingly change the binding affinity for APC, affecting the normal function of the APC/Axin complex, which would lead to the accumulation of β-catenin.
The change from Tyr305 into stop codon in exon 2 would lead to truncation of the Axin product that lacks binding domains for MEKK1, GSK-3β, β-catenin and the DIX domain for heterodimerization with Dishevelled and Axin homodimerization. It has been shown that Axin downregulated β-catenin in SW480 cells but not the Axin mutant that lacks the β-catenin-binding site.29 Similarly, the truncation mutant Axin cannot bind to GSK-3β that is vital for the down-regulation of β-catenin.29 Furthermore, Axin without the MEKK1-interacting domain has a dominant-negative effect on JNK activation by wild-type Axin.23 Taken together the fact that JNK activation by Axin has also been implicated in the regulation of apoptosis, the mutation of Y305 to stop codon definitely renders Axin defective in tumor-suppressing activity.
In conclusion, from 54 colon cancer samples, we detected 6 single alterations that resulted in amino acid changes, with a mutation rate of 11%. The results indicate that Axin is relatively frequently mutated in the colorectal cancer. It could provide a new way for the diagnosis and therapy of colorectal cancers that do not result from mutations of the APC gene. Further studies will be directed to address how these mutations in Axin may alter its function in both the Wnt signaling pathway and the JNK pathway.