Disease-associated casein kinase I δ mutation may promote adenomatous polyps formation via a Wnt/β-catenin independent mechanism

Abstract

The Wnt signaling pathway is critical for embryonic development and is dysregulated in multiple cancers. Two closely related isoforms of casein kinase I (CKIδ and ϵ) are positive regulators of this pathway. We speculated that mutations in the autoinhibitory domain of CKIδ/ϵ might upregulate CKIδ/ϵ activity and hence Wnt signaling and increase the risk of adenomatous polyps and colon cancer. Exons encoding the CKIϵ and CKIδ regulatory domains were sequenced from DNA obtained from individuals with adenomatous polyps and a family history of colon cancer unaffected by familial adenomatous polyposis or hereditary nonpolyposis colorectal cancer (HNPCC). A CKIδ missense mutation, changing a highly conserved residue, Arg324, to His (R324H), was found in an individual with large and multiple polyps diagnosed at a relatively young age. Two findings indicate that this mutation is biologically active. First, ectopic ventral expression of CKIδ(R324H) in Xenopus embryos results in secondary axis formation with an additional distinctive phenotype (altered morphological movements) similar to that seen with unregulated CKIϵ. Second, CKIδ(R324H) is more potent than wildtype CKIδ in transformation of RKO colon cancer cells. Although the R324H mutation does not significantly change CKIδ kinase activity in an in vitro kinase assay or Wnt/β-catenin signal transduction as assessed by a β-catenin reporter assay, it alters morphogenetic movements via a β-catenin-independent mechanism in early Xenopus development. This novel human CKIδ mutation may alter the physiological role and enhance the transforming ability of CKIδ through a Wnt/β-catenin independent mechanism and thereby influence colonic adenoma development. © 2006 Wiley-Liss, Inc.

The Wnt/β-catenin pathway is crucial in multiple aspects of development in multicellular organisms.1 This pathway regulates cell proliferation and differentiation in part by controlling the level of nuclear accumulation of β-catenin.2 In addition, β-catenin-independent functions of Wnt and frizzled receptors are increasingly described. The planar cell polarity (PCP) pathway in Drosophila and the related mechanism regulating convergent extension movements in vertebrates require subsets of Wnt, frizzled and dishevelled (Dvl) molecules for proper hair cell polarity and body axis elongation during gastrulation.3, 4, 5 Downstream of Dvl, activation of JNK (c-Jun N-terminal kinase) and small GTPases, including Rho and Rac, are implicated in the PCP pathway.6 Another noncanonical Wnt pathway, the Wnt/Ca²⁺ pathway leads to activation of protein kinase C and calcium or calmodulin-dependent protein kinase II and regulates cell fate and gastrulation movements as well.7 This pathway has also been found to activate NF-AT, which antagonizes Wnt/β-catenin pathway and promotes ventral cell fate in Xenopus embryos.8

Four isoforms of casein kinase I function in the regulation of β-catenin stability. Casein kinase I α (CSNK1A, CKIα) is a constitutively active negative regulator of the Wnt/β-catenin pathway, interacting with axin and directly phosphorylating β-catenin on serine 45, priming it for regulated phosphorylation by GSK3.9, 10 CKIγ (CSNK1G), a membrane-bound member of the CKI family, phosphorylates LRP5/6 following priming phosphorylation (perhaps by GSK3) after Wnt binds to the cell surface.11, 12 CKIδ (CSNK1D) and the closely related CKIε (CSNK1E) are positive regulators of the Wnt/β-catenin pathway.13, 14 Overexpression of CKIδ or CKIε promotes the expression of Wnt-responsive genes in Xenopus embryos, leading to the formation of ectopic dorsal axis,13 whereas inactivation and knockdown of CKIε blocks β-catenin-dependent signaling.15 Interestingly, CKIε also plays a negative role in Wnt/β-catenin pathway by decreasing the interaction between the Wnt coreceptor LRP6 and Axin, thereby increasing the Axin-mediated degradation of β-catenin.16

Potential mechanisms for the way in which CKIδ and CKIε positively regulate the canonical Wnt pathway have been proposed. CKIε can phosphorylate APC, Dvl, axin, and TCF, destabilizing the β-catenin destruction complex and stabilizing β-catenin/Tcf interaction15, 17, 18, 19, 20, 21 and regulating the ability of Dvl to activate the JNK activity in the noncanonical Wnt pathway.22, 23 Given the evidence for an important role of CKIδ/ε in the Wnt pathway, the regulation of these kinases is of great interest. Intramolecular autophosphorylation of the 13 kDa carboxyl-terminal autoregulatory domain leads to marked inhibition of CKIδ/ε activity, while removal of the inhibitory phosphoryl groups by protein phosphatases or mutation of the sites causes a marked increase in kinase specific activity.24, 25 Furthermore, Wnt signaling also regulates the activity of CKIε and CKIδ. Addition of Wnt ligand triggers dephosphorylation of the autoinhibitory domain of CKIε, leading to the rapid activation of CKIε.26 Mutation of CKIε inhibitory autophosphorylation sites creates an activated CKIε that causes both axis duplication and a distinct phenotype, when ectopically expressed in Xenopus embryos, and is a more potent activator than wild-type CKIε for β-catenin-dependent transcription. Loss of kinase activity or deletion of the autoregulatory domain of CKIε disrupts its ability to induce a secondary axis in Xenopus embryos, implying that both are required for CKIε to positively regulate the canonical Wnt pathway.14

Multiple human cancers are associated with aberrant Wnt signaling. Gain-of-function mutations of Wnt and β-catenin and loss-of-function mutations in APC and axin are found in human epithelial cancers.19, 27, 28, 29, 30, 31 Because CKIδ/ε are positive regulators of the Wnt/β-catenin pathway and their kinase activities are regulated by their carboxyl-terminal autoinhibitory domain, we speculated that mutations of the autoinhibitory domain of CKIδ/ε might constitutively activate the kinase, thereby stimulating Wnt signaling and contributing to the development of adenomatous polyps and colon cancer.

Here we report the identification of a germline missense mutation in human CSNK1D that changes Arg324 to His in a patient with early onset adenomatous polyps. CSNK1D(R324H) is biologically distinct from wildtype CKIδ, as ectopic expression of the mutant but not wild-type gene causes both axis duplication and a distinct phenotype with gastrulation defect in Xenopus embryos. Additionally, the mutant CKIδ is more active than wildtype CKIδ in promoting cellular transformation as assessed by anchorage-independent growth of RKO cells. However, the R324H mutation does not significantly enhance global CKIδ kinase activity in in vitro kinase assays, or increase the potency of Wnt/β-catenin signal transduction as assessed by a Lef-1 reporter assay. The gastrulation defects caused by R324H mutation are not blocked by disrupting the β-catenin pathway. The data suggest that this novel mutation stimulates the activity of CKIδ in a Wnt/β-catenin independent pathway that contributes to colonic polyp formation.

Material and methods

Genomic DNA resources

All aspects involving research subjects were reviewed and approved by the Institutional Review Board of the University of Utah.

The individuals from kindred with familial colon cancer were participants in a high-risk colon cancer research clinic at the Huntsman Cancer Institute at the University of Utah. Pedigrees came from 2 resources: (i) families with multiple colorectal cancers and no known genetic colorectal syndromes were referred, and (ii) families identified as having a significantly increased risk of colorectal cancer (p < 0.01) over the general population through the Utah Population Database (UPDB). UPDB includes over 3 million genealogies abstracted from the Utah Family History Library linked to state-wide cancer records from the Utah Cancer Registry and Cancer Data Registry of Idaho, Utah death certificates and Utah birth certificates.32, 33 Clinic participants were provided with education and genetic counseling on colon cancer risks and underwent colonoscopic evaluation when medically indicated. During clinic visit, a peripheral blood sample was taken from patients for germline genomic DNA analysis. State cancer records, death certificates, and primary medical records were obtained to confirm clinical history of cancer and adenomatous polyps in all participants. Families with known genetic polyposis syndromes were excluded by medical record review and sequencing of the APC gene. Families with hereditary nonpolyposis colorectal cancer (HNPCC) were excluded by sequencing of the MLH1 and MSH2 genes and microsatellite instability testing followed by immunohistochemistry of MLH1, MSH2, and MSH6 in available colorectal cancers.34, 35, 36

Polymerase chain reaction and genomic DNA sequencing

Primer pairs flanking the exons encoding the autoinhibitory domain of human CSNK1D and CSNK1E were designed based on exon structure predicted by Ensembl (www.ensembl.org). CSNK1D exon 7 primer sequences were 5′-GGT TGC TTT CTG CCA AAG CTG GTT CCA TTC CG-3′ and 5′-GTC ACC CCA GAG CCA GCC CCA GAG-3′. CSNK1D exon 8 primer sequences were 5′-CTC TCA TAG AAA GCC CCG AGT GAG AGG CAG C-3′ and 5′-GGT AGC CCG AGG CCC AGC GC-3′. CSNK1D exon 9 primer sequences were 5′-GGC TCC CCC AGC CCC CAT GC-3′ and 5′-GGA CGT GTC ACT AGT AAA GCC ATT GGT AAC AGA GTA GAT CAG CC-3′. CSNK1D alternative exon 9 primer sequences were 5′-GGA TGC CCA TCT GGA CAT GGA GCC GAC AC-3′ and 5′-CAC TCA CCC AGT GCT GCC TTC CGA TGG-3′. CSNK1E exon 8 primer sequences were 5′-GGC TGG AGC AGA AGG TTT CTG AGA TAT CCA CCT GG-3′ and 5′-CCT GCT CAC CAG CCG GCT GGA TGC-3′. CSNK1E exon 9 primer sequences were 5′-GGC TCT GTT TGT GCC TAC TCT GAG GGT GGC-3′ and 5′-GGT CTC TAA CTC AGT TCT GAG GCC CAG AGG GAC-3′. CSNK1E exon 10 primer sequences were 5′-GGA CTT GAC TTG AAG CTT GTA CCT CCC TCT CTC TCC C-3′ and 5′-GTC ATG GAC TCA CCT AAG CAA ACA CTG GTC CAA TGG-3′.

Fifty microliter PCR reactions were set up with 50 ng of patient genomic DNA template (extracted from peripheral blood using the Purgene DNA purification kit (Gentra Systems, Minneapolis, MN)), 1 μM of each forward and reverse primers, 500 μM dNTPs, 1.5 mM MgCl₂ and 3.75 units Taq Polymerase (Qiagen). DNA was amplified for 30 cycles with annealing temperatures optimized for each primer set ranging from 65 to 68°C in an MJ Research Peltier Thermal Cycler Model PTC-200. Samples of PCR products were visualized on a 2% agarose gel and purified with an automated Biomek 2000 system using Millipore 96-well filter purification plates. Five microliters of purified PCR product was then combined with 3.2 pmol primer for each sequencing reaction. Sequencing was done on an automated ABI sequencer in the University of Utah sequencing core facility, and chromatograms of sequences were computer analyzed and visually inspected for mutations.

Plasmids and antibodies

The expression plasmid pCS2-CKIδ was constructed by cloning the full-length CKIδ cDNA into pCS2+ vector.37, 38 Plasmid pCS2-CKIδ(R324H) was then prepared by mutating R324 to H with QuikChange Site-Directed Mutagenesis Kit (Stratagene). pCS2-MT-CKIδ(WT) and pCS2-MT-CKIδ(R324H) with 6 amino-terminal Myc-epitope tags were constructed by cloning the DNA fragment of CKIδ(WT) and CKIδ(R324H) from pCS2-CKIδ(WT) and pCS2-CKIδ(R324H) into pCS2-MT vector. Retroviral plasmids pBABEpuro-MTCKIδ(WT) and pBABEpuro-MTCKIδ(R324H) were constructed by directly cloning a BamH I-SnaB I fragment from pCS2-MT-CKIδ(WT) and pCS2-MT-CKIδ(R324H) into pBABEpuro. Expression plasmids pCS2+XWnt-8, TOPOFLASH, pEV3S-Lef-1, and pRL-SV40 were kindly provided by Randy Moon, Hans Clevers, Marian Waterman, and Don Ayer, respectively. The DN-TCF expression construct was prepared by cloning TCF-DN67 fragment lacking the 30 amino-terminal β-catenin binding residues into the pCS2+ vector. Anti-Myc (9E10) was obtained from Santa Cruz Biotechnology. The anti-CKIδ monoclonal antibody 128A was a generous gift from Icos, Bothell, WA. Myc-CKIδ and CKIδ(R324H) immunoprecipitation-kinase assays were performed as previously described,26 using casein as a substrate.

Xenopus injections and analysis of phenotypes

Sense mRNA of CKIδ(WT), CKIδ(R324H) and DN-TCF were prepared with mMessage mMachine kit (Ambion) using linearized plasmid as templates. RNA was purified with RNAeasy (Qiagen). Kinase RNA (3 to 3.75 ng) or 50 pg of DN-TCF RNA were microinjected into the ventral side of four-cell stage blastomere embryos as previously described.17 The embryos were scored 1 and 3 days after injection.

Colony formation assay

RKO cells were transduced with pBABE-Myc empty vector, pBABE-Myc-CKIδ(WT) or pBABE-Myc-CKIδ(R324H). After selecting with puromycin (2.5 μg/ml) for 2 weeks, the cells were collected and directly subjected to soft agar colony formation assay. Base agar was made with DMEM containing 10% FBS and 0.8% agarose; and top agar was made with DMEM containing 10% FBS and 0.4% agarose. Two hundred and fifty cells were seeded into each of 6 35-mm dishes for each construct, and the colonies were counted after culturing for 1 week.

LEF1-luciferase reporter assay

Twenty-five to 100 ng of pCS2-CKIε(WT), pCS2-CKIε(MM2) or pCS2 empty vector plasmid was transfected into HEK 293 cells in 35-mm dishes along with TOPFLASH(500 ng), pEV3S-Lef-1(100 ng) and pRL-SV40 (100 ng). At 46-h post-transfection, cells were harvested with passive lysis buffer (Dual-Luciferase Reporter assay system, Promega), followed by the firefly luciferase and Renilla luciferase assay measured with the Dual-Luciferase Reporter assay system (Promega) and Microtiter Plate Luminometer (Dynex Technologies) according to the manufacturer's instructions. To standardize for transfection efficiency, the luciferase activities of all transfected cells were divided by the Renilla luciferase activities. The fold increase of Lef-1 dependent activity was defined by comparing the normalized level observed from cells transfected with pCS2. Data are presented as mean ± SD from 2 separate experiments.

Results

Identification of a heterozygous mutation in exon 7 of CKId

The activity of CKIδ and CKIε both in vitro and in vivo is regulated by autophosphorylation of a carboxyl-terminal autoinhibitory domain. Additionally, this domain may be involved in the interaction of the kinase with specific components in the Wnt/β-catenin pathway.14 Both kinases are positive regulators of the Wnt/β-catenin signaling pathway. Upregulation of Wnt/β-catenin pathway plays a causal role in the development of some adenomatous polyps. Therefore, we hypothesized that germline mutations in the inhibitory domain of CKIε or CKIδ that blocked autoinhibition might lead to constitutive rather than regulated kinase activity and that this increase in kinase activity might stimulate Wnt/β-catenin signaling and contribute to the development of adenomatous polyps or colon cancer.

To test this hypothesis, we sequenced germline genomic DNA from individuals in 13 different kindreds with familial colon cancer not due to HNPCC or mutations in APC.39 As index cases, individuals with advanced colonic neoplasms (colon cancer, adenomatous polyps ≥10 mm, or adenomatous polyps with high grade dysplasia) in each pedigree for whom genomic DNA samples were available were selected (n = 23 individuals). Primers flanking each of the exons encoding the autoinhibitory domains of CKIδ and CKIε were used to amplify the exons, and the resulting PCR products were directly sequenced (Fig. 1a). Silent mutations in both CSNK1D (Pro358(CCC→CCT) and Arg389(CGA→CGG)) and CSNK1E (Thr407(ACA→ACG)) were found. A heterozygous missense mutation was identified in 1 individual in exon 7 of CSNK1D: CGC→CAC, resulting in histidine replacing arginine at position 324 (R324H) (Fig. 1b). This region was reamplified and resequenced on 3 separate occasions to confirm the result.

Figure 1. A heterozygous missense mutation is found in CKIδ. (a) Schematic representation of human CKIδ and CKIε. The autoinhibitory domain of CKIδ and CKIε are encoded by exons 7–9 or alternative splicing exon 9 and by exons 8–10, respectively. The DNA of each exon was amplified by PCR using the primers indicated by the arrows. The amplified PCR products were directly subjected to DNA sequencing. (b) Representative DNA sequence from the sample with the G/A mutation in exon 7 of CKIδ. The DNA sequence corresponding to nucleotides 71–104 of CKIδ exon 7 is shown here. The red arrow indicates the site of the heterozygous mutation. The codon change, from CGC to CAC, changes residue 324 from Arg to His. (c) R324 is highly conserved in vertebrate CKIδ and CKIε. Shown is a selected region of the CLUSTALW alignment of the indicated CKI autoregulatory domains.

This missense mutation is unlikely to be a polymorphism. First, R324 is completely conserved in vertebrates including zebrafish, chicken, rodents, frog and human CKIδ and CKIε (Fig. 1c) consistent with a significant role in CKIδ/ε function. Second, the mutant sequence is not found in BLAST searches of mammalian EST databases including 67 human CSNK1D and 100 human CSNK1E sequences. Finally, this alteration has not been found in 75 control individuals (Ying-Hui Fu and Louis Ptacek, personal communication).

R324H mutation in CKIδ is present only in the patient with prominent polyp formation

The CSNK1D(R324H) mutation was identified in an individual with early onset multiple adenomatous polyps in a pedigree with a high incidence of colon cancer and its precursor lesion, adenomatous polyps. To determine whether a germline mutation of CSNK1D was the cause of familial adenomatous polyps and colon cancer in this pedigree, genomic DNA from individuals on both the paternal and maternal branches of the family was sequenced. As Figure 2 shows, the proband (indicated by arrow) has an earlier onset of polyps at a substantially younger age (46 years old) and developed more and larger polyps than his 2 affected siblings (58 and 55 years old). The R324H mutation was not present in either sibling, nor in 4 second- and 8 third-degree relatives in either the maternal and paternal branches. DNA was not available on the mother and father, who both died of colon cancer. Hence, the R324H mutation may increase the penetrance of polyp and cancer risk in the proband (as reflected in earlier onset and larger polyp size) but is not the cause of the adenomatous polyps and colon cancer in other members of this pedigree.

Figure 2. The mutation of R324H in CKIδ is present in the patient with prominent polyp formation. Pedigree of the family containing the patient with the R324H mutation. The arrow designates proband described containing the R324H mutation, who developed more and larger polyps and at a younger age than the siblings. Colorectal cancer cases are shaded black and other cancers are shaded gray with the cancer and the age at diagnosis listed. Adenomatous polyps are represented as triangles with a dot. The number of adenomas for each size category, and the age when the adenoma was identified are listed for phenotyped individuals. Sequencing results for the CSNK1D mutation is indicated as wild-type (wt) or R324H. Individual sexes are masked to enhance confidentiality; full pedigrees are available upon request.

Biological effects of CKIδ(R324H)

Given the high conservation of R324 and its proximity to T323, a residue implicated in the regulation of CKIε activity,24, 26 we asked what effect the CKIδ(R324H) mutation had on the Wnt/β-catenin signaling pathway in vivo. Wnt/β-catenin signaling normally induces dorsalization during Xenopus embryogenesis. Ectopic stabilization of β-catenin by Wnt signaling in ventral cells alters the cell fate specificity, directing ventral cells to dorsal cells and thereby resulting in the formation of a secondary dorsal axis. CKIδ, a positive transducer of Wnt/β-catenin signaling pathway, mimics the effect of Wnt signaling, leading to the formation of a secondary axis in Xenopus embryos.13 To test if the R324H mutation alters the biological effect of CKIδ on Xenopus development, we microinjected mRNA encoding myc-tagged wildtype or R324H into one of the ventral cells of 4-cell-stage embryos. The embryos were allowed to develop for 3 days at room temperature and the resultant phenotype was scored at 1 and 3 days. As previously reported, ventral expression of wildtype CKIδ RNA induces the formation of an ectopic dorsal axis (Fig. 3a (ii)). However, expression of CKIδ(R324H) causes secondary axis formation and also results in a distinct phenotype in a reproducible and significant number of embryos (Fig. 3a (iii)). These embryos displayed a gastrulation defect and commonly had a shortened and bent tail along with a shortened anterior–posterior axis. This phenotype was seen only rarely at the highest dose of wildtype CKIδ injection. Similar results were observed after the injection of untagged kinases (data not shown), implying that the myc tag did not affect the teratogenic action of CKIδ. The gastrulation defect was due to a change in CKIδ(R324H) function rather than abundance, since immunoblotting of the Xenopus embryo extracts showed no obvious difference in the abundance of Myc-CKIδ(WT) and Myc-CKIδ(R324H) protein (Fig. 3b). These results suggest that the R324H mutation alters the biochemical properties of CKIδ in Xenopus development.

Figure 3. The biological effects of CKIδ(R324H). (a) Forced expression of CKIδ(R324H) causes a distinct phenotype in Xenopus embryogenesis. mRNA (3.75 ng) encoding CKIδ(WT) or CKIδ(R324H) was microinjected into one of the ventral cells of 4-cell Xenopus embryos. Embryos were scored and photographed at 24 and 72 hr of postinjection (stage 18–20 and stage 37–40). (i) Representative uninjected control. (ii) CKIδ expression leads to axis duplication. Black arrowheads indicate the secondary axis. (iii) Expression of CKIδ(R324H) causes a distinct phenotype in approximately 40% of tadpoles. (iv) Kinase-dead CKIδ loses the ability to induce axis duplication. (v) Kinase-dead CKIδ(R324H) loses the ability to induce the distinct phenotype. The representative pictures of tadpoles are shown in the upper panel and the scoring data from stage 18 are shown in the bottom panel. Black arrowhead, double axis; white arrowhead, gastrulation defect. This experiment was repeated 3 times with similar results. The scoring data shown here is from 1 experiment with a total of 170 embryos (35 uninjected controls, 32 with CKIδ(WT), 33 with CKIδ (R324H), 35 with CKIδ(K38A), and 35 with CKIδ(R324H/K38A)). (b) Immunoblotting of 1-day embryo extracts demonstrates equal expression of wildtype and mutant CKIδ.

As mentioned in the Introduction, kinase-inactive CKIε(K38A) loses the ability to induce secondary axis formation in Xenopus embryos, suggesting that kinase activity is required for secondary axis formation. To further understand the gastrulation defect induced by CKIδ(R324H), we asked whether kinase activity is required. To assess this question, a CKIδ mutant with both R324H and K38A mutations was injected into the ventral cell of 4-cell stage Xenopus embryos. As shown in Fig. 3a (iv) and (v), kinase activity is required both for secondary axis formation and development of the gastrulation defect, as the CKIδ(K38A/R324H) mutant did not produce developmental defects.

Given the evidence that the R324H mutation alters the activity of CKIδ, we next tested if this mutation confers transforming ability on CKIδ. One characteristic of transformed cells is the ability to proliferate in suspension cultures or in a semisolid medium without attachment to a surface.

To test whether mutant CKIδ expression in fact conferred anchorage-independent growth, RKO cells were transduced with a retrovirus encoding either Myc-CKIδ(WT) or Myc-CKIδ(R324H). Following selection in puromycin for 2 weeks, bulk cultures were collected and tested for their ability to form colonies in soft agar. RKO cells transformed with empty pBabe-puro vector rarely formed colonies (Fig. 4a). Forced expression of CKIδ(WT) and CKIδ(R324H) lead to growth of approximately 6 and 15 colonies per 250 cells plated, respectively, indicating that first, wildtype CKIδ expression has transformation abilities, and second, CKIδ(R324H) is more potent than wild type kinase in promoting cell-anchorage independence. Wildtype and mutant kinases were expressed at equal levels in the pooled cultures (Fig. 4b), indicating again that the difference in colony formation is due to a change in kinase function rather than kinase abundance.

Figure 4. Expression of CKIδ(R324H) promotes anchorage-independent growth of RKO cells. (a) Soft agar colony formation assay. RKO cells were transduced with pBABE-puro-Myc-CKIδ(WT), pBABE-puro-Myc-CKIδ(R324H) or empty vector. After selection in puromycin, 250 cells were seeded in soft agar in a 35-mm dish. One week later, the soft agar plates were stained with crystal violet and the cell colonies were counted under microscope. Six plates were seeded with each pool of 250 transduced RKO cells, and the experiment was performed on 2 separate occasions with similar results. The data shown here is the mean ± SD from 6 plates. (b) Immunoblot of transduced RKO cells demonstrates equal expression of wildtype and mutant Myc-CKIδ.

Taken together, these data suggest that R324 is an important residue in CKIδ. The mutation of R324H alters the function of CKIδ, leading to a distinct phenotype in Xenopus embryos via a kinase-dependent mechanism and enhanced transforming activity reflected in its ability to increase anchorage-independent cell growth.

The R324H mutation does not alter the general kinase activity or change the autophosphorylation status of the C-terminal inhibitory domain

The ability of CKIδ(R324H) to cause novel developmental abnormalities and to promote colony formation in soft agar suggests the point mutation causes a gain of function. One potential consequence of the mutation could be decreased autoinhibition of CKIδ due to decreased autophosphorylation of an inhibitory site, such as T323. To test this, we performed in vitro phosphopeptide mapping on autophosphorylated CKIδ and CKIδ(R324H). No differences in the in vitro autophosphorylation pattern were observed (data not shown).

CKIε is rapidly dephosphorylated and activated in vivo by addition of Wnt-3A or Wnt-8. Because CKIδ is regulated by autophosphorylation, we next tested if the R324H mutation altered the activation of CKIδ by Wnt. An immunoprecipitation-kinase (IP-kinase) activity assay was performed with Myc-CKIδ and Myc-CKIδ(R324H) expressed without or with Wnt-8 in HEK293 cells. As shown in Fig. 5a, both CKIδ(WT) and CKIδ(R324H) had low basal activity and their kinase activity increased 5-fold by coexpression of Wnt-8. These data suggest that the R324H mutation does not enhance the basal activity nor alter the activation of CKIδ by Wnt signaling, at least when casein is used as an in vitro substrate. Taken together, the phosphopeptide mapping and IP-kinase data suggest that the R324H mutation does not affect kinase autophosphorylation or its in vivo activation by Wnt signaling.

Figure 5. R324H mutation does not affect in vivo kinase regulation nor Lef-1 dependent transcription. (a) Kinase activity assay. Myc-CKIδ(WT) and Myc-CKIδ(R324H) were expressed or coexpressed with Wnt-8 in HEK293 cells and immunoprecipitated by Myc Ab (9E10). The kinase activity of Myc-CKIδ(WT) and Myc-CKIδ(R324H) were then assessed by incubating with γ-³²P-ATP and using casein as a substrate. Samples were then analyzed by SDS-PAGE and autoradiography. The data shown here were normalized by the level of immunoprecipitated kinase and subtracted out the background kinase activity in the absence of Myc-kinase expression. Data are presented as mean ± SD. (b) Lef-1 reporter assay. Activation of the pTOPFLASH reporter plasmid by CKIδ(WT) and CKId(R324H) was measured as described (Material and Methods).

CKIδ(R324H) alters development in a β-catenin independent manner

CKIδ is a positive regulator of the Wnt/β-catenin signaling pathway and forced expression of CKIδ results in the secondary axis formation in Xenopus embryos. Because ectopic expression of CKIδ(R324H) additionally leads to a gastrulation defect (Fig. 3a (iii)), we tested if the R324H mutation enhanced the activity of CKIδ in the Wnt/β-catenin pathway. Activation of the Wnt/β-catenin signaling pathway leads to transcription from Lef-1/Tcf responsive promoters. Therefore, we compared the ability of wild-type and mutant CKIδ to drive expression of a Lef-1 reporter plasmid. As Figure 5b shows, the expression of CKIδ and CKIδ(R324H) each caused an increase in Lef-1 reporter activity in a dose-dependent manner. However, the transactivating activities of CKIδ(WT) and CKIδ(R324H) are not significantly different, indicating that CKIδ(R324H) is no more potent than wildtype CKIδ in the activation of β-catenin dependent gene transcription.

The above data suggest that the ability of CKIδ(R324H) to enhance polyp formation in the proband and cause a gastrulation defect in Xenopus development might be independent of its role in the Wnt/β-catenin signaling pathway. To directly test this, CKIδ and CKIδ(R324H) were expressed in Xenopus embryos, and downstream β-catenin signaling was disrupted by coexpression of dominant negative TCF (DN-TCF). DN-TCF lacks the first 30 amino acids required for β-catenin binding,40 binds to DNA, but is not activated by increasing β-catenin abundance. As expected, coexpression of DN-TCF blocks the formation of the secondary axis induced by CKIδ, demonstrating that CKIδ is upstream of β-catenin and requires β-catenin for its function (Fig. 6a). Notably, coexpression of DN-TCF4 blocks the secondary axis induced by CKIδ(R324H) but does not block formation of the gastrulation defect. This indicates that the gastrulation defect is due to a β-catenin independent pathway.

Figure 6. R324H mutation may alter the physiological roles of CKIδ in a β-catenin independent manner. (a) DN-TCF blocks the Wnt/β-cat pathway but not the distinct phenotype. CKIδ(WT) or CKIδ(R324H) mRNA (3 ng) was coinjected with DN-TCF mRNA (50 pg) into one of the ventral cells of Xenopus embryos at 4-cell stage. Embryos were scored after 24 and 72 hr and photomicrograph of representative tadpoles after 72 hr of development are shown. This experiment was repeated twice with similar results. The scoring data shown here is from 1 experiment with a total of 135 embryos (25 uninjected controls, 26 with CKIδ (WT), 28 with CKIδ(WT)/DN-TCF, 27 with CKIδ(R324H) and 28 with CKIδ(R324H)/DN-TCF). (b) R324H mutation causes a severe defect in blastopore closure. Three nanograms of CKIδ(WT) or CKIδ (R324H) mRNA was injected into one of the ventral cells of Xenopus embryos at the 4-cell stage. Photomicrographs of embryos were taken at stages 12, 13 and 18. The embryos were scored at stage 18 for blastopore closure. Red, black and white arrowheads indicate the blastopores, the secondary axis and the leakage of yolk cells, respectively.

We then asked what additional pathways the R324H mutation might alter. CKIδ and CKIε are regulators of convergent extension, a key Wnt-regulated but β-catenin independent process of cell migration and shape change required for anterior–posterior (AP) axis development.23 Blastopore closure defects, a subset of gastrulation defects, commonly accompany defects in convergent extension movements. Because the embryos with the distinct phenotype have a short anterior–posterior axis, we speculated that the R324H mutation might cause defects of blastopore closure. CKIδ(R324H) expression caused a significant blastopore closure defect, with 21/43 injected embryos displaying an open blastopore, compared with only 2/43 embryos injected with wildtype CKIδ (Fig. 6b). Additionally, CKIδ(R324H) induced defect of blastopore closure caused deadherence and leakage of yolk cells at stage 13 and 18. These data suggest the CKIδ(R324H) mutation alters cell migration and shape change required for convergent extension.

Discussion

Casein kinase I family members have been implicated in regulation of the Wnt/β-catenin signaling pathway and act to inhibit the β-catenin-independent convergent extension pathway. Here we identify a germline mutation in CSNK1D that appears to increase the number and size of adenomatous polyps in a relatively young individual. The mutant allele does not appear to be a polymorphism, since it was not identified in 75 control individuals, EST databases, or other colon cancer pedigrees. Furthermore, it alters a residue that is highly conserved within all vertebrate CKIδ and CKIε proteins. Given the effect of the mutation on colony formation in soft agar and the requirement for kinase activity to produce the developmental phenotype in Xenopus, it appears to be a gain of function, although we cannot exclude a dominant negative effect of an overexpressed protein. Fuja et al. reported somatic mutation of CKIε associated with loss of heterozygosity (LOH) in breast cancer, although functional studies have not been reported.41 Although it would be desirable to examine polyps in our patient for evidence of LOH, the tissue blocks were not available.

R324 is a highly conserved residue in the autoregulatory domain of CKIδ. Although the mutation to histidine is a conservative change, it clearly has significant biological effects. Ectopic expression of the mutant kinase in Xenopus embryos caused the delay of blastopore closure, shedding of yolk cells, and tadpoles with a short anterior–posterior axis. Expression of the mutant kinase in the RKO colon cancer cell line led to an increase in anchorage-independent colony formation, consistent with a role for this mutation in promoting transformation of cells.

The autoregulatory domain of CKIδ and CKIε not only mediates the response of the kinase to extracellular signaling such as Wnt, but may also play a role in protein–protein interactions with various substrates in the cell. Although our initial speculation was that mutations in autophosphorylation sites would cause continuous kinase activation, the R324H mutation did not cause a change in kinase autophosphorylation nor in basal- or Wnt-stimulated activity. The mutation may alter the interaction of CKIδ with signaling proteins and thus alter the rate of phosphorylation of critical substrates that regulate convergent extension in Xenopus embryos and transformation in the mammalian cells. One potential target in this cascade is the dishevelled protein. Dvl is involved in both canonical and noncanonical Wnt signaling pathways and is known to both bind to and be phosphorylated by CKIε and CKIδ. However, wild type and mutant CKIδ bound to and phosphorylated Dvl-1 equally well in transient expression studies (data not shown).

Because the CKIδ mutation did not function through the Wnt/β-catenin pathway, we considered that its pathologic effect might be mediated through alteration of the Wnt/PCP pathway instead. Aberrant regulation of Wnt/PCP signaling has been found to be associated with malignant progression in human cancers, leading to the abnormal tissue polarity, invasion and metastasis.42, 43, 44 Therefore, our finding that the Xenopus embryos expressing CKIδ(R324H) display an open blastopore suggests that the R324H mutation alters cell–cell interaction and invasion potency.

Potential targets of the R324H mutant kinase are suggested by its role in both abnormal gastrulation in Xenopus embryos and transformation. Gastrulation, which involves narrowing the mediolateral axis (convergence) and extending anterior–posterior axis (extension), is one of the critical steps during vertebrate development. Convergent extension (CE) movements are stimulated by Wnt/PCP signaling driving cytoskeleton reorganization. A similar Wnt-regulated mechanism has been found to regulate cell morphology, cell–cell interaction, and cell motility.45, 46 To explore the role of CKIδ(R324H) in the regulation of the Wnt/PCP pathway, we tested for changes in the downstream effects JNK and RhoA. However, no differential effect of mutant CKIδ was observed in those experiments. We speculate that either (i) the effects of the R324H mutation are biochemically subtle and the phenotypic consequences evolve over many years and, therefore, may be difficult to detect in our single-point assays or that (ii) CKIδ(R324H) may affect Wnt/PCP pathway through an unknown downstream pathway, other than RhoA and JNK. We also assessed if the R324H mutation changes the interaction between CKIδ and NF-AT, which is activated by Wnt/Ca²⁺ pathway.8, 47 Again, no significant difference was observed. Clearly, further study is needed to better elucidate the downstream targets of the CKIδ(R324H) mutation.

Our inability to find changes in the activity of the mutant CKI relate to our ignorance of the relevant substrate. Our recent studies in an unrelated CKIε circadian rhythms mutant suggest that kinase point mutations can alter kinase activity in an extremely substrate dependent manner.48 These alterations in activity or specificity may also have relatively subtle biochemical consequences that take years to demonstrate a phenotype, and the biochemical difference between wild type and mutant kinase may be hard to detect in short-term assays. Less subtle and more active germline mutations in CKI family members may produce more lethal phenotypes. Supporting this, in a study of the Drosophila homolog of CKIε and CKIδ, double-time (also known as disks overgrown), most mutations cause severe developmental defects related to effects on cell proliferation and survival that might not be tolerated in human development.49

In summary, we report here identification of a germline mutation of the regulatory domain of CKIδ as a potential modifier of polyposis in an individual from a relative with increased cancer risk. The mutation is biologically active, enhancing anchorage independent growth and altering cell movements required for blastopore closure and anterior–posterior axis formation in Xenopus. We speculate that the mutation affects kinase interaction with β-catenin-independent substrates and alters the process of convergent extension. In future studies, it may be important to examine additional cancers for both inherited and acquired mutations in the CKIδ/ε regulatory domains.

Acknowledgements

We thank Ms. Amy Lee for expert assistance with clinical samples, and Mr. Prof. David Jones and Mr. Prof. Scott Kuwada for critical reading of the manuscripts. This research was supported by the Huntsman Cancer Foundation, the Willard Snow Hansen Chair for Cancer Research (D.M.V), NIH R01 CA40641 and P01 CA73992 to R.W.B, and NIH R01 CA80809 to D.M.V. DNA sequencing and oligonucleotide synthesis was supported by Cancer Center Support Grant P30 CA42014. This research is also supported by the Utah Cancer Registry, which is funded by contract No. N01-PC-3541 from NCI with additional support from the Utah Department of Health and the University of Utah.