Research ArticleCell Biology

A Cancer-Associated Mutation in Atypical Protein Kinase Cι Occurs in a Substrate-Specific Recruitment Motif

See allHide authors and affiliations

Science Signaling  17 Sep 2013:
Vol. 6, Issue 293, pp. ra82
DOI: 10.1126/scisignal.2004068

Abstract

Atypical protein kinase Cι (PKCι) has roles in cell growth, cellular polarity, and migration, and its abundance is frequently increased in cancer. We identified a protein interaction surface containing a dibasic motif (RIPR) that bound a distinct subset of PKCι substrates including lethal giant larvae 2 (LLGL2) and myosin X, but not other substrates such as Par3. Further characterization demonstrated that Arg471 in this motif was important for binding to LLGL2, whereas Arg474 was critical for interaction with myosin X, indicating that multiple complexes could be formed through this motif. A somatic mutation of the dibasic motif (R471C) was the most frequent mutation of PKCι in human cancer, and the intact dibasic motif was required for normal polarized epithelial morphogenesis in three-dimensional cysts. Thus, the R471C substitution is a change-of-function mutation acting at this substrate-specific recruitment site to selectively disrupt the polarizing activity of PKCι.

Introduction

Atypical protein kinase Cs (aPKCs) are serine/threonine protein kinases composed of two closely related human homologs, PKCι and PKCζ, that play roles in cell growth, proliferation, migration, and survival (14). Increased abundance of PKCι mRNA or protein or both is found in various human epithelial malignancies and is correlated with poor patient outcome or cancer grade (512). Loss of apical-basal polarity is a hallmark of aggressive cancers (13), and the abundance of PKCι-interacting polarity proteins is correlated with human malignancy, although this is context-dependent (1416). This relationship with malignancy is supported by fly and mouse models that implicate decreases in the abundance of polarity proteins in tumorigenesis (15, 17, 18). In view of the disease linkage and previous successes at inhibiting kinases in cancer therapy, aPKC itself is an attractive therapeutic target, and several groups have reported anticancer effects of aPKC-directed interventions (1921).

aPKC complexes are evolutionarily conserved from invertebrates to humans (22). aPKC binds to and phosphorylates the polarity protein Par3 (23) and engages Par6 through the Phox Bem1 (PB1) domain. These interactions contribute to the formation of the trimeric, apically located, polarity Par complex (2, 22, 24). Normal apico-basal polarity is achieved by mutual antagonism between the Par complex and the Scribble complex, a basolateral determinant that contains three tumor suppressor proteins: Lethal Giant Larvae, Discs Large, and Scribble (SRB). For example, the scaffold protein Lethal Giant Larvae (single fly isoform, LGL; two human isoforms, LLGL1 and LLGL2) can disrupt the apically localized Par3 complex by competing with Par3 for Par6-aPKC binding (2528). Conversely, aPKC can phosphorylate LLGL within an extended surface loop, leading to translocation from the cell cortex to the cytoplasm, and thus exclusion from aPKC-containing apical regions (29). However, the LLGL interaction site on aPKC has remained elusive.

Functionally validated aPKC substrates include Par3 and LLGL2 as well as the kinases ROCK1 and MARK2 and the Hippo pathway component Kibra (2933). These substrates have basic residues flanking a phosphoacceptor site consistent with basophilic AGC kinase consensus sites (https://www.kinexus.ca). However, short peptide substrates (4 to 14 residues) bearing such motifs are often poor surrogates for intact protein substrate counterparts presented in a cellular context. The recognition elements outside of these motifs that facilitate aPKC substrate recruitment are not known. More generally, protein kinase–substrate interactions must balance the need for both high affinity and specificity against the demands of efficient substrate turnover and phospho-protein release (34). Some serine/threonine kinases have substrate-docking sites that are remote from the phospho-acceptor site, and serve to increase substrate affinity and specificity (35, 36).

Here, we identified a docking site motif in subdomain X of the kinase fold motif of aPKC by inspecting lattice contacts in a PKCι inhibitor crystal structure. This invariant dibasic motif, Arg-Ile-Pro-Arg (RIPR), which is also present in other AGC kinases, is not involved in trans-activation but serves instead as a recruitment platform for LLGL2 but not Par3 substrates. Mutagenesis of the RIPR motif in PKCι leads to defective epithelial polarity in three-dimensional (3D) cyst morphogenesis, and this motif is mutated in human cancers (R474C), suggesting a mutation-driven loss of polarizing capacity in malignancy.

Results

Structural analysis reveals an invariant dibasic RIPR motif within the PKCι kinase domain

We previously determined the crystal structure of a bis-phosphorylated PKCι kinase domain bound to a selective adenosine 5′-triphosphate (ATP)–competitive inhibitor (37). Inspection of this structure revealed an interesting lattice contact between adjacent pairs of PKCι kinase domain molecules [Fig. 1A: Protein Data Bank (PDB) code 3ZH8] not seen in previous PKCι crystal structures (23, 38, 39). Residues 469 to 485 and 510 to 514 from the C lobe of one kinase molecule (Fig. 1A, magenta) engage N-lobe residues 279 to 287 and 558 to 561 from a symmetry-related molecule (Fig. 1A, green). Central to this contact is an Arg-Ile-Pro-Arg (RIPR) motif (amino acids 471 to 474) that precedes helix H/subdomain X from the C lobe of one kinase domain. This RIPR motif interacts with the B helix from the N lobe of a second kinase domain to form an asymmetric homodimer (Fig. 1B). Each of the three independent copies of the kinase domain within the crystal lattice forms the same asymmetric homodimer through RIPR motif contacts, which in turn leads to open helical assembly of kinase domains. This attracted our attention for several reasons. First, the arginine side chains are invariant in all aPKCs from vertebrates to invertebrates (Fig. 1C), although they are spatially located far from the catalytic site of the kinase and on the surface of the molecule. Second, we noticed that the homodimeric interaction involved an extended network of side chain–mediated hydrogen bonds (closer than 3.0 Å). The RIPR motif hydrogen-bonding network extends from the phospho-Thr403 in the activation loop at one extremity of the interface, with both arginine side chains flanking an intervening glutamate side chain from the acidic PKCι B-helix on the other side (Fig. 1B). Third, a RIPR consensus motif that considered both local sequence and secondary structure was conserved in a subset of AGC kinases (PKCι, PKCζ, Akt1, Akt2, PKN1, and PKN2), suggesting a more generalized function (Fig. 1D). Finally, we noticed that the RIPR motif in subdomain X has a similar main-chain backbone conformation to the equivalent RLPQ motif in epidermal growth factor receptor (EGFR), which contributes to an interface within an asymmetric EGFR homodimer that functions during trans-activation (40) (Fig. 1D and fig. S1A).

Fig. 1 The RIPR motif of PKCι is a conserved region with structural equivalence to a trans-activating region of EGFR.

(A) The RIPR motif (magenta) is within the C lobe of the PKCι kinase domain (PDB code 3ZH8) and lies spatially distant from the nucleotide pocket occupied by an ATP-competitive chemical inhibitor (orange). (B) The RIPR motif (magenta) makes contact with the B-helix (green) of a neighboring PKCι kinase domain molecule within the crystal lattice, forming an asymmetric dimer pair. (C) Sequence conservation of aPKC homologs close to the RIPR motif. Residues that contribute to the lattice contact are indicated (arrows). Bottom: Structure-based sequence alignment of the RIPR motif in PKCι with the EGFR subdomain X (residues 470 to 488 of PKCι and 908 to 927 of EGFR), highlighting the equivalent contacts (arrowheads) made within their respective asymmetric dimers. (D) Conservation of the RIPR motif among related AGC kinase subfamily members mapped onto a phylogenetic tree derived from Manning et al. (74). Sequence database search was made taking into account both the local sequence and the secondary structural context of the RIPR motif to define the consensus motif I-X(5)-R-X-P-R-X(5)-A-X(9)-K-X(4)-R-X-G.

We considered two possibilities: either the motif could influence higher-order assembly that consequently regulated kinase activity, or it could serve as a protein interaction motif for unidentified partner(s).

The RIPR motif is dispensable for PKCι activity but necessary for phosphorylation of LLGL2

To investigate whether the RIPR motif regulated kinase activity, we tested the catalytic potential of wild-type PKCι (PKCι-WT) and a double alanine AIPA mutant (PKCι-AIPA) both in cells and in vitro. In human embryonic kidney (HEK) 293 cells cotransfected with LLGL2 and either PKCι-WT, PKCι-AIPA, or the predicted catalytically inactive PKCι-D368N mutant (41), PKCι-AIPA had lower activity than PKCι-WT toward the full-length LLGL2 substrate as determined by immunoblotting with a phosphospecific LLGL1/2 antibody (Fig. 2, A and B). To determine whether the RIPR motif functions as a trans-activating region, analogous to the corresponding motif in EGFR (42), we attempted to rescue the activity of the PKCι-AIPA mutant using PKCι-D368N, a catalytically inactive mutant that has an intact RIPR motif. We created high localized PKCι protein concentrations by adding a membrane-targeting myristoylation sequence (myr) to the PKCι constructs (43), simulating the conditions used to promote EGFR trans-activation (42). Coexpression of the kinase domain versions of myr-PKCι-AIPA or myr-PKCι-D368N with full-length myr-PKCι-WT resulted in a dominant-negative effect on LLGL2 phosphorylation. This effect did not occur when full-length mutants were expressed with kinase domain myr-PKCι-WT, likely because of the reduction in the protein abundance of full-length PKCι that occurred with coexpression of the kinase domain constructs (protein ratio of full-length/kinase domain, 1:5). However, the myr-PKCι-D368N construct did not rescue myr-PKCι-AIPA–induced phosphorylation of exogenous LLGL2 (Fig. 2, C and D), which suggests that activation of PKCι does not occur by trans-activation in a RIPR-dependent fashion.

Fig. 2 The RIPR motif is involved in protein-protein interaction rather than catalytic activity.

GFP-PKCι(WT), GFP-PKCι(AIPA) (RIPR>AIPA mutant), GFP–PKC-D368N, and empty vector, as full-length (FL) or kinase domains (KD), were cotransfected with LLGL2 in HEK293 cells. (A and B) Phosphorylation of LLGL2 was revealed by Western blot with pLLGL1/2 antibody (A). Representative immunoblots of three independent experiments are shown. (B) For quantification, pLLGL was normalized to the total PKCι in each lysate and then to the PKCι(WT) control. (C and D) Full-length and kinase domain cDNA of GFP-PKCι/PKCι mutants were cotransfected with LLGL2, and the phosphorylation of LLGL2 was measured (C). PKCι(AIPA) activity was not rescued by PKCι(D368N) (see arrowheads). Representative immunoblots of four independent experiments are shown. Quantification of the pLLGL2 immunoblots in (D). (E and F) The specific activity of recombinant PKCι-WT and PKCι-AIPA protein was evaluated in an in vitro kinase assay with either PKCε pseudosubstrate synthetic peptide (E) or full-length LLGL2 (F) as the substrate. Bar graphs present data amalgamated from three independent experiments. Molecular weight markers in kilodaltons are displayed on the left of the Western blots. Means ± SD are shown. Differences are compared with PKCι-WT control. ***P < 0.001; **P < 0.01; *P < 0.05; ns, nonsignificant.

We tested the ability of purified, baculovirus-expressed PKCι-WT and PKCι-AIPA kinase domains to catalyze the phosphorylation of an established PKC synthetic peptide substrate (based on an A-to-S point mutation of the PKCε pseudosubstrate sequence) (44) and found that PKCι-WT and PKCι-AIPA displayed the same kinetic properties (Fig. 2E and fig. S1, A to C).

To address the inability of the PKCι-AIPA mutant to phosphorylate LLGL2 in cells despite comparable enzyme kinetics toward the synthetic peptide substrate, we determined its ability to catalyze LLGL2 phosphorylation in vitro. The efficiency of the PKCι-AIPA mutant in phosphorylating full-length LLGL was only 40% of PKCι-WT (Fig. 2F and fig. S1D). Hence, catalytic potential toward a synthetic peptide was normal, but recognition of LLGL2 protein was compromised by the RIPR>AIPA mutation. The fact that the RIPR motif contributes to but is not sufficient for phosphorylation of LLGL in vitro suggests that it adds specificity but is not an absolute requirement. This result indicates that the RIPR motif may play a specific role in substrate recognition.

LLGL2 interacts with the RIPR motif on PKCι

To test whether the interaction of LLGL2 with PKCι required the RIPR motif, we assessed the presence of LLGL2 in GFP (green fluorescent protein)–PKCι-WT and GFP–PKCι-AIPA immunocomplexes. Less LLGL2 bound to the PKCι-AIPA mutant than to PKCι-WT (Fig. 3A). Par3 and Par6B formed an oligomeric complex with aPKC (22, 45), and both of these proteins bound to similar extents to both PKCι-WT and PKCι-AIPA, indicating that these interactions did not require the RIPR motif (Fig. 3A). This suggested that LLGL2, but not Par3, bound the RIPR motif, which was confirmed by loss of the immunoreactive band in GFP–PKCι-WT immunocomplexes from cells pretreated with small interfering RNA (siRNA) directed against LLGL2 (Fig. 3B).

Fig. 3 LLGL2 interacts with the RIPR motif of PKCι.

(A) GFP immunoprecipitates (GFP-IP) from HCT116 cells expressing GFP, GFP-PKCι(WT), or GFP-PKCι(AIPA) and lysates were immunoblotted with the indicated antibodies. Representative immunoblots of three independent experiments are shown. (B and C) GFP immunoprecipitates and lysates from HCT116 cells reverse-transfected with siRNA-LLGL2 or scrambled siRNA, and cells expressing GFP-PKCι were immunoblotted for LLGL2, Par3, and Par6B (B). Representative immunoblots of three independent experiments are shown. Molecular weight markers in kilodaltons are displayed on the left of the Western blot. (C) Quantification of the amount of the coimmunoprecipitated proteins normalized to that in the lysates (both LLGL2 IP values were normalized to siSCRAM lysate input) and the PKCι(WT) complex. Means ± SD are shown. **P < 0.01; ns, not significant.

Mutation of the RIPR motif in PKCι leads to a disrupted epithelial morphology

Madin-Darby canine kidney (MDCK) cells embedded in extracellular matrix gels form cysts that broadly recapitulate the morphological features of the renal collecting ducts from which they are derived (46, 47). Development of a normal cyst with an apical (central) lumen requires regulated polarity and mitotic spindle alignment. Genetic depletion of polarity proteins such as aPKC, Par6B, and cdc42 in mammalian epithelial cysts disrupts the apical positioning and thus the polarized epithelial morphogenesis of cysts but does not affect the apical-basal polarity per se (48, 49). The development of multiple lumens has been used to indicate defective epithelial morphogenesis (4951). To assess the role of the RIPR motif on epithelial morphogenesis, we generated MDCK cells that stably expressed PKCι-WT, PKCι-AIPA, or empty vector. PKCι-AIPA MDCK cysts developed a multicystic appearance (Fig. 4A and fig. S2), and MDCK cells expressing PKCι-AIPA produced 58% fewer cysts with normal apical lumens compared to those expressing PKCι-WT (Fig. 4B) despite comparable protein abundance (Fig. 4C). To ensure that the observed multilumen phenotype was not due to a dominant-negative effect of the overexpressed PKCι-AIPA protein, we depleted endogenous PKCι by siRNA in the cells that stably expressed siRNA-resistant complementary DNAs (cDNAs) for PKCι-WT or PKCι-AIPA. In the control MDCK cells, siRNA-PKCι treatment led to an induction of the multilumen phenotype (Fig. 4D). Expression of exogenous PKCι-WT, but not PKCι-AIPA, at a comparable abundance to that of the endogenous protein rescued the multilumen phenotype (Fig. 4, D to F). This suggests that the RIPR motif not only is necessary for efficient phosphorylation of LLGL2 but also contributes to the ability of PKCι to support a normal polarized morphology.

Fig. 4 The RIPR motif in PKCι is required for polarized growth.

(A) MDCK cells stably expressing empty vector, PKCι(WT), or PKCι(AIPA) were cultured in Matrigel to form cysts. Representative single confocal images of actin staining (white) are presented. (B) Quantification of the number of single apical lumens. (C) Western blot showing the abundance of exogenous and endogenous PKCι in the mutant cell lines. (D) MDCK cells stably expressing empty vector, PKCι(WT), or PKCι(AIPA) were reverse-transfected with siPKCι or siSCRAM, cultured in Matrigel to form cysts, stained with phalloidin and Hoechst dyes, and evaluated by confocal microscopy. (E) Quantification of the number of single apical lumens. (F) Western blot showing the abundance of exogenous and endogenous PKCι in the mutant cell lines. Scale bar, 50 μm. (B and D) At least 100 spheroids were counted per condition for each replicate. The means ± SD of at least five replicates are presented; ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. (C and F) Representative immunoblots of three independent experiments are shown.

The RIPR motif in PKCι recruits selected substrates

To investigate whether other proteins were involved in RIPR-dependent interactions and polarity, we assessed proteomes associated with PKCι-WT and PKCι-AIPA. HCT116 cells are human colonic cancer cells that are normally polarized and have detectable amounts of both LLGL2 and PKCι proteins. Proteins in immunoprecipitates of GFP-tagged PKCι-WT, GFP-tagged PKCι-AIPA, or GFP from HCT116 cells were identified by mass spectrometry. The RIPR motif–dependent binding proteins were selected on the basis of a relative coverage of more than threefold in the PKCι-WT complexes compared to PKCι-AIPA (Table 1). From this screen, we validated a series of known PKCι partners: Par3, Par6, LLGL2, myosin X, and sequestosome-1. Myosin X is an unconventional myosin with roles in filopodia formation and spindle assembly and orientation (5254), and sequestosome-1 is a multifunctional scaffold protein that binds the PB1 domain of atypical PKCs (55). An intact RIPR motif was required for the ability of PKCι to efficiently immunoprecipitate LLGL2 and myosin X, but not Par3, Par6, and sequestosome-1 (Fig. 5A). For LLGL2 and myosin X, we confirmed this pattern of interaction with reciprocal pulldown experiments (Fig. 5, B to E). GFP-LLGL2 and GFP–myosin X immunoprecipitated PKCι-WT but not PKCι-AIPA.

Table 1 Relative peptide coverage of proteins identified by mass spectrometry in PKCι-WT compared to PKCι-AIPA immunocomplexes.

Gray background: RIPR motif binders; relative coverage (no. unique peptides WT/AIPA) ≥3. White background: non-RIPR motif binders; relative coverage (no. unique peptides WT/AIPA) <3.

View this table:
Fig. 5 LLGL2 and myosin X bind to the PKCι RIPR motif and show differential binding affinities for distinct arginine mutations.

(A) GFP immunoprecipitates (GFP-IP) from HCT116 cells expressing GFP, GFP-PKCι(WT), or GFP-PKCι(AIPA) and lysates were immunoblotted for LLGL2, Par3, Par6B, myosin X (MyoX), and sequestosome-1 (Sqstm1). Representative immunoblots of three independent experiments are shown. (B and C) GFP immunoprecipitates from HCT116 cells expressing GFP–myosin X in combination with empty vector, untagged PKCι, or PKCι(AIPA) and lysates were immunoblotted for PKCι and GFP (B). Representative immunoblots of three independent experiments are shown. (C) Quantification of the amount of the coimmunoprecipitated (coIP) PKCι normalized to that in the lysates. (D and E) GFP immunoprecipitates from HCT116 cells expressing GFP-LLGL2 in combination with empty vector, untagged PKCι, or PKCι(AIPA) and lysates were immunoblotted for PKCι and GFP (D). Representative immunoblots of three independent experiments are shown. (E) Quantification of the amount of the coimmunoprecipitated PKCι normalized to that in the lysates. (F and G) GFP immunoprecipitates from HCT116 cells expressing GFP, GFP-PKCι(WT), GFP-PKCι(AIPA), GFP-PKCι(AIPR), or GFP-PKCι(RIPA) and lysates were immunoblotted with the indicated antibodies. Representative immunoblots of three independent experiments are shown. (G) Quantification of the amount of the coimmunoprecipitated proteins normalized to that in the lysates and the PKCι(WT) complex. Means ± SD are shown. ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant.

To refine the definition of the binding motif, we expressed single arginine residue mutants (AIPR and RIPA) of PKCι in HCT116 cells and blotted PKCι immunocomplexes for LLGL2 and myosin X (Fig. 5, F and G). The AIPR mutant displayed significantly reduced LLGL2 binding, indicating that Arg471 is the dominant binding determinant for LLGL2. The RIPA single-point mutant had a stronger effect on the immunoprecipitation of myosin X than that of LLGL2. The amounts of both binding partners were diminished in AIPA double mutant immunocomplexes (although myosin X was not completely absent). This differential behavior indicates that myosin X and LLGL2 have independent binding requirements and are not mutually dependent on each other, and that multiple complexes must form through this dibasic motif.

A cancer-associated loss-of-function mutation is found within the RIPR motif

The detection of a short motif in PKCι, which when mutated leads to impaired polarity through the loss of target protein binding, led us to search for somatic mutations in human cancers that corresponded to disruption of the RIPR motif. During the course of this work, 56 separate mutations of PKCι in human samples were published and collated on the COSMIC database (http://www.sanger.ac.uk/genetics/CGP/cosmic/), of which 6 were missense mutations leading to a substitution of the first Arg of the RIPR motif [1411C>T (3 patients), 1411C>A, 1412G>A, and 1412G>T]. These mutations were identified in four patients with adenocarcinoma of the intestine, one patient with esophageal carcinoma, and one patient with serous ovarian carcinoma. The most frequent mutation was 1411C>T, which leads to a substitution of Cys for the first Arg of the RIPR motif (R471C) (56, 57).

To assess the molecular relevance of the somatic RIPR>CIPR mutation, we determined whether it interfered with PKCι interactions (Fig. 6, A and B). The CIPR mutation impaired the efficient immunoprecipitation of LLGL2 and myosin X (although to a lesser extent) but did not affect that of Par3, Par6, or sequestosome-1. To further assess the effects on PKCι function related to regulation of polarity, we generated a stable MDCK population expressing PKCι-CIPR. Consistent with the role identified for this motif, the PKCι-CIPR–expressing cells generated 48% fewer cysts with single apical lumens than those expressing PKCι-WT (Fig. 6, C and D), although they expressed comparable amounts of exogenous protein (fig. S3), suggesting that the cancer-associated R>C mutation in PKCι does not support polarized behavior.

Fig. 6 Effect of the R471C mutation on protein binding and epithelial morphogenesis.

(A and B) GFP immunoprecipitates from HCT116 cells expressing GFP, GFP-PKCι(WT), or GFP-PKCι(CIPR) and lysates were immunoblotted for LLGL2, Par3, Par6B, myosin X (MyoX), and sequestosome-1 (Sqstm1) (A). Representative immunoblots of three independent experiments are shown. (B) Quantification of the amount of the coimmunoprecipitated proteins normalized to that in the lysates and the PKCι(WT) complex. The graph shows means ± SD. (C) MDCK cells stably expressing GFP, GFP-PKCι(WT), or GFP-PKCι(CIPR) were cultured in Matrigel. Representative confocal images of actin staining (white) are shown. Scale bar, 50 μm. See also fig. S3. (D) Quantification of the number of single apical lumens. At least 100 spheroids were counted per condition for each replicate. The means ± SD of five replicates are presented. ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant.

Discussion

Signaling downstream of aPKC is important for polarizing morphogenesis and is driven by the selection of aPKC-specific substrates. Atypical PKC substrates may be targeted by subcellular colocalization with aPKC as well as by substrate-specific sequences flanking phospho-acceptor sites. To date, substrate specificity determinants of aPKC (and AGC kinases more generally) have been relatively unexplored. Here, we demonstrate that a short sequence motif within PKCι (RIPR; amino acids 471 to 474) is required for interaction with LLGL2 and myosin X and the PKCι polarizing function in mammalian epithelial cells. This functionality is defective in a recurrent mutation in human cancer.

The double arginine motif (RXXR) identified as a critical contact within the PKCι crystal lattice is conserved in several other protein kinases including PKCζ, PKCβ, Akt1 and Akt2, and PKN1 and PKN2, raising the possibility that these kinases also use this motif to recruit substrates. Here, we demonstrated that the RIPR motif and the EGFR subdomain X used for trans-activation are structurally but not functionally equivalent. We revealed that the PKCι substrate LLGL2 was not phosphorylated by a double arginine mutant (PKCι-AIPA) in cells or in vitro. However, the PKCι-AIPA mutant phosphorylates short peptide substrates with kinetics comparable to PKCι-WT. This suggests that the RIPR motif functions as a selective substrate recognition site, reminiscent of the MAPK CD domain (36), rather than an allosteric regulator of catalysis as reported for short C-terminal peptides of classical PKCs (58). It is likely therefore that the RIPR motif plays an analogous role in helping to recruit selective PKCι substrates.

Using the 3D cyst growth of MDCK cells in Matrigel as a model for polarized morphogenesis, we demonstrated that stable cell lines ectopically expressing PKCι mutated at the first or both of the arginines of the RIPR motif have a multilumen phenotype, demonstrating an important physiological function of the RIPR motif in controlling the recruitment of proteins required for normal epithelial morphogenesis.

Comparison of the proteomes of PKCι-AIPA and PKCι-WT immunocomplexes indicated that other PKCι substrates, in addition to LLGL2, rely on the RIPR motif for binding. The dependence on the RIPR motif for interaction was validated by reciprocal immunoprecipitation experiments for LLGL2 and myosin X, another PKCι binding protein. Further interrogation of this docking site revealed that Arg471 and Arg474 were key determinants of the interaction with LLGL2 and myosin X, respectively. The LLGL-PKCι complex binding appears to be independent of the myosin X–PKCι complex binding and vice versa.

Both LLGL2 and myosin X have been previously characterized as PKCι-interacting proteins (27, 59) and have since been reported as playing a key role in the polarized function of epithelial cells (52, 60). Myosin X is an unconventional myosin that can move on actin filaments (61), induces the formation of filopodia (62, 63), binds to integrins (53), and is required for normal spindle assembly and orientation (54, 64, 65). In MDCK cells, altered abundance of LLGL2 protein (66) or depletion of myosin X leads to defective lumen formation (52).

Loss of polarity and spindle misorientation are both associated with human malignancy (13, 67, 68). Genetic deletions of the polarity proteins SRB, Par4 (also known as LKB1), and Par3 cooperate with oncogenes to promote tumor formation in mice (15, 17, 18). Recurrent mutations causing a substitution of the first arginine of the RIPR motif have been identified, the most prevalent of which is 1411C>T, which leads to an R471C substitution. These findings implicate a mutation associated with human cancer in aberrant polarity control.

Although LGL was the first tumor suppressor gene to be described in the fly (69), its role in human cancer is not well defined. LLGL1 and LLGL2 have predominantly been reported as tumor suppressor proteins (14, 7072), but this is incongruent with gene expression arrays of control-matched cancer samples that show overexpression of LLGL2 in epithelial malignancies (EMBL-EBI database; http://www.ebi.ac.uk). Phosphorylation of LLGL2 by aPKC causes it to translocate from the cell cortex to the cytoplasm (25, 29). Disruption of this PKCι-LLGL2 interaction may either neutralize a tumor suppressor function or facilitate an oncogenic one.

In addition to the oncogenic role of PKCι, for which gene amplification leads to a growth advantage in malignant cells, the data presented here suggest that PKCι also has tumor suppressor functions. The identified mutation does not affect the intrinsic catalytic activity of the kinase but disrupts regulation of a pathway output, presumably unbalancing outputs to favor proliferative or invasive behaviors. Evidently, our understanding of mutations revealed by whole-genome sequencing requires a thorough understanding of the structure-function relationships of the mutated protein if we are to draw meaningful conclusions.

In summary, we have identified a cancer-associated mutation of PKCι that disrupts binding to the aPKC substrate LLGL2 and leads to abnormal epithelial morphogenesis. This finding has implications for our understanding of aPKC and polarity complexes in cancer.

Materials and Methods

Cell culture and 3D cystogenesis assay

MDCK and HEK293 cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin. Isoenzyme and SNIP analysis was performed in MDCK cells to authenticate their origin. HCT116 cells were cultured in McCoy’s 5A medium supplemented with 25 mM Hepes, 10% FBS, and penicillin-streptomycin. The 3D culture of MDCK cells in Matrigel was performed as previously described (46). In brief, cells in log-phase growth were trypsinized and resuspended in standard medium supplemented with 2% low growth factor Matrigel (BD) at 2 × 104 cells/ml. Each well of an eight-well chamber slide (BD) was precoated with 30 μl of 100% Matrigel, to which 400 μl of the cell suspension was added. Medium–2% Matrigel was changed on alternate days for 5 days.

Plasmids

For mammalian studies, human PKCι cDNA (a gift from T. Biden) was subcloned into pEGFP-C1 vector (Clontech) as full-length or kinase domain PKCι (residues 239 to 587), incorporating a 5′ Myc-tag sequence and using 5′ Sal I and 3′ Bam HI restriction sites. For insect cell expression, the kinase domain of PKCι was inserted into a pBacPAK-His3 vector (Clontech) using Bam HI and Xho I restriction sites. The vector was modified to have a glutathione S-transferase (GST) followed by a 3C protease cleavage site upstream of the kinase domain. The Entrez Nucleotide accession number of PKCι is NM_002740.5. The cDNA contains two start codons at base pairs 1 to 3 and base pairs 28 to 30, with the second methionine denoted as the first amino acid of the protein. GFP-PKCι (full-length and kinase domain) mutants were generated with the QuikChange system. Membrane targeting of GFP-tagged PKCι was achieved by fusing the myristoylation sequence from Lyn kinase to the N terminus of GFP (MGCIKSKGKD) in GFP-PKCι expression constructs (43).

Transfection

Transient reverse transfections of cDNA for HEK293 and MDCK cells were performed on poly-l-lysine–precoated plates with Lipofectamine 2000 as per the manufacturer’s instructions (Invitrogen). Cotransfections of LLGL2 and PKCι were at a 5:1 DNA ratio. HCT116 cells were transfected 24 hours after plating using FuGENE HD as per the manufacturer’s instructions (Roche). siRNA reverse transfections of MDCK and HCT116 were performed with Lullaby and poly-l-lysine–precoated plates. In a six-well plate, 25 nM siRNA and 10 μl of Lullaby were mixed together in Opti-MEM, incubated for 20 min, and added to 2 × 105 to 3 × 105 cells. Medium was refreshed at 20 hours, and typically, further manipulation took place 72 hours after transfection. siRNA reverse transfections of MDCK were performed using Lullaby (Oz Biosciences) and poly-l-lysine–precoated plates. In a six-well plate, 37.5 nM siRNA and 10 μl of Lullaby were mixed together in Opti-MEM, incubated for 20 min, and added to 2 × 105 to 3 × 105 cells. After 24 hours, cells were transferred into Matrigel for cytogenesis assays or onto six-well plates for assessment of knockdown efficiency. Knockdown efficiency was analyzed 72 hours after transfection by Western blot and immunostaining of PKCι compared to the α-tubulin loading control. Target sequences were, for canine PKCι, 5′-AGTTCTGTTGGTGCGATTA-3′ (cfPKCι) and, for scrambled control, 5′- AATGAGTGAGTAGTCTTTGCT-3′ (cfSCRAM).

Immunoblotting

For immunoblotting, lysates or immunoprecipitates were resolved by SDS–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes. The membranes were blocked in 3% BSA–TBST (bovine serum albumin and tris containing 0.1% Tween 20) and probed with primary antibodies as indicated. The primary antibodies used in this study were phospho-LLGL1/2 (S650/S654) (Abnova), LLGL2 (Abcam), GFP (Santa Cruz Biotechnology), PKCι/λ (BD Transduction Laboratories), Par3 (Millipore), Par6B (H-64 Santa Cruz Biotechnology), sequestosome-1/p62 (BD Transduction Laboratories), myosin X (SDIX), and phospho-(Ser)-PKC substrate (Cell Signaling). After incubation with appropriate secondary antibodies, bands were visualized by enhanced chemiluminescence (ECL, Amersham). Horseradish peroxidase–conjugated secondary antibodies were obtained from Sigma. The quantification of the bands was performed using the Gel function on ImageJ version 1.40g.

Immunoprecipitation

Immunoprecipitation of GFP-tagged proteins was performed with GFP-TrapM magnetic beads (Chromotek). Thirty-six hours after transfection, HCT116 cells were washed twice in cold Dulbecco’s phosphate-buffered saline (PBS) and then extracted from the dish with coimmunoprecipitation lysis buffer [1% Triton X-100, 20 mM tris (pH 8), 130 mM NaCl, 1 mM dithiothreitol (DTT), 10 mM NaF, protease inhibitors (Complete, Roche), and phosphatase inhibitors (Calbiochem, II and IV)]. The lysate was incubated on ice for 10 min and then spun at 13,200g for 10 min at 4°C. The supernatant was transferred to a precooled Eppendorf tube, from which a small aliquot was retained for analysis by Western blot. The GFP-TrapM beads and control beads were washed twice in wash buffer [1% Triton X-100, 20 mM tris (pH 8), 200 mM NaCl, 1 mM DTT, and protease inhibitor]. Lysates were precleared with 20 μl (original bead volume) of control beads for 1 hour at 4°C, followed by incubation with 20 μl of GFP-TrapM beads for 90 min at 4°C. After five washes with wash buffer, the beads were resuspended in 2× LDS loading buffer.

Immunofluorescence microscopy

MDCK cells grown in Matrigel on chamber slides were fixed with 2% formaldehyde in PBS, washed in PBS, and then permeabilized with 0.5% Triton X-100 in PBS. After an additional wash, cells were directly stained with phalloidin–Alexa Fluor 546 (1:200; Invitrogen) and Hoechst dyes (1:5000; Invitrogen). After a quick rinse with PBS, cultures were mounted with ProLong Gold hard-set mounting medium (Invitrogen). MDCK cysts were visualized with a confocal microscope (Zeiss 510). Images were acquired with 25× and 40× oil objectives. Typically, the images were taken of the middle of a cyst where the lumen is its largest. To enhance image quality, the following parameters were used: 1 airy unit pinhole, low-speed imaging, averaging image ×8, and a pixel resolution of at least 512 × 512. For apical lumen assessment, actin staining of MDCK cysts was visualized under a confocal microscope (Zeiss 510). The middle of a cyst in the z plane was identified, and the following criteria were applied to determine whether the cyst has a “predominantly single lumen”: (i) there must be a clear continuous F-actin–defined lumen; (ii) the luminal actin staining must be more intense than the basal (outside) staining; (iii) the unidimensional measurement of this lumen must be at least one-third of the cyst diameter; (iv) the unidimensional measurement of the lumen must be at least twice the size of any other luminal structure.

Mass spectrometry

Mass spectrometry was performed by the Protein Analysis and Proteomics Laboratory at CRUK London Research Institute. Coimmunoprecipitation samples were separated by mass on a polyacrylamide gel (Invitrogen), and protein was stained with GelCode Blue (Thermo Scientific). Single differential bands or whole lanes were cut from the gel and subjected to analysis. Polyacrylamide gel slices (1 to 2 mm) containing the purified proteins were prepared for mass spectrometric analysis with the Janus liquid handling system (PerkinElmer, UK) followed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) and ultraperformance liquid chromatography (UPLC). MS/MS data were validated with the Scaffold program (Proteome Software Inc.). All data were additionally interrogated manually. After identification of peptides in individual samples, a subtractive approach was used to identify proteins that were specifically purified in the PKCι or PKCι-AIPA pulldowns and not in the control pulldowns. Data were then consolidated by identifying proteins that were both reproducible in two separate experiments and for which more than two separate unique peptides were identified.

Recombinant protein expression and purification

Viruses used for infection were obtained with standard protocol (Oxford Expression Technology UK). For PKCι kinase domain purification, viruses were used to infect Hi5 cells with a multiplicity of infection of 2. Cells were grown in shaker flasks in SFIII medium (Life Technologies) and gentamicin (10 mg/ml; Life Technologies). Cells were co-infected with PDK-1 virus at a multiplicity of infection of 1 to increase the phosphorylation state of PKCι to the fully active form. Cells were harvested 72 hours after infection and resuspended in lysis buffer {20 mM Hepes (Sigma) (pH 7.4), 150 mM NaCl (Sigma), 10 mM benzamidine (Sigma), 0.2 mM AEBSF [4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride] (Melford Laboratories), 1 mM EDTA (Sigma), and 1 mM DTT (Melford Laboratories)}. All purification steps were done at 4°C or on ice. Cells were lysed by sonication and spun down at 30,000g for 30 min. PKCι was purified with glutathione–Sepharose 4B beads (Amersham Biosciences), followed by removal of the GST affinity tag with GST-3C protease (PreScission Protease, Amersham Bioscience) and ion-exchange chromatography (Hi-Trap Q column, GE Healthcare). The protein was dialyzed into the final buffer [25 mM tris (pH 8.0), 150 mM NaCl, 1 mM EDTA, and 1 mM DTT] and concentrated before use.

Kinase assay

Kinase assays were performed as previously described (73). The final reaction mix contained 200 μM ATP, 10 mM MgCl2, PKCι or PKCι-AIPA (2.5 μg/ml), and substrate (250 μg/ml). A master mix containing ATP, magnesium, and substrate was spiked with 0.25 μl of [γ-32P]ATP (~0.125 μCi). Thirty microliters of this mix was added to 10 μl of the enzyme to start the reaction and incubated at 30°C with shaking (Thermoshaker) for 10 min. This procedure was applied to the different samples sequentially, the reaction mix was then spotted onto precut p81 filter paper tabs, and the reaction was terminated by washing in 30% acetic acid. The p81 tabs were placed in a scintillation vial, and incorporated γ-32P was measured over 1 min in a Beckman Scintillation LS6000IC machine. A known volume of the spiked master mix was placed directly onto the p81 tab, and scintillations were counted so that the specific activity of ATP in the mix could be calculated.

In vitro kinetic assay

The ADP (adenosine diphosphate) Quest Kit (DiscoveRx) was used to determine the catalytic constant (kcatapp) and Michaelis-Menten constant (Kmapp) for the wild-type and mutant PKCι. The assay uses a coupled reaction to convert ADP to a product that has a fluorescence excitation at 530 nm and emission at 590 nm. A synthetic PKCε pseudosubstrate peptide (ERMRPRKQGSVRRRV) was used as substrate. For the kcatapp and Kmapp reactions, the ATP concentration ranged from 0 to 200 μM, and the synthetic peptide substrate was kept constant at 200 μM to avoid product inhibition. The reactions were measured every 2 min for 30 min in a 384-well plate with a Safire2 plate reader (Tecan).

Statistical analysis

All statistical analysis was performed with Prism (GraphPad Prism, version 5.0d, GraphPad Software). Statistical differences between two groups were assessed with Student’s t test, and for three or more groups, one-way analysis of variance (ANOVA) with Newman-Keuls correction for multiple comparisons was used. A significance threshold of P < 0.05 was used. The kinetic constants of the ADP Quest assay were determined by fitting the data to the Michaelis-Menten equation.

Supplementary Materials

www.sciencesignaling.org/cgi/content/full/6/293/ra82/DC1

Fig. S1. Kinetic studies of the PKCι suggest a role in protein-protein binding rather than trans-activation.

Fig. S2. The RIPR motif in PKCι alters the pattern of phalloidin staining in MDCK cysts.

Fig. S3. Western blot showing exogenous and endogenous PKCι in the MDCK cell lines used in Fig. 6.

References and Notes

Acknowledgments: We thank M. Skehel for assistance with mass spectrometry and subsequent data analysis, R. Mitter for help with mutational databases, G. Kelly for statistical advice, C. Swanton for useful discussions, and N. Brownlow for critical reading of the manuscript. We are grateful to Cancer Research Technology for gifting the CRT0066854 tool compound. Funding: We thank Cancer Research UK (CRUK) for funding this work. M.L. was funded by a CRUK Clinical Research Training Fellowship and by the National Institute for Health Research/Royal Marsden Hospital Biomedical Research Center. M.S.-G. was funded by the Spanish Ministry of Education, Programa Nacional de Movilidad de Recursos Humanos del Plan Nacional de I-D+i 2008-2011. Author contributions: M.L. and M.S.-G. designed and carried out all the experiments unless otherwise stated; this included LLGL2 phosphorylation assays in cells, in vitro kinase assays, polarized morphogenesis assay, and PKCι interaction screen, cloning, and microscopy. E.S. purified LLGL2 and carried out the IMAP (immobilized metal ion affinity fluorescence polarization)–based kinetic assays of PKCι and PKCι-AIPA. Y.Z. helped perform the polarized morphogenesis assay. S.K. purified the LLGL2 protein and assembled and crystallized the CRT0066854 inhibitor–PKCιk complex. A.P. collected the x-ray crystallography data. P.K. purified LLGL2 and assessed its phosphorylation by PKCι. P.R. helped perform the PKCι protein interaction screen. C.R. assisted with the characterization of the PKCι mutants in cells. A.C. contributed to the experimental design. P.J.P. and N.Q.M. planned the project and designed the experiments. M.L., M.S.-G., N.Q.M., and P.J.P. prepared the figures and wrote the paper. Competing interests: P.J.P. advises on an atypical PKC inhibitor drug development program. The other authors declare that they have no competing interests. Data and materials availability: The raw data from the mass spectrometry screen have been deposited at PeptideAtlas with an identifier of PASS00310 and a data set tag of MDL2004068. The myosin X plasmid requires a material transfer agreement (MTA) from J. Ivaska at VTT Medical Biotechnology and University of Turku (Turku, Finland). The LLGL2 plasmid requires an MTA from Mount Sinai Hospital (Toronto, Ontario, Canada). The CRT0066854 compound requires an MTA from Cancer Research Technology Limited (London, UK).
View Abstract

Stay Connected to Science Signaling

Navigate This Article