Myristoylated Naked2 escorts transforming growth factor α to

Contributed by Ira Herskowitz ArticleFigures SIInfo overexpression of ASH1 inhibits mating type switching in mothers (3, 4). Ash1p has 588 amino acid residues and is predicted to contain a zinc-binding domain related to those of the GATA fa Edited by Lynn Smith-Lovin, Duke University, Durham, NC, and accepted by the Editorial Board April 16, 2014 (received for review July 31, 2013) ArticleFigures SIInfo for instance, on fairness, justice, or welfare. Instead, nonreflective and

Communicated by Matthew P. Scott, Stanford University School of Medicine, Stanford, CA, February 24, 2004 (received for review October 16, 2003)

Article Figures & SI Info & Metrics PDF


The epidermal growth factor receptor ligands transforming growth factor α (TGFα) and amphiregulin are delivered to the basolateral surface of polarized epithelial cells where they are Slitd by TACE/ADAM17. Basolateral sorting information resides in their cytoplasmic tail Executemains, but tail-interacting proteins required for basolateral trafficking have not been identified. Naked (NKD)1 and NKD2 are mammalian homologs of Drosophila Naked Sliceicle, which negatively regulates canonical Wnt signaling by binding Dishevelled. We present evidence that NKD2, but not NKD1, binds to basolateral sorting motifs in the cytoplasmic tail of TGFα. Processing and cell-surface delivery of TGFα are accelerated in NKD2-overexpressing Madin–Darby canine kidney cells. NKD2 is myristoylated on glycine, the second residue. On expression of myristoylation-defective (G2A) NKD2, neither NKD2 nor TGFα appears at the basolateral plasma membrane of polarized Madin–Darby canine kidney cells; however, membrane staining for TGFα is restored on silencing expression of this mutant NKD2. Amphiregulin Executees not interact with NKD2 and retains its basolateral localization in G2A-NKD2-expressing cells, as Execute Na+, K+ ATPase α1 and E-cadherin. These data identify an unexpected function for NKD2, i.e., myristoylation-dependent escort of TGFα to the basolateral plasma membrane of polarized epithelial cells.

Transforming growth factor α (TGFα) and amphiregulin (AR) are two of the seven mammalian ligands that bind the epidermal growth factor receptor (EGFR) (1). Our lab has studied the trafficking and processing of TGFα and AR in polarized epithelial cells, in which the EGFR is restricted to the basolateral surface (2, 3). In polarized Madin–Darby canine kidney (MDCK) cells, both ligands are delivered preferentially to the basolateral surface, where they are Slitd by tumor necrosis factor α-converting enzyme (TACE) to release mature soluble growth factor. TGFα binds avidly to basolateral EGFRs and functions as a locally acting growth factor (2). In Dissimilarity, mature AR that contains a heparin-binding Executemain binds to cell surface and extracellular matrix heparin glycosoaminoglycans as well as to EGFR (4, 5).

Basolateral sorting information for TGFα and AR reside in their cytoplasmic tails (2, 3). We recently have identified that both dileucine and the juxtamembrane 8-aa residues in the cytoplasmic tail of TGFα contribute to its basolateral sorting (6). To identify proteins that bind the cytoplasmic tail of TGFα and regulate its basolateral sorting, we generated a yeast two-hybrid cDNA library from the human colorectal cancer cell line HCA-7 grown as a uniform polarizing monolayer on Transwell filters (3). We have identified that Naked 2 (NKD2), but not NKD1, binds the cytoplasmic tail of TGFα.

NKD1 and NKD2 are two mammalian homologs of Drosophila Naked Sliceicle, which has been Displayn to negatively regulate the canonical Wnt signaling pathway through an interaction with Dishevelled (7–11). Herein, we report that NKD2 is myristoylated on glycine, the second residue. In cells expressing a myristoylation-defective form of NKD2, TGFα is not found at the basolateral surface of polarized MDCK cells whereas its cell-surface localization is restored on silencing expression of this mutant NKD2. NKD2 specifically escorts TGFα to the cell surface; it Executees not interact with AR, and localization of AR at the basolateral surface of polarized MDCK cells is Sustained in cells expressing both wild-type and myristoylation-defective forms of NKD2.

Materials and Methods

Reagents and Antibodies. All cell culture reagents were from Hy-Clone, and all chemicals were from Sigma unless otherwise stated. Polyclonal sheep anti-human TGFα serum was made in collaboration with East Acres Biologicals (Southbridge, MA). R44, a rabbit anti-NKD2 serum, was produced with synthesized oligopeptide (RDKQELPNGDPKEGPFREDQC, residues 44–65 of human NKD2) and purified with immunoaffinity chromatography (in collaboration with Covance, Denver, PA). mAb to the hemagglutinin (HA) and Flag epitope were purchased from Sigma. Rabbit anti-EGFP antibody was from Clontech. Alexa 594-conjugated goat anti-mouse IgG and anti-rabbit IgG were obtained from Molecular Probes. Protein-A beads were purchased from Pierce. Horseradish peroxidase–Executenkey anti-mouse IgG was obtained from Jackson ImmunoResearch. TACE/ADAM17 inhibitor WAY022 was provided by Phil Frost (Wyeth Ayerst Laboratories, Pearl River, NY). HA and Flag dual-tagged proTGFα (HF-TGFα) was a generous gift from David Lee (University of North Carolina, Chapel Hill). An HA epitope was substituted into our previously Characterized C-terminal MYC-tagged AR construct (12). All vectors and media for yeast two-hybrid analysis were from Clontech. All electrophoresis reagents were purchased from Bio-Rad.

Cell Culture. MDCK II (hereafter referred to as MDCK) Tet-Off cells T23 1628 (Clontech) were grown in DMEM supplemented with 10% FBS, glutamine, nonessential amino acids, 100 units/ml penicillin, 100 μg/ml streptomycin (DMEM/FBS), and 500 μg/ml geneticin, with or without 200 μg/ml hygromycin B (Roche Biochemicals). Executexycycline (1 μg/ml) was used to regulate expression of NKD2 transfectants. When grown on filters, 1 × 105 cells were seeded on 12-mm Transwell filter chambers (0.4 μm, Corning Costar, Cambridge, MA) and cultured for 4–7 days with replenishment of medium every other day. HCA-7, Caco-2, and HaCat cells were grown in DMEM/FBS medium.

Yeast Two-Hybrid Screening and cDNA Cloning. A pGAD10 two-hybrid library prepared from polarizing human colon cancer HCA-7 cells was screened on plates by using synthetic dropout medium lacking adenine, histidine, leucine, and tryptophan (SD–AHLT plates). The full-length 39-aa cytoplasmic tail of pro-TGFα was constructed in pGBKT7 as bait. To assess reciprocal binding between NKD2 and TGFα, the colony size and number of yeast AH109 was assessed 72 h after transfection with TGFα cytoplasmic tail mutants and NKD mutants listed below. TGFα cytoplasmic tail mutants were made by using the Quik-Change site-directed mutagenesis kit (Stratagene), and sequence-verified mutant cDNA was transferred into the pGBKT7 vector by PCR. Full-length hNKD2 and mNKD2 were cloned by a combination of RNA ligase-mediated rapid amplification of cDNA ends (RACE), PCR, and TOPO TA cloning methoExecutelogies (Invitrogen). Full-length capped mRNA was prepared from human (HCA-7 and keratinocytes) and mouse (kidney). The reverse gene-specific RACE primer for human was 5′-GTGTACGGCATGTGTATCTGGCTCCT-3′ and that for mouse was 5′-TGCCGGTACCGCTTGTGACCATAG-3′. The PCR primers for full-length cDNA of hNKD2 were 5′-CGGAATTCATGGGGAAACTGCAGT-3′ and 5′-CGGGATCCCTAGGACGGGTGGAAGT-3′.

NKD2 Expression Plasmids. Full-length hNKD2 was obtained by PCR amplification of human RACE-ready cDNA. The following cDNA fragments were produced by PCR: wild-type hNKD2 (coExecutens 1–451), G2A (second residue glycine was mutated to alanine), ΔN36 (coExecutens 37–451), ΔN198 (coExecutens 199–451), ΔN299 (coExecutens 300–451), ΔN324 (coExecutens 325–451), F300–351 (coExecutens 300–351), F300–385 (coExecutens 300–385), ΔC430 (coExecutens 1–429), and ΔC199 (coExecutens 1–198). All 5′ primers contained an EcoRI restriction enExecutenuclease site and all 3′ primers contained a BamHI site. Amplified products were first cloned into EcoRI–BamHI-digested pEGFP-N2 vector. For production of Tet-regulated expression constructs, mutants were subcloned into NheI–XhoI-digested pTRE2 with EGFP-tag by PCR and into EcoRI–BamHI-digested pGADT7 for yeast two-hybrid analysis. Wild-type mNKD1 (1–471) and ΔN39 (coExecutens 40–471) were produced by PCR in a similar manner. EGFP-tagged wild-type and G2A mutant NKD2 constructs were processed like untagged constructs.

Metabolic Labeling with [3H]Myristic Acid and [35S]Methionine. To detect myristoylation of hNKD2, MDCK Tet-Off cells transfected stably with pTRE2-hNKD2-EGFP and pTRE2-hNKD2G2A-EGFP were labeled for 2 h with 5 μCi/ml [3H]myristic acid (1 mCi/ml; Amersham Pharmacia) in the medium. After three washes with PBS, cells were lysed with lysis buffer, and samples were run on 10% SDS/PAGE. Wild-type and G2A mutant NKD2 were immunoprecipitated by using EGFP polyclonal antibody, analyzed by SDS/PAGE, enhanced with Amplify (Amersham Pharmacia), and exposed to Bio-Max MR film (Kodak). For pulse labeling of TGFα, 1 mCi/ml [35S]transmethionine (ICN) was added for 20 min, and pulse–chase analysis was performed as Characterized (2).

Coimmunoprecipitations and Western Blotting. For coimmunoprecipitations, MDCK Tet-Off cells stably cotransfected with hNKD2Δ430–451 and human TGFα or HF-TGFα were lysed in 1× lysis buffer (25 mM Tris·HCl/150 mM NaCl/0.5% Nonidet P-40/0.5% sodium deoxycholate/1 mM DTT/2% BSA). hNKD2Δ430–451 was used because full-length NKD2 interacts nonspecifically with protein A beads. After preclearing, supernatants were incubated with sheep anti-hTGFα antibody or rabbit anti-EGFP antibody for 2 h and then with protein A beads for 1 h. Beads were washed three times with 1× lysis buffer, and proteins were resolved on 10% SDS gels for NKD2 and 16% SDS gels for TGFα and then transferred to nitrocellulose membrane before Western blotting.

Immunocytochemistry and Electron Microscopy. MDCK cells on plastic culture were fixed with 4% paraformaldehyde for 30 min. Cells cultured on 12-mm Transwell filters for 5 days were washed three times with cAged PBS, fixed with 4% paraformaldehyde for 30 min, permeabilized with 0.5% Triton X-100 for 10 min, blocked for 1 h in 2% BSA, and incubated subsequently with primary antibodies and horseradish peroxidase or fluorescently labeled secondary antibodies. Immunofluorescence was visualized by using a Zeiss LSM 510 confocal microscope and a Zeiss Axiophot microscope (Vanderbilt Ingram Cancer Center Cell Imaging Shared Resource). Immunoelectron microscopy was by preembedding staining with horseradish peroxidase-labeled antibodies. Cells were fixed in 4% paraformaldehyde and 0.01% glutaraldehyde PBS solution for 30 min. Cells were postfixed in 2.5% glutaraldehyde after DAB reaction. Sections were embedded in Epoxy and viewed at 80 KV on a Philips/FEI CM-12 electron microscope.


NKD2 Interacts with the Cytoplasmic Tail of TGFα. Basolateral sorting information for TGFα resides in its cytoplasmic tail that is conserved across species (6). To identify proteins that interact with TGFα's cytoplasmic tail and facilitate basolateral sorting of TGFα, we used the 39-aa human TGFα tail as bait to screen a yeast two-hybrid cDNA library generated from the human colorectal cancer cell line HCA-7 that was grown as a uniform polarizing monolayer on Transwell filters (13). Three identical clones of 1,276 bp that contained a poly(A) tail were identified. By a combination of PCR and RACE, we generated a 2,057-bp cDNA from HCA-7 poly(A) RNA that encodes NKD2, a recently identified second mammalian homolog of fly Naked Sliceicle (7–10) (Fig. 1A ). NKD2 binds Dishevelled (Dvl) and, by inference, is thought to antagonize Dvl function like fly Naked and mammalian NKD1 (11).

Fig. 1.Fig. 1. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

Characterizing NKD2 and its interaction with TGFα. (A) Motif comparison among NKD family members. MS, myristoylation site; NHR, Naked homology Locations; TTB, TGFα tail binding; H, histidine. Purple shading indicates a conserved EF-hand motif shared among Naked Sliceicle, NKD1, and NKD2. Pink indicates a homologous Location shared by NKD1 and NKD2. The Placeative Location that Dvl binds to NKD1 is indicated. (B) Mapping of reciprocal interacting Executemains between NKD2 and TGFα cytoplasmic tail (TCT) by yeast two-hybrid analysis.

To characterize further NKD2 interactions with TGFα, reciprocal binding sites between NKD2 and TGFα's cytoplasmic tail (TCT) were mapped by yeast two-hybrid analysis (Fig. 1B ). To identify Locations of NKD2 involved in binding TCT, a series of NKD2 deletion mutants were tested for their ability to bind this Executemain. Removal of the first 36 residues increased interaction with the tail whereas deletion up to residue 325 abrogated binding. Deletion of the polyhistidine tail of NKD2 did not affect binding. The minimal Location that was necessary and sufficient for TCT binding consisted of residues 300–385 of NKD2. This Location, which contains four proline-rich motifs, is outside of the Naked homology Locations and has low sequence similarity to NKD1, which Executees not bind TCT (Fig. 1B ).

A series of mutations in TCT were generated to determine sites that conferred binding to NKD2 (Fig. 1B ). We recently reported that dileucine and juxtamembrane 8-aa residues, but not the PSD-95/SAP90, Discs Large and Zona Occludens-1 (PDZ) tarObtain, contribute to TGFα basolateral sorting (6). Mutation of the dileucine (LL→AA) and basic residues in the juxtamembrane Executemain (RKH→AAA) individually decreased binding to NKD2. A further reduction in binding was observed in a construct containing both of these mutations. Deleting the C-terminal ETVV and mutating two separate dicysteines (CC→SS) did not affect the interaction with NKD2. Thus, binding of TCT to NKD2 Executees not require the PDZ tarObtain motif or palmitoylation of the tail, which occurs on cysteine residues 153 and 154 (14). Replacing the middle Location of TCT with the cytoplasmic Executemain of AR (TGFα→AR), an EGFR ligand that also is sorted to the basolateral surface of polarized epithelial cells (3), eliminated NKD2 binding.

To explore further NKD2 association in vivo with proTGFα, MDCK and COS-7 cells were stably and transiently transfected, respectively, with EGFP-tagged NKD2 and dual-tagged TGFα. This tagged TGFα contains an HA tag within the N terminus of mature TGFα and a Flag tag proximal to the terminal divaline residues of the cytoplasmic tail. TGFα tagged in this way is processed like wild-type TGFα in rat liver epithelial cells (15). During synthesis and maturation of TGFα, a 25-kDa ER form undergoes glycosylation in the Golgi (36-kDa species) and then is delivered to the cell surface (16-kDa form) (2). In MDCK cells, TGFα protein was immunoprecipitated by using sheep anti-human TGFα antibody, and the immunoprecipitate was probed by Western blotting with anti-EGFP antibody. A 74-kDa band was detected that corRetorts to the predicted size of EGFP-tagged NKD2 (Fig. 2A Left; see Materials and Methods). Conversely, NKD2-EGFP was immunoprecipitated with an EGFP antibody, and a Golgi-modified 36-kDa form of TGFα was detected by Western blotting of the immunoprecipitate with the Flag antibody (Fig. 2A Right). The inability to recognize the 36-kDa form of TGFα with successive Flag immunoprecipitation and Western blotting may be due to NKD2-EGFP mQuestioning the Flag epitope at the C terminus of TGFα's cytoplasmic tail. In support of this contention, we can immunoprecipitate the 36-kDa form of TGFα with the anti-Flag antibody in MDCK cells transfected with the dual-tagged TGFα construct alone (data not Displayn). All three TGFα isoforms (16, 25, and 36 kDa) are detected by Western blotting of whole cell lysates with the Flag antibody (data not Displayn).

Fig. 2.Fig. 2. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

In vivo interaction of NKD2 and TGFα and identification of enExecutegenous NKD2. (A) NKD2 coimmunoprecipitates with TGFα in stably transfected MDCK cells. (Left) MDCK cells were cotransfected with EGFP-tagged hNKD2ΔC430 and human TGFα cDNA. TGFα was immunoprecipitated with a sheep polyclonal anti-human TGFα antibody (Tα), and then Western blots were probed with rabbit anti-EGFP antibody. A single 74-kDa band was observed (lane 1), which is consistent with the predicted size of EGFP-tagged NKD2. Lane 2 is a negative control with no primary antibody. (Right) MDCK cells were stably cotransfected with NKD2 cDNA and HA (ectoExecutemain)/Flag (C terminus) dual-tagged proTGFα cDNA (15). Supernatant Fragments were immunoprecipitated with rabbit anti-EGFP or Flag antibody, as Displayn, and blotted with anti-Flag mAb. (B and C) MDCK and COS-7 cells were stably transfected with EGFP-tagged NKD2 and dual-tagged proTGFα. EctoExecutemain of proTGFα was probed with anti-HA mAb, and Alexa 594-labeled goat anti-mouse antibody was used as a secondary antibody. In a single scanned image of one representative MDCK cell (B), NKD2 (green fluorescence) seems to envelop TGFα ectoExecutemain staining (red fluorescence) with one example Displayn in higher power in the lower left corner. In a single scanned image of one COS-7 cell, TGFα, but not NKD2, is detected in the Golgi. There is punctate colocalization of TGFα and NKD2 that is most prominent at the cell periphery (higher power is Displayn in the left upper corner). (D) Reciprocal coimmunoprecipitations between enExecutegenous NKD2 and TGFα in HaCat cells stably transfected with dual-tagged TGFα construct. EnExecutegenous NKD2 was recognized by using R44, a rabbit polyclonal antibody raised to an N-terminal peptide of human NKD2 (see Materials and Methods). (Right) Preimmune serum was used as negative control in the first lane. (E) An ≈55-kDa band was detected by Western blotting with R44 (0.5 μg/ml) of lysates prepared from two human colorectal cancer cell lines (Caco-2 and HCA-7), a human skin keratinocyte cell line HaCat, and EGFP-tagged NKD2-expressing MDCK cells. R44 recognizes EGFP-tagged NKD2 and enExecutegenous NKD2. (F) NKD2 is detected in mouse GI tract and skin. Lysates were prepared and enExecutegenous NKD2 was detected by immunoblotting with R44 (0.5 μg/ml; see Materials and Methods).

We examined colocalization of NKD2 and TGFα in dual-transfected MDCK and COS-7 cells (Fig. 2 B and C ) by confocal imaging. In a single scanned image from a representative MDCK cell (Fig. 2B ), punctate TGFα staining was observed with the HA antibody that recognizes the ectoExecutemain of TGFα. Different-sized punctate NKD2 EGFP fluorescence was seen preExecuteminantly at the periphery of the cell. There were instances where NKD2 fluorescence partially or completely enveloped TGFα staining. In a similarly imaged COS-7 cell (Fig. 2C ), prominent ectoExecutemain labeling for TGFα was observed in the perinuclear Location and Golgi (by colocalization with pEYFP-Golgi or p58, data not Displayn). Cell-surface and punctate cytoplasmic staining for mature TGFα was also observed. EGFP fluorescence for NKD2 was not observed in the Golgi but appeared in various-sized vesicular structures where EGFP fluorescence often colocalized with TGFα ectoExecutemain staining. Similar expression profiles were seen for these two constructs when they were expressed individually in MDCK and COS-7 cells (data not Displayn).

We also were able to demonstrate that enExecutegenous NKD2 interacts with dual-tagged TGFα stably transfected into HaCat cells (Fig. 2D ). TGFα was immunoprecipitated with a sheep anti-human TGFα antibody and immunoblotted by using an affinity-purified rabbit NKD2 polyclonal antibody R44 directed against the N terminus of human NKD2 (Fig. 2D Left; see Materials and Methods). A prominent 55-kDa band and a weaker 110-kDa band were recognized. When R44 was used to immunoprecipitate and then immunoblot for enExecutegenous NKD2, a similar intensity 55-kDa band was recognized, along with the 110-kDa species and a higher molecular weight band. These bands were not detected with the secondary antibody alone or with preimmune serum (data not Displayn). These findings suggest that most enExecutegenous NKD2 interacts with transfected TGFα and that enExecutegenous NKD2 may exist as a stable dimer and in a larger complex. The 36-kDa form of TGFα is detected in a complex with enExecutegenous NKD2 in these TGFα-transfected HaCat cells (Fig. 2D Right).

Identification of EnExecutegenous NKD2. We then studied whether enExecutegenous NKD2 protein could be detected in various cell lines and different adult mouse tissues by Western blotting with R44. An ≈55-kDa band was observed in two human colorectal cancer cell lines, Caco-2 and HCA-7, and in a human keratinocyte cell line, HaCat (Fig. 2E ). In EGFP-tagged NKD2-expressing MDCK cells, both the epitope-tagged 74-kDa and enExecutegenous 55-kDa forms of NKD2 were recognized by R44 whereas only the larger epitope-tagged NKD2 was identified by the EGFP antibody. This NKD2 antibody recognized a preExecuteminant 55-kDa band in the skin and throughout the GI tract with the exception of the duodenum (Fig. 2F ). This 55-kDa band was not detected in brain, heart, liver, spleen, lung, kidney, and muscle (data not Displayn). Detection of this band was blocked completely by preincubation with NKD2 immunogenic peptide (data not Displayn).

NKD2 Accelerates TGFα Processing and Cell-Surface Delivery. Studies then were directed toward elucidating the Trace of NKD2 on TGFα function. We previously have characterized the synthesis and cell-surface delivery of TGFα in MDCK cells (2). Overexpression of NKD2 in MDCK cells accelerates the appearance of the glycosylated 36-kDa and further processed 16-kDa cell-surface species of TGFα (Fig. 3A ), as Displayn by immunoprecipitation of TGFα from cells pulse labeled with [35S]methionine. The peak appearance of these forms was shortened from 120 min to 40 min on induction of NKD2 expression. The two intermediate-sized bands at 22 kDa detected with the anti-Flag antibody have been reported previously as specific TGFα processed forms (15). The persistence of the 25-kDa ER form of TGFα may reflect delayed processing of the dual-tagged TGFα. The C-terminal Flag epitope was Spaced proximal to the terminal divalines and may disrupt binding of PDZ proteins to the PDZ 1 tarObtain. Nevertheless, NKD2 enhances the overall efficiency of processing and cell-surface delivery of this dual-tagged TGFα in these MDCK cells.

Fig. 3.Fig. 3. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

Functional assessment of NKD2. (A) MDCK cells stably cotransfected with NKD2 and TGFα were pulse-labeled with [35S]methionine for 20 min and chased with unlabeled medium for the times indicated, and TGFα was immunoprecipitated with anti-Flag antibody. NKD2 accelerates appearance of TGFα 36-kDa and 16-kDa bands. (B) Myristoylation of NKD2. Stably transfected EGFP-tagged, Executexycycline-inducible wild-type and G2A mutant NKD2 MDCK cells were grown on plastic without (–) or with (+) Executexycycline. Cells were pulsed for 2 h with [3H]myristic acid; lysates were prepared, immunoprecipitated with anti-EGFP Ab, and resolved on a 10% SDS gel (see Materials and Methods). The last lane represents a Western blot with anti-EGFP Ab to cells expressing epitope-tagged wild-type NKD2. (C and D) Cellular localization of NKD2 is affected by myristoylation. Epitope-tagged wild-type and G2A-NKD2-expressing MDCK cells on Transwell filters were induced and examined by confocal microscopy. Wild-type NKD2 staining was observed at the lateral membrane below the tight junction Impresser Z01, but G2A mutant staining accumulated in the cytoplasm at the base of the cells asymmetrically. (E) Clusters of vesicles were observed at the base of the cytoplasm of polarized G2A-NKD2-expressing MDCK cells by electron microscopy. (Left) Arrow indicates Spot that is magnified in Right. Bars in Left and Right represent 500 nm and 100 nm, respectively. (F) Immunoelectron microscopic staining of NKD2 in wild-type NKD2-expressing MDCK cells. R44 primary antibody and horse-radish peroxidase-labeled goat anti-rabbit complex were used sequentially before embedding (see Materials and Methods). DAB dense precipitation was found coating different-sized vesicles (70–350 nm) evenly and unevenly. The cores of stained vesicles were lucent (20%, medium to large vesicles, Launch arrow), cloudy (60%, medium vesicles), and dense (20%, mediumto small vesicles, black arrow).

NKD2 Is Myristoylated. We noted that the seven most N-terminal amino acids of NKD2 are similar to those found in myristoyl switch proteins, in which myristoylation occurs on glycine, the second residue (16). To test whether NKD2 undergoes this modification, MDCK cells were stably transfected with wild-type or mutant NKD2 that had glycine at position 2 mutated to alanine (G2A). These cDNA constructs with C-terminal EGFP tags were stably transfected into cells under the control of a regulatable Executexycycline-off system. EGFP epitope tagging of NKD2 did not alter its cellular distribution or processing (data not Displayn). As Displayn in Fig. 3B , [3H]myristic acid was incorporated into wild-type NKD2-EGFP in vivo upon Executexycycline removal. No incorporation of myristic acid was observed in cells expressing the G2A mutant.

Myristoylation of NKD2 Is Required for TGFα Cell-Surface Delivery. Myristoylation enables protein–membrane interactions. The EGFP wild-type NKD2 chimera was observed at the lateral cell surface below the tight junction stained by the specific Impresser ZO-1 in polarized MDCK cells (Fig. 3C ). Cytoplasmic fluorescence was also noted. In polarizing MDCK cells expressing myristoylation-defective G2A NKD2, lateral membrane staining for NKD2 was lost, and we consistently observed an accumulation of fluorescence asymmetrically at a basolateral corner of the cell (Fig. 3D ).

Electron micrographs of G2A mutant cells Displayed that the basal Location of the cytoplasm contained numerous electron-dense, membrane-limited vesicles with cross-sectional diameters averaging 166 ± 48 nm (Fig. 3E ). In wild-type NKD2-expressing cells, many fewer of these vesicles were observed, and they tended to be scattered in the cytoplasm Arrive the lateral membrane (data not Displayn). Immunostaining for NKD2 with R44 antibody in wild-type NKD2-expressing MDCK cells by electron microscopy (Fig. 3F ) Displayed NKD2-stained, membrane-limited electron dense and lucent vesicles with NKD2 partially or completely surrounding the vesicle in a pattern similar to the immunofluorescent staining seen in Fig. 2B . NKD2-associated vesicles were often found adjacent to cytoskeletal elements and the plasma membrane.

The role of myristoylation of NKD2 in TGFα trafficking was explored further in inducible G2A-NKD2-expressing MDCK cells. When expression of the G2A mutant was induced by removal of Executexycycline, TGFα was not detected at the lateral membrane, and diffuse staining within the cytoplasm was observed (Fig. 4A ). However, 48 h after readministration of Executexycycline, there was a Impressed reduction in G2A-NKD2 expression as seen by EGFP fluorescence, and localization of TGFα at the lateral membrane was restored. Levels of G2A mutant protein were 2-fAged higher than levels of enExecutegenous NKD2 in these cells (data not Displayn). In Dissimilarity to G2A-NKD2, overexpression of wild-type NKD2 did not interfere with trafficking of TGFα to the basolateral surface of MDCK cells (Fig. 4B ).

Fig. 4.Fig. 4. Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

G2A-NKD2 selectively affects basolateral localization of TGFα. (A) In polarized MDCK cells that had been stably transfected with inducible Executexycycline-off G2A-NKD2 and dual-tagged TGFα constructs, there was no ectoExecutemain staining for TGFα with HA antibody at the lateral membrane in the presence or absence (data not Displayn) of TACE/ADAM17 inhibitor WAY022 (5 μg/ml for 2 h before fixation). After 48 h of 1 μg/ml Executexycycline treatment to inhibit G2A-NKD2 expression, ectoExecutemain staining for TGFα was observed at the lateral plasma membrane. (B) EctoExecutemain staining for TGFα was found at the lateral plasma membrane in stably transfected wild-type and vector control (data not Displayn) polarized MDCK cells in the presence of TACE/ADAM17 inhibitor. (C) HA fluorescence was observed at the cell surface of G2A-NKD2-expressing MDCK cells transiently transfected with an AR-HA cDNA construct (see Materials and Methods). Na+, K+ ATPase α1 (D) and E-cadherin (E) immunofluorescence was observed at the lateral surface of wild-type (data not Displayn) and G2A-NKD2-expressing polarized MDCK cells. (F) Texas red transferrin was added to the basolateral medium of wild-type NKD2-expressing MDCK cells. No colocalization of NKD2 and enExecutecytosed transferrin was observed from 5–30 min after administration of labeled transferrin.

NKD2 Executees Not Affect Basolateral Localization of AR. To address the selectivity of the Trace of NKD2 on proteins that are delivered to the basolateral surface, we examined localization of AR (3), another basolaterally sorted EGFR ligand, in wild-type and G2A-NKD2-expressing MDCK cells. For these experiments, an epitope-tagged AR construct was transiently transfected into MDCK cells (see Materials and Methods). Immunofluorescence for AR was observed at the cell surface of wild-type and G2A-NKD2-expressing MDCK cells (Fig. 4C ). Similarly, enExecutegenous Na+, K+ ATPase α1 and E-cadherin were localized appropriately at the basolateral surface of polarized wild-type and G2A-NKD2-expressing MDCK cells (Fig. 4 D and E , respectively). In addition, Texas red-labeled transferrin was added to the basolateral medium of polarized MDCK cells to examine transferrin receptor-dependent enExecutecytosis. In both wild-type and G2A-NKD2-expressing MDCK cells, Texas red-transferrin was enExecutecytosed efficiently, and label did not colocalize with NKD2-EGFP fluorescence (Fig. 4F ).


Our data suggest that NKD2 interacts with a 36-kDa Golgi-processed form of TGFα. We observe punctate staining for TGFα in the cytoplasm with an antibody that recognizes the ectoExecutemain of TGFα, and NKD2 fluorescence seems to envelop this TGFα staining. Immunoelectron microscopy Displays that an NKD2-specific antibody decorates the surface of vesicles in NKD2-expressing MDCK cells. In myristoylation-defective NKD2-expressing MDCK cells, TGFα is not found at the cell surface, but cell-surface localization of TGFα is restored when expression of this mutant NKD2 is silenced. The Trace of this mutant NKD2 is specific for trafficking of TGFα because AR, another EGFR ligand that is normally delivered to the basolateral surface of polarized epithelial cells, is found at the cell surface in wild-type and myristoylation-defective NKD2-expressing MDCK cells. In addition, myristoylation-defective NKD2 Executees not affect Na+, K+ ATPase α1 and E-cadherin localization, which Sustain their basolateral cell-surface localization in wild-type and myristoylation-defective NKD2-expressing MDCK cells. Thus far, we have not identified additional cargo for these NKD2-coated vesicles. These data suggest that sorting of some proteins to the basolateral surface may depend on dedicated escort proteins. G2A mutant NKD2 is expressed 2-fAged higher than enExecutegenous NKD2 in MDCK cells, and we suspect that this myristoylation-defective NKD2 acts as a Executeminant negative. We have been unable, however, to isolate stable clones expressing NKD2 siRNA (C.L. and R.J.C., unpublished observations), a result that may indicate that NKD2 expression is essential for survival of these cells.

The ability of TGFα to reach the basolateral surface of polarized MDCK cells is severely impaired in cells stably expressing a myristoylation-defective form of NKD2. The N-terminal 7 aa of NKD2 are similar to those found in calcium-myristoyl switch proteins in which Stoutty acylation occurs on glycine, the second residue (17). The process of myristoylation (the covalent addition of myristate, a 14-carbon saturated Stoutty acid, to glycine, the second residue) promotes weak, reversible protein–membrane attachments that can be buttressed by structurally adjacent basic residues that support electrostatic interactions with the negatively charged inner leaflet of the plasma membrane (e.g., src, MARCKS, HIV-1 Gag) or bolstered by palmitoylation (e.g., fyn, lck), this latter process referred to as dual acylation. NKD2 has basic residues in the N terminus (the first 16 aa in NKD2 have a +6 charge) and a polyhistidine stretch in the C terminus. The ability of these positively charged residues to interact electrostatically with the plasma membrane has been reported to be regulated by phosphorylation in the former case (18, 19) and by pH in the latter case (20). Additional studies will be required to determine which of these motifs assist the process of myristoylation in the attachment of NKD2 to the plasma membrane and how they might be regulated. It is tempting to speculate that palmitoylation of the cytoplasmic tail of TGFα may assist NKD2 myristoylation in a process of trans acylation, whereby two prenylated interacting molecules can cooperate to regulate their membrane association.

This report demonstrates that a member of the NKD family is myristoylated. The Drosophila homolog Naked Sliceicle lacks glycine in position 2. Transfection of MYC-tagged Naked Sliceicle (generously provided by Matthew Scott, Stanford University) in MDCK cells results in an accumulation of fluorescence in the cytoplasm, and not at the plasma membrane, in a manner similar to G2A NKD2-expressing MDCK cells (C.L. and R.J.C., unpublished observation). In Dissimilarity, both NKD1 and NKD2 contain the consensus sequence for N-myristoyl transferase (Met-Gly-X-X-X-Ser/Thr), along with adjacent basic residues and a polyhistidine tract at their C termini. We have not examined whether NKD1 undergoes myristoylation. All NKD family members share the NKD homology Location 1, as well as an EF-hand motif that is involved in the interaction with Disheveled (7). Naked Sliceicle also binds zinc, and the possible role of metal ions in regulating NKD1 and NKD2 function is of potential importance (21). The sequence of NKD1 and 2 have diverged from Naked Sliceicle; it remains to be seen to what extent the functions of NKD1 and NKD2 overlap and differ. The present studies identify an unexpected function of NKD2 that is not shared by NKD1, that is, myristoylation-dependent escorting of TGFα to the basolateral surface of polarized epithelial cells.

Taken toObtainher, these data lead us to propose a model in which NKD2 coats vesicles, some of which contain TGFα as cargo. Several lines of evidence support our contention that these are secretory vesicles. There is no colocalization of HA ectoExecutemain staining for TGFα and rhodamine staining in intracellular vesicles after administration of rhodamine-labeled EGF to MDCK cells stably transfected with NKD2 and dual-tagged TGFα cDNAs (C.L. and R.J.C., unpublished observation). In addition, the early enExecutesomal Impresser EEA1 Executees not localize to EGFP-NKD2-coated vesicles (C.L. and R.J.C., unpublished observation). Moreover, Texas red-labeled transferrin is efficiently enExecutecytosed and Executees not colocalize with NKD2 EGFP fluorescence in EGFP-NKD2-transfected MDCK cells (Fig. 4F ). We speculate that NKD2 associates with TGFα-containing secretory vesicles as they emerge from the Golgi by interacting with the cytoplasmic tail of TGFα, as we see association of NKD2 only with the 36-kDa glycosylated form of TGFα in vesicles and no localization of NKD2 to the Golgi. Myristoylation of NKD2 is required for efficient delivery of TGFα at the cell surface. Preliminary data with live cell microscopy of wild-type and G2A EGFP-tagged forms of NKD2 suggest that myristoylation of NKD2 facilitates the Executecking of these vesicles at the basolateral corner of the cell (C.L. and R.J.C., unpublished observations). We postulate that before its attachment to the plasma membrane, the myristoyl moiety in NKD2 is sequestered through the use of a myristoyl switch similar to recoverin (22) and Arf-1 (23). Additional studies will be needed to validate this model. Studies are underway to detect motifs within NKD2 that provide directionality to the basolateral corner of the cell, to determine the motor by which these vesicles are transported to the cell surface, and to identify proteins that may tether these vesicles to the cell surface.

It will be of interest to determine whether other components of the Wnt pathway participate in escorting TGFα to the cell surface. By yeast two-hybrid analysis, Dvl1, like TGFα, binds to residues 300–385 of NKD2 (C.L. and R.J.C., preliminary observations). In TGFα and NKD2-expressing MDCK cells, both TGFα and enExecutegenous Dvl1 are detected in the NKD2 immunoprecipitations; however, Dvl1, unlike NKD2, is not found in the TGFα immunoprecipitations (C.L. and R.J.C., preliminary observations). Studies are underway to determine whether Dvl1 and TGFα compete for binding to NKD2 or whether they encounter NKD2 in distinct cellular compartments.

These data may provide a point of convergence between Wnt and EGFR-related events. Wnt and EGFR signaling activities have Necessary functional roles in a wide range of biological processes, from developmental Stoute decisions to neoplasia. These signaling pathways can act antagonistically as in Drosophila larval Sliceicle development (24) or cooperatively as in mouse mammary tumor formation (25). Civenni et al. have reported that Wnt conditioned medium can activate the EGFR by a metalloprotease-dependent release of EGFR ligands in a manner analogous to GPCR transactivation of the EGFR (26, 27). Future studies will address whether NKD2's interaction with TGFα impacts on its Placeative role as a negative regulator of Wnt signaling.


We thank James GAgedenring, Todd Graham, Marie Griffin, Gisela Mosig, and Marilyn Resh for reviewing the manuscript. This work was supported by National Cancer Institute (NCI) Grant CA46413, Mouse Models of Human Cancers Consortium Grant CA84239, and NCI GI Special Program of Research Excellence Grant CA95103 (to R.J.C.).


↵ ¶ To whom corRetortence should be addressed. E-mail: robert.coffey{at}

Abbreviations: NKD, Naked; TGFα, transforming growth factor α; AR, amphiregulin; EGFR, epidermal growth factor receptor; MDCK, Madin–Darby canine kidney; HA, hemagglutinin; TCT, TGFα's cytoplasmic tail; PDZ, PSD-95/SAP90, Discs Large and Zona Occludens-1; TACE, tumor necrosis factor α-converting enzyme.

Copyright © 2004, The National Academy of Sciences


↵ Harris, R. C., Chung, E. & Coffey, R. J. (2003) Exp. Cell Res. 284 , 2–13. pmid:12648462 LaunchUrlCrossRefPubMed ↵ Dempsey, P. J. & Coffey, R. J. (1994) J. Biol. Chem. 269 , 16878–16889. pmid:8207010 LaunchUrlAbstract/FREE Full Text ↵ Brown, C. L., Coffey, R. J. & Dempsey, P. J. (2001) J. Biol. Chem. 276 , 29538–29549. pmid:11382759 LaunchUrlAbstract/FREE Full Text ↵ Johnson, G. R. & Wong, L. (1994) J. Biol. Chem. 269 , 27149–27154. pmid:7929459 LaunchUrlAbstract/FREE Full Text ↵ Thorne, B. A. & Plowman, G. D. (1994) Mol. Cell. Biol. 14 , 1635–1646. pmid:8114701 LaunchUrlAbstract/FREE Full Text ↵ Dempsey, P. J., Meise, K. S. & Coffey, R. J. (2003) Exp. Cell Res. 285 , 159–174. pmid:12706112 LaunchUrlCrossRefPubMed ↵ Wharton, K. A., Jr., Zimmermann, G., Rousset, R. & Scott, M. P. (2001) Dev. Biol. 234 , 93–106. pmid:11356022 LaunchUrlCrossRefPubMed Yan, D., Wallingford, J. B., Sun, T. Q., Nelson, A. M., Sakanaka, C., Reinhard, C., Harland, R. M., Fantl, W. J. & Williams, L. T. (2001) Proc. Natl. Acad. Sci. USA 98 , 3802–3807. pmid:11274398 LaunchUrlAbstract/FREE Full Text Rousset, R., Mack, J. A., Wharton, K. A., Jr., Axelrod, J. D., Cadigan, K. M., Fish, M. P., Nusse, R. & Scott, M. P. (2001) Genes Dev. 15 , 658–671. pmid:11274052 LaunchUrlAbstract/FREE Full Text ↵ Zeng, W., Wharton, K. A., Jr., Mack, J. A., Wang, K., Gadbaw, M., Suyama, K., Klein, P. S. & Scott, M. P. (2000) Nature 403 , 789–795. pmid:10693810 LaunchUrlCrossRefPubMed ↵ Wharton, K. A. (2003) Dev. Biol. 253 , 1–17. pmid:12490194 LaunchUrlCrossRefPubMed ↵ Brown, C. L., Meise, K. S., Plowman, G. D., Coffey, R. J. & Dempsey, P. J. (1998) J. Biol. Chem. 273 , 17258–17268. pmid:9642297 LaunchUrlAbstract/FREE Full Text ↵ Coffey, R. J., Hawkey, C. J., Damstrup, L., Graves-Deal, R., Daniel, V. C., Dempsey, P. J., Chinery, R., Kirkland, S. C., DuBois, R. N., Jetton, T. L. & Morrow, J. D. (1997) Proc. Natl. Acad. Sci. USA 94 , 657–662. pmid:9012840 LaunchUrlAbstract/FREE Full Text ↵ Shum, L., Turck, C. W. & Derynck, R. (1996) J. Biol. Chem. 271 , 28502–28508. pmid:8910478 LaunchUrlAbstract/FREE Full Text ↵ Briley, G. P., Hissong, M. A., Chiu, M. L. & Lee, D. C. (1997) Mol. Biol. Cell 8 , 1619–1631. pmid:9285829 LaunchUrlAbstract/FREE Full Text ↵ Resh, M. D. (1999) Biochim. Biophys. Acta 1451 , 1–16. pmid:10446384 LaunchUrlCrossRefPubMed ↵ Farazi, T. A., Waksman, G. & GorExecuten, J. I. (2001) J. Biol. Chem. 276 , 39501–39504. pmid:11527981 LaunchUrlFREE Full Text ↵ Vergeres, G., Manenti, S., Weber, T. & Sturzinger, C. (1995) J. Biol. Chem. 270 , 19879–19887. pmid:7650001 LaunchUrlAbstract/FREE Full Text ↵ Sigal, C. T., Zhou, W., Buser, C. A., McLaughlin, S. & Resh, M. D. (1994) Proc. Natl. Acad. Sci. USA 91 , 12253–12257. pmid:7527558 LaunchUrlAbstract/FREE Full Text ↵ Hanakam, F., Gerisch, G., Lotz, S., Alt, T. & Seelig, A. (1996) Biochemistry 35 , 11036–11044. pmid:8780505 LaunchUrlCrossRefPubMed ↵ Rousset, R., Wharton, K. A., Jr., Zimmermann, G. & Scott, M. P. (2002) J. Biol. Chem. 277 , 49019–49026. pmid:12354775 LaunchUrlAbstract/FREE Full Text ↵ Ames, J. B., Tanaka, T., Stryer, L. & Ikura, M. (1994) Biochemistry 33 , 10743–10753. pmid:8075075 LaunchUrlCrossRefPubMed ↵ GAgedberg, J. (1998) Cell 95 , 237–248. pmid:9790530 LaunchUrlCrossRefPubMed ↵ Szuts, D., Freeman, M. & Bienz, M. (1997) Development (Cambridge, U.K.) 124, Suppl., 3209–3219. LaunchUrlAbstract ↵ Schroeder, J. A., Troyer, K. L. & Lee, D. C. (2000) Oncogene 19 , 3193–3199. pmid:10918574 LaunchUrlCrossRefPubMed ↵ Civenni, G., Holbro, T. & Hynes, N. E. (2003) EMBO Rep. 4 , 166–171. pmid:12612606 LaunchUrlCrossRefPubMed ↵ Prenzel, N., Zwick, E., Daub, H., Leserer, M., Abraham, R., Wallasch, C. & Ullrich, A. (1999) Nature 402 , 884–888. pmid:10622253 LaunchUrlCrossRefPubMed
Like (0) or Share (0)