Differential Stimulation of PKC Phosphorylation of Potassium Channels by ZIP1 and ZIP2

See allHide authors and affiliations

Science  03 Sep 1999:
Vol. 285, Issue 5433, pp. 1565-1569
DOI: 10.1126/science.285.5433.1565


Targeting of protein modification enzymes is a key biochemical step to achieve specific and effective posttranslational modifications. Two alternatively spliced ZIP1 and ZIP2 proteins are described, which bind to both Kvβ2 subunits of potassium channel and protein kinase C (PKC) ζ, thereby acting as a physical link in the assembly of PKCζ-ZIP-potassium channel complexes. ZIP1 and ZIP2 differentially stimulate phosphorylation of Kvβ2 by PKCζ. They also interact to form heteromultimers, which allows for a hybrid stimulatory activity to PKCζ. Finally, ZIP1 and ZIP2 coexist in the same cell type and are elevated differentially by neurotrophic factors. These results provide a mechanism for specificity and regulation of PKCζ-targeted phosphorylation.

Protein phosphorylation is important for the regulation of neuronal excitability and many other non-neuronal processes. Phosphorylation of ion channels, which could change channel expression as well as kinetic properties, is thought to be one of the key regulatory mechanisms of membrane excitability (1). The existence of signaling complexes containing both ion channels and closely associated protein kinases and phosphatases has been inferred from biochemical and electrophysiological studies (2). Such ion channel–kinase/phosphatase complexes are thought to provide the necessary macromolecular organization for signaling specificity and regulation (3, 4). Many ion channels are phosphorylated by serine-threonine kinases such as PKC and PKA (5). The mechanisms for targeting the specificity and regulating the activity of PKC or PKA to ion channel proteins are not well understood.

Potassium channels are important for controlling neuronal excitability (6). Shaker-type potassium channels have been implicated in associative memory in model systems (7). The auxiliary Kvβ subunits specifically interact with a subset of Kvα subunits, which, in some cases, results in modulatory effects (8–10). The mammalian Kvβ2 is an abundant subunit found in both excitable and nonexcitable cells, and the native neuronal Shaker-type potassium channel has an estimated stoichiometry of (Kvα)4(Kvβ)4(11).

To identify proteins that functionally interact with potassium channels, a yeast two-hybrid library made from rat hippocampal mRNA was screened against full-length Kvβ2. The interacting clones include multiple partial cDNA fragments of Kvβ2 (12). In addition, we found two cDNA fragments, B20 and B24, encoding two previously unknown Kvβ2-binding proteins. Sequence comparison of full-length cDNAs showed that B20 and B24 are identical except for a stretch of 27 residues missing in B24. Although B24 has never been reported, B20 was identical to a previously reported gene known as ZIP (PKC-zeta interacting protein) that is isolated as a protein interacting with an atypical PKCζ isozyme (13). Human homologs for this protein, known as A170 or p62, have also been identified (14). Thus, B20 and B24 were termed ZIP1 and ZIP2.

We generated primers flanking the insertion site, which allows polymerase chain reaction (PCR) to amplify DNA covering the putative splicing site, thereby producing DNA fragments of distinct sizes (15). Reverse transcription (RT)–PCR using rat cerebellum mRNA amplified two DNA fragments (Fig. 1C, lane 5). They possessed the same mobility as the PCR products obtained using cloned ZIP1 and ZIP2 as templates and corresponded individually to the coding sequences of ZIP1 and ZIP2 (Fig. 1C, lanes 2 through 4). These data provide evidence that ZIP1 and ZIP2 resulted from alternative splicing, and both transcripts are present in the brain. The two forms contain a number of protein domains suggestive of their biochemical features (Fig. 1, A and B). The NH2-terminal acidic motif is a novel domain homologous to a segment of CDC24 in yeast. They were found in diverse proteins of evolutionarily distant species ranging from bacteria to mammal. All share a characteristic sequence pattern of (Y/W)XDXXGD(L/F) (V/I). The zinc finger domain contains three CXXC motifs and a DYDL signature (16, 17). Both PEST and ubiquitin associated (UBA) domains are involved in regulation of protein turnover by either coding protein stability or directing proteins to degradation pathways. In addition, the PEST sequence is implicated in interaction with calmodulin (18).

Figure 1

(A) A schematic diagram showing the domain organization of the ZIP protein. (B) Primary structure of ZIP1 and ZIP2. Putative protein kinase C sites are marked with an asterisk. The protein domains and the 27–amino acid regions lacking in ZIP2 are highlighted. (C) RT-PCR identification of ZIP1 and ZIP2 transcripts (15). PCR products were separated on polyacrylamide gel. The DNA templates are cloned cDNAs of ZIP1 (lane 2) and ZIP2 (lane 3), a 1:1 mixture of ZIP1 and ZIP2 cDNA (lane 4), and RT-PCR products from cerebellar mRNA (lane 5). (D) Immunoblot detection of one or more ZIP proteins from tissues. Total protein extracts from the indicated tissues were prepared and 100 μg of protein from each tissue were loaded. The ZIP polypeptides were detected with rabbit anti-ZIP antibody (8, 26). In most tissues, a 69-kD polypeptide was detected. The signal could be specifically blocked with purified GST-ZIP1 protein. Lower panel, RT-PCR detection of mRNA from the indicated tissues was carried out as described in (15). The ZIP1/ZIP2 ratios are shown at the bottom. (E) Identical amounts of cerebellar protein extracts from different days of postnatal rat were prepared and separated on SDS-PAGE and followed by immunoblot detection using anti-ZIP antibody. The postnatal stages of each sample are shown on the top. The ZIP1/ZIP2 ratios are shown at the bottom.

ZIP was identified from rat brain cDNA library, but its expression was not restricted to the nervous system. A 69-kD polypeptide was detected in all tissues tested (Fig. 1D). Different from most tissues, regions in the central nervous system gave rise to two bands, 69 and 66 kD, both of which could be completely blocked with purified GST-ZIP1 fusion protein (19). The sizes of the two polypeptides are consistent with those of ZIP1 and ZIP2, although we cannot rule out that the two species result from differential posttranslational modifications. RT-PCR was employed to detect the expression and relative ratio of ZIP1 and ZIP2 transcripts. In nonexcitable tissues, ZIP1 was present in significantly higher amounts than ZIP2, whereas in regions of the central nervous system, the ZIP1/ZIP2 ratio was closer to 1 (Fig. 1D, lower panel). The expression of the ZIP proteins was also temporally regulated. In the cerebellum, an area with a high expression level of ZIP (Fig. 2C), the expression of the ZIP proteins peaked at postnatal day 13 (Fig. 1E). The two ZIP polypeptides differed both in expression level and in relative amount during postnatal development. RT-PCR also detected an increase of ZIP1 transcripts (Fig. 1E, lower panel).

Figure 2

(A) Summary of the yeast two-hybrid test. Transformants with both GAL4-DB and GAL4-TA plasmids were selected and allowed to grow in His media. Growth (+) and no growth (–) are indicated. The β-galactosidase activity is tested using X-gal, where blue was observed within the first 30 min (indicated by +++) and within the first 2 hours (++). (B) Affinity binding of GST-ZIP1 or GST-ZIP2 to the HA-Kvβ2. Soluble lysates (lane 3) from HA-Kvβ2 transfected cells were incubated with Affi-gel cross-linked with GST-ZIP1, GST-ZIP2, or control GST-PDZ3 of PSD95 (27). The unbound (lanes 4, 5, 8, and 9) and bound (lanes 6, 7, 10, and 11) materials were separated on SDS-PAGE and detected with anti-HA antibody. The samples are marked on the top of gel. (C) In situ hybridization detection of mRNAs encoding Kvβ2, ZIP1/2, and PKCζ (28). Horizontal sections of adult rat brain were hybridized with DIG-labeled antisense (left panels) and sense (right panels) probes. The various ribo-probes were generated using full-length cDNAs as indicated on the left. (D) Antisera against GST fusion proteins of Kvβ2 and ZIP1 were generated and affinity-purified. In immunoblot analyses, these two antibodies and commercial anti-PKCζ antibody detected polypeptides of Kvβ2, ZIP1/ZIP2, and PKCζ of predicted sizes. The immunoblot signals could be blocked by the corresponding antigens (29). Coimmunoprecipitation was performed using adult soluble cerebellar extracts (30). The antibody-precipitated materials were separated by SDS-PAGE followed by the immunoblot detection. Each lane is identified by antibody that was used for precipitation. The starting material, soluble cerebellar extract, was also loaded for comparison. (Right) Antibodies used in immunoblot. (Left) Molecular size standards.

The yeast two-hybrid method was employed to further test the interaction among ZIPs and various potassium channel subunits. The formation of the α-β complex of potassium channels is mediated by the T1 domain of Kv1α subunits and the conserved core regions of Kvβ−subunits (8, 10). ZIP1 and ZIP2 interacted with both Kvβ1 and Kvβ2, but not with the NH2-terminal domain of the Kv1α subunits of ShB (NShB) (Fig. 2A), which contains the Kvβ binding site. This suggests that ZIP binds to potassium channels via the Kvβ subunits. ZIP1 and ZIP2 interacted both homomerically and heteromerically (Fig. 2A). Nondetergent soluble protein complexes from rat brain containing ZIP migrated with a Stoke's radius larger than 6.1 × 10−9 m which corresponds to a 640-kD globular protein (20). The purified GST-cdc homology domain of ZIP alone behaved as an oligomer (13).

To directly demonstrate the interaction between Kvβ2 and ZIP, we tested for binding of GST-ZIPs expressed in Escherichia coli to HA-tagged Kvβ2 expressed in mammalian tissue culture cells by incubating the GST-ZIP1-Affigel or GST-ZIP2-Affigel resin with an excess amount of HA-Kvβ2 (15). While the control GST-PDZ3-Affigel resin displayed no detectable binding (lanes 6 and 10), both GST-ZIP1-Affigel and GST-ZIP2-Affigel precipitated the HA-tagged Kvβ2 (lanes 7 and 11) (Fig. 2B).

The ability of ZIP to interact with both PKCζ and Kvβ2 led us to hypothesize that ZIPs act as a link in the PKCζ-ZIP-Kvβ2 complex. Sequential horizontal sections of rat brain were probed by in situ hybridization with either sense control (right panels) or antisense (left panels) probes of Kvβ2, ZIP1, and PKCζ (Fig. 2C). Their mRNA localization was overlapping, especially among pyramidal cells in the hippocampus and Purkinje cells in the cerebellum, where transcripts for the three proteins were found to be more concentrated than in other parts of the brain.

Immunoprecipitation was carried out using detergent-solubilized protein lysate prepared from adult rat cerebellum. Antibodies specific to Kvβ2, ZIP, and PKCζ were used to precipitate protein complexes. Each antibody was capable of precipitating protein complexes containing Kvβ2, ZIP, and PKCζ. In contrast, translin, an unrelated protein abundant in the cerebellum, was present in the crude soluble lysate, but none of the antibodies was able to precipitate it. ZIP1 is thought to bind specifically to PKCζ (13). Consistent with the results, PKCβII could not be precipitated by any of the above antibodies (Fig. 2D). Although the stoichiometry of the Kvβ2-ZIP-PKCζ complexes is not known, within the soluble fraction of cerebellar extracts the 69-kD species was coimmunoprecipitated more efficiently, suggestive of a potential binding difference of the two ZIP polypeptides to Kvβ2 and PKCζ.

Complementary DNAs encoding Kvβ2 and PKCζ were expressed as HA-tagged fusion proteins in mammalian COS7 cells cultured in phosphate-free media (21). Soluble protein lysates of either PKCζ or Kvβ2-transfected COS7 cells were prepared, mixed, and precipitated with anti-HA monoclonal antibody. Activation of PKCζ in the presence of [33P]-γ-ATP induced basal phosphorylation of the Kvβ2 subunit. When purified, recombinant GST-ZIP1 was added to the reaction, the Kvβ2 phosphorylation was stimulated by more than ninefold, while GST-ZIP2 showed less than twofold stimulation (n = 5,Fig. 3A). Using myelin basic protein (MBP), a common experimental substrate of the PKCζ enzyme, similar levels of MBP phosphorylation were found in the presence or absence of GST-ZIP1, GST-ZIP2, or a control GST fusion protein, GST-PDZ3 (Fig. 3B). Thus, the ZIP-mediated stimulation of PKCζ is substrate-selective, presumably restricted to the ZIP interacting proteins such as Kvβ2. The ratios of the ZIP1 and ZIP2 messages varied significantly among different tissues, suggestive of a distinct function for ZIP1 and ZIP2. Indeed, GST-ZIP1 is significantly more potent than GST-ZIP2 in its ability to stimulate Kvβ2 phosphorylation. The time course of their activity could be best fit with two exponential time constants (Fig. 3C and legends).

Figure 3

Stimulation of PKCζ phosphorylation of the Kvβ2 subunit. (A) In vitro phosphorylation of Kvβ2. Anti-HA precipitated Kvβ2 and PKCζ were tested for phosphorylation in the presence of 1 μM purified GST proteins (21). The phosphorylated proteins were visualized by autoradiography. The protein inputs for each reaction are listed on the top of the gel. The Kvβ2 signal was quantified by phosphoimaging, normalized against Kvβ2 signal in the absence of PKCζ and GST fusion proteins, and listed at the bottom. (B) Phosphorylation of myelin basic protein (MBP) was carried out using the procedures (21). (C) ZIP-mediated differential stimulation of the Kvβ2 phosphorylation by PKCζ. The time course of Kvβ2 phosphorylation was performed as described (21) except that the reaction was carried out at 4°C in the presence of 1 μM purified GST-ZIP1 and GST-ZIP2. The data were best fit with two time constants τ1 and τ2 in the equation: A[1–exp(–t/τ1)] + B[1–exp(–t/τ2)]. For GST-ZIP1 (n = 3), A = 0.33 ± 0.040, B= 0.69 ± 0.037, τ1 = 1.3 ± 0.35 min and τ2 = 26 ± 3.9 min; for GST-ZIP2 (n = 3), A = 0.12 ± 0.037, B = 0.10 ± 0.059, τ1 = 40 ± 34 min and τ2 = 4.3 ± 3.0 min.

Under fixed amounts of Kvβ2 and PKCζ, the phosphorylation of Kvβ2 was measured with increasing concentrations of purified GST-ZIP1 or GST-ZIP2. The EC50values of GST-ZIP1 and GST-ZIP2 were comparable: 1.3 ± 0.4 μM (n = 4) and 2.1 ± 0.4 μM (n = 4), respectively. In contrast, the maximal stimulation by GST-ZIP1 was 7.4-fold higher than that by GST-ZIP2. ZIP1 and ZIP2 behaved similarly in binding to Kvβ2 and PKCζ as tested in the yeast two-hybrid system (Fig. 2A) (13). The differential stimulation could thus result from their potential different binding affinity influenced by differential posttranslational modifications of ZIP1 and ZIP2 as suggested by the coimmunoprecipitation experiments (Fig. 2D).

ZIP-mediated stimulation showed a reproducible decline phase following the peak stimulation at high concentrations of GST-ZIP. The maximal ZIP-mediated stimulation was neither enhanced nor reduced by the addition of the same or a higher concentration of a control GST-fusion protein such as GST-PDZ3 of PSD-95, indicating that the decline phase did not result from an excess of GST. Other supportive evidence includes that changes of inputs of PKCζ or Kvβ2 induced a shift of ZIP concentration for the maximal stimulation of the Kvβ2 phosphorylation (22). These data are consistent with the notion that the ZIP-mediated stimulation was achieved through the formation of a stoichiometrically defined reaction complex involving PKCζ, ZIP, and Kvβ2. Thus, instead of reaching saturation, the higher concentration of ZIP would disrupt the optimal ratio for a functional PKCζ-ZIP-Kvβ2 complex, for example, by assembly of less active forms of oligomeric ZIP.

ZIP1 and ZIP2 are two alternatively spliced transcripts with distinct ability to stimulate the phosphorylation of Kvβ2. The ratio of these two messages may be regulated in response to hormone stimulation and may thus be a way to modulate the intracellular phosphorylation level. Neurotrophins, such as nerve growth factor (NGF), are involved in regulation of synaptic plasticity and ion channel activity (23). PC12 cells respond to NGF stimulation (24), providing a model system to test the hypothesis.

The mRNAs of both ZIP1 and ZIP2 in PC12 cells were detectable in the absence of NGF. In the presence of NGF, the ZIP1/ZIP2 ratio increased from 0.2 to 1.0 within 24 hours (Fig. 4A). At the protein level, the expression of ZIP was increased by approximately 30-fold. Upon NGF stimulation, the signal for the putative 66-kD ZIP2 intensified, which was accompanied by an increase in the 69-kD putative ZIP1 polypeptide (Fig. 4B). Consistent with the changes in the ZIP1/ZIP2 mRNA ratio, the ratio of 69 kD to 66 kD increased from 0.1 in the absence of NGF to 0.8 after 72 hours of the stimulation. In contrast, at the protein level Kvβ2 showed almost no change and PKCζ showed only a modest change in response to NGF stimulation (Fig. 4B). These results show that both ZIP1 and ZIP2 were present in the same cell type, suggesting the potential for the formation of heteromultimers and regulation of stoichiometry of the ZIP1-ZIP2 heteromultimers.

Figure 4

(A) RT-PCR results using untreated and NGF-stimulated PC12 cells. PC12 cells were treated with NGF and collected at 24, 48, and 72 hours after treatment (31). Bottom: ZIP1/ZIP2 ratios of corresponding lanes. (B) Immunoblot analyses of Kvβ2, ZIP1/2, and PKCζ proteins. Identical amounts (100 μg) of total protein were loaded from untreated and NGF-stimulated PC12 cells collected after different times as indicated. Bottom: Primary antibodies used for the detection. (C) Stimulation of the Kvβ2 phosphorylation by GST-ZIP1 in the presence of fixed concentrations of GST-ZIP2 protein as indicated (n = 4). (D) A simulated curve (solid line) was created by adding the signal of GST-ZIP2 at 1 μM with the signals of GST-ZIP1 at different concentrations. The curve is plotted with the experimental data under the same conditions.

Phosphorylation of Kvβ2 was assayed with increasing concentrations of purified GST-ZIP1 supplemented with fixed concentrations of GST-ZIP2. In the presence of GST-ZIP2, the overall EC50 value gradually shifted to the lower concentration of GST-ZIP1 (Fig. 4C). For example, when a fixed amount of 1 μM GST-ZIP2 was included, the EC50 value of GST-ZIP1 reduced by approximately 10-fold, from 1.3 ± 0.4 μM (n = 4) to 0.18 ± 0.05 μM (n = 4) (Fig. 4C). GST-ZIP2, by itself, did show stimulatory activity, although at a considerably lower level. Figure 4D compares experimental data of GST-ZIP1 + 1.0 μM GST-ZIP2 with a simulated curve obtained by digital addition of the signal of 1.0 μM GST-ZIP2 to signals of GST-ZIP1 at different concentrations. Both the reduction of EC50 and the shape change of the dose response curve indicate that the stimulatory activity was not a result of linear addition of activities contributed independently by ZIP1 and ZIP2. The heteromultimeric interactions detected between ZIP1 and ZIP2 by the yeast two-hybrid test (Fig. 2A) were consistent with the notion of a functional interaction between GST-ZIP1 and GST-ZIP2.

Our results suggest that ZIP acts as a link that targets the activity of PKCζ to Kvβ2. ZIP1 and ZIP2, two alternatively spliced protein products, possess distinct activities in stimulating PKCζ phosphorylation of Kvβ2. Their ability to interact with each other to form homo- and heteromultimeric complexes provides an explanation for the synergistic stimulatory activity seen only in the presence of both ZIP1 and ZIP2 (Fig. 4C). The regulation of ZIP activity in various tissues and cell types is not limited to the differential temporal and spatial expression; the ZIP1/ZIP2 ratios are also tissue-specific. Furthermore, the ratio of ZIP1 and ZIP2 within the same cell type is dynamically regulated in response to stimulation by neurotrophic factors.

Redistribution of PKC proteins upon hormone stimulation is a key step involved in specifying and activating PKC activity (25). Our data suggest that one of downstream steps after PKC translocation may involve specific and dynamic assembly of protein complexes critical for the translation of hormonal signals into targeted covalent modifications. The identification of ZIPs and their biochemical function has provided a mechanism by which the PKC phosphorylation potential to an ion channel protein can be regulated and finely tuned transcriptionally and posttranslationally.

  • * Present address: Aurora Biosciences Corporation, San Diego, CA 92121, USA.

  • Present address: The Salk Institute for Biological Studies, San Diego, CA 92186, USA.

  • To whom correspondence should be addressed. E-mail: minli{at}


View Abstract

Navigate This Article